Close
Logo
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected 
Invert selection
Deselect all
Deselect all
 Click here to refresh results
 Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Three papers on dynamic externalities and regional growth
(USC Thesis Other) 

Three papers on dynamic externalities and regional growth

doctype icon
play button
PDF
 Download
 Share
 Open document
 Flip pages
 More
 Summarize
 Download a page range
 Download transcript
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content






THREE PAPERS ON DYNAMIC EXTERNALITIES AND REGIONAL GROWTH


by
Eunha Jun



A Dissertation Presented to the  
FACULTY OF THE USC GRADUATE SCHOOL
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY

December 2016
 
ii



ACKNOWLEDGEMENTS
I would like to thank Dr. Richard Green, the chair of my dissertation committee, for his attentive
advice and guidance. I am also grateful to the support and encouragement from the other
committee members Dr. Marlon Boarnet and Dr. James Moore. Most importantly, I would like to
thank to my family - my parents, my husband Junyoug, and my daughter Seoyeon – for their
love and patience throughout my long academic journey.  

 
iii



TABLE OF CONTENTS
ACKNOWLEDGEMENTS................................................................................................................. ii
TABLE OF CONTENTS ................................................................................................................... iii
LIST OF TABLES .............................................................................................................................v
ABSTRACT ................................................................................................................................... viii
CHAPTER ONE: DYNAMIC EXTERNALITIES AND EMPLOYMENT GROWTH IN U.S. CITIES ...1
Introduction ................................................................................................................................1
Literature Review........................................................................................................................3
Theories of Growth ..............................................................................................................3
Empirical Studies on Agglomeration Externalities ..................................................................4
Data and Variables ......................................................................................................................6
Construction of the Data Set .................................................................................................6
Description of the Data .........................................................................................................9
Variables and Descriptive Statistics ..................................................................................... 10
Regression Results .................................................................................................................... 13
Employment Growth in Cities ............................................................................................. 13
Growth in Cities and National Growth ................................................................................. 15
Employment Growth and Manufacturing Industry ................................................................ 17
Employment Growth of Smaller Industries .......................................................................... 19
Conclusion................................................................................................................................ 20
CHAPTER TWO: DYNAMIC EXTERNALITIES AND WAGE GROWTH IN U.S. CITIES............... 59
Introduction .............................................................................................................................. 59
Literature Review...................................................................................................................... 60
Dynamic Externalities and Wage Growth ............................................................................ 60
Theories of Productivity Growth ......................................................................................... 61
Data and Variables .................................................................................................................... 62
Construction of the Data Set ............................................................................................... 62
iv



Description of the Data ....................................................................................................... 64
Variables and Descriptive Statistics ..................................................................................... 65
Regression Results .................................................................................................................... 67
Wage Growth in Cities ....................................................................................................... 67
The Impact of Dynamic Externalities by Industry ................................................................. 70
Conclusion................................................................................................................................ 74
CHAPTER THREE: DYNAMIC EXTERNALITIES AND REGIONAL GROWTH IN SOUTH KOREA
........................................................................................................................................................ 86
Introduction .............................................................................................................................. 86
Empirical Studies in Korea ................................................................................................. 87
Regional Industry in South Korea........................................................................................ 88
Characteristics of South Korean Industries ........................................................................... 90
Data and Variables .................................................................................................................... 98
Construction of the Dataset ................................................................................................. 98
Variables and Descriptive Statistics ..................................................................................... 99
Regression Results .................................................................................................................. 104
Glaeser et al.’s (1992) Externality Indicators...................................................................... 104
Alternative Externality Indicators ...................................................................................... 105
National Growth and Other Control Variables .................................................................... 107
Conclusion.............................................................................................................................. 110
REFERENCES............................................................................................................................... 122

 
v



LIST OF TABLES
Table 1.1.1A. Largest and Smallest Cities in 1986 ................................................................ 23
Table 1.1.1B. Largest and Smallest Cities in 1998 ................................................................ 25
Table 1.1.1C. Largest and Smallest Cities in 2012 ................................................................ 27
Table 1.1.2A. 20 Largest City-Industries in 1986 ................................................................. 29
Table 1.1.2B. 20 Largest City-Industries in 1998.................................................................. 30
Table 1.1.2C. 20 Largest City-Industries in 2012.................................................................. 31
Table 1.2.1A. Variable Means and Standard Deviations, 1986-2012 (All 20 Industries) .......... 32
Table 1.2.1B. Variable Means and Standard Deviations, 1998-2012 (All 20 Industries) .......... 33
Table 1.2.1C. Variable Means and Standard Deviations, 1986-1998 (All 20 Industries) .......... 34
Table 1.2.2A. Variable Means and Standard Deviations, 1986-2012 (Top 6 Industries)........... 35
Table 1.2.2B. Variable Means and Standard Deviations, 1998-2012 (Top 6 Industries) ........... 36
Table 1.2.2C. Variable Means and Standard Deviations, 1986-1998 (Top 6 Industries) ........... 37
Table 1.3.1.1A. City-Industry Employment Growth between 1986 and 2012 (All 20 Industries)
.................................................................................................................................. 38
Table 1.3.1.1B. City-Industry Employment Growth between 1998 and 2012 (All 20 Industries)
.................................................................................................................................. 39
Table 1.3.1.1C. City-Industry Employment Growth between 1986 and 1998 (All 20 Industries)
.................................................................................................................................. 40
Table 1.3.1.2A. City-Industry Employment Growth between 1986 and 2012 (Top 6 Industries)
.................................................................................................................................. 41
Table 1.3.1.2B. City-Industry Employment Growth between 1998 and 2012 (Top 6 Industries)
.................................................................................................................................. 42
Table 1.3.1.2C. City-Industry Employment Growth between 1986 and 1998 (Top 6 Industries)
.................................................................................................................................. 43
Table 1.3.2.1A. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1986 and 2012 (All 20 Industries)................................................................... 44
Table 1.3.2.1B. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1998 and 2012 (All 20 Industries)................................................................... 45
vi



Table 1.3.2.1C. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1986 and 1998 (All 20 Industries)................................................................... 46
Table 1.3.2.2A. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1986 and 2012 (Top 6 Industries) ................................................................... 47
Table 1.3.2.2B. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1998 and 2012 (Top 6 Industries) ................................................................... 48
Table 1.3.2.2C. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1986 and 1998 (Top 6 Industries) ................................................................... 49
Table 1.4.1. Manufacturing Industry Employment in Cities between 1986 and 2012 ............... 50
Table 1.4.2. Manufacturing among City-Industries between 1986 and 2012 ........................... 50
Table 1.5.1. Employment by Industry between 1986 and 2012 .............................................. 51
Table 1.5.2. Employment Share by Industry (in percentage) between 1986 and 2012 .............. 52
Table 1.6.1A. Employment Growth of Smaller Industries between 1986 and 2012 (All 20
Industries) .................................................................................................................. 53
Table 1.6.1B. Employment Growth of Smaller Industries between 1998 and 2012 (All 20
Industries) .................................................................................................................. 54
Table 1.6.1C. Employment Growth of Smaller Industries between 1986 and 1998 (All 20
Industries) .................................................................................................................. 55
Table 1.6.2A. Employment Growth of Smaller Industries between 1986 and 2012 (Top 6
Industries) .................................................................................................................. 56
Table 1.6.2B. Employment Growth of Smaller Industries between 1998 and 2012 (Top 6
Industries) .................................................................................................................. 57
Table 1.6.2C. Employment Growth of Smaller Industries between 1986 and 1998 (Top 6
Industries)................................................................................................................... 58
Table 2.1.1. 10 Largest Real Wage Cities ............................................................................ 75
Table 2.1.2A. 20 Largest Real Wage City-Industries in 1998 ................................................ 76
Table 2.1.2B. 20 Largest Real Wage City-Industries in 2012................................................. 77
Table 2.1.3. 20 Most Grown Real Wage Cities, 1998 and 2012 ............................................. 78
Table 2.2.1. Variable Means and Standard Deviations (All 20 Industries) .............................. 79
Table 2.2.2. Variable Means and Standard Deviations (Top 6 Industries) ............................... 80
vii



Table 2.3.1. City-Industry Nominal Wage Growth between 1998 and 2012 (All 20 Industries) 81
Table 2.3.2. City-Industry Real Wage Growth between 1998 and 2012 (All 20 Industries) ...... 82
Table 2.4. City-Industry Nominal Wage Growth between 1998 and 2012 (Top 6 Industries) ... 83
Table 2.5.1. Employment Growth and Dynamic Externalities by Industry between 1998 and
2012 .......................................................................................................................... 84
Table 2.5.2. Real Wage Growth and Dynamic Externalities by Industry between 1998 and 2012
.................................................................................................................................. 85
Table 3.1.1. Industries in Korea, 2008 and 2013 ................................................................. 112
Table 3.1.2. Industry Growth in Korea, between 2008 and 2013 .......................................... 113
Table 3.2.1. Regional Employment in Korea, 2008 and 2013 .............................................. 114
Table 3.2.2. Regional Employment Growth in Korea, between 2008 and 2013 ..................... 115
Table 3.3. Concentration of Industries in Seoul Metropolitan Area in Korea, 2008 and 2013 . 116
Table 3.4. Small and Medium Sized Firm Employment and Establishment by Industry in Korea,
2008 and 2013 .......................................................................................................... 117
Table 3.5. Variable Means and Standard Deviations, Korea ................................................ 118
Table 3.6.1. Regional Industry Employment Growth in South Korea between 2008 and 2013 119
Table 3.6.2. Regional Industry Employment Growth in South Korea between 2008 and 2013 120
Table 3.7. Competition(R) and Competition(N) indicator by Industry in Korea, 2008 ........... 121
 
viii



ABSTRACT
The first paper in Chapter one examines the effects of three types of dynamic externalities on
employment growth in U.S. cities. By constructing an extensive dataset for city industries
between 1986 and 2012, this study tests the theories of economic growth proposed by Marshall,
Porter, and Jacobs. The analysis result finds that competition and diversification encourage
employment growth in cites, whereas specialization does not. The evidence supports Jacob’s
theory that knowledge spillovers contribute to the economic growth of cities through the
exchange of ideas between different industries. The result further draws attention to the
importance of nationwide economic performance in investigating the growth of cities.  
The second chapter is a continuing investigation into the effect of three types of dynamic
externalities on U.S. city growth that is in parallel with the first paper. The theories of economic
growth by Marshall, Porter, and Jacobs are tested through wage-based regressions. The analysis
results suggest that diversity has positive effects on the growth of city-industry wages.
Specialization is found to have an effect on the wage growth of the largest city-industries. Unlike
in the employment growth analysis, no evidence for competition is found in the wage growth
analysis. The additional regression for individual industry also shows interesting findings.  
Chapter three examines how dynamic externalities have affected recent regional growth in South
Korea. The effect of industrial structure on local employment growth between 2008 and 2013 is
tested with various measures for externalities. The analysis results do not confirm any of the
theories of Marshall, Porter, or Jacobs. This finding suggests that different characteristics of
industries may affect the role of dynamic externalities. The evidence indicates that diversity of
industries impedes employment growth in South Korea. Regarding competition, regions grow
when the share of small and medium-sized firms in regional industry is high. However, at the
ix



same time, a lack of large firm makes a region hard to grow. No evidence is found regarding the
effect of specialization. It is also found that the influence of nationwide growth is dominant in
determining regional growth in South Korea.  


 
1



CHAPTER ONE: DYNAMIC EXTERNALITIES AND EMPLOYMENT GROWTH IN
U.S. CITIES  
Introduction
Over the past several decades, scholars have searched for the source of growth. Cities are
often said to be the engine of economic growth for regions and countries. Urban economists
typically attribute the importance of cities to the presence of agglomeration externalities. In cities,
knowledge is easily exchanged and transferred across people and firms due to geographical
proximity.  Diverse ideas floating around cities are combined and modified, and then new ideas
are produced. This dynamic view of cities, from the new growth theory in particular, stresses the
importance of technological change facilitated by knowledge spillovers as it contributes to
endogenous growth and, hence, economic growth (e.g. Lucas 1988; Romer 1994).  
Agglomeration externalities associated with such knowledge spillovers are often
distinguished into several types, which are Marshall-Arrow-Romer (MAR) externalities, Porter
externalities, and Jacobs externalities. MAR externalities are agglomeration benefits that come
from the local concentration of a specialized industry, and they are often treated interchangeably
with localization externalities (e.g. Henderson, Kuncoro, & Turner, 1995).  MAR externalities
are external to firms but internal to an industry within a city.  This type of externality is
sometimes distinguished from Porter externalities (e.g., Glaeser, Kallal, Scheinkman, & Shleifer,
1992).  Porter externalities also arise from specialized and geographically concentrated industries;
however, as opposed to MAR theory, which prefers local monopoly, Porter (1990) argued that
local competition helps to sustain industrial growth.  Jacobs externalities, unlike MAR and Porter
externalities, are benefits that come from the diversity of geographically proximate industries.  
Jacobs (1969) also favored local competition.  
2



The existence of agglomeration externalities and their impact on economic growth have
attracted a great deal of attention from scholars in the various fields of social science. A large
amount of empirical and theoretical literature on these externalities has been produced from
various perspectives. In spite of this solid theoretical foundation, however, our understanding of
which externalities are the most beneficial to growth is still limited. One reason for this is that
the evidence from the empirical literature is often contradictory. For instance, Henderson (1986)
found positive MAR externalities, whereas Glaeser, Kallal, Scheinkman, and Shleifer (1992)
found the same to be negative.  For Jacobs externalities, Jaffe, Trajtenberg, and Henderson (1993)
and Glaeser et al. (1992) found supportive evidence, but Henderson (2003a) and Combes (2000)
argued that diversity can negatively affect some industries.  While Henderson (1986) found
evidence of urban diseconomies on productivity growth, Nakamura (1985) and Moomaw (1988)
suggested that positive urban externalities exist. Another reason for our limited understanding in
this area is the lack of continuous empirical works. Although there are a good number of
empirical studies, it is not easy to find works that are comparable to or that complement each
other. Knowledge spillover and externalities are hard to capture or measure in nature. Moreover,
researchers take various approaches with different data for different countries. Because most
empirical works are based on the data for particular industry sectors, periods, or regions, the
usefulness or applicability of the findings in establishing public policy is limited. Methodological
differences among studies also tend to cause results to differ. Beaudry and Schiffauerova (2009)
showed that measurement and methodological issues (e.g., the levels of industrial and
geographical aggregation and the choice of performance measures) are the main causes that
make it difficult for the empirical studies on knowledge externalities to achieve consensus.  
3



In light of this, the present study aims to provide follow-up research to a seminal work by
Glaeser et al. (1992) on growth in cities. Glaeser et al. tested theories of economic growth by
Marshall-Arrow-Romer, Porter, and Jacobs by looking at the growth of large industries in U.S.
cities between 1956 and 1987. The sources of spillovers were classified into specialization,
diversification, and competition according to these theories. The evidence they found supported
Jacobs’ theory that diversity and competition contributes to growth.  
The present empirical study updates their work by reproducing it with recent data, from
1986 to 2012. The findings of this study confirm Glaeser et al.’s arguments in general. The
evidence suggests that industry employment grows in competitive and diverse cities, rather than
in specialized cities. This study further highlights the importance of nationwide economic
situations in studying city growth.  
Literature Review
Theories of Growth
A body of theory related to long-run economic growth emerged in the 1980s.  This new
growth theory questioned the neoclassical view of capital accumulation as an engine of growth.  
The new growth theory distinguished itself by emphasizing economic growth as an endogenous
outcome of an economic system. Positive agglomeration externalities were emphasized as a
major source of endogenous growth. Dense urban environments facilitate interaction between
people, and knowledge is disseminated through the exchange of ideas and learning from other
people, often without payment. Such knowledge spillovers result in a technological increase that
contributes to city growth. Thus, agglomeration externalities that are related to knowledge
spillovers are categorized as dynam ic externalities, in contrast to static urban externalities.  
4



Though there seems to be consensus on the importance of dynamic externalities for
growth, theories of city growth differ in two aspects. The first of these is whether knowledge
spillovers are generated within the same industries or between other industries, and the second is
how local competition influence knowledge spillovers.
According to Marshall (1890), Arrow (1962), and Romer (1986), specialization of a
particular industry in a region helps growth, as firms and workers in the same industry can
effectively accumulate and exchange common knowledge and experiences. However, the
presence of too many firms in a region may trigger competition. Firms then may be able to
internalize the externalities by monopolizing the market (Romer, 1994). MAR externalities are
often treated interchangeably with localization externalities (e.g. Henderson et al., 1995).  
Porter (1990) agrees with MAR about the importance of within-industry knowledge
spillovers but argues competition is better for growth than monopoly. Local competition
motivates firms and then promotes innovation and technological advances. Porter argues that the
benefit of competition is greater than any harm, such as the decline of returns to the innovator.  
Unlike MAR and Porter, Jacobs (1969) thinks that the diversity of geographically
proximate industries promotes the growth of cities, because the most important innovations often
come from outside the industry. Jacobs also favors competition. Her prediction is that locally
competitive cities should grow more quickly as competition facilitates interaction and induces
faster transmission of technologies and innovations.      
Empirical Studies on Agglomeration Externalities
Along with theories, empirical studies have investigated the source of externalities related
to specialization, diversification, and competition. Despite the fact that empirical literature has
5



generally found the positive effects of agglomeration economies on economic growth, the
debates on the effects of dynamic externalities remain inconclusive.  
Glaeser et al. (1992) set up an empirical growth analysis that tested, among other points,
which dynamic externalities are more important. They analyzed U.S. city-industry data in 1956
and 1987. Regional economic performance was measured in terms of employment. To test the
theories of MAR, Porter, and Jacobs, they suggested a way to measure the level of specialization,
competition, and diversification of city-industries, which has since been adopted in many later
empirical studies, including the present study. Their findings supported Jacobs externalities, as
their regression results suggested that local competition and diversity of industries, rather than
specialization, encourage employment growth in cities.          
In a study focusing on similar questions, Henderson, Kuncoro, and Turner (1995) argued
that the kind of industry and the stage of product development are important in determining the
effect of dynamic externalities. They also analyzed employment growth in U.S. metropolitan
areas between 1970 and 1987, but they examined only five manufacturing industries. They found
that MAR externalities, but not Jacobs externalities, are effective for mature capital goods
industries, whereas both MAR and Jacobs externalities work for new high-tech manufacturing
industries. They thus concluded that Jacobs externalities are important in attracting new
industries, while MAR externalities are important for retaining the industry.
Building on these two works, many later empirical studies investigated the relationship
between industrial structure and economic growth, with some modifications and particular
emphases depending on the purpose and the interest of each study. The majority of these studies
have been conducted in the U.S. (Acs & Armington, 2004; Feldman & Audretsch, 1999;
6



Harrison, Kelley, & Gant, 1996; Malpezzi, Seah, & Shilling, 2004), while a few have focused on
European countries like the UK (Baptista & Swann, 1998), Italy (Cingano & Schivardi, 2004;
Paci & Usai, 2000), France (Combes, 2000), and the Netherlands (van der Panne, 2004). Studies
in Asian countries, such as South Korea (Henderson, Lee, & Lee, 2001; Lee, Kim, & Hong,
2005), Japan (Dekle, 2002), China (Mody & Wang, 1997), and other developing countries are
relatively scarce.  
The resulting evidence is mixed and often conflicting. Most recently, there has been an
attempt to explain the variations in empirical study results through meta-analysis of the existing
literature (Beaudry & Schiffauerova, 2009; de Groot, Poot, & Smit, 2007; Melo, Graham, &
Noland, 2009). After reviewing and analyzing dozens of papers, all of these efforts found that
study characteristics affect the results of the empirical research. Evidence favoring Jacobs
externalities is found more often than evidence supporting MAR or Porter externalities. However,
the effects of specialization, competition, and diversity tend to be detected differently according
to the sectoral, temporal, and spatial choices of the study. The design of the externality measures
also matters, as does the choice of dependent variables. Overall, the evidence observed in these
studies calls for more refined empirical research.  
Data and Variables
Construction of the Data Set
Following Glaeser et al. (1992), the data set was constructed from the County Business
Patterns (CBP) data, produced by the U.S. Census Bureau. The CBP provides annual subnational
economic data by industry, at various geographical levels. CBP data include employment, annual
payroll, and number of establishments. The information is reported for 4-digit SIC codes or 3-, 4-,
5-, and 6-digit NAICS codes for every county in the United States from 1986 to 2012. Glaeser et
7



al. chose the 1956 and 1987 data because they were the first and the last year available at the
time of the study. In the present study, the 1986, 1998, and 2012 data were used. 1986 was the
first year of downloadable data available, 1998 was the first year that CBP was tabulated based
on NAICS, and 2012 was the last year available.  
One difficulty associated with using historical industrial data is that the compatibility of
data has been affected by changes in the industry classification system. The biggest change
occurred in 1998, when the North American Industry Classification System (NAICS) replaced
the Standard Industrial Classification System (SIC). While some of the individual SIC industries
directly correspond to industries classified according to the NAICS system, many SIC industries
split into several NAICS industries.  
The Bureau of Labor Statistics provides the ratios of employment from SIC to NAICS
and from NAICS to SIC for 2-, 3-, and 4-digit data series. These ratios are prepared from the
Quarterly Census of Employment and Wages Program and can be used to evaluate time series
breaks and correspondence between SIC and NAICS in the Current Employment Statistics (CES)
series. The SIC to NAICS ratios for 2-digit SIC series were applied to the CBP dataset in order
to make the 1986 data (which is tabulated in SIC) comparable to the 1998 and 2012 data (which
are tabulated in NAICS) for the purposes of this study. As not only employment data but also
annual payroll and the number of establishment data have been converted based on the same
ratios of employment, the data correspondence is expected be more accurate in employment data
than in the other forms of data.          
After conversion, the dataset contains the information on employment, annual payroll,
and number of establishments by every 2-digit NAICS industry with FIPS State and County
8



codes for every county in the United States for the years 1986, 1998, and 2012. Employment
data includes data suppression flags, which report employment size in ranges instead of exact
numbers in the case of data withheld to avoid disclosure or because data do not meet publication
standards. Em ploym ent size classes for these flags are 0-19, 20-99, 100-245, 250-499, 500-999,
1,000-2,499, 2,500-4,999, 5,000-9,999, 10,000-24,999, 25,000-49,999, 50,000-99,999, and
100,000 or more. When employment data are represented by a data suppression flag, the
midpoint value of the range was used. For example, if a flag indicates employment in a county is
between 0 and 19 then employment was set as 10. When the range is 100,000 or more, the value
was set as 100,000. Wage data was obtained by dividing annual payroll by employment. In some
cases, annual payroll or employment data include missing values or zero values. The data were
excluded in these instances to avoid errors. The number of observations of the dataset thus
decreased from 7,660 to 6,719.        
In terms of the geographical level, this study focused on cities rather than counties.
Glaeser et al. adopted the standard metropolitan areas (SMAs) in 1956 and constructed 170 cities.
In this study, Centers for Medicare & Medicaid Services’ (CMS) SSA to FIPS CBSA to county
crosswalk for 2012, available at the National Bureau of Economic Research, was used to
aggregate counties to 383 cities. Thus, the dataset contains employment, number of
establishments, and wage data for 1986, 1998, and 2012 by 20 two-digit NAICS industries per
each 383 CBSA.
Glaeser et al. examined the six largest two-digit industries in terms of 1956 payroll only,
for each city, due to the limitation of hand-collected data and their interest in regionally
specialized industries. This choice of industry, however, results in a bias against small or young
industries, as Glaeser et al. noted. With the new dataset, this study ran regressions for both the
9



top six and all twenty industries. Both cases show consistent results overall, except that R
squares are lower for the top six industries.  
Description of the Data
The dataset reveals trends related to the change of major industries in cities. Tables
1.1.1A to 1.1.1C show the six largest and the six smallest cities in terms of employment, along
with the number of employment and the six largest industries, in 1986, 1998, and 2012. Tables
1.1.2A to 1.1.2C are the 20 largest city-industries, where the size is measured by the employment
each year. New York-White Plains-Wayne, NY-NJ; Los Angeles-Long Beach-Santa Ana, CA;
and Chicago-Joliet-Naperville, IL, remain the biggest cities in the U.S. during the past 27 years,
but the largest industries in those cities have changed considerably over time.
In 1956, the year from which Glaeser et al.’s data were drawn, apparel was the largest
city-industry. The largest industries in cities showed great diversity. Besides apparel, metal
products, durable/nondurable wholesale trade, transportation equipment and others were
commonly found to be the largest industries in the largest cities. In the smallest cities, on the
other hand, auto dealers, general merchandise, special trade contractors, and durable/nondurable
wholesale trade and others were the largest city-industries.  
The updated dataset for this study shows dramatic changes in industrial structure. In 1986,
manufacturing and retail trade were the most common largest industries in both the largest and
the smallest cities. Other large city-industries were health care and social assistance, wholesale
trade, finance and insurance, and professional, scientific, and technical services.  
Between 1986 and 1998, manufacturing employment dropped in the largest cities. In its
place, health care and social assistance became the largest industry in the largest cities after 1998.
10



Retail trade retained its high rank. Professional, scientific, and technical services became larger
city-industries than before, while manufacturing and wholesale trade declined steadily from 1986
to 2012.
Variables and Descriptive Statistics
Glaeser et al. designed their empirical analysis to measure the growth of the same sectors
in different cities in order to test the effects of the externalities. The choice of the variables and
the definition of the indicators of externalities in the present study followed the work of Glaeser
et al. Tables 1.2.1A to 1.2.2C present variable means and standard deviations between 1986 and
2012, 1998 and 2012, and 1986 and 1998, respectively.  
Growth theories suggest that economic growth may depend on regional specialization,
competition or diversity of industries in cities. The dependent variable is employment growth,
which is defined as the log of the employment in the end year divided by the employment in the
beginning year.
employment growth = ln
𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑒𝑒𝑒 𝑒𝑒 𝑦 𝑦 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑏 𝑒 𝑏 𝑖 𝑒𝑒 𝑖 𝑒 𝑏 𝑒𝑒 𝑦 𝑦
The indicators of externalities are defined below. The specialization indicator utilizes the
location quotient, a ratio that is widely used to compare an area’s distribution of employment by
industry to a reference area’s employment distribution. The indicator is defined as the city-
industry’s share of city employment relative to the U.S. industry’s share of the U.S. employment.  
specialization =
𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑐 𝑖𝑒𝑒 𝑒 𝑒 𝑒 𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑐 𝑖𝑒𝑒 ⁄
𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑈 . 𝑆 . 𝑒 𝑒 𝑒 𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑈 . 𝑆 . ⁄

11



This is a static measure. If the indicator value is one, it means that the city and the nation
have the same percentage of that industry in terms of employment. If the value is smaller than
one, it means that the industry has a smaller representation in the city compared to the nation. A
value greater than one, on the other hand, means the industry has a greater representation in the
city than in the nation, which can be interpreted to mean that the industry is specialized in the
city. The mean of specialization indicator of Glaeser et al.’s dataset was 3.367 in 1956. The value
calculated in this study was 1.2683 in 1986 and 1.2922 in 1998 for the top six industries. It is
evident from these findings that cities have become less specialized. A substantial decrease of
the specialization indicator mean may reflect a fundamental change in industries, such as the
decline of mining and manufacturing.    
The competition indicator is the number of firms per worker in the city-industry relative
to the number of firms per worker in the industry in the U.S.  
competition =
𝑓 𝑖𝑦 𝑒 𝑖 𝑖 𝑒 𝑐 𝑖𝑒𝑒 − 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑖 𝑒 𝑐 𝑖𝑒𝑒 − 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 ⁄
𝑓 𝑖𝑦 𝑒 𝑖 𝑖 𝑒 𝑒 ℎ 𝑒 𝑈 . 𝑆 . 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑖 𝑒 𝑒 ℎ 𝑒 𝑈 . 𝑆 . 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 ⁄

By comparing a city to the nation, we can measure how concentrated an industry is in a
city. When industries yield a value greater than one, this can be interpreted to mean that firms in
those industries are more concentrated in a city compared to the nation. Therefore, those
industries are locally more competitive. Alternatively, the value can be greater than one when
firms of an industry in a city are only smaller than their national average. Because it is very hard
to distinguish these two cases, Glaeser et al.’s approach was to use the data of only the six
biggest industries in each city to decrease the possibilities of the latter case. Still, the biggest
industries can be made up of many small firms. The mean of the competition indicator of Glaeser
12



et al.’s dataset was less than 1 (0.752) in 1956.   However it was 1.0895 in 1986 and 1.0681 in
1998 for the top six industries, suggesting that these cities have become more competitive.    
Diversity is measured as the share of total city employment held by a city’s other top five
industries.
diversity =
𝑐 𝑖𝑒 𝑒 ′
𝑖 𝑒𝑒 ℎ 𝑒 𝑦 𝑒 𝑒 𝑒 𝑓 𝑖 𝑓 𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑐 𝑖𝑒𝑒
Glaeser et al. tried to measure a variety of industries by looking at the large industry
employment in a city other than the industry in question. In a diverse city, even the largest
industries may not take a big share as in a specialized city. In this case, a low value of the
diversity indicator may suggest that a city is diverse. The mean of the diversity indicator of
Glaeser et al.’s dataset was 0.351 in 1956. The indicator mean was 0.5859 in 1986 and 0.5866 in
1998 for the top six industries. According to this indicator, cities have become less diverse.  
In addition to these variables, the national employment growth or wage growth in an
industry, the wage and employment of the city-industry in the initial year, and a dummy variable
indicating that a city is located in the South are included in the regression analysis as controls.
These variables are expected to affect local employment growth.  
When employment in an industry grows quickly in the whole country, employment in
that industry in specific cities may also grow quickly. Therefore, the national employment
growth is expected have a positive sign. The South dummy variable is also expected to have a
positive sign, because cities located in the South tend to grow faster than other cities. The effect
of the initial wage and employment is unclear.
13



Regression Results
As this study’s primary aim is to update the paper by Glaeser et al. (1992), all of their
regression result tables are reproduced in the same framework for the three test periods. The first
period is from 1986 to 2012, which is the longest period with available data. The second period
is from 1998 to 2012, as 1998 is the first year for which data were produced in NAICS. Because
historical data in different classification systems were bridged, the second period can serve as an
indirect means to check the validity of the dataset. Also, it is simply a recent period. The third
period, from 1986 and 1998, is used to test the difference between it and the second period.
Employment Growth in Cities
Tables 1.3.1.1A to 1.3.1.2C present the reproduced regression analysis results on the
employment growth in city industries. In each table, equations 1 to 3 include one of each
externality indicator, respectively, and equation 4 includes all three measures of externalities.
The overall results tend to be in accord with Glaeser et al.’s test results, and the estimation
results are consistent across test periods.  
Tables 1.3.1.1A and 1.3.1.2A show the estimated coefficients are statistically significant
and consistent in tests both for all 20 and for the top 6 industries. Regarding the externality
indicators, the coefficients of specialization and diversification indicators are negative, whereas
the coefficient of competition is positive. This suggests that if industries’ employment share is
greater in cities than that of the U.S., then employment grows slower in the city-industry. This
evidence contradicts the MAR externality, as it suggests that specialization does not promote
growth; on the contrary, a more competitive environment is related with higher growth of city-
industry employment. This result is in favor of Porter’s and Jacob’s theory. Also, cities whose
largest industries take up a relatively small proportion of the total city employment grow faster.
14



This evidence suggests that diversification induces growth, as Jacob argued. Rather than having
only a few dominant industries, having many smaller industries helps cities to grow. Overall,
cities grow with competitive and diverse industries, not with specialized industries.    
All control variables are also statistically significant and consistent. City-industries grow
when national employment grows as a whole, when initial wage is high, or when cities are
located in the South. High initial employment, however, has negative effects on the employment
growth of city-industries.  
Tables 1.3.1.1B and 1.3.1.2B show similar results. As the first and the second test periods
produced consistent results, we might be able to say that the reliability of the data conversion
method is acceptable.      
A noticeable result here is the estimates of the national growth. In table 1.3.1.1A, the
coefficients of the national growth are as large as around 0.90. This indicates that a one-percent
increase in the national growth would yield about a 0.9 percent increase in the employment
growth in cities throughout the test period (between 1986 and 2012). The employment growth in
city industries strongly follows the employment growth in the country as a whole.  
In table 1.3.1.1B, the coefficients of the national growth are about 0.77. A one-percent
increase in the national employment growth would thus bring about a 0.77 percent increase in the
employment growth in cities in the later period (between 1998 and 2012).  The growth of
employment in city industries tracks the growth of national employment, but their relationship is
much weaker than that of the period between 1986 and 2012.    
The same analysis for the earlier period, shown in table 1.3.1.1C, provides a clue to the
difference. Between 1986 and 1998, the coefficients of the national growth are about 0.97. A
15



one-percent increase in the national growth would thus yield about a 0.97 percent increase in the
employment growth in cities. This indicates that the national growth virtually determined the
growth of the city-industry employment during the period from 1986 to 1998.  
In this set of regressions to explore the effects of externalities on growth, the three
externality indicators showed consistent results, and all of the other dependent variables were
also significantly related with the employment growth in city-industries. However, the results
also revealed that what actually played a dominant role in determining the growth in cities was
the growth of the nation, more than a certain type of externalities, especially up to the 1990s.
This evidence requires further investigation.
Growth in Cities and National Growth
This study was intended to investigate which type of externality has contributed to the
growth in cities after the period Glaeser et al. studied. However, the updated data and the
reproduced results have challenged the goal of the study itself. The resulting view of this study is
that the growth of cities has been largely dependent on the growth of the nation.  
In a critical review of the economic growth literature, Polèse (2005) also suggests seeing
the big picture. Though there is much positive evidence of a link between cities and economic
growth, it is unclear whether cities are the cause or the product of economic growth. Polèse
argues that the socioeconomic processes that make cities productive operate primarily at the
national/societal level, and that cities are captors of national economic growth.
Glaeser et al. studied the growth of cities between 1956 and 1987. The U.S. enjoyed
postwar prosperity and significant economic growth by the early 1970s, then suffered from the
following recession and inflation. The economy of the U.S. has kept growing with fluctuations.
16



The period of 27 years from 1986 to 2012 is not an extremely long period, but the change in the
socio-economic environment during this period was enormous. Populations and cities have
significantly expanded, and current technology- and information-oriented industries are not
comparable to the manufacturing-oriented industries of the 1980s.  
As national economies grow, economic structure changes as industries grow through their
life cycle. In some stages, national growth leads to city growth. Through political and socio-
economical change over time, certain components of a nation, such as cities, play more
important roles in the national economy.  
The results in tables 1.3.1.1A to 1.3.1.2C suggest that national economic performance
affects city growth, but this phenomenon was stronger in the past. Most U.S. cities initially
emerged as industrial cities. Natural or transportational advantage was essential in the prior
growth of cities. Along with successful industrial development, mainly in manufacturing, the
growth of cities meant the growth of the national economy, and the fruits of national economic
growth were distributed relatively well among cities in the 1980s and 1990s. Today, however,
the growth of cities, especially big cities, is not closely related with the performance of the
national economy as it once was. The influence of national growth on cities has decreased along
with industrial structure changes. Other factors have become more important instead. Certain city
attributes, such as diversity and the presence of highly educated or creative individuals, are often
referred to as significant in growth. Location matters more now.  
The geographic concentrations of economic activities make cities efficient places that
facilitate knowledge spillovers. To benefit from dynamic externalities of agglomeration,
knowing the source of those externalities is critical, and many attempts have been made to
17



identify it. However, we need to focus upon the recent period, because the study results might be
more reliable and the effects of externalities might be clearer under the lower influence of
national growth.    
Employment Growth and Manufacturing Industry
The revealed trend in city-industries over three decades shows dramatic shifts in
industrial structure. The change of the largest industry from manufacturing to health care and
social assistance suggests the changing role of manufacturing industry. This section provides an
additional set of regression results to examine the role of manufacturing in employment growth
in U.S. cities.
Manufacturing is a traditional engine of city growth because it creates jobs and actually
produces various materials. By creating output that has economic value, manufacturing
contributes to growth.
Table 1.4.1 shows that manufacturing employment was stable from 1986 to 1998 and has
decreased considerably since then, while total city-industry employment has continued to grow.
Manufacturing employed 13.7 million workers in 1986, representing about 19.9 percent of total
employment in cities. The importance of manufacturing is also shown in the mean and median of
manufacturing employment compared to total city employment. In 1986, the mean and median
of manufacturing employment were 35,921 and 14,003, respectively, while those of all city-
industries were 9,056 and 1,942. Clearly, the size of manufacturing employment is much greater
than that of other industries in general. In 2012, manufacturing employed 8.8 million workers, or
8.9 percent of total city employment. Its mean and median were 23,224 and 10,845, while those
18



of all other city industries were 12,992 and 2,875. Thus, we can see that, even decreased,
manufacturing is still a large employer.
The significance of the manufacturing industry is also revealed in table 1.4.2.
Manufacturing was the biggest industry in as many as 253 (66 percent) and 188 (49 percent) out
of 383 total cities, in 1986 and 1998, respectively. Further, it was in top five industries in terms
of employment in 362 cities (94.5% of all cities) in 1986 and 346 cities (90% of all cities) in
1998. In 2012, manufacturing was still in the top five industries in 298 cities (77.8%) and the
biggest industry in 60 cities (15.6%) cities.      
Though in serious decline, given its size and importance in overall city economies,
manufacturing should have played a particularly important role in growth in cities compared to
other industries. Therefore, regressions have been run with a manufacturing dummy variable. All
the variables for employment growth regression were used in the same format. The added
variable is a dum m y variable indicating that the size of manufacturing in a city is above the
median of manufacturing employment of all cities in the initial year. Tables 1.3.2.1A to 1.3.2.2C
report the results.  
The addition of the manufacturing dummy barely affected the estimated coefficients of
other variables or R-square values. The estimated coefficients of the manufacturing dummy are
statistically significant and have a negative sign. These results suggest that large manufacturing
industry hurts employment growth in cities. Similarly, Glaeser et al. (1995) found that the initial
share of manufacturing employment had a negative relationship with population or income
growth in U.S. cities between 1960 and 1990.  
19



One important point to note from this study is the difference in the estimated coefficients
of the manufacturing dummy among the three test periods. The size of the manufacturing
coefficients in table 1.3.2.1A is about twice as great as those in tables 1.3.2.1B and 1.3.2.1C.  
Manufacturing affected city growth least in the earliest period, between 1986 and 1998.
Perhaps this is because manufacturing employment was relatively stable in this period, as
manufacturing was the major industry in most cities. However, the hindering effect on growth
appears well in the period between 1986 and 2012. During this period, the U.S. economy
experienced a rapid decline in the manufacturing industry. The manufacturing sector declined
continuously after its cyclical peak in 1998 and lost 4.6 million jobs between 1998 and 2012.
Heavily manufacturing-oriented cities must have suffered more severely. However, the results
for the growth between 1998 and 2012 show that this effect became weaker in recent years. The
most plausible reason for this would be the diminishing share of manufacturing industry. Table
1.5.1 and table 1.5.2 show the distribution of city industries in 1986, 1998, and 2012. Overall, the
U.S. city industry was extensively manufacturing-centered in 1986 but has become more
heterogeneous over years.      
The scale of manufacturing in city industry has affected growth in cities along with the
industrial structure that generates externalities.  
Employment Growth of Smaller Industries
The previous analyses were conducted to test the role of MAR, Porter, and Jacobs
externalities. These externalities are referred to as dynamic externalities, in that they explain
knowledge spillovers and technology dissemination that contribute to city growth. In contrast,
there are externalities that are static, rather than dynamic, such as urbanization externalities.
20



Urbanization externalities are advantages stemming from large market size and strong
knowledge infrastructure.  
Glaeser et al. ran another simple test to examine urbanization externalities focusing on
the growth of smaller industries. Tables 1.6.1A to 1.6.2C present the same test on the
employment growth of smaller industries using the updated dataset. The dependent variable is
small industry growth, which is measured as employment growth of industries other than the
four largest industries in a city. The independent variables are initial wage and initial
employment outside these four largest industries and employment growth in the four largest
industries.        
The estimation results are similar to those of Glaeser et al. Like the evidence found in the
employment growth analysis of all city industries, initial employment has negative effects on
employment growth of city industries other than the four largest industries. Initial wages have no
effect. The estimated coefficient is statistically significant but close to zero. The coefficient of
employment growth of the four biggest industries is positive and statistically significant. The
overall result suggests that the growth of large industries induces the growth of small industries
in cities. This indicates the existence of urbanization externalities due to demand spillover.  
Conclusion
This study reproduces the empirical study of Glaeser et al. (1992). The effect of three
types of knowledge externalities on city growth is investigated through employment growth in
city industries. An updated dataset is constructed to test whether their arguments, based on data
from 1956 and 1987, are still valid. By bridging historical data in different classification systems
through ratio-based conversion, the new dataset allows for empirical research over a longer
21



period, from 1986 to 2012, which has rarely been studied. In addition to updating the study
period, the present study significantly increased the number of observations and the accuracy of
the dataset.  
The study results confirmed Glaeser et al.’s arguments. The evidence suggests that
industry employment grows in competitive and diverse cities, rather than specialized cities. The
evidence is consistent over the three test periods between 1986 and 2012. As in Jacobs’ theory, it
appears that knowledge spillovers contribute to the economic growth of cities through the
exchange of ideas across different industry sectors.    
A few points should be noted in relation to these results. The results show that the
effectiveness of the competitive and diverse urban environment in economic growth extends
from the 1950s to the early 21th century. This might be due in part to the fact that this study
adopted the methodology of Glaeser et al. As mentioned earlier, the choice of dependent and
independent variables, the definition of externality measures, the level of industrial
agglomeration and many other methodological choices tend to affect the result of empirical
research. For instance, it might be suspected that the 2-digit NAICS industries are too general to
reveal the effect of specialization. Also, there may be measures that are more likely to detect the
effects of externalities.  
The results of this study further draw attention to the importance of nationwide economic
performance in the growth of cities. The evidence shows that national growth has played a
dominant role in determining the growth of cities. The influence of the national economic
performance is weaker in the most recent period than it was previously. This implies that the
study period matters. To investigate which type of knowledge externality contributes to
22



economic growth in cities, researchers should focus on the recent period, or their results may be
misleading. In the same manner, the historically dominant industry of manufacturing affects the
growth of city industries in the U.S. Therefore, the findings of this study on the work of
knowledge externalities do not necessarily extend to the growth of other countries.  
Although there is wide recognition of the importance of knowledge spillovers in
economic growth, their intangible nature and data limitations make empirical research difficult.
Therefore, most empirical work on this subject has focused on only part of the phenomenon.
Researchers have often investigated the mechanisms and the effects of knowledge spillovers by
looking into specific industries or regions over a relatively short period of time. This study does
not show how knowledge spillovers contribute to growth; however, as a nationwide cross-
sectional empirical study for all industries, the findings of this study can provide a useful
overview. On this basis, the next step will be to devise useful measures of knowledge
externalities and to search for implications beyond the observed phenomenon.

 
 
 
Table 1.1.1A. Largest and Smallest Cities in 1986
Six Largest Cities in 1986
  Employment  

 CBSA 1986 1998 2012
Six Largest Industries
1 New York-White Plains-Wayne, NY-NJ       4,054,208        4,360,151        4,704,254  
Manufacturing, Finance and insurance, Health care
and social assistance, Retail trade, Wholesale trade,
Administrative and support and waste management
and remediation services  
2 Los Angeles-Long Beach-Santa Ana, CA       3,286,367        3,671,562        3,663,306  
Manufacturing, Retail trade, Health care and social
assistance, Accommodation and food services,
Wholesale trade, Administrative and support and
waste management and remediation services  
3 Chicago-Joliet-Naperville, IL       2,694,339        3,418,756        3,334,229  
Manufacturing, Retail trade, Health care and social
assistance, Finance and insurance, Wholesale trade,
Accommodation and food services  
4 Minneapolis-St. Paul-Bloomington, MN-WI       1,586,599        2,290,400        2,421,747  
Manufacturing, Retail trade, Health care and social
assistance, Accommodation and food services,
Finance and insurance, Wholesale trade  
5 Philadelphia, PA       1,407,628        1,661,645        1,761,606  
Manufacturing, Health care and social assistance,
Retail trade, Finance and insurance,
Accommodation and food services, Wholesale trade  
6 Houston-Sugar Land-Baytown, TX       1,291,201        1,873,037        2,335,772  
Retail trade, Manufacturing, Construction,
Accommodation and food services, Health care and
social assistance, Wholesale trade  
   
 



24



Table 1.1.1A. Largest and Smallest Cities in 1986 (Cont.)
Six Smallest Cities in 1986
  Employment  

 CBSA 1986 1998 2012
Six Largest Industries
1 Palm Coast, FL             3,972              9,386            15,919  
Manufacturing, Retail trade, Construction, Real
estate and rental and leasing, Accommodation and
food services, Health care and social assistance  
2 Hinesville-Fort Stewart, GA             4,807              8,729            12,719  
Retail trade, Accommodation and food services,
Manufacturing, Finance and insurance, Health care
and social assistance, Information  
3 St. George, UT             8,765            23,290            37,780  
Retail trade, Accommodation and food services,
Health care and social assistance, Manufacturing,
Construction, Transportation and warehousing  
4 Madera-Chowchilla, CA           12,502            21,958            25,946  
Manufacturing, Retail trade, Health care and social
assistance, Accommodation and food services,
Construction, Wholesale trade  
5 Carson City, NV           12,850            21,703            21,570  
Manufacturing, Retail trade, Accommodation and
food services, Health care and social assistance,
Construction, Arts, entertainment, and recreation  
6 Hanford-Corcoran, CA           13,430            17,350            23,313  
Manufacturing, Retail trade, Accommodation and
food services, Health care and social assistance,
Wholesale trade, Construction  
         
 

 
25



Table 1.1.1B. Largest and Smallest Cities in 1998
Six Largest Cities in 1998
  Employment  
 CBSA 1986 1998 2012 Six Largest Industries
1 New York-White Plains-Wayne, NY-NJ       4,054,208        4,360,151        4,704,254  
Health care and social assistance, Finance and
insurance, Retail trade, Professional, scientific, and
technical services, Manufacturing, Administrative
and support and waste management and remediation
services  
2 Los Angeles-Long Beach-Santa Ana, CA       3,286,367        3,671,562        3,663,306  
Manufacturing, Professional, scientific, and
technical services, Health care and social assistance,
Retail trade, Administrative and support and waste
management and remediation services,
Accommodation and food services  
3 Chicago-Joliet-Naperville, IL       2,694,339        3,418,756        3,334,229  
Manufacturing, Retail trade, Health care and social
assistance, Administrative and support and waste
management and remediation services, Professional,
scientific, and technical services, Accommodation
and food services  
4 Minneapolis-St. Paul-Bloomington, MN-WI       1,586,599        2,290,400        2,421,747  
Manufacturing, Health care and social assistance,
Retail trade, Administrative and support and waste
management and remediation services,
Accommodation and food services, Finance and
insurance  
5 Atlanta-Sandy Springs-Marietta, GA       1,171,239        1,886,121        2,063,001  
Retail trade, Manufacturing, Administrative and
support and waste management and remediation
services, Health care and social assistance,
Accommodation and food services, Wholesale trade  
6 Houston-Sugar Land-Baytown, TX       1,291,201        1,873,037        2,335,772  
Retail trade, Manufacturing, Administrative and
support and waste management and remediation
services, Health care and social assistance,
Accommodation and food services, Construction  
     

26



Table 1.1.1B. Largest and Smallest Cities in 1998 (Cont.)
Six Smallest Cities in 1998
  Employment  

 CBSA 1986 1998 2012
Six Largest Industries
1 Hinesville-Fort Stewart, GA             4,807              8,729            12,719  
Retail trade, Health care and social assistance,
Accommodation and food services, Manufacturing,
Other services (except public administration),
Construction  
2 Palm Coast, FL             3,972              9,386            15,919  
Retail trade, Manufacturing, Accommodation and
food services, Health care and social assistance,
Construction  
3 Hanford-Corcoran, CA           13,430            17,350            23,313  
Retail trade, Health care and social assistance,
Manufacturing, Accommodation and food services,
Construction, Administrative and support and waste
management and remediation services  
4 Lewiston, ID-WA           14,888            19,983            21,101  
Retail trade, Manufacturing, Health care and social
assistance, Accommodation and food services,
Construction, Finance and insurance  
5 Fairbanks, AK           16,753            21,022            26,847  
Retail trade, Health care and social assistance,
Accommodation and food services, Transportation
and warehousing, Construction, Other services
(except public administration)  
6 Carson City, NV           12,850            21,703            21,570  
Manufacturing, Retail trade, Health care and social
assistance, Accommodation and food services,
Construction, Administrative and support and waste
management and remediation services  
         
 
 
27



Table 1.1.1C. Largest and Smallest Cities in 2012
Six Largest Cities in 2012
  Employment  
 CBSA 1986 1998 2012 Six Largest Industries
1 New York-White Plains-Wayne, NY-NJ       4,054,208        4,360,151        4,704,254  
Health care and social assistance, Retail trade,
Finance and insurance, Professional, scientific, and
technical services, Accommodation and food
services, Administrative and support and waste
management and remediation services  
2 Los Angeles-Long Beach-Santa Ana, CA       3,286,367        3,671,562        3,663,306  
Health care and social assistance, Retail trade,
Manufacturing, Accommodation and food services,
Professional, scientific, and technical services,
Administrative and support and waste management
and remediation services  
3 Chicago-Joliet-Naperville, IL       2,694,339        3,418,756        3,334,229  
Health care and social assistance, Retail trade,
Administrative and support and waste management
and remediation services, Manufacturing,
Accommodation and food services, Professional,
scientific, and technical services
4 Minneapolis-St. Paul-Bloomington, MN-WI       1,586,599        2,290,400        2,421,747  
Health care and social assistance, Manufacturing,
Retail trade, Accommodation and food services,
Finance and insurance, Administrative and support
and waste management and remediation services  
5 Houston-Sugar Land-Baytown, TX       1,291,201        1,873,037        2,335,772  
Health care and social assistance, Retail trade,
Accommodation and food services, Manufacturing,
Professional, scientific, and technical services,
Administrative and support and waste management
and remediation services  
6 Atlanta-Sandy Springs-Marietta, GA       1,171,239        1,886,121        2,063,001  
Retail trade, Health care and social assistance,
Accommodation and food services, Administrative
and support and waste management and remediation
services, Professional, scientific, and technical
services, Wholesale trade  
   

28



Table 1.1.1C. Largest and Smallest Cities in 2012 (Cont.)
Six Smallest Cities in 2012
  Employment  

 CBSA 1986 1998 2012
Six Largest Industries
1 Hinesville-Fort Stewart, GA             4,807              8,729            12,719  
Health care and social assistance, Retail trade,
Manufacturing, Accommodation and food services,
Professional, scientific, and technical services,
Transportation and warehousing  
2 Palm Coast, FL             3,972              9,386            15,919  
Accommodation and food services, Retail trade,
Health care and social assistance, Administrative
and support and waste management and remediation
services, Other services (except public
administration), Manufacturing  
3 Lewiston, ID-WA           14,888            19,983            21,101  
Health care and social assistance, Retail trade,
Manufacturing, Accommodation and food services,
Finance and insurance, Construction  
4 Carson City, NV           12,850            21,703            21,570  
Health care and social assistance, Retail trade,
Manufacturing, Accommodation and food services,
Administrative and support and waste management
and remediation services, Arts, entertainment, and
recreation  
5 Hanford-Corcoran, CA           13,430            17,350            23,313  
Health care and social assistance, Retail trade,
Accommodation and food services, Manufacturing,
Transportation and warehousing, Wholesale trade  
6 Danville, IL           27,372            28,650            24,119  
Manufacturing, Arts, entertainment, and recreation,
Retail trade, Accommodation and food services,
Wholesale trade, Finance and insurance  
           
                                                                                                   29                                                                                                                                                          

Table 1.1.2A. 20 Largest City-Industries in 1986
20 Largest City-Industries in 1986
 
CBSA Industry Employment  
1 Los Angeles-Long Beach-Santa Ana, CA
Manufacturing
         
814,641  

2 New York-White Plains-Wayne, NY-NJ
Manufacturing
         
553,546  

3 Chicago-Joliet-Naperville, IL
Manufacturing
         
549,744  

4 New York-White Plains-Wayne, NY-NJ
Finance and insurance
         
477,502  

5 New York-White Plains-Wayne, NY-NJ
Health care and social assistance
         
438,252  

6 New York-White Plains-Wayne, NY-NJ
Retail trade
         
406,971  

7 Los Angeles-Long Beach-Santa Ana, CA
Retail trade
         
364,867  

8 Chicago-Joliet-Naperville, IL
Retail trade
         
346,065  

9 Minneapolis-St. Paul-Bloomington, MN-WI
Manufacturing
         
339,865  

10 New York-White Plains-Wayne, NY-NJ
Wholesale trade
         
311,958    
11 New York-White Plains-Wayne, NY-NJ
Administrative and support and waste
management and remediation services
         
273,838  

12 Los Angeles-Long Beach-Santa Ana, CA
Health care and social assistance
         
269,577  

13 Philadelphia, PA
Manufacturing
         
257,510  

14 Chicago-Joliet-Naperville, IL
Health care and social assistance
         
244,376  

15 San Jose-Sunnyvale-Santa Clara, CA
Manufacturing
         
231,698  

16 Los Angeles-Long Beach-Santa Ana, CA
Accommodation and food services
         
230,933  

17 Santa Ana-Anaheim-Irvine, CA
Manufacturing
         
224,638  

18 New York-White Plains-Wayne, NY-NJ
Professional, scientific, and technical
services
         
218,712  

19 New York-White Plains-Wayne, NY-NJ
Information
         
213,134  

20 New York-White Plains-Wayne, NY-NJ
Accommodation and food services
         
212,510    
   

 
30



Table 1.1.2B. 20 Largest City-Industries in 1998
20 Largest City-Industries in 1998
 
CBSA Industry Employment  
1 New York-White Plains-Wayne, NY-NJ
Health care and social assistance
         
680,840  

2 Los Angeles-Long Beach-Santa Ana, CA
Manufacturing
         
638,389  

3 Chicago-Joliet-Naperville, IL
Manufacturing
         
529,678  

4 New York-White Plains-Wayne, NY-NJ
Finance and insurance
         
450,066  

5 Minneapolis-St. Paul-Bloomington, MN-WI
Manufacturing
         
394,438  

6 New York-White Plains-Wayne, NY-NJ
Retail trade
         
389,937  

7 Los Angeles-Long Beach-Santa Ana, CA
Professional, scientific, and technical
services
         
383,371  

8 Chicago-Joliet-Naperville, IL
Retail trade
         
379,450  

9 Chicago-Joliet-Naperville, IL
Health care and social assistance
         
369,041  

10 Los Angeles-Long Beach-Santa Ana, CA
Health care and social assistance
         
360,703    
11 New York-White Plains-Wayne, NY-NJ
Professional, scientific, and technical
services
         
358,272  

12 Los Angeles-Long Beach-Santa Ana, CA
Retail trade
         
349,666  

13 New York-White Plains-Wayne, NY-NJ
Manufacturing
         
331,578  

14 New York-White Plains-Wayne, NY-NJ
Administrative and support and waste
management and remediation services
         
315,277  

15 Los Angeles-Long Beach-Santa Ana, CA
Administrative and support and waste
management and remediation services
         
311,818  

16 New York-White Plains-Wayne, NY-NJ
Wholesale trade
         
308,920  

17 Chicago-Joliet-Naperville, IL
Administrative and support and waste
management and remediation services
         
292,380  

18 Minneapolis-St. Paul-Bloomington, MN-WI
Health care and social assistance
         
269,953  

19 Minneapolis-St. Paul-Bloomington, MN-WI
Retail trade
         
265,739  

20 Philadelphia, PA
Health care and social assistance
         
265,624    
   

 
31



Table 1.1.2C. 20 Largest City-Industries in 2012
20 Largest City-Industries in 2012
 
CBSA Industry Employment  
1 New York-White Plains-Wayne, NY-NJ
Health care and social assistance
         
894,233  

2 Los Angeles-Long Beach-Santa Ana, CA
Health care and social assistance
         
488,872  

3 Chicago-Joliet-Naperville, IL
Health care and social assistance
         
486,806  

4 New York-White Plains-Wayne, NY-NJ
Retail trade
         
483,643  

5 Washington-Arlington-Alexandria DC-VA
Professional, scientific, and technical
services
         
424,401  

6 New York-White Plains-Wayne, NY-NJ
Finance and insurance
         
408,256  

7 New York-White Plains-Wayne, NY-NJ
Professional, scientific, and technical
services
         
404,283  

8 New York-White Plains-Wayne, NY-NJ
Accommodation and food services
         
393,266  

9 Los Angeles-Long Beach-Santa Ana, CA
Retail trade
         
386,821  

10 Minneapolis-St. Paul-Bloomington, MN-WI
Health care and social assistance
         
379,610    
11 Chicago-Joliet-Naperville, IL
Retail trade
         
360,431  

12 Los Angeles-Long Beach-Santa Ana, CA
Manufacturing
         
354,273  

13 Los Angeles-Long Beach-Santa Ana, CA
Accommodation and food services
         
353,954  

14 Philadelphia, PA
Health care and social assistance
         
335,711  

15 Los Angeles-Long Beach-Santa Ana, CA
Professional, scientific, and technical
services
         
331,182  

16 Los Angeles-Long Beach-Santa Ana, CA
Administrative and support and waste
management and remediation services
         
323,414  

17 Chicago-Joliet-Naperville, IL
Administrative and support and waste
management and remediation services
         
315,205  

18 Chicago-Joliet-Naperville, IL
Manufacturing
         
303,169  

19 Chicago-Joliet-Naperville, IL
Accommodation and food services
         
299,814  

20 Houston-Sugar Land-Baytown, TX
Health care and social assistance
         
292,654    
   


 
32



Table 1.2.1A. Variable Means and Standard Deviations, 1986-2012 (All 20 Industries)
 Variable Mean
Standard
Deviation
Number of
Observations
 
     

Employment growth, 1986-2012
   
 
Log(employment in 2012/employment in 1986)
in the City-Industry
0.1993 1.0974                   6,719

     

National employment growth, 1986-2012
   
 
Log(U.S. employment in 2012/
U.S. employment in 1986)
in the industry outside the city
0.1814 1.0090 20


Wage in the city-industry  
in 1986 (in thousands)
95.6953 158.2463                   6,719


Employment in the city-industry
in 1986 (in millions)
               0.0098                  0.0278                    6,719


Specialization indicator
   
 
City-industry's share of city employment
relative to U.S. industry's share of U.S.
employment in 1986
1.0051 1.1037                   6,719


Competition indicator
   
 
Establishments per employee in the city-
industry relative to establishments per
employee in the U.S. industry in 1986
1.3399 0.7064
                                   
6,719  

Diversification indicator
   
 
City's other top five industries' share of 1986
total city-industry
0.6232 0.0880                   6,719

     

Wage growth, 1986-2012
   
 
Log(wage in 2012/wage in 1986) in the City-
Industry
0.2090 1.0273                   6,719


National wage growth, 1986-2012
   
 
Log(U.S. wage in 2012/
U.S. wage in 1986)
in the industry outside the city
0.1877 0.7837 20


Employment growth of smaller industries,  
1986-2012    
 
Log(employment in 2012/employment in 1986)
in the city outside the four biggest industries
0.4172 0.3372 383


Employment outside the four biggest industries in
1986 (in millions)
               0.0888                  0.1939  383


Wage outside the four biggest industries  
in 1986 (in thousands)
1030.5576 1022.0850 383


Employment growth in the four biggest industries,
1986-2012
0.3627 0.3558 383

             
 
33



Table 1.2.1B. Variable Means and Standard Deviations, 1998-2012 (All 20 Industries)
 Variable Mean
Standard
Deviation
Number of
Observations
 
     

Employment growth, 1998-2012
   
 
Log(employment in 2012/employment in 1998)
in the City-Industry
0.0281 0.5441                   6,719

     

National employment growth, 1998-2012
   
 
Log(U.S. employment in 2012/
U.S. employment in 1998)
in the industry outside the city
0.0133 0.4209 20


Wage in the city-industry  
in 1998 (in thousands)
62.5442 75.0528                   6,719


Employment in the city-industry
in 1998 (in millions)
0.0132                  0.0327                    6,719


Specialization indicator
   
 
City-industry's share of city employment
relative to U.S. industry's share of U.S.
employment in 1998
1.0659 1.8912                   6,719


Competition indicator
   
 
Establishments per employee in the city-
industry relative to establishments per
employee in the U.S. industry in 1998
1.4120 0.8744                   6,719


Diversification indicator
   
 
City's other top five industries' share of 1998
total city-industry
0.6154 0.1063                   6,719

     

Wage growth, 1998-2012
   
 
Log(wage in 2012/wage in 1998) in the City-
Industry
0.3780 0.3595                   6,719


National wage growth, 1998-2012
   
 
Log(U.S. wage in 2012/
U.S. wage in 1998)
in the industry outside the city
0.3879 0.1764 20


Employment growth of smaller industries,  
1998-2012    
 
Log(employment in 2012/employment in 1998)
in the city outside the four biggest industries
0.0758 0.1485 383


Employment outside the four biggest industries in
1998 (in millions)
               0.1256                  0.2543 383


Wage outside the four biggest industries  
in 1998 (in thousands)
944.7414 997.6255 383


Employment growth in the four biggest industries,
1998-2012
0.0595 0.1695 383

             
34



Table 1.2.1C. Variable Means and Standard Deviations, 1986-1998 (All 20 Industries)
 Variable Mean
Standard
Deviation
Number of
Observations
 
     

Employment growth, 1986-1998
   
 
Log(employment in 1998/employment in 1986)
in the City-Industry
0.1642 0.8063                   6,719  

     

National employment growth, 1986-1998
   
 
Log(U.S. employment in 1998/
U.S. employment in 1986)
in the industry outside the city
0.1680 0.6400 20


Wage in the city-industry  
in 1986 (in thousands)
95.6953 158.2463 6,719


Employment in the city-industry
in 1986 (in millions)
               0.0098                  0.0278                   6,719


Specialization indicator
   
 
City-industry's share of city employment
relative to U.S. industry's share of U.S.
employment in 1986
1.0051 1.1037 6,719


Competition indicator
   
 
Establishments per employee in the city-
industry relative to establishments per
employee in the U.S. industry in 1986
1.3399 0.7064 6,719


Diversification indicator
   
 
City's other top five industries' share of 1986
total city-industry
0.6232 0.0880                   6,719

     

Wage growth, 1986-1998
   
 
Log(wage in 1998/wage in 1986) in the City-
Industry
-0.1316 1.0750 6,719


National wage growth, 1986-1998
   
 
Log(U.S. wage in 1998/
U.S. wage in 1986)
in the industry outside the city
-0.2002 0.7531 20


Employment growth of smaller industries,  
1986-1998    
 
Log(employment in 1998/employment in 1986)
in the city outside the four biggest industries
0.3414 0.2575 383


Employment outside the four biggest industries in
1986 (in millions)
               0.0888                  0.1939  383


Wage outside the four biggest industries  
in 1986 (in thousands)
1030.5576 1022.0850 383


Employment growth in the four biggest industries,
1986-1998
0.3032 0.2755 383

             
 
35



Table 1.2.2A. Variable Means and Standard Deviations, 1986-2012 (Top 6 Industries)
 Variable Mean
Standard
Deviation
Number of
Observations
 
     

Employment growth, 1986-2012
   
 
Log(employment in 2012/employment in 1986)
in the City-Industry
0.3327 0.5287                   2,290

     

National employment growth, 1986-2012
   
 
Log(U.S. employment in 2012/
U.S. employment in 1986)
in the industry outside the city
0.2990 0.4052 6


Wage in the city-industry  
in 1986 (in thousands)
164.1937 230.4780 2,290


Employment in the city-industry
in 1986 (in millions)
               0.0199                  0.0432 2,290


Specialization indicator
   
 
City-industry's share of city employment
relative to U.S. industry's share of U.S.
employment in 1986
1.2683 1.6411 2,290


Competition indicator
   
 
Establishments per employee in the city-
industry relative to establishments per
employee in the U.S. industry in 1986
1.0895 0.3460
                                   
2,290  

Diversification indicator
   
 
City's other top five industries' share of 1986
total city-industry
0.5859 0.0878 2,290

     

Wage growth, 1986-2012
   
 
Log(wage in 2012/wage in 1986) in the City-
Industry
-0.4369 0.9000 2,290


National wage growth, 1986-2012
   
 
Log(U.S. wage in 2012/
U.S. wage in 1986)
in the industry outside the city
-0.4374 0.8116 6


Employment growth of smaller industries,  
1986-2012    
 
Log(employment in 2012/employment in 1986)
in the city outside the four biggest industries
0.4136 0.3559 383


Employment outside the four biggest industries in
1986 (in millions)
               0.0850                  0.1889  383


Wage outside the four biggest industries  
in 1986 (in thousands)
1001.3469 992.0132 383


Employment growth in the four biggest industries,
1986-2012
0.3631 0.3576 383

             
 
36



Table 1.2.2B. Variable Means and Standard Deviations, 1998-2012 (Top 6 Industries)
 Variable Mean
Standard
Deviation
Number of
Observations
 
     

Employment growth, 1998-2012
   
 
Log(employment in 2012/employment in 1998)
in the City-Industry
0.0232 0.3101                   2,290

     

National employment growth, 1998-2012
   
 
Log(U.S. employment in 2012/
U.S. employment in 1998)
in the industry outside the city
0.0302 0.2394 6


Wage in the city-industry  
in 1998 (in thousands)
64.7561 80.4298 2,290


Employment in the city-industry
in 1998 (in millions)
0.0252                  0.0473                    2,290


Specialization indicator
   
 
City-industry's share of city employment
relative to U.S. industry's share of U.S.
employment in 1998
1.2922 2.0322                   2,290


Competition indicator
   
 
Establishments per employee in the city-
industry relative to establishments per
employee in the U.S. industry in 1998
1.0681 0.3483                   2,290


Diversification indicator
   
 
City's other top five industries' share of 1998
total city-industry
0.5866 0.0902                   2,290

     

Wage growth, 1998-2012
   
 
Log(wage in 2012/wage in 1998) in the City-
Industry
0.3998 0.1791                   2,290


National wage growth, 1998-2012
   
 
Log(U.S. wage in 2012/
U.S. wage in 1998)
in the industry outside the city
0.4035 0.0430 6


Employment growth of smaller industries,  
1998-2012    
 
Log(employment in 2012/employment in 1998)
in the city outside the four biggest industries
0.0749 0.1516 383


Employment outside the four biggest industries in
1998 (in millions)
               0.1201                 0.2476 383


Wage outside the four biggest industries  
in 1998 (in thousands)
915.1049 967.9145 383


Employment growth in the four biggest industries,
1998-2012
0.0588 0.1714 383

             
37



Table 1.2.2C. Variable Means and Standard Deviations, 1986-1998 (Top 6 Industries)
 Variable Mean
Standard
Deviation
Number of
Observations
 
     

Employment growth, 1986-1998
   
 
Log(employment in 1998/employment in 1986)
in the City-Industry
0.3094 0.3332                   2,290

     

National employment growth, 1986-1998
   
 
Log(U.S. employment in 1998/
U.S. employment in 1986)
in the industry outside the city
0.2688 0.1900 6


Wage in the city-industry  
in 1986 (in thousands)
164.1937 230.4780 2,290


Employment in the city-industry
in 1986 (in millions)
               0.0199                 0.0432                   2,290


Specialization indicator
   
 
City-industry's share of city employment
relative to U.S. industry's share of U.S.
employment in 1986
1.2683 1.6411 2,290


Competition indicator
   
 
Establishments per employee in the city-
industry relative to establishments per
employee in the U.S. industry in 1986
1.0895 0.3460 2,290


Diversification indicator
   
 
City's other top five industries' share of 1986
total city-industry
0.5859 0.0878                   2,290

     

Wage growth, 1986-1998
   
 
Log(wage in 1998/wage in 1986) in the City-
Industry
-0.8377 0.8769 2,290


National wage growth, 1986-1998
   
 
Log(U.S. wage in 1998/
U.S. wage in 1986)
in the industry outside the city
-0.8409 0.7848 6


Employment growth of smaller industries,  
1986-1998    
 
Log(employment in 1998/employment in 1986)
in the city outside the four biggest industries
0.3387 0.2723 383


Employment outside the four biggest industries in
1986 (in millions)
               0.0850                  0.1889 383


Wage outside the four biggest industries  
in 1986 (in thousands)
1001.3469 992.0132 383


Employment growth in the four biggest industries,
1986-1998
0.3043 0.2747 383

             
 
38



Table 1.3.1.1A. City-Industry Employment Growth between 1986 and 2012 (All 20 Industries)
     
Log(Employment in 2012/Employment in 1986)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0427
***
-0.2776
***
-0.0212

-0.1163
*



         

Log(U.S. employment in 2012/
U.S. employment in 1986)
in the industry outside the city

0.9128 *** 0.8995 *** 0.9154 *** 0.8999 ***
         

Wage in the city-industry  
in 1986 (in thousands)

0.0001 ** 0.0002 *** 0.0002 *** 0.0002 ***
         

Employment in the city-
industry
in 1986 (in millions)

-1.0830 *** -0.5154

-1.1739 *** -0.6272 *
         

Dummy variable indicating
presence in the South  
0.0435 *** 0.0528 *** 0.0408 *** 0.0523 ***

         

Specialization indicator

-0.0303 ***
   
-0.0048
 
       

Competition indicator
 
0.2056 ***
 
0.2077 ***

       

Diversification indicator
   
-0.0128

-0.2488 ***

       


         

R Square

0.7086

0.7244

0.7077

0.7247
 

Adjusted R Square

0.7084

0.7242

0.7075

0.7245
 

Standard Error

0.5926

0.5763

0.5935

0.5761
 

Observation

6,719

6,719

6,719

6,719
                         
Note: *** p<.01 **p<.05 *p<.10
 
39



Table 1.3.1.1B. City-Industry Employment Growth between 1998 and 2012 (All 20 Industries)
     
Log(Employment in 2012/Employment in 1998)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0306 *** -0.2166 ***
-0.0264

-0.1352 ***

           

Log(U.S. employment in 2012/
U.S. employment in 1998)
in the industry outside the city

0.7894 *** 0.7679 *** 0.7893 *** 0.7690 ***
         

Wage in the city-industry  
in 1998 (in thousands)

0.0000

0.0002 ** 0.0000

0.0001 **
         

Employment in the city-industry
in 1998 (in millions)

-0.1551

0.4075 ** -0.1377

0.3310 *
       


Dummy variable indicating
presence in the South

0.0224 ** 0.0234 ** 0.0215 ** 0.0244
**

       



Specialization indicator

-0.0194 ***
   
-0.0077
***

       


Competition indicator
 
0.1479 ***


0.1465 ***

       

Diversification indicator
   
0.0578

-0.1125
**

       
           

R Square

0.3780

0.4272

0.3736

0.4283
 

Adjusted R Square

0.3776

0.4268

0.3531

0.4277
 

Standard Error

0.4293

0.4119

0.4308

0.4116
 

Observation

6,719

6,719

6,719

6,719
                         
Note: *** p<.01 **p<.05 *p<.10
 
40



Table 1.3.1.1C. City-Industry Employment Growth between 1986 and 1998 (All 20 Industries)
     
Log(Employment in 1998/Employment in 1986)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0088

-0.2452
***
0.0296

-0.0703




         

Log(U.S. employment in 1998/
U.S. employment in 1986)
in the industry outside the city

0.9821 *** 0.9713 *** 0.9872 *** 0.9713 ***
         

Wage in the city-industry  
in 1986 (in thousands)

0.0002 *** 0.0002 *** 0.0002 *** 0.0002 ***
         

Employment in the city-
industry
in 1986 (in millions)

-0.9081 *** -0.4717 * -1.0187 *** -0.5886 **
         

Dummy variable indicating
presence in the South  
0.0182

0.0253 ** 0.0157

0.0249 **

         

Specialization indicator

-0.0256 ***
   
-0.0064
 
       

Competition indicator
 
0.1617 ***
 
0.1636 ***

       

Diversification indicator
   
-0.0723

-0.2668 ***

       


         

R Square

0.6151

0.6332

0.6140

0.6339
 

Adjusted R Square

0.6149

0.6329

0.6137

0.6335
 

Standard Error

0.5004

0.4886

0.5012

0.4881
 

Observation

6,719

6,719

6,719

6,719
                         
Note: *** p<.01 **p<.05 *p<.10

41



Table 1.3.1.2A. City-Industry Employment Growth between 1986 and 2012 (Top 6 Industries)
     
Log(Employment in 2012/Employment in 1986)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0578
***
-0.3257
***
0.3084 *** 0.1645
***



         

Log(U.S. employment in 2012/
U.S. employment in 1986)
in the industry outside the city

0.9423 *** 0.9465 *** 0.9835 *** 1.0071 ***
         

Wage in the city-industry  
in 1986 (in thousands)

0.0001 * 0.0001 ** 0.0001

0.0000

         

Employment in the city-
industry
in 1986 (in millions)

-1.0718 *** -0.6668 *** -1.3028 *** -0.9763 ***
         

Dummy variable indicating
presence in the South  
0.0607 *** 0.0766 *** 0.0573 *** 0.0778 ***

         

Specialization indicator

-0.0180 ***
   
-0.0117 ***

       

Competition indicator
 
0.3149 ***
 
0.3514 ***

       

Diversification indicator
   
-0.4753 *** -0.8855 ***

       


         

R Square

0.5360

0.5740

0.5376

0.5894
 

Adjusted R Square

0.5350

0.5731

0.5365

0.5882
 

Standard Error

0.3605

0.3454

0.3599

0.3393
 

Observation

2,290

2,290

2,290

2,290
                         
Note: *** p<.01 **p<.05 *p<.10
 
42



Table 1.3.1.2B. City-Industry Employment Growth between 1998 and 2012 (Top 6 Industries)
     
Log(Employment in 2012/Employment in 1998)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

-0.0112

-0.1611 ***
0.0924
*** 0.0000
 
           

Log(U.S. employment in 2012/
U.S. employment in 1998)
in the industry outside the city

0.8977 *** 0.9048 *** 0.9048 *** 0.9186 ***
         

Wage in the city-industry  
in 1998 (in thousands)

0.0001

0.0002 ** 0.0001

0.0002 **
         

Employment in the city-industry
in 1998 (in millions)

-0.1188

-0.0353

-0.2186 * -0.1986 *
       


Dummy variable indicating
presence in the South

0.0131

0.0212 ** 0.0143

0.0245
**

       



Specialization indicator

-0.0012
   
0.0011


       


Competition indicator
 
0.1293 ***


0.1464 ***

       

Diversification indicator
   
-0.1743 *** -0.3024
***

       
           

R Square

0.4745

0.4947

0.4767

0.5011
 

Adjusted R Square

0.4734

0.4935

0.4755

0.4995
 

Standard Error

0.2251

0.2207

0.2246

0.2194
 

Observation

2,290

2,290

2,290

2,290
                         
Note: *** p<.01 **p<.05 *p<.10
43



Table 1.3.1.2C. City-Industry Employment Growth between 1986 and 1998 (Top 6 Industries)
     
Log(Employment in 1998/Employment in 1986)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0609
***
-0.2277
***
0.1187 *** 0.0168




         

Log(U.S. employment in 1998/
U.S. employment in 1986)
in the industry outside the city

0.9607 *** 0.9943 *** 1.0228 *** 1.0329 ***
         

Wage in the city-industry  
in 1986 (in thousands)

0.0001 * 0.0001 ** 0.0001 ** 0.0000

         

Employment in the city-
industry
in 1986 (in millions)

-0.8603 *** -0.5943 *** -0.9777 *** -0.7361 ***
         

Dummy variable indicating
presence in the South  
0.0464 *** 0.0574 *** 0.0441 *** 0.0582 ***

         

Specialization indicator

-0.0164 ***
   
-0.0094 ***

       

Competition indicator
 
0.2256 ***
 
0.2400 ***

       

Diversification indicator
   
-0.1613 ** -0.4272 ***

       


         

R Square

0.3361

0.3831

0.3314

0.3936
 

Adjusted R Square

0.3346

0.3817

0.3299

0.3917
 

Standard Error

0.2718

0.2620

0.2728

0.2599
 

Observation

2,290

2,290

2,290

2,290
                         
Note: *** p<.01 **p<.05 *p<.10

 
44



Table 1.3.2.1A. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1986 and 2012 (All 20 Industries)
     
Log(Employment in 2012/Employment in 1986)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.1115
***
-0.2259
***
0.0573

-0.0765




         

Log(U.S. employment in 2012/
U.S. employment in 1986)
in the industry outside the city

0.9101 *** 0.8992 *** 0.9130 *** 0.8993 ***
         

Wage in the city-industry  
in 1986 (in thousands)

0.0002 *** 0.0003 *** 0.0002 *** 0.0002 ***
         

Employment in the city-
industry
in 1986 (in millions)

-0.6359 * -0.3204

-0.7341 ** -0.4135

         

Dummy variable indicating
presence in the South

0.0434 *** 0.0518 *** 0.0404 *** 0.0517 ***

         
Manufacturing industry
dummy
-0.1465 *** -0.0787 *** -0.1411 *** -0.0781 ***  
         

Specialization indicator

-0.0348 ***
   
-0.0087
 
       

Competition indicator
 
0.1935 ***
 
0.1939 ***

       

Diversification indicator
   
0.0275

-0.2207 ***

       


         

R Square

0.7127

0.7255

0.7115

0.7258
 

Adjusted R Square

0.7124

0.7253

0.7112

0.7255
 

Standard Error

0.5885

0.5752

0.5859

0.5750
 

Observation

6,719

6,719

6,719

6,719
                         
Note: *** p<.01 **p<.05 *p<.10
 
45



Table 1.3.2.1B. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1998 and 2012 (All 20 Industries)
     
Log(Employment in 2012/Employment in 1998)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0609 *** -0.2076 ***
0.0096

-0.1213 ***

           

Log(U.S. employment in 2012/
U.S. employment in 1998)
in the industry outside the city

0.7869 *** 0.7676 *** 0.7873 *** 0.7687 ***
         

Wage in the city-industry  
in 1998 (in thousands)

0.0001 * 0.0002 *** 0.0001 * 0.0002 **
         

Employment in the city-industry
in 1998 (in millions)

0.0371

0.4433 ** 0.0254

0.3743 **
       


Dummy variable indicating
presence in the South

0.0171

0.0222 ** 0.0167

0.0229
**

       


Manufacturing industry  
dummy
-0.0736 *** -0.0164  -0.0671 *** -0.0207 *  
       



Specialization indicator

-0.0207 ***
   
-0.0082
***

       


Competition indicator
 
0.1462 ***


0.1442 ***

       

Diversification indicator
   
0.0427

-0.1152
**

       
           

R Square

0.3819

0.4274

0.3768

0.4286
 

Adjusted R Square

0.3814

0.4269

0.3763

0.4280
 

Standard Error

0.4280

0.4119

0.4297

0.4115
 

Observation

6,719

6,719

6,719

6,719
                         
Note: *** p<.01 **p<.05 *p<.10
 
46



Table 1.3.2.1C. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1986 and 1998 (All 20 Industries)
     
Log(Employment in 1998/Employment in 1986)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0457
***
-0.2304
***
0.0484

-0.0596




         

Log(U.S. employment in 1998/
U.S. employment in 1986)
in the industry outside the city

0.9789 *** 0.9709 *** 0.9846 *** 0.9706 ***
         

Wage in the city-industry  
in 1986 (in thousands)

0.0002 *** 0.0003 *** 0.0002 *** 0.0002 ***
         

Employment in the city-
industry
in 1986 (in millions)

-0.6690 ** -0.4158

-0.7911 *** -0.5309 *
         

Dummy variable indicating
presence in the South

0.0181

0.0250 ** 0.0155

0.0248 **

         
Manufacturing industry  
dummy
-0.0780 *** -0.0224 * -0.0728 *** -0.0209 *  
         

Specialization indicator

-0.0281 ***
   
-0.0075
 
       

Competition indicator
 
0.1583 ***
 
0.1599 ***

       

Diversification indicator
   
-0.0514

-0.2593 ***

       


         

R Square

0.6173

0.6333

0.6158

0.6340
 

Adjusted R Square

0.6169

0.6330

0.6155

0.6336
 

Standard Error

0.4991

0.4885

0.5000

0.4881
 

Observation

6,719

6,719

6,719

6,719
                         
Note: *** p<.01 **p<.05 *p<.10

 
47



Table 1.3.2.2A. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1986 and 2012 (Top 6 Industries)
     
Log(Employment in 2012/Employment in 1986)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.1046
***
-0.2754
***
0.3192 *** 0.1886
***



         

Log(U.S. employment in 2012/
U.S. employment in 1986)
in the industry outside the city

0.9589 *** 0.9564 *** 0.9961 *** 1.0116 ***
         

Wage in the city-industry  
in 1986 (in thousands)

0.0001 *** 0.0001 *** 0.0001 *** 0.0001

         

Employment in the city-
industry
in 1986 (in millions)

-0.7318 *** -0.5138 ** -0.9741 *** -0.8329 ***
         

Dummy variable indicating
presence in the South

0.0619 *** 0.0757 *** 0.0582 *** 0.0772 ***

         
Manufacturing industry
dummy
-0.1304 *** -0.0721 *** -0.1197 *** -0.0600 ***  
         

Specialization indicator

-0.0210 ***
   
-0.0136 ***

       

Competition indicator
 
0.2914 ***
 
0.3280 ***

       

Diversification indicator
   
-0.4266 *** -0.8427 ***

       


         

R Square

0.5488

0.5777

0.5483

0.5919
 

Adjusted R Square

0.5476

0.5766

0.5471

0.5905
 

Standard Error

0.3556

0.3440

0.3558

0.3383
 

Observation

2,290

2,290

2,290

2,290
                         
Note: *** p<.01 **p<.05 *p<.10
 
48



Table 1.3.2.2B. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1998 and 2012 (Top 6 Industries)
     
Log(Employment in 2012/Employment in 1998)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0123

-0.1323 ***
0.1214
*** 0.0320
 
           

Log(U.S. employment in 2012/
U.S. employment in 1998)
in the industry outside the city

0.9041 *** 0.9084 *** 0.9111 *** 0.9229 ***
         

Wage in the city-industry  
in 1998 (in thousands)

0.0002 *** 0.0002 *** 0.0002 ** 0.0002 ***
         

Employment in the city-industry
in 1998 (in millions)

0.0667

0.0967

-0.0380

-0.0697

       


Dummy variable indicating
presence in the South

0.0122

0.0198 ** 0.0134

0.0232
**

       


Manufacturing industry  
dummy
-0.0638 *** -0.0475 *** -0.0640 *** -0.0473 ***  
       



Specialization indicator

-0.0024
   
-0.0001


       


Competition indicator
 
0.1185 ***


0.1347 ***

       

Diversification indicator
   
-0.1861 *** -0.3035
***

       
           

R Square

0.4831

0.4993

0.4853

0.5056
 

Adjusted R Square

0.4817

0.4980

0.4840

0.5039
 

Standard Error

0.2233

0.2198

0.2228

0.2185
 

Observation

2,290

2,290

2,290

2,290
                         
Note: *** p<.01 **p<.05 *p<.10
 
49



Table 1.3.2.2C. City-Industry Employment Growth with Manufacturing Dummy Variable
between 1986 and 1998 (Top 6 Industries)
     
Log(Employment in 1998/Employment in 1986)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0842
***
-0.2065
***
0.1228 *** 0.0268




         

Log(U.S. employment in 1998/
U.S. employment in 1986)
in the industry outside the city

0.9825 *** 1.0054 *** 1.0421 *** 1.0376 ***
         

Wage in the city-industry  
in 1986 (in thousands)

0.0001 *** 0.0001 *** 0.0001 *** 0.0001 *
         

Employment in the city-
industry
in 1986 (in millions)

-0.6689 *** -0.5274 *** -0.7903 *** -0.6706 ***
         

Dummy variable indicating
presence in the South

0.0470 *** 0.0570 *** 0.0447 *** 0.0579 ***

         
Manufacturing industry  
dummy
-0.0751 *** -0.0327 *** -0.0702 *** -0.0278 **  
         

Specialization indicator

-0.0177 ***
   
-0.0102 ***

       

Competition indicator
 
0.2148 ***
 
0.2292 ***

       

Diversification indicator
   
-0.1343 * -0.4071 ***

       


         

R Square

0.3467

0.3850

0.3406

0.3949
 

Adjusted R Square

0.3450

0.3834

0.3389

0.3928
 

Standard Error

0.2697

0.2617

0.2709

0.2596
 

Observation

2,290

2,290

2,290

2,290
                         
Note: *** p<.01 **p<.05 *p<.10



 
50



Table 1.4.1. Manufacturing Industry Employment in Cities between 1986 and 2012  
   Manufacturing Employment in Cities   Total City-Industry Employment  
   1986 1998 2012   1986 1998 2012  
         

Mean 35,921  35,258  23,224  

9,056  12,136  12,992  


Median 14,003  15,214  10,845  

1,942  2,628  2,875  


Standard Deviation 70,993  62,328  38,140  

26,772  31,514  33,625  


Sum 13,757,826  13,503,972  8,894,794  

69,371,877  92,963,741  99,521,278  

                   





Table 1.4.2. Manufacturing among City-Industries between 1986 and 2012
   1986 1998 2012  
     

The Biggest Industry 253 188 60


In Top 3 Industries 334 282 172


In Top 5 Industries 362 346 298

           
Note: the total number of cities in the dataset is 383.

 
51



Table 1.5.1. Employment by Industry between 1986 and 2012
 
NAICS Industries 1986 1998 2012
 
     

Agriculture, forestry, fishing and hunting 380,528 107,470 99,589


Mining, quarrying, and oil and gas extraction 326,815 248,948 332,921


Utilities 502,683 578,688 553,016


Construction 4,162,565 5,088,494 4,568,193


Manufacturing 13,757,826 13,503,972 8,894,794


Wholesale trade 4,184,719 5,317,487 5,109,637


Retail trade 9,550,211 11,987,395 12,582,687


Transportation and warehousing 2,154,477 3,055,813 3,693,007


Information 2,527,932 2,889,906 2,961,711


Finance and insurance 4,184,821 5,325,830 5,420,148


Real estate and rental and leasing 1,305,405 1,658,174 1,771,186


Professional, scientific, and technical services 2,765,284 5,655,452 7,386,782


Management of companies and enterprises 1,078,377 2,608,461 2,938,949


Administrative and support and waste management
and remediation services 3,496,488 7,237,831 7,211,582


Educational services 1,410,431 2,073,170 3,216,587


Health care and social assistance 6,746,827 11,726,831 15,896,498


Arts, entertainment, and recreation 1,053,904 1,404,909 1,854,970


Accommodation and food services 5,947,430 8,028,241 10,374,149


Other services (except public administration) 3,076,437 4,394,199 4,643,550


Industries not classified 758,718 72,470 11,322

     

Sum 69,371,877 92,963,741 99,521,278
           




52



Table 1.5.2. Employment Share by Industry (in percentage) between 1986 and 2012
 
NAICS Industries 1986 1998 2012
 
     

Agriculture, forestry, fishing and hunting 0.55 0.12 0.01


Mining, quarrying, and oil and gas extraction 0.47 0.27 0.33


Utilities 0.72 0.62 0.56


Construction 6.00 5.47 4.59


Manufacturing 19.83 14.53 8.94


Wholesale trade 6.03 5.72 5.13


Retail trade 13.77 12.89 12.64


Transportation and warehousing 3.11 3.29 3.71


Information 3.64 3.11 2.98


Finance and insurance 6.03 5.73 5.45


Real estate and rental and leasing 1.88 1.78 1.78


Professional, scientific, and technical services 3.99 6.08 7.42


Management of companies and enterprises 1.55 2.81 2.95


Administrative and support and waste management
and remediation services 5.04 7.79 7.25


Educational services 2.03 2.23 3.23


Health care and social assistance 9.73 12.61 15.97


Arts, entertainment, and recreation 1.52 1.51 1.86


Accommodation and food services 8.57 8.64 10.42


Other services (except public administration) 4.43 4.73 4.67


Industries not classified 1.09 0.08 0.01
           


 
53



Table 1.6.1A. Employment Growth of Smaller Industries between 1986 and 2012 (All 20
Industries)
     
Log(Employment in 2012/Employment in 1986)
in the City Outside the Four Biggest Industries
   
 Variables        
     

Intercept

0.2982 ***

     

1986 employment outside the four
biggest industries (in millions)  
-0.1079 ***

 
 

1986 wage outside the four  
biggest industries (in thousands)

0.0000 ***

   

Employment growth in
the four biggest industries

0.2686 ***

   
     

R Square

0.0861
 

Adjusted R Square

0.0857
 

Standard Error

0.3224
 

Observation

6,719
 
           
Note: *** p<.01 **p<.05 *p<.10

 
54



Table 1.6.1B. Employment Growth of Smaller Industries between 1998 and 2012 (All 20
Industries)
     
Log(Employment in 2012/Employment in 1998)
in the City Outside the Four Biggest Industries
   
 Variables        
     

Intercept

0.0511 ***

     

1998 employment outside the four
biggest industries (in millions)  
-0.0258 ***

   

1998 wage outside the four  
biggest industries (in thousands)

0.0000 *

   

Employment growth in
the four biggest industries

0.4121 ***

   
     

R Square

0.2232
 

Adjusted R Square

0.2228
 

Standard Error

0.1309
 

Observation

6,719
 
           
Note: *** p<.01 **p<.05 *p<.10
 
55



Table 1.6.1C. Employment Growth of Smaller Industries between 1986 and 1998 (All 20
Industries)
     
Log(Employment in 1998/Employment in 1986)
in the City Outside the Four Biggest Industries
   
 Variables        
     

Intercept

0.2943 ***

     

1986 employment outside the four
biggest industries (in millions)  
-0.0933 ***

 
 

1986 wage outside the four  
biggest industries (in thousands)

0.0000 ***

   

Employment growth in
the four biggest industries

0.0941 ***

   
     

R Square

0.0161
 

Adjusted R Square

0.0156
 

Standard Error

0.2555
 

Observation

6,719
 
           
Note: *** p<.01 **p<.05 *p<.10

 
56



Table 1.6.2A. Employment Growth of Smaller Industries between 1986 and 2012 (Top 6
Industries)
     
Log(Employment in 2012/Employment in 1986)
in the City Outside the Four Biggest Industries
   
 Variables        
     

Intercept

0.2869 ***

     

1986 employment outside the four
biggest industries (in millions)  
-0.1056 **

 
 

1986 wage outside the four  
biggest industries (in thousands)

0.0000 ***

   

Employment growth in
the four biggest industries

0.2846 ***

   
     

R Square

0.0870
 

Adjusted R Square

0.0858
 

Standard Error

0.3403
 

Observation

2,290
 
           
Note: *** p<.01 **p<.05 *p<.10

 
57



Table 1.6.2B. Employment Growth of Smaller Industries between 1998 and 2012 (Top 6
Industries)
     
Log(Employment in 2012/Employment in 1998)
in the City Outside the Four Biggest Industries
   
 Variables        
     

Intercept

0.0496 ***

     

1998 employment outside the four
biggest industries (in millions)  
-0.0273 *

   

1998 wage outside the four  
biggest industries (in thousands)

0.0000
 
   

Employment growth in
the four biggest industries

0.4135 ***

   
     

R Square

0.2208
 

Adjusted R Square

0.2197
 

Standard Error

0.1339
 

Observation

2,290
 
           
Note: *** p<.01 **p<.05 *p<.10
 
58



Table 1.6.2C. Employment Growth of Smaller Industries between 1986 and 1998 (Top 6 Industries)
     
Log(Employment in 1998/Employment in 1986)
in the City Outside the Four Biggest Industries
   
 Variables        
     

Intercept

0.2858 ***

     

1986 employment outside the four
biggest industries (in millions)  
-0.0914 **

 
 

1986 wage outside the four  
biggest industries (in thousands)

0.0000 ***

   

Employment growth in
the four biggest industries

0.1086 ***

   
     

R Square

0.0175
 

Adjusted R Square

0.0162
 

Standard Error

0.2701
 

Observation

2,290
 
           
Note: *** p<.01 **p<.05 *p<.10



 




 
59



CHAPTER TWO: DYNAMIC EXTERNALITIES AND WAGE GROWTH IN U.S.
CITIES
Introduction
This is the second part of the study reproducing the empirical research of Glaeser, Kallal,
Scheinkman, and Shleifer (1992). The first part examined the effects of different dynamic
externalities on growth through employment-based regressions. This second part investigates the
same research question through wage-based regressions.  
This study explores the effects of agglomeration externalities related to knowledge
spillovers on wage growth in city industries. Wage refers to average annual payroll per employee
in the city industry. Glaeser et al. conducted a wage growth regression as a brief secondary
analysis to their employment growth analysis. They primarily measured industry growth using
employment numbers, but they noted that a better measure would be productivity growth and a
proxy might be wage growth. Glaeser et al.’s work suffered from the limited availability of data
for their study period (1956 to 1987). They had the data for only the six largest city industries,
and the small size of their sample (833 observations) led to concerns about the possibility of
sample selection bias. The updated dataset used in the present study is constructed with
significantly higher accuracy and coverage. The dataset contains 6,719 observations for all
industry sectors in U.S. cities between 1998 and 2012. There are still not many empirical studies
using nationwide data. Parallel regressions using employment and wages will provide evidence
to compare the analysis results of different measures of economic growth.    
60



Literature Review
Dynamic Externalities and Wage Growth
The majority of empirical studies on regional economic development have focused on
employment growth. However, the appropriateness of employment-based studies has been
questioned. The most obvious issue is the association between employment growth and local
condition. Employment growth analysis assumes that productivity increases due to
agglomeration economies resulting in proportional employment growth and that labor is a
homogenous input. However, the growth of employment is also affected by the local industrial
and social environment, such as resource endowments, weather, regulatory conditions, and other
factors unrelated to the productivity of a region (e.g., Cingano & Schivardi, 2004; Dekle, 2002;
Henderson, 2003).  
Another issue is that dynamic externalities might have different impacts on employment
and wage growth. Duranton and Puga (2001) suggested the product life cycle as a reason for
such a difference. They argued that diversity in a city stimulates innovations suited to the early
stage of a product, whereas specialization can be an advantageous environment to fully
developed products. This implies that employment growth is more likely to benefit from
diversification, while productivity growth would benefit from specialization. Moreover, some
studies have shown that the choice of employment and wage growth affect analysis results.
Cingano and Schivardi (2004) investigated the source of dynamic externalities using total factor
productivity (TFP) data from Italy. Concerning the issues employment-based growth analysis
generate, they ran two sets of regressions based on TFP and employment for comparison. The
TFP-based regression yielded evidence of specialization, but not for competition or diversity,
while the opposite result was found for the employment growth analysis. Almeida (2007) also
61



found evidence of specialization in some sectors, but no evidence of competition or diversity in
most sectors, in a study on regional adjusted wage growth in Portugal. The opposite evidence
was found for employment growth.  
Beaudry and Schiffauerova (2009) reviewed the evidence from a large amount of
empirical literature on dynamic externalities. They confirmed that employment growth is popular
as a dependent variable because of data availability. Jacobs externalities are most frequently
found to be effective in employment growth regressions, whereas MAR externalities show
positive effects more often in productivity growth regressions.        
Henderson (2003) estimated plant-level production functions for machinery and high-
tech manufacturing sectors using U.S. wage data. He found positive effects of localization/MAR
externalities, but no evidence of diversity/Jacobs economies. Malpezzi, Seah, and Shilling (2004)
used growth in total earnings as a measure of productivity. By examining wage growth in U.S.
metropolitan areas between 1970 and 1999, they found that both urbanization and localization
economies contribute local growth. Their evidence reveals that local productivity is more
important than industry mix. They also identified negative effects of initial earnings per capita on
growth.  
Theories of Productivity Growth
Economic growth can be decomposed into the growth of factor inputs and the growth of
total factor productivity (TFP). This paper makes a connection to the theories of productivity
growth, in that it presents an investigation into wage growth. Following Solow (1956) and Swan
(1956), the neoclassical growth models explained the growth of output incorporating the growth
of productivity. They explain economic growth by capital accumulation, population growth, and
62



increase in productivity which can be explained by technological progress. In Solow’s (1956)
model of long-run growth, marginal productivity determines the growth path of real wages.
Growth can be achieved only through technological progress. The model predicts convergence.  
Economists have empirically investigated whether per capita income converges across
regions. The studies examining the U.S. and Western Europe generally find evidence for
convergence (Barro & Sala-i-Martin, 1991; Carlino & Mills, 1993). Baumol (1986) and De Long
(1988) found that income convergence exists only for a group of industrialized countries,
whereas there is no universal convergence. Many other studies have shown the importance of
political and social factors for growth (Barro, 1991; De Long & Shleifer, 1993; Glaeser,
Scheinkman, & Shleifer, 1995).  
Data and Variables
Construction of the Data Set
This study uses the same dataset that was used for the U.S. employment growth analysis,
presented earlier. The dataset contains employment, number of establishments, and annual
payroll data for 1986, 1998, and 2012 by 20 two-digit NAICS industries per each 383 CBSA.
The source of the data is the County Business Patterns (CBP), which is produced by the U.S.
Census Bureau.
Annual payroll data by place of work from CBP is an annual extension of the Census
Bureau’s quinquennial economic censuses.  Payroll includes all forms of compensation paid
during the year to all employees, such as salaries, wages, commissions, dismissal pay, bonuses,
vacation allowances, sick-leave pay, and employee contributions to qualified pension plans.
Payroll is reported before deductions for social security, income tax, insurance, union dues, and
so on.  
63



The CBP data are derived from Federal administrative records and survey information of
business establishments. The CBP data exclude data on self-employed individuals, employees of
private households, railroad employees, agricultural production employees, and most
government employees.  Still, it covers a good part of economic activities in the country.
Wage data is obtained by dividing annual payroll by employment, as in Glaeser et al.
Thus, wage is, specifically, average annual payroll in city-industry. What is referred to as ‘total
wage’ in this study is equal to annual payroll in the corresponding industry or region.  
Glaeser et al. (1992) tested the effects of dynamic externalities on the growth of nominal
city-industry wage. The payroll data from CBP is nominal. This study tests real wage along with
nominal wage. The real values were calculated by applying corresponding GDP deflators
provided by the Bureau of Economic Analysis. The values of GDP deflators are 58.395 for 1986,
78.859 for 1998, and 105.214 for 2012. The base year is 2009.    
The employment growth analysis used the data for 1986, 1998, and 2012 to examine
change over the longest period possible. Since 1998, CBP has been tabulated according to the
North American Industry Classification System (NAICS). Data for prior periods were tabulated
based on the Standard Industrial Classification System (SIC). To make the 1986 data comparable
to the later data, the SIC-to-NAICS employment conversion ratios from the Bureau of Labor
Statistics were applied. However, because employment and annual payroll are different in nature,
it would not be appropriate to use payroll data converted by employment ratios for wage growth
analysis. Wage growth is therefore examined only for the period from 1998 to 2012, which
consistently follows NAICS.  

64



The original CBP data include many missing or zero values. When data do not meet
publication standards, data are withheld for confidentiality, and employment or payroll data is set
to zero. In these cases, the data were excluded from the dataset, causing the number of
observations to decrease from 7,660 to 6,719. Most of such instances happen for only a few
industries, including agriculture, forestry, fishing and hunting, mining, quarrying, oil and gas
extraction, utilities, and industries not classified. Such missing values in these industries would
not seriously harm the purpose of this study, which is examining wage growth in city industries.        
Description of the Data
Tables 2.1.1 to 2.1.3 show the changing trend of wages between 1998 and 2012. Table
2.1.1 presents the ten largest cities in terms of total real wage. Total wage refers to the sum of
annual payroll paid in the city industry. The listed cities are also the largest cities in terms of
employment, in general. The amount of total wage paid in the city was the largest in New York-
White Plains-Wayne, NY-NJ, in both 1998 and 2012, followed by Atlanta-Sandy Springs-
Marietta, GA, and Minneapolis-St. Paul-Bloomington, MN-WI.  
Tables 2.1.2A and 2.1.2B show the largest city industries measured by total real wage.
The change from 1998 to 2012 is similar to that of the largest city-industry employment.
Manufacturing and wholesale trade are traditionally the largest industries across cities. However,
the data show that these industries are declining not only in terms of employment but also in
terms of total wage. The other largest city industries are finance and insurance; professional,
scientific, and technical services; and health care and social assistance.
65



In terms of wage growth, Table 2.1.3 shows that the largest wage growth between 1998
and 2012 occurred mostly in small cities. The regression results presented below also indicate
that initial city-industry wage has a negative relationship with city wage growth.    
Variables and Descriptive Statistics
Wage growth is analyzed in the same format as employment growth regression. Growth
is measured using wage data. The definition of externality indicators and the choice of
independent variables rem ain the sam e. Tables 2.2.1 and 2.2.2 present variable m eans and
standard deviations between 1998 and 2012 for all 20 industries and for the top six industries,
respectively.  
The dependent variable is wage growth, which is defined as the log of the wage in the
end year divided by the wage in the beginning year.
wage growth = ln
wage 𝑖 𝑒 𝑒 ℎ 𝑒 𝑒𝑒𝑒 𝑒𝑒 𝑦 𝑦 wage 𝑖 𝑒 𝑒 ℎ 𝑒 𝑏 𝑒 𝑏 𝑖 𝑒𝑒𝑖 𝑒 𝑏 𝑒𝑒 𝑦 𝑦
The mean of nominal wage growth is 0.3948, and the mean of real wage growth is 0.1065.
The same indicators of externalities defined following Glaeser et al. are used as the
employment growth analysis. The specialization indicator is defined as the city industry’s share
of city employment relative to the U.S. industry’s share of the U.S. employment.  
specialization =
𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑒𝑒 𝑒𝑒𝑒 𝑖 𝑒 𝑐 𝑖𝑒𝑒 𝑒 𝑒 𝑒 𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑐 𝑖𝑒𝑒 ⁄
𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑈 . 𝑆 . 𝑒 𝑒 𝑒 𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑈 . 𝑆 . ⁄

An indicator value greater than one means that the industry has a greater representation in
the city than in the nation. This can be interpreted to mean that the industry is specialized in the
city.
66



The competition indicator is the number of firms per worker in the city industry relative
to the number of firms per worker in the industry in the U.S.  
competition =
𝑓 𝑖𝑦 𝑒 𝑖 𝑖 𝑒 𝑐 𝑖𝑒𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦 𝑖 𝑖 𝑒 𝑐 𝑖𝑒𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 ⁄
𝑓 𝑖𝑦 𝑒 𝑖 𝑖 𝑒 𝑒 ℎ 𝑒 𝑈 . 𝑆 . 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑖 𝑒 𝑒 ℎ 𝑒 𝑈 . 𝑆 . 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 ⁄

By comparing a city to the nation, we can measure how concentrated an industry is in a
city. When industries yield a value greater than one, this can be interpreted to indicate that firms
in those industries are more concentrated in the city compared to the nation. Therefore, those
industries are locally more competitive. Alternatively, the value can be greater than one when
firms of an industry in a city are smaller than their national average. Because it is very hard to
distinguish these two cases, Glaeser et al.’s approach was to use the data of only the six biggest
industries in each city to decrease the possibilities of having the value greater than one just
because firms are small. However, the biggest industries can be made up of many small firms.
The mean of the competition indicator of Glaeser et al.’s dataset was less than 1 (0.752) in 1956.  
However it was 1.0895 in 1986 and 1.0681 in 1998 for the top six industries, suggesting that
cities have become more competitive.    
Diversity is measured as the share of total city employment of the city’s other top five
industries.
diversity =
𝑐 𝑖𝑒 𝑒 ′
𝑖 𝑒𝑒 ℎ 𝑒 𝑦 𝑒 𝑒 𝑒 𝑓 𝑖 𝑓 𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑐 𝑖𝑒𝑒
Glaeser et al. tried to measure a variety of industries by looking at the large industry
employment in a city other than the industry in question. In a diverse city, even the largest
industries may not take a big share as in a specialized city. In this case, the low value of the
67



diversity indicator may suggest that a city is diverse. The mean of the diversity indicator in
Glaeser et al.’s dataset was 0.351 in 1956. The indicator mean was 0.5859 in 1986 and 0.5866 in
1998 for the top six industries. According to this indicator, cities have become less diverse.  
In addition to these variables, the national employment growth or wage growth in an
industry, the wage and the employment of the city industry in the initial year, and a dummy
variable indicating that a city is located in the South are included in the regression analysis as
controls. These variables are expected to affect local employment growth.  
When employment in an industry grows quickly in the whole country, employment in
that industry in cities may also grow quickly. Therefore, the national employment growth is
expected to have a positive sign. The South dummy variable is also expected to have a positive
sign, because cities located in the South tend to grow faster than other cities. The effect of the
initial wage will be negative, and the effect of the initial employment will be positive, if the data
support convergence theory.    
Regression Results
The regression result is presented for both nominal and real wage growth between 1998
and 2012 for all twenty industries. As in the employment growth analysis, the result for the top
six city industries is also presented for comparison with Glaeser et al.’s results (as noted earlier,
Glaeser et al. analyzed only the six largest industries in each city, due to data limitations).  
Wage Growth in Cities
Tables 2.3.1 to 2.4 present the reproduced analysis results of the wage growth in city-
industries. The regression result using nominal and real wage for all industries is almost the same,
as shown in tables 2.3.1 and 2.3.2. Table 2.4 shows the result of the regression for the top six
68



industries. It is consistent with the result for all industries in general. However, it should be noted
that all three regressions show very low R-square and adjusted R-square values.
As expected, there are some differences in the estimation result for wage growth versus
that for employment growth. This result is different from Glaeser et al.’s wage growth result as
well. Glaeser et al. found that specialization has no effect on wage growth, while competition
reduces wage growth and diversity helps growth. Their findings contradict Porter’s but confirm
Jacobs’s theory.  
In this study, only diversification indicators have statistically significant coefficients
throughout every wage growth regression. In the regressions for all industries, the coefficients of
specialization and competition are not statistically significant, whereas the coefficients of
diversification are highly significant. The negative sign of the diversity coefficients, in tables
2.3.1 and 2.3.2, suggests that cities grow more when their largest industries do not take a large
proportion of the total city employment. Diversification of city-industry induces the growth of
wages as well as the growth of employment. This result supports Jacobs’s view. No evidence of
MAR or Porter’s view is found.
On the other hand, in the regression for the top six industries, shown in table 2.4, the
coefficients of all three externality indicators are statistically significant. The coefficients of
diversification are negative and statistically significant. The sign of the coefficients of
competition are mixed. It is noticeable that the specialization indicator is positive and statistically
significant. The specialization indicator has consistently shown a negative impact on
employment growth and all-industry wage growth. No evidence for specialization was found by
Glaeser et al., either. The evidence in table 2.4, on the contrary, is favorable to MAR, as it
69



suggests that specialization contributed to wage growth of the largest city industries, between
1998 and 2012. However, it also should be noted that the R-square and adjusted R-square values
of table 2.4 are extremely low.  
   The overall result in tables 2.3.1 to 2.4 suggests that wage, or more accurately average
annual payroll per employee in the city industry, grows in cities with more diverse industries.
Specialization in a city also contributes to wage growth of the largest city industries.      
With respect to the control variables, initial wage and initial employment show opposite
results from the employment growth regressions. The result of all other variables remains the
same. This is consistent with Glaeser et al.’s result. City-industry wage growth has positive
relations with national wage growth, initial employment, and location in the South. Initial wage
in city industry is negatively related with wage growth.    
The coefficients of the national growth are not only positive and statistically significant
but also extremely large. A one-percent increase in the national wage growth would yield about a
0.8 percent and about a 0.99 percent increase in the wage growth of all city industries and the top
six city industries, respectively. The nationwide growth has significant impacts on city growth in
terms of wages. This is the case with the employment growth regressions as well. However, the
impact on wage growth is greater than the impact on employment growth for the same period.
This could be due to the fact that the wage difference across cities is smaller than the difference
in the number of employment.    
In contrast to the impact on employment growth, initial wage in the city industry has
negative effects on wage growth, whereas initial employment in the city industry has positive
effects. This result is consistent with several previous studies. Glaeser, Scheinkman, and Shleifer
70



(1995) found that population and income growth tend to move together. Barro and Sala-i-Martin
(1991), Carlino and Mills (1993), and Malpezzi et al. (2004) found negative effects of high initial
wage on growth. The negative relation between initial wage and wage growth corresponds to the
prediction of convergence theory.    
The Impact of Dynamic Externalities by Industry
While a full analysis to explain the industry-specific function of agglomeration
externalities is beyond the scope of this paper, estimating the separate impact of externalities for
each industry might provide some useful information, along with the all-industry results. Thus,
the same regressions were run for each industry sector to see how the impact of dynamic
externality indicators varies by industry.  
In the all-industry analysis for employment growth (presented earlier), com petition and
diversification showed a positive impact on city industry employment growth, whereas
specialization showed a negative impact. In the all-industry analysis for wage growth, shown in
tables 2.3.1 and 2.3.2, only diversification shows a positive impact.  
Tables 2.5.1 and 2.5.2 present the effects of dynamic externalities for each city-industry
on the growth of employment and wage, respectively. The tables list the signs of the estimated
coefficients of each externality indicator with its statistical significance. Therefore, as in the
regression result tables below, a positive sign of the specialization and competition indicators
points to positive impacts, and a negative sign of the diversification indicator also points to
positive impacts.      
The overall results shown in the two tables are similar to, and tend to correspond with,
the all-industry analysis results. It is noticeable that the positive effect of the competition
71



indicator is quite consistent across industries in both cases. The diversification indicator also
bears a negative sign in several cases, which indicates a positive effect. On the other hand,
specialization shows negative effects in a majority of the cases.  
This individual industry analysis result shows the difference between employment and
wage growth analysis better than the all-industry analysis result. The previous literature has
argued that the choice of performance measure might affect the analysis result and therefore
inferences based on it. For example, the evidence for specialization is often found in wage-based
analysis, as opposed to the result of employment-based analysis (e.g., Almeida, 2007; Cingano &
Schivardi, 2004; Henderson, 2003; Malpezzi, Seah, & Shilling, 2004).  
The evidence is similar in this study. Although specialization shows a negative impact on
employment growth, and no evidence for specialization is found in the all-industry wage growth
analysis, specialization is found to be effective on wage growth for six of the industries when
examined individually. The influence is negative only for one industry sector and is actually
positive for five city industries on the wage growth. Competition has a statistically significant
impact on the greatest number of industries. The impact of competition is consistently positive
on employment growth. Its impact on wage growth is negative for a few industries, which
include agriculture, forestry, fishing and hunting and mining. Those are industries that typically
benefit from economies of scale. Diversification shows positive effects for many industries on
both employment and wage growth.  
The different characteristics of industries are also observable from the results. Among the
empirical studies on agglomeration externalities, not many consider all industry sectors. It is
common to rely on the analysis of only one or a few industry sectors, mainly due to problems
72



with data availability. This difference in industry sector choice could be one reason for the
inconsistency in the empirical literature, because it has often been suggested that the effects of
agglomeration externalities are likely to vary across industries. By examining manufacturing
employment, for instance, Henderson, Kuncoro, and Turner (1995) found that only MAR
externality performs in mature capital goods industries, whereas evidence of both MAR and
Jacobs externalities is found in young high-tech industries. Duranton and Puga (2001) also
argued that the effects of agglomeration externalities differ according to the stages of a product’s
life cycle. They found that diversity in a city helps new industries to grow, whereas a specialized
environment is better for mature industries. Beaudry and Schiffauerova (2009) investigated the
effect of MAR and Jacobs externalities in the literature according to type of industries. When
categorizing industries into low-tech, medium-tech, high-tech, and service industries, the effect
of MAR externalities tend to be found to be positive more often in low-tech sectors. On the other
hand, the effect of the Jacobs externality is found to be positive in overall industry and seems to
be stronger in industries related with higher technology.      
Industries can also be categorized into tradable and non-tradable sectors according to the
tradability of their output. Tradable sectors are regional industries that sell their products and
services across regions. Non-tradable sectors, on the other hand, are industries whose products
and services are consumed locally. This study utilizes 2-digit NAICS industry classification,
which is a relatively broad aggregation. Therefore, only a rough distinction can be made.
Manufacturing commonly consists in large part of tradable sectors. Agriculture, mining, and
wholesale may also fall in the category. Non-tradable sectors include m ost service industries,
such as administrative, educational, health care, and food services. In addition, utilities,
construction, retail, and real estate could also be included.
73



 Scholars have remarked on the importance of the tradable industry sector in the regional
economy. Porter (2003) studied the economic performance of U.S. regions over the period from
1990 to 2000. Among industries classified into traded, local, and resource-dependent, traded
industries account for only about one-third of employment. However, average wage, productivity,
and wage growth are much higher in traded industries. The author thus argued that clusters of
traded industries strongly influence regional economic performance. They shape local wages and
drive local employment. Moretti (2010) estimated local multipliers at the city level using U.S.
data from 1980, 1990, and 2000. The author found that each new job in the tradable sector in a
given city generates 1.6 new jobs in the non-tradable sector in the same city. This multiplier is
also significantly larger for skilled jobs, as the wage in such jobs is higher. In industry-specific
estimation, the multipliers of high-tech industries are the largest.  
In tables 2.5.1 and 2.5.2, the results of the two types of sectors seem to be distinguishable.
Whereas all three externalities tend to have an impact on the growth of the non-tradable sector,
only specialization and competition affect the growth of the tradable sector. No evidence for
diversification is found for the tradable sector.    
Diversity is a feature of urban environments that is very frequently found to be effective
in creating positive externalities. This study also found evidence for diversification on both
employment and wage growth of city industry in the analysis using all-industry data. If industrial
diversity in a city works differently in the tradable sector, as this result is based on rough
classification, it requires further examination of the role of the tradable sector in the regional
economy.    
74



Conclusion
This study explores the effects of three types of knowledge externalities on wage growth
in U.S. cities between 1986 and 2012. This parallels the externality analysis on employment
growth presented in the first part. Glaeser et al.’s (1992) wage regression is reproduced with the
recent data. Wage is a better measure than employment for assessing economic growth. The
wage data for this study cover a shorter period than the employment data cover. However,
comparing the analysis results of two different economic performance measures allows for a
more comprehensive investigation into the work of dynamic externalities.  
This study finds evidence of diversification, as did Glaeser et al.’s results. Diversity
shows positive effects on the growth of city-industry wage. Specialization is found to have an
effect on wage growth in the largest city industries. Unlike the employment growth analysis, no
evidence for competition is found in the wage growth analysis. The choice of economic
performance measure thus seems to have affected the analysis result.  
The additional regressions for individual industry suggest the impact of dynamic
externalities may differ depending on the characteristics of each industry. When distinguishing
industries into tradable and non-tradable industries, the growth of tradable sector is affected only
by specialization and competition. Diversification shows no impact on the employment and wage
growth of the tradable sector.  
 
75



Table 2.1.1. 10 Largest Real Wage Cities

Total Real Wage (in billions)  
 
CBSA  1998 2012  

   

1 New York-White Plains-Wayne, NY-NJ
                 
2,126.37                  2,495.32  

2 Atlanta-Sandy Springs-Marietta, GA
                 
1,530.40                  1,748.40  

3 Minneapolis-St. Paul-Bloomington, MN-WI
                 
1,361.48                  1,574.12  

4 Chicago-Joliet-Naperville, IL
                     
997.38                  1,066.98  

5 Washington-Arlington-Alexandria DC-VA
                     
786.88                  1,738.92  

6
Waterloo-Cedar Falls, IA
                     
786.88                  1,738.92  

7
Houston-Sugar Land-Baytown, TX
                     
606.24                      921.26  

8 St. Louis, MO-IL
                     
543.21                      551.13  

9 Dallas-Plano-Irving, TX
                     
446.09                      541.76  

10 Philadelphia, PA
                   
358.41                      424.76  
 
     

 
76



Table 2.1.2A. 20 Largest Real Wage City-Industries in 1998

 
Total Real Wage
(in billions)

 CBSA Industry 1998 2012  



 
1 New York-White Plains-Wayne, NY-NJ
Finance and insurance
351.07 439.07  
2 New York-White Plains-Wayne, NY-NJ
Health care and social assistance
318.24 427.96  
3 Minneapolis-St. Paul-Bloomington, MN-WI
Manufacturing
307.21 256.63  
4 New York-White Plains-Wayne, NY-NJ Professional, scientific, and
technical services
211.21 287.36  
5 Washington-Arlington-Alexandria DC-VA Professional, scientific, and
technical services
207.69 620.79  
6 Atlanta-Sandy Springs-Marietta, GA
Manufacturing
203.42 124.25  
7 Chicago-Joliet-Naperville, IL
Manufacturing
198.66 130.30  
8 New York-White Plains-Wayne, NY-NJ
Wholesale trade
181.54 177.41  
9 Atlanta-Sandy Springs-Marietta, GA
Wholesale trade
168.30 158.10  
10 New York-White Plains-Wayne, NY-NJ
Manufacturing
158.10 78.15  
11 New York-White Plains-Wayne, NY-NJ Management of companies and
enterprises
147.96 167.63  
12 Atlanta-Sandy Springs-Marietta, GA
Retail trade
144.28 148.93  
13 Minneapolis-St. Paul-Bloomington, MN-WI
Health care and social assistance
142.46 237.74  
14 Atlanta-Sandy Springs-Marietta, GA Professional, scientific, and
technical services
138.55 195.89  
15 New York-White Plains-Wayne, NY-NJ
Information
136.20 134.31  
16 Atlanta-Sandy Springs-Marietta, GA
Health care and social assistance
132.36 234.28  
17 Minneapolis-St. Paul-Bloomington, MN-WI
Finance and insurance
127.76 172.03  
18
Atlanta-Sandy Springs-Marietta, GA
Administrative and support and
waste management and
remediation services
122.44 130.56

19 Minneapolis-St. Paul-Bloomington, MN-WI
Wholesale trade
118.93 139.91  
20 St. Louis, MO-IL
Manufacturing
113.57 68.99  
 
 
77



Table 2.1.2B. 20 Largest Real Wage City-Industries in 2012

 
Total Real Wage
(in billions)

 CBSA Industry 1998 2012  



 
1
Washington-Arlington-Alexandria DC-VA
Professional, scientific, and
technical services 207.69 620.79

2 New York-White Plains-Wayne, NY-NJ Finance and insurance 351.07 439.07  
3 New York-White Plains-Wayne, NY-NJ Health care and social assistance 318.24 427.96  
4
New York-White Plains-Wayne, NY-NJ
Professional, scientific, and
technical services 211.21 287.36

5 Minneapolis-St. Paul-Bloomington, MN-WI Manufacturing 307.21 256.63  
6 Minneapolis-St. Paul-Bloomington, MN-WI Health care and social assistance 142.46 237.74  
7 Atlanta-Sandy Springs-Marietta, GA Health care and social assistance 132.36 234.28  
8
Atlanta-Sandy Springs-Marietta, GA
Professional, scientific, and
technical services 138.55 195.89

9 New York-White Plains-Wayne, NY-NJ Wholesale trade 181.54 177.41  
10 Washington-Arlington-Alexandria DC-VA Health care and social assistance 61.71 175.88  
11 Minneapolis-St. Paul-Bloomington, MN-WI Finance and insurance 127.76 172.03  
12
New York-White Plains-Wayne, NY-NJ
Management of companies and
enterprises 147.96 167.63

13 Atlanta-Sandy Springs-Marietta, GA Wholesale trade 168.30 158.10  
14 Chicago-Joliet-Naperville, IL Health care and social assistance 103.30 157.51  
15 Houston-Sugar Land-Baytown, TX Manufacturing 103.24 149.64  
16 Atlanta-Sandy Springs-Marietta, GA Retail trade 144.28 148.93  
17 New York-White Plains-Wayne, NY-NJ Retail trade 112.67 145.38  
18 Minneapolis-St. Paul-Bloomington, MN-WI Wholesale trade 118.93 139.91  
19 New York-White Plains-Wayne, NY-NJ Information 136.20 134.31  
20 Atlanta-Sandy Springs-Marietta, GA Finance and insurance 111.80 132.39  

78



Table 2.1.3. 20 Most Grown Real Wage Cities, 1998 and 2012

Growth
Total Real Wage
(in billions)

 
CBSA
Log(total real wag e i n
2012/total real wag e i n 1998)

1998 2012  

   

1 Midland, TX 0.9343  1.62 4.12
2 Odessa, TX 0.8511  1.18 2.77
3 Hinesville-Fort Stewart, GA 0.8011  0.22 0.48
4 Washington-Arlington-Alexandria DC-VA 0.7929  786.88 1738.92
5 Waterloo-Cedar Falls, IA 0.7929  786.88 1738.92
6 Casper, WY 0.7543  0.72 1.54
7 Palm Coast, FL 0.6278  0.21 0.40
8 St. George, UT 0.6151  0.57 1.06
9 McAllen-Edinburg-Mission, TX 0.6144  2.37 4.38
10 Bismarck, ND 0.5379  2.06 3.53
 
11 Anchorage, AK 0.5290  9.02 15.30
12 College Station-Bryan, TX 0.5066  2.92 4.84
13 Greeley, CO 0.5052  1.62 2.68
14 Morgantown, WV 0.4914  1.71 2.79
15 San Antonio-New Braunfels, TX 0.4813  111.10 179.78
16 Sioux Falls, SD 0.4787  9.18 14.81
17 Bakersfield-Delano, CA 0.4765  4.68 7.54
18 Las Cruces, NM 0.4730  0.85 1.37
19 Austin-Round Rock-San Marcos, TX 0.4728  74.76 119.95
20 Lafayette, LA 0.4682  6.40 10.22
 
     
 
79



Table 2.2.1. Variable Means and Standard Deviations (All 20 Industries)
 Variable Mean
Standard
Deviation
Number of
Observations
 
     

Nominal wage growth, 1998-2012
   
 
Log(nominal wage in 2012/ nominal wage in
1998) in the City-Industry
0.3948 0.3583                   6,719

Real wage growth, 1998-2012    
 
Log(real wage in 2012/ real wage in 1998) in
the City-Industry
0.1065 0.3586 6,719

     

National nominal wage growth, 1998-2012    

 
Log(U.S. nominal wage in 2012/
U.S. nominal wage in 1998)
in the industry outside the city
0.3998 0.1465 20

National real wage growth, 1998-2012    
 
Log(U.S. real wage in 2012/
U.S. real wage in 1998)
in the industry outside the city  
0.1115 0.1465 20

     

Nominal wage in the city-industry  
in 1998(in thousands)
63.8009 76.3142                   6,719


Real wage in the city-industry  
in 1998 (in thousands)
80.9050 96.7730 6,719  

Employment in the city-industry
in 1998 (in millions)
               0.0138                  0.0333  6,719

     

Specialization indicator
   
 
City-industry's share of city employment
relative to U.S. industry's share of U.S.
employment in 1998
1.0636 1.9071                   6,719


Competition indicator
   
 
Establishments per employee in the city-
industry relative to establishments per
employee in the U.S. industry in 1998
1.3973 0.8288
                                   
6,719  

Diversification indicator
   
 
City's other top five industries' share of 1998
total city-industry
0.6134 0.1057                   6,719

     
             
 
80



Table 2.2.2. Variable Means and Standard Deviations (Top 6 Industries)
 Variable Mean
Standard
Deviation
Number of
Observations
 
     

Nominal wage growth, 1998-2012
   
 
Log(nominal wage in 2012/ nominal wage in
1998) in the City-Industry
0.3998 0.1791                   2,290

     

National nominal wage growth, 1998-2012
   
 
Log(U.S. nominal wage in 2012/
U.S. nominal wage in 1998)
in the industry outside the city
0.4035 0.0430 6

     

Nominal wage in the city-industry  
in 1998 (in thousands)
64.7561 80.4298 2,290


Employment in the city-industry
in 1998 (in millions)
0.0252                  0.0473                    2,290

     

Specialization indicator
   
 
City-industry's share of city employment
relative to U.S. industry's share of U.S.
employment in 1998
1.2922 2.0322                   2,290


Competition indicator
   
 
Establishments per employee in the city-
industry relative to establishments per
employee in the U.S. industry in 1998
1.0681 0.3483                   2,290


Diversification indicator
   
 
City's other top five industries' share of 1998
total city-industry
0.5866 0.0902                   2,290

             
 
81



Table 2.3.1. City-Industry Nominal Wage Growth between 1998 and 2012 (All 20 Industries)
     
Log(Nominal wage in 2012/ Nominal wage in 1998)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0605 *** 0.0665 *** 0.1824 *** 0.1833 ***

           

Log(U.S. nominal wage in 2012/
U.S. nominal wage in 1998)
in the industry outside the city

0.8128 *** 0.8140 *** 0.8246 *** 0.8251 ***
         

Nominal wage in the city-industry
in 1998 (in thousands)

-0.0006 *** -0.0006 *** -0.0006 *** -0.0006 ***
         

Employment in the city-industry
in 1998 (in millions)

0.8127 *** 0.7984 *** 0.6734 *** 0.6717 ***
         

Dummy variable indicating
presence in the South

0.0184 ** 0.0184 * 0.0195 ** 0.0195 **

         

Specialization indicator

0.0004
   
-0.0003
 
       

Competition indicator
 
-0.0040
 
-0.0006
 
       

Diversification indicator
   
-0.2000 *** -0.1996 ***

       
           

R Square

0.0973

0.0973

0.1004

0.1004
 

Adjusted R Square

0.0966

0.0967

0.0997

0.0995
 

Standard Error

0.3405

0.3305

0.3399

0.3300
 

Observation

6,719

6,719

6,719

6,719
                         
Note: *** p<.01 **p<.05 *p<.10
 
82



Table 2.3.2. City-Industry Real Wage Growth between 1998 and 2012 (All 20 Industries)
     
Log(Real wage in 2012/ Real wage in 1998)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0065

0.0128

0.1318 *** 0.1328 ***

           

Log(U.S. real wage in 2012/
U.S. real wage in 1998)
in the industry outside the city

0.8128 *** 0.8140 *** 0.8246 *** 0.8251 ***
         

Real wage in the city-industry in
1998 (in thousands)  

-0.0004 *** -0.0004 *** -0.0005 *** -0.0005 ***
         

Employment in the city-industry
in 1998 (in millions)

0.8127 *** 0.7984 *** 0.6734 *** 0.6717 ***
         

Dummy variable indicating
presence in the South

0.0184 ** 0.0184 ** 0.0195 *** 0.0195 **

         

Specialization indicator

0.0004
   
-0.0003
 
       

Competition indicator
 
-0.0040
 
-0.0006
 
       

Diversification indicator
   
-0.2000 *** -0.1996 ***

       
           

R Square

0.0973

0.0973

0.1004

0.1004
 

Adjusted R Square

0.0966

0.0967

0.0997

0.0995
 

Standard Error

0.3405

0.3405

0.3399

0.3300
 

Observation

6,719

6,719

6,719

6,719
                         
Note: *** p<.01 **p<.05 *p<.10
 
83



Table 2.4. City-Industry Nominal Wage Growth between 1998 and 2012 (Top 6 Industries)
     
Log(Nominal wage in 2012/ Nominal wage in 1998)
in the City-Industry  
 Variables  
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0106

-0.0457

0.0556

0.0206  

           

Log(U.S. nominal wage in 2012/
U.S. nominal wage in 1998)
in the industry outside the city

0.9596 *** 0.9355 *** 1.0788 *** 1.0103 ***
         

Nominal wage in the city-industry  
in 1998 (in thousands)

-0.0001 ** -0.0001

-0.0001 *** -0.0001 **
         

Employment in the city-industry
in 1998 (in millions)

0.2118 ** 0.2450 *** 0.1440

0.1601 *
         

Dummy variable indicating
presence in the South

0.0014

0.0056

0.0032

0.0076
 
         

Specialization indicator

0.0030
   
0.0038 **

       

Competition indicator
 
-0.0610 ***
 
0.0734 ***

       

Diversification indicator
   
-0.1474 *** -0.1908 ***

       
           

R Square

0.0560

0.0682

0.0592

0.0779
 

Adjusted R Square

0.0539

0.0662

0.0571

0.0750
 

Standard Error

0.1742

0.1731

0.1739

0.1722
 

Observation

2,290

2,290

2,290

2,290
                         
Note: *** p<.01 **p<.05 *p<.10

 
84



Table 2.5.1. Employment Growth and Dynamic Externalities by Industry between 1998 and 2012

Dynamic Externality Indicators  
 
NAICS Industries Specialization Competition Diversification  

   

Agriculture, forestry, fishing and hunting  +  
 Mining, quarrying, and oil and gas
extraction
+ +  
 Utilities -  -
 Construction  + -
 Manufacturing - +  
 Wholesale trade  +  
 Retail trade  + -
 Transportation and warehousing  +  
 Information  +  
 Finance and insurance  +  
 
Real estate and rental and leasing  + -
 Professional, scientific, and technical
services
+ -
 Management of companies and enterprises - +  
 Administrative and support and waste
management and remediation services
-  
 Educational services - + -
 Health care and social assistance  +  
 Arts, entertainment, and recreation - + -
 Accommodation and food services -  -
 Other services (except public
administration)
- + -
 Industries not classified - +  
 
     

85



Table 2.5.2. Real Wage Growth and Dynamic Externalities by Industry between 1998 and 2012

Dynamic Externality Indicators  
 
NAICS Industries Specialization Competition Diversification  

   

Agriculture, forestry, fishing and hunting  -  
 Mining, quarrying, and oil and gas
extraction
+ +  
 Utilities    
 Construction + - -
 Manufacturing  +  
 Wholesale trade + +  
 Retail trade    
 Transportation and warehousing  + -
 Information    
 Finance and insurance   -
 
Real estate and rental and leasing    
 Professional, scientific, and technical
services
- - -
 Management of companies and enterprises    
 Administrative and support and waste
management and remediation services
+  
 Educational services    
 Health care and social assistance  + -
 Arts, entertainment, and recreation   -
 Accommodation and food services  + -
 Other services (except public
administration)
+  
 Industries not classified  -  
 
     

86



CHAPTER THREE: DYNAMIC EXTERNALITIES AND REGIONAL GROWTH IN
SOUTH KOREA
Introduction
This paper estimates the impact of agglomeration externalities associated with knowledge
spillovers on regional growth in South Korea. The estimation uses the latest data available, from
2008 to 2013. The analysis results are expected to show how each externality performs in the
context of South Korea in recent years. The results will also provide comparative evidence on
whether the effect of externalities is different in this Asian country, compared to the
conventional findings from empirical studies on Western countries.  
Because the choice of measurement and methodology could be critical to the analysis
results, the basic framework of this study follows the work of Glaeser, Kallal, Scheinkman, and
Shleifer (1992), as did the earlier papers on economic growth in the U.S. In addition, this study
adopts several alternative externality measures. The same analysis will be conducted using
different externality measures. Glaeser, Kallal, Scheinkman, and Shleifer’s (1992) work is
seminal in the economic growth literature. They examined the effect of industrial structure on the
economic growth of U.S. cities. The externality measures they used have becoming a staple of
later studies. The analysis with their measures and alternative measures will allow for an
examination of whether different measures produce different results.
Knowledge spillovers are locally available externalities generated from agglom eration.
Like individuals, firms also benefit from geographic proximity to other firms. However, the
question remains as to what form of local industries is most effective in contributing to economic
growth. Marshall (1890) believes that firms learn from locally adjacent firms in only their own
industry. Agglomeration benefits that come from local specialization of industry are called
87



Marshall-Arrow-Romer (MAR) externalities. This type of externality prefers local monopoly.
Porter (1990) agrees with MAR that specialization helps growth, but Porter favors local
competition between firms. An alternative view is that firms learn from all other firms in a
region. In this vein, Jacobs (1969) argues that diversity of industries as well as local competition
induces growth.  
Empirical Studies in Korea
A large body of empirical literature on the relationship between agglomeration
externalities and economic growth has been developed in studies in the U.S. (Feldman &
Audretsch, 1999; Glaeser, Kallal, Scheinkman, & Shleifer, 1992; Henderson, Kuncoro, & Turner,
1995) and a few European countries (Cingano & Schivardi, 2004; Combes, 2000). Although
agglomeration economies have been generally found to contribute to growth, previous studies
have found mixed and often conflicting evidence on the role of each type of externality.  
Studies in Asian or other developing countries remain relatively scarce. Henderson, Lee,
and Lee (2001) examined how agglomeration externalities worked in South Korea between 1983
and 1993. In this period, South Korea was a fast-growing developing country, and industries
were being de-concentrated from the Seoul Metropolitan Area according to the national
development plan. The examination results were similar to traditional U.S. results. The study
found that Jacobs externalities are significant only for high-tech manufacturing, while MAR
externalities are significant for all other manufacturing industries.  
In Korea, only a small number of empirical studies on the subject have been conducted in
recent years. Most of the studies are built upon Glaeser et al. (1992) and examine industrial
structure and employment growth in South Korea in the 1990s and the 2000s. The choice of
88



externality indicators is similar to those of Glaeser et al. The location quotient is used as a
specialization indicator in most of the cases. The number of firms per employment is a popular
competition indicator, and the inverse Herfindahl index serves as a diversification indicator.
Despite investigating growth over similar periods, however, the findings are not consistent. Kim
and Min (2010) and Kim and Ko (2009) found that specialization helps growth. Lee (2014)
argues that MAR theory works for existing firms while Porter’s works for new firms in the
manufacturing industry. Cheon (2009) and Lee, Kim, and Hong (2005) agree with Jacobs; they
found that diversification and competition promote growth in South Korea. Yim and Kim (2003)
found that diversity is effective on growth only in high-tech manufacturing industry, whereas
competition contributes to the growth of various industries.  
Regional Industry in South Korea  
The economy of South Korea is known for its rapid growth, especially from the 1960s to
the 1990s. Having almost no natural resources and a small domestic market, South Korea has
pursued an export-oriented economy with manufacturing-centered industrialization. According
to the World Bank, the nominal GDP of South Korea ranked 30
th
largest in the world in 1960,
28
th
in 1980, and 12
th
in 2000.
Table 3.1.1 presents the national industry employment of South Korea in order of
employment size. It shows that manufacturing and wholesale and retail trade are still the biggest
industries in South Korea in terms of employment. Manufacturing accounts for 20.12 percent,
while wholesale and retail trade accounts for 15.62 percent, of the national employment in 2008.
Interestingly, as shown in table 3.1.2, the industry with the greatest growth between 2008 and
2013 is human health and social work activities.  In the U.S., manufacturing and retail trade were
traditionally the largest city industries, in terms of employment. However, since the late 1990s,
89



human health and social assistance became the largest industry in the largest cities. The next
industries that have shown the most growth in South Korea are membership organizations, repair
and other personal services; professional, scientific and technical activities; information and
communications; and business facilities management and business support services. These
industries take only about five percent of the national employment each, but they are growing
fast. On the other hand, the biggest industries, such as manufacturing, wholesale and retail trade,
accommodation and food service activities, and education have grown less than the national
average. These data suggest that the industrial structure of Korea is changing in a way that is
similar to that of the U.S.  
Table 3.2.1 shows regional employment and the three largest industries in the region.
Seoul is the capital and the largest region in South Korea in terms of employment. The second
largest is Gyeonggi, a metropolitan region that surrounds most of Seoul, geographically. The
table shows that manufacturing, wholesale and retail trade, and accommodation and food service
activities are the biggest industries in most of the region.  
Table 3.2.2 shows that the region with the most growth between 2008 and 2013 is
Chungnam. This is, however, considered a result of an unusual event, the creation of a new city
in Chungnam, rather than spontaneous growth. A special administrative district called Sejong
City opened in 2012, integrating parts of two regions, Chungnam and Chungbuk. All or at least a
part of forty central administrative agencies and fifteen government-funded research institutions
in the Seoul Metro Area were relocated to Sejong. This caused the migration of more than ten
thousand government employees and their families to Chungnam.    
90



Jeju is an island with a population of about half a million people. Tourism, agriculture,
and fishery have been major industries of the region. The warm weather and beautiful natural
environment of Jeju has recently attracted not only tourists, but also young migrants and firms.
Jeju has experienced a steep increase of in-migration since around 2010. In 2013, its net
migration rate was the second highest in the nation, behind only Sejong. Its regional employment
growth between 2008 and 2013 also ranked the second highest. Regional employment grew in all
industries, except Transportation, during this period in Jeju.    
Employment growth was highest in Gyeonggi. As a part of the Seoul Metropolitan Area,
Gyeonggi is a convenient and affordable location for both employees and firms. Small and large
firms in various industries are located in this region. Continuous development of several
industrial complexes contributes to regional employment growth.
Other big regions, on the other hand, such as Seoul and Busan, showed the lowest
employment growth.      
Characteristics of South Korean Industries  
Regional concentration. Some characteristics of South Korean industries need to be
highlighted in order to understand regional growth. First of all, the industries and population of
South Korea are heavily concentrated in the Seoul Metropolitan Area. This characteristic may
diminish the effect of industrial structure on regional growth. The Seoul Metropolitan Area
incorporates Seoul, the capital, and its two surrounding regions, Gyeonggi and Incheon. The
Korean government has indicated that the regional economic and population disparity due to the
concentration in this area is the first issue to be addressed in the national development
(Government of the Republic of Korea, 2001).  
91



The Seoul Metropolitan Area constitutes only 11.8 percent of the national area but
accounted for 49.4 percent of the total population in 2013. As shown in table 3.3, about half of
the national employment and establishments are located in this area. Moreover, central
management function and knowledge production is even more concentrated in the region. A total
of 69.4 percent of the central government agencies and 83.3 percent of the government
investment agencies are located in the Seoul Metropolitan Area. The headquarters of 95 of the
100 largest corporate firms in South Korea are also located in this area. Moreover, the region
housed 42.3 percent of all universities and 40.5 percent of enrolled university students in 1992
(Government of the Republic of Korea, 2001, p. 19). This regional concentration is a long-
standing circumstance but is believed to be the consequence of the modern government-led
national development policy.  
The central location in the Korean Peninsula, the adjacency to the sea, and the presence
of a large river throughout the city provide natural locational advantages to Seoul. It has been a
center of socio-cultural and political importance ever since prehistoric times. Since 1394, this
region has been the capital of the country, though it has only been called Seoul since 1947 (refer
to Seoul Metropolitan Government, 2009, for a brief history of Seoul).  
Korea suffered from Japanese colonial rule from 1910 to 1945, followed by the Korean
War from 1950 to 1953. The war left little social infrastructure. The postwar recovery in the
1950s was an attempt at development without a systematic plan, depending upon foreign aid.
The 1960s was a preparatory stage for planned development. The government sought
quantitative growth through modernization and industrialization in selected regions. The
investment was concentrated in the capital of Seoul and the Southeast coastal regions, which
have harbor for export. Those regions became urbanized and industrialized rapidly, while the rest
92



of the country remained barely developed. The economic boom in Seoul attracted many people
from rural areas. The population of Seoul was about two million in the early 1960s but increased
to five million by the late 1960s.  
The economic development of modern South Korea has been controlled by the
government with a comprehensive national development plan. To achieve efficiency in a short
time, South Korea adopted an unbalanced top-down growth strategy in the early phase of the
national development planning. This resulted in regional disparity, which still requires much
effort to alleviate (refer to Moon et al., 2013, for more information and discussion).  
The first Comprehensive National Territorial Plan (CNTP) was implemented in the 1970s.  
It continued the targeted growth strategy for Seoul and the Southeast coastal regions. Industrial
complexes and the first highway connecting these two regions were built. As employment
opportunities were concentrated in these regions, the population concentration there continued as
well. Seoul grew socially, econom ically, and geographically. The population of Seoul reached
seven million in the 1970s.  
In response to the severe regional disparity and diseconomies of urbanization in the
growth poles, the South Korean government changed the direction of its growth strategy. From
the 1980s, they sought regionally balanced growth. They initiated growth management of the
Seoul Metropolitan Area. The growth plans for less developed regions were mainly physical,
such as building facilities to foster local industries.  
The second CNTP in the 1980s was intended to disperse population and industries from
the growth poles. Nonetheless, along with rapid economic growth, the concentration in the
capital area continued. In the 1980s, the population of Seoul was about ten million. Thus, about
93



35 percent of the total population was in the Seoul Metropolitan Area. The third CNTP in the
1990s and the current fourth CNTP also aimed at regionally balanced development, by fostering
the economies of less developed regions and alleviating concentration in the capital area.  
In spite of the continuing pursuit of regionally balanced development, many policies to
induce industries to move out of the Seoul Metropolitan Area and to revitalize other regional
economies have been unsuccessful. The well-established economic and perceptual advantages of
being around the capital are difficult to substitute. In the two growth poles, the share of the
population grew from 43.8 percent in 1960 to 73.3 percent in 1995. The share of manufacturing
in those regions also grew, from 56 percent in 1960 to 80 percent in 1995. Agglomeration
economies in the growth poles have contributed to fast quantitative growth. However, such
regional concentration is now often perceived as a hindrance to nationwide growth (e.g. Hong,
1999; Cho, 2000; Jung, 2001, Kang, 2006).
Neither firms nor workers are easily relocated from the Seoul Metropolitan Area to other
regions. Kwon and Lee (1995) found that the spatial pattern in South Korea is one of intra-
regional migration rather than inter-regional migration. The Seoul Metropolitan Area has been
the only area with net in-migration since 1970. Choi (2004) also found that the concentration of
the population in the Seoul Metropolitan Area continues. In addition, the author found that the
characteristics of in-migrants to the Seoul Metropolitan Area differ according to their age. People
in their twenties account for the biggest share of in-migrants of the region. This suggests that a
major trigger for migration is disparity in educational opportunities. The author argues that it
further suggests selective migration of highly educated people to the Seoul Metropolitan Area,
resulting in a brain drain in other regions.    
94



The recent creation of Sejong City as a special administrative district was a part of the
policy for de-concentration from the Seoul Metropolitan Area and regionally balanced
development. The relocation of about two thirds of central government agencies to Sejong
resulted in the migration of a large number of government employees and rapid growth of the
region in a short period of time. There has been support for this policy in terms of economic
growth of a less developed region (e.g., Cho, 2010; Kang, 2006; Yuk, 2009). However, there
also has been a lot of criticism regarding the effectiveness of planned growth and the irrationality
of dividing the central government (e.g. Choi, 2009; Chun, 2010; Jang, 2003; Lim, 2011).
Large business groups. Another factor that should be noted is the major role of large
business groups. There is a general public perception that the economic power of South Korea is
greatly concentrated on large business groups.  
In South Korea, in the course of rapid economic growth through industrialization, a
number of leading companies formed large business groups. Those companies constituted the
main engine of the government-led economic growth plan from the 1960s to the 1980s. These
large business groups often gained monopolistic or oligopolistic status with great market power
during this period. Many of them are still dominant in various industries.      
What generates criticism about the large business groups of South Korea is their unique
organizational or management structure. Companies in many Western countries tend to grow
into multi-divisional conglomerates. The multi-divisional form, which is particularly popular in
the U.S., is an organizational structure by which a company comprises the parent company and
the smaller companies that make use of the brand and name of the parent company. The parent
company generally owns the smaller companies. Even though the whole organization is
95



ultimately controlled by the central management, each division (smaller company) is semi-
autonomous and responsible for its own operations. The multi-divisional form has been
considered an efficient way of diversifying with economies of scale.
On the other hand, South Korean conglomerates tend to prefer to grow into business
groups by adding affiliates. Large business groups exercise their influence upon an extensive
range of industries, while they are still largely controlled by their founding families through
direct and indirect means. Affiliates belonging to the same business group are usually tied to
each other in a complex chain of cross-equity holding. Circular equity investment among
affiliates enables the owners of large enterprise groups to take control of the entire group holding
a small share.  
The Korea Fair Trade Commission (KFTC) designates business groups with total assets
of at least five trillion Korean won (about 4.15 billion USD) of domestic affiliates belonging to
the same group as the large business groups. Those large business groups are subject to the
Monopoly Regulation and Fair Trade Act (MRFTA), which provides a system to prevent adverse
effects of the concentration of economic power and impediment of market competition by large
business groups. The substances of MRFTA include prohibition of new circular investment,
limitation of mutual investment, restriction of debt guarantee and holding-company system, and
publication of current status of groups. In 2013, according to the KFTC, 62 large business groups
in South Korea had a total of 1,746 affiliates. The internal shareholding ratio of those large
business groups has risen continuously. For the 43 large business groups that have chief
executives, the internal shareholding ratio was 54.79 percent in 2013. The ratio of chief
executives was 2.09 percent, and the ratios of their families and the affiliates were 4.36 percent
and 48.15 percent, respectively.    
96



According to the KFTC, the concentration of economic power in the largest firms is
continuing in South Korea. The KFTC provides annual concentration ratios of the three largest
firms (CR
3
) for the mining and manufacturing industries at the national level. Due to data
limitations, information on service sectors is not provided. The concentration ratio measures the
percentage share of output produced by a given number of firms in the industry. A low
concentration ratio would indicate greater competition among the firms in that industry, whereas
a high concentration ratio would indicate concentration in the largest firms and suggest the
possibility of an oligopoly or monopoly. The CR
3
of sub-classification (5-digit) industries in
mining and manufacturing has increased steadily since 2002. The simple average of CR
3
was
40.7 in 2002, 45.4 in 2008, and 45.2 in 2011. The average CR
3
weighted by industry share was
47.6 in 2002, 55.3 in 2008, and 56.1 in 2011.    
Within the Korean economy, as in most industrial countries, small and medium-sized
enterprises (SMEs) account for the vast majority of employment and establishments. SMEs also
account for approximately half the gross domestic product (GDP) generated by nonagricultural
sectors. The criteria for SMEs are set by law. They are defined by the number of employees,
sales, capital, and independence of ownership. The employment size criteria differ according to
industry. For example, the size criterion for manufacturing is 300 employees or less, while that
for wholesale and retail trade is 200 employees or less. Criteria for all other industries are found
in the data and variables section below.  
Table 3.4 presents SMEs’ employment and establishments by industry in South Korea.
According to the dataset constructed for this study, most SME employment and establishments
are found in manufacturing, wholesale and retail trade, and accommodation and food service
activities, as are national industry employment and establishments. In 2008, SMEs accounted for
97



86.80 percent of total employment and 99.88 percent of total establishments in South Korea. In
2013, the employment share of SMEs decreased to 86.11 percent. The establishment share also
decreased slightly, to 99.87 percent. As the concentration ratios do, table 3.4 also suggests that
the economic concentration in the large firms continues. In manufacturing, which is the largest
industry, SME employment share in total employment decreased between 2008 and 2013, while
SME establishment share increased. In the same period, SME employment share decreased in the
two next largest industries as well.      
In the U.S., according to the U.S. International Trade Commission (USITC), 26.8 million
SMEs (firms that employ fewer than 500 employees) made up 99.9 percent of all firms and
employed about 50 percent of private sector employees in 2006. SMEs accounted for 40 percent
of U.S. firm sales in 2002 and approximately half of U.S. non-farm GDP in 2004. Compared to
SMEs in South Korea, SMEs in the U.S account for a similarly high percentage of
establishments, but their share of employment is much smaller, although significant. Furthermore,
the U.S. SME employment is heavily concentrated in the service sectors, followed by
manufacturing.  
There is continuing concern about the large business groups in South Korea because their
influence on the national economy could be larger than the scale reported to the public. Many
SMEs are subcontractors of the large business groups. For this reason, the decisions of the
largest business groups often seem to affect legally unrelated SMEs and even regional economic
growth.  
98



SMEs play a vital role in the national economy. They not only make up an absolute
majority of the national businesses, but they are the primary source of jobs. Further, it is widely
believed that SMEs drive innovation and boost industry competition and dynamics.  
It is not known whether this role of SMEs in South Korea is being disrupted because of
the large business groups. It is also not clear whether the economic power of large business
groups in South Korea over regional economies is meaningfully different from the situation in
other countries. Still, this characteristic needs to be considered in interpreting the relationship
between industrial structure and regional growth in South Korea.  
Data and Variables
Construction of the Dataset
The dataset is collected from two sources. The Census on Establishments produced by the
Korean Statistical Information Service (KOSIS) includes annual employment and establishment
data. The Labor Cost of Enterprise Survey by the Ministry of Employment and Labor of Korea
(MOEL) provides annual wage data. Both sets of data are available by industry, region, and
employment size annually.  
Regarding industry classification, the highest-level industry sections, which roughly
correspond to the 2-digit NAICS of the U.S, are selected. These include a total of nineteen
industries, of which three are removed due to data restrictions. The Labor Cost Survey does not
collect wage data for agriculture and administration, and the wage data of mining and quarrying
has some zero values. The Korean Standard Industry Classification System (KSIC) has been
updated; the current system is the 9
th
version (KSIC 9), which was announced in 2008. The
Census on Establishments began to follow KSIC 9 since 2006 and the Labor Cost of Enterprise
Survey began to follow KSIC 9 since 2008. Because classification systems are not designed to be
99



convertible, only the data from 2008 to 2013, which both data follow KSIC 9, are used for this
study. The data thus cover a relatively short period of time (only five years), but they are the
most recent data that have not been used in previous studies. Also, this period matches the term
of Lee Myung-bak, the 10
th
president of South Korea. This means that the nation was under
relatively continuous government policies during the study period.  
The regions included in the data cover the whole country. Even though the geographical
unit is administrative, these metropolitan areas are economic units as well and are relevant to
serve our interest in cities. One region was created during the study period. Sejong City opened
in July 2012 as a special administrative district integrating parts of two regions, Chungnam and
Chungbuk. The Census on Establishments provides the employment and establishment data of
Sejong separately beginning in 2012. The wage data from the Labor Cost Survey, however,
continue to follow the regional classification from before 2012. The data of Sejong are thus
incorporated into the data of Chungnam, as most of the Sejong region was previously part of
Chungnam geographically. To construct a consistent dataset, the employment and establishment
data of Sejong are also added to that of Chungnam.  
Employment size is categorized into 5 classes: 5 to 9, 10 to 29, 30 to 99, 100 to 299, and
300 or more employees.  
The combined dataset contains employment, establishment, and wage data from 2008 to
2013 for 16 regions and 16 industries of South Korea.  
Variables and Descriptive Statistics
The first regression of this study follows Glaeser et al.’s (1992) choice of variables and
externality indicators to provide comparative results to those of the U.S. case of a similar period.
100



The specialization, competition, and diversification indicators used by Glaeser et al. to measure
dynamic externalities have been a staple of the following studies. Several studies on Korea have
also adopted those indicators. However, none of them use wage data. Beaudry and Schiffauerova
(2009) reviewed empirical studies about the relationship between knowledge externalities and
urban growth to show that measurement and methodological differences tend lead to different
results. They found that the levels of industrial and geographical aggregation and the choice of
performance measures are the main causes of inconsistent results. The second regression of this
study utilizes other popular externality indicators and tests their effects on the results.  
Economic growth of a region is measured by employment growth. The dependent
variable is the log of employment in the end year divided by employment in the beginning year,
as defined by Glaeser et al.  
employment growth = ln
𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑒𝑒 𝑒 𝑒𝑒 𝑦 𝑦 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑏 𝑒 𝑏 𝑖 𝑒𝑒 𝑖 𝑒 𝑏 𝑒𝑒 𝑦 𝑦
Knowledge externalities are believed to promote the economic growth of cities when
industries are specialized, competitive, or diversified. The specialization indicator utilizes the
location quotient, which is defined as the regional industry’s share of regional employment
relative to the national industry’s share of the total national employment.  
specialization (LQ) =
𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑦 𝑒𝑏 𝑖𝑒 𝑒 𝑒 𝑒 𝑒 𝑦 𝑒 𝑦 𝑒𝑏𝑖 𝑒 𝑒𝑦𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 ⁄
𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑒𝑦𝑒𝑖 𝑒 𝑒 𝑒 𝑒 𝑒 𝑦 𝑒 𝑒𝑦𝑒𝑖 𝑒 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 ⁄

This is a static but widely used measure to compare an area’s distribution of employment
by industry to a reference area’s employment distribution. When the indicator value is one, the
region and the nation have the same percentage of employment in that industry. When an
101



industry has a smaller representation in the region compared to the nation, the indicator value is
less than one. A value greater than one, on the other hand, means that the industry has a greater
representation in the region than in the nation, which can be interpreted to mean that the industry
is specialized in the region.  
The competition indicator used by Glaeser et al. is the number of firms per worker in the
regional industry relative to the national ratio of that industry.  
competition (G) =
𝑓 𝑖𝑦 𝑒 𝑖 𝑖 𝑒 𝑦 𝑒𝑏𝑖 𝑒 𝑒𝑦𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑖 𝑒 𝑦 𝑒𝑏𝑖 𝑒 𝑒𝑦𝑒 𝑖 𝑒 𝑒 𝑖𝑖𝑒𝑦 𝑒 ⁄
𝑓 𝑖𝑦 𝑒 𝑖 𝑖 𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑒𝑦𝑒𝑖 𝑒 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑖 𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑒𝑦𝑒𝑖 𝑒 𝑒 ⁄

This indicator measures competitiveness in an industry by comparing firms per worker in
the region to the national ratio. When firms in an industry are more concentrated in the region
compared to the nation in general, the indicator has a value greater than one. Such industry can
be interpreted as locally competitive. However, when the regional industry is composed of firms
that are smaller in size than their national average, the value can also be greater than one. These
two cases are hard to distinguish.  
Glaeser et al.’s competition measure is popular in Korean empirical studies as well.
However, as discussed earlier, the influence of large firms and the role of small and medium-
sized firms in Korea have been controversial in terms of competitiveness of firms. For this
reason, this study tests two additional measures related to the share of small and medium-sized
firms.  
competition (R)
=
𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑒𝑓 𝑒 ℎ 𝑒 𝑖 𝑒 𝑦𝑒𝑒 𝑦𝑒𝑒 𝑒 𝑒𝑒 𝑖 𝑖𝑒 − 𝑖 𝑖 𝑠 𝑒𝑒 𝑓 𝑖𝑦 𝑒 𝑖 𝑖 𝑒 𝑦 𝑒𝑏𝑖 𝑒 𝑒𝑦𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑖 𝑒 𝑦 𝑒𝑏𝑖 𝑒 𝑒𝑦𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒
102



competition (N)
=
𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑒𝑓 𝑒 ℎ 𝑒 𝑖 𝑒 𝑦𝑒𝑒 𝑦𝑒𝑒 𝑒 𝑒𝑒 𝑖 𝑖𝑒 − 𝑖 𝑖 𝑠 𝑒𝑒 𝑓 𝑖𝑦 𝑒 𝑖 𝑖 𝑒 𝑦 𝑒𝑏𝑖 𝑒 𝑒𝑦𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑖 𝑒 𝑦 𝑒𝑏𝑖 𝑒 𝑒𝑦𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑒𝑓 𝑒 ℎ 𝑒 𝑖 𝑒 𝑦𝑒𝑒 𝑦𝑒𝑒 𝑒 𝑒𝑒 𝑖 𝑖𝑒 − 𝑖 𝑖 𝑠 𝑒𝑒 𝑓 𝑖𝑦 𝑒 𝑖 𝑖 𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑒𝑦𝑒𝑖 𝑒 𝑒 𝑤 𝑒 𝑦 𝑤 𝑒𝑦𝑖 𝑖 𝑒 𝑖 𝑒𝑒 𝑖𝑖𝑒𝑦 𝑒 𝑖 𝑒 𝑒 ℎ 𝑒 𝑒𝑦𝑒𝑖 𝑒 𝑒
These indicators measure the share of small and medium-sized enterprise (SME)
employment in regional industry. One measures simple regional share, and the other measures
regional share divided by the national share.  
This study categorizes the employment of SMEs in each industry according to the
employment size criterion. The size criterion is 300 or fewer employees for manufacturing;
construction; transportation; information and communications; business facilities management
and business support services; professional, scientific and technical activities; and human health
and social work activities. The criterion is 200 or fewer employees for electricity, gas, steel and
water supply; wholesale and retail trade; accommodation and food service activities; financial
and insurance activities; and arts, sports and recreation-related services. It is 100 or fewer
employees for sewage, waste management, materials recovery and remediation activities;
Education; and membership organizations, repair and other personal services. Finally, the
criterion is only 50 or fewer employees for real estate activities and renting and leasing.
The dataset built for this study includes employment, establishment, and wage data by the
size of employment. The employment size classes are 1-4, 5-9, 10-19, 20-49, 50-99, 100-299,
300-499, 500-999, and 1,000 or more. Because data for 200 or fewer employees are unavailable,
the values up to the 100-299 class are summed in this case.  
Glaeser et al. tried to measure a variety of industries utilizing the large industry
employment in a city other than the industry in question. In a diverse city, even the largest
103



industries may not take a big share, but they may in a specialized city. Thus, they defined the
diversity of a region as the share of total city employment of a city’s other top five industries.
diversity (G) =
𝑦 𝑒𝑏 𝑖𝑒 𝑒 ′ 𝑖 𝑒𝑒 ℎ 𝑒 𝑦 𝑒 𝑒 𝑒 𝑓 𝑖 𝑓 𝑒 𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑦 𝑒𝑏 𝑖𝑒 𝑒
A low value of this diversity indicator may suggest that a region is diverse. However, the
number of the largest industries that makes up the major share of a region’s industry may differ
according to region, and thus the choice of five is rather arbitrary. A more commonly used
diversity measure is the Herfindahl index.      
The Herfindahl index (HI) is a widely accepted measure of market concentration, which
is defined as the sum of the squares of the market shares of the firms within the industry. When a
market is close to perfect competition, the HI value is close to zero. In contrast, a very high HI
value is indicative of a monopoly. In this study, the inverse form of the Herfindahl index (IH) is
calculated using the employment data of each industry in a region. If a few industries are major
employers of a region, the IH value would be very low. On the other hand, if workers in a region
are relatively evenly hired across all industries, the IH value would be high, indicating that the
industries of that region are diversified.  
diversity (IH) =
1
∑ �
𝑖 𝑒𝑒 𝑖𝑖 𝑒𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑦 𝑒𝑏 𝑖𝑒 𝑒 𝑒 𝑒 𝑒 𝑦 𝑒 𝑒𝑒 𝑒 𝑒 𝑒 𝑒 𝑒 𝑒𝑒 𝑒 𝑖 𝑒 𝑦 𝑒𝑏 𝑖𝑒 𝑒 �
2
𝑖 𝑖 𝑖 𝑖 𝑖 𝑖 𝑖 𝑖 𝑖 𝑖
Other independent variables included in the regression are the national employment
growth and the wage and employment in the initial year. The national employment growth is
expected to have a significant effect on regional growth. The effect of initial wage and initial
employment is inconclusive. There are arguments that firms move to low-wage areas while
104



workers move to high-wage areas. An area that is already big can be either attractive due to
economies of scale or unattractive due to diseconomies of scale.
The means and the standard deviations of all variables are listed in table 3.5. The means
of the externality indicators that are defined following Glaeser et al. are similar to those of the
U.S.
Regression Results
Glaeser et al.’s (1992) Externality Indicators
Table 3.6.1 presents the first set of regression results, which follows the choice of
variables and the definition of externality indicators from Glaeser et al. (1992). Equations 1 to 3
include one of each of the externality indicators respectively, while equation 4 includes all three
indicators. The estimated coefficients show consistent results across equations. The results
indicate that the national employment growth helped regional employment growth, whereas the
initial wage and diversification did the opposite, in South Korea between 2008 and 2013.  The
coefficients of specialization, competition, and initial employment of a regional industry do not
show statistical significance.  
Unlike the U.S. case, the result suggests that diversification does not help employment
growth in South Korea. This evidence contradicts Jacobs’s theory that diversification induces
growth. On the contrary, these findings indicate that regions grow when their largest industries
take up a large share of the total regional employment. Thus, in South Korea, having dominant
industries contributes to the growth of a region, rather than having many small industries. No
evidence for the impact of specialization or competition is found.    
105



Alternative Externality Indicators
Table 3.6.2 shows the second set of regression results, which test alternative externality
indicators. Equation 1 is the original form that includes Glaeser et al.’s three indicators.
Equations 2 to 4 include each of alternative competition or diversification indicators,
respectively, instead of Glaeser et al.’s. The substitution of indicators slightly increased the value
of R-square and adjusted R-square. The estimated coefficients and the statistical significance of
Glaeser et al.’s indicators and control variables are hardly affected.  
Glaeser et al.’s specialization and competition indicators, Specialization (G) and
Competition (G), are still statistically insignificant. Glaeser et al.’s diversification indicator,
Diversification (G), remains positive and statistically significant (except in equation 3) with
alternative indicators.
In equation 4 of table 3.6.2, the inverse Herfindahl index is used as an alternative
diversification index, Diversification (IH). Its coefficient is negative and highly significant. This
evidence suggests that diversification hurts regional employment growth in South Korea. It
appears that a region grows when a few industries are major employers and therefore the region
is less diversified. This confirms the estimation results of Diversification (G).  
The estimation results of alternative competition indicators, Competition (R) and
Competition (N), are interesting. In equations 2 and 3 of table 3.6.2, the coefficients of both are
statistically significant but have opposite signs. Both indicators measure the share of small and
medium-sized enterprise (SME) employment in regional industry to test the role of small and
large firms in South Korea. Competition (R) measures the simple regional share of SME
employment, and Competition (N) measures the regional share relative to the national share.
106



Competition (R) has a negative impact on regional growth, whereas Competition (N) has a
positive impact.    
Table 3.7 compares Competition (R) and Competition (N) by industry. The table shows
the two indicators’ industry averages across regions and their ranks. Industries are listed in order
of employment size. The listed numbers reveal a different picture of the two indicators.
Industries that rank high in Competition (R) tend to rank low in Competition (N).
The negative sign of the Competition (R) coefficient suggests that a high share of SME
employment in a region slows regional employment growth. Detailed investigation into the data
reveals that a high value of Competition (R) occurs because of the regions that have no large
firms in the industry at all. Some regional industries are totally composed of SMEs. For instance,
the value of Competition (R) is the highest on average for accommodation and food service
activities. Accommodation and food service activities is the third largest industry in South Korea,
and its share of SME employment in the total national employment is also the third largest, as
shown in table 3.4. The regional share of SME employment for the accommodation and food
service activities industry is high overall, and the industry’s regional share of SME employment
is actually 1.0000 in eight regions out of sixteen. In many regions, there is no large firm in the
industry. Likewise, the wholesale and retail trade industry, which is ranked second, is composed
solely of SMEs in four regions. On the contrary, there are very few such regions for industries
that ranked low for Competition (R). Thus, rather than suggesting that a high share of SME
employment harms regional growth, the findings appear to indicate that a lack of large firm
makes a region hard to grow.  
107



The average values of Competition (N) tend to be low for large industries, which have
not grown much (see table 3.1.2). The three industries ranked the highest for Competition (N)
also ranked high in national industry growth. The positive coefficient of Competition (N)
indicates that regional employment grows if the regional industry’s SME employment share is
greater than that of the nation.        
Corresponding evidence has been found in previous studies on South Korean industry
growth. Kim and Min (2010) argued that the share of small firms in a region, which corresponds
to Competition (R) in this study, has a negative impact on the employment growth of every
industry. However, they counted the employment of very small firms of 50 employees or less.
Lee et al. (2005) examined productivity growth in manufacturing. Their evidence suggests that
competition among small firms contributes to the growth of all manufacturing industries in South
Korea.
This study tested two alternative competition indicators along with Glaeser et al.’s,
considering the characteristics of Korean industries. Glaeser et al.’s competition indicator,
Competition (G), measures the number of firms per employees. Its coefficient has a positive sign
but lacks statistical significance. The two alternative competition indicators measure competition
incorporating the size of firms. The analysis results suggest that small firms are conducive to
regional growth in South Korea, but not in the absence of large firms.  
National Growth and Other Control Variables  
The estimated coefficients of control variables also show consistent results throughout
regressions. Regional employment grows when national employment grows as a whole. High
108



initial wage in a regional industry reduces employment growth.  The impact of initial
employment is statistically insignificant.  
The estimated coefficients of the national employment growth are not only positive and
statistically significant but also extremely large. A one percent increase in the national growth
would yield about a 0.97 percent increase in regional employment growth. The impact of
national growth was dominant in regional growth in South Korea between 2008 and 2013.  
This resembles the U.S. case between 1986 and 1998. As argued in the previous paper on
the employment growth of U.S cities, the influence of national growth on regional growth may
decrease along with industrial structure changes. At an early stage of industrialization, the
performance of the regional and national economy would be closely related. National growth
may lead to regional growth during vigorous industrialization. The U.S. industries have grown
and changed in structure. In the past, manufacturing was a major industry nationwide.
Employment in manufacturing was stable until the 1990s, but it has decreased considerably since
then. In its place, service and technology-related industries have grown. Health care and social
assistance is now the largest industry in the largest cities (see tables 3.5.1 and 3.5.2 of the first
paper).        
In the previous paper, the estimates of the national employment growth in the regressions
are about 0.97 and 0.77 for the periods between 1986 and 1998 and between 1998 and 2012,
respectively (see tables 1.3.1.1B and 1.3.1.1C of the first paper). The decrease of the estimate
over time suggests that the influence of national growth has been weakened, while regional
attributes, including industrial structure, may have become more important than before.    
109



The large size of the national growth coefficient in the Korean case could indicate that
the industrial structure of South Korea has not changed much from its government-led and
manufacturing-centered old structure. On the other hand, the five-year study period might be too
short to reveal regional effect on growth. There are studies about the timing of the impact of
externalities. Blien, Suedekum, and Wolf (2006) point out that, in economic growth research, it
is assumed that a historical pattern from several decades ago affects current growth. Several
empirical studies on static and dynamic externalities, however, have found that it is the current
and very recent economic structure that affects growth, rather than the historical environment
(e.g., Blien, Suedekum, & Wolf, 2006; Combes, Magnac, & Robin, 2004; Henderson, 1997).  
Contrary to the results of the U.S. case, the coefficients of initial wage in this study are
negative and statistically significant. In the previous paper, initial wage showed a positive effect
on the employment growth of U.S. cities. U.S. workers tend to move to high-wage areas.  Rather
than suggesting that Korean workers have the opposite tendency, the present result might be
indicative of geographical differences. South Korea is much smaller geographically than the U.S.,
and there are less than one hundred cities of all sizes. To get a meaningful economic unit close to
a metropolitan area, this study classified the whole area of the country into 16 regions. In
contrast, the number of regions used for the U.S. growth analysis is 383. Further, more than half
of the total population is concentrated in only a handful of regions in South Korea. The fact that
most of these largest regions, whose wage is the highest, have the least growth may have led to
the negative sign of initial wage variable.  
Additional regressions excluding the Seoul Metropolitan Area have been run to see the
impact of the regional concentration of industries. The test produced very similar results to those
110



of the all-region regressions. Regional concentration does not appear to affect the relation
between industrial structure and regional growth in South Korea.  
Conclusion
The theories of economic growth have emphasized the importance of agglomeration and
knowledge spillover. However, the source of externalities is a subject that is still under debate.
This study explores the effects of knowledge externalities on regional growth in South Korea
between 2008 and 2013. This study is intended to fill the gap in the empirical literature, which is
heavily weighted towards the examination of a few Western countries. The industry of South
Korea has developed rapidly since the 1960s. Though the government policy for development
almost controlled the formation of the early structure of regional industries, each region is now
striving to find an effective way to achieve economic growth. The investigation of the work of
knowledge externalities in recent years would be useful for setting up a development plan in
South Korea as well.    
The overall analysis results do not support any of the theories of MAR, Porter, or Jacobs.
No evidence regarding the impact of specialization has been detected. Contrary to Jacobs,
diversity does not induce growth in South Korea. Regarding competition, having many small and
medium-sized firms may help regional growth, but having no large firms may harm growth. The
evidence also indicates that the regional growth of South Korea is still under the significant
influence of the national growth. This could be because of the state of industrialization in South
Korea as well as the influence of government policy.  
It has been argued that the definition of externality measures tend to affect analysis
results regarding their impact (Beaudry & Schiffauerova, 2009; de Groot, Poot, & Smit, 2007;
111



Melo, Graham, & Noland, 2009). This study adopts three different measures for competition and
two measures for diversification. Holding other measures and methods the same, the use of
different externality measures does not change the result of this study.      
The empirical analysis result suggests that the impact of knowledge externalities on
regional growth in South Korea is counter to the conventional expectation. None of the three
types of externalities shows positive effects. Characteristics of South Korean industries could
have caused this result. Firms and employees prefer to be located around the capital and do not
move much between regions. The significant role of large firms could have weakened the impact
of externalities. The result could also have been impacted by the relatively short period of time
covered in the study. Data limitations made it difficult to build a consistent dataset for a decade-
long time interval to analyze. A follow-up study examining an extended period would contribute
to a more comprehensive understanding.
 
 
112



Table 3.1.1. Industries in Korea, 2008 and 2013
  National Employment by Industry  
Industries  2008 2013  
     
Manufacturing             3,277,271          3,802,218  
Wholesale and retail trade             2,544,849           2,879,955  
Accommodation and food service activities             1,727,882           1,991,476  
Education             1,311,869           1,492,354  
Transportation               927,042           1,325,849  
Hum an health and social work activities               889,988           1,040,207  
Construction               872,821           1,014,030  

Mem bership organizations, repair and other
personal services
             795,813             943,283  
Professional, scientific and technical activities               689,741             936,830  
Financial and insurance activities               666,466              861,716  

Business facilities management and business
support services
             661,749              700,421  
Administration               575,148              644,981  
Real estate activities and renting and leasing               434,607              516,208  
Information and communications               420,129              466,719  
Arts, sports and recreation related services               314,394              360,621  
Electricity, gas, steel and water supply                 68,029                77,910  

Sewage, waste management, materials recovery
and remediation activities
               62,895                68,297  
Agriculture, forestry and fishery                 29,140                34,527  
Mining and quarrying                 18,447                15,872  
     
All Industries           16,288,280         19,173,474  
     

 
113



Table 3.1.2. Industry Growth in Korea, between 2008 and 2013
  National Industry Employment Growth  
Industries  
Log(2013
employment/2008
employment)
2013 employment –  
2008 employment

     
Hum an health and social work activities  0.3578 435,861  

Mem bership organizations, repair and other
personal services
0.3062 141,017  
Professional, scientific and technical activities  0.2569 171,975  
Information and communications  0.1721 96,079  

Business facilities management and business
support services
0.1700 281,534  
Agriculture, forestry and fishery  0.1696 5,387  
Construction  0.1560 167,386  
Transportation  0.1500 86,988  
Manufacturing  0.1486 524,947  
Accommodation and food service activities  0.1420 263,594  
Arts, sports and recreation related services  0.1372 46,227  

Sewage, waste management, materials recovery
and remediation activities
0.1356 15,015  
Education  0.1289 180,485  
Wholesale and retail trade  0.1237 335,106  
Administration  0.1146 69,833  
Real estate activities and renting and leasing  0.1052 32,112  
Electricity, gas, steel and water supply  0.0824 268  
Financial and insurance activities  0.0568 33,955  
Mining and quarrying  -0.1503 -2,575  
     
All Industries           0.1631  2,885,194  
     

 
114



Table 3.2.1. Regional Employment in Korea, 2008 and 2013
  Regional Employment



Region  2008 2013

Three Largest Regional Industries, 2008

       
Seoul   4,079,277   4,585,090  

Wholesale and retail trade, Accommodation
and food service activities, Professional,
scientific and technical activities

Gyeonggi   3,438,570   4,259,215  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Busan   1,165,574   1,297,862  
Wholesale and retail trade, Manufacturing,
Accommodation and food service activities

Gyeongnam   1,101,580   1,275,688  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Gyeongbuk   844,659   1,004,067  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Incheon   765,014   895,657  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Daegu   739,022   849,631  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Chungnam   641,731   834,710  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Jeonnam   535,252  623,407  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Jeonbuk   512,017            624,407  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Chungbuk   498,337   591,509  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Gangwon   466,538   551,182  
Accommodation and food service activities,
Wholesale and retail trade, Education

Gwangju   464,104            529,113  
Wholesale and retail trade, Manufacturing,
Accommodation and food service activities

Daejeon   450,857            536,181  
Wholesale and retail trade, Accommodation
and food service activities, Manufacturing

Ulsan   404,866            488,627  
Manufacturing, Wholesale and retail trade,
Accommodation and food service activities

Jeju   180,882            226,734  
Wholesale and retail trade, Accommodation
and food service activities, Education

     


 
115



Table 3.2.2. Regional Employment Growth in Korea, between 2008 and 2013
  Regional Employment Growth  
Region  
Log(2013 employment/
2008 employment)
2013 employment -
2008 employment

     

Chungnam

                             
0.2629  
                           
192,979  


Jeju

                             
0.2259  
                             
45,852  


Gyeonggi

                             
0.2140  
                           
820,645  


Jeonbuk

                             
0.1984  
                           
112,390  


Ulsan

                             
0.1880  
                             
83,761  


Daejeon

                             
0.1733  
                             
85,324  


Gyeongbuk

                             
0.1729  
                           
159,408  


Chungbuk

                             
0.1714  
                             
93,172  


Gangwon

                             
0.1667  
                             
84,644  


Incheon

                             
0.1577  
                           
130,643  


Jeonnam

                             
0.1531  
                             
88,549  


Gyeongnam

                             
0.1467  
                           
174,108  


Daegu

                             
0.1395  
                           
110,609  


Gwangju

                             
0.1311  
                             
65,009  


Seoul

                             
0.1169  
                           
505,813  


Busan

                             
0.1075  
                           
132,288  

     

 
116



Table 3.3. Concentration of Industries in Seoul Metropolitan Area in Korea, 2008 and 2013
 
Regional Employment
(Share of the National Employment)

Regional Establishment
(Share of the National
Establishment)

 
2008 2013  2008 2013

       


Seoul

4,079,277 4,585,090   719,687   785,094  



(25.04%) (23.91%)  (22.04%) (21.35%)


Incheon

765,014  895,657    157,980   177,990  



(4.70%) (4.67%)  (4.84%) (4.84%)


Gyeonggi

3,438,570  4,259,215    651,428   773,216  



(21.11%) (22.21%)  (19.95%) (21.03%)



     


Seoul Metro Total

8,282,861  9,739,962    1,529,095   1,736,300  

  (50.85%) (50.80%)  (46.84%) (47.22%)

       

 
117



Table 3.4. Small and Medium Sized Firm Employment and Establishment by Industry in Korea,
2008 and 2013
 
Small and Medium
Sized Firm Employment
(Share of the National
Employment)

Small and Medium
Sized Establishment
(Share of the National
Establishment)

 
2008 2013  2008 2013

 

   

Manufacturing

2,619,239   3,050,730    319,424   369,164  

 
(16.72%) (16.55%)  (9.83%) (10.11%)


Wholesale and retail trade

2,471,599   2,813,874    859,640   958,653  

 
(15.78%) (15.27%)  (26.46%) (26.24%)


Accommodation and food
service activities

1,708,020   1,964,114    623,878   684,521  

 
(10.90%) (10.66%)  (19.20%) (18.74%)


Education

1,052,134   1,123,637    158,958   171,831  

 
(6.72%) (6.10%)  (4.89%) (4.70%)


Transportation

847,932   926,860    340,387   370,938  

 
(5.41%) (5.03%)  (10.48%) (10.15%)


Construction

793,528   863,182    94,561   116,536  

 
(5.07%) (4.68%)  (2.91%) (3.19%)


Membership organizations, repair
and other personal services

770,655   900,981    364,532   397,526  

 
(4.92%) (4.89%)  (11.22%) (10.88%)


Hum an health and social work
activities

723,327   1,118,613    93,855   125,231  

 
(4.62%) (6.07%)  (2.89%) (3.43%)


Financial and insurance activities

576,281   603,791    37,355   41,270  

 
(3.68%) (3.28%)  (1.15%) (1.13%)


Professional, scientific and
technical activities

521,444   632,150    66,277   87,294  

 
(3.33%) (3.43%)  (2.04%) (2.39%)


Business facilities management
and business support services

401,457   597,727    30,853   45,791  

 
(2.56%) (3.24%)  (0.95%) (1.25%)


Real estate activities and renting
and leasing

376,252   411,286    124,612   131,720  

 
(2.40%) (2.23%)  (3.84%) (3.61%)


Information and communications

339,988   409,743    23,578   35,335  

 
(2.17%) (2.22%)  (0.73%) (0.97%)


Arts, sports and recreation
related services

292,178   337,518    100,860   103,741  

 
(1.87%) (1.83%)  (3.10%) (2.84%)


Sewage, waste management,
materials recovery and
remediation activities

56,563   71,249    4,848   6,900  

 
(0.36%) (0.39%)  (0.15%) (0.19%)


Electricity, gas, steel and water
supply

47,420   47,560    1,386   1,632  

 
(0.30%) (0.26%)  (0.04%) (0.04%)

       

Total

13,598,017 15,873,015  3,245,004   3,648,083

 
(86.80%) (86.11%)  (99.88%) (99.87%)

       
118



Table 3.5. Variable Means and Standard Deviations, Korea
 Variable Mean
Standard
Deviation
Number of
Observations
 
     

Employment growth, 2008-2013
   


Log(employment in 2013/employment in
2008) in the regional industry
0.1706 0.1462                256

     

National employment growth, 2008-2013
   


Log(employment in 2013/employment in
2008) in the national industry
0.1643 0.0990 16


Wage in the regional industry in 2008 (in
millions)
2.5249 0.7932                256


Employment in the regional industry in
2008 (in thousands)
         
61.1935
           
105.373 1
                 
256  
     

Specialization (G), 2008  
 




Regional industry's share of regional
employment relative to the national
industry's share of the total national
employment  
0.9800 0.3552                256


Competition (G), 2008
   


Establishments per employee in the
regional industry relative to establishments
per employee in the national industry  
1.1336 0.3024 256  

Competition (R), 2008
 

 
Small and medium sized firm employment
share in the regional industry
0.8973 0.1142 256

Competition (N), 2008
 



Small and medium sized firm employment
share in the regional industry relative to the
national industry share
1.0628 0.1363 256


Diversification (G), 2008    



Other top five industries' employment share
of the regional employment
0.6070 0.0709 256                  

Diversification (IH), 2008
 


The inverse Herfindahl index
9.1064 1.9984 16


Specialization (G), 2013  0.9855 0.3602 256


Competition (G), 2013 1.1228 0.2866
               
256  
Competition (R), 2013 0.8958 0.1228 256  

Competition (N), 2013 1.0659 0.1446 256


Diversification (G), 2013 0.6060 0.0679 256

Diversification (IH), 2013 9.1709 1.9739 16  
     
             
119



Table 3.6.1. Regional Industry Employment Growth in South Korea between 2008 and 2013
 
Dependent Variable  
 
Log(Employment in 2013/Employment in 2008)
in the Regional Industry
 
 

 
(1)   (2)   (3)   (4)  
 
       
 


Intercept

0.0926
***
0.0575

-0.0246

-0.0801




 






 

Log(Employment in
2013/Employment in 2008)
in the National Industry

0.9805 *** 0.9696 *** 0.9718 *** 0.9622 ***

 








Wage in the regional
industry (in millions), 2008

-0.0279 *** -0.0280 *** -0.0287 *** -0.0290 ***

 



 


Employment in the regional
industry (in thousands), 08

-0.0001  0.0000

0.0000

0.0000


 

   



       


Specialization(G)  

-0.0055  
   
0.0176

 
Regional industry's share
of regional employment
relative to the national
industry's share of the
total national
employment, 2008          

       


Competition(G)  
 
0.0271
 
0.0314



Establishments per
employee in the regional
industry relative to
establishments per
employee in the national
industry, 2008
       


Diversification(G)  
   
0.1842 * 0.1913 *


Other top five industries'
employment share of the
regional employment, 08
       



         

R Square

0.5465

0.5491

0.5523

0.5553



Adjusted R Square

0.5392

0.5491

0.5451

0.5445



Standard Error

0.0999

0.0996

0.0992

0.0993



Observation

256

256

256

256


                       
Note: *** p<.01 **p<.05 *p<.10
120



Table 3.6.2. Regional Industry Employment Growth in South Korea between 2008 and 2013
 
Dependent Variable  
 
Log(Employment in 2013/Employment in 2008)
in the Regional Industry  
 

 (1)   (2)   (3)   (4)    
       
 


Intercept

-0.0801  0.1397  -0.1556

0.1081
**



         

Log(Employment in
2013/Employment in 2008)
in the National Industry

0.9622 *** 0.9066 *** 0.9451 *** 0.9706 ***

 








Wage in the regional industry
(in millions), 2008

-0.0290 *** -0.0389 *** -0.0292 *** -0.0287 ***

 

   


Employment in the regional
industry (in thousands), 08

0.0000  0.0000

0.0000

0.0000


       


         


Specialization(G)  

0.0176  -0.0104

0.0246

0.0208



Regional industry's share
of regional employment
relative to the national
industry's share of the total
national employment,2008          

Competition(G)  

0.0314
   
0.0378



Establishments per
employee in the regional
industry relative to
establishments per
employee in the national
industry, 2008
       
 Competition(R)     -0.1676 **      
 
Small and medium sized
firm employment share in
the regional industry, 2008          

Competition(N)       0.1191 **    
 
Small and medium sized
firm employment share in
the regional industry
relative to the national
industry share, 2008          


Diversification(G)  

0.1913 * 0.2381 ** 0.1596

 


Other top five industries'
employment share in
regional employment,2008
       

Diversification(IH)  
      -0.0089 ***

 
The inverse Herfindahl
index, 2008          



       
 

R Square

0.5553

0.5641

0.5604

0.5633
 

Adjusted R Square

0.5445

0.5536

0.5498

0.5528
 

Standard Error

0.0993

0.0983

0.0987

0.0984
 

Observation

256

256

256

256
 
                       
Note: *** p<.01 **p<.05 *p<.10
121



Table 3.7. Competition(R) and Competition(N) indicator by Industry in Korea, 2008  
  Competition(R)  Competition(N)  
Industries  Average Rank  Average Rank  
       
Manufacturing  0.7969 14

0.9976 16  
Wholesale and retail trade  0.9869 2

1.0161 11  

Accommodation and food service
activities
0.9931 1

1.0046 14  
Education  0.8214 13

1.0241 10  
Transportation  0.9445 6

1.0327 9  

Human health and social work
activities
0.8509 12

1.0469 8  
Construction  0.9520 5

1.0472 7  

Membership organizations, repair
and other personal services
0.9753 3

1.0072 13  

Professional, scientific and
technical activities
0.8883 11

1.1750 2  
Financial and insurance activities  0.9583 4

1.1083 4  

Business facilities management
and business support services
0.7746 15

1.2768 1  

Real estate activities and renting
and leasing
0.9082 9

1.0491 6  
Information and communications  0.9170 8

1.1331 3  

Arts, sports and recreation related
services
0.9435 7

1.0152 12  

Electricity, gas, steel and water
supply
0.7433 16

1.0663 5  

Sewage, waste management,
materials recovery and
remediation activities
0.9033 10

1.0045 15  
       
All Industries  0.8973  

1.0628  
       

 
122



REFERENCES

Acs, Z. & Armington, C. (2004). Employment growth and entrepreneurial activity in cities.
Regional Studies, 38, 911–927.
Almeida, R. (2007). Local economic structure and growth. Spatial Economic Analysis, 2(1), 65-
90.
Arrow, K. (1962). The economic implications of learning by doing. Review of Economic Studies,
29, 155–172.
Baptista, R. & Swann, P. (1999). A comparison of clustering dynamics in the US and UK
computer industries. Journal of Evolutionary Economics, 9, 373–399.
Barro, R. J. and Sala-i-Martin, X. (1991). Convergence across States and Regions. Brookings
Papers on Economic Activity 1: 107-182.
Baumol, W. J. (1986). Productivity growth, convergence, and welfare: what the long-run data
show. The American Economic Review, 1072-1085.
Beaudry, C. & Schiffauerova, A. (2009). Who’s right, Marshall or Jacobs? The localization
versus urbanization debate. Research Policy, 38, 318-337.
Blien, U, Suedekum, J., & Wolf, K. (2006). Local employment growth in West Germany: a
dynamic panel approach. IZA Discussion Papers, No. 1723.  
Carlino, G. A. and Mills, L. (1993). Are U.S. Regional Incomes Converging? A Time Series
Analysis. Journal of Monetary Economics 32, 335-346.
123



Cheon, B. Y. (2009). 도시의 산업 특성과 고용 성과 [Industrial structure of cities and
employment growth]. Quarterly Journal of Labor Policy, 9, 29-52.  
Cho, E. S. (2000). 수도권 집중화에 따른 지역격차 문제와 해소방안 [The study on the
problems and solution of regional disparity due to concentration in the capital area].
Local Government Studies, 8, 185-216.
Cho, M. R. (2010). 행정중심복합도시 촉진방안 [Strategies for administrative multifunctional
City]. The Chungnam Review, 52, 8-26.
Choi, E. Y. (2004). 선택적 인구이동과 공간적 불평등의 심화: 수도권을 중심으로
[Migration selectivity and growing spatial inequality: in case of the Seoul Metropolitan
Areas]. Journal of the Korean Urban Geographical Society, 7, 57-69.  
Choi, M. J. (2009). 행정중심복합도시( 세종시), 어떻게 할 것인가 [Comments on Sejong
city].The Korean Association for Public Administration Conference Discussion Paper,
69-71.
Chun, Y. P. (2010). 국정운영의 측면에서 본 세종시 ‘ 정부 분할 이전’ 의 문제점 [Critical
review on Sejong city project in the context of administrative governance system]. The
Journal of Social Science, 17, 75-95.  
Cingano, F. & Schivardi, F. (2004). Identifying the sources of local productivity growth. Journal
of the European Economic Association, 2, 720–742.
124



Combes, P. (2000). Economic structure and local growth: France, 1984-1993. Journal of Urban
Economics, 47, 329-355.  
Combes, P., Magnac, T., & Robin, J. (2004). The dynamics of local employment in France.
Journal of Urban Economics, 56, 217-243.
De Long, J. B. (1988). Productivity growth, convergence, and welfare: comment. The American
Economic Review, 78(5), 1138-1154.
De Long, J. B., & Shleifer, A. (1993). Princes and merchants: European city growth before the
industrial revolution. National Bureau of Economic Research, No. w4274.
De Groot, H. L., Poot, J., & Smit, M. J. (2007). Agglomeration, innovation and regional
development: Theoretical perspectives and Meta-analysis. Tinbergen Institute Discussion
Paper, No. 07-079/3.
Dekle, R. (2002). Industrial concentration and regional growth: Evidence from the prefectures.
Review of Economics and Statistics, 84, 310–315.
Duranton, G. and Puga, D. (2001). Nursery cities: urban diversity, process innovation, and the
life cycle of products. The American Economic Review, 91(5), 1454-1477.
Feldman, M. & Audretsch, D. (1999). Innovation in cities: science-based diversity, specialization
and localized competition. European Economic Review, 43, 409–429.  
Glaeser, E. L., Kallal, H. D., Scheinkman, J. A., & Shleifer, A. (1992). Growth in cities. Journal
of Political Economy, 100, 1126-1152.
125



Glaeser, E. L., Scheinkman, J., & Shleifer, A. (1995). Economic growth in a cross-section of
cities. Journal of monetary economics, 36(1), 117-143.
Government of the Republic of Korea. (2001). The Fourth Comprehensive National Territorial
Plan in Korea (2000-2020). Ministry of Construction and Transportation and Korea
Research Institute for Human Settlements.
Harrison, B., Kelley, M. R., & Gant, J. (1996). Specialization versus diversity in local economies:
the implications for innovative private-sector behaviour. A Journal of Policy
Development and Research, 2, 61–93.
Henderson, V. (1986). Efficiency of resource usage and city size. Journal of Urban Economics,
19, 47-70.
Henderson, V. (2003a). Marshall’s scale economies. Journal of Urban Economics, 53, 1-28.
Henderson, V. (2003b). The urbanization process and economic growth: the so-what question.
Journal of Economic Growth, 8, 47-71.
Henderson, V., Kuncoro, A., & Turner, M. (1995). Industrial development in cities. Journal of
Political Economy, 103, 1067-1090.
Henderson, V., Lee, T., & Lee, Y. J., (2001). Scale externalities in Korea. Journal of Urban
Economics, 49, 479-504.
Hong, J. H. (1999). 90 년대 우리나라 지역격차의 실태분석 [Regional disparity in South
Korea in the 1990s]. Korean Public Administration Research, 8, 48-78.
Jacobs, J. (1969). The Economy of Cities. New York: Random House.
126



Jaffe, A. B., Trajtenberg, M., & Henderson, R. (1993). Geographic localization of knowledge
spillovers as evidenced by patent citations. The Quarterly Journal of Economics, 108,
577-598.
Jang, S. H. (2003). 수도권 문제, 집중과 분산의 동화- 행정수도 건설 문제를 중심으로
[Dynamics of centralization and decentralization of Seoul Metropolitan Area]. Economy
& Society, 12, 40-66.  
Jung, W. S. (2001). 지방자치시대의 도시간 지역격차의 실태와 영향요인 분석 [The
empirical analysis of factors to the interurban disparity in the age of local government].
Journal of Korean Local Government Studies, 13, 141-160.  
Kang, H. S. (2006). 수도권 계획적 관리의 개념과 필요성 (2) [Planned management for Seoul
Metropolitan Areas: concept and need]. Urban Information Service, 296, 35-47.  
Kim, K.-S. & Ko, S.-C. (2009). 직접경제가 지역 고용성장에 미친 영향 [The effects of
agglomeration economy on regional employment growth]. Journal of the Korean
Planning Association, 44, 43-59.  
Kim, K. S. & Min, I. S. (2010). 직접경제가 지역- 산업 고용성장에 미친 영향: System GMM
추정방법의 활용 [The effect of agglomeration economy on the growth of local-industry
employment: using system GMM estimators]. Journal of the Korean Planning
Association, 45, 227-246.  
127



Kwon, Y. W. and Lee, J. W. (1995). 수도권 인구이동의 공간적 특성에 관한 연구 [Spatial
patterns of migration in the Seoul Metropolitan Area]. Journal of Korea Planning
Association, 30, 21-39.  
Lee, B. S., Kim, S., & Hong, S. H. (2005). Sectoral manufacturing productivity growth in
Korean regions. Urban Studies, 42, 1201-1219.
Lee, S. H. (2014). 공간패널 모형을 이용한 산업집적의 고용효과 분석 [Agglomeration and
local employment growth: A spatial panel approach]. Korean Journal of Labor Studies,
20, 107-148.
Lim, H. B. (2011). 행정중심복합도시 건설에 대한 비판적 고찰과 제언 [Critical thoughts on
special administrative multifunctional city]. The Journal of Korean Policy Studies, 11,
235-257.
Lucas, R.E. (1988). On the mechanics of economic development. Journal of Monetary
Economics, 22, 3-42.
Malpezzi, S., Seah, K-Y., & Shilling, J. D. (2004). Is it what we do or how we do it? New
evidence on agglomeration economies and metropolitan growth. Real Estate Economics,
32, 265-295.
Marshall, A. (1890). Principles of Economics. London: MacMillan.
Melo, P. C., Graham, D. J., & Noland, R. B. (2009). A meta-analysis of estimates of urban
agglomeration economies. Regional Science and Urban Economics, 39, 332-342.
128



Mody, A. & Wang, F-Y. (1997). Explaining industrial growth in coastal China: Economic
reforms … and what else? World Bank Economic Review, 11, 293-325.  
Moomaw, R. L. (1988). Agglomeration Economies: Localization or Urbanization? Urban
Studies, 25, 150-161.
Moon, J. H. et al. (2013). National Territorial and Regional Development Policy: Focusing on
Comprehensive National Territorial Plan. Supervised by Ministry of Land, Infrastructure
and Transport, Republic of Korea, Prepared by Korea Research Institute for Human
Settlement (KRIHS). Government Publications Registration Number 11-7003625-
000071-01.  
Moretti, E. (2010). Local multipliers. The American Economic Review, 100(2), 373-377.
Nakamura, R. (1985). Agglomeration economies in urban manufacturing industries: a case of
Japanese cities. Journal of Urban Economics, 17, 108–124.
Paci, R. & Usai, S. (2000). The role of specialisation and diversity externalities in the
agglomeration of innovative activities. Rivista Italiana Degli Economisti, 2, 237-268.  
Polèse, M. (2005). Cities and national economic growth: A reappraisal. Urban Studies, 42, 1429-
1451.
Porter, M. (1990). The Competitive Advantage of Nations. London: Macmillan.
Porter, M. (2003). The economic performance of regions. Regional studies, 37, 545-546.  
Romer, P. M. (1986). Increasing returns and long-run growth. Journal of Political Economy, 94,
1002–1037.
129



Romer, P. M. (1994). The origins of endogenous growth. The Journal of Economic Perspectives,
8, 3-22.  
Seoul Metropolitan Government (2009). Urban Planning of Seoul. Seoul Metropolitan
Government.
Solow, R. M. (1956). A contribution to the theory of economic growth. The Quarterly Journal of
Economics, 70, 65-94.
Swan, T. W. (1956). Economic growth and capital accumulation. Economic Record, 32, 334-361.
van der Panne, G. (2004). Agglomeration externalities: Marshall versus Jacobs. Journal of
Evolutionary Economics, 14, 593–604.
Yim, C. H. & Kim, J. S. (2003). 산업집적의 외부효과가 도시경제성장에 미치는 영향
[Impact of dynamic externalities on urban economic growth]. Journal of the Korean
Planning Association, 38, 187-201.  
Yuk, D. I. (2009). 행정중심복합도시는 왜 계속 추진되어야 하는가 [The reason why special
administrative multifunctional city should be carry forward]. The Korean Association for
Public Administration Conference Paper No.15, 1-23. 
Abstract (if available)
Abstract The first paper in Chapter one examines the effects of three types of dynamic externalities on employment growth in U.S. cities. By constructing an extensive dataset for city industries between 1986 and 2012, this study tests the theories of economic growth proposed by Marshall, Porter, and Jacobs. The analysis result finds that competition and diversification encourage employment growth in cites, whereas specialization does not. The evidence supports Jacob’s theory that knowledge spillovers contribute to the economic growth of cities through the exchange of ideas between different industries. The result further draws attention to the importance of nationwide economic performance in investigating the growth of cities. ❧ The second chapter is a continuing investigation into the effect of three types of dynamic externalities on U.S. city growth that is in parallel with the first paper. The theories of economic growth by Marshall, Porter, and Jacobs are tested through wage-based regressions. The analysis results suggest that diversity has positive effects on the growth of city-industry wages. Specialization is found to have an effect on the wage growth of the largest city-industries. Unlike in the employment growth analysis, no evidence for competition is found in the wage growth analysis. The additional regression for individual industry also shows interesting findings. ❧ Chapter three examines how dynamic externalities have affected recent regional growth in South Korea. The effect of industrial structure on local employment growth between 2008 and 2013 is tested with various measures for externalities. The analysis results do not confirm any of the theories of Marshall, Porter, or Jacobs. This finding suggests that different characteristics of industries may affect the role of dynamic externalities. The evidence indicates that diversity of industries impedes employment growth in South Korea. Regarding competition, regions grow when the share of small- and medium-sized firms in regional industry is high. However, at the same time, a lack of large firm makes a region hard to grow. No evidence is found regarding the effect of specialization. It is also found that the influence of nationwide growth is dominant in determining regional growth in South Korea. 
Linked assets
University of Southern California Dissertations and Theses
doctype icon
University of Southern California Dissertations and Theses 
Conceptually similar
Essays on congestion, agglomeration, and urban spatial structure
PDF
Essays on congestion, agglomeration, and urban spatial structure 
Essays on the economics of cities
PDF
Essays on the economics of cities 
The flexible workplace: regional tendencies and daily travel implications
PDF
The flexible workplace: regional tendencies and daily travel implications 
Growth, trade and structural change in low income industrializing economies
PDF
Growth, trade and structural change in low income industrializing economies 
Urban spatial structure, commuting, and growth in U.S. metropolitan areas
PDF
Urban spatial structure, commuting, and growth in U.S. metropolitan areas 
Essays on fiscal outcomes of cities in California
PDF
Essays on fiscal outcomes of cities in California 
The demand for reliable travel: evidence from Los Angeles, and implications for public transit policy
PDF
The demand for reliable travel: evidence from Los Angeles, and implications for public transit policy 
Healthy mobility: untangling the relationships between the built environment, travel behavior, and environmental health
PDF
Healthy mobility: untangling the relationships between the built environment, travel behavior, and environmental health 
The role of leadership and collaboration as a catalyst for regional economic development: a grounded theory study
PDF
The role of leadership and collaboration as a catalyst for regional economic development: a grounded theory study 
Essays on family and labor economics
PDF
Essays on family and labor economics 
Environmental justice in real estate, public services, and policy
PDF
Environmental justice in real estate, public services, and policy 
An empirical analysis of the relationship between real estate investment and regional economy in China
PDF
An empirical analysis of the relationship between real estate investment and regional economy in China 
Location of warehouses and environmental justice: Three essays
PDF
Location of warehouses and environmental justice: Three essays 
Talent migration: does urban density matter?
PDF
Talent migration: does urban density matter? 
Governance networks: internal governance structure, boundary setting and effectiveness
PDF
Governance networks: internal governance structure, boundary setting and effectiveness 
Urban universities' campus expansion projects in the 21st century: a case study of the University of Southern Calfornia's "Village at USC" project and its potential economic and social impacts on...
PDF
Urban universities' campus expansion projects in the 21st century: a case study of the University of Southern Calfornia's "Village at USC" project and its potential economic and social impacts on... 
The impact of mobility and government rental subsidies on the welfare of households and affordability of markets
PDF
The impact of mobility and government rental subsidies on the welfare of households and affordability of markets 
Three essays on the identification and estimation of structural economic models
PDF
Three essays on the identification and estimation of structural economic models 
Curating gastronomy: restaurants and social media in the cultural economy
PDF
Curating gastronomy: restaurants and social media in the cultural economy 
The impact of minimum wage on labor market dynamics in Germany
PDF
The impact of minimum wage on labor market dynamics in Germany 
Asset Metadata
Creator Jun, Eunha (author) 
Core Title Three papers on dynamic externalities and regional growth 
School School of Policy, Planning and Development 
Degree Doctor of Philosophy 
Degree Program Policy, Planning, and Development 
Publication Date 09/23/2016 
Defense Date 07/13/2016 
Publisher University of Southern California (original), University of Southern California. Libraries (digital) 
Tag dynamic externalities,economic growth,industry structure,OAI-PMH Harvest,South Korea,U.S.,wage growth 
Format application/pdf (imt) 
Language English
Contributor Electronically uploaded by the author (provenance) 
Advisor Green, Richard (committee chair), Boarnet, Marlon (committee member), Moore, James (committee member) 
Creator Email eunhajun@gmail.com,eunhajun@usc.edu 
Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-304737
Unique identifier UC11280418 
Identifier etd-JunEunha-4795.pdf (filename),usctheses-c40-304737 (legacy record id) 
Legacy Identifier etd-JunEunha-4795.pdf 
Dmrecord 304737 
Document Type Dissertation 
Format application/pdf (imt) 
Rights Jun, Eunha 
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
Tags
dynamic externalities
economic growth
industry structure
U.S.
wage growth