University of Southern California Dissertations and Theses

Effects of economic incentives on creative project-based networks: communication, collaboration and change in the American film industry, 1998-2010

(USC Thesis Other)

Effects of economic incentives on creative project-based networks: communication, collaboration and change in the American film industry, 1998-2010

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

EFFECTS OF ECONOMIC INCENTIVES ON CREATIVE PROJECT-BASED
NETWORKS: COMMUNICATION, COLLABORATION AND CHANGE IN THE
AMERICAN FILM INDUSTRY, 1998-2010

By
Nina F. O’Brien

________________________________________________

A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(COMMUNICATION)

August 2013

Copyright 2013 Nina F. O’Brien
i

ACKNOWLEDGEMENTS
I have always enjoyed the observation by David Hurst in Jane Smiley’s novella The Age
of Grief that graduate study can sometimes feel like a very large meal that one has to eat
all by oneself. I am grateful to have shared my table with family, friends and colleagues
who commiserated while I chewed my Brussels sprouts, and now share with me in the
enjoyment of dessert. Among them are my husband, Tomás Cabral, my mother, Angie
O’Brien, my extended family in Germany and Portugal, and friends from near and far, all
of whom have in one way and another challenged, nurtured and supported me throughout
my journey. I am also indebted to mentors (both official and informal) and colleagues,
whose fine minds and powerful insights have contributed to make me a better scholar
than I could ever have hoped to become without their influence. Among these, I wish to
thank especially Peter Monge, Janet Fulk, Mark Young, Patricia Riley, Larry Gross,
Thomas Valente, and the entire faculty and staff at the Annenberg School for
Communication and Journalism at the University of Southern California. For their
invaluable advice and crucial assistance on particular issues related to data collection and
methodology I also wish to thank Katya Ognyanova, Poong Oh, Ayushman Dutta, and
Christianne Lane.

ii

TABLE OF CONTENTS
ACKNOWLEDGEMENTS ................................................................................................. i
LIST OF TABLES .............................................................................................................. v
LIST OF FIGURES ........................................................................................................... vi
ABSTRACT ....................................................................................................................... ix
INTRODUCTION .............................................................................................................. 1
CHAPTER 1: THE ORGANIZATIONAL NETWORK ECOLOGY OF PROJECT-
BASED INDUSTRIES ....................................................................................................... 5
Economic Development Theory ....................................................................... 5
Project-based enterprises ................................................................................ 10
Organizational Ecology of Project-Based Industries ...................................... 17
Population Ecology ....................................................................................... 18
Community Ecology ..................................................................................... 30
Project-based industries as community ecologies ......................................... 34
Project-based Ecologies as Interorganizational Networks ............................ 37
CHAPTER 2: THE AMERICAN FILM INDUSTRY AS A PROJECT-BASED
NETWORK ECOLOGY .................................................................................................. 52
The Film Project Process ................................................................................ 52
Divestment, Diversity, Diffusion .................................................................... 54
Deregulation and Spread ................................................................................. 58
iii

Incentivizing the Film Industry ....................................................................... 60
Incentive Program Evaluation ......................................................................... 65
CHAPTER 3: HYPOTHESIS DEVELOPMENT: PROMOTING ECONOMIC
DEVELOPMENT IN PROJECT-BASED ECOLOGICAL NETWORKS ...................... 75
CHAPTER 4: RESEARCH DESIGN AND METHOD ................................................... 92
Data Sources and Measures ............................................................................ 92
Constructing the American Film Industry Project Network ......................... 95
Variables ......................................................................................................... 96
Independent Variables .................................................................................. 96
Dependent Variables ................................................................................... 101
Analyses ........................................................................................................ 102
CHAPTER 5: INCENTIVE, ECOLOGICAL AND NETWORK EFFECTS ON
ECONOMIC DEVELOPMENT OUTCOMES: STUDY RESULTS ............................ 105
Descriptive Results ....................................................................................... 105
Incentives .................................................................................................... 107
Ecological Factors ....................................................................................... 111
State Network Factors ................................................................................. 116
National Network Factors ........................................................................... 118
Tests of Hypotheses ...................................................................................... 125
Mixed Effects of Predictors ........................................................................ 127
iv

Post-Hoc Analyses ........................................................................................ 143
Collinearity ................................................................................................. 143
Outliers ........................................................................................................ 145
Post-Hoc Descriptive Results ..................................................................... 146
Post-Hoc Tests: Mixed Effects of Predictors .............................................. 161
CHAPTER 6: DISCUSSION AND CONCLUSION ..................................................... 169
Incentives ...................................................................................................... 169
Ecological Factors ......................................................................................... 174
Network Factors ............................................................................................ 179
REFERENCES ............................................................................................................... 183

v

LIST OF TABLES
Table 1: Summary of hypotheses, organized by dependent variable .............................. 126
Table 2: Summary statistics, all states except Iowa (n = 650) ........................................ 127
Table 3: Effects of incentives, ecological factors and network factors on
industrial activity, 1998 - 2010 ....................................................................................... 130
Table 4: Effects of incentives, ecological factors and network factors on
employment, 1998-2010 ................................................................................................. 136
Table 5: Effects of incentives, ecological factors and network factors on
establishments, 1998 - 2010 ............................................................................................ 140
Table 6: Dependent variable means for California, New York, and other states,
1998 - 2010 ..................................................................................................................... 145
Table 7: Percent change in intercept and coefficient for the effect of time with
and without California and New York ............................................................................ 149
Table 8: Coreness statistics with and without California and New York ....................... 161
Table 9: Summary statistics, excluding California and New York (n = 624) ................ 162
Table 10: Effects of incentives, ecological factors and network factors on
industrial activity (excluding California and New York), 1998 - 2010 .......................... 166
Table 11: Effects of incentives, ecological factors, and network factors on
employment (excluding California and New York), 1998 - 2010 .................................. 167
Table 12: Effects of incentives, ecological factors and network factors on
establishments (excluding California and New York), 1998 - 2010 .............................. 168

vi

LIST OF FIGURES
Figure 1: Feature films produced and released in the United States, 1998 - 2010 ......... 106
Figure 2: Average employment in motion picture and video production
(NAICS 5121), 1998 - 2010............................................................................................ 106
Figure 3: Average establishments in motion picture and video production
(NAICS 5121), 1998 - 2010............................................................................................ 107
Figure 4: Film industry incentive program introductions by US states,
1997 - 2010 ..................................................................................................................... 108
Figure 5: Average value of incentives for film production (in millions of US
dollars), 1998 - 2010 ....................................................................................................... 109
Figure 6: Average incentives offered by US states and average number of films
produced and released, 1998 - 2010 ............................................................................... 110
Figure 7: Average incentives offered by US states and average film
industry employment ...................................................................................................... 111
Figure 8: Average incentives offered by US states and average number of
film industry establishments ........................................................................................... 111
Figure 9: Companies in the dominant organizational population, 1998 …..…………….113
Figure 10: Companies in the dominant organizational population, 2010 ....................... 113
Figure 11: Organizational diversity by state, 1998 ......................................................... 115
Figure 12: Organizational diversity by state, 2010 ......................................................... 115
Figure 13: Number of states with an observed active network, 1998 - 2010 ................. 117
Figure 14: Average E-I values for US states, 1998 - 2010 ............................................. 119
vii

Figure 15: Aggregate relations between US states, 1998. Line weight reflects
the number of company-company connections between states ...................................... 122
Figure 16: Aggregate relations between US states, 2007. Line weight reflects
the number of company-company connections between states ...................................... 122
Figure 17: Aggregate relations between US states, 2010. Line weight reflects
the number of company-company connections between states ...................................... 123
Figure 18: Coreness scores for top five states, 1998 ...................................................... 124
Figure 19: Coreness scores for top five states, 2007 ...................................................... 124
Figure 20: Coreness scores for top five states, 2010 ..................................................... 124
Figure 21: Average number of films produced and released, 1998 - 2010 ..................... 147
Figure 22: Average employment in motion picture and video production
(NAICS 5121), 1998 - 2010............................................................................................ 147
Figure 23: Average establishments in motion picture and video production
(NAICS 5121), 1998 - 2010............................................................................................ 148
Figure 24: Value of film incentive programs by US states, 1998 - 2010 ....................... 150
Figure 25: Average size of the dominant organizational population, 1998 - 2010 ......... 151
Figure 26: Share of total companies in the dominant population, 1998 - 2010 .............. 151
Figure 27: Small world measures for California and New York, 1998 - 2010 ............... 153
Figure 28: Small world values for California, with 95% confidence intervals .............. 154
Figure 29: Small world values for New York, with 95% confidence intervals .............. 154
Figure 30: Company-company collaboration network, New York, 1998 ...................... 155
Figure 31: Company-company collaboration network, California, 1998 ....................... 156
Figure 32: Company-company collaboration network, New York, 2010 ...................... 156
viii

Figure 33: Company-company collaboration network, California, 2010 ....................... 157
Figure 34: E-I Values for California and New York, 1998 - 2010 ................................. 158
Figure 35: E-I values for California, with 95% confidence intervals ............................. 159
Figure 36: E-I values for New York, with 95% confidence intervals ............................ 159
Figure 37: Percentage of external and internal ties, New York, 1998 - 2010 ................. 160
Figure 38: Percentage of external and internal ties, California, 1998 - 2010 ................. 160

ix

ABSTRACT
Traditional theories of economic development suggest that industry-specific tax
rebates, credits and other incentives will promote sector-specific industrial activity,
employment, and establishments in the jurisdictions where they are applied. However,
project-based industries, which assemble for the completion of a specific task, are both
temporary and highly mobile, presenting a challenge for jurisdictions seeking to draw
these kinds of industries through incentive programs. This dissertation employs insights
from the theories of organizational ecology and interorganizational networks to explain
the uneven success of tax incentive programs targeting the film industry in the United
States.
Film production is a project-based industry that is characterized by the interaction
of specialized firms in a dynamic interorganizational network. This study examines the
relationships among organizations which collaborate in the production of feature films to
determine whether incentives targeting this industry produce development outcomes like
increased in-state filming, film-sector employment and an increase in film-sector
establishments in the states that offer them. The dissertation explores ecological and
social network predictors of economic development, and tests the hypotheses that states
which have more diverse organizational networks, and a larger number of dominant firms
enjoy higher rates of filming, employment and establishment. States whose networks
demonstrate specific structures associated with successful collaboration, including the
balance of ties the state’s firms have with others within and beyond the state, the
networks’ small-worldness and the degree of network coreness relative to other states, are
also predicted to result in more significant and stable development outcomes.
x

This study uses mixed-effects models to examine the effects of incentive
programs, organizational diversity, the presence of dominant firms, and networks of
communication and collaboration on economic development outcomes in the
contemporary American film industry between 1998 and 2010. The results of the study
show that states offering an incentive program increase the amount of filming in their
state significantly. States offering an incentive program also increase film-industry
employment and establishment, but the size of the effect is more modest for employment
and establishment than it is for filming. The amount of money a state offers does not
appear in these analyses to make a statistically significant difference, suggesting that
states do not need to outspend one another to achieve positive development outcomes.
However, post-hoc analysis reveals collinearity among the variables which measure
incentives, and suggests that the claim that the presence of incentives is more important
for outcomes than the dollar value of those incentives cannot be fully supported until the
question of collinearity is resolved.
An even more significant predictor of positive outcomes for incentivizing states is
their organizational diversity: having a more diverse organizational community is
associated with gains in in-state filming, employment and establishment. The presence of
industrially dominant firms, including those engaged in development, marketing and
sales activities is related to employment and establishment, suggesting that having
indigenous firms of this type promotes industrial stability. The interorganizational
network characteristics for states did not significantly predict development outcomes, in
part because of the enduring concentration of so much of the national production network
in California and New York, and the persistence of these states as dominant centers of
xi

production. This persistence suggests that rumors of the demise of these states’ film
industries have been greatly exaggerated. Implications for theory and policy are
discussed.
1

INTRODUCTION
This dissertation explains the uneven effects of economic development policies
which target project-based industries by examining the effects of organizational diversity,
the presence of dominant firms, and networks of communication and collaboration in the
American film industry. While traditional theories of economic development suggest that
state-level fiscal, tax and other nonfinancial economic development tools will promote
development outcomes including increases in industrial activity, employment and
establishments, the success of these incentives is experienced unevenly in different states.
The current research employs insights from organizational ecology and theories of
organizational networks to explain the uneven success of tax incentive programs in
promoting project-based industry in the United States. This study examines the
relationships among organizations in the American film industry to investigate how states
with incentives, diverse organizational communities, and collaborative network structures
associated with creativity and collective innovation promote significant and stable
development outcomes.
The theory of organizational ecology examines organizations and organizational
populations in the context of the resource environments within which they interact with
other organizations and organizational populations. Relationships among these
organizations and populations occur in an institutional context in which regulatory and
policy frameworks impact resource environments, enabling some relationships while
constraining others. Changes in regulatory frameworks and policies therefore create new
opportunities and constraints for the ways organizations and populations relate to one
another, sometimes producing unforeseen or even undesirable consequences.
2

For project-based enterprises, in which organizations come together in temporary
coalitions to execute projects or contracts, the effects of policy changes are felt both at
the level of individual project networks, and across the larger networks constituted by
organizations grouping and regrouping as they collaborate on multiple projects over time.
This dissertation advances the argument that state-level policy change which increases
the resources available to project-based industries in a community promotes development
outcomes by increasing the diversity and productivity of production networks where the
policy is applied.
The dissertation proceeds as follows: The first chapter lays out the theoretical
framework. An overview of economic development theory establishes traditional
expectations about the effects of policy instruments on economic development outcomes.
The challenges in applying insights from economic development theory to industries
characterized by short-term, project-based collaboration are described, and the project-
based ecological network is advanced as a framework which better captures the effects of
policy intervention on these dynamics.
Chapter two presents the American film industry as the context for empirical
analysis, and explores the implementation and consequences of tax rebate and credit
incentive programs introduced by US states to attract film productions to locate their
short-term motion picture projects in their states. An excellent example of an industry
organized around short-term production teams, the film industry’s dynamic grouping and
regrouping of organizational project participants provides an apt context for exploring the
interaction of incentive initiatives, ecological patterns, and interorganizational network
structures.
3

Chapter three advances and justifies hypotheses about the ecological processes
and network structures associated with development outcomes, and chapter four
describes the data, measures and analytical models used to test the hypotheses. The
effects of ecological and network structures on development outcomes are tested by
examining the distribution, attributes and relationships of film production companies
active in the United States between 1998 and 2010. The study examines the networks of
collaboration which emerged as these film companies worked together on projects, and
advances the hypotheses that characteristics such as tie balance, small-worldness, and
structures like core-periphery arrangements are associated with economic development
outcomes. The hypotheses are tested in a series of negative binomial mixed effects
models.
Results of the analyses are reported in chapter five. The study finds that states
with an incentive program of any size experience higher rates of film sector activity,
employment and establishment than states without an incentive program in place. Other
significant predictors are the diversity of an organizational population within a state, as
well as the proportion of industrially dominant companies involved in high-value
activities such as distribution, sales and marketing. While these ecological factors are
found to have significant effects, social network structures are not found to significantly
influence economic development outcomes. The results also suggest that the effects of
diversity and dominant companies are strongest for promoting industrial activity, more
moderate for promoting employment, and weakest for promoting increases in industrial
establishments, suggesting that these incentive programs are most effective at promoting
short-term outcomes.
4

Chapter six draws the dissertation to its conclusion by discussing the implications
of the study, addressing its limitations, and establishing an agenda for continued research
in this area.

5

CHAPTER 1: THE ORGANIZATIONAL NETWORK ECOLOGY OF PROJECT-
BASED INDUSTRIES

“Hollywood is an extraordinary kind of temporary place”
- John Schlesinger, Film Director
Economic Development Theory
Governments at national, state and local levels combine fiscal, tax and
nonfinancial tools to promote sustained growth in industrial activity, employment, and to
increase the number of business establishments within their jurisdictions (Koven &
Lyons, 2010; Mathur, 1999). The use of these tools to promote development outcomes
derives from the view that in a mixed-market economy, public investments which make
an environment more favorable to business interests will produce opportunities for
investment, which in turn benefit the growth and development of that community overall
(Blakely & Leigh, 2010; Koven & Lyons, 2010). Economic development initiatives may
include supply-side interventions, such as grants, loans, subsidies and tax incentives
which benefit producers and suppliers, as well as demand-side programs which target
workers and consumers through workforce development and the building of civil society
institutions and community social capital (Koven & Lyons, 2010; Flora, 1998).
Economic development initiatives are frequently advanced to foster
improvements in three areas: (a) economic and industrial activity, (b) employment, and
(c) the number of business or manufacturing establishments (Bennett & Giloth, 2008;
6

Seidman, 2004; Stimson, Stough & Roberts, 2006). A substantial body of research has
investigated the effectiveness of different economic development tools on these
outcomes. Research on the effects of incentives on employment show small but
significant increases in the number of jobs and employed individuals where incentives are
offered and economic development programs instituted (Bartik, 1991; Goss & Phillips,
1995; 1997). Studies which examine growth in the establishment of factories and
businesses show that high tax rates in a state tend to dampen the development of industry
and the births of new firms (Papke, 1989). Conversely, tax incentives in the form of
rebates and credits are associated with growth in employment and establishment, as
research by Gabe and Kraybill (2002) demonstrates. Non-tax incentives have also been
shown to produce positive effects on employment and business establishments
(O’hUllachain and Satterhwaite, 1992; Bartolome & Spiegel, 1997). Meta-analyses by
Bartik (1991) and Goss and Phillips (1995) confirm these findings.
Variations on the general theory of economic development differ in the emphasis
they place on alternative models of economic growth. Location models evaluate place-
based advantages such as proximity to markets and costs of local labor (Martin &
Ottaviano, 1999; Ericksson & Lindgren, 2009), while export models advocate
development and innovation in a jurisdiction’s export sector (Giles & Williams, 1999;
Rees & Stafford, 1986; Weinstein, Gross & Rees, 1985). Innovation models emphasize
the importance of avant-garde process and product development, and associate economic
growth with developments in high-technology sectors and intellectual property regimes
(Cameron, 1996; Malecki, 1997; Boekema, 2000). These frameworks for understanding
economic development, as well as the tools and strategies which are used to promote
7

outcomes, have changed over time in concert with broader changes in the ways
organizations operate and in response to broader economic trends. Blakely and Bradshaw
(2002) distinguish three overlapping economic development regimes in the United States,
each characterized by a different dominant view of the primary driver of economic
development, as well as by the use of distinct strategies and tools associated with the
achievement of those development goals.
After the Second World War, first-wave economic development strategies,
colloquially described as “smokestack-chasing,” focused on attracting large corporations
to relocate their physical plants to a target jurisdiction. Direct relocation payments,
subsidized loans, as well as tax abatements, credits and exemptions were commonly used
strategies to attract relocation of large manufacturing companies (Blakely & Bradshaw,
2002). In the United States this system of industrial recruitment was dominant through
the 1980s, and is exemplified by the well-publicized efforts of multiple US states to
attract General Motors’ new Saturn assembly plant in 1985 (Bartik, Becker, Lake &
Bush, 1987; Fulton, 2010).
During the 1980s, this first wave of economic development was joined by a
second wave characterized by more indirect forms of assistance to promote the health of
firms already operating within jurisdictions. In contrast to first-wave strategies of
recruitment, second-wave strategies are more often described as leveling the playing
field, and employ below-market loans, incremental tax structures and other instruments to
assist extant firms in building out, and local entrepreneurs in starting up, new enterprises
(Blakely & Bradshaw, 2002; Hembd, 2008). Second-wave policies focus on promoting
and maximizing the performance of existing industries through public-private
8

partnerships, investment in entrepreneurship, and emphasis on high-technology and high-
value industries. This program of growth promotion emerged in response to increasing
global competition, the decline in American manufacturing (Eisinger, 1988), and cuts in
programs for economic development at the federal level (Clarke & Gaile, 1992).
An example of second-wave development programs is the emergence in the 1950s
and 1960s of science and technology parks (for example the Stanford Research Park,
established 1951, and the Research Triangle Park, established 1959). The success of these
early efforts to integrate the contributions of universities, industry and the government
sector, and to promote knowledge and technology transfer among them, led to expansion
of science parks in the 1980s (Link & Scott, 2003). In the United States, while 20 science
and technology parks were created between 1951 and 1981, in the following ten years an
additional 80 parks were established, peaking in 1986 (Link & Scott, 2003). Plosila
(2004) situates the rise of the technology park in the 1980s squarely in the context of
second-wave economic development trends, noting that “state science and technology
efforts began before this time, in the 1960s, but the 1980s began the close integration of
state science and technology efforts with new directions in economic development
practice and planning” (p. 113).
The current economic development framework is characterized by third-wave
strategies which emphasize regional competitiveness and the development of thriving
industrial clusters (Boeckelman, 1999; Eisinger, 1993). The causal perspective of
development has shifted from the notion that providing benefits to individual firms will
promote community level effects, in favor of a more structural view that resources should
be directed at building supportive frameworks and infrastructures, to which companies
9

will be attracted, and from which many organizations can benefit. Strategies associated
with third-wave economic development continue to include the use of fiscal and tax
instruments of previous waves, but additionally emphasize the development of
collaborative infrastructures, the building of supportive institutions, and the promotion of
regional, cross-sector innovation and entrepreneurship. In addition to cutting back on
direct state spending to specific companies and industries, third-wave programs have also
moved to include the participation of non-profit sector and community groups, for
example in training and education initiatives.
The current economic development environment reflects the accretion of
strategies and programs from all three waves. As Zheng and Warner (2010) note, “instead
of one wave substituting for another, local governments now tend to employ all three
approaches simultaneously” (n.p.). Fosler (1992) describes third-wave policy as a shift in
priorities in which “states are concerned with the ways in which workers and businesses
interact in networks and clusters. And they are interested in the dynamics among those
economic entities and related social and political institutions within the context of
specific regions and communities” (p. 5).
The relational emphasis of the third wave has been especially important in terms
of our understanding of economic activity which takes place outside the factories,
corporations and formal, permanent, firms targeted by first- and second-wave programs.
For example, project-based enterprises (Goodman & Goodman, 1976) have long been
common in industries characterized by one-off production, as in shipbuilding (Winch,
1986), construction (Cheng, Dainty & Moore, 2007; Keegan & Turner, 2002) and
creative production, including in the theater (Goodman & Goodman, 1976; Uzzi & Spiro,
10

2005) and film production (Faulkner & Anderson, 1987; DeFilippi and Arthur, 1998;
Ferriani, Cattani & Baden-Fuller, 2007, 2009). Because organizations in these industries
are constituted and reconstituted over time in response to changing production needs,
these project-based industries are structured in ways that both take advantage, and
facilitate the ongoing development, of productive local clusters or agglomerations.
Indeed, the shifts in economic development theory and practice described above reflect a
growing acknowledgement of the importance of project-based industries to promote local
economies, as well as the challenges associated with promoting project-based enterprises.
Project-based enterprises
Research on project-based organizing explores the mechanisms which generate
stability and productivity in industries characterized by temporary, task-oriented
organizing patterns. As DeFillippi and Arthur (1998) describe, project-based enterprises
are ‘typically found where complex, non-routine tasks require the temporary employment
and collaboration of diversely skilled specialists” (p. 125). Unlike project teams which
assemble within the boundaries of formal and permanent organizations (Guzzo &
Dickson, 1996; Kozzlowski & Bell, 2001), project-based enterprises are autonomous,
temporary coalitions of individuals and organizations which coordinate their activities
around the execution of a particular task (Goodman & Goodman, 1976; Grabher 2002).
While project-based organizations vary in composition and structure according to the
specific task for which they are assembled, as a distinct organizational form, project-
based organizations share a number of salient characteristics.
11

First, project-based enterprises exist only as long as necessary for the completion
of their task. In other words, they employ “institutionalized termination” (Lundin &
Sonderholm, 1995). Organizational benchmarks are largely task-dependent, and derive
from production requirements, rather than overarching organizational imperatives, such
as organizational survival or profitability. Vital events for participants in these project-
based enterprises often correspond to deadlines or deliverables which signal changes in
project phases (Wenger & Snyder, 2000; Lundin & Soderholm, 1995). For example, in
film production, the project phases of preproduction, principal photography and post-
production constitute vital events which correspond with the delivery of specific project
elements, such as script, raw footage, and the final cut (Wasko, 2003; Faulkner &
Anderson, 1987).
Second, while permanent organizations emphasize long-term goals and expect
those goals to unfold over time and across the various activities which take place within a
firm (Simon, 1964), project-based organizations are oriented around the accomplishment
of a particular task-based goal, or the fulfillment of a specific contract (Lundin &
Soderholm, 1995). Because project-based institutions are temporary, their horizon for
future activity is short, and the setting of long-term goals unnecessary. Research on
project groups in research and development (Katz, 1982), as well as on airline flight-deck
teams (Weick and Roberts, 1993) has shown that, compared with individuals in
permanent organizations, members of temporary organizations are more attentive to
specific tasks than to long-term goals. Indeed, because project-based organizations are
often constructed as pass-through structures which do not themselves produce or accrue
revenues or profits (DeFillippi & Arthur, 1998), even standard organizational goals such
12

as survival and profitability are not as germane to the project-based organization or to its
participants as they are in more traditional, permanent organizations. For example, in the
film industry, it is common for film producers to create separate limited liability
companies (LLCs) for each project. The LLC structure exists as a short-term corporate
form, which affords the project tax, liability, control and financing options which are
available to corporations but not to individuals. Individuals and organizations brought on
to a film project are hired and compensated through the LLC, which is often dissolved
after financial obligations and returns are realized (Lawyers for the Creative Arts, 2006).
A third characteristic of project-based enterprises is that, because not all project
activities are simultaneous, and because projects are characterized by a division of
specialized labor, it is possible that a project participant may actively engage in different
phases of multiple projects at any given time, playing distinct or similar roles on those
projects (Bechky, 2006). For example, in the film industry it is common for a producer to
be involved in the development of a script in the preproduction stage of one project while
also “closing out” the post-production phase of another project. Also in the film industry,
it is common for organizations such as those which produce special and visual effects, to
work on several projects simultaneously (Curtin & Shattuc, 2009; Lucas, 2011). In
traditional firms, individuals may move from team to team or project to project and may
engage in simultaneous overlapping projects (Grabher, 2002), but these activities are all
coordinated under the aegis of a single organizational entity. In contrast, project-based
entities are independent, and work roles and task organization are less stable. For
example, in his analysis of the advertising industry, Grabher (2002) differentiates
between advertising agencies as permanent organizations within which project teams are
13

regularly shuffled, and advertising projects, which include participants from outside the
agency itself. The distinction is one between a team operating within a traditional firm,
and a project which draws on the contributions of multiple firms and participants in the
construction of a new, independent and temporary structure operating beyond the
boundaries of a single participating firm.
For all these reasons, project-based enterprises present a challenge to traditional
views of the organization, which draw a sharp distinction between coordinated activity
which occurs within the boundary of a firm, on the one hand, and activity which takes
place in open markets, on the other. In the traditional view of the firm as hierarchy
(Coase, 1937), the firm internalizes and coordinates economic activity through its control
of capital and human resources. This dichotomous view of hierarchies and markets has
been relaxed by Williamson (1981, 1991) and others, acknowledging that many
structures now observed in the organizational landscape lie somewhere between these
two poles. Powell (1990) noted that “the history of modern commerce, whether told by
Braudel, Polyani, Pollard or Wallerstein, is a story of family businesses, guilds, cartels
and extended trading companies – all enterprises with loose and highly permeable
boundaries” (Powell, 1990, p. 298). Project-based organizational forms are not properly
described as intermediate or hybrid forms that fall between markets and hierarchies, but
represent a fundamentally distinct logic based on the networks which emerge out of the
relationships among organizations engaged in a common field of activity. Whereas in
permanent organizations, financial and human resources are contained within the firm, in
project-based industries, resources reside in individuals and firms in the larger
community, and are recombined and reconfigured as necessary to meet the demands of
14

specific tasks. In addition to hosting and circulating financial and human resources,
organizational networks also generate resources in the form of the social capital that
individuals and organizations enjoy by virtue of their position in a community (Lin 2001;
Gulati, 2007). Like financial or human capital, social capital is a resource that can be
deployed, but unlike human or financial capital, social capital issues not from the
personal attributes of individuals or organizations, but from the relationships within
which they are engaged (Bourdieu, 1986; Lin, 1999).
Project-based organization also presents a challenge to traditional views of
economic development, which have tended to target permanent and stable organizations,
and structure development expectations around the anticipated growth associated with
large corporate tenants, as in first-wave strategies. However, the relational turn of third-
wave development strategies complicates matters, both for policy-makers seeking to
develop industrial clusters, as well as for researchers who aim to explain the causal
mechanisms through which economic development activity generates advantageous long-
term outcomes (Fosler, 1992). While the analytic and theoretical tools offered by
economic development perspectives are helpful for assessing the effect of individual
economic development tools on a narrow set of goals on a case-by-case basis (Clarke &
Gaile, 1992; Reese & Fasenfest, 1997), making informed judgments about the expected
effectiveness of community interventions requires a theoretical and analytical framework
which explicitly concerns itself with global relational processes (Bathlet & Glucker,
2003; Yeung, 2004).
Economic development tools are carefully selected and strategies are tailored to
accomplish specific jurisdictional goals, such as promoting white-collar jobs, attracting
15

tourism, or establishing a high-technology cluster (Goss & Phillips, 1995; Koven &
Lyons, 2010). These tools also reflect a community’s specific portfolio of environmental
endowments and disadvantages (such as the available workforce, climate and economic
base). The specificity of these goals and affordances makes effective evaluation and
comparison of economic development programs a challenge for program developers,
legislators and researchers (Hantry, Fall, Singer & Liner, 1990).
As Koven and Lyons (2010) describe, the diversity of policy instruments, the
variety of possible measures and indicators of success, and the subjectivity with which
these indicators are selected for evaluation make economic development program success
“difficult to define” and can be “a double-edged sword” (p. 10). The authors illustrate
these challenges by citing the legalization of gambling activity as an exemplary case:
[J]obs can be created through legislation that legalizes certain activities such as
casino gambling. If job creation is the only measure [of interest], legalization of
gambling could be described as a successful economic development strategy.
However, other consequences – the flow of revenue to company stockholders
rather than to workers, spread of crime, increases in healthcare costs – call into
the question the success of the gambling strategy (Koven & Lyons, 2010, p. 10).

The challenges of comparative program evaluation are even more acute. As
Clarke and Gaile (1992) describe, “[a]ttempting to assess the effects of local economic
development strategies is a quagmire of good intentions and bad measures. An interest in
finding out whether policies have the desired effect is laudable, but there is little
consensus on appropriate measures of success or impact” (p. 193).
In order to understand how contemporary economic development policies operate
beyond specific cases, it is necessary to examine the interactions of organizations within
16

and across jurisdictions, and to attend to the differing environmental contexts these
relationships traverse. Several studies have emphasized the important role policy changes
play in the development of organizational populations and communities (Swaminathan,
1995; Tucker, Singh & Meinhardt, 1990). For example, Dobbin and Dowd’s (1997)
research on Massachusetts railway firms showed that when the state of Massachusetts
made funds available for the establishment of businesses which would support and supply
railroad operations, the increase in overall resources for the industry rose, and foundings
of railway-related organizations rose with it. Not only direct infusions of capital, but
incentives in the form of state-issued public bonds also contributed to the increase in
railroad-related industry in the state. These findings demonstrate that state policy which
augments the resources available to an organizational community can be successful in
stimulating that community.
Staber’s (1989) examination of the effect of changes in the Canadian tax code on
the establishment of cooperative firms is also instructive. Staber used tax rates to
categorize government support for the cooperative form as either munificent (years
during which cooperatives were exempt from tax liability) or restrictive (years for which
exemptions were repealed) and found that a munificent tax environment was associated
with increased cooperative organization founding.
Scheiber’s (2001) analysis of the United States as a federation describes the fifty
US states as a kind of organizational population, linked by commensalist relations.
Scheiber focuses on the competitive, or ‘rivalistic’ relations among states as they
compete with one another for businesses. His discussion is worth quoting at length:
17

Examples of this rivalry include corporation law, in which states compete for
investment by giving corporations legal advantages and immunities; taxation law,
in which special concessions are given as subsidies to investors and companies,
inducing them to locate in the state rather than elsewhere, in rival states; and
above all, in the continuing true diversity of politics represented in state law in
such fields as labor relations, education, criminal justice and taxation (Scheiber,
2001, p. 69).
By focusing on the competitive and collaborative relationships among
organizations, as well as by highlighting the specificity of their environmental contexts,
the organizational ecology framework provides a profitable and pertinent approach to
comparative analysis of economic development programs.
Organizational Ecology of Project-Based Industries
Employing concepts from biological science, the theory of organizational and
community ecology explains how organizations of various kinds interact as they struggle
to secure the resources necessary for their survival. Described by Hawley (1986) as “the
study of the relation of organisms to their environment” (p.1), ecology has been used as a
model for describing the relationship of firms to one another and to their industrial
environments. Like biological organisms, organizations operate within bounded resource
environments which they must share with others (Hannan & Freeman, 1977, 1984). The
struggle to obtain resources shapes relationships among parties, so that some firms are in
competition with one another, while others collaborate, and still others interact in ways
between these two extremes, or not at all (Aldrich & Ruef, 2006). An advantage of the
ecological perspective of organizations is that firms are examined within the larger
context of their multiple interdependencies, allowing researchers to explore
simultaneously the resources in an environment, the way these resources structure
relationships among organizations, and how environmental processes shape and change
18

these interactions over time (Hannan & Freeman, 1977, 1984; Baum & Amburgey, 2005;
Rao, 2005). The following section introduces the population ecology framework, which
examines interdependencies among groups of similar organizations.
Population Ecology
Hannan and Freeman (1977, 1984) pioneered the application of ecological
principles to organizations, observing that “ecological analysis is conducted at three
levels: individual, population, and community. Events at one level almost always have
consequences at other levels” (Hannan & Freeman, 1977, p. 933). Advocating a shift
from analysis of individual organizational cases to the population level of analysis,
Hannan and Freeman define organizational populations as aggregate sets of organizations
which share a “unit character” (1977, p. 934). Analogous to biological species, an
organizational population is defined in terms of the similarities among its constituent
organizations. This similarity may be reflected in a shared organizational form, which
Hannan and Freeman (1977) describe as a population’s shared “blueprint for
organizational action” (p. 935). Organizations sharing a common form are distinguished
by their activities, as well as by the formal structures which facilitate their effective
engagement in those activities. For example, the organizational form “hospital” can be
distinguished from the organizational forms “day-care center” or “bank” by examining
what each type of organization does (its activities) and how it is arranged to accomplish
its objectives (its structure).
An organizational population may also be distinguished in terms of the role it
plays in an environment or community. This role is referred to by organizational
19

ecologists as the niche the population and its organizations occupy (Freeman & Hannan,
1983). The specific set of resources required by organizations in a population constitutes
the population’s fundamental niche (Hannan and Freeman, 1977). For example, local
newspaper organizations require journalists, access to a press, and local consumers. This
set of resources is common to, and marks as similar in resource requirements, all local
newspapers, suggesting that “local newspaper” is an identifiable organizational
population (Dimmick, 2003). Organizations sharing a fundamental niche rely on the same
resources, and are therefore likely to compete over these resources in their shared
environment (Carroll, 1985; Hannan & Carroll, 2003). The presence of competitors
potentially constrains the ability of organizations to acquire all the resources they need.
For example Hsu (2006) shows that, in their common dependence on audiences, film
production companies share a fundamental niche and engage in competition with one
another for those audiences. The concept of the realized niche accounts for this
competitive constraint, and refers to “the subset of the fundamental niche in which an
entity can sustain itself in the presence of given competitors” (Hannan, Carroll & Polos,
2003, p. 310). Returning to the example of film production companies, Hsu (2006)
describes the way film companies produce films in different genres, in order to address
this competitive constraint.
The division of the resource space into distinct niches means that organizational
populations are distinguished from one another by the part of the resource space they
seek to exploit. Generalist organizations cast a wide net and correspondingly have a
broad niche, whereas specialist organizations target a narrower set of resources and
therefore have a narrower niche (Carroll, 1985). Operating at the “center” of the resource
20

space, where a wide variety of resources are available, generalists compete with other
generalists over a diverse set of heterogeneous resources (Carroll, 1985). Specialists,
seeking a narrower range of resources tend not to compete head-to-head with generalists
at the crowded center of the resource space, but target resources at the periphery, which
remain open and unexploited by generalists (Freeman & Hannan, 1983). Hsu’s (2006)
study of film production companies showed that generalist film production companies
targeted a wide audience demographic and produced films in a variety of genres, whereas
specialist film production companies focused on specific audiences by producing films in
a single genre. Similarly, Carroll (1985) demonstrated that a small newspaper
organization may sustain itself by appealing to a small, ethnically specific audience, for
which larger, generalist newspapers would not compete directly. While both types of
newspapers require the same resource types, by adopting a specialist strategy, the ethnic
newspaper need not engage in direct competition with firms serving larger markets.
This process of resource partitioning also explains why concentration in industries
leads, not to fewer organizations, but to more organizations. The convergence of
organizations at the center of the resource space actually frees up more resources at the
periphery, which can be targeted and exploited by new, specialist organizations. Carroll
(1985) illustrated this process in his examination of the newspaper industry, revealing
that consolidation among generalist newspapers created opportunities for smaller
newspapers to gain resources from more narrowly defined audiences based on geography,
ethnicity and reader interests. By neglecting the specialized corners of the market in their
efforts to capture the general resources at the center, generalists created resource
opportunities for specialists to thrive. Research in population ecology has shown that this
21

practice of partitioning a resource space into generalist and specialist niches is
widespread across industries and countries, and can be observed in populations of
American labor organizations (Hannan and Freeman, 1989), in the media industries
(Dimmick, 2003), among semi-conductor manufacturers (Podolny, Stuart & Hannan,
1995) and in populations of restaurants (Freeman & Hannan, 1983) and producers of beer
and wine (Carroll & Swaminathan, 1992; Swaminathan 1995, 2001).
The niche is a particularly useful concept because it explicitly defines a
population in terms of its resource requirements, which also structures its interactions
with other organizations in the population (Hannan & Carroll, 2003; Freeman & Hannan,
1983). The realized niche suggests a competitive relationship between similar
organizations, but in fact there are a range of possible interactions organizations may
have with other members of their population. Hawley (1986) defined these relationships
within a population of similar organizations as commensalist, because the organizations
draw on the same set of resources, or “eat from the same table” (p. 39). Aldrich and Ruef
(2006) describe the range of commensalist relationships as a continuum ranging from full
competition, in which resources represent a zero-sum game and gains by one
organization represent losses by another, to full mutualism, in which organizations
benefit from the presence of their competitors, despite the fact that they pursue the same
resources. Between these two extremes, organizations may engage in partial competition,
partial mutualism, or have a neutral relationship in which they do not interact at all.
This range of possibilities for interorganizational interaction results from the fact
that, even within a population of similar organizations, not all firms will require precisely
the same resources as others. A niche is multidimensional to the degree that variations
22

exist in the specific resource requirements needed by the organizations in a population to
survive. Within a population, organizations may have more or less specialized
requirements, the variety of which is described as the niche’s width (Carroll, 1985).
Returning to the population of hospitals as an example, organizations in this population
all require patients, nurses, doctors and facilities to persist. However, within the
population of hospitals, those specializing in cancer treatment will require a different
portfolio of specific resources from those required by a generalist hospital. The width and
specific dimensions of an organization’s niche differentiates it from others in its
population, and structures its range of possible relationships. Two organizations in the
same population which have precisely the same niche dimensions and width will engage
in full competition for those resources. On the other hand, for the specialist and generalist
hospitals described above, while the organizations’ niche dimensions are the same (each
requiring doctors, patients, nurses and facilities), the general hospital has a wider niche
than the specialist hospital, which suggests that the two hospitals will engage in partial
competition where their niches overlap (Bruggeman, Grunow, Leenders, Vermulen &
Kuilman, 2012). As well, supposing that the general hospital is ill-equipped to handle
oncology, and the specialist cancer hospital ill-equipped to tackle gunshot wounds, the
two organizations may have a mutualistic relationship in which each derives a benefit
from the existence of the other.
A crucial factor influencing the relationships among organizations in a population
is the relative abundance or scarcity of resources in their shared environment. Within the
ecological framework, the viability of individual organizations and organizational
populations depends on their ability to collect resources from the environment, which
23

may be seen as a pool from which organizational populations draw sustenance (Specht,
1993). The “scarcity or abundance of critical resources needed by (one or more) firms
operating within an environment” is termed environmental munificence and influences
organizational relationships in a number of ways (Castrogiovanni,1991, p. 542).
First, environmental munificence influences relationships by encouraging or
discouraging organizational founding. Venter and colleagues (2012) found that the influx
of economic resources associated with South Africa’s hosting of the 2010 FIFA World
Cup Championship in soccer promoted entrepreneurship in the country’s informal
economic sector, and the formation of new firms in services, manufacturing and retail
trading. Munificence in the resource environment not only encourages local founding, but
Kirca et al. (2012) find in a meta-analysis of 154 studies that firms are more likely to
become multinational, that is, to found new branches in other countries, when their home
environments have high levels of resources.
In addition to increases in founding, the abundance of resources in an
environment is also associated with organizational survival. Castrogiovanni (1996)
proposed a positive relationship between environmental resource munificence and the
survival of new businesses. Bamford, Dean and McDougal (1999) found that resource
munificence at the time of bank foundings accounted for a significant amount of variation
in their performance for five to six years following their establishment. Banks founded
under conditions of resource abundance significantly outperformed those founded under
more lean initial conditions. Carroll and Hannan (1989) found that the level of resources
in an organization’s environment during its founding can have long-term consequences.
The authors demonstrated that, across five industries in three countries, organizations
24

which were founded under conditions of environmental scarcity suffered permanently
higher rates of failure.
Romanelli (1989) argued that resource munificence is particularly salient for new
and young firms, and tested this hypothesis through empirical analyses of survival rates
in the micro-computing industry. The results demonstrated that that resource abundance
have direct positive effects on the survival of new and young firms. An allied hypothesis,
that increasing competitive concentration among these firms would have negative effects
on survival, was not supported. This led Romanelli to conclude that, for new and young
firms, positive survival effects of environmental munificence overshadow the effects of
the competition they face from rivals.
The abundance of environmental resources is also associated with what Cassia
and Minola (2012) describe as “hyper-growth.” After exploring entrepreneurial
orientation and management style as potential contributors to this extreme growth,
(defined as a minimum of twenty percent growth by initially small enterprises for a
minimum of four consecutive years), the authors determined that access to deep resource
pools is the primary common feature among the media, transportation and shipbuilding
enterprises they examine. As they note, despite heterogeneity in management styles and
organizational factors specific to each case, “there is a consistency among the sample of
hyper-growth firms; that is their extraordinary access to resources” (Cassia & Minola,
2012, p. 192).
Environmental munificence is also associated with the success of novel and
emergent organizational forms. Adner and Levinthal (2002) show that the evolution of
25

both the wireless radio and internet industries depended significantly on the release into
the environment of massive new resource caches. The authors (2002) describe that some
of these resources were released speculatively, on the basis of expectations that radio and
internet technologies would, in time, open up profitable new markets. In other words, the
mere promise of an influx of resources through new markets and applications at some
unspecified later point in time was sufficient to prompt investors to release funds into the
environment earlier, augmenting the resources available for the development of the
industry in its early stages.
Research on industrial districts also supports the view that resource munificence
leads to positive outcomes for organizational founding and success, suggesting that some
cities or regions are advantageous because natural, human and financial resources are
concentrated within limited geographic distances. For example, in the fields of
biotechnology (Sorenson & Stewart, 2008) and high-technology (Lee, Lee & Pennings,
2001) entrepreneurs are shown to found their operations in locations where essential
resources like labor and venture capital are concentrated. Crone (2012) demonstrated that
the resource environment in Ireland in the late 1990s and early 2000s was abundant with
the specific human and financial resources essential to software firms, and that the
presence of these resources promoted firm founding during that period.
In addition to financial capital and human resources, institutional resources such
as legitimacy and social support also promote organizational activity (Mayer & Rowan,
1977; Baum & Oliver, 1996). Wan and Hoskisson (2003) assert that effective political
institutions, legal institutions and societal institutions all constitute important resources
which promote organizational activity. The authors examine the propensity of
26

organizations to establish international branches and find that home country differences
in institutional resources such as product and intellectual property protection had
powerful effects on founding, at home and abroad. Firms from more institutionally
munificent home environments were more likely to diversify internationally, as well as to
gain greater performance advantages from doing so.
Baum and Oliver (1991) also found that munificent institutional environments
have positive effects on the development of organizational populations. Their study of
child care centers in Toronto revealed that in the early development of the private child-
care industry in that city, formal relationships with government agencies promoted
founding, and diminished failure rates, of child care organizations. Moreover, as these
formal relationships became more common across the child-care population, their
positive survival effects extended even to organizations which did not themselves engage
in these formal institutional relationships. Not only did direct institutional links provide
tangible resources to specific organizations in the population, these links generated
legitimacy benefits for the child-care community as a whole.
Galaskiewicz, Bielefeld and Dowell (2006) found similar results in their study of
not-for-profit organizations. Relationships between this population and elite donors were
paramount, not only because of the access to tangible resources the relationships
provided, but also because these relationships enhanced the non-profits’ reputation and
legitimacy. Pontikes and Barnett (2010) provide further support for the value of
institutional resource munificence and suggest that infusions of capital to specific
organizations in a population may promote additional foundings across novel market
areas, not only because those resources are available to newcomers, but because the
27

action of committing resources to a representative organization generates legitimacy, or
“blesses” the market overall. The value of institutional munificence and legitimacy
appears to generalize beyond particular organizations to others belonging to the same
population or community.
Because the population ecology framework examines organizations in their
environmental and relational context, research has also explored how the failure of some
organizations and populations may generate opportunities for others. The failure of a
competitor within an organizational population generates opportunities for extant
organizations in that population to exploit the resources released by the firm’s demise
(Astley, 1985). As well, organizational failures also promote the founding of new
organizations which can capture the newly freed-up resources made available by failed
organizations. Delacroix and Carroll (1983) find that newspaper failures occasioned by
political turmoil made resources available for the foundation of new newspaper
organizations. In a similar vein, Stuart and Sorenson (2003) find that liquidity events in
the form of acquisition and initial public offering behavior by firms in the biotechnology
industry released resources into their environment which served to accelerate the
founding of new organizational forms in that sector. The consolidation of generalist
organizations in a population through competition and merger at the center of the
resource space (described on pps. 19-21) also leaves resources at the periphery of the
resource space open for exploitation by specialists. This means that, as organizations
struggle to acquire resources available in their environments, failures and other
competitive activities may make them the source of resources for others in their
population, too.
28

The munificence of the resource environment clearly exerts an important direct
influence on the dynamics of organizational populations. However, the level of available
resources for population use may also have indirect effects on organizational founding,
survival and failure. Organizational foundings in a population are argued to promote
subsequent additional foundings because of the strong signals these foundings provide to
other potential entrepreneurs that a fertile niche exists for exploitation by new
organizations (Carroll & Hannan, 2000; Delacroix & Rao, 1994; Hannan & Carroll,
1992). However, this positive effect of prior foundings has a ceiling, because the addition
of each new organization to the environment also increases competition for resources.
Even a munificent resource environment or niche is limited in the total level of resources
it provides. This limit is referred to as the environment’s carrying capacity, and describes
the total number of organizations that can be supported by the resource environment
(Hannan & Freeman, 1977; Carroll & Hannan, 2000). As the demands of a growing
number of firms begins to approach the finite limit of an environment’s resource
capacity, additional foundings lead to accelerating resource scarcity, increased
competition and eventual failures (Hannan & Freeman, 1977). As a population begins to
approach the limits of the resource environment, the risks for younger and weaker
organizations increase as well. As Barron (1999) notes, “the gap between the survival
chances of the most robust organizations and the most frail organizations should widen as
the population density reaches the carrying capacity” (p. 429). While the resulting
failures may in turn release usable resources into the environment, promoting a new
round of foundings, Carroll and Delacroix (1983) warn that too many failures may
indicate an inhospitable or unpredictable environment, curtailing founding.
29

These dynamics have two related implications: not only does the number of
organizations in a population alter the level of resources available across a population, it
also impacts the degree of institutional and cultural support organizations can expect
from their environment. Defined by Suchman (1995) as “a generalized perception or
assumption that the actions of an entity are desirable, proper or appropriate within some
socially constructed system of norms, values beliefs and definitions,” (p. 574) legitimacy
rises and falls with population density. The greater the number of organizations that take
on a particular form, the more that form becomes taken for granted as a normal,
reasonable and sensible way to coordinate activity. Legitimacy promotes the expansion of
that form. These density-dependent dynamics provide additional support for the view that
environmental munificence comprises both tangible resources and institutional resources
like legitimacy, and that these are interdependent in their effects on population growth
(Delacroix & Rao, 1994; Freeman & Audia, 2011; Petersen & Koput, 1991).
Legitimacy may also explain why, in their analyses of failure rates in cement and
semiconductor industries, Anderson and Tushman (2001) did not find a direct
relationship between declines in resource munificence and failures. They suggest that
firms may “develop routines and structures that cope with the harsh worlds of the ups and
downs of the business cycle and with resource-scarce environments” (p. 700). Their
analyses suggest that uncertainty and unpredictability about the resource environment is
an even more significant predictor of firm exit than the actual level of resources available
to firms. While the abundance or growth of organizational forms, and the devotion of
resources to their survival signals legitimacy, uncertainty and unpredictability about the
resource environment challenges it. Anderson and Tushman’s finding that stability is
30

more important than the absolute level of resources in the environment is echoed by
Elbanna and Alhwarai (2012), who find that, while managers were more uncertain about
fiscal policies than about the relative abundance of resources in the United Arab Emirates
and in Egypt, both types of uncertainty were associated with lower organizational
performance.
Resource munificence clearly has both direct and indirect effects on the
competitive and collaborative relationships among similar organizations interacting in a
resource environment. The population ecology framework clarifies these commensalist
relationships among similar organizations, and finds consistent support for the resource
and legitimacy dynamics which characterize their development. However, while
organizations interact with one another in populations, populations themselves also
interact in the context of larger communities. It is to the elaboration of these relationships
that we now turn.
Community Ecology
Beyond the population level of organizational analysis, the ecological framework
also extends to the dynamics of larger organizational communities (Astley, 1985; Astley
& Fombrun, 1983; Hannan & Freeman, 1977). A community-level approach to
organizational ecology shares the same basic assumptions as the population approach,
including its focus on the interaction of entities in the context of shared resource
environments. While population ecology examines similar organizations as distinct
populations, a community perspective on organizational ecology takes a wider view, and
examines the relationships among those different organizational populations as they
31

interact in larger communities (Astley, 1985). An ecological community consists of
multiple organizational populations which are related to one another through “their
orientation to a common technology, normative order, or legal-regulatory regime”
(Aldrich, 1999, p. 301). In this way, organizational communities are analogous to
organizational fields, which DiMaggio and Powell (1983) describe as the set of “those
organizations that, in the aggregate, constitute a recognized area of institutional life: key
suppliers, resource and product consumers, regulatory agencies and other organizations
that produce similar services and products” (p. 148).
For example, in their study of the children’s television community, Bryant and
Monge (2008) examine the interactions among eight related populations: entertainment
content creators, education content creators, programmers, merchandisers, advertisers,
advocacy groups, governmental bodies and philanthropic organizations. While each of
these populations has a distinct form and plays a unique role, together they represent the
major constituents of the larger children’s television community (Bryant & Monge,
2008).
The community level of ecological analysis examines how multiple populations
interact. Some relationships between populations whose niches overlap are commensalist.
For example, Bryant and Monge (2008) distinguish education and entertainment content
programmers on the basis of important differences in the products they create, though the
two populations overlap to some degree, and do compete in a limited way to provide their
content to programmers. Inter-population commensalism of this kind has also been
observed among populations of American labor organizations. In their analysis of
industrial and craft labor unions, Hannan and Freeman (1987) find that increases in the
32

founding of industrial labor unions constrained the founding of craft labor unions, in part
because of the competition for members occasioned by the overlap of the two
populations’ resource niches.
However, because populations often occupy distinct niches, their relations may
not be structured around resource competition. Rather, relations between populations are
often symbiotic in nature. Symbiosis refers to mutual dependencies between populations
(Hawley, 1986). For example, in the children’s television community, the relationship
between content creators and programmers can be described as symbiotic because of the
ways these populations each rely on the other to gain the resources they need: without
content, programmers would not be able to acquire the audiences that are necessary to
their survival. Conversely, without programmers to distribute their material, content
creators would not be able to continue making new programs (Bryant & Monge, 2005).
An advantage of the community ecology framework is the opportunity it affords for
challenging the idea that the environment is “a more or less intractable externality, a
predefined context that ultimately establishes what is feasible” (Astley & Fomburn, 1983,
p. 576). Rather, a community ecology perspective encourages the view that organizations
and populations themselves constitute endogenous sources of opportunity. In other
words, as different populations in the community engage in relationships arrayed around
their use of resources in the environment, the community itself becomes a source of
resources.
One population in a community may constitute the source of essential resources
for another population. For example, in the children’s television community,
programmers are a source of resources for content producers, who purchase their creative
33

content. Research by Audia, Freeman and Reynolds (2006) describes similar dynamics in
the specialized instruments community, detailing how manufacturers and purchasers of
these specialized instruments maintain symbiotic relationships in which each requires
essential resources from the other, exchanging financial for material resources. All the
while competition within these populations remains fierce. Indeed, communities which
interact as parts of a supply chain are good examples of both commensalism and
symbiosis, because multiple competing suppliers in one population engage in exchange
relationships with suppliers and providers up and down the chain (Hugos, 2011).
In addition to the commensalist and symbiotic relationships which may shape
interactions between populations, another important relation is that of dominance
(Aldrich & Ruef, 2006). In some organizational communities, one population may enjoy
disproportionate power, particularly if it controls the flow of resources to other
populations, producing a hierarchy of influence and power (Mizruchi, 1996). As Hawley
(1986), describes, “certain functions are by their nature more influential than others; they
are strategically placed in the division of labor and thus impinge directly upon a larger
number of other functions” (p. 221).
Community ecological analysis allows researchers to investigate resources, not as
latent environmental variables, but as endogenous, having their source within the
community itself. Indeed, as organizations and populations in a community interact with
one another over time, they may develop relational patterns and norms which make them
what Astley (1985) calls “functionally integrated systems of interacting populations; they
are emergent identities that, over time, gain a degree of autonomy from their
environments” (p. 234). As patterns of relations among organizations and populations in
34

a community stabilize, organizations and populations become more interdependent. If
these relations generate sufficient opportunities for meeting the community members’
various resource requirements, these communities may become relatively autonomous
from the larger environment. This coherence may buffer a community from
environmental shocks (Kelly & Amburgey, 1991).
However, the equilibrium that develops as populations settle into stable
relationships within an organizational community is fragile, and potentially temporary.
The stability of a community can make it more rigid in the face of environmental change.
For example, Cooke (2002) attributes the collapse of the community of organizations
involved in silk production in central England to its “calcification,” and resistance to new
entrants. The relational stability and structure which developed among silk-making
organizations over many years proved fatal when that industry became more globalized,
and faced competition from foreign competitors. Thus, while a community may be
optimized for efficient activity under a particular set of environmental conditions, it may
be rendered ineffective or inefficient if its stability prevents it from adapting in the face
of change. In the context of economic development strategies, while the introduction of
resources may promote economic activity and development, and over time may see the
development of a stable network of participants and activities, the same community may
not be robust to environmental changes (Baum, 1990; Dobrev, Kim & Carroll, 2003).
Project-based industries as community ecologies
In ecological approaches, researchers explore the ways organizations and
populations interact with one another and within a bounded resource environment.
35

Research in population and community ecology has developed a comprehensive analytic
framework and research program for the examination of populations and communities
consisting of formal and permanent organizations and institutions. This framework can
be adapted to the study of project-based organizational fields by viewing the permanent
organizations involved in temporary project enterprises as organizational members of
interacting populations within a larger industrial community (Grabher, 2002).
Project-based populations consist of those groups of firms which share a unit
character. Project leaders assemble temporary coalitions of organizations who fill
particular roles in the execution of a project task (Cattani, Ferriani & Baden-Fuller, 2009;
Goodman & Goodman, 1976). Among the primary responsibilities of a project manager
or project entrepreneur is the filling of project positions with organizations that have the
requisite experience, knowledge and abilities to execute the tasks associated with their
role or project position. A project-based population, then, consists of all those
organizations which are eligible for the fulfillment of a specific project role. A shared
organizational form (Aldrich & Ruef, 2006) indicated by a shared job or task description
may be used to define individuals as members of the same population (Bechky, 2006).
Just as organizations within a population may interact commensalistically as competitors,
collaborators, or in various ways between, so do the organizations within a population in
a project ecology. The term project ecology was first used by Grabher (2002) to describe
“the interdependencies between projects and the particular firms, personal relations,
localities and corporate networks from which these projects draw essential sources” (p.
246). The project ecology concept is also adopted by Johns (2010) in her analysis of the
television and film production community in Manchester, England. However, while both
36

authors employ the term ecology as a productive metaphor, neither makes the crucial link
between the term and the theoretical framework of organizational ecology. This research
seeks to deepen our understanding of project ecologies by explicitly joining it with the
theoretical framework of organizational and community ecology.
While the populations in a project ecology may interact commensalistically, they
may also engage in symbiotic relationships with one another. Because project-based
enterprises engage “a diversely skilled set of people working together on a complex
task,” (DeFillippi & Arthur, 1998, p. 125), most project organizations will consist of
representatives of several organizational populations (Cattani, Ferriani & Baden-Fuller,
2007, 2009; Uzzi & Spiro, 2005). The diversity of organizations participating in a project
based enterprise marks an important difference between project teams, which reassemble
and reconfigure the assets of a single firm toward the execution of a task, and project
organizations or project enterprises which assemble these resources from the larger
community (Whitley, 2006; Edmonson & Nembhard, 2009). And, because, in the context
of a specific project, they are engaged in a collaborative enterprise around the completion
of a common task, interactions among these project participants will often be symbiotic
in nature, as well as collaborative and mutualistic.
The environment in which these project-based enterprises emerge also shapes
organizational interactions. Commensalist and symbiotic relationships among project
participants unfold against a background of financial, human and network resources,
which are sought and assembled anew for each project. The uneven distribution of
financial, human and network resources in the environment will create uneven growth
across a project-based organizational community. As described above (see pps. 22-28),
37

communities in more munificent environments are more likely to support more
organizations, more organizational populations, and more project teams and activity.
Project-based Ecologies as Interorganizational Networks
In the preceding section, project-based industries were described as organizational
communities of interacting firms, tied to one another through project collaboration as
well as through dependencies which emerge as they struggle to capture resources in their
shared environment. However, because project-based organization is defined by the serial
creation and dissolution of teams, the analysis of founding and failure rates common in
the study of organizational ecology does not capture the complex population and
community dynamics in project-based industries. Organizational birth and death, the
variables of greatest interest in organizational ecology studies, are routine occurrences in
project-based industries, rather than vital events. As Kenis, Janowicz-Panjaitan and
Cambre (2009) note, explaining disbanding and survival rates in populations of
organizations only makes sense if one assumes that organizations intend to survive over
time and will not be intentionally terminated, for example, after accomplishing a task.
Analyses of organizational demographics, common in studies of ecological population
and community transformation, fail to capture more subtle indicators of stability inherent
in project-based industries characterized by organizational fluidity. Previous efforts to
examine project-based or temporary organizations through an ecological lens have either
examined these kinds of organizations as a distinct organizational population (Kenis,
Janowicz-Panjaitan & Cambre, 2009), or have used the ecological framework as a
structuring metaphor, without developing explicitly ecological hypotheses or analyses
(Grabher, 2002).
38

However, the ecological focus on relational connections among permanent
organizational actors is shared by researchers who adopt the theoretical perspectives and
analytical methodologies of social network analysis. Incorporating a social networks
approach to the study of organizational communities resonates with the ecological
perspective’s focus on relationships among organizations and their environment. This
allows the researcher to emphasize the interactions among community members, and the
organizational structures which emerge and evolve as connections among organizations
form and dissolve over time. While traditional ecological methods are based on
demographic shifts across relatively stable casts of organizational characters, by focusing
on temporary relational structures and patterns among a set of permanent organizations,
social network approaches are more amenable to the analysis of communities of
organizations in which relationships are dynamic, and where turnover among a changing
roster of organizations is the norm.
By shifting from a focus on the specific organizations in a community to the
structure and patterns of the ties among populations of participants, social network
analysis provides a tool for examining the population and community processes of
project-based ecologies. In addition to providing a methodological framework for
examining ecological processes, the social network perspective also suggests a number of
specific structural signatures which clarify more precisely how patterns of ties between
organizations at a local level might produce far-ranging consequences across the
communities to which they belong. Structural signatures reflect common social and
organizational behaviors and practices which give rise to particular relationships and
arrangements among individuals and organizations (Robins, 2011). For example
39

sociologists have extensively examined the tendency for human social relations to be
organized around the principle of reciprocity (Myers, 2000; Worchel, Cooper, Goethals,
& Olson, 2000). Reciprocity norms, in which individuals tend to seek balance in social
relationships by returning favors and exchanging gifts, appear in social networks as a
balancing signature. A tie initiated by one person or organization is frequently observed
to be reciprocated by the other. The extent to which this tendency cumulates across the
many ties in a network allows the researcher to determine whether pairwise decisions to
reciprocate produce network-wide tendencies (see, for example, research on the
emergence of generalized exchange in microfinance lending networks by Larance, 2001
and analysis of favor exchange networks by Jackson, Rodriguez-Barraquer & Tan, 2010).
In the context of this study, the structural signatures of interest include the extent
to which organizations and groups of organizations are embedded in an organizational
network, the extent to which a network demonstrates properties associated with small
worlds, and the existence of core-periphery patterns. The following section addresses
each of these network structures in turn.
Embeddedness
A fundamental premise of the social network perspective is that social, economic
and organizational activity take place within the context of relationships, but is not
wholly determined by them. As Granovetter (1985) describes,
Actors do not behave or decide as atoms outside a social context, nor do they
adhere slavishly to a script written for them by the particular intersection of social
categories that they happen to occupy. Their attempts at purposive action are
instead embedded in concrete, ongoing systems of social relations (p. 487).

40

In the context of project-based ecologies, each project-based enterprise is itself a
network consisting of the individuals and organizations engaged in the execution of a
task, and the collaborative ties which result from their joint participation in a project.
However, as Engwall (2003) reminds, no project is an island independent of its
environment. These project-based organizations are themselves embedded in larger
networks, made up of the combination and recombination of participants across multiple
projects over time. This layering of relationships is clarified by the distinction between
relational embeddedness and structural embeddedness (Granovetter, 1985; Gulati, 1998;
Uzzi, 1996).
Relational embeddedness describes the relationships and interactions in which
organizations directly participate. Through direct interactions, organizations develop
expectations about one another’s behaviors, establish trust, and generate dependencies
(Gulati, 2007). Relational embeddedness establishes norms of interaction which reduce
uncertainties about how others are likely to behave. Uzzi (1997) describes relational
embeddedness in his study of garment workers as the “special” or “close” relationships
between interacting partners who are known to one another, who interact over time, who
exchange information and who work together to solve problems.
Relational embeddedness is contrasted with structural embeddedness, which
accounts for the larger social environment in which an individual or organization is
positioned, beyond the individual ties they participate in directly. Gulati (1998) describes
structural embeddedness as going “beyond the immediate ties of firms and emphasize[s]
the informational value of the structural position these partners occupy in the network.
Information travels not only through proximate ties in networks but through the structure
41

of the network itself” (p. 296). While relational embeddedness speaks to a set of
relationships a specific organization has with other organizations, structural
embeddedness reflects the position an organization occupies in the larger network,
relative to others with which is it connected directly or indirectly.
Drawing on Granovetter (1985), Uzzi (1997) articulates a contrast between
underembeddedness, in which individuals or organizations are insufficiently tied to
others who would be able to provide them opportunities and information, and
overembeddedness, in which the ties are redundant, or generate unproductive, or even
counterproductive, dependencies. For example, Granovetter (1973) explored the way
distant interpersonal ties of individuals influenced the ease with which job-seekers found
employment. The finding that an individual’s strong, close ties were less helpful than
their less intimate, weak ties for accessing information about job opportunities suggests
that close, strong ties are sources of relational depth, while distant ties provide breadth by
providing access to novel and unique sources of information. Uzzi (1997) notes that
“overembeddedness can also stifle economic action if the social aspects of exchange
supersede the economic imperatives.” Uzzi (1997) demonstrated that friendships between
independent contractors in the garment industry sometimes encouraged exchange
decisions which might seem unwise when considered purely from an economic, profit-
maximizing perspective, but which make sense when trust, interdependence and the
potential for future interactions is taken into account. However, overembeddedness may
also promote inertia, or an inability of a tightly-knit community to respond adequately to
changing conditions, if the closeness of ties prevents actors from seeking opportunities
beyond their circle of close relationships (Hannan & Freeman, 1984; Singh, 1986). The
42

balance of relationships in a network to maximize opportunities and minimize constraints
is therefore an important consideration for members of organizational networks. While a
closely-connected group of organizations may develop trust, consistent expectations, and
norms of reciprocity through their strong ties, the very redundancy and consistency which
promotes cohesion may also constitute a constraint. Without diverse ties, a closely
connected group does not have access to novel information or new sources of resources.
As with Granovetter’s (1973) job seekers, strong ties within a group must be balanced
with weak ties between groups to invite novelty opportunity into the network.
The potential for groups with too many strong ties and too few weak ties to
become isolated from the opportunities present in the wider environment is analogous to
the process of community closure described in ecological theory. As a set of
organizations deepens its relationships through multiple and overlapping ties, it may
experience the kind of buffering from the external environment described by Astley
(1985) (see pps. 33-34). This may have the positive effect of insulating the community
from environmental shocks, but risks shielding it from access to novelty and innovation
which may originate from outside the local network. Krackhardt and Stern (1988)
demonstrated the perils of community closure in the context of organizational work
groups. Drawing on the expectation that strong friendship ties among individuals within
their work group would make them more isolated from others in a firm, the authors
posited that groups in which most individuals had close connections within their work
group would be less effective in response to an organizational emergency than groups in
which individuals had more close friendships which crossed work group boundaries. To
explore this hypothesis, the authors created an external-internal or E-I index of group
43

cohesion, which summarizes the connections within and across groups as a ratio. The
results of experimental simulation demonstrate that work groups with more external links
were more resilient to crisis, while those groups with more internal links were more
autonomous, but less effective in responding to an unanticipated organizational
emergency.
Subsequent research by McGrath & Krackhardt (2003) found that a high index of
ties across subunits facilitated the cooperation that was necessary for coordinated action
across the organization as a whole, but that this arrangement of ties was unlikely to
emerge spontaneously, because individuals interacted most frequently with others in their
subunits, promoting stronger tie formation (friendships). These findings have also been
found to apply in the context of distributed work teams. Applying the E-I index in an
empirical study of collaborators on academic research papers, Gonzalez, Veloso &
Krackhardt (2008) found that research teams whose participants had a higher index of
collaboration across academic disciplines were more productive, generating more
academic papers. While the E-I index has been used to characterize the balance of
internal and external ties within organizations, the expectations which drive its
application to members of teams within organizational units are equally appropriate for
the analysis of organizations within project-based enterprises. The theoretical basis of the
E-I index presumes that frequency of interaction and closeness of relations within and
across groups shape the dynamics of tie formation. These factors also structure the
dynamics relevant to project-based ecological networks. Drawing on Scott (1998),
Bathelt, Malmberg & Maskell (2004) suggest that regional production systems require
the “right mix of local and nonlocal transactions and that strong growth can only result if
44

external markets are linked to the production cluster” (p. 41). In his empirical study of the
media production cluster around Leipzig, Germany, Bathelt (2005) provides evidence for
this claim, demonstrating that the inability of densely clustered local organizations to
make meaningful connections beyond Leipzig was associated with a lack of innovation
and growth in the cluster overall. It is expected that these outcomes will also apply in the
context of project-based ecological networks, and that striking the right balance between
internal and external ties will be advantageous for project-based organizational
communities.
Small World Networks
Project-based ecological networks are created as organizations forge ties with one
another in the context of projects in which the organizations are collaborators. Research
on the structures associated with collaborative networks reveals that, whether the
collaboration produces an academic paper (Newman, 2001), a tool for scientific research
(Huang, Huang, Margolin, Ognyanova, Shen & Contractor , 2010) or a Hollywood film
(Rossman, Esparza & Bonacich, 2010; Cattani & Ferriani, 20088), these networks share
the signatures of small world networks. These signatures include shorter than average
paths between nodes in the network, and a higher than average clustering of nodes
(Watts, 1999). Small world networks have densely connected clusters of nodes, which are
also connected to one another through few intermediaries, qualities which have been
associated with efficiency, creativity and innovation (Fleming & Marx, 2006; Iravani,
Kolfal & Van Oyen, 2007; Uzzi & Spiro, 2005). As Fleming, King and Judah (2007)
note, these qualities “enable dense and clustered relationships to coexist with distant and
more diverse relationships. The dense and clustered relationships enable trust and close
45

collaboration, while distant ties bring fresh and non-redundant information to the cluster”
(p. 938). Recent research by Uzzi and Spiro (2005) and Schilling and Phelps (2007)
illustrate these ramifications of small world structures for collaborative and creative
networks.
In their analysis of Broadway musical production networks, Uzzi and Spiro
(2005) demonstrate the effects of small world structures on collaboration in project-based
enterprises. They find that the levels of connectedness and cohesion across a creative
production network leads to the circulation of creative ideas and is associated with
financial success and critical acclaim. However, the positive effect of increased cohesion
has a limit. Above a certain threshold, intense connectivity among participants in project-
based networks was shown to produce homogenizing effects and to reduce the creative
and financial success of creative projects. The authors found an inverted U-shaped
relationship between small world characteristics and creative performance.
Schilling and Phelps’ (2007) examination of interorganizational alliance networks
in high-technology manufacturing also found that high degrees of clustering and short
path lengths among a wide variety of partners contributed to greater innovative output by
organizations across 11 industries. Schilling and Phelps’ study is of particular interest in
the context of project-based networks because their study examines a network, not of
individual contributors to projects, but of temporary alliances among organizational
actors, like the project-based enterprises which are the subject of this study. As in the
discussion of embeddedness above, their research suggests that too little and too much
small-worldness can be deleterious. For this reason, moderate levels of the small world
indicators are associated with positive outcomes for network members.
46

Core-Periphery Structures
As the discussion of small world networks suggests, the nodes and links in most
real-world networks are not equally or randomly distributed. Social networks of all kinds
demonstrate significantly higher clustering than has been found in simulated random
networks (Barabasi, Jeong, Neda, Ravasz, Schubert & Viseck, 2008). Social network
theories differentiate a number of plausible hypotheses regarding how this clustering
occurs in networks, including theories of homophily (McPherson, Smith-Lovin & Cook,
2001), which suggest that ties are more likely to be forged between actors sharing some
common trait or attribute (such as language, gender, age, etc.), theories of proximity
(Blau, 1977; Sorensen, 2003), in which ties are expected to be more common between
actors at smaller distances, and theories of preferential attachment (Barabasi & Albert,
1999; Newman, 2001), in which nodes having many links are more likely to continue to
add additional links, relative to less popular nodes.
One object of social network analysis is to examine how the aggregation of
pairwise decisions to form relationships results in global structures which affect the
network at large, as well as how these structures, in turn, promote or constrain
opportunities for future relationships. One such pattern of ties is described as a core-
periphery structure. This network pattern reflects a split between those organizations
which occupy important and advantageous central network positions and other
organizations whose ties position them in less desirable positions, making them
dependent on their more central counterparts.
47

As Borgatti and Everett (1999) describe, “the core-periphery model consists of
two classes of nodes, namely a cohesive subgraph (the core) in which actors are
connected to each other in some maximal sense, and a class of actors that are more
loosely connected to the cohesive subgraph, but lack any maximal cohesion with the
core” (p. 377). Barnett (2001) uses the metaphor of a wheel with spokes but no outer rim
to illustrate the core-periphery phenomenon. The arrangement of networks into core-
periphery patterns will have important consequences for the organizations that make up
these networks. Organizations in a network’s core enjoy a privileged position because
their relational ties provide them access to more diverse partners, resources, and
information. In their study of an industrial manufacturing community in Japan, Takeda,
Kajikawa, Sakata and Matsushima (2008) found a core-periphery structure dominated by
what they termed “hub” firms, which served to bridge different industrial sectors into a
cohesive network. Morrison’s (2008) analysis of a furniture manufacturing community
found that organizations in that network’s core experienced information advantages,
which allowed them to function as gatekeepers, securing important advantages relative to
other, more peripheral, organizations in the community. While core actors gain
advantage, organizations located on the network’s periphery are relatively
disenfranchised by their positions in the network, largely because they are structurally
dependent on few network ties for access to the core and its resources. For example,
research on international trade patterns shows that countries in the periphery of exchange
networks have greater incidences of poverty, weaker governments and institutions, and
tend to be more dependent on their core partners (Wallerstein, 2004; Fujita, Krugman and
Venables, 2001).
48

The advantages and disadvantages of core and peripheral network positions have
been demonstrated in a number of studies which examine clustering patterns in the global
economy, and find that activities associated with knowledge production, design, research
and development and distribution are concentrated in core network positions, while
activities associated with physical and material production and assembly are dispersed
across a network’s periphery (Castells, 2010; Hozic, 2001; Piore & Sabel, 1984). The
literature on economic globalization addresses this split in terms of the clustering of high-
value companies in wealthier nations, while non-knowledge-intensive production and
support processes are outsourced to branches or partners in low cost locations. This split
pattern has been documented in a number of industries, including the information
technology sector (Bardhan, Whittaker & Mithas, 2006), manufacturing (Gilly, Greer &
Rasheed, 2004) and entertainment (Hozic, 2001).
In addition to core and peripheral positions, some networks demonstrate a
semiperiphery, which may emerge over time as links are forged among peripheral actors
or organizations. In the global economic network, the emergence of a set of
semiperipheral countries is associated with geopolitical changes attending the decline of
colonial powers after World War One. Sometimes described as “emerging” economies,
research has found countries like Brazil, India and South Africa have emerged into more
semiperipheral positions in networks of global trade and commerce by forging trade
relationships with other peripheral states (Wallerstein, 1974; Chirot & Hall, 1982).
Research on the global telecommunications network documents the emergence of
semiperipheral clusters over time. A study by Monge and Matei (2004) found increases in
communication ties between peripheral states, promoting the emergence of semi-
49

peripheral clusters. A subsequent study by Lee, Monge, Bar and Matei (2007) using the
same longitudinal telecommunication data showed that the network demonstrated an
increase in tie formation between peripheral countries and that what appeared to be a
core-periphery pattern when examined at only one point in time, actually represented the
transition from a core-periphery structure to one with an emergent semiperipheral sector.
However, the global hyperlink network studied by Barnett and colleagues
(Barnett & Park, 2005; Park, Bennett & Chung, 2011) suggests a different possibility.
Barnett (2001) found evidence of a stable core-periphery pattern of ties within the global
telecommunications network between 1978 and 1996, as well as in global hyperlink
networks in 2003 and 2009. However, the 2009 hyperlink network had grown even more
centralized, suggesting that countries in the core experienced more growth and became
increasingly dominant relative to their peripheral counterparts. Interestingly, while the
2003 network showed more hyperlinks between peripheral states, and the existence of an
emergent semi-peripheral group, only two semiperipheral countries had maintained their
improved network positions in the 2009 network. Together, these results led Barnett and
colleagues to conclude that that the international hyperlink network became more
unequal over time. These divergent findings regarding the stability and direction of
change in core periphery structures are echoed in empirical analyses of organizational
communities as well. On the one hand, the existence and stability of core-periphery
structures in industrial networks is confirmed by Orsenigo and colleagues (Orsenigo,
Pammolli, Riccaboni, Bonacorsi and Turchetti, 1998; Orsenigo, Pammolli & Riccaboni,
2001). In this pair of papers about the biotechnology industry the authors find that the
majority of biotechnology organizations which began in a core position in the industry’s
50

interorganizational network in 1973 maintained this position though 1998, as did the
majority of organizations which began in peripheral positions. Despite increases in the
overall size and connectedness of the network over time, the prevailing core periphery
structure of the network was resistant to change.
On the other hand, research has also documented the emergence of semi-
peripheries in organizational communities, for example in creative industries like
television and film production. Cattani and Ferriani (2008) analyzed the network of core
crew members (directors, writers, producers, editors, cinematographers, production
designers and composers) involved in film and television production over a 10-year
period. Examination of the core-periphery structure in this collaborative network
revealed core, peripheral and semiperipheral groups of participants, and further showed
that the individuals and teams earning the most professional recognition were not those in
the network’s core, but those in the semiperiphery. The authors suggest that individuals
in core positions in this network are overembedded, and individuals in the periphery
underembedded. What is most interesting in this analysis is the observation that an
individual’s over- or under-embeddedness could be compensated by working on teams
that collectively have a balance of core and peripheral participants. These results indicate
that in the network of individual participants, there is a defined semiperiphery associated
with network advantages, and that network participants are able to improve their
positions through strategic tie formation which places them in the semiperiphery of the
collaborative network.
Taken together, organizational embeddedness, small world qualities and core-
periphery patterns facilitate our understanding of the structures which emerge in project-
51

based ecological community networks. Project-based collaborative networks are expected
to benefit from small world characteristics, as well as by achieving balance between
cohesion and embeddedness. The distribution of these qualities across a network
produces opportunities and constraints for network participants whose positions shape
their participation in the network.
The foregoing discussion draws together insights from economic development,
ecological and network theories into a framework for analyzing project-based industries.
Conceptualizing project-based industries as an ecological system allows us to explore
how these organizations and populations interact in their search for resources, and to
approach the question of how resources in the form of incentives impacts these
organizational relationships. Further, by examining these ecological systems as
interorganizational networks, we can better understand how these dynamics unfold when
organizations themselves are temporary and highly mobile. One such industrial context is
the American film industry, to which we turn in chapter two.

52

CHAPTER 2: THE AMERICAN FILM INDUSTRY AS A PROJECT-BASED
NETWORK ECOLOGY
The empirical context for understanding the relationship between incentives,
ecological patterns and network structures advanced in the first chapter is the
contemporary American film industry. The film industry is a project-based industry
characterized by the short-term assembly of teams of specialized workers and firms
organized to accomplish the production of a motion picture (Faulkner & Anderson, 1987;
DeFillippi & Arthur, 1998; Cattani, Ferriani & Baden-Fuller, 2007, 2009). Applying the
definitional framework of project-based enterprises, we can see that films are produced
by temporary organizations which are especially assembled for the execution of a
specific, limited task, and then terminated. As Ferriani, Cattani and Baden-Fuller (2009)
describe, while film production used to occur predominately under the aegis of a handful
of major film studios, it is now more common for independent production companies
“assuming primary responsibility for organizing overall production tasks” to work closely
with these major studios or distributors, or for independent producers to work
“autonomously, approaching majors as distributors rather than partners” (p. 5).
The Film Project Process
A film project begins with what Ferriani and colleagues call “project
entrepreneurs” or what DiFillippi and Arthur (1998) call “principal lieutenants”
assembling the necessary elements for production to take place. In contrast to traditional
firms in which most or all of the required materials and capital required for production
are owned and managed by the firm, in project-based enterprises these assets are
53

assembled ad-hoc (Turner & Keegan, 1999). This is the case in film-making, in which
elements are “rented” for the duration of a film production: labor is hired by the week,
hour or project (DeFillippi & Arthur, 1998), and financing is provided through cash-
flows according to a budget, which describes a bounded, project-specific financial
investment by backers. The physical assets associated with production are also assembled
on a short-term basis, with equipment, facilities, talent and locations hired for short-term
use (DeFellippi and Arthur, 1998).
As a film project ramps up, a relatively small group of individuals and
organizations participate in the planning, selection of staff and participants, and the
collection of the necessary physical, locational and other assets which constitute the
preproduction phase of a film. Physical production is the next and largest phase of the
filmmaking process. Physical production (also called shooting or principal photography)
engages the largest number of individual and organizational participants, and may take
place in one or a number of locations over a period of days or months (Wasko 2003;
Schatz, 2010). Although the term post-production suggests that they take place after
shooting, in fact, editing, the creation of special visual and computer-generated effects,
composition and inclusion of music, and the recording of additional scenes and dialogue,
usually begin simultaneously with filming, but extend beyond the conclusion of principal
photography (Lucas, 2011). Once all the requisite elements have been collected,
assembled, and delivered to distributors, whose job it is to arrange for the exhibition and
sale of a film, a film project is considered “wrapped” (Wasko, 2003).
While preproduction, principal photography and post-production engage different
individuals and organizations, and the size of a production grows and shrinks over time
54

as it moves through these stages, the entire process constitutes a single, bounded project,
which undergoes institutionalized termination once complete. As in other project-based
enterprises, the unfolding of different aspects of production over time means that
participants are generally involved in multiple, overlapping projects (DeFillippi and
Arthur, 1998; Bechky, 2006).
Divestment, Diversity, Diffusion
Contemporary film production in the United States has settled on this pattern of
project-based organization as a result of a series of historical and policy developments
reaching back to the 1930s and 1940s. What is sometimes described as Hollywood’s
“golden age” refers to a surprisingly brief period during which a small number of
vertically-integrated companies engaged in all aspects of film financing, production,
distribution and exhibition (Balio, 1996; Bordwell, Staiger & Thompson, 1985). During
this period, the creation of films followed a mass-production model, with each studio
managing the production, distribution and exhibition of films in-house (Gomery, 2008;
Schatz, 1997). Financial, technological and regulatory pressures (including the
Paramount Decree in 1948, which compelled the studios to restructure), led the big
studios at the end of the 1940s to begin to divest themselves of significant portions of
their production and exhibition operations, maintaining tight control over distribution
(Balio, 1996; Schatz, 1997; Scott, 2002; Waterman, 2005).
This strategy proved lucrative, as the studios rid themselves of the riskier
portions of the film production chain and focused instead on the acquisition, sale and
rental of finished products. This decision to focus on distribution had the consequence of
55

making the studios at once less monopolistic, as well as more powerful (Aksoy &
Robbins, 1992; Gomery, 2008; Wasko, 2003). Without the guaranteed market for films
they previously enjoyed as exhibitors, studios were justified in making fewer of them.
Restricting the supply of films in this way allowed distributors to charge exhibitors
higher rates. The studios simultaneously decreased the costs of production and exhibition
while increasing their revenues, cementing their positions as crucial gatekeepers in the
film industry’s value chain.
The collapse of the studio system also resulted in the emergence of interacting
populations of independent specialist organizations and a variety of contractual and other
relations among them (Christopherson & Storper, 1989; Caves 2000). The process of
transformation of the film production process from an activity occurring in integrated
firms to a project-based network of specialized contractors is well documented (Aksoy &
Robbins, 1992; Christopherson & Storper, 1986, 1989; Christopherson, 2006; Schatz,
1997; Scott, 1998, 2002; Veron, 1999). Christopherson (2002) elegantly summarizes
how regulatory changes in the 1940s and 1950s shaped the industry into what
Christopherson and Storper described as a “complex of firms which is tied together by
an elaborate structure of transactions, including exchanges of information, material input-
output flows and personal contacts” (1989, p. 331). In other words, the contemporary
American film industry can be described as a project-based ecological network (Jones,
1996; Alvarez & Svejnova, 2002).
The American film industry is a well-documented example of the industrial shift
in the late twentieth century from a dominant regime of mass production to a
disaggregated process Piore and Sabel (1984) have termed flexible specialization. The
56

authors define flexible specialization as the “inverse” of large-scale mass production, and
describe instead a dynamic series of interactions among highly specialized, independent,
small organizations. In an early test of the hypothesis that the film industry had become
flexibly specialized Storper (1989) found that both the number and diversity of
companies in the motion picture industry in Los Angeles had increased between 1960 and
1980. While the number of films produced over the course of this period remained
steady, the number of companies involved in their production nearly tripled, rising from
563 to 1473. This expansion is reflective of a transition from the studio era model of
vertically integrated production to a transactions-intensive network of independent,
interacting production firms. In addition, Storper (1989) showed that the companies
serving the industry also became more diverse, further confirming the hypothesis that
vertical disintegration of the studios led to the emergence and specialization of a
disaggregated set of interdependent companies. For example, though only two lighting
services firms served the industry in 1966, 23 did so in 1981, and the number of editing
houses rose from four to 113 over the same period.
In related research, Christopherson and Storper (1986) analyzed national
employment data to compare the degree of labor concentration in Los Angeles with that
elsewhere in the United States, finding evidence for concentration of the industry in Los
Angeles. The Bureau of Labor Statistics data Christopherson and Storper employ
demonstrate that employment in services allied to motion picture production has always
been highly concentrated in both New York and Los Angeles (Christopherson & Storper,
1986). These employment and establishment findings support the claim that vertical
57

disintegration in the late 1940s and 1950s promoted growth and specialization in the
industry, and that this growth was significantly concentrated in and around Los Angeles.
However, using data from trade magazines about the location of production
activity, Christopherson and Storper (1989) found that activity associated with physical
production, or filming, did not concentrate, but actually dispersed. Between 1960 and
1984, while productions which were shot in California dropped from 50 percent of
productions to 38 percent of productions, filming increased in New York and in other
locations across the United States. These results suggest that the growth of the film
industry followed the split pattern found in manufacturing industries by Piore and Sabel
(1984) and Bardhan, Whittaker and Mithas, (2006), in information technology (Gilly,
Greer & Rasheed, 2004) and in entertainment (Faulkner, 1987; Hozic, 2001).
As vertical integration gave way to contracting networks of autonomous
organizations and independent contractors engaged in project-based organization, the
geography of production bifurcated. As Leamer and Storper (2001) describe, “the 20
th

century tendency toward geographical fragmentation of the chain of production was
accompanied by the spatial agglomeration of certain parts of the chain, particularly the
intellectual, immaterial activities such as accounting, strategy, marketing, finance and
legal work” (p. 642).
Aksoy and Robins (1992) locate the industry’s real power in these “intellectual,
immaterial activities” as well as in distribution, and argues that these have remained
concentrated in Los Angeles, as Leamer and Storper suggest. As Hawley describes,
“because certain functions are by their nature more influential than others; they are
58

strategically placed in the division of labor and thus impinge directly upon a larger
number of other functions” (1986, p. 221). The power of distributors in this industry
persists through the present. The major Hollywood studios remain central in the financing
and distribution of films which are increasingly made by third party producers (Scott,
2002; Hozic, 2001). For example, of the top 25 grossing films in the US and Canada in
2011, 23 were distributed by the major studios (Warner Brothers, Paramount, Disney,
Universal, 20
th
Century Fox, and Sony) (Motion Picture Association of America, 2011).
While these majors continue to play a central role in the distribution, sales and marketing
of films, new players are also emerging, particularly independent and online distributors
many of whom take advantage of online exhibition platforms as an alternative to
theatrical distribution through a major studio (Squires, 2013; Parks, 2012; Pennington,
2011). Whether established or independent, these powerful distribution, marketing and
sales organizations constitute the high-capital link in the industry’s value chain, and
control the flow of resources in the industry. If distributors are dominant in the film
community, it is their relation to other organizational populations, and their command of
the resource space, that makes them so.
Deregulation and Spread
While regulation in the form of the Paramount Decision promoted the shift to the
industry’s characteristic flexibly-specialized organizational structure, Christopherson
(2002) links the spread of routine production activity to locations beyond Los Angeles
directly to the deregulation of the entertainment industries which took place in the 1980s
and 1990s. The full implications of this trend have developed over time.
59

During the 1980s and early 1990s, the regulations which had previously limited
cross-ownership in the industry were substantially relaxed (Baker, 2007; Holt, 2001;
McChesney, 1999). The big studios began to recapture their holdings in exhibition and to
deepen their production and distribution activities by acquiring major outlets in network
television, cable and the internet (Christopherson, 2002; Curtin & Shattuc, 2009). At the
same time, the studios began to expand into their current forms as multinational
entertainment conglomerates through international mergers and acquisitions, as well as
investment in lower-cost locations for production. It is in this context of deregulation and
expansion that US studios began to make more serious investments in mergers and
acquisitions which expanded their companies into what McChesney (1999) has deemed a
global media oligopoly. The studios also began in this period to seek out new locations
like Vancouver, where lower-cost labor, favorable exchange rates and more flexible labor
organizations combined to reduce costs of production.
FOX’s 1989 purchase of the Steven J. Cannell studios in Canada was instrumental
in establishing Canada as a profitable off-site location for the production of American
films. Writers about Cannell and the success of his media empire attribute it in part to his
efforts in promotion subsidy legislation which established Canada as a premier location
for US film and entertainment production (Carter, 2010; Gross, 1996). Following
Canada’s lead, and capitalizing on the development of inexpensive, lightweight and
mobile equipment, as well as declines in the cost of transportation which promote film
production mobility, other countries began to promote themselves as low-cost “satellite”
production centers for US film and media production including Australia, Mexico,
60

Bulgaria and the Czech Republic (Ulrich & Simmens, 2001; Friday, 2010; Elmer &
Gasher, 2005).
Incentivizing the Film Industry
Observing that the financial incentives to shoot in low-cost locations were
successful in recruiting American film productions to Canada, beginning in the late
1990s, economic incentives began to be offered by US states to promote alternative
domestic locations as sites for film production activity in the United States. Adapting the
use of tax shelter and tax incentive programs implemented by the Canadian Film
Development Corporation as early as the 1960s (Vipond, 2011; Bracey, n.d. ) the use of
tax instruments to promote domestic motion picture production in the United States
began in 1997 with Arkansas’ small film incentive, and was quickly joined by similar
incentives offered by Minnesota and Hawaii in 1998. Based on the success of early
programs, as part of the 2004 American Jobs Creation Act the Federal Government
introduced an incentive for investment in, and production of, films in the US, contained
in section 181 of the Federal Tax Code (Cohen & Kolstad, 2009). An important
consequence of the Federal Incentive’s passage was the legitimating effect it had on the
use of tax instruments to promote film activity. Following the enactment of the Federal
incentives, incentives offered by states took off rapidly, tripling from 8 states with
incentives in 2003 to 24 states incentivizing production in 2005. Another 9 states added
incentives in 2006, with new incentives tapering off thereafter. By 2010, 45 US states
offered incentives for motion picture production (Luther, 2010, Chianese, Codrova &
Rosenfeld, 2010). This pattern of a federal program promoting state-level policy adoption
is common (Walker, 1969; Welsch & Thompson, 1980), and follows the diffusion of
61

innovations pattern which characterizes adoption rates in all manner of arenas, from
communications technologies to farming innovations to inoculations (Rodgers, 2003
[1962]; Valente, 2005).
The explosion of state-level film industry incentive adoption over this period has
produced a cottage industry of tax and location specialists, brokers and advisers who
assist film producers in determining what locations offer the best creative and economic
advantages for filming. Indeed, the two major production services and payroll companies
in the industry, Cast and Crew and Entertainment Partners each provide software and
services which allow producers to integrate tax and other incentive assumptions into their
projections, and to directly compare the tax and other benefits of filming in one state
against another. Insurance, finance and lending organizations serving the film industry
now offer incentive-specific services covering everything from location-specific weather
insurance, to guaranteeing state-sponsored incentive funds (for example, Tax Credits
LLC, AON, The Film Incentives Group, and Gallagher Entertainment Insurance).
States use a broad range of instruments to promote industrial and economic
development. In the film industry, production incentives may take the form of rebates, tax
exemptions, tax credits as well as a variety of non-financial inducements (National
Conference of State Legislatures, 2011). Rebates (and grants) are cash sums which are
paid by the state to qualifying film productions to offset the costs of goods, services,
accommodation and wages purchased in the state (Chianese, Codrova & Rosenfeld,
2011).
62

States may also offer exemptions from taxes on goods, services, labor and hotel
occupancy as part of their incentives packages. For example, New York State’s
exemption language states that,
The creation of a feature film, television film, commercial and similar film and
video production is considered a manufacturing activity that results in the
production of tangible personal property. Accordingly, a person producing a film
for sale is afforded the same exemptions available to New York’s manufacturers.
New York’s manufacturing exemption covers purchases of machinery,
equipment, parts, tools, and supplies used in production. The exemption also
covers services like installing, repairing and maintaining production equipment
and the fuel and utility services used for production. In addition, goods and
services purchased for resale are exempt from tax. This means that film and video
productions get a sales tax exemption for all production consumables and
equipment rentals and purchases as well as related services. These exemptions
cover just about every aspect of film and video productions and postproduction -
from sets, props, wardrobe and makeup to cameras, lighting, sound, special
effects, edition and mixing (New York State Business Information Center).

These exemptions, and the logic that supports them, are widespread across US
states. New York’s description of its exemptions and their justification is consistent with
that found on the websites and in the promotional materials for the majority of
incentivizing states. These exemptions from tax liability are often joined by tax credits,
which promote spending by promising reductions in taxes due to the state. Virtually all
states offering incentives for film production do so through tax credits of between 5% and
45% of production spending in-state on goods, services and wages (Chianese, Codrova &
Rosenfeld, 2011). Film productions must meet specific criteria established by the state
film board or commission as a qualified production company through an application
process. The acceptance of an application results in the film commission authorizing the
63

state tax commission to issue a tax credit to be applied against the taxes owed by the
production company (Chianese, Codrova & Rosenfeld, 2011).
For a project-based industry in which subsidized companies are both temporary
and highly mobile, the use of these instruments raises two important concerns. First,
because a film production may spend only a short time in a particular state, its tax
liability may be quite small, making even a quite generous credit against tax liability
negligible in practice (Goodman, 2012). Second, film production companies are often
established as temporary limited liability pass-through structures, which are not expected
to exist as long-term or ongoing concerns (DeFillippi & Arthur, 1998). As a consequence,
states have introduced refundable and transferrable tax instruments. In a refundable credit
program, when a film production company’s tax credit is larger than its tax obligation the
state pays the difference to the company (Chianese, Codrova & Rosenfeld, 2011;
Goodman, 2012). Under a transferrable credit program, a production company earning a
tax credit in excess of its tax liability may sell that credit to another company, which may
then apply the credit against its own tax liability.
These structures have encouraged a number of lenders, ranging from banks to
industry-specific financial brokers, to offer advances to temporary film production
companies against the value of credits (Goodman, 2012). As well, tax incentive
exchanges which match sellers and potential buyers of tax credits have emerged in recent
years. Louisiana has been particularly aggressive in this arena, pioneering an online tax
incentive exchange (OIX, 2012; Goodman, 2012) which allows tax credit holders and
seekers to trade credits directly. As well, Louisiana and a handful of other states will
directly buy back incentive credits from productions at between eighty-five and ninety-
64

four cents on the dollar (Alan Bailey, personal communication). It should be noted that
the use of transferable tax credits is not unique to the entertainment industry but is also
common in other industries which feature large initial investments. The OIX website lists
among its sellers entertainment, information technology, real estate, and traditional and
renewable energy companies. Among its buyers are listed investment and commercial
banks, hedge funds, insurance companies and other, mostly large, organizations which
are permanently located and have high tax burdens (OIX About, 2013).
A final set of incentives sometimes offered by states to promote filming are non-
tax incentives such as the use of municipal services at cost or for free (ticketing,
permitting, police and fire services, use of municipal energy and water resources),
incentives for employing low-cost student labor, industry-specific job training, and loans
or guaranteed completion costs for independent or low-budget projects (Chianese,
Codrova & Rosenfeld, 2011). Services offered by state film commissions may also
include assistance in finding qualified crew, locations, and equipment providers, and
local “fixers,” who are tasked with liaising between visiting production and local
providers of goods and services.
These incentives represent a significant financial investment by states. A special
report on industry subsidy and incentive programs in the United States by the New York
Times counted $1.51 billion in incentives specifically targeted at the film industry in
2012 (Story, Fehr & Watkins, 2012). The total reported amount of incentives offered by
states for all industries in 2010 was $80.4 billion (Story, Fehr & Watkins, 2012). This
money, the Economist reminds us, is “volunteered on behalf of taxpayers,” an
arrangement which has raised criticism on two fronts. Tax incentive programs for the
65

film industry have been critiqued as being bad for the states which institute them, as well
as bad for the industry overall (Economist, 2011).
Incentive Program Evaluation
Those who critique the use of public funds by US states to incentivize the film
industry most often focus on the lack of evidence states are able to provide that their
incentive programs are economically sound. First, as Friday (2010) notes, detractors
claim that “little or no evidence can be presented to suggest that motion picture
incentives, which come at the expense of tax-payer dollars, actually pay for themselves in
aggregate dollar terms” (p. 25). In their evaluation of state incentive programs,
Christopherson and Rightor (2009) cite the challenges for those wishing to evaluate tax
incentive programs, including a lack of transparency by states, and states’ failure to
demand transparency from the productions they support, as well as the use of non-
standard models “using a set of assumptions about what kind of expenditures are made by
productions shooting in a state, and what the economic ‘ripple effects’ will be” (p. 3).
In their own economic assessments of incentive program effectiveness, states
incorporate multipliers which reflect the (usually positive) externalities attributed to tax
incentive programs. These are particularly difficult to incorporate in a comparative
analysis, or one that seeks to understand national trends. Miller (2012) describes
multipliers this way: “A multiplier summarizes the total impact that can be expected
from change in a given economic activity. For example, a new manufacturing facility or
an increase in exports by a local firm are economic changes which can spur ripple effects
or spin-off activities” (p. 1). In the context of film incentives, for example, states often
66

cite increases in tourism spending as a consequence of film incentive funding. A
promotional film by the Colorado Film Incentives agency describes the multiplier effect
this way:
Each dollar spent stimulates the economy as it changes hands: people buy more
stuff, and the stuff-makers hire more people to make more stuff. Film production
creates a massive influx of spending over a huge number of industries. This
tsunami-like ripple spreads out from the direct providers and affects all of the
ancillary industries: lodging, tourism, marketing, retailers, medical providers,
entertainment options, restaurants, real estate and on and on and on (Colorado,
2012).

The accounting of film industry multipliers is critiqued on several grounds. First,
although the economic multiplier for film production is reported at $1.92 (which means
that for each dollar given in rebates or credits or other incentives, an additional 92 cents
is generated in revenue for the state), this is comparable to the multiplier for hotels,
automotive manufacturing and power plants (Luther, 2010). As Luther elaborates, “while
South Carolina officials boast that their incentives program generated $2.38 in economic
activity for every dollar spent in 2006 and 2007, this is less impressive when one realizes
that many other industries achieve higher multipliers with invested funds” (2010, p. 10).
While incorporating multipliers may permit a state to more persuasively advocate for
renewal or increases in their incentives, doing so makes comparison across states a
challenge. As well, critics like Christopherson and Rightor (2009) suggest that non-
transparency, a lack of consistent measures for evaluation, and the use of inconsistent
assumptions and multipliers makes it difficult for outside auditors to replicate state
findings, or to conduct truly independent analyses. The evidence that does exist, they
claim, actually shows a negative impact for most states. The authors cite reports from the
67

state fiscal review authorities of Michigan, Rhode Island, Wisconsin and Connecticut, all
of which suggest that film incentives have negative effects on state balance sheets. For
example, Michigan’s incentive program was expected to cost $127 million in 2008, but
only to collect $10 million in income and sales tax earnings, and Rhode Island’s program
was calculated to lose 72 cents on every dollar spent on its tax program (Christopherson
& Rightor, 2009). The Connecticut Department of Community and Economic
Development found that “the state will not receive enough additional revenue from
increased [film related] economic activity to pay for the estimated $16.5 million in tax
credits applied for in 2007” (in Christopherson & Rightor, 2009, p. 3). The bill analyses
which are conducted by state economic departments and filed alongside the legislation
for these and other states often address these issues, which are argued to be short-term
costs of a long-term investment.
Programs which report gains in tax revenue and employment have been
challenged to prove that these are lasting effects, and not merely “momentary glitter,”
(Cieply, 2008). Critics claim that the newness, instability and lack of loyalty by producers
who are being seduced by increasingly numerous and attractive incentives all over the
country (and all over the world) make any gains in revenues or employment temporary.
Michigan Senator Nancy Cassis is on record saying that in her state “These are not long-
term jobs,” Ms. Cassis said. “If just one state offers more, they’ll be out of here before you
can say ‘lickety-split.’” (in Cieply, 2008). When asked for his assessment of the generous
incentive program in New Mexico, Representative Dennis Kintigh told an Albuquerque
Journal reporter, “I think we are losing money. I think it hurts the taxpayers” (in Friday,
2010, p. 97). The loss to tax payers is echoed by Friday (2010) who notes that limited state
68

budget coffers mean that money which goes to incentivizing the film industry is not available
for other “chronically underfunded” education and infrastructure programs. Using the
expressions “Hollywood handouts” and “welfare for the wealthy” to characterize the
incentive program in Louisiana, Mayer and Goldman (2010) challenge the claims that their
state sees employment and tax revenue increases as a consequence of its program. Further,
they argue that the state has “outsourced public resources needed to build a local film
industry” to private interests, including the tax brokers and companies who buy-back tradable
credits. Their concluding remark echoes critics of programs across the US: “The chase for
private sector capital in film production has generated profit, not jobs, and not a sustainable
film industry” (Mayer & Goldman, 2010, n.p.).
Another concern about the stability of gains in employment and revenues stems from
the legislative process itself. Detractors claim that the potential for upheavals in leadership
every two years, and the short terms of incentive programs which typically “sunset” after 3 to
5 years, contributes to the disloyalty of film producers. Without a consistent track-record of
support for incentives, states new to the game are particularly vulnerable to the location
whims of a fickle production community. Take, for example, Kansas, which introduced
incentive legislation in 2007. After just 18 months, the state suspended its program because
of allegations of reporting irregularities, criticism from constituents, and citing the need to
prioritize other programs in rebalancing its budget (Patton, 2010; Tax Foundation, 2011;
Center on Budget and Policy Priorities, 2010). While the program was reinstated in 2011,
industry insiders suggest that legislative see-saws of this kind may permanently harm the
state’s chances of landing future productions (Patton, 2010).
Instability of this kind has encouraged film insurance companies to expand the
packages they offer productions to include “incentives insurance” which covers productions
69

in the event of retroactive changes in legislation or in cases where a state is unable to pay
promised credits (Gallagher, 2012). Citing the case of Michigan which “changed overnight”
when Rick Snyder made reducing tax breaks and incentives for Michigan businesses his first
priority as the new Governor of the state (Bomey, 2011; Halpert, 2011; Linebauch, 2011),
Bob Jellen a broker at Gallagher Insurance and President of Entertainment Specialty
Insurance Brokers, said in an incentives information session held in Los Angeles in 2010,
that the mere announcement in February 2011 that Michigan’s program would be cut had “a
huge effect even before the new cap went into effect. The intention alone scared films away.”
A final point of contention for those opposed to tax incentive programs for film
regards the competition this has engendered among incentivizing states. Citing a “race to
the bottom” Mayer and Goldman (2010) and others have noted that, as new incentives are
offered domestically and internationally, states must offer more generous packages to
remain competitive for mobile production dollars. Adrian McDonald of Film LA, argues
that,
having all fifty states competing with each other is not only counterproductive,
but could prove to be financially devastating to numerous state governments
unable to sustain the huge amount of funds needed to pay for production
incentives. Any hope that film and television production will remain in states with
no history in the industry once production incentives cease is wishful thinking.
The race to the bottom, certainly in the United States, must end (McDonald, 2009)

The claim that interstate competition occasions an interstate incentive “arms race”
is echoed in a New York Times report calling US production hopefuls “the new
Bulgaria,” likening the completion for production dollars in the US to the attractiveness
of low-cost locations in Eastern Europe. This “economic war among states” (Moore,
70

2009) also includes critics in California, like McDonald and his not-for-profit
organization, Film L.A., which endeavors to bring production “home” to southern
California. As a state with a historically strong and productive film production industry, it
is argued that California especially stands to lose as a consequence of film production
moving to other US states and international locations. Reports commissioned by the
industry itself suggest that shooting days, employment and overall activity in Los
Angeles are down as a consequence of what some call “runaway production” or “film
flight” (Klowden, Chatterjee and Hynek, 2010). Citing “rival locations” as the cause, the
Monitor Company reported in 1999 that the number of films shot, as well as the state’s
share of movie and video industry employment have fallen dramatically as a consequence
of films choosing to shoot in international locations, or in US locations outside Southern
California. The report states, “Economic runaway film and television productions are a
persistent, growing, and very significant issue for the U.S.” (Monitor,1999, p. 4).
However, it should also be noted that the Monitor report was commissioned by the
Screen Actors Guild, and the Directors Guild of America, groups which represent the
industry’s creative labor. A similar report conducted by the Milliken Institute in 2010
found the same results, and warned that “a significant number of workers in film
production continue to maintain permanent residence in the state, but they spend
increasing amounts of time in other locations. If production losses continue, the workers
themselves will relocate altogether with increasing consequences for California’s
revenues and its pool of human capital” (Klowden, Chatterjee & Hynek, 2010, p. 2).
Indeed, the state of California has also begun to introduce its own tax incentives
to support filmmakers wishing to produce their films in southern California. The
71

introduction of California’s incentive in 2009 was promoted as a “targeted economic
stimulus package to increase production spending, jobs and tax revenues in California”
(California State Assembly Bill 2026, 2011). Offering $100 million in credits and rebates
annually, the incentive is argued not to “compete” with incentives from other states, but
rather to “stop the bleeding” (Vary, 2012). In fact, relative to the size of its industry,
California’s incentive program is so small that an annual “lottery” determines which of
the thousands of applications for incentive program rewards will receive them. As
Dominic Pattern of Deadline Hollywood explains, the 2012 lottery process looked like
this:
Producers and filmmakers had between 9 AM-3 PM to drop off applications
packets at the commission’s Hollywood Blvd office. Each packet was given a
lotto number. Those numbers will be put in a bowl, according to Film
Commission executive director Amy Lemisch, and picked out by an on-hand CHP
[California Highway Patrol] officer. Once the $100 million is used up, remaining
projects that didn’t get funding will be put on a waiting list. If already approved
projects drop off the winners’ list due to scheduling or production delays, those on
the waiting list will take their place and credits. The low-tech lotto method has
been used by the Film Commission annually to allocate the credits since 2009.
The idea is that a random process removes any advantage for any one company
(Patten, 2012).

Legislators in California have described tax incentive legislation as a retention
measure. California senator Ron Calderon, a sponsor of California’s tax incentive
legislation, has said, “When [Hollywood industry representatives] came to me four or
five years ago when we did our first tax credit, [they said], ‘Just give us a sign that you
want us here — give us some reason to stay here ….We wanted to stop the bleeding and
create some jobs, bottom line” (Vary, 2012). As well, Jason Sisney, of the California’s
Legislative Analyst's Office described California’s incentive as noncompetitive with
72

those offered by other states, saying “California's film and television industries are so
large that it would be cost prohibitive to have a very broad-based credit applying to most
productions. So (we have) this very limited one” (quoted in Longwell, 2012, n.p.).
Whether these incentives have been effective retention measures for California is
an empirical question remaining to be answered. While the Los Angeles county
Economic Development Commission estimated the program had produced $3.8 billion in
economic output and added over 20,000 local jobs, as in other states’ economic reports,
the calculation of multipliers and add-on effects makes evaluation of California’s
program difficult. However, in 2011, the California incentive was extended despite the
report of the Franchise Tax Board that the authorizing legislation would generate revenue
losses of $5.1 million in fiscal year 2014 to 2015, $22 million in fiscal years 2016-2017,
and as much as $161 million in losses “as a result of extended tax credit benefits”
(California State Assembly, 2011). California’s incentives program is subject to the same
criticisms levied at programs in other states: with ballooning state budget deficits and
cuts effecting transportation, education and infrastructure programs state-wide, “there’s
no denying it’s a tough time to afford any additional tax breaks,” as Klowden, Chaterjee
and Hynek, note (2010, p. 27). As well, California’ relatively healthy industry is itself a
sticking point for critics who do not interpret gains across the country as losses to
California in terms of employment, particularly. A critique of the California program
from the other side of the political divide is that the program doesn’t go far enough. As
one industry insider is reported to have said, “the pot is too small and the lottery system is
too random to be able to rely on, that’s what still driving production out of state” (in
Patten, 2012, n.p.).
73

Despite claims that they generate revenue, employment, and development of a
clean local industry, film incentives are roundly criticized by some constituents, scholars
and legislators even as they are applauded by others. Among the challenges for
evaluating these competing claims are the transparency issues raised by Christopherson
and Rightor (2009) and Mayer and Goldman (2010). In addition to the conflicting reports
issuing from motivated or commissioned sources (such as the industry labor
organizations’ sponsorship of the Monitor Report, research generated by interest groups
like the Tax Foundation, or the 2012 Motion Picture Association of America Report
conducted by Ernst and Young), disagreement about the interpretation of these reports is
rife. Friday (2010) states that “opposing sides on this issue draw radically different
conclusions from the same body of knowledge” (p. 95). Christopherson and Rightor
(2010) show that “advocates of film incentives find big job gains and strong multipliers,
while analysts such as those working for the US Federal Reserve Bank or for state
revenue departments and legislative find only modest job creation” (2010, p. 3). Still
other fiscal impact studies “overwhelmingly conclude that the subsidies have a negative
impact on state revenues, particularly if they take the form of saleable tax credits”
(Christopherson & Rightor, 2010, p. 2). These divergent interpretations of the same data
reflect the challenges described in chapter one about economic development program
evaluation.
How, then, can this complex system of incentive tools and their outcomes be
understood? The range and diversity of film production incentive programs, and the
affordances for production that exist in different states presents real challenges for
estimating the success of individual programs, and makes comparison across states
74

particularly difficult. Examining the system as an interdependent ecology, as well as
analyzing the temporary project networks upon which this mobile industry is built
represents a small but important step towards understanding the effects of incentives on
project-based industries, and the consequences for the states which offer them.

75

CHAPTER 3: HYPOTHESIS DEVELOPMENT: PROMOTING ECONOMIC
DEVELOPMENT IN PROJECT-BASED ECOLOGICAL NETWORKS
Examining project-based communities as ecological networks allows us not only
to describe collaborative organizational relationships and the structures which result from
them, but it also allows us to investigate in a systematic way the effect of economic
development programs which target these project-based organizations and industries. To
paraphrase Fosler’s description of third-wave economic development strategies only
slightly (see p. 17), states are now using the tools of economic development to target and
transform interorganizational network dynamics. By focusing on the development of
collaborative infrastructures and forging productive industrial communities, economic
development programs may be viewed as strategies of network interventions (Bartik,
2011). While traditional economic development theory hypothesizes direct effects of
economic development tools and strategies on outcomes, the incorporation of ecological
and network perspectives suggests that economic development tools will be effective to
the extent that they are offered in the context of specific ecological patterns and network
structures which are associated with development outcomes.
Economic development theory suggests a direct relationship between the
introduction of economic development tools like the film incentive packages described
here, and industrial activity, employment, and establishment of businesses in a state.
Research on the effects of incentives provides support for the view that the introduction
of targeted incentive packages by a state should promote activity, employment, and
establishment in that state.
76

Though differences among incentive programs make comparative analysis a
challenge, as Buss describes (2001), sufficient support for these programs exists to make
incentives extremely popular among policymakers. Buss argues that, while positive
effects of tax and other incentives are not definitive from a purely economic standpoint,
economic development spending is widely considered good politics. He argues,
“businesses receiving [incentives] are most supportive, whereas taxpayers funding them
are largely unaware or indifferent. There is little risk to politicians when incentives fail,
because failure can be blamed on economics, market forces or dysfunctional corporate
behavior. Political dividends during economic good times are great because policy
makers can claim credit for intervening” (p. 92). While challenges to comparative
program analysis abound, this study takes as a point of departure the traditional view that
the introduction of incentive programs will have positive effects on each of the dependent
variables:
H1a: Incentives targeting an industry will directly promote greater levels of
industrial activity in the state where the incentives are applied.
H1b: Incentives targeting an industry will directly promote higher rates of sector-
specific employment in the state where the incentives are applied.
H1c: Incentives targeting an industry will directly promote increases in the
number of sector-specific business establishments in the state where the
incentives are applied.
Evaluation of the effects of economic development often makes use of such a
diverse set of measures, multipliers, and subjective decisions about program goals that
77

consensus about program effectiveness is often difficult to reach (Hatry, et al., 1990;
Koven & Lyons, 2010). Comparative analyses of economic development interventions
across multiple jurisdictions are infrequent, and, because the mix of program tools and
goals is so varied, they run the risk of drawing erroneous conclusions based on mis-
matched or ill-fitting comparisons. It is theoretically unlikely, even if all jurisdictions
were to adopt precisely the same economic development strategy, oriented to precisely
the same goals, that jurisdictions would experience the same outcomes (Bradshaw &
Blakely, 1999). Variations in local affordances such as the availability and cost of local
labor, access to essential infrastructure, and availability of relevant markets would
produce significant unevenness in development outcomes. Since jurisdictions vary on all
these dimensions, adopt different portfolios of economic development tools, and offer
varying levels of support to target organizations and industries, the expectation of
economic development theory, and this research, is that the effects of incentive programs
on economic development outcomes will be uneven (Fosler, 1992; Reese & Fasenfest,
1997; Koven & Lyons, 2012).
As the ecological framework suggests, changes in resources available in one part
of an industrial environment may produce consequences which reach beyond that
environment, producing far-reaching effects across a larger industrial system (Rao,
2005). Insofar as economic development strategies represent changes in a state’s resource
environment, we might view economic development interventions as effective only
within that boundary because fiscal, tax and nonfinancial incentives will be available to
only those organizations operating within the boundary of that state. However, because
industries operate at a national level, the state-level network is itself embedded in a
78

larger, national set of relations which may also be affected by state-level intervention.
The effects of economic development policy in one place may reverberate throughout the
industry. Drawing appropriate boundaries represents a theoretical and methodological
challenge in the analysis of networks for precisely these reasons. This research, by
focusing on the relationships among organizations in industries targeted by economic
development strategies, permits us to examine networks at two levels: as state-level
networks, which include only those organizations subject to the economic development
tools offered by the state in which they are located, and at the level of the national
network, which comprises all organizations active in an industry, across states.
States which are home to more of an industry’s dominant organizations are also
expected to see gains in economic development outcomes. From an ecological
perspective, organizations which control the flow of resources in a community occupy a
position of power, and are advantaged by this aspect of resource control. Hawley (1986)
introduced the concept of ecological dominance by differentiating between key and
contingent functions in an ecological system. He argued that, for example, in agricultural
systems, key functions are often synonymous with the transformation of raw materials,
but qualified this statement to allow that the functions which are key in a given system
are “determined by the kind of environmental input that is of most critical importance to
the system. Where, for example a product of local resources is systematically traded for
the products of other systems, the key function is determined by the comparative
importance of production and of trade as sources of sustenance.” (p, 35). A dominant
population in a community may exercise control over the community’s resources
(Duncan, et al 1960), or may exercise power in the community (Friedland & Palmer,
79

1984), or both. Indeed, Hawley describes the relationship between resource control and
power as interdependent. As Audia, et al. (2006) describe “standing higher in the flow of
resources of a community itself creates hierarchy. Other actors standing downstream in
the resource flow depend on those above them, and are subject to extortionate demands
by dominant actors” (p. 394).
Populations which are dominant in a community have been found to grow more
rapidly than subordinate populations. For example Audia, et al (2006) find higher levels
of founding of instruments manufacturers in geographic locales where that population has
the largest share of resources. The authors also found that in locations where the
dominant population was unrelated to instruments manufacturers, this population suffered
lower rates of founding. The authors conclude that this may result from the lack of
legitimacy instruments manufacturers have as an organizational form in places where a
different dominant industry enjoys high legitimacy. For example, they cite the frustration
of software engineers in New York, who lament that the strong association of software
engineering with Silicon Valley and Boston undermines their own legitimacy.
The association between dominant populations and economic development
appears in the development and policy literature as the anchor-tenant hypothesis
(Pashigian & Gould, 1998). This argument derives from the strategy used in the
development of shopping mall and other consumer centers, of “anchoring” a community
of shops around large and highly recognizable tenants, which serve to attract traffic
which spills over to the other shops in the mall or area. Adapting this beyond the retail
case, Agrawal and Cockburn (2002) explore how the presence of industrial research and
development tenants serve to anchor emerging industrial communities and clusters. Their
80

analysis of university and private R&D tenants in the fields of medical imaging, neural
networks and signal processing provide limited support for the hypothesis that these
R&D organizations stimulate the development and cohesion of those industrial
communities. While anchor firms and dominant populations are not interchangeable,
Agrawal and Cockburn’s (2002) findings suggest that the concept of the anchor tenant
may be usefully applied in a community ecology context.
The presence of dominant populations is expected to generate the legitimacy and
founding effects described by Audia, et al. (2006). Because dominant populations control
the flow of community resources as well as enjoy positions of power, the presence of
organizations belonging to these dominant populations signals that a state enjoys some
degree of resource control, power or autonomy. Therefore, this research advances the
hypothesis that,
H2a: Increases in the number of industrially dominant companies in a state will be
associated with gains in industrial activity.
H2b: Increases in the number of industrially dominant companies in a state will
be associated with gains in sector-specific employment.
H2c: Increases in the number of industrially dominant companies in a state will be
associated with increases in the number of sector-specific business
establishments.
Project-based ecological networks are comprised of interacting organizations and
populations, the diversity of which has important consequences for the possible
81

collaborative networks that can emerge. As studies of research and development
networks demonstrate, innovation results from access to new knowledge, skills and
resources, which may be provided through connections to diverse organizational partners
(Powell, Koput & Smith-Doerr, 1996; Owen-Smith & Powell, 2004). As a network of
interacting participants, a project-based organization also requires the contributions of
many types of organizations to accomplish its goals. For example, in the advertising
industry, Grabher (1980; 2002) found that interorganizational diversity was associated
with creativity and project success. A state hosting only a single organizational
population will require the contributions of more partners from beyond the state to satisfy
the needs of a project than one which is home to a diverse set of organizational actors.
For example, Johns’ (2010) work on the Manchester television and film production
community found that the diversity of the production networks increased over time, and
that this increased the community’s ability to participate in more activities in the industry,
for example adding more preproduction activity over time. States with a more diverse
community constitute an environment in which more advantageous production networks
can be assembled. Research on clusters and agglomerations reaching back to Marshall
(1890) emphasizes the importance of diversity for economic growth and development.
The colocation of individuals and organizations leads to what Marshall and others have
described as spillovers. These knowledge “spillovers” are defined by Breschi and Lissoni
(2001) as a process through which “one or a few agents investing in research or
technology development will end up facilitating other agents’ innovation efforts (either
unintentionally as it happens when inventions are imitated, or intentionally, as it may
happen when scientists divulge the results of their research” (p. 975).
82

The Marshall-Arrow-Romer (MAR) theory of knowledge spillovers argues that
knowledge produced by one firm in an industry generates benefits for other firms in the
same industry, and that the effects of these externalities are especially acute when the
firms are collocated (Glaeser et al., 1992). Porter (1990) suggested that knowledge
spillover and innovation are the result of increased competition among proximate firms in
the same industry. Competition, Porter (1990) argues, is the specific mechanism which
promotes more rapid and consistent efforts at innovation by firms, as well as more rapid
adoption by firms of the innovations of others. This speaks to the commensalist
relationships within organizational populations.
Jacobs (1969), whose focus was on the diffusion of knowledge throughout cities,
emphasized the importance of organizational diversity within a location, and
hypothesized that innovation and growth result from a diverse set of organizations and
industries all interacting in a bounded geographical area. Jacobs describes the decline of
Manchester, where organizations were narrowly focused on textile production,
comparing it to the growth of Birmingham, a more industrially diverse community. She
concludes that regions benefit from hosting a diversity of industrial tenants, a hypothesis
that is also borne out in the economic analyses of city growth by Glaeser, Kallal,
Scheinkman & Schleifer (1992). In this study of 170 US cities which tests competing
hypotheses that specialization or diversity will be more predictive of growth, the authors
found support for the effects of diversity but not specialization.
By emphasizing the development of innovative and productive industrial
communities, contemporary economic development strategies also promote
organizational diversity as strategy for long-term growth (Brockelman 1999). For
83

example, Conceicao and Heitor (2001) conclude from their work on the role of
educational institutions in Europe that educational organizations beyond the traditional
university should also be targeted for participation in economic development campaigns.
They argue that perceived challenges to the institutional integrity of universities when
they participate in private-sector relationships could be ameliorated by including other
educational partners, such as trade and vocational schools. In their conclusions, the
authors recommend diversity, this time on the basis of the gains in legitimacy they expect
it to generate.
Incentive programs are expected to be more successful in those states where
organizational communities in an industry are more diverse, and include multiple
organizational populations relevant to the incentivized production community. This view
leads to the following hypotheses:
H3a: Higher levels of industry-specific diversity in a state’s organizational
community will promote greater levels of industrial activity.
H3b: Higher levels of industry-specific diversity in a state’s organizational
community will promote higher rates of sector-specific employment.
H3c: Higher levels of industry-specific diversity in a state’s organizational
community will promote increases in the number of sector-specific business
establishments.

84

Moving from ecological to network predictors of economic development
outcomes, empirical studies of embeddedness in organizational networks suggest that
underembeddedness is associated with a lack of opportunities and reliable information
which are common benefits of close ties and cluster participation (Kogut, 2000; Kogut &
Zander, 1992). On the other hand, overembeddedness carries risks of isolation and
insulation from innovations, knowledge and opportunities that might be provided by
more distant relations (Burt, 1992; Gulati, Nohria & Zaheer, 2000). Organizational
networks which have a balance of ties within and beyond their state should see more
positive development outcomes. Evans’ (1998) examination of the development of
biotechnology industries in Brazil, India and Korea led him to argue that in the
international context, only those states with a balance of global embeddednes and local
autonomy could become competitive on the global stage. Research on the development of
industrial clusters in advertising (Grabher 2002) and biotechnology have similarly shown
that the “buzz” of activity generated by the interaction of firms within a cluster (Bathelt,
Malmberg & Maskell, 2004, p. 38) is amplified when “network pipelines” also connect
these clusters to remote sources of innovation and information (Owen-Smith & Powell,
2002). As Bathelt, Malmberg and Maskell (2004) describe, there is an “intrinsic tradeoff
between too much inward-looking, and too much outward looking organizational
structure” in industrial clusters (p. 47). Following Krackhardt and colleagues’
(Krackhardt & Stern, 1988; McGrath & Krackhardt, 2003; Gonzalez, Veloso &
Krackhardt, 2008) insights about the benefits of a moderate ratio of external to internal
ties for network resiliency and productivity, this research adapts to a project-based
85

context the view that a balance of internal and external ties is advantageous for economic
development outcomes in a state. Specifically,
H4a: State-level organizational networks with a ratio of external to internal ties
closer to zero will enjoy higher rates of sector-specific activity.
H4b: State-level organizational networks with a ratio of external to internal ties
closer to zero will enjoy greater rates of sector-specific employment.
H4c: State-level organizational networks with a ratio of external to internal ties
closer to zero will be associated with increases in the number of sector-specific
business establishments.
Consistent with the expectations that moderate levels of embeddedness will
augment the benefits associated with economic development strategies (Evans, 1995;
Woolcock, 1998) , project-based industrial networks which demonstrate small world
characteristics are also hypothesized to account for some of the variation in economic
development outcomes. The small-world-ness of a network describes the influence of
clustering, path-length, and the interaction of these two network properties. As discussed
in chapter one, clustering is argued to promote creativity and innovation for reasons
consistent with the view of embeddedness: close, repeated and strong ties promote
“collaboration, resource pooling and risk sharing” (Uzzi & Spiro, 2005). The description
of third-wave development strategies, clusters and agglomeration echoes these
arguments, drawing on Marshall’s (1890) view that geographic clustering promotes
spillovers and positive externalities by creating the conditions under which a diverse set
of organizations and individuals are encouraged to interact. Kogut and Walker (2001)
86

make the connection between the network concept of the small world and localized or
clustered economic networks in their study of interorganizational ownership links in
Germany. They find that Germany’s business ownership network demonstrates small
world qualities, and that this quality permits Germany’s business network to remain
clustered nationally, as well as connected globally.
Path length, the measure of the average number of “steps” between any two nodes
in a network, describes the ability of organizational actors to reach one another. The
existence of just a few ties which cross clusters reduces path lengths significantly,
facilitating access to remote parts of a network. As Fleming, King and Judah (2007) put
it, “short path lengths in a network indicate that distant information – where distance can
be logical, organizational, or geographical – is surprisingly short in social space” (p.
941). A corollary of short path lengths is that connections across different clusters in the
network draws distinct components of the network into contact (Fleming, King & Juda,
2007; Watts and Strogatz, 1998). As clusters are brought together through cross-cutting
ties, formerly isolated portions of the network are incorporated into a connected
component (Watts, 1999).
Empirical research favors the evaluation of the interaction of path length and
clustering over their independent effects. As Fleming, King and Judah explain, “the heart
of a small-world and creativity argument lies not in the first-order effects but the in the
interaction of increased clustering and decreased path length. A small world network
should be more creative to the extent that its clusters are tighter and the path lengths
between clusters are shorter” (p. 942). However, consistent with the argument that both
ovembeddedness and underembeddedness produce negative effects, the benefits of small
87

world structures also reach a limit beyond which higher levels of cohesiveness and short
paths between nodes are counterproductive. The analyses of Fleming, King and Judah
(2007) and Uzzi and Spiro (2005) examine the effects of these small world structures on
organization-level outcomes, such as number of patents, financial success, and critical
acclaim. This research adapts this hypothesis to suggest that moderate levels of small-
worldness promote economic development outcomes.
H5a: State-level organizational networks with moderate levels of small world
structures will enjoy higher rates of sector-specific activity.
H5b: State-level organizational networks with moderate levels of small world
structures will enjoy greater rates of sector-specific employment.
H5c: State-level organizational networks with moderate levels of small world
structures will be associated with increases in the number of sector-specific
business establishments.
While hypotheses 1 through 5 address the effects of ecological and network
structures within specific states, the introduction of incentives by multiple states is also
expected to produce changes across the larger network constituted by the interactions of
companies across states, allowing meaningful comparison of the effects of policy
introduction. By seeking to relocate production and organizations from one place to
another, economic development strategies necessarily produce effects beyond the border
of an incentivizing state. The widespread use of expressions such as “brain drain,”
“runaway production,” and “outsourcing” by detractors to describe the potentially
deleterious consequences of incentives of this kind provides a colloquial reminder that
88

the promise of development in one area or sector carries with it the threat of decline in
another.
Moving from an examination of the project organization networks of individual
states to the interaction of these networks in an overarching national framework allows us
to explore higher-level consequences of local economic development interventions. The
analysis of a project-based production network in terms of its conformity to a core-
periphery pattern allows us to determine how processes in specific states contribute to
higher-level changes.
Efforts by legislators and economic development practitioners to relocate
organizations and industries are particularly challenging when target industries have an
abiding historical association with a particular place (Audia, Freeman & Reynolds, 2006).
In the United States, examples include the association of the automotive industry with
that sector’s once dominant cluster in Detroit, Michigan, the connection between carpet
manufacturing and the city of Dalton, Geogia, the link between the high-technology
sector and the Silicon Valley region of California, and the linguistic slippage which
makes Hollywood a descriptor, not of a neighborhood, but of a national, and even global,
media production industry (Hozic, 2001).
Studies of industrial location and agglomeration going back to Marshall (1890)
provide various theories of why industries took root and thrived in particular locations.
Location-specific production networks in industries of this kind are likely to demonstrate
geographic clustering, with a tightly connected set of organizations occupying a specific
geographic location (Cooke, 2002). However, these industries are unlikely to be
89

completely contained within a specific geographic boundary, and will have links beyond
the cluster, to suppliers of raw materials, to transportation centers, and to partners
covering a wide geographic space. To the extent that these far-flung partners do not
themselves engage in significant interaction with one another, a network of interacting
jurisdictions is likely to conform to a core-periphery pattern (Krugman & Venables,
1995; Smith & White, 1992).
Whatever factors contribute to the emergence of an identifiable industrial cluster,
and however long an industry remains rooted in a specific place, Hanson (1996) notes
that “while localization is a generalized phenomenon, the stability of industrial location is
not. Over time, activities tend to disperse from industry centers, giving rise to complex
patterns of industrial organization” (p. 256). Economic development programs seek to
promote this dispersion, and to influence the patterns that shape industrial organization.
In other words, state-level economic development programs which target geographically
clustered industries can be viewed as efforts to promote the degradation of core-periphery
structures, and to promote the emergence of industrial clusters in new places (Copes,
2001; Krugman, 1991). Legislators promoting their states as an attractive new place for
cluster development are, in effect, promoting the movement of their state from a non-
participant in an industrial network to a participant, or else promoting the movement of
their state from a peripheral to a semi-peripheral network position.
In research on the global division of industrialized labor, the core-periphery
distinction has been demonstrated to reinforce a power differential between centers of
production and centers of distribution (Fujita, Krugman & Venables, 2001). Because the
financial returns for distribution activities are much higher than those for production
90

activities, the dominance of core network positions are reinforced over time, as is the
structural dependence of peripheral states on those in the core (Baldwin & Forslid, 2000;
Krugman & Venables, 1995). This persistent finding suggests that, in concert with the
importance of hosting dominant organizational populations, the movement by a state
from a peripheral to a semiperipheral network position should also be associated with
positive economic development outcomes for that state. Specifically,
H6a: The improvement of a state’s position from a peripheral to a semiperipheral
position within the national network will be associated with increases in industrial
activity.
H6b: The improvement of a state’s position from a peripheral to a semiperipheral
position within the national network will be associated with increases in sector-
specific employment.
H6c: The improvement of a state’s position from a peripheral to a semiperipheral
position within the national network will be associated with increases in sector
specific business establishments.
Examining project-based industry through an ecological network lens allows us to
evaluate the efficacy of using traditional incentive programs to promote sector-specific
activity, employment and establishments in a project-based organizational context. The
hypotheses developed here argue that the success of economic incentives depends on the
diversity and composition of a state’s organizational community, as well as on the
structure and embeddedness of organizational networks within and across its borders. In
order to test these hypotheses, this study examines the networks of collaboration in the
91

American film industry and the use of tax incentive programs by US states which seek to
promote filming, film-sector employment, and the establishment of companies involved
in film production activities within their borders. The next chapter presents a detailed
account of the processes and analyses through which the hypotheses are empirically
tested in the context of the contemporary American film industry.

92

CHAPTER 4: RESEARCH DESIGN AND METHOD
The research presented here provides a much needed systematic understanding of
the effects of economic incentives which target project-based industries. The study
examines the role that ecological patterns and network structures play in that relationship.
The following sections spell out the data sources, measures and analyses which will be
used to test the hypotheses advanced in chapter three.
Data Sources and Measures
A primary source of data for the study is the Internet Movie Database (IMDB)
and its partner site IMDBPro. IMDB is an online accessible database of information
about film, television, videogame and other entertainment media products, as well as the
actors, crew and organizations involved in their creation. IMDB’s data are collected and
verified both by paid research staff, as well as by users of the site (IMDB, 2013). IMDB
data are drawn from on-screen credits, press-kits, interviews, autobiographies and the
trade press (IMDB, 2013). Data from IMDB have been used extensively in empirical
research, particularly to generate relational matrices for network analysis of actors and
creators of media content (Ansari, et al. 2000; Cattani & Baden-Fuller, 2007; Ferriani et
al. 2005; Hui & Png, 2002; Jensen, Neville & Hay, 2003; Rossman, Esparza & Bonacich,
2010; Zuckerman, 2005; Sorensen & Waguespack, 2006; Uzzi & Spiro, 2005).
The second major source of data is the Essential Guide to US and International
Production Incentives (Chianese, Cordova & Rightor, 2010), produced quarterly by
Entertainment Partners, a production services provider based in Burbank California. In
conjunction with their payroll and scheduling services, Entertainment Partners also
93

provides guidance for productions on selecting shooting locations based on up-to-date
information about the tax and other incentives offered by US states and international
jurisdictions. The Guide has been published since 2008, facilitating the comparison of
incentive benefits from year to year. In addition to the EP guide, an online directory of
incentive programs in the United States is maintained by the rival entertainment services
organization, Cast & Crew (Cast and Crew, 2013), which was also consulted for
information on incentive legislation. For years prior to those covered by these sources,
data are collected from the states’ departments of economic development, the states’ film
and television commissions, and from the states’ legislative records, which represent the
official record of the implementation and subsequent changes to film and television
policies. The relevant economic development, tourism or film offices for each state were
contacted to confirm the accuracy of incentives data culled from secondary sources.
Incentives are authorized by states through the legislative process. The current,
active film industry incentive legislation (House and Senate Bills) for each state was
pulled from the Cast and Crew incentives comparison database and examined. All House
and Senate bills include extensive information about when legislation was introduced, the
process of a bill’s adoption and the dates for which the legislation will be active. The
legislative record also includes extensive notes and citations which indicate what
previous legislation the current document replaces or modifies, making it possible to
track the legislative history of film incentive programs to their landmark legislation by
pulling previously enacted bills from the online archives of each state’s legislature. When
states introduce incentive legislation, the final legislative record normally includes a
specific allocation which funds the program. This is expressed as a dollar amount
94

committed by each state for each year. Data on the complete legislative history of film
incentives were collected for each state, and information about the dollar amount of the
allocation recorded.
Additional data for analysis are drawn from the United States Census Bureau
which collects and archives population and economic data about the United States and
individual states within it. Data on the size of state populations, the number of workers in
individual industries by state, as well as the number of establishments in individual
industries by state are publicly available through the Census Bureau’s online web portal
(US Census Bureau, 2012). Across the various databases it manages, the US Census
Bureau maintains a consistent classification system for categorizing economic and
industrial data, the North American Industry Classification System (NAICS)
classification scheme. The “NAICS is the standard used by Federal statistical agencies in
classifying business establishments for the purpose of collecting, analyzing, and
publishing statistical data related to the U.S. business economy” (US Census Bureau,
2012). The NAICS code for motion picture and video production industries is NAICS
5121, and “comprises establishments primarily engaged in producing, or producing and
distributing motion pictures, videos, television programs, or television commercials” (US
Census Bureau, 2012). Data on population size, state gross domestic product (GDP),
employment and establishments in NAICS 5121 were all drawn from the US Bureau of
the Census.
In addition to formal data collection, I also attended a number of industry events,
panels and demos offered by the Association of Film Commissioners International, Ease
Entertainment Services, Entertainment Partners, Media Services and the Producers Guild
95

of America’s FilmUSA working group. I also engaged in informal conversations with
representatives from film commissions, film production companies, and incentives
services companies which were helpful in developing an understanding about how
incentives are approached and experienced by practitioners.
Constructing the American Film Industry Project Network
The data sources described above were employed in the creation of a large,
relational database of films produced in the United States between 1998 and 2010. For
each year, all English language feature films produced and distributed within the United
States appearing in the IMDB database were recorded. For each of these films, data from
its IMDB entry were collected on all listed shooting locations, as well as the names,
addresses and screen credits of all companies which participated in a film’s production.
In all, 14,180 films met the initial selection criteria. After films for which no location data
could be collected were excluded, or for which location data did not include locations in
the United States, the sample of films was reduced to 9,056 films, on which 13,138
companies were collaborators. These data were used to construct 13 bipartite actor-event
matrices relating each film to the organizational actors who contributed to its production
(1998-2010). These bipartite graphs were subsequently transformed into single-mode
networks, in which two organizations which both participated in a film project were
linked by a non-directed tie. The networks were constructed and analyzed in R using the
“SNA”, “Network” and “iGraph” packages (Butts, 2012; Butts, Handcock, & Hunter,
2012; Csardi & Nepusz, 2012).

96

Variables
Independent Variables
Incentives. This research seeks to understand how the introduction of resources in
the form of economic incentives affects increases in activity, employment and
establishments, and the extent to which these effects are moderated by network structures
and ecological processes. While specific programs vary from state to state, the economic
development programs of all states targeting the film industry include a tax rebate or
credit as their primary tool of economic development (Chianese, Codrova & Rosenfeld,
2010). While a number of states also offer loans as well as non-economic incentives, this
research focuses on tax rebate and credit tools only, so that comparisons can be made
across all states. The variable reflects the dollar amount a state had committed to film
incentive programs in a given year. To facilitate interpretation in the analyses and for
improved readability of tables, the values are divided by $1 million, so that a $50 million
program appears as 50. A zero value indicates that no incentive existed in that year, or
that no funds were allocated to the program. A small number of states did not include a
“cap” on their incentive program, but left the question of funding to the discretion of the
film commissions and the state budget process. For the purposes of this study, incentives
with no cap were valued at $300 million dollars (300), an amount reflecting a ten percent
increase over the highest cap offered over the entire observation period. In order to
ensure accuracy, the sequence of legislative bills and the allocation amounts they
authorized were checked with officials in each state’s film commission.
97

Diversity. Information about the role an organization played as a participant on
film projects was collected for analysis. The raw data included 9,081 unique IMDB
credits which indicated the role a company played on a film production. This set of
credits was recoded into nine categories, according to conventions used by film
production and accounting professionals. These nine categories reflect distinct
organizational populations, and distinguish companies, first according to the part of the
film production process they serve. The literature on film production processes suggested
the initial distinction among parts of the production process: preproduction, production,
post-production and distribution (Wasko, 2003; Coe and Johns, 2004). Within these four
process-driven categories, an additional distinction was drawn based on the type of
contribution the company makes, or what the company’s role is on a film set. Three
dominant categories emerged: companies which provide facilities, companies which
provide services, and companies which provide equipment. The complete set of
categories include (1) preproduction, (2) producing and production office, (3) production
facilities, (4) production services, (5) production equipment, (6) post-production services,
(7) post-production facilities, (8) post-production equipment, and (9) distribution,
marketing, and sales companies. The coding scheme was discussed with three industry
professionals (a film industry finance and budget expert, a post-production supervisor,
and a film production coordinator) to confirm that the categories reflect commonly-held
distinctions by film production practitioners. Intercoder reliability was assessed post-hoc
by having two additional coders code a random 10% sample of the data (930 credits). The
intercoder reliability between the additional coders was excellent (Kappa = 0.909).
Intercoder reliability between each of these secondary coders and the codes applied by
98

the primary researcher was also very good (Kappa 0.871, and Kappa 0.881), indicating
very high reliability of the coding scheme. The number of distinct firm types found in
each state was counted, and that value divided by 9 (the total number of categories). The
resulting value reflects the amount of diversity in each state’s organizational community.
Dominant Population. In addition to organizational diversity, this study also seeks
to understand the role of dominant populations. In the film industry, companies involved
in distribution, sales and marketing are positioned dominantly in terms of their control
over the flow of resources to the community. This measure of the presence of industrially
dominant companies is calculated as the percentage of total companies in a state which
fall into the “distribution, marketing and sales” category described above.
External-Internal (E-I) Index. As developed by Krackhardt and Stern (1988), the
E-I index measures the relationship between the ties members of a group have within
their group and the ties they have beyond the group. The index is calculated in the
following way:
Number of External Ties – Number of Internal Ties
E-I Index = Number of External Ties + Number of Internal Ties

The result is a value between -1 and +1. A score closer to -1 reflects a structure in
which most ties are between individuals within a group. A score closer to +1 reflects a
structure in which most ties are external to the group. To evaluate the effects of resource
introduction in the form of state-level tax incentives and rebates, this study applies the
E-I index at the level of the state. Instead of intraorganizational group ties, I calculate the
E-I Index as:
99

Number of Interstate Ties – Number of Intrastate Ties
E-I Index = Number of Interstate Ties + Number of Intrastate Ties

A score closer to -1 reflects a structure in which most ties occur within the
boundary of the state, while a score closer to +1 reflects a structure in which ties mostly
extend beyond the jurisdiction of the state. Because a state’s balance is indicated by the
distance of the score from zero, the absolute value of the E-I index is used for analyses.
When a state did not have a network, or did not have connections within or across its
boundaries that state’s score is zero, which is an accurate, if somewhat misleading
interpretation of the E-I measure, requiring care in interpretation.
Small World. The small world measure is calculated by dividing the network’s
clustering coefficient by the network’s path length (Watts, 1999; Fleming, King & Judah,
2007). The small world measure was estimated by calculating the clustering coefficient
for the largest component of each state’s network (using the routine “Cohesion >
Clustering Coefficient” in UCINET 6.4), as well as the network’s average path length
(using the “Cohesion > Distance” routine in UCINET 6.4). The clustering coefficient was
divided by the average path length to determine the network’s small world indicator.
Plans to calculate the small world quotient on these networks presented a challenge: The
calculation of the small world measure is the clustering coefficient divided by the average
path length in a network’s largest fully connected component. For most state-year
combinations (604 out of 663), the networks are simply too small, and the small world
calculation produces a non-integer result. Of the remaining 59 networks for which
calculation of the small world measure is possible, 26 of these observations are in New
100

York and California, and the remaining 33 spread across a changing roster of states over
the 13 years of observation. Outside New York and California, small world measures are
too infrequent to make any reliable predictions about the effect of the measure on a
state’s filming, employment or establishment outcomes. Unlike the E-I measure, in which
no ties between states and no ties within a state reflects a true (if misleading) indicator of
network balance, imputation of the small world measure is not appropriate. The small
world-ness of a network for which the clustering coefficient cannot be divided by the
average path length is not simply missing, but without meaning: it is not possible to
interpret the arrangement of nodes and links in extremely small networks as more or less
small world-like.
Transitivity. In an effort to capture some of the theoretical significance of the
small world measure, the simple transitivity of each network was calculated. Transitivity
captures the number of triangles present in a network, and provides a loose
approximation of the numerator value in the small world calculation (Hanneman &
Riddle, 2005; Wasserman & Faust, 1994). Because the number of triangles is a count,
even small or sparse networks which do not meet the minimum threshold for producing
triangles can nonetheless be included in the model (a network with no triangles would
have a transitivity value of zero) (Latora & Marchiori, 2003).
Core-Periphery. To assess the extent to which the film industry network conforms
to a core-periphery structure in each year, continuous and discrete core-periphery
analyses were conducted on a projection of the national network which collapses activity
for each state so that each state is treated as a single node weighted by the strength of its
companies’ connections to those in another state. A continuous core-periphery
101

measurement, calculated using the “Core/Periphery>Continuous” routine in UCINET 6.4,
produces a global measure which reflects the network’s overall conformity to a core-
periphery structure. Normalized coreness values can be compared year to year to
determine how well the network fits a core-periphery pattern, and whether this changes
over time. Discrete core-periphery models (“Core/Periphery>Discrete”) produce
individual coreness measures for individual states, which are included in the analyses as
the state’s Coreness measure. Not all states are participants in the national network in
every year. These states are given the coreness score zero.
Dependent Variables
The hypotheses advanced in chapter three make predictions about economic
development outcomes. As described earlier (p. 22), economic development research
employs a wide range of dependent variables to assess the success of individual
programs. In order to make comparisons across programs, it is necessary to use measures
which are consistently recorded for all states across all years.
Industrial Activity. In project-based industries, the completion of a project or
product represents the goal around which organizational activity is coordinated. In the
case of the film industry, the completion of a motion picture production represents the
completion of a project. By limiting the analysis to films which have seen release, the
study includes distribution, sales and marketing activities as part of the industrial process.
Because the sampling method for selecting films for analysis described above excludes
films which were not released, only completed projects are included in the data. States
which are successful in promoting film production are expected to host more filming
102

activity. Industrial activity is measured as the number of films which were shot in each
state. One way to account for the location of production is to regard a film as being
“made” in a state if that state is listed among its shooting locations. While shooting
constitutes only one part of the production process in the creation of motion pictures, the
number of films shot in state is a reasonable measure of activity, particularly because
filming is required by states for productions to receive incentive benefits.
Employment. Levels of film sector employment for each state are available from
the United States Census Bureau for the years 1998 to 2010 inclusive. These data
represent counts of paid employees in each state in NAICS sector 5121: Motion Picture
and Video Industries.
Establishments. The United States Census Bureau defines an establishment as “a
single physical location at which business is conducted and/or services are provided”
(USCB, 2012). Data on the total number of establishments in NAICS 5121: Motion
Picture and Video Industries are available for each state from the United States Census
Bureau for the years 1998 to 2010 inclusive. Data for the state of Iowa were excluded
from all analyses because an ongoing series of criminal investigations into that state’s
incentive program prohibit verification of incentive program data (see Verrier, 2011).
Analyses
The hypotheses advanced in chapter three were tested using a series of negative
binomial mixed effects models. All modeling was performed with version 21 of the
statistical software package SPSS. To systematically examine the relationships between
incentives, ecological and social network variables and the dependent variables, separate
103

models were estimated for each dependent variable (activity, employment, and
establishment). Because the 50 cases (50 states, minus Iowa, plus the District of
Columbia) were observed at 13 points in time, mixed effects using repeated measures
with a time-varying covariate were modeled. This strategy is robust to data which exhibit
correlation and non-constant variability, as occurs with repeated measures (Little,
Pendergast & Natar, 2000; Tabachnik,& Fidell, 2006).
Because each of the dependent variables represent counts (number of films shot in
state, number of individuals employed, number of establishments), Poisson regression
models were explored. However, all three of the dependent variables in this study
demonstrate overdispersion, violating the assumption for Poisson models that variable
variance is less than or equal to the mean. Efforts to transform the data using log and
exponential normalization techniques were not successful in producing normalized
distributions (rejection of Kologrov-Smirnov and Shapiro-Wilk tests at p<.05), so models
using a negative binomial distribution were preferred. The negative binomial distribution
is also favored because of the assumption that Poisson count processes are “memoryless,”
or that a count value observed at one time is independent of the count value observed at a
later time (Walters, 2007). This assumption is violated in the repeated measures analyses
conducted here. Particularly for the dependent variables employment and establishments,
it is reasonable to expect that values in a given year are not independent of values
observed in the previous year(s). The negative binomial model addresses this dependence
by incorporating a random parameter which Gardner, Mulvey and Shaw (1995) describe
as reflecting “the uncertainty about the true rates at which events occur for individual
104

cases” (p. 399). The negative binomial model is favored for analysis of these data, which
violate both assumptions of the Poisson model.
A distinguishing feature of the negative binomial distribution is that the
dispersion coefficient is not constrained to zero, as is the case in a Poisson model. A
check of the appropriateness of a negative binomial model is the estimate of the
dispersion coefficient (Negative Binomial Parameter). Values greater than zero confirm
overdispersion in the dependent variable and the appropriateness of modeling under the
assumption of a negative binomial distribution. As well, the parameter’s 95% confidence
interval should not include zero. These requirements were satisfied for all models,
confirming that the negative binomial model is indeed appropriate. Model fitting began
with the unconditional (intercept-only) model and added controls, ecological and network
variables in a step-wise process. Each model was evaluated against the unconditional
model, and successive nested models compared with likelihood ratio tests.

105

CHAPTER 5: INCENTIVE, ECOLOGICAL AND NETWORK EFFECTS ON
ECONOMIC DEVELOPMENT OUTCOMES: STUDY RESULTS
Descriptive Results
Film is a dynamic and growing industry, which has considerably increased output
of feature length films, and has witnessed more modest gains in establishment and
employment over the 13-year period between 1998 and 2010. Figure 1 shows the number
of English language feature-length films which were produced and released in the United
States over the study’s observation period. The number of films made increased from
512 films in 1998 to a high of 1,701 films in 2009, before declining to 1,452 films in the
year 2010. The growth in filming is significant: the results of a simple comparison of the
mean number of films per state for 1998 and 2010 using a paired samples t-test is
significant (t = -3.66, p =.001). Figure 2 illustrates growth in film industry employment
between 1998 and 2010. While average employment in the film industry increases from
5,094 in 1998 to 6,087 in 2007 before declining to 5,797 in 2010, a comparison of means
using a paired samples t-test shows that this does not represent a significant change (t = -
.944, p. < .05). Figure 3 reflects the average number of film industry establishments in the
United States between 1998 and 2010. Gains in the number of establishments are also
modest, increasing from 393 in 1998 to 414 in 2007 before declining to 398 in 2010.
Paired t-tests on the mean number of establishments is non-significant (t= -.411, p. > .05)
indicating that the number of establishments by state does not change significantly over
the observation period.
106

Figure 1: Feature films produced and released in the United States, 1998 - 2010

Figure 2: Average employment in motion picture and video production (NAICS 5121), 1998 -
2010
0
200
400
600
800
1000
1200
1400
1600
1800
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
Feature Films, by
date of release
0
1000
2000
3000
4000
5000
6000
7000
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Average
Employment in
NAICS 5121
107

Figure 3: Average establishments in motion picture and video production (NAICS 5121), 1998
- 2010
Incentives
This study examines the effects of film production incentive programs on patterns
of activity, employment and establishment in the film industry. Incentives targeting film
production began to be introduced in US states in 1997, and were implemented in 45
states over the following 13 years. Figure 4 illustrates the pattern of incentive adoption in
the US between 1997 and 2010. The sharp increase in number of adoptions per year in
2005 and 2006 follows the introduction of a federal program incentivizing film
production in the US described on page 60.

340
350
360
370
380
390
400
410
420
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Average
Establishments in
NAICS 5121
108

Figure 4: Film industry incentive program introductions by US states, 1997 - 2010
The amount of money dedicated to these tax incentive programs by different
states varies widely. In 1997 Arkansas mounted a very small incentive program which
spent only $25,024 in its first year, and has never spent more than $26,747 in any fiscal
year. In contrast, Minnesota’s program, the second domestic program to be introduced,
spent $9,919,828 in 1998, and has spent between $5 million and $20 million annually.
The actual amount of money spent is not available for all states or years, but a good
indicator of the size of an incentive program is the allocation that the legislation earmarks
for a program. Generally, legislation specifies a total program budget, which may be
applied annually or may cover a specified number of years. In several states, legislatures
have elected to install incentive programs without specifying a cap on allocations
(Hawaii, Louisiana, Illinois, Maine, Massachusetts and Montana), and others have
reformed their policies to include a cap after initially introducing programs with
unlimited funds (Michigan, and New Mexico). Capped allocations range from less than
$100,000 dollars per year to New Mexico’s high of $274.9 million in 2008.
0
5
10
15
20
25
30
35
40
45
50
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Number of
Incentives
Introduced
Cumulative
Incentives
Introduced
109

Figure 5 shows the increase in the average dollar amount committed to incentives
between 1998 and 2010 (in millions of US dollars). It should also be noted that a very
small number of states have never offered film incentives, either because legislation to
create a program has never been introduced or passed (Delaware, Indiana, New
Hampshire, Nevada, Nebraska, and North Dakota) or because a program was established
through legislation, but was never funded (South Dakota).

Figure 5: Average value of incentives for film production (in millions of US dollars), 1998 -
2010
A preliminary look at the relationship between spending on incentives and
economic development outcomes is illustrated in Figures 6, 7 and 8, which chart the
average level of film incentives spending in the US against changes in industrial activity,
employment and establishments, respectively. Visual inspection suggests that incentive
spending appears to track closely with filming until 2009, at which point additional
incentive dollars do not promote additional filming activity. Incentives appear to have a
more modest relationship with employment, particularly after 2004, however additional
0
10
20
30
40
50
60
70
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Millions $
Average Incentives
1998-2010
110

incentive dollars do not continue to promote employment after 2008. Early incentives
appear not to have an effect on average number of establishments, until 2004, at which
point incentive spending and establishments track closely, until 2007, after which time
additional incentives do not appear to contribute to further increases in establishments.
Because of overdispersion in the dependent variables, zero-order spearman correlations
were calculated to evaluate the relationships between incentives and each of the
dependent variables over time. The rank-order relationship between incentives and
industrial activity was significant (p < .05, r = .358) as were the relationships between
incentives and employment (p <.05, r = .118) and incentives and establishment (p < .001,
r = .135). When the effect of time was partialed out, the correlations remain significant.
The rank-order correlation between incentives and activity decreased (p < .001, r = .209)
and the correlations between incentives and employment (p < .001, r = .138) and between
incentives and establishment (p < .001, r = .163) remain relatively stable.

Figure 6: Average incentives offered by US states and average number of films produced and
released, 1998 - 2010
0
5
10
15
20
25
30
35
40
45
0
10
20
30
40
50
60
70
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Millions
Incentive
Films
111

Figure 7: Average incentives offered by US states and average film industry employment

Figure 8: Average incentives offered by US states and average number of film industry
establishments
Ecological Factors
In addition to the effect of incentives, this research also considers the degree to
which ecological patterns and network structures promote filming, employment and
establishment of companies in the film industry. Two ecological factors are considered.
0
1000
2000
3000
4000
5000
6000
7000
0
10
20
30
40
50
60
70
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Millions
340
350
360
370
380
390
400
410
420
0
10
20
30
40
50
60
70
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Millions
Incentive
Establishment
112

First, the presence of dominant populations, companies that are involved in the higher-
profit end of the film production chain. In the case of film production these are
companies involved in distribution, marketing and sales activities. Figures 9 and 10 show
changes in the locational pattern of these types of companies across the US between 1998
and 2010. In 1998, of the 156 companies classified as distribution, sales or marketing
companies, 118 were located in California, an additional 14 in New York, and the
remaining 24 companies distributed across 15 states. In 1998 no distribution companies
were located in the remaining 33 states or the District of Columbia.
By 2010 both the number and range of states with distribution, sales and
marketing companies increased. Of the 424 companies belonging to this category, 272
were headquartered in California, and 54 in New York. The remaining 86 companies are
spread across 27 states. No distribution, sales or marketing companies were found in the
remaining 21 states or the District of Columbia. Over the observation period, 27 states
increased the number of these companies within their borders, with the largest increases
occurring in California (+154), New York (+40), Florida (+10), Arizona (+7), Illinois
(+6), Texas (+5) and in Michigan, Minnesota, New Jersey, Pennsylvania and Tennessee,
each of which added 4 companies in this population.
113

Figure 9: Companies in the dominant organizational population, 1998

Figure 10: Companies in the dominant organizational population, 2010
114

The presence of these powerful, high-value companies is joined in these analyses
by a measure of organizational diversity. Because project-based teams assemble
specially-skilled individuals and organizations, this research argues that the presence of
more organizational populations in a state should promote positive development
outcomes. As described previously (p. 90) all the companies extracted from the IMDB
data were grouped into one of 9 populations. Gains in the total number of populations
represented by the state’s company tenants between 1998 and 2010 were made by 33
states, among which Missouri added four new organizational types, and Virginia, Ohio,
New Mexico and Michigan each added three new organizational types over the study
period. Figures 11 and 12 illustrate these trends in the organizational diversity of the
states.

Figure 11: Organizational diversity by state, 1998

Figure 12: Organizational diversity by state, 2010
1
115
116

State Network Factors
The final set of predictors evaluated in this research concern the structure of
relationships among companies which collaborate in the production of films. The set of
companies engaged in film production collaborations and their relationships can be
profitably examined as social networks. When two companies both participate on the
same film project, they are considered to have a relationship, or a tie to one another. In
this study, company collaboration networks are analyzed at two levels: the networks of
companies which are co-located in particular states, as well as the national network which
is comprised of the interactions among companies across these different states.
In order to count as a network, a minimum of two companies must share at least
one connection. Even this low threshold for defining a network was not satisfied for
many states in the sample, particularly in early years of observation. For example, in
1998 only three states (California, New York and Washington) were found to have active
networks operating within their borders, and Washington’s network just meets the
minimum criteria with 2 companies engaged in a single common project. The number of
active networks within states reached a high of 18 for the years 2007 and 2008. Even
using the most liberal definition of a network, in 20 states and the District of Columbia,
networks are not observed at any point between 1998 and 2010. Among states with
networks, only California and New York’s are present in every year of the study. In the
state of Arizona, networks can be observed in 12 of 13 years, and in Texas networks are
observed in 10 of the 13 years. In Utah, North Carolina and Florida networks are in
operation in 6 of 13 years each. Figure 13 shows the number of states with an active
production network per year. The number of observed networks jumps dramatically in
117

concert with the introduction of the Federal Incentive in 2005 and the spike in
introduction of state incentives which followed.

Figure 13: Number of states with an observed active network, 1998 - 2010
Plans to calculate the small world quotient on these networks presented a
challenge: The calculation of the small world measure is the clustering coefficient
divided by the average path length in a network’s largest fully connected component. For
most state-year combinations (604 out of 663), the networks are simply too small, and the
small world calculation produces a non-integer result. Of the remaining 59 networks for
which calculation of the small world measure is possible, 26 of these observations are in
New York and California, and the remaining 33 spread across a changing roster of states
over the 13 years of observation. Outside New York and California, small world
measures are too infrequent to make any reliable predictions about the effect of the
measure on a state’s filming, employment or establishment outcomes.

0
2
4
6
8
10
12
14
16
18
20
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
US States
118

National Network Factors
In addition to evaluating the networks of companies within individual states, this
research also examines the larger national network which is comprised of the relations
among companies as they interact across state boundaries. The E-I index, (described on
pps. 42-43) estimates the extent to which a state’s network consists of activity within or
across its borders. Figure 14 illustrates the range of values for states in terms of their E-I
Index. States on the left-hand side of the bar chart are characterized by companies having
more interstate ties than ties within the state, whereas states on the right-hand side of the
bar chart have more ties among companies within the state than with companies beyond
the state. While zero scores suggest a “perfect balance” between internal and external
ties, closer examination reveals that these zero values for Maine, South Dakota and North
Dakota are actually a consequence of those states having no ties at all. The states with the
average measures of internal to external ties closest to zero are Oregon, the District of
Columbia, Utah, Florida, Nevada and Kentucky.

Figure 14: Average E-I values for US states, 1998 - 2010
1
119
120

A final assessment of the network structures associated with company
collaborations in the film industry concerns the extent to which, as a consequence of the
connections their companies make to other companies, a given state network is
positioned in the core or in the periphery of the national network. A traditional core-
periphery structure would resemble a wheel with spokes but no rim, in which one (or
perhaps a small group) of states having many connections to one another occupy the core
or center of the network. This state (or group of core states) would also be well-connected
to states beyond the core. If the network follows the core-periphery pattern, those
peripheral states would not have many direct connections to one another.
By taking the sum of all company-company connections between two states and
using that as an aggregate measure of the strength of the connection between the two
states, we can examine the national network of state interactions and assess its core-
periphery structure. The core-periphery algorithm in UCINET estimates the correlation of
the observed network data against an “ideal” core periphery structure in which core states
would be completely connected to one another as well as connected to peripheral states.
In the ideal pattern, peripheral states would be completely disconnected from all but the
core states. While the number and identity of the states active in the film collaboration
network changes from year to year, the best estimate of the core and peripheral members
of the network are remarkably consistent across the period of observation: between 1998
and 2007 only California and New York appear in the core of the network of states. This
means that many ties exist between California and New York, and that these two states
are each linked to many other states through collaboration on film productions. The
remaining states have very few connections to one another, revealing an imperfect but
significant core periphery structure. In 2007, California and New York were joined in the
115
121

core by Washington, Utah, Florida and Massachusetts. However, their membership in the
core is fleeting. Although Florida remained in the core of the national network in 2008,
by 2009 only California and New York remain. There is no lasting change in the core-
periphery structure of the network over the 13 year period of observation, and no
peripheral states develop sufficient ties to one another to constitute a semiperipheral
group.
Network visualizations of the network of states for the years 1998, 2007 and 2010
are presented in figures 15, 16 and 17 The dark-colored squares represent New York and
California, and the heavy black line between them reflects the larger number of
collaborative ties between the two states than exist between others. All remaining states
have a direct connection to California, New York or both. A few interesting connections
among peripheral states exist: for example in 1998, the ties between Tennessee and
Minnesota, Minnesota and Washington, and Washington and Wisconsin suggest the
potential for the emergence of a semiperpheral group, however, the necessary ties to
make this set of states cohere into a stable pattern of collaborative exchanges do not,
ultimately, materialize. A similar potential, given the periphery-periphery ties between
Delaware, Michigan, Utah and Florida also remains unrealized in later years. By 2007,
and later in 2010, more ties between peripheral states have emerged, but these are not
sufficient to alter the strong and consistent core-periphery pattern of California and New
York’s exclusive membership in the network’s core.

122

Figure 15: Aggregate relations between US states, 1998. Line weight reflects the number of
company-company connections between states
.

Figure 16: Aggregate relations between US states, 2007. Line weight reflects the number of
company-company connections between states
123

Figure 17: Aggregate relations between US states, 2010. Line weight reflects the number of
company-company connections between states

Another way of estimating the core-periphery structure of a network is to examine
the coreness scores of its individual nodes. The continuous coreness model estimates
normalized individual coreness of each state. This continuous model also estimates a Gini
coefficient of coreness inequality. If all states had the same coreness score, the inequality
in the data would be zero. If one state has a coreness score of 1 and all other states zero,
then the Gini coefficient would be 1. Equality in coreness scores theoretically ranges
between zero and 1, and in these networks ranges between .790 and .933, suggesting high
and persistent inequality. Figures 18, 19 and 20 provide another look at this inequality by
showing the individual coreness scores for the 5 states with the highest values, as well as
a summary value for the remaining states. The Gini inequality coefficient is also reported.
The year 2007 had the largest number of states participating in the network, as well as the
lowest Gini coefficient, though California’s dominance nonetheless remains clear.
124

Figure 18: Coreness scores for top five states, 1998

Figure 19: Coreness scores for top five states, 2007

Figure 20: Coreness scores for top five states, 2010
CA (0.996)
NY (0.082)
OTHERS (0.035)
WI (0.011)
MN (0.010)
MI (0.006)
24 states total
Gini Coefficient = 0.923

CA (0.976)
NY (0.211)
OTHER (0.162)
FL (0.018)
MN (0.015)
NM (0.015)
41 states total
Gini Coefficient = 0.887

CA (0.991)
NY ( 0.128)
OTHER (0.117)
PA (0.019)
TX (0.019)
FL (0.015)
40 states total
Gini Coefficient = 0.904

125

Tests of Hypotheses
The analyses presented below test the hypotheses advanced in chapter three,
which suggest significant relationships between tax incentive programs, ecological and
network predictors, and the dependent variables filming, employment and establishment.
In chapter three, hypotheses are grouped according to the predictor variable. For example
hypotheses 1a, 1b and 1c predict the effect of a single predictor (incentives) on each of
the three dependent variables (activity, employment and establishment). The procedure
for testing these hypotheses estimates three separate repeated measures mixed-effects
models with a time-varying covariate, one for each of the dependent variables. The
hypotheses presented in chapter three are summarized in Table 1, below, which regroups
them according to the dependent variable. This reflects the order in which predictors were
added to models, as well as the order in which results are presented in the remainder of
the chapter.

126

Table 1: Summary of hypotheses, organized by dependent variable

Dependent Variable - Industrial Activity (Filming)
H1a Incentives targeting an industry will directly promote greater levels of industrial activity in the state
where the incentives are applied.
H2a Increases in the number of industrially dominant companies in a state will be associated with gains in
industrial activity.
H3a Higher levels of industry-specific diversity in a state's organizational community will promote greater
levels of industrial activity.
H4a State-level organizational networks with a ratio of external to internal ties closer to zero will be enjoy
greater levels of industrial activity
H5a State-level organizational networks with moderate levels of small world structures will enjoy greater
levels of industrial activity
H6a The improvement of a state’s position from a peripheral to a semiperipheral position within the
national network will be associated with increases in industrial activity.
Dependent Variable - Sector-Specific Employment
H1b Incentives targeting an industry will directly promote higher rates of sector-specific employment in the
state where the incentives are applied.
H2b Increases in the number of industrially dominant companies in a state will be associated with gains in
sector-specific employment
H3b Higher levels of industry-specific diversity in a state's organizaitonal community will promote higher
rates of sector-specific employment.
H4b State-level organizational networks with a ratio of external to internal ties closer to zero will enjoy
higher rates of sector employment
H5b State-level networks with moderate levels of small world structures will enjoy greater rates of sector-
specific employment
H6b The improvement of a state’s position from a peripheral to a semiperipheral position within the
national network will be associated with increases in sector-specific employment.
Dependent Variable - Sector-Specific Establishment
H1c Incentives targeting an industry will directly promote increases in the number of sector-specific
business establishments in the state where the incentives are applied.
H2c Increases in the number of industrially dominant companies in a state will be associated with increases
in the number of sector-specific business establishments.
H3c Higher levels of industry-specific diversity in a state's organizational community will promote increases
in the number of sector-specific business estavlishments.
H4c State-level networks with a ratio of external to internal ties closer to zero will be associated with
increases in the number of sector-specific business establishments.
H5c State-level networks with moderate levels of small world structures will be associated with increases
in the number of sector-specific business establishments.
H6c The improvement of a state’s position from a peripheral to a semiperipheral position within the
national network will be associated with increases in sector specific business establishments.
127

Mixed Effects of Predictors
Summary statistics for the data including all states (except Iowa) are presented in
Table 2.The dependent variables industrial activity, employment and establishment all
represent counts, and demonstrate significant overdispersion, making negative binomial
mixed effects models appropriate for modeling these data.
Table 2: Summary statistics, all states except Iowa (n = 650)
Variable Mean Std. Dev. Min. Max.
Industrial Activity 23.50 60.82 0.00 597.00
Employment 5694.14 16154.51 175.00 150634.00
Establishment 396.58 912.62 27.00 6729.00
Incentives 27.13 78.61 0.00 300.00
GDP 2.38 2.85 0.00 18.00
Population 58.00 64.50 5.00 373.00
Dominant Pop.. 0.27 0.34 0.00 1.00
Diversity 0.15 0.17 0.00 1.00
E-I Index 0.26 0.33 0.00 1.00
Core-Periphery 2.48 13.93 0.00 99.80
Small World*
0.37 0.29 0.05 1.03
Transitivity
0.04 0.17 0.00 1.00
* n for Small World = 60

Model fitting began with the unconditional (intercept-only) model and added
controls, network and ecological variables one at a time in a step-wise process. Each
model was evaluated against the unconditional model, and successive nested models
compared with likelihood ratio tests.
As discussed above (p. 111) calculating the small world quotient was impossible
for all but a few isolated state-year combinations, including New York and California.
Unlike the E-I and Core Periphery measures, for which imputation is justifiable, for the
128

small world measure, imputation does not make theoretical sense. Unfortunately, this
means that too few small world values were available for analysis. This parameter cannot
be estimated in these models. As described above (p. 111) the simpler measure of
transitivity was calculated and used to capture some of the theoretical significance of the
small world measure. The measure of transitivity was included in negative binomial
models but its effect was never significant and did not improve model fit. Models
excluding the effect of transitivity are reported throughout.
Industrial Activity
Tax incentive programs targeted at the film industry are designed to promote
industrial activity. In this study, industrial activity is operationalized as the shooting,
completion and release of motion pictures. The primary requirement that must be met by
a film project team in order to take advantage of tax incentives offered by an
incentivizing state is that filming must take place in the state. Because in-state filming is
the minimum threshold which must be met for tax incentive eligibility, it is also expected
to see the largest gains as a result of incentives. Hypotheses 1a, 2a, 3a, 4a and 6a which
predict the effects of incentives, ecological and network predictors on industrial activity
were modeled by adding these variables in 4 nested models. Hypothesis 5a could not be
tested because of the problems with the measurement of this variable, explained on page
99.
The baseline model (Model 1) includes the control variables GDP and Population,
as well as the predictor variable incentives, which reflects the dollar value of incentives
offered by states. Model 2 adds the dichotomous indicator of whether a state offered an
incentive of any size (Incentive, dichotomous). In model 3, the ecological predictors
129

(presence of a dominant organizational population, and the diversity of the organizational
community) were added. The final model (Model 4) includes the control variables (GDP,
Population), ecological predictors (dominant population and diversity) and network
predictors (E-I index and Core-Periphery). The results of the analyses are summarized in
table 3.

Table 3: Effects of incentives, ecological factors and network factors on industrial activity, 1998 - 2010
Predictor Variables Model 1 Model 2 Model 3 Model 4
B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B)
Incentive and Controls
Incentive (dichotomous) 2.23 0.07 0.00 9.34 1.77 0.06 0.00 5.89 1.62 0.07 0.00 5.03
Incentive ($ millions) 0.01 0.00 0.00 1.01 0.00 0.00 0.42 1.00 0.00 0.00 0.95 1.00 0.00 0.00 0.70 1.00
GDP 0.74 0.15 0.00 2.09 0.49 0.06 0.00 1.63 0.27 0.06 0.00 1.31 0.41 0.06 0.00 1.51
Population 0.00 0.01 0.45 1.00 -0.01 0.00 0.00 0.99 0.00 0.00 0.07 1.00 -0.01 0.00 0.00 0.99
Ecological Predictors
Dominant Population 0.24 0.08 0.00 1.27 0.12 0.08 0.13 1.13
Diversity 3.14 0.26 0.00 23.05 3.44 0.26 0.00 31.12
Network Predictors
E-I Index 0.15 0.08 0.05 1.17
Core-Periphery -0.20 0.39 0.62 0.82
Log Likelihood -2543.83 -2118.06 -2029.69 -2005.82
Pseudo R2 0.35 0.46 0.48 0.49
Neg. Bin. Param. 1.61 0.40 0.27 0.24
Model LL -2543.83 -2118.06 -2029.69 -2005.82
Prev Model LL -3898.71 -2543.83 -2118.06 -2029.69
DifLL 1354.88 425.77 88.37 23.87
Δ DF 3 2 2 3
>ChiSq yes yes yes yes
UC LL -3898.71
UC Neg. Bin. Param. 23.38
Model 1: Baseline model with controls and incentives
Model 2: Baseline model with controls, incentives and dichotomous incentives indicator
Model 3: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors
Model 4: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors and network factors
1
130
131

Hypothesis 1a posited that incentives incentivizing an industry would promote
industrial activity in that state. The effect of incentives on filming activity was evaluated
by estimating the main effect of incentives (Incentives) on the number of films shot in the
state (Industrial Activity). Also included as controls in the first model were measures of
state wealth (GDP) and size (Population). In the first model, the dollar amount of
incentives is a positive and significant predictor of filming, (Incentives, B = .001, p <.00,
Exp (B) = 1.01). Thus, H1a was supported, though the effect size is small.
Because of the large number of states without incentive programs in early years, a
dichotomous indicator distinguishing between states with (Incentive, dichotomous =1)
and without incentives (Incentive, dichotomous = 0) was added in Model 2. The
dichotomous indicator was positive and significant in Model 2 (B = 2.23, p < .001) and
in subsequent models (Model 3, B = 1.77, p < .001, Model 4, B = 1.62, p < .001),
suggesting that the presence of an incentive program is a significant predictor of filming
activity in a state. The odds ratio for having an incentive in the best fitting model (Model
4) is Exp(B) = 5.03, indicating that states with an incentive are associated with an
increase in filming of just over five times, compared to states without an incentive. When
the dichotomous indicator is included, the model fit is considerably improved, but the
effect of the dollar value of incentives disappears (Incentives, B = 0, p = .42). While these
results suggest that the dollar value of money committed to an incentive program is less
important for promoting filming activity than is the presence of an incentive in any
amount, the post-hoc discovery of collinearity among the continuous and binary
incentives variables calls that interpretation into question. Variable collinearity may have
the effect of underestimating the contribution of the dollar amount of incentives in these
models (see p. 143 for a discussion of the post-hoc investigation of this issue). Overall,
however, these findings provide support for H1a.
132

Ecological predictors were added in model 3. Hypothesis 2a argued that the
representation of dominant (distribution, sales and marketing companies) in a state would
have a positive and significant effect on filming. The predictor is positive and significant
(Dominant Population, B = .24, p < .01, Exp(B) = 1.27), providing support for
Hypothesis 2a. These effects were modeled together with Hypothesis 3a, which predicted
that increases in the diversity of companies in a state would have a positive and
significant effect on filming. This parameter was also positive and highly significant
(Diversity, B = 3.14, p < .001, Exp(B) = 23.05) providing support for Hypothesis 3a.
Because the diversity measure is a ratio of how many company types are present in a
state out of the nine possible categories, I divide the Beta coefficient by nine and then
exponentiate the result to capture the size of the effect. An odds ratio of 23.05 means that
for each additional company type present in a state, filmmaking increases by a factor of
1.42 times. The dichotomous network indicator remains positive and significant, and the
effect of the dollar value of incentives remains non-significant (see H1a).
The network predictors, E-I Index and Core Periphery, were both included in
Model 4. To account for the large number of states without networks in the data, a
dichotomous indicator of network presence was also tried, but this parameter was never
significant and did not improve the model fit compared to models without the network
indicator. The more parsimonious models without the network indicator are presented.
Because values of the E-I index closest to zero are associated with gains, Hypothesis 4a
suggested that the E-I index would be negative and significant. The coefficient is
significant, but the direction of the effect is positive (EI, B = .15, p <.05, Exp(B) = 1.17).
Thus, H4a was not supported. As indicated earlier, H5a was not tested because of
problems with the measure. Hypothesis 6a predicted that higher coreness scores for a
state would be associated with gains in filming activity. The coefficient is non-significant
133

(Core Periphery, B = -.20, p = .39). Network predictors do not appear to have a
significant effect on filming, and H6a is not supported As well, when these indicators are
included, the effect of the percentage of distribution, sales and marketing companies
disappears (Dominant Population, B = .12, p = .13) leaving Hypothesis 2a unsupported.
However, in Model 4, which represents the best fit to the data and accounts for 49% of
the variance, the presence of an incentive remains positive and significant (Incentive,
dichotomous, B = 1.62, p < .001, Exp(B) = 5.03). The effect of organizational diversity
(Diversity, B = 3.44, p > .001, Exp(B) = 31.12) is also positive and significant. The best
predictors of filming in a state appear to be the existence of an incentive, and the
diversity of the state’s organizational population: states with an incentive in any amount
see an increase in filming of 5.03 times compared with non-incentivizing states, and the
addition of a new organizational population is associated with an increase in filming of
1.47 times. The final model provides support for hypotheses 1a and 2a but does not
support hypotheses 3a, 4a, 6a. Hypothesis 5a was not tested so no conclusion can be
drawn.
Employment
Proponents of economic development programs targeting the film industry argue
that sector-specific employment will increase with the use of economic development
instruments promoting in-state production. While film incentive programs do not require
that local labor be used on the films shot in-state, most states offer a higher level of tax
credits and exemptions to productions that hire locally. Local hires may also represent a
cost savings for productions over importing labor from out of state (Screen Actors Guild,
2012; Bryant, n.d.; Line Producing, 2013). The effects of incentives, ecological and
network predictors on film industry employment were modeled by adding these sets of
134

variables in 4 nested models which test hypotheses 1b, 2b, 3b, 4b, and 6b. Hypothesis 5b
could not be tested because of measurement problems.
The models testing the effects of predictors on employment were constructed
following the same pattern as those testing the effects on industrial activity, above. Model
1 included the control variables (GDP and Population) as well as the dollar value of
incentives offered by states (Incentives). Model 2 added the dichotomous incentives
indicator (Incentives, dichotomous). Model 3 introduced the ecological variables
(Dominant Population and Diversity). The final model (Model 4) included the control
variables (GDP and Population), ecological predictors (Dominant Population and
Diversity) and the network predictors (E-I Index and Core-Periphery). The results of the
analyses are summarized in Table 4.
Hypothesis 1b predicted that the introduction of sector-specific incentives would
promote employment in that state’s incentivized sector. I evaluated the effect of
incentives on employment by estimating the main effect of incentives (Incentives) on the
number of individuals employed in film sector work (Employment). Controls for the
wealth (GDP) and size (Population) of states were included. In the first employment
model, the amount of incentives is a positive and significant predictor of employment
(Incentives, B = .01, p< .001), but the effect size is small (Exp(B) = 1.01), providing
support for Hypothesis 1b.
The dichotomous variable indicating whether or not a state offered an incentive
(Incentive, dichotomous, =1/0) was added in employment Model 2. The dichotomous
indicator was positive and significant in Model 2 (Incentive, dichotomous, B = 7.02, p <
.001, Exp(B) = 1120.38), suggesting that the presence of an incentive is associated with
an additional 1120 individuals employed in film sector work. This finding provides
additional support for Hypothesis 1b.
135

When the dichotomous indicator is included, the model fit is considerably
improved, and the dollar amount of incentives remains significant and small, but the
direction of the effect becomes negative (Incentives, B = -0.0008, p = .01, Exp(B) =
.999). This suggests that the amount of money committed to an incentive program has a
negligible, but overall negative effect on employment. Larger amounts of money
committed to tax incentives do not appear to correspond to employment growth.
However, the simple presence of an incentive in any amount remains a positive and
significant predictor of employment. As in the previous model which predicted the
effects of predictors on industrial activity, the post-hoc discovery of collinearity calls into
question the interpretation that incentives in any amount are more a more significant
predictor of employment than the dollar value attached to those incentives (see p. 143 for
additional discussion of post-hoc analyses). However, the results of the dichotomous
predictor of incentives continue to provide support for Hypothesis 1b.
Ecological predictors were added in employment Model 3. Hypothesis 2b argued
that the percentage of distribution, sales and marketing companies in a state would have a
positive and significant effect on employment. The effects of the predictor is positive and
significant (Dominant Population, B = .41, p < .001, Exp(B) = 1.51), supporting
Hypothesis 2b. These effects were modeled together with the diversity measure.
Hypothesis 3b predicted that increases in the diversity of companies in a state would have
a positive and significant effect on employment. This parameter was also positive and
significant (Diversity, B = .56, p = .01, Exp(B) =1.75). Each additional organizational
population in a state is associated with an increase in employment of 1.06 times. The
dichotomous network indicator remains positive and significant, and the effect of the
dollar value of incentives becomes non-significant. Hypothesis 3b was supported.

Table 4: Effects of incentives, ecological factors and network factors on employment, 1998-2010

Predictor Variables Model 1 Model 2 Model 3 Model 4
B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B)
Incentive and Controls
Incentive (dichotomous) 7.02 0.05 0.00 1120.38 6.84 0.06 0.00 935.53 6.68 0.06 0.00 793.49
Incentive ($ millions) 0.01 0.00 0.00 1.01 -0.0008 0.00 0.01 0.999 0.00 0.00 0.09 1.00 0.00 0.00 0.23 1.00
GDP 5.18 0.55 0.00 177.47 0.00 0.05 0.94 1.00 -0.05 0.05 0.36 0.95 0.08 0.06 0.15 1.09
Population 0.12 0.02 0.00 1.13 0.02 0.00 0.00 1.02 0.02 0.00 0.00 1.02 0.01 0.00 0.00 1.01
Ecological Predictors
Dominant Population 0.41 0.06 0.00 1.51 0.29 0.06 0.00 1.33
Diversity 0.56 0.22 0.01 1.75 0.69 0.22 0.00 2.00
Network Predictors
E-I Index 0.19 0.06 0.00 1.21
Core-Periphery -1.39 0.39 0.00 0.25
Log Likelihood -7475.91 -5420.65 -5387.53 -5352.42
Pseudo R2 0.32 0.51 0.51 0.52
Neg. Bin. Param. 20.19 0.28 0.25 0.23
Model LL -7475.91 -5420.65 -5387.53 -5352.42
Prev Model LL -11042.56 -7475.91 -5420.65 -5387.53
DifLL 3566.65 2055.25 33.12 35.11
Δ DF 3 2 2 3
>ChiSq yes yes yes yes
UC LL -11042.56
UC Neg. Bin. Param. 5724.94
Model 1: Baseline model with controls and incentives
Model 2: Baseline model with controls, incentives and dichotomous incentives indicator
Model 3: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors
Model 4: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors and network factors
1
136
137

The network predictors E-I Index and Core Periphery were added together in
model 4. Because values of the E-I index closest to zero are associated with gains,
Hypothesis 4b predicts that the effect of the E-I variable will be negative and significant.
The coefficient is significant, but the direction of the effect is positive, against
expectations (EI, B = .19, p < .001). Hypothesis 4b was not supported by the results.
Hypothesis 5b was not tested because of measurement problems described earlier.
Hypothesis 6b predicted that higher coreness scores for a state would be associated with
gains in filming activity. The coefficient is significant but its direction is negative (CP, B
= -1.30, p <.001). Network predictors do not appear to have the hypothesized effect on
employment.
In this model, which represents the best fit to the data, and accounts for 52% of
the variance, the presence of an incentive remains positive and significant (Incentive,
dichotomous, B = 6.68, p < .001, Exp(B) = 793.49), as do the effects of the percentage of
distribution, sales and marketing companies (Dominant Population, B = .29, p < .001,
Exp(B) = 1.33) and organizational diversity (Diversity, B = .69, p > .001, Exp(B) = 2.0).
The existence of an incentive is associated with gains of 793 employed individuals in a
state, while a unit increase in the percentage of distribution, sales and marketing
companies in a state is associated with growth in employment by a factor of 1.33. The
diversity of the state’s organizational population is a much weaker predictor of
employment than it was for filming. Because the diversity measure is divided by nine to
capture the effect size the odds ratio 2.0 means that for each additional organizational
population a state hosts, employment increases by a factor of 1.08. The analyses showed
support for hypotheses 1b, 2b and 3b. Hypotheses 4b and 6b were not supported and 5b
was not tested.
138

Establishment
In addition to the effects of incentive programs on filming and employment, this
research also examines longer-term effects of incentives by modeling their effect on
establishments in the industry. Increases in establishments are a sign of developing
stability for an industry as it takes root in a new, or previously underdeveloped, location.
Hypotheses 1c, 2c, 3c, 4c, and 6c, which predict the effects of incentives, ecological and
network predictors on film-industry establishments were estimated by modeling these
variables in 4 nested models. Hypothesis 5c could not be tested because of measurement
issues. As in the analyses of the effects of these variables on activity and employment,
the first model included the control variables (GDP and Population) as well as the
variable Incentives, which captures the dollar value of state-offered incentives. Model 2
adds the binary incentives indicator to the previous model. Model 3 incorporates the
ecological variables (Dominant Population and Diversity), and Model 4 adds network
predictors (E-I Index and Core-Periphery). The full set of variables tested in Model 4
includes the controls (GDP and Population) the two measures of incentives (Incentives,
Incentives, dichotomous), the ecological predictors (Dominant Population and Diversity)
and network variables (E-I Index and Core-Periphery). The results of the analyses are
summarized in Table 5.
Hypothesis 1c predicted that the use of economic incentives in a state would
promote industry-specific establishments in that state. The effect of incentives on
establishments was assessed by estimating the main effect of incentives (Incentives) and
number of film industry establishments (Establishments). Controls for the wealth of a
state (GDP) and its size (Population) were also included. In establishment model 1, the
139

dollar amount of incentives was a positive and significant predictor of establishments
(Incentives, B = .001, p < .001) but the effect size was again very small (Exp (B) = 1.01).
The dichotomous indicator distinguishing between states with (Incentive,
dichotomous =1) and without incentives (Incentive, dichotomous = 0) was added in
Model 2. The analysis shows a positive and significant effects of the dichotomous
incentives indicator (B = 4.59, p < .001) in model 2 and in subsequent models (Model 3,
B = 4.42, p < .001, Model 4 B = 4.26, p < .001). As in the analyses of the other
dependent variables, the presence of an incentive program, regardless of its size, is also a
significant predictor of establishments. In the best fitting model (Model 4) the odds ratio
for offering an incentive of any size is Exp(B) = 71.08, indicating that states with an
incentive are associated with an increase in establishments by a factor of more than 71
times, compared with states that do not offer an incentive at all.

Table 5: Effects of incentives, ecological factors and network factors on establishments, 1998 - 2010

Predictor Variables Model 1 Model 2 Model 3 Model 4
B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B)
Incentive and Controls
Incentive (dichotomous) 4.59 0.05 0.00 98.82 4.42 0.05 0.00 83.20 4.26 0.05 0.00 71.08
Incentive ($ millions) 0.01 0.00 0.00 1.01 -0.0009 0.00 0.00 1.00 0.00 0.00 0.02 1.00 0.00 0.00 0.07 1.00
GDP 1.94 0.28 0.00 6.98 0.10 0.05 0.02 1.11 0.05 0.05 0.24 1.06 0.20 0.05 0.00 1.23
Population 0.06 0.01 0.00 1.06 0.01 0.00 0.00 1.01 0.01 0.00 0.00 1.01 0.01 0.00 0.00 1.01
Ecological Predictors
Dominant Population 0.36 0.06 0.00 1.44 0.24 0.05 0.00 1.27
Diversity 0.62 0.19 0.00 1.85 0.77 0.18 0.00 2.17
Network Predictors
E-I Index 0.15 0.05 0.01 1.16
Core-Periphery -0.88 0.32 0.01 0.41
Log Likelihood -5353.37 -3738.33 -3701.32 -3649.36
Pseudo R2 0.31 0.52 0.52 0.53
Neg. Bin. Param. 0.21 0.19 0.16
Model LL -5353.37 -3738.33 -3701.32 -3649.36
Prev Model LL -7749.56 -5353.37 -3738.33 -3701.32
DifLL 2396.19 1615.05 37.00 51.96
Δ DF 3.00 2.00 2.00 3.00
>ChiSq yes yes yes yes
UC LL -7749.56
UC Neg. Bin. Param. 8.20
Model 1: Baseline model with controls and incentives
Model 2: Baseline model with controls, incentives and dichotomous incentives indicator
Model 3: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors
Model 4: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors and network factors
1
140
141

The inclusion of the dichotomous indicator improves the fit of the model to the
data substantially, but the effect of the amount of incentives disappears (Incentives, B =
-0.0009, p =< .001). This suggests that the amount of money committed to an incentive
program has a negligible, but overall negative, effect on employment. Higher levels of
incentive investment are not shown to correspond to gains in establishments, however, as
in the previous two models, the post-hoc discovery of collinearity undermines this claim
(see p. 143 for additional discussion of post-hoc analyses). Because the simple presence
or absence of an incentive in any amount continues to be a positive and significant
predictor of establishments, Hypothesis 1c remains supported overall.
Ecological predictors were added in model 3. Hypothesis 2c argued that the
percentage of distribution, sales and marketing companies in a state would have a
positive and significant effect on establishments. The predictor is positive and significant
(Dominant Population, B = .36, p < .001, Exp(B) = 1.44), and provides support for
Hypothesis 2c. These effects were modeled together with hypothesis 3c, which predicted
that increases in the diversity of companies in a state would have a positive and
significant effect on establishments. This parameter was also positive and significant
(Diversity, B = .62, p < .001, Exp(B) =1.85), revealing that for each additional
organizational population present in a state, establishments increase by a factor of 1.07
times. Hypothesis 3c is therefore supported. While the effect of the dollar value of
incentives remains non-significant, the positive and significant result of the dichotomous
indicator provides continuing support for Hypothesis 1c.
The network predictors E-I Index and Core Periphery were added together in
establishment Model 4. Because values of the E-I index closest to zero are associated
with gains, Hypothesis 4c suggested that the E-I index would be negative and significant.
142

The coefficient is significant, but the direction of the effect is positive (E-I, B = .15, p
<.001, Exp(B) = 1.16), so Hypothesis 4c is not supported. H5c was not tested because of
measurement issues. Hypothesis 6c predicted that higher coreness scores for a state
would be associated with gains in establishment. The coefficient was significant, but
negative (CP, B = -.88, p = < .001), providing no support for the hypothesis. Network
predictors do not appear to have the hypothesized effect on establishments. However, in
this model, which represents the best fit to the data and explains 53% of the variance, the
presence of an incentive (Incentive, dichotomous, B = 4.26, p < .001, Exp(B) = 71.08),
the presence of organizations in the dominant population (Dominant Population, B = .24,
p < . 001, Exp (B) = 1.27) and organizational diversity (Diversity, B = .77, p > .001,
Exp(B) = 2.17) are all positive and significant. For each additional organizational
population present in a state, establishments increase by a factor of 1.08 times. The best
predictors of establishment in a state appear to be the existence of an incentive, the
percentage of distribution, sales and marketing companies, and the diversity of the state’s
organizational population, providing support for hypotheses, 1c, 2c and 3c. Hypotheses
4c and 6c are not supported.
Summary. The three statistical analyses (one for each dependent variable) of the
four models described above examined the effects of incentives, ecological and network
variables on the dependent variables industrial activities, employment and establishment
for all US states (except Iowa) and the District of Columbia. The dollar amount of
incentives had no persistent significant effect on any of the dependent variables, but the
simple presence of an incentive in a state was a significant predictor for all three
dependent variables. While the effect of the presence of an incentive is clear, the post-hoc
discovery of collinearity among the binary and continuous incentives variables means
143

that the effect of the dollar amount of incentives in these models may be underestimated.
Further analysis is required to tease out the additional effect of the dollar amount of
incentives on outcomes. For all dependent variables, the diversity of a state’s
organizational community was a significant predictor, and for employment and
establishment, the percentage of companies involved in distribution, sales and marketing
was also significant, though this relationship did not hold for industrial activity. This
suggests that the presence of dominant populations is more important for employment
and establishment than for production activity.
Effect sizes were largest for industrial activity (Incentive, dichotomous, Exp(B) =
5.03; Diversity, Exp(B) = 31.12), confirming the intuition that increases in filming are the
lowest hanging fruit. More modest effects were found for employment (Incentive,
dichotomous, Exp(B) = 793.49; Diversity, Exp(B) = 2.00; Dominant Population, Exp(B)
= 1.33) and establishment (Incentive, dichotomous, Exp(B) = 71.08, Diversity, Exp(B) =
2.17; Dominant Population, Exp(B) = 1.27).
Post-Hoc Analyses
Collinearity
The finding that the presence of an incentive program was a significant predictor
of all three dependent variables, but that the dollar value of those incentives was not, was
surprising, and prompted an investigation of potential collinearity between the two
incentives predictors. For the continuous incentives variable for every state-year, the
dollar amount of incentives was recorded. In contrast, the dichotomous indicator assigned
a 1 to any state-year combination in which any amount of incentive was offered, and a
zero to any state-year combination in which zero dollars of incentive were offered, or no
144

incentive program was in effect. Thus, the choice made here in modeling the construct of
incentive is similar to that of a dummy variable where the reference group is no incentive,
and the estimates include first the effect of any incentive, and then if there is an incentive,
what the impact of that incentive is on the outcome. Two issues arise: first, a large part of
the data is exactly predicted one variable to the next (no incentive = $0), raising the
potential for collinearity. Post-hoc tests of correlation revealed that this is indeed the
case. Parametric correlations are not advised because of the overdispersion of the data,
but the Spearman’s rank-order correlation between the dichotomous and continuous
variables was very high (r = .975, p < .001) confirming collinearity. In this case,
collinearity could have the effect of underestimating the effect of the continuous variable,
potentially giving a spurious non-significant result for the dollar value of incentives
(Hayes, 2005). This post-hoc finding calls the interpretation of the result into question,
and requires further investigation of the collinearity problem before the true effect of the
dollar amount of incentives can be definitely reported.
A second consideration is the fact that six states (Delaware, Indiana, Nebraska,
Nevada, New Hampshire and North Dakota) never offered an incentive in any year, and
therefore receive zeroes for all years in the binary variable. This might also bias the
distribution, making the effect of the amount of incentives more difficult to observe. An
alternative is to reevaluate the models with these six states removed. However, because
the number of cases in these analyses is the number of states, removing 6 of the 50 cases
(12%) results in a significant loss of power, preventing model convergence in all three
models. Taken together, these two issues make interpretation of the results of the
continuous incentives variable tenuous, pending resolution of the collinearity problem in
145

the present data, or testing the relationship between the dichotomous and continuous
variables on new data which do not demonstrate collinearity.
Outliers
Examination of the data revealed that California and New York are outliers: both
states host significantly more filming, employment and establishment activity than any
other state, as seen in Table 6.
Table 6: Dependent variable means for California, New York, and other states, 1998 - 2010

California New York Other Other Max. Other Min.
Mean Films 383.77 164.77 12.89 49.69 1.15
Mean Employment 112,513.31 24,956.46 3,040.73 335.54 14,180.46
Mean Establishments 6,241.69 2,373.15 232.10 30.77 1,064.54

The introduction of tax incentive programs in California and New York also
follows a different logic than incentive introduction in states which have not traditionally
had an active film sector, as described in chapter two. While other states seek to forge
and develop a local film production industry, and use incentive programs to achieve those
ends, in California and New York the incentives are more prophylactic in nature.
Incentives introduced elsewhere in the country, and particularly their aggressive increase
after 2004, have put these traditional production centers on the defensive.
In order to determine whether the outsized contributions of California and New
York generate significant changes in the pattern of prediction, sensitivity analyses were
performed by repeating the modeling strategy performed above on a restricted sample of
the data. An attempt was made to fit models to the data for only New York and
146

California. However, the limited range imposed by a model consisting of 26 observations
of two cases made model convergence problematic and did not produce usable results.
Instead, models excluding the contributions of California and New York (13 observations
of 624 cases), were analyzed.
Post-Hoc Descriptive Results
Because California and New York play such an important role in the industry, it is
valuable to understand whether their contributions produce changes in the pattern of
prediction. Revisiting simple descriptive statistics on filming, establishment and
employment, we find that the contributions of California and New York do make a
difference. Figure 21 shows the average number of films shot by all states (excluding
Iowa), as well as the average when New York and California and excluded from the
mean calculation. The pattern is remarkably consistent, but the levels change
dramatically. Figure 22 illustrates growth in film industry employment between 1998 and
2010. The solid line reflects average employment for all states (excluding Iowa), and the
dotted line reflects employment when California and New York are excluded. Figure 23
illustrates growth in film industry establishment between 1998 and 2010. Again, the solid
line reflects average establishments for all states (excluding Iowa), while the dotted line
reflects establishments when California and New York are excluded.
147

Figure 21: Average number of films produced and released, 1998 - 2010

Figure 22: Average employment in motion picture and video production (NAICS 5121), 1998 -
2010

0
5
10
15
20
25
30
35
40
45
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Average # Films
Shot Per State
Average# Films
Shot Per State
(Excluding CA and
NY)
0
1000
2000
3000
4000
5000
6000
7000
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Average
Employment in
NAICS 5121
Average
Employment in
NAICS 5121
Excluding CA and
NY
148

Figure 23: Average establishments in motion picture and video production (NAICS 5121),
1998 – 2010

A direct comparison of the statistical significance between models including and
excluding California and New York are precluded, due to the fact that one is nested
within the other. However, descriptive negative binomial models of the unadjusted linear
trends across time were computed, first on the data including California and New York,
and again with those states excluded from the analysis. The slopes across time were
compared between the two models by examining the percent change from the full model
(including California and New York) and the restricted model (full model – restricted
model / full). If the trends across time are comparable between models, it would be
expected that the intercepts would be different (because California and New York are
outliers), but that the slopes would remain similar.
Table 7 shows that for all three dependent variables, the intercept declines when
California and New York are excluded, consistent with expectations. An examination of
the beta coefficients for the effect of time show that while the percent change in the slope
0
50
100
150
200
250
300
350
400
450
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Average
Establishments in
NAICS 5121
Average
Establishments
NAICS 5121
Excluding CA and
NY
149

for industrial activity is small (14.6%) changes for employment (123.5%) and
employment (75%) are substantial. This suggests that the effect of the outlier states may
exert a considerable influence on employment and establishment outcomes. Further
analysis of these restricted data, with the full set of predictors, is appropriate.
Table 7: Percent change in intercept and coefficient for the effect of time with and without
California and New York
Dependent Var. All States Excluding CA & NY Percent Change
Indust. Activity
Intercept B 2.36 1.64 0.305
Time B 0.103 0.118 0.146
Employment
Intercept B 8.529 8.058 0.055
Time B 0.017 -0.004 1.235
Establishment
Intercept B 5.956 5.447 0.085
Time B 0.004 0.001 0.750

Incentives
New York introduced its incentive in 2004, whereas California did not begin to
incentivize film production until 2009. In the context of incentives overall, the
contributions of these two states is relatively small. Figure 24 shows the increase in the
average dollar amount committed to incentives between 1998 and 2010. The solid line
reflects the contributions of all states to incentives. The superimposed dashed line
illustrates the relatively minor contribution made by New York and California’s
incentives to the national average: New York’s incentive does not register as distinct
from the contributions of others until 2008. The large contribution of California (in the
amount of $100 million) does not take effect until 2009, at which point the impact of its
$100 million dollar per-year program becomes noticeable.
150

Figure 24: Value of film incentive programs by US states, 1998 - 2010
Ecological Factors
As the descriptive analysis of the ecological factors on pages 111-115 makes
clear, California and New York host the largest numbers of dominant population
organizations. Figure 25 shows how many companies in the dominant population
(distribution, sales and marketing) are located in California, New York and in other
states. Figure 26 shows the share of total companies in this population for California,
New York and all other states as a percentage of the total. Because the presence of
dominant firms is hypothesized to exercise a significant effect on the dependent
variables, and because California and New York are overrepresented in this regard,
accounting for the extent to which these states contribute to the effects of dominant
populations on outcomes is desirable.

0
10
20
30
40
50
60
70
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2020
Millions $
Average
Incentives 1998-
2010
Average
Incentives
Excluding CA and
NY
151

Figure 25: Average size of the dominant organizational population, 1998 - 2010

Figure 26: Share of total companies in the dominant population, 1998 - 2010
California and New York also have the most diverse organizational communities
among the states. As described in the descriptive analysis of the full set of data,
California and New York are the most diverse states in terms of the organizational
populations they host throughout the study period. California averages 8.2 of the 9
organizational populations over the study period, and never hosts fewer than 6
populations in any year. New York averages just under 6 of 9 possible organizational
populations, with no fewer than 4 populations represented in any given year. The next
0
50
100
150
200
250
300
350
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
California
New York
Other
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Average
%OTHER
%NY
%CA
152

most diverse states host between two and three organizational populations in a given
year, and the average across all states (except California, New York and Iowa) ranges
between 1.36 populations (in 1999) and 2.275 (in 2009). Because the degree of diversity
in a state’s organizational population is predicted to have a significant effect on activity,
employment and establishment outcomes, accounting for the extent to which California
and New York’s diversity influences the overall results is useful.
State Network Factors
As described on page 116 California and New York were the only states which
had active networks in every year of the analysis. This means that the calculation of
network measures for these states for the duration of the observation was possible. While
calculation of the small world measure was impossible on many of the state’s small
networks, for California and New York the small world measure is meaningful. As figure
28 illustrates, the small world quotient for California’s networks declines over time. After
starting at a value of .09 in 1998 the index increased to slightly over .12, and then
declined over the years, with a small spike in 2005 to a value of .06 in 2010. This may
indicate that California is becoming more small-worldly, even in the face of the
industry’s expansion outside the state.
153

Figure 27: Small world measures for California and New York, 1998 - 2010
An examination of the 95% confidence intervals for California’s small world
measure in figure 28 provides support by showing that the confidence intervals for the
small world values in 1998 and 2010 do not overlap. However, New York’s small world
values are consistently smaller than those of California, and are more variable. The
confidence interval for the small world measure is significantly higher in 2003 than it was
in 1998, but falls again, before a spike between 2006 and 2007. After 2007 the small
world measure for New York declines again. Looking just at the start and end of the
observation period, the overlap in confidence intervals in 1998 and 2010 shown in figure
29 reveal that New York’s small world measure has undergone substantial change during
the observation period, but with no lasting effect.
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
California
New York
154

Figure 28: Small world values for California, with 95% confidence intervals

Figure 29: Small world values for New York, with 95% confidence intervals
On the other hand, New York’s small world statistics are more erratic and
difficult to interpret. One possible explanation of this finding is that California’s network
is more cohesive than New York’s, an intuition which can be confirmed by comparing
the network visualizations of these networks. Figures 30 and 31 compare California and
New York’s networks for the year 1998. In that year, the largest connected component in
New York was made up of 8 companies, whereas the size of California’s largest
component was 211. Figures 32 and 33 compare the network again for 2010. The size of
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
155

New York’s largest connected component has grown to 91, and California’s to 258.
Besides being much smaller than California’s networks, while New York hosts many
pairs and trios of collaborators, the network has more disconnected components.
California’s network is more cohesive, overall, even though more collaborative groups
disconnected from the largest component are visible in the visualization of the 2010
California network.

Figure 30: Company-company collaboration network, New York, 1998

156

Figure 31: Company-company collaboration network, California, 1998

Figure 32: Company-company collaboration network, New York, 2010
157

Figure 33: Company-company collaboration network, California, 2010

National Network Factors
It is interesting to note that California’s E-I index is the highest among the states
at +.76, and that New York’s measure is -.18, which places it in the middle of the range
of states with a negative E-I. Changes in the E-I index for California and New York over
time are shown in figure 34, which shows that the two states’ networks are quite different
in terms of their balance of external and internal ties throughout the 13 years of
observation, with California’s values consistently above and New York' s consistently
below (or near) zero.
158

Figure 34: E-I Values for California and New York, 1998 - 2010
New York is observed to enjoy slightly more interstate than internal ties, with an
average E-I score of –0.18 over the 13 observations. In contrast, California’s average E-I
value is +.76, indicating a higher number of collaborative ties occur within the state than
between states. This difference in magnitude and direction of the E-I value between
California and New York (the states with the largest numbers of films, employed
individuals and establishments) suggests that while California and New York are both
outliers in the national production system, their networks have very different structures.
California’s network is characterized by intense internal connectivity, whereas New
York’s is constructed around interstate connections. However, California and New York
each make improvements in their E-I scores between 1998 and 2010. Figures 35 and 36
below show the 95% confidence intervals for the E-I values for each year for California
and New York. The lack of overlap in confidence intervals in 1998 and 2010 suggests a
significant change in each state’s E-I value. While the direction of change is different in
each state, in both cases movement is towards zero, or towards greater balance among
external and internal ties.
-0.6
-0.4
-0.2
0
0.2
0.4
0.6
0.8
1
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
California
New York
159

Figure 35: E-I values for California, with 95% confidence intervals

Figure 36: E-I values for New York, with 95% confidence intervals
A final look at the differences in California and New York’s E-I values is
presented in figures 37 and 38, which show the pattern of ties which occur within and
across the boundary of each state as a percentage of the total number of ties among film
production companies. These figures emphasize visually the differences between the two
states in their pattern of external and internal organizational ties.

0.5
0.55
0.6
0.65
0.7
0.75
0.8
0.85
0.9
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
California
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0
0.1
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
New York
160

Figure 37: Percentage of external and internal ties, New York, 1998 - 2010

Figure 38: Percentage of external and internal ties, California, 1998 - 2010
As described on pages 120 - 121, the core-periphery structure of the national
network is remarkably stable over the period of observation, with only New York and
California consistently in the Core of the network. The persistent dominant position of
these two states is a fundamental characteristic of the network structure. By excluding
them from the analysis, the huge amount of variance New York and California contribute
to the coreness scores is reduced substantially, as can be seen in Table 8. By excluding
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
External Ties
Internal Ties
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
External Ties
Internal Ties
161

California and New York, it is hoped that the variance in coreness scores among the
remaining states can be better estimated.
Table 8: Coreness statistics with and without California and New York

Post-Hoc Tests: Mixed Effects of Predictors
Because California and New York host so much filming, film-sector employment
and so many film-sector establishments compared with states in the rest of the country,
the analyses presented for the full set of 50 states above were repeated on data which
excluded California and New York. As discussed in the beginning of the chapter, both the
distributions of the dependent variables, and the theoretical argument about incentives
promoting film as a new or relatively small industry within the state, suggest that the
contributions of California and New York constitute a separate, and potentially
confounding, set of processes. Table 9 lists the summary statistics for the variables when
California and New York are excluded from the sample.

Variable Mean Std. Dev. Min Max
Core-Periphery (with CA & NY) 2.48 13.93 0 99.8
Core-Periphery (excluding CA & NY) 0.26 0.48 0 3.1
162

Table 9: Summary statistics, excluding California and New York (n = 624)
Variable Mean Std. Dev. Min. Max.
Industrial Activity 13.05 14.56 0.00 100.00
Employment 3067.45 2977.66 175.00 16507.00
Establishments 233.63 212.60 27.00 1191.00
Incentives 26.96 79.19 0.00 300.00
GDP 1.96 1.86 0.00 11.00
Population 49.05 45.32 5.00 251.00
Dominant Pop. 0.27 0.35 0.00 1.00
Diversity 0.13 0.13 0.00 0.56
E-I 0.26 0.33 0.00 1.00
Core-Periphery 0.26 0.48 0.00 3.10
Small World* 0.58 0.20 0.21 1.03
Transitivity 0.03 0.16 0.00 1.00
*Small world n = 34
The hypotheses were retested on the restricted data set by following precisely the
same procedures used to test the full data set. For each dependent variable, four nested
negative binomial regression models were fitted. Model 1 included the control variables
(GDP and Population) and the dollar value of incentives offered by states (Incentives).
Model 2 added the dichotomous incentive variable distinguishing states offering an
incentive of any size from states without an incentive program of any type. Model 3
added the ecological variables (Dominant Population and Diversity) and the final model
(Model 4) included control variables (GDP and Population), both incentive variables
(valued and dichotomous), the ecological predictors (Dominant Population and Diversity)
as well as the network predictors (E-I Index and Core-Periphery).
Industrial Activity
The exclusion of California and New York from the mixed effects model of the
dependent variable industrial activity does not provide significantly different results from
those which include the two dominant states. Results of the analysis are summarized in
163

Table 10. In the final model estimating the effects of the predictors on industrial activity
(this time excluding California and New York) the direction and significance of the
effects of incentives and the ecological variables are the same, though the odds ratios
shift slightly. When California and New York are excluded, the odds ratio of the
dichotomous incentives effect falls from 5.03 to 4.70, and of the effect of diversity from
31.12 to 25.75. The effects of the network variables are less consistent. The significance
of the E-I variable increases from p < .05 to p < .001, and the odds ratio increases from
1.17 to 1.25, but the sign of the effect is positive (against the expectations of the
hypothesis). The Core Periphery parameter becomes significant when California and
New York are excluded (p < .001), but again, the sign of the effect is negative, against
expectations. The interpretation of the results in terms of the hypothesis tests remains the
same: significant predictors of filming in a state include the existence of an incentive (but
not its dollar value) and the diversity of a state’s organizational population. As in the
model including California and New York, the statistical analyses continue to provide
support for hypotheses 1a and 3a but Hypotheses2a, 4a and 6a are not supported.
Hypothesis 5a was not tested because of measurement issues.
Employment
As we learned from the results of the models predicting industrial activity,
excluding California and New York from the mixed effects model of the dependent
variable employment does not produce substantively different results from those
including the two dominant states. Results are summarized in Table 11. In the final
employment model excluding California and New York the effects of incentives and the
ecological variables are positive and significant, though the odds ratios shift slightly from
164

the models with all states included. The exclusion of California and New York does not
substantively change the log odds for the binary incentive variable, but does cause the
effect diversity to increase (Exp(B) for all states = 2.0; Exp(B) excluding California and
New York = 4.53). The effects of the network variables remain inconsistent with the
hypotheses. For these variables, when California and New York are excluded network
predictors are non-significant, and the direction of the effects remains in the opposite
direction of that hypothesized. The results of the analysis excluding California and New
York are not substantively different from the models using data from all states: the
significant predictors of employment in a state include the existence of an incentive (but
not its dollar value), the percentage of distribution, sales and marketing companies in the
state, and the diversity of a state’s organizational population. However, the increase in
the effect size of the diversity parameter when California and New York are excluded
suggests that the effect of diversity is particularly important outside traditional production
centers. Support for hypotheses 1b, 2b and 3b are confirmed, but Hypotheses 4b and 6b
are not supported. Hypothesis 5a was not tested because of measurement issues.
Establishment
The exclusion of California and New York from the mixed effects model of the
dependent variable establishment does not produce significantly different results from
those obtained when California and New York are included. The results are summarized
in Table 12, In the final establishment model in which California and New York are
excluded from the analysis, the direction and significance of the effects of incentives and
the ecological variables are the same, though the odds ratios shift again. The exclusion of
California and New York does not significantly change the log odds for the binary
165

incentive variable, but does cause the effect size of diversity to increase (Exp(B) for all
states = 2.17; Exp(B) excluding California and New York = 3.45). The effects of the
network variables remain inconsistent. For these variables, when California and New
York are excluded, the E-I predictor becomes non-significant, and the direction of its
effect remains positive, against expectations. The Core Periphery predictor becomes non-
significant. The interpretation of the results in terms of the hypothesis tests remains the
same: significant predictors of establishment in a state include the existence of an
incentive (but not it’s dollar value), the percentage of distribution, sales and marketing
companies in the state, and the diversity of a state’s organizational population. Support
for hypotheses 1c and 2c and 3c is confirmed. Hypotheses 4c and 6c are not supported.
Hypothesis 5a was not tested because of measurement issues.

Table 10: Effects of incentives, ecological factors and network factors on industrial activity (excluding California and New York),
1998 - 2010
Predictor Variables Model 1 Model 2 Model 3 Model 4
B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. B SE B Sig. Exp(B)
Incentive and Controls
Incentive (dichotomous) 2.15 0.07 0.00 8.57 1.58 0.07 0.00 1.55 0.07 0.00 4.70
Incentive ($ millions) 0.00 0.00 0.00 1.00 0.00 0.00 0.55 1.00 0.00 0.00 0.72 0.00 0.00 0.63 1.00
GDP 0.83 0.10 0.00 2.29 0.70 0.09 0.00 2.01 0.54 0.07 0.00 0.51 0.07 0.00 1.66
Population -0.02 0.00 0.00 0.98 -0.01 0.00 0.00 0.99 -0.01 0.00 0.00 -0.01 0.00 0.00 0.99
Ecological Predictors
Dominant Population 0.12 0.08 0.13 0.11 0.08 0.17 1.11
Diversity 3.78 0.28 0.00 3.25 0.33 0.00 25.75
Network Predictors
E-I Index 0.22 0.08 0.01 1.25
Core-Periphery -0.24 0.07 0.00 0.79
Log Likelihood -2004.56 -1948.27 -1845.77 -1832.22
Pseudo R2 0.09 0.11 0.16 0.16
Neg. Bin. Param. 0.53 0.41 0.25 0.23
Model LL -2004.56 -1948.27 -1845.77 -1832.22
Prev Model LL -2192.76 -2004.56 -1948.27 -1845.77
DifLL 188.20 56.29 102.50 13.56
DF (# new pred) 3 2 2 3
>ChiSq yes yes yes yes
UC LL -2192.76
UC Neg. Bin. Param. 1.04
Model 1: Baseline model with controls and incentives
Model 2: Baseline model with controls, incentives and dichotomous incentives indicator
Model 3: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors
Model 4: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors and network factors
1
166

Table 11: Effects of incentives, ecological factors, and network factors on employment (excluding California and New York),
1998 - 2010
Predictor Variables Model 1 Model 2 Model 3 Model 4
B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B)
Incentive and Controls
Incentive (dichotomous) 6.85 0.05 0.00 944.02 6.67 0.06 0.00 785.97 6.67 0.06 0.00 792.04
Incentive ($ millions) 0.01 0.00 0.00 1.01 0.00 0.00 0.03 1.00 0.00 0.00 0.24 1.00 0.00 0.00 0.25 1.00
GDP 6.28 0.65 0.00 531.80 0.23 0.07 0.00 1.26 0.14 0.07 0.04 1.15 0.16 0.07 0.02 1.17
Population 0.10 0.03 0.00 1.10 0.01 0.00 0.00 1.01 0.01 0.00 0.00 1.01 0.01 0.00 0.00 1.01
Ecological Predictors
Dominant Population 0.31 0.06 0.00 1.36 0.27 0.06 0.00 1.32
Diversity 0.88 0.23 0.00 2.42 1.51 0.28 0.00 4.53
Network Predictors
E-I Index 0.07 0.07 0.29 1.08
Core-Periphery 0.82 5.76 0.89 2.26
Log Likelihood -6986.97 -5096.81 -5065.72 -5050.96
Pseudo R2 0.31 0.50 0.50 0.50
Neg. Bin. Param. 16.97 0.26 0.24 0.23
Model LL -6986.97 -5096.81 -5065.72 -5050.96
Prev Model LL -10123.10 -6986.97 -5096.81 -5065.72
DifLL 3136.13 1890.16 31.08 14.77
Δ DF 3 2 2 3.00
>ChiSq yes yes yes yes
UCLL -10123.10
UC Neg. Bin. Param. 3043.23
Model 1: Baseline model with controls and incentives
Model 2: Baseline model with controls, incentives and dichotomous incentives indicator
Model 3: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors
Model 4: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors and network factors
1
167

Table 12: Effects of incentives, ecological factors and network factors on establishments (excluding California and New York),
1998 – 2010
Predictor Variables Model 1 Model 2 Model 3 Model 4
B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B) B SE B Sig. Exp(B)
Incentive and Controls
Incentive (dichotomous) 4.44 0.05 0.00 85.01 4.27 0.05 0.00 71.70 4.27 0.05 0.00 71.71
Incentive ($ millions) 0.01 0.00 0.00 1.01 0.00 0.00 0.00 1.00 0.00 0.00 0.06 1.00 0.00 0.00 0.07 1.00
GDP 2.67 0.32 0.00 14.40 0.31 0.06 0.00 1.36 0.23 0.06 0.00 1.26 0.23 0.06 0.00 1.26
Population 0.04 0.01 0.00 1.04 0.00 0.00 0.05 1.01 0.01 0.00 0.01 1.01 0.01 0.00 0.00 1.01
Ecological Predictors
Dominant Population 0.27 0.05 0.00 1.31 0.24 0.05 0.00 1.27
Diversity 0.86 0.19 0.00 2.36 1.24 0.23 0.00 3.45
Network Predictors
E-I Index 0.08 0.06 0.19 1.08
Core-Periphery 3.20 4.87 0.51 24.49
Log Likelihood -4964.14 -3471.72 -3436.41 -3424.53
Pseudo R2 0.29 0.51 0.51 0.51
Neg. Bin. Param. 6.93 0.19 0.17 0.16
Model LL -4964.14 -3471.72 -3436.41 -3424.53
Prev Model LL -7020.36 -4964.14 -3471.72 -3436.41
DifLL 2056.21 1492.42 35.31 11.89
Δ DF 3 2 2 3
>ChiSq yes yes yes yes
UC LL -7020.36
UC Neg. Bin. Param. 228.69
Model 1: Baseline model with controls and incentives
Model 2: Baseline model with controls, incentives and dichotomous incentives indicator
Model 3: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors
Model 4: Baseline with controls, incentives, dichotomous incentives indicator, ecological factors and network factors
1
168
169

CHAPTER 6: DISCUSSION AND CONCLUSION
This study investigated the effects state-level incentive programs have had on
industrial activity, employment and establishment in the film industry, arguing that these
outcomes are linked not only to incentives, but also to ecological and network factors.
Over the past twenty years, the use of these incentives by US states has become
incredibly popular (Story, Fehr & Watkins, 2010; Berr, 2012), but their effects have
remained unclear (Longwell, 2012; Friday, 2010). This lack of clarity contributes to a
larger debate over the use of public funds to support private business concerns
(Economist, 2011).
This debate is especially interesting in the context of project-based industries,
because of their transient and mobile nature. Incentives for the film industry have been
critiqued on the one hand as wasteful and ineffective (Mayer & Goldman, 2010;
Christopherson & Rightor, 2010) and on the other as so effective they pose a risk to
traditional production centers (McDonald, 2011). Studies of incentives conducted by state
legislative agencies as well as by policy and academic sources have produced conflicting
results, in part because they examine individual programs, making it difficult to
generalize the effects of incentives overall (Christopherson & Rightor, 2010). This study
seeks to intervene in this debate by offering a view of incentive programs as they have
developed over time and across the US as a whole.
Incentives
The study began by introducing the expectations of economic development
programs, and examining the challenges which emerge when incentives are used to
promote project-based, rather than traditional and permanent, organizational forms. The
170

film industry was advanced as a particularly apt example of a project-based industry
which has been heavily recruited through incentive programs. Looking first at the effects
of incentives, the study found support for the hypothesis that state incentive programs
targeting the US film industry have positive effects on the amount of industrial activity,
the number of people employed, and the number of business establishments associated
with filmmaking in incentivizing states. Of course, the size of incentives offered varies
dramatically from state to state, with some incentivizing states offering less than
$100,000 in incentives per year, and others offering hundreds of millions of dollars
annually. While states have offered larger and larger incentives to filmmakers, and the
results of the analyses offered here show that the simple presence of an incentive of any
size is a significant predictor of activity, employment and establishments, the presence of
collinearity between the continuous and binary incentives predictors means the additional
effect of the price-tag attached to a program is not clear. While the results appear to
suggest that non-financial incentives, and positive externalities not captured by the
amount of legislative spend may play a significant role in outcomes, additional evaluation
of the relationship between the incentives variables is required before any substantive
claims can be made. Future work in this area should focus on this question. In principle,
the amount of money offered in incentives should be associated with the size of gains in
the dependent variables, and while this research cannot definitively claim otherwise, the
question remains an important empirical one.
The role of institutional legitimacy provides a plausible rationale for why the
dollar amount of incentives might indeed have a negligible additional effect on the
dependent variables. The ecological literature emphasizes the importance of financial and
economic environmental munificence, but also acknowledges the crucial value of
171

institutional resources like legitimacy (Mayer & Rowan, 1977; Galaskiewicz, Bielfeld &
Dowell, 2006; Pontikes & Barnett, 2010). Indeed, resource munificence and legitimacy
are closely intertwined, such that increases in a population or community’s resources or
its size are frequently attended by increases in their legitimacy, which in turn promote
further growth. The introduction of incentives by a state through the legislative process
clearly introduces financial resources into the community by making public funds
available for projects and businesses to do their work in that state. Tucker, Meinhardt and
Singh (1990) showed that a Canadian policy initiative focused on youth promoted the
founding of voluntary organizations in that state. Baum and Oliver’s (1992) study of
Toronto’s day care population found that an increase the social services budget in that
city stemmed the failure of day-care centers.
The effects of legitimacy may also cross state borders. Wade, Swaminathan and
Saxon (1998) found that state-level prohibitions on the sale of alcohol had effects on
organizational founding both within the states where the regulations were imposed, as
well as in neighboring states. In their study, the authors found that non-uniformity in
regulation between states produces interesting opportunities and dynamics. For example,
states with laws prohibiting the sale of alcohol saw resources flow to neighboring non-
prohibiting states as individuals crossed state borders to acquire the product.
In addition to the resources policy and legislation may introduce to a population
or community, these processes also generate legitimacy. Institutional theorists like
Edelman (1990) describe the normative and socially constructive effects of legislation on
a community, finding, for example in the context of the Civil Rights legislation in the
1960s that the adoption of formal legislation and policy “encouraged employers to adopt
172

formal grievance procedures for employees, even though the existing laws did not
mandate the adoption of such practices” (in Wade, Swaminathan & Saxon, p. 906).
In the case of film incentive programs, it is plausible that the introduction of a
specific state policy instrument to incentivize filmmaking activity, employment and
establishment of businesses exerts a legitimating effect beyond that which is captured by
the size of the state’s financial investment. Not only does the enactment of an incentive
through legislation generate financial resources for those who would participate in the
film industry, institutional legitimacy also provides normative support to the community
overall. In addition to attracting out-of-state production, institutional resources also
encourage local individuals and organizations to engage in film sector activity.
The introduction of an incentive program in a state may generate important
institutional and legitimacy resources within the state, which contribute to local activity,
employment and establishments. As well, legislative decisions by individual states may
also generate legitimacy beyond the state. Research on the diffusion of policy instruments
has shown that while policies and laws are sometimes adopted through a rational
decision-making process, or in response to a specific and deliberately tacked social issue,
“adoption can occur based on how well the policy confirms to social and cultural norms,
which can increasingly favor the policy as more states adopt the policy and it gains
legitimacy” (Jensen, 2003). Studies of lottery, civil rights, welfare and education policies
by US states all show that the enactment of successful policies by US states tends to
follow a classic diffusion of innovations pattern; as more states adopt a particular
legislative policy, that policy becomes more legitimate, and more likely to be introduced
and adopted by other states (Gray, 1973; Jensen 2003; Walker, 1969). Incentive
173

legislation therefore may have legitimating effects which extend beyond the boundary of
an individual state.
Welch and Thompson (1980) examined the extent to which federal legislation
played a role in this process. They argue that state policy based on similar federal
legislation should diffuse more quickly among states than legislation which does not have
a counterpart at the federal level. Welch and Thompson find that the federal government
is usually not the source of innovation, but rather “almost all major [federal incentives]
seek to implement in all states what some states have already achieved … more usually
after the policy has been tried in one or more states and found to be successful, the idea
becomes established in federal legislation” (p. 717-18). In turn, federal legislation
provides legitimacy for states without a local policy to introduce one. Further
investigation of the effects of interstate legitimation on the pattern of state adoption of
film industry incentives illustrated on page 108 is a profitable direction for future
research.
The legitimating function of incentives may play a role in the tenuous finding of
this study that the size of an incentive targeted at the film industry is less significant than
the fact that a state offered an incentive of any size. If confirmed, this would suggest that
simply outspending one another is not the best strategy for states seeking to maximize the
effects of their incentives. Recent analyses by Christopherson and Rightor (2009) and
Mayer and Goldman (2010), as well as popular criticism by the Economist (2011) lament
the arms race in spending by states which seek to outcompete other states in attracting
film industry activity. The results presented here suggest that de-escalation may be
appropriate: outspending does not generate the most significant increases in the outcomes
174

of interest. Further investigation of the relationship between the baseline effect of having
an incentive and the additional effect of the size of the incentive are a fertile avenue for
empirical research and theoretical development.
Ecological Factors
This study sought to understand how incentives operate on project-based
industries by incorporating insights from ecological theory. The concept of a project
ecology (Grabher, 2002) was oriented toward two important ecological constructs:
dominance and diversity. Ecological theory suggests that some organizational
populations are more powerful and dominant than others because of their control over, or
their position in the flow of, resources in the community (Hawley, 1986; Aldrich & Ruef,
2006). In the film industry, those organizations focused on the high-value parts of the
production chain (distribution, marketing and sales) are dominant. As Johns (2010)
describes, these kinds of companies are “higher-up” in the film industry project hierarchy
(p. 1068). For states seeking to develop a local and sustainable film industry, the presence
of dominant firms is essential. The results of this study show that having more of these
dominant companies contributes to significant gains in employment and increases in film
business establishments. The effects on industrial activity are non-significant, which
suggests that the presence of a dominant population is not essential for the production of
films, which is highly mobile, but is important for local industry development. States
hoping to promote themselves as centers of production for project-based industries, rather
than as what Lukinbeal (2004) calls merely “occasional sites” should benefit from
focusing their recruitment strategies on attracting these dominant populations especially.
175

With the exception of studies by Audia and colleagues (Audia & Kurkoski, 2011;
Audia, Freeman & Reynolds, 2006; Freeman & Audia, 2006), recent studies of
dominance in organizational ecology are relatively rare. It is hoped that this study and its
findings about the importance of dominant populations in cluster development help to
encourage further work in this area. In their work on the specialized instruments industry,
Audia, Freeman and Reynolds (2006) examine dominance from a community
perspective, and hypothesize that industries may be dominant in various places. They
argue that the dominance of one industry in a location may dampen the possibilities for
growth by other industries, in part because of the high degree of legitimacy the dominant
industry enjoys. The authors find that if the dominant population in a location is related to
a focal industry, foundings in the focal industry increase, whereas the dominance of an
unrelated industry causes foundings to decline. This suggests an interesting line of future
research on these film industry data, which could explore the effect of the dominance of
the film industry in California on the growth of other sectors in that state.
As well, the use of incentives by states to promote film industry activity,
employment and establishments may be constrained by the degree to which an
incentivizing state hosts an unrelated dominant industry. For example, we might ask
whether Michigan’s historical association with the automotive industry, or Georgia’s
textile and carpet manufacturing legacy, makes any difference in terms of the success of
these states in attracting and maintaining a healthy film production sector. Another
related consideration is the role of non-film industry dominant companies in states which
allow film production companies to trade their tax credits and rebates. Examining the
inter-population dynamics within these incentive exchanges provides an interesting
176

domain for developing and testing theory about dominance and resource exchange in
organizational community ecologies.
In addition to the presence of the industry’s dominant population, this study also
finds that the diversity of a state’s organizational community makes a significant
difference in terms of its film industry activity, employment and establishments. A film
production hoping to take advantage of incentives and cost savings related to the hiring of
local talent and engaging the services of local companies is best able to do so in a state
with a larger number of film production organizational populations. During a meeting of
producers based in Los Angeles which I attended in 2011, a recurrent theme was the
available “crew depth” in different states. One producer remarked that many locations
which “look great on paper might be, unless there’s another show in town” (Ease
Entertainment Services Summer Production Incentives Symposium, 2011). Without
diversity and depth, there are not enough labor, service and facility resources available
for multiple project teams to operate at the same time without bringing in outside crew or
sending work back to California or New York. Part of the advantage ascribed to keeping
their projects located in California, particularly is the fact that the depth and diversity of
its organizational populations can fulfill most production requirements, even when many
projects are underway simultaneously.
The effect of diversity on the dependent variables in this study was large and
significant. This suggests that, rather than outspend one another by giving away money in
the form of tax credits and rebates, states might find better results by tailoring their
incentives to promote organizational diversity. The post-hoc analysis of the data which
excluded California and New York showed even larger effects of diversity than when
177

those states are included, making an even stronger case for other states aspiring to
promote local production communities to redouble their efforts to expand the diversity of
their organizational community. Indeed, some states are already acknowledging this, and
have begun to explicitly include video game production, commercial production and
visual effects production in their incentive programs (see, for example, the most recent
legislation by New Mexico, Colorado and Georgia, which offer incentives to producers of
digital and interactive media, as well as the summary of digital and interactive media
incentives programs by Beck, Reed & Riden, 2011).
The current study, by focusing exclusively on film production does not capture
the organizational diversity that these related industries might contribute to the
production ecology. It is possible that if television, commercial or video game production
were included in the analysis, a more diverse set of organizational populations might be
represented in the data, and the effect size of the diversity variable would not be as
strong. Another qualification is raised by the coding of organizations into nine distinct
populations in this study. While the coding process which generated these populations
derived from a close understanding of the film industry, and reflects a categorical system
which resonates with practitioners, a better validated coding scheme would provide
additional support for the claims advanced here. Such a coding system would also permit
greater confidence in comparing results across future studies examining these related
sectors. An industrially-specific coding scheme is particularly important if television,
commercial, videogame and other related industries are also to be included.
An expansion of the data collection to include television production projects and
their company networks, especially, is likely to overcome some of the challenges which
178

emerged in testing the network hypotheses advanced in this study. The incentives
described here provide benefits to organizations and individuals who also work in
episodic television, commercial, video game and other multimedia entertainment content.
While some incentives explicitly target these kinds of productions, other benefits emerge
because the same organizations may provide facilities, services or equipment to them. By
restricting the analysis presented here to feature films, and excluding production for these
other media, the size, composition and diversity of the organizational networks which
were constructed for analysis may underestimate the actual size of these production
communities.
Episodic television is perhaps the closest relative of film production in terms of
the similarity in the services, facilities and equipment they require (Curtin & Shattuc,
2009), and anecdotal evidence suggests that the episodic, series nature of television
production provides important stability for nascent production communities. While a film
shoot may unfold over a number of days or weeks, a successful episodic television
program may remain in a location for months and even years. Including episodic
television in these analyses might provide additional information about the effects of
incentives on activity, employment and establishment. For example, in 2011 the feature
film The Hunger Games spent 84-days shooting in the state of North Carolina (Vlessing,
2003), compared with six seasons of filming in the state for television series Dawson’s
Creek (1998-2003) and One Tree Hill (2003-2012). The North Carolina Film Office
credits each episode of filming for One Tree Hill with generating $1 million dollars in
economic impact and employing a stable crew of more than 150 local production workers
(North Carolina Film Office, 2011). While these numbers have not been independently
verified, they nevertheless suggest that including television production in future work on
179

the effects of incentives would likely capture greater stability of effects than analysis of
feature films alone.
Network Factors
Results for the tests of the network hypotheses in this study were disappointing.
The E-I index of balance among external and internal ties was not a significant predictor
of activity, employment or establishments in the film industry. One possibility is that
balance of local and extra-local ties among organizations is simply unrelated to
outcomes. However, the empirical findings of Bathelt, Malmberg and Maskell (2004) and
Bathelt (2005) suggest otherwise. California and New York have very high rates of
filming, employment and establishments, but differ on their E-I score, with the majority
of California’s ties occurring within the state, while the majority of New York’s ties cross
state boundaries. Rather than conclude from these divergent results that balance is not an
important factor for successful industrial development, it is possible that there is more
than one optimal strategy for balancing organizational relationships, and that these might
depend on factors related to the individual features of the state. A detailed examination of
the qualities of states with high and low E-I measures might provide an indication of
whether and under what conditions more internal or external ties might be of value in
generating activity, employment and establishments.
Another possibility is that the E-I measure may require refinement in its
application to an interorganizational context. The E-I index was originally designed to
capture the balance of external and internal friendship ties in intra-organizational work
groups. While the principles of embeddedness and frequency of potential contact which
guide the application of the measure to an interorganizational context are sound, factors
180

like the scale of a national network of interacting organizations across thousands of miles
may make it necessary to incorporate a parameter which would account for physical
distance between partners, for example. As well, the E-I analyses presented here
aggregated company-company ties to the state level, which sacrifices the granularity and
detail of the company-company links, and may result in a rather more blunt instrument
than is ideal.
The smallness of many of the individual state networks, which is a consequence
of restricting the data to feature films, made the small world measure impossible to
calculate in all but a handful of cases. A search of the Internet Movie Database using the
same criteria used to collect the data of feature films in this study, but applied to
television series, returns 3,820 television series, produced and released in the United
States, in the English language, for which shooting location information is available.
These series account for 9,469 individual episodes. The inclusion of these productions
and the companies which come together to work on them would likely augment the data
set, and the resulting company-company networks, sufficiently to allow analysis of small-
world signatures for many states.
In addition, the hypotheses advanced here suggested that increasing small-
worldness of a state would result in increased activity, employment and establishments
within the state. Within the ecological literature, work by Audia and colleagues(Audia &
Kurkoski, 2011; Audia, Freeman & Reynolds, 2006), as well as the work in interstate
effects of policy on organizational ecology by Wade, Swaminathan and Saxon (1998)
suggest that limiting the analysis to the boundary of the state may be too restrictive.
Incentives may well generate effects beyond the border of the state which offers it. For
181

example, companies located in New Jersey may see a benefit when New York offers an
incentive, because interorganizational activity within the tri-state area is not uncommon
or difficult. Although the analyses conducted here do not offer insights as to whether
incentives might promote activity, employment or establishments in neighboring states,
under what conditions, and at what distances, that is a valuable line of future research in
this area.
A final set of hypotheses tested in this dissertation concerns the power of
incentives to promote changes in the national structure of filmmaking networks. The
aggregation of ties among companies in different states was hypothesized to promote
movement by some states into semi-peripheral positions, and disrupt the persistent core-
periphery pattern of the film production network. The analyses presented here show that
even after twenty years of aggressive incentives to promote alternative locations, the
primacy of California and New York have not been undermined. From an organizational
network perspective, the expression “runaway” production does not seem to accurately
capture the changes happening in the film industry. Nor do the results presented here
suggest that the use of incentives to promote filming elsewhere has had a deleterious
effect on the industry as it continues to operate in California and New York.
California’s response to the introduction of incentives by other states has been to
enter into incentives competition, by introducing $100 million in incentives annually
between 2009 and the present. However, the results of this study suggest that this may
not be the best strategy for shoring up California’s production ecology. California’s share
of dominant companies, it’s incredibly high number of ties among businesses within the
state, and its persistent diversity suggest that capitalizing on these qualities in a targeted
182

way might be a better way forward than distributing resources through a piecemeal
lottery system.
This dissertation represents a first step towards understanding how incentive
programs targeting project-based industries operate, and under what conditions they are
successful in promoting activity, employment and increases in the number of business
establishments. The strong findings for the importance of dominant populations and
community diversity suggest that examining these processes as a complex ecological
system is productive. While the analysis of network measures were not successful in this
case, it is anticipated that a data collection process which includes more of the industry’s
participants, and reflects a more realistic view of the larger visual entertainment
community would generate better results. An advantage of an ecological approach to
organizational communities is the many possibilities it affords for drilling down into
smaller, local ecologies to better understand the networks in regions, states and cities.
The ecological approach also encourages zooming out to examine how these processes
unfold in the context of a global entertainment industry. The efforts put forth in this
research represent an important first step.

183

REFERENCES
Adner, R. & Levinthal, D.E. (2002). The emergence of emerging technologies. Preprint
manuscript accessed online June 6 2012 at
www.users.unimi.it/exaptation2009/docs/Adner/%20&@20Levinthal.pdf
Aksoy, A. & Robbins, K. (2992). Hollywood for the 21
st
century: Global competition for
critical mass in image markets. Cambridge Journal of Economics, 16, 1-22.
Aldrich, H. E. (1999). Organizations evolving. Thousand Oaks, CA: Sage.
Aldrich, H. E., & Ruef, M. (2006). Organizations evolving. London: Sage.
Alvarez, J. L. & Svejenova, S. (2002). Symbiotic Careers in Movie Making: Pedro and
Agustín Almodóvar. In Maury Peiperl, Michael Arthur, and N. Anand (Ed.).
Career Creativity: Explorations in the remaking of Work (pp. 183-208). Oxford,
United Kingdom: Oxford University Press.
Amaral, L. A. N., Scala, A., Bartelemy, M., and Stanley, H.E. (2000). “Classes of Small-
World Networks.” Proceedings of the National Academy of Science 111:49–52.
Anderson & Tushman 2001
Ansari, A., Essegaier, S. & Kohli, R. (2000). Internet Recommendation Systems. Journal
of Marketing Research, 37(3), pp, 363-375.
Argote, L. (2011). Organizational learning research: Past, present and future.
Management Learning , 42(4), 439-446.
Arrow, K.J. (1962). The economic implications of learning by doing. Review of
Economic Studies, 29(3), 155-173.
Astley, W. G. (1985). The two ecologies: Population and community perspectives on
organizational evolution. Administrative Science Quarterly , 30 (2), 224-241.
Astley, W. G., & Fombrun, C. J. (1983). Collective strategy: Social ecology of
184

organizational environments. Academy of Management Review , 8 (4), 576-587.
Audia, P. G., Freeman, J. H., & Reynolds, P. D. (2006). Organizational foundings in
community context: Instruments manufacturers and their interrelationship with
other organizations. Administrative Science Quarterly , 51, 381-419.
Badaracco, J. (1991). The knowledge link: How firms compete through strategic
alliances. Cambridge: Harvard University Press.
Baker, C. E. (2007). Media concentration and democracy. Cambridge: Cambridge
University Press.
Baker, W. E., & Faulkner, R. R. (1991). Role as resource in the Hollywood film industry.
American Journal of Sociology , 97 (2), 279-309.
Baldwin, R E., Forslid, R (2000). The core-periphery model and endogenous growth:
stabilizing and destabilizing integration. Economica, 67: 307-324.
Balio, T. (1996). Adjusting to the new global economy: Hollywood in the 1990s. In A.
Moran (Ed.) Film Policy: Intenrational, National and Regional Perspectives.
London: Routledge, pp. 23-38.
Bamford, C.E, Dean, T.J. & McDougal, P.P. (1999). An examination of the impact of
Journal of Business Venturing, 15, 253-277.
Barabasi, A.L. & Albert, R. (1999). Emergence of scaling in random networks. Science,
286 (5439), 509-512.
Barabasi, A.L., Jeong, H., Neda, Z, Ravasz, E. Schubert, A. & Viseck, T. (2002).
Evolution of the social networks of scientific collaborations. Physica A:
Statistical Mechanics and its Applications, 311, 3-4, 590-614.
Bardhan, I., Whittaker, J., & Mithas, S. (2006). Informatio n technology, production
185

process outsourcing and manufacturing plant performance. Journal of
Management Information, 23(2), 13-40.
Barnett G. A. & Park, H. W. (2005). The structure of international internet hyperlinks and
bilateral bandwidth. Annales des Telecommunications, 60(9-10), pp. 1115-1132.
Barnett, G. (2001). A Longitudinal analysis of the international telecommunications
network: 1978-1996. American Behavioral Scientist, 44, pp. 1638-1655.
Bartik, T. J. 2011. "State Economic Development Policies: What Works?" Presented at
the 19th Annual State Fiscal Policy Conference, Center on Budget and Policy
Priorities, Washington, D.C., November 30, 2011.Accessed April 4, 2013 from
http://research.upjohn.org/presentations/27
Bartik, T.J., Becker, C., Lake, S. & Bush, J. (1987). Saturn and state economic
development. " Forum for Applied Research and Public Policy 2(1)(Spring): 29-
40. Accessed April 4, 2013 from http://research.upjohn.org/jrnlarticles/147
Bathelt, H. (2005). Cluster relations in the media industry: Exploring the 'distanced
neighbor' paradox in Leipzig. Regional Studies, 39(1), 105-127.
Bathelt, H., Malmberg, A. & Maskell, P. (2004). Clusters and knowledge: Local buzz,
global pipelines and the process of knowledge creation. Progress in Human
Geography, 28(1), 31-56.
Baum, J.A.C. (1999). Rise of chain nursing homes in Ontario, 1971-1996. Social Forces,
78(2), 543-583.
Baum, J. A. (1990, August). Inertial and adaptive patterns in organizational change.
Academy of Management Proceedings (Vol. 1990, No. 1, pp. 165-169). Academy
of Management.
Baum, J.A.C. & Amburgey, T.L. (2005). Organizational ecology.In Baum, J. A. C. (Ed.).
186

The Blackwell companion to organizations. Malden, MA: Blackwell
Baum, J.A.C. & Oliver, C. (1992). Institutional embeddedness and the dynamics of
organizational populations. American Sociological Review, 57, 540-559.
Baum, J.A.C. & Oliver, C. (1991). Institutional linkages and organizational mortality.
Administrative Science Quarterly, 36, 187-218.
Baum, J.A.C & Oliver, C. (1996). Toward an institutional ecology of organizational
founding. Academy of Management Journal, 39, 1378-1427.
Baum, J.A.C., Shipilov, A.V.& Rowley, T.J.(2003). Where do small worlds come from?
Industrial and Corporate Change, 12, 697-725.
Baumann, S. (2001). Intellectualization and art world development: Film in the United
States. American Sociological Review, 66,404–26.
Bechky, B.A. (2006). Gaffers, gofers and grips: Role-based coordination in temporary
organizations. Organization Science, 17(1), 3-21.
Bielby, W. T., & Bielby, D. D. (1999). Organizational mediation of project-based labor
markets: Talent agencies and the careers of screenwriters. American Sociological
Review , 64 (1), 64-85.
Blakely, E.J. & Bradshaw, T.K. (2002). Planning local development. Thousand Oaks,
CA: Sage.
Blakely, E.J. & Leigh, N.G. (2010). Planning local development: Theory and Practice.
Thousand Oaks, CA: Sage.
Bloomberg (2012, November). Incentives watch: Monetizing state tax credits? There's an
exchange for that. Retrieved April 7, 2013 from http://www.bna.com/incentives-
watch-monetizing-b17179870903/
Bockelman, K. (1999). Administration of state economic development policy. In Gargan,
187

J. (Ed.) Handbook of State Government Administration. New York: Marcel
Dekker Inc.
Boekema, F. (2000). Knowledge, Innovation, and Economic Growth: The Theory and
Practice of Learning Regions. Mawah, NJ: Edward Elgar Publishing.
Bomey, N. (2011). Rick Snyder proposes big cuts to business taxes, elimination of credits
for film, high-tech industries. Ann Arbor.com Accessed April 4, 2013 from
http://www.annarbor.com/business-review/rick-snyders-budget-proposal-
includes-massive-cuts-to-business-taxes-credits-for-film-high-tech-batt/
Bordwell, D., Staiger, J. & Thompson, K. (1985). The classical Hollywood cinema: Film
style and mode of production to 1960. New York: Columbia University Press.
Borgatti, S.P. & Everett, M.G. (1999). Models of core/periphery structures. Social
Networks 21 , 375-395.
Bourdieu, P. (1986). The forms of capital. In J. Richardson (Ed.) Handbook of Theory
and Research for the Sociology of Education (pp. 241–258). New York:
Greenwood.
Bracey, L. (n.d.). Rich part, bright future. Business in Focus. Accessed April 5, 2013
from http://www.businessinfocusmagazine.com/2013/03/canadian-film-
production/
Bruggeman, J., Grunow, D., Leenders,M.A., Vermeulen, I., & Kuilman, J.G. (2012)
Market positioning: the shifting effects of niche overlap. Industrial and Corporate
Change, 21, 1451-1477.
Bryant, J. A., & Monge, P. R. (2008). The evolution of the children's television
community, 1953-2003. International Journal of Communication , 2, 160-192.
Bryant, P. (n.d.). Everything you ever wanted to know about booking a video crew.
188

Accessed April 5, 2013 from http://www.gototeam.com/video-producer-guide/
Burt, R. S. (1992). Structural holes: The social structure of competition. Cambridge, MA:
Harvard University Press.
Burt, R.S. ( 2004). Structural holes and good ideas. American Journal of Sociology, 110
(2), 349-399.
Butts, C. T. (2012). sna: Tools for Social Network Analysis (Version 2.2) [R package].
Irvine, CA. Retrieved from CRAN.R-project.org/package=sna
Butts, C. T., Handcock, M., & Hunter, D. R. (2012). network: Classes for Relational Data
(Version 1.7-1) [R package]. Irvine, CA. Retrieved from CRAN.R-
project.org/package=network
California State Assembly (2011) Assembly Bill 2026 Bill Analysis. Accessed April 5,
2013 from http://www.leginfo.ca.gov/pub/11-12/bill/asm/ab_2001-
2050/ab_2026_cfa_20120815_175735_asm_floor.html
Cameron, G. (1996). Innovatoin and economic growth. Center for Economic
Performance Discussion Paper. Accessed April 5, 2013 from
http://eprints.lse.ac.uk/20685/1/Innovation_and_Economic_Growth.pdf
Carroll, G. R. (1985). Concentration and specialization: Dynamics of niche width in
populations of organizations. American Journal of Sociology , 90 (6), 1262-1283.
Carroll, G.R. & Hannan, M.T. (2000). The Demography of Corporations and Industries.
Princeton: Princeton University Press.
Carroll, G. R., & Hannan, M. T. (1989). Density Dependence in the Evolution of
Populations of Newspaper Organizations. American Sociological Review , 54 (4),
524-541.
189

Carroll, G.R. & Swaminathan, A. (1992). The organizational ecology of strategic groups
in the American brewing industry from 1975 to 1990. Industrial and Corporate
Change, 1(1), 65-97.
Carter, B. (2010). Stephen J. Cannell, prolific TV writer dies at 69. New York Times.
Accessed April 5 2013 from
http://www.nytimes.com/2010/10/02/arts/television/02cannell.html
Castells, M. (2010). The rise of the network society (2nd ed.). Malden, MA: Wiley-
Blackwell.
Castrogiovanni, G.J. (1991). Environmental munificence: A theoretical assessment.
Academy of Management Review, 16(3), 542-565.
Castrogiovanni, G.J. (1996). Pre-startup planning and the survival of new small
businesses: Theoretical linkages. Journal of Management, 22(6), 801-822.
Cassia, L. & Minola, T. (2012). Hyper-growth of SMEs: Toward a reconciliation of
entrepreneurial orientation and strategic resources. International Journal of
Entrepreneurial Behaviour & Research, Vol. 18(2), 179 - 197
Cattani, G., & Ferriani, S. (2008). A core/periphery perspective on individual creative
performance: Social networks and cinematic achievements in the Hollywood film
industry. Organization Science , 19 (6), 824-844.
Cattani, G., Ferriani, S., Negro, G., & Perretti, F. (2008). The structure of consensus:
Network ties, legitimation and exit rates of US feature film producer
organizations. Administrative Science Quarterly , 53 (1), 145-182.
Caves, R. E. (2000). Creative industries: Contracts between art and commerce.
Cambridge, MA: Harvard University Press.
190

Center on Budget and Policy Priorities. (2013). State Budget Crises and the Economy.
Retrieved April 4, 2013, from
http://www.cbpp.org/slideshows/?fa=stateFiscalCrisis
Chianese, J., Cordova, M., & Rosenfeld, B. (2010). The essential guide to US and
international production incentives. Los Angeles: Entertainment Partners.
Cheng, M., Dainty, A. & Moore, D. (2007). A multifaceted performance excellence
framework for project-based organizations. Journal of Human Resources
Development and Management, 7(3), 254-275.
Chirot, D. & Hall, T.D. (1982). World system theory. Annual Review of Sociology, 8, 81-
106.
Christopherson, S. (2006). Behind the scenes: How transnational forms are constructing a
new international division of labor in media work. Geoforum, 37, 739-751.
Christopherson, S. (2002). Project work in context: Regulatory change and the new
geography of media. Environment and Planning A, 34, 2003-2015.
Christopherson, S., & Rightor, N. (2010). The Creative Economy as “Big Business”:
Evaluating State Strategies to Lure Filmmakers. Journal of Planning Education
and Research , 29 (3), 1-5.
Christopherson, S., & Storper, M. (1986). The city as studio; the world as back lot: The
impact of vertical disintegration on the location of the motion picture industry.
Environment and Planning , 4, 305-320.
Christopherson, S., & Storper, M. (1989). The effects of flexible specialization of
industrial politics and the labor market: The motion picture industry. Industrial
and Labor Relations Review , 42 (3), 331-347.
Cieply, M. (2008). States' film production incentives cause jitters. New York Times.
191

Accessed on April 5, 2013 from
http://www.nytimes.com/2008/10/12/us/12incentives.html?pagewanted=all&_r=0
Clarke, S.E. & Gaile, G.L. (1992). The next wave: Postfederal local economic
development strategies. Economic Development Quarterly, 6,2, 198-198.
Coase, R.H. (1937). The nature of the firm. Economica, 4(16), 386-405.
Cohen & Kolstad (2009).
Colorado Incentives Office (2012). “The Ripple Effect”, Colorado Film Incentives.Video
Retrieved from http://www.youtube.com/watch?v=DwtAu1QKNp8&feature
=youtube_gdata_player
Cooke, P. (2002). Knowledge economies: Clusters, learning and cooperative advantage.
London: Routledge.
Crone, M. (2012). New venture internationalisation and the cluster life cycle:
Interdisciplinary insights on Ireland's indigenous software industry. Academy of
International Business (UK & Ireland Chapter) 39th Annual Conference,
University of Liverpool, 29-31 March 2012.
Csardi, G. & Nepusz, T. (2006) The igraph software package for complex
network research, InterJournal, Complex Systems.
Curtin, M., & Shattuc, J. (2009). The American television industry. London: British Film
Institute.
Davis, G.F., Yoo, M., & Baker, W. (2003). The small world of the American corporate
elite, 1982–2001.” Strategic Organization, 3, 301–26.
Delacroix, J., & Carroll, G. R. (1983). Organizational foundings: An ecological study of
the newspaper industries of Argentina and Ireland. Administrative Science
Quarterly , 28, 274-291.
192

Delacroix, J., & Rao, H. (1994). Externalities and ecological theory: Unbundling density
dependence. Evolutionary dynamics of organizations, 255, 268.
DeVany, A. (2004). Hollywood economics: How extreme uncertainty shapes the film
industry.New York: Routledge.
DiMaggio, P. J., & Powell, W. W. (1983). The iron cage revisited: Institutional
isomorphosm and collective rationality in organizational fields. American
Sociological Review , 48, 147-160.
DiFillippi, R.J. & Arthur, M.B. (1998). Paradox in project-based enterprise: The case of
film making. California Management Review, 40(2), 125-139.
Dimmick, J. W. (2003). Media competition and coexistence: The theory of the niche.
Mahwah, NJ: Lawrence Erlbaum Associates.
Dobbin, F. & Dowd, T.J. (1997). How policy shapes competition: Early railroad
foundings in Massachusetts. Administrative Science Quarterly, 42(3), 501-529.
Dobrev, S.D., Kim, T.Y. & Carroll, G.R. (2003). Shifting gears, shifting niches:
Organizational inertia and change in the evolution of the US automobile industry,
1885-1981. Organization Science, 14(3), 264-282.
Dornisch, D. (2002). The evolution of post-socialist projects: Trajectory shift and
transitional capacity in a Polish region. Regional Studies.36(3), 2002.
Economist. (2011, June 9). Film industry subsidies: Unilateral disarmament. Accessed
April 4, 2013 from http://www.economist.com/node/18805941
Edmondson, A. C., & Nembhard, I. M. (2009). Product Development and Learning in
Project Teams: The Challenges Are the Benefits. Journal of Product Innovation
Management, 26(2), 123–138.
Eisinge, P.K. (1993). State venture capitalism, state politics and the world of high risk
193

investment. Economic Development Quarterly, 7(2), 131-139.
Eisinger, P.K. (1988). Rise of the entrepreneurial state. Madison: The University of
Wisconsin Press.
Elbanna, S. & Alhwarai, M. (2012). The influence of environmental uncertainty and
hostility on orgnization performance. UAEU-FBE-Working Paper Series.
Retrieved June 6, 2012 from:
www.fbe.uaeu.ac.ae/research/docs/wp2012_01_updated.pdf
Elmer, G. & Gasher, M. (2005). Catching up to runaway productions, in G. Elmer and M.
Gasher (Eds.), Contracting out Hollywood, Maryland: Rowman & Littlefield.
Engwall, M. (2003). No project is an island: Linking projects to history and context.
Research Policy, 32, 789-808.
Eriksson, R. & Lindgren, U. (2009).Localized mobility clusters: Impacts of labour market
externalities on firm performance. Journal of Economic Geography, 9(1), 33-53.
Faulkner, R. R., & Anderson, A. B. (1987). Short-term projects and emergent careers:
Evidence from Hollywood. American Journal of Sociology , 92 (4), 879-909.
Ferriani, S. Corrado, R. & Borschetti, C. (2005) Organizational learning under
organizational impermanence: Collaborative ties in film project firms. Journal of
Management and Governance, 9(3), 257-285.
Ferriani, S. Cattani, G. & Baden-Fuller, C. (2007). Fitness determinants in creative
industries: A longitudinal study on the Hollywood film-making industry, 1992-
2003. In U. Cantner & F. Malerba (Eds.) Innovation, Industrial Dynamics and
Structural Transformation: Schumpeterian Legacies, pps. 209-237.
Ferriani, S., Cattani, G. and Baden-Fuller, C. (2009). The relational antecedents of
194

project-entrepreneurship: Networks centrality, team composition and project
performance. Research Policy, 38(10), 1545-1558.
Fleming, L., King, C. & Juda, A.I. (2007). Small worlds and regional innovation.
Organization Science, 18(6), 938-954.
Fleming, L. & Marx, M. (2006). Managing creativity in small worlds. California
Management Review, 48(4), 6-27.
Fosler, R.S. (1992). State economic policy: The emerging paradigm. Economic
Development Quarterly, 6(1), 3-13.
Freeman, J., & Audia, P. G. (2011). Community context and founding processes of
banking organizations. Research in the Sociology of Organizations, 33, 253–282.
Freeman, J., & Hannan, M. T. (1983). Niche width and the dynamics of organizational
populations. American Journal of Sociology , 88 (6), 1116-1145.
Friday, S.D. (2011). Behind the scenes of 'Hollywood South': A case study of state
motion picture incentive policy in Shreveport, Louisiana. Unpublished
Manuscript.
Flora, J.L. (1998). Social capital and communities of place. Rural Sociology, 63,4,481-
506.
Fujita, M., Krugman, P. & Venables, A.J. (2001). The spatial economy: Cities, regions
and international trade. Cambridge, MA: MIT Press
Fulton, W. (2010). City and state smokestack chasing blows mostly smoke. Accessed
April 4, 2013 from http://citiwire.net/columns/city-and-state-smokestack-chasing-
blows-mostly-smoke/
Galsakiewicz, J., Bielefeld, W. & Dowell, M. (2006). Networks of organizational growth:
195

A study of community-based nonprofits. Administrative Science Quarterly, 51,
337-380.
Gardner, W. Mulvey, E.P. & Shaw, E.C. (1995). Regression analyses of counts and rates:
Poisson and overdispersed Poisson and negative binomial models. Quantitative
Methods in Psychology, 118(3), 392-404.
Gargan, J.J., (1999). Handbook of State Government Administration. New York: Marcel
Dekker Inc.
Gilley, K.M., Greer, C.R. & Rasheed, A.A. ( 2004). Human resource outsourcing and
organizational performance in manufacturing firms. Journal of Business
Research, 57(3), 232-240.
Giles, J.A. & Williams, C.L. (1999). Export-led growth: A survey of the empirical
literature and some noncasuality results. Econometrics Working Paper. Accessed
on April 5, 2013 from
http://www.web.uvic.ca/econ/research/papers/pdfs/ewp9901.pdf
Girard, M. & Stark, D. (2002). Distributing intelligence and organizing diversity in new
media projects. Environment and Planning A, 34, 1927-1949.
Glaeser, E., Kallal, H.D., Scheinkman, J.A. & Schleifer, A. ( 1992). Growth in cities.
Journal of Political Economy, 100(6), 1126-1152.
Gomery, D. (2008). The Hollywood studio system: A history. London: The British Film
Institute.
Goodman, J. (2012, December 14). Tax breaks for sale: Transferable tax credits
explained . Retrieved April 4, 2013, from The Pew Charitable Trusts: State and
Consumer Initiatives: http://www.pewstates.org/projects/stateline/headlines/tax-
breaks-for-sale-transferable-tax-credits-explained-85899436534
196

Goodman, R.A. & Goodman, L.P. (1976). Some management issues in temporary
systems: A study of professional development and manpower: The theater case.
Administrative Science Quarterly, 21, 494-501.
Gonzalez, C.N., Veloso, F.M. & Krackhardt, D. (2008). Social capital and the creation of
knowledge. Industry Studies Association Working Pa:pers. Retrieved June 6,
2012 from www.papers.ssrn.com/sol3/papers.cfm?abstract_id=1120977
Grabher, G. (2002). The project ecology of advertising: Tasks, talents and teams.
Regional Studies, 36(3), 245-262.
Grabher, G.( 2002). Cool projects, boring institutions: Temporary collaboration in social
context. Regional Studies, 36(3), 205-214.
Granovetter, M. (1985). Economic action and social structure: The problem of
embeddedness. American Journal of Sociology , 91 (3), 481-510.
Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology , 78
(6), 1360-1380.
Greenwood, R. Sudabby, R. & Hinings, C.R. (2002). Theorizing change: The role of
professional associations in the transformation of institutionalized fields.
Academy of Management Journal, 45(1), 58-80.
Gross, J. (1997, March 30). "Average" family struggles on edge of financial collapse.
Retrieved April 5, 2013, from The Los Angeles Times:
http://www.southcoasttoday.com/apps/pbcs.dll/article?AID=/19960330/NEWS/30
3309968&cid=sitesearch&template=printart
Guiliani, E. & Bell, M. (2005). The micro-determinants of meso-level learning and
innovation: Evidence from a Chilean wine cluster. Research Policy, 34, 47-68.
Gulati, R. (1998). Alliances and networks. Strategic Management Journal , 19 (4), 293-
197

317.
Gulati, R. (2007). Managing network resources: Alliances, affiliations and other
relational assets. Oxford: Oxford University Press.
Galaskiewicz, J., Bielefeld, W. & Dowell, M. (2006). Networks and organizational
growth: A study of community based nonprofits. Administrative Science
Quarterly, 51, 337-380.
Guzzo, R.A. & Dickson, M.W. (1996). Teams in organizations: Recent research on
performance and effectiveness. Annual Review of Pschology, 47, 307-338.
Halpert, J. (2011, March 21). Michigan's Next Big Blow: The Film Industry. Retrieved
April 4, 2013, from CNN Money:
http://money.cnn.com/2011/03/21/news/companies/michigan_film_industry.fortu
ne/index.htm
Hannan, M.T. & Carroll, G.R. ( 2003). The organizational niche. Sociological Theory,
21(4), 309-340.
Hannan, M.T. & Carroll, G.R. (1992). Dynamics of Organizational Populations. Oxford:
Oxford University Press.
Hannan, M. T., & Freeman, J. (1989). Organizational ecology. Cambridge, MA: Harvard
University Press.
Hannan, M. T., & Freeman, J. (1987). The ecology of organizational founding: American
labor unions 1836-1985. American Journal of Sociology , 92 (4), 910-943.
Hannan, M. T., & Freeman, J. T. (1984). Structural inertia and organizational chance.
American Sociological Review , 49, 149-164.
Hannan, M. T., & Freeman, J. T. (1977). The population ecology of organizations.
American Journal of Sociology , 85, 929-964.
198

Hannan, M. T., Carroll, G. R., & Polos, L. (2003). The organizational niche. Sociological
Theory, 21 (4), 309-340.
Hanson, G.H. (1996). Agglomeration, dispersion, and the pioneer firm. Journal of Urban
Economics, 39, 225-281.
Hatry, H., Fall, M., Singer, T.O. & Liner, E.B. (Eds). (1990). Monitoring the Outcomes
of Economic DevelopmentPrograms. Washington, DC: The Urban Institute.
Hawley, A. H. (1986). Human ecology: A theoretical essay . Chicago: University of
Chicago Press.
Hayes, A.F. (2005). Statistical Methods for Communication Science. Mahwah, NJ:
Lawrence Erlbaum Associates.
Hembd, J. (2008) Sustainable development: Is it going mainstream? Building
Communities Webinar Education Series Presentation, Northern Center for
Community and Economic Development, University of Wisconsin. Retrieved
from http://www.uwex.edu/ces/cced/communities/documents/Series9.pdf.
Hesmondhalgh, D. (2007). The Cultural Industries. Thousand Oaks CA: Sage.
Holt, J. (2001). In Deregulation We Trust: The Synergy of Politics and Industry in
Reagan-Era Hollywood. Film Quarterly, 55, 2, 22-29.
Hozic, A. A. (2001). Hollyworld: Space, power and fantasy in the American economy.
Ithaca: Cornell University Press.
Hsu, G. (2006). Jacks of all trades and masters of none: Audience's reactions to spanning
genres in feature film production. Administrative Science Quarterly , 51, 420-250.
Huang, M., Huang, Y., Margolin, D., Ognyanova, K., Shen, C., & Contractor, N. (2010).
199

The Effects of Diversity and Repeat Collaboration on Performance In Distributed
Nanoscientist Teams. Academy of Management (AOM) Annual Conference,
Montreal, QC, Canada
Hui, K. L. & Png, I.P.L.(2002). One the supply of creative labor: Evidencce from the
movies. American Economic Review, 92(2), pp. 217-220.
Hugos, M. (2011). Essentials of Supply Chain Management. Hoboken, NJ: Wiley .
Investing in Film Using Irs Section 181 for Federal Tax Credits Vs. New Markets Real
Estate Credits. (n.d.). ArticlesBase.com. Retrieved April 7, 2013, from
http://www.articlesbase.com/business-articles/investing-in-film-using-irs-section-
181-for-federal-tax-credits-vs-new-markets-real-estate-credits-746745.html
Iravani, S.M.R., Kolfal, B. & Van Oyen, M.P. (2007). Call Center Labor Cross-Training:
It's a Small World After All. Management Science, 53, 1102--1112.
Jackson, M. O., Rodriguez-Barraquer, T., & Tan, X. (2012). Social capital and social
quilts: Network patterns of favor exchange. The American Economic Review,
102(5), 1857–1897.
Jensen, N., Neville, J. & Hay, M. (2003). Avoiding bias when aggregating relational data
with degree disparity. Proceedings of the 20
th
Annual International Conference on
Machine Learning, Washington DC. Retrieved June 6, 2012 from:
www.cs.purdue.edu/homes/neville/papers/jenses-et-al-icml2003.pdf
Jones, C. (2011). Coevolution of entrepreneurial careers, institutional rules and
competitive dynamics in American film, 1895-1920. Organization Studies, 22(6),
911-944.
Jones, C. (1996). Careers in project networks: The case of the film indsutry. In M.B.
200

Arthur and D.M. Rousseau (Eds.) The boundaryless career: A new
employmentprinciple for a new organizational era. Oxford: Oxford University
Press, pps 58-75.
Jones, C. & Walsh, K. (1997). Boundaryless Careers in the US Film Industry. Industrielle
Beziehungen, 4(1), 58-73.
Kansas Department of Revenue (n.d.). Film Production Credit. Retrieved Aptil 7 2013
from http://www.ksrevenue.org/taxcredits-film.html
Katz, R. (1982). The effects of group longevity on project communication and
performance. Administrative Science Quarterly, 27(1), 81-104
Keegan, A. & Turner, J.R. (2002).The management of innovation in project-based firms.
Long Range Planning, 35(4), 367-388.
Kelly, D. and Amburgey, T. (1991). Organizational inertia and momentum: a dynamic
model of strategic change. Academy of Management Journal, 34(3): 591-612.
Kenis, P., Janowicz-Panjaitan, M & Cambre, B. (2009). Temporary Organizations:
Prevalence, Logic and Effectiveness. Mawah, NJ: Edward Elgar.
Kirca, A.H., Hult, T.M., Deligonul, S., Perryy, M.Z. & Cavusgil, S.T. (2012). A
multilevel examination of the drivers of firm multinationality: A meta-analysis.
Journal of Management, 38(2), 502-530.
Klowden, K., Chatterjee, A., & Hynek, C. F. (2010). Film flight: Lost production and its
economic impact on California. Santa Monica: The Miliken Institute.
Klowden, K., Shen, I., & Ho, K. W. (2012). Fighting production flight: Improving
California's filmed entertainment tax credit program. The Miliken Institute.
Kogut, B. & Walker, G. (2001). The small world of Germany and the durability of
national networks. American Sociological Review, 66(3), 317-335.
201

Kraatke, S. (2002). Network analysis of production clusters: The Potsdam / Babelsberg
film industry as an example. European Planning Studies, 10(1), 27-54.
Koven, S.G. & Lyons, T.S. (2010). Economic development: Strategies for state and local
practice. Washington D.C.: ICMA Press.
Kozolowski, S.W.J. & Bell, B.S.(2001). Work groups and teams in organizations.
Articles and Chapters. Paper 389. Retrieved June 6, 2012 from:
www.digitalcommons.ilr.cornell.edu/articles/389.
Krackardt, D. & Stern, R.N. (1988). Informal networks and organizational crises: An
experimental simulation. Social Psychology Quarterly, 51(2), 123-140.
Krugman, P. (1991). Increasing returns and economic geography. The Journal of
Political Economy, 99(3), 483-499.
Larance, L. Y. (2001). Fostering social capital through NGO design: Grameen Bank
membership in Bangladesh. International Social Work, 44(1), 7–18.
Lawyers for the Creative Arts (2006). Legas issues in film production: Business and
investment structure. Retrieved April 24, 2013 from http://www.law-
arts.org/pdf/Legal_Issues_in_Film_Production.pdf
Leamer, E. E., & Storper, M. (2001). The economic geography of the internet age.
Journal of International Business Studies , 32 (4), 641-665.
Lee, S., Monge, P., Bar, F., & Matei, S. A. (2007). The emergence of clusters in the
global telecommunications network. Journal of Communication , 57 (3), 415-434.
Lee, C., Lee, K. & Pennings (2001). Internal capacilities and external networks and
performance: A study on technology-based ventures. Strategic Management
Journa l, 22, 615-640.
Litman, B. R. (1998). The motion picture mega-industry. Needham Heights, MA: Allyn
202

& Bacon.
Lin, N. (1999). Building a network theory of social capital. Connections, 22(1), 28–51.
Lin, N., Cook, K.S. & Burt, R.S. (2001). Social Capital: Theory and Research. New
Brunswixk, NJ: Transaction Publishers.
Line Producing. (2013). Crew Rates in the USA for I.A.T.S.E. Members. Retrieved April
6, 2013, from LineProducing.com: http://www.lineproducing.com/index.php/iatse
Linebaugh, K. (2011, February 17). Michigan wants to cut film-industry credits.
Retrieved April 5, 2013, from The Wall Street Journal:
http://online.wsj.com/article/SB1000142405274870465770457615042355801160
8.html
Link, A. N., & Scott, J. T. (2003). US science parks: the diffusion of an innovation and
its effects on the academic missions of universities. International Journal of
Industrial Organization, 21(9), 1323–1356.
Littel, R.C., Pendergast, J. & Nata R. (2000). Tutorial in biostatistics: Modeling
covariance structure in the analysis of repeated measures data. Statistics in
Medicine, 19, 1793-1819.
Longwell, T. (2012, October 5). The biz's taxing situation: Net benefit of incentive
programs is complicated, controversial. Retrieved April 5, 2013, from Variety:
http://variety.com/2012/film/news/the-biz-s-taxing-situation-1118060338/
Lucas,R. C. (2011). Crafting digital cinema: Cinematographers in contemporary
Hollywood. Unpublished Dissertation.
Lukinbeal, C. (2004). The rise of regional film production centers in North America,
1984-1997. GeoJournal , 59, 307-321.
Lundin, R.A. & Soderholm, A. (1995). A theory of the temporary organization.
203

Scandinavian Journal of Management, 11(4), 437-455.
Luther, W. (2010). Movie production incentives: Blockbuster support for lackluster
policy. Washington D. C.: The Tax Foundation.
Malecki, E. (1997). Technology and economic development: The dynamics of local,
regional and national change. University of Illinois at Urbana-Champaign's
Academy for Entreprenerial Leadership Historical Research Conference in
Entrepreneurship. Retrieved June 6, 2012 from http://ssrn.com/abstract=1496226
Marshall, A. (1890). Principles of Economics. Retrieved June 6, 2012 from:
www.econlib.org/library/Marshall/marP.html
Martin, P. & Ottaviano, G.I.P. (1999). Growing locations: Industry location in a model of
endogenous growth. European Economic Review, 43, 281-302.
Maskell, P. & Malmberg, A. (1999). Localised learning and industrial competitiveness.
Cambridge Journal of Economics, 23(2), 167-185.
Mayer, J.W. & Rowan, B. (1977). Institutionalized organizations: Formal structure as
myth and ceremony. American Journal of Sociology, 83(2), 340-363.
Mayer, V., & Goldman, T. (2010). Hollywood handouts: Tax credits in the age of
economic crisis. Jump Cut , 52, Retrieved October, 3, 2001 from
http://www.ejumpcut.org/currentissue/mayerTax/text.html.
McChesney, R.W. (1999). Rich media, poor democracy: Communication politics in
dubious times. New York: The Free Press.
McDonald, A. (2011). Down the rabbit hole: The madness of state film incentives as a
solution to runaway production. University of Pennsylvania Journal of Business
Law. Accessed April 6 2013 at http://ssrn.com/abstract=1495091
McDonald, A. (2009). Down the rabbit hole: The madness of state film incentives as a
204

solution to runaway production. Accessed April 5, 2013 from
http://works.bepress.com/adrian_mcdonald/3
McGrath, C. & Krackhardt, D. (2003). Network conditions for organizational change.
Journal of Applied Behavioral Scienve, 39(3), 324-336.
McPherson, M., Smith-Lovin, L., & Cook, J. M. (2001). Birds of a feather: Homophily in
social networks. Annual Review of Sociology, 415–444.
Messallam, A.A. (1998). The organizational ecology of investment firms in Egypt:
Organizational founding. Organization Studies, 19(2), 23-
Mezias, J. M., & Mezias, S. J. (2000). Resource partitioning, the founding of specialist
firms, and innovation: The American feature film industry, 1912-1929.
Organization Science , 11 (3), 306-322.
Miller, W. P. (n.d.). Economic multipliers: How communities can use them for planning.
Fayetteville, AR: The University of Arkansas Department of Community and
Economic Development.
Mizruchi, M.S. (1996). What do interlocks do? An analysis, critique and assessment of
research on interlocking directorates. Annual Review of Sociology, 22, 271-298.
Monge, P. & Matei, S.A. (2004). The role of the global telecommunications network in
bridging economic and political divided, 1989 to 1999. Journal of
Communication, 54(3), 511-531.
Monitor Company. (1999). US runaway film and television production study report.
Santa Monica: Monitor Company .
Moore, S. M. (2009). Commentary: Could federal tax credit end economic war among
states? The Hollywood Reporter. Retrieved April 1 from
205

http://www.hollywoodreporter.com/news/commentary-federal-tax-credit-
economic-84707
Moroney, T. (2010). Strapped states to Hollywood: Stay home. Retrieved April 4, 2013,
from Bloomberg Businessweek:
http://www.businessweek.com/magazine/content/10_49/b4206031973534.htm
Morrison, A. (2008). Gatekeepers of knowledge within industrial districts: Who they are,
how they interact. Regional Studies, 42(6), 817-835.
Motion Picture Association of America (2011). Theatrical market statistics, 2011.
Retrieved April 5, 2013 from http://www.mpaa.org/resources/5bec4ac9-a95e-
443b-987b-bff6fb5455a9.pdf
Motti, R. (1994). Producing artistic value: The Case of rock music. The Sociological
Quarterly, 35,85–102.
National Conference of State Legislatures (2011). State film production incentives and
programs. Retrieved April 7, 2013 from http://www.ncsl.org/issues-
research/budget/state-film-production-incentives-and-programs.aspx
Newman, M.E.J. (2000). Models of the small world. Journal of Statistical Physics, 101,
819-841.
Newman, M.E.J. ( 2001). Scientific collaboration networks, II: Shortest paths, weighted
networks and centrality. Physical Review E, 64, 1-7.
Newman, M.E.J., Strogatz, S.H. & Watts, D.J.(2001). Random graphs with arbitrary
degree distributions and their applications. Physical Review E. Retrieved Jun 6,
2012, from: www.arxiv.org/pdf/cond-mat/0007235v2.pdf
206

New York State Business Information Center. (n.d.). New York State Sales Tax
Exemptions for Film and Video Production. Retrieved March 16, 2013, from NY
Loves Film: http://www.nylovesfilm.com/pdf/SalesTaxExemption.pdf
North Carolina Film Office. (2011, July 13). Cameras In Wilmington Roll On Series'
Final Season. North Carolina Film Office. Retrieved April 30, 2013, from
http://www.ncfilm.com/news/557/115/Cameras-In-Wilmington-Roll-On-Series-
Final-Season.html
Orsenigo, L., Pammolli, F. Riccaboni, M., Bonaccorsi, A. & Turchetti, G. (1998). The
evolution of knowledge and the dynamics of an industry network. Journal of
Management and Governance, 1, 147-175.
Orsenigo, L., Pammolli, F. & Riccaboni, M. (2001). Technological change and network
dynamics: Lessons from the pharmaceutical industry. Research Policy, 30, 485-
508.
Park, H.W., Barnett, G. A., & Chung, J. C.(2011). Structural changes in the 2003–2009
global hyperlink network. Global Networks, 11(4), pp. 522–42.
Parks, S. (2012). The insider's guide to independent film distribution. Waltham, MA:
Focal Press.
Patten, D. (2012, September). Governor OKs $100M Film-TV Tax Credit Until 2017; Is
It Enough To Keep Up With Rival States? Deadline.com. Retrieved April 6, 2013,
from http://www.deadline.com/2012/09/governor-oks-california-film-tv-tax-
credit-program-until-2017-jerry-brown-production-georgia-louisana/
Pendakur, M. (2008). Hollywood and the state: The American film industry cartel in the
age of globalization. In P. McDonald, & J. Wasko (Eds.), The contemporary
Hollywood film industry (pp. 182-194). Malden, MA: Blackwell Publishing.
207

Pennington, A. (2011, October). The future of online movie distribution. Retrieved April
6, 2013 from http://www.csimagazine.com/csi/The-future-of-online-movie-
distribution.php
Petersen, T., & Koput, K. W. (1991). Density dependence in organizational mortality:
Legitimacy or unobserved heterogeneity? American Sociological Review, 399–
409.
Phillips, A., Kline, R. & Fox, W. (2012). Evaluating the effectiveness of state film tax
credit programs. Commissioned by the Motion Picture Association of America.
Retrieved June 6, 2012 from: www.deadline.com/vimg.net/wp-
content/uploads/2012/05/Motion-Picture-assoc.-film-credit-
study_120510071748.pdf
Piore, M., & Sabel, C. (1984). The second industrial divide. New York: Basic Books.
Plosila, W. H. (2004). State science-and technology-based economic development policy:
History, trends and developments, and future directions. Economic Development
Quarterly, 18(2), 113–126.
Podolny, J. M., Stuart, T. E., & Hannan, M. T. (1996). Networks, knowledge and niches:
Competition in the worldwide semiconductor industry, 1984-1991. American
Journal of Sociology , 103 (3), 659-689.
Pontikes, E.G. & Barnett, W.P. (2010). When to be a nonconformist entrepreneur.
Unpublished manuscript draft. Retrieved June 6, 2012 from www.faculty-
gsb.stanford.edu/barnett/documents/IconicEventsApril142010_ep.pdf
Polyani, M. (1983). The Tacit Dimension. Chicago: University of Chicago Press.
Porter, M.E. (1998) Clusters and the New Economics of Competitiveness. Harvard
Business Review, December, pp. 77-90.
208

Powell, W. W. (1990). Neither market nor hierarchy. Research in Organizational
Behavior , 12, 295-336.
Rao, H. (2005). Interorganizational ecology. In J.A.C. Baum (Ed.) The Blackwell
Companion to Organizations.Malden, MA: Blackwell.
Ratman, Y. (2009, January 29). Investing in film using IRS section 181 for Federal tax
credits vs. New Markets real estate credits. Retrieved April 5, 2013, from
ArrticlesBase: http://www.articlesbase.com/business-articles/investing-in-film-
using-irs-section-181-for-federal-tax-credits-vs-new-markets-real-estate-credits-
746745.html
Rees, J. & Stafford, H. (1986). Theories of regional growth and industrial locaiton: Their
relevance for understanding high-technology complexes. In J. Rees (Ed.)
Technology, Regions and Policy. Towtoa, NJ: Rowman & Littlefield.
Robins, G. (2011) Exponential graph models for social network analysis. In Scott, J., &
Carrington, P. J. (2011). The SAGE Handbook of Social Network Analysis. SAGE.
Robbins, J.A. (1993). Organization as strategy: Restructuring production in the film
industry. Strategic Management Journal, 14, 103-118
Rogers, E.M. (2003 [1962]). Diffusion of Innovations. New York: The Free Press.
Romanelli, E. (1989). Environments and strategies of organization start-up: Effects on
early survival. Administrative Science Quarterly, 34(3), 369-387.
Rossman, G., Esparza, N., & Bonacich, P. (2010). I'd like to thank the Academy: Team
spillovers and network centrality. American Sociological Review , 75 (1), 31-51.
Schatz, T. (2010). The Genuis of the System. Minneapolis: University of Minnesota Press.
209

Schatz, T. (1997). The return of the Hollywood studio system. In A. Schiffrin (Ed.),
Conglomerates and the media (pp. 73-106). New York: The New Press.
Scheiber, H. N. (2001). American federalism and a working system and as a
constitutional model. In Federalism and decentralization: Perspectives for the
transformation process, pps 51-70. Rose, J. & J. C. Traut, Eds. New York:
Palgrave.
Schilling, M.A. & Phelps, C.C. (2007). Interfirm collaboration networks: The impact of
large-scale network structure on firm innovation. Management Science, 53,7,
1113-1126.
Scott, A.J. (1998). New industrial spaces: flexible production organization and regional
development in North America and western Europe. London: Pion.
Scott, A. J. (2002). A new map of Hollywood: The production and distribution of
American motion pictures. Regional Studies , 39 (9), 957-975.
Scott, A. J. (2005). On Hollywood: The place, the industry. Princeton: Princeton
University Press.
Simon, H.A. (1964). On the concept of organizational goal. Administrative Science
Quarterly, 9(1), 1-22.
Singh, J. (1986). Performance, slack and risk-taking in organizational decision-making.
Academy of Management Journal, 29(3), 562-585.
Soderlund & Andersen 1998
Sorensen, O. (2003). Social networks and industrial geography. Journal of Evolutionary
Economics, 13, 513-527.
Sorensen, O. & Stewart, T.E. (2008) Bringing the context back in: Settings for syndicate
210

partners in venture capital investment networks. Administrative Science
Quarterly, 53, 266-294.
Sorensen, O. & Waguespack, D.M. (2006). Social structure and exchange: Self-
confirming dynamics in Hollywood. Administrative Science Quarterly, 51(4),
560-589.
Soutar, P. (2011, June). Kansas Reloads Film Production Tax Credit After Two-Year
Suspension. Retrieved April 7, 2013, from http://watchdog.org/36942/ks-kansas-
reloads-film-production-tax-credit-after-two-year-suspension/
Specht, P.H. (1993). Munificence and carrying capacity of the environment and
organization formation. Entrepreneurship: Theory and Practice, 17(2), 77-
Squires, A. (2013, January 27). Online Venture Takes On Film Distribution
Heavyweights : NPR. Retrieved April 8, 2013, from
http://www.npr.org/2013/01/27/170408129/online-venture-takes-on-film-
distribution-heavyweights
Storper, M. (1989). The transition to flexible specialization in the US film industry:
External economies, the division of labor and the crossing of industrial divides. In
A.Amin (Ed). Post-Fordism: A Reader. London: Blackwell.
Story, L., Fehr, T., & Watkins, D. (2012). The United States of subsidies. New York
Times. Retrieved April 7, 2013 from http://www.nytimes.com/interactive/2012/
12/01/us/government-incentives.html?_r=0
Suchman, M. C. 1995. Managing legitimacy: Strategic and institutional approaches.
Academy of Management Review, 20: 571–610.
Swaminathan, A. (2001). Resource Partitioning and the Evolution of Specialist
211

Organizations: The Role of Location and Identity in the U.S. Wine Industry.
Academy of Management Journal , 44 (6), 1169-1185.
Swaminithan, A. (1995). The proliferation of specialist organizations in the American
wine industry, 1941-1990. Administrative Science Quarterly , 40, 653-680.
Tabachnick, B. G. & Fidell, L.S. (2006). Using multivatiate statistics (5
th
Edition).
Boston: Pearson.
Takeda, Y., Kajikawa, Y., Sakata, I. & Matsushima, K. (2008). An analysis of
geographical agglomeraton and modularized industrial networks in a regional
cluster: A case study at Yamagata prefecture in Japan. Technovation, 28, 531-539.
Tannenwald, R. (2010, December 9). State film subsidies: not much bang for too many
bucks. Retrieved April 5, 2013, from Center on Budget and Policy Priorities:
http://www.cbpp.org/cms/?fa=view&id=3326
Tannenwald, R., Shure, J., & Johnson, N. (2011). Tax flight is a myth: Higher state taxes
bring more revenue, not more migration. Retrieved April 4, 2013, from The
Center on Budget and Policy Priorities:
http://www.cbpp.org/cms/index.cfm?fa=view&id=3556
Uzzi, B. (1997). Social structure and competition in interfirm networks. Administrative
Science Quarterly , 42 (1), 35-67.
Uzzi, B., & Spiro, J. (2005). Collaboration and creativity: The small world problem.
American Journal of Sociology , 111 (2), 447-504.
Ulrich, P.C. & Simmens, L. (2001). Motion picture production: To Run or Stay made in
the USA? Entertainment Law Review, 21,3, 357-370.
Vary, A.B. (2012). Can California remain the world’s movie capital? Filmmakers, execs
212

weigh in on tax incentives Retrieved April 7, 2013, from
http://insidemovies.ew.com/2012/10/05/california-tax-breaks/
Valente, T. W. (2005).Network models and methods for studying the diffusion of
innovations. In P. Carrington, J. Scott & S. Wasserman (Eds.) Models and
Methods in Social Network Analysis: New York: Cambridge University Press,
pps. 98-116.
Venter, R.B., Rogerson, C.M., Semens, A. & Myres, K.( 2012). FIFA World Cup 2010:
Implications for and effects on the entrepreneurial performance of South Africa's
informal sector. Africal Journal of Business Management, 6(1), 449-458.
Veron, L. (1999). The competitive advantage of Hollywood industry. Working paper,
Center for International Studies, University of Southern California. Available at
https://www.cc.columbia.edu/sec/dlc/ciao/wps/vel01/.
Verrier, R. (2011, January 19). Iowa film tax credit program racked by scandal. Los
Angeles Times. Retrieved from
http://articles.latimes.com/2011/jan/19/business/la-fi-ct-onlocation-20110119
Vipond, M. (2011). The Mass Media in Canada. Ontario: James Lorimer & Co.
Von Ende, C. N. (2001). Repeated measures analysis: Growth and other time-dependent
measures. In S.M. Scheiner & J. Guervitch, Eds. Design and Analysis of
Ecological Experiments. Oxford: Oxford University Press.
Wade, J.B., Swaminathan, A.& Saxon, M.S. (1998). Normative and resource flow
consequences of local regulations in the American brewing industry, 1845-1918.
Administrative Science Quarterly, 43(3), 905-935.
Walker, Jack. 1969. The diffusion of innovations among the American states. American
Political Science Review, 63 (September 1969): 880-899.
213

Wallerstein, I. (2004). World-Systems Analysis: An Introduction. Durham, North
Carolina: Duke University Press.
Wallerstein, I. (1974). The Modern World-System, vol. I: Capitalist Agriculture and the
Origins of the European World-Economy in the Sixteenth Century. New
York/London: Academic Press.
Walters, G.D. (2007). Using Poisson class regression to analyze count data in
correctional and forensiz psychology: A relatively old solution to a relatively new
problem. Criminal Justice and Behavior, 34(12), 1659-1674.
Wan, W.P. & Hoskinsson, R.E. (2003). Home country environments, corporate
diversification strategies and firm performance. Academy of Management
Journal, 46(1), 27-45.
Wasko, J. (2003). How Hollywood works. London: Sage.
Waterman, D. (2005). Hollywood's Road to Riches. Cambridge, MA: Harvard University
Press.
Watts, Duncan J. 1999. Small Worlds: The Dynamics of Networks between Order and
Randomness. Princeton, N.J.: Princeton University Press.
Weick, K.E. & Roberts, K.H. (1993). Collective mind in organizations: Heedful
interrelating on flight decks. Administrative Science Quarterly, 38(3), 357-381.
Wegner, E.C.& Snyder, W.M. ( 2000). Communities of practice: The organizational
frontier. Harvard Business Review,78(1), 139-
Weinstein, B. L., Gross, H. T. & rees, J. (1985). Regional Groeth and Decline in the
United States. New York: Praeger.
Welch, S. & Thompson, K. (1980). The impact of feceral incentives on state policy
innovation. American Journal of Political Science, 24(4), 715-729.
214

Whitley, R. (2006). Project-based firms: new organizational form or variations on a
theme? Industrial and Corporate Change, Volume 15, Number 1, pp. 77–99
Williamson, O.E. (1991). Comparative economic organization: The analysis of discrete
structural alternatives. Administrative Science Quarterly, 36(2), 269-296.
Williamson, O. E. (1985) The Economic Institutions of Capitalism: Firms, Markets and
Relational Contracting. Free Press, New York.
Williamson, O.E. (1981). The economics of organization: The transaction cost approach.
American Journal of Sociology, 87(3), 548-577.
Winch, G. M. (1986) The labour process and labour market in construction, International
Journal of Sociology & Social Policy 6, 103–16.
Zheng Warner (2010). Business incentive use among US local governments:
Accountability and policy learning. Economic Development Quarterly, 24(4),325-
336.
Zuckerman, E. W. (2005). Do firms and markets look different? Repeat collaboration in
the feature film industry, 1935-1995. Unpublished manuscript.

Abstract (if available)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The role of capabilities in new alliance creation and performance: a study of the biotechnology industry

PDF

The coevolution of multimodal, multiplex, and multilevel organizational networks in development communities

PDF

The evolution of multidimensional and multilevel networks in online crowdsourcing

PDF

Understanding alliance building and communication across multiple technology platforms among California's immigrant-serving NGOs

PDF

The evolution of knowledge creation online: what is driving the dynamics of Wikipedia networks

PDF

Organizational mimicry in American social movement communities: an analysis of form communication effects on the evolution of crisis pregnancy centers, 1989-2009

PDF

The evolution of scientific collaboration networks

PDF

Staying ahead of the digital tsunami: strategy, innovation and change in public media organizations

PDF

Communicating organizational knowledge in a sociomaterial network: the influences of communication load, legitimacy, and credibility on health care best-practice communication

PDF

A multitheoretical multilevel explication of crowd-enabled organizations: exploration/exploitation, social capital, signaling, and homophily as determinants of associative mechanisms in donation-...

PDF

Understanding the varied effects of leadership on employee retention in high stress work environments

PDF

Collaboration: is it worth it? The Magnolia Community Initiative from the perspective of initiative partner participants

PDF

Quango reforms and challenges in South Korea: social relations, informal networks, and hidden actions

PDF

Developing countries in the digital era: state and business interactions for industrial upgrading

PDF

Hollywood's 2015 diversity problem and the powerful impact of film and television

PDF

Pathway programs for HBCU students in television & film…

PDF

Homegrown violent extremism: designing a community-based model to reduce the risk of recruitment and radicalization

PDF

The functions of the middleman: how intermediary nonprofit organizations support the sector and society

PDF

The future of work: defining a healthy ecosystem that closes skills-gaps

PDF

The significance of investigating the absence of Black decision-makers in television and feature films

Asset Metadata

Creator O'Brien, Nina (f.)

Core Title Effects of economic incentives on creative project-based networks: communication, collaboration and change in the American film industry, 1998-2010

School Annenberg School for Communication

Degree Doctor of Philosophy

Degree Program Communication

Publication Date 07/27/2015

Defense Date 06/03/2013

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag economic development,film industry,OAI-PMH Harvest,organizational ecology,organizational networks,project-based industries

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Monge, Peter R. (committee chair), Fulk, Janet (committee member), Young, S. Mark (committee member)

Creator Email nfobrien@gmail.com,nobrien@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-301775

Unique identifier UC11292797

Identifier etd-OBrienNina-1857.pdf (filename),usctheses-c3-301775 (legacy record id)

Legacy Identifier etd-OBrienNina-1857.pdf

Dmrecord 301775

Document Type Dissertation

Format application/pdf (imt)

Rights O'Brien, Nina, F.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA