University of Southern California Dissertations and Theses

Locating the need for financial education

(USC Thesis Other)

Locating the need for financial education

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

LOCATING THE NEED FOR FINANCIAL EDUCATION

by

David James Breeding

_____________________________________________________________________________

A Thesis Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(GEOGRAPHIC INFORMATION SCIENCE AND TECHNOLOGY)

JUNE 2014

Copyright 2014 David James Breeding
i

DEDICATION

For Ali the Adventurer, I'm ready to go.

ii

ACKNOWLEDGEMENTS
The author is indebted to a number of people whose insight and support have made this thesis
possible. First, my thesis committee at the University of Southern California, Dr. Robert Vos, Dr.
Daniel Warshawsky, and Dr. Katsuhiko Oda (committee chair) of the Spatial Sciences Institute,
each provided excellent insight and guidance during the formation of my nascent interest in
foreclosure trends and financial education. Also from the Spatial Sciences Institute, I would like
to thank Dr. Yao-Yi Chiang, Dr. Karen Kemp, and Dr. Jennifer Swift, for their enthusiasm and
academic challenges that sustained my interest in the spatial sciences over 3 years of study.
Credit also goes to the City of Malden’s GIS Department, Assessing Department, and Treasury
Department for their assistance extracting and providing the information required to provide
local significance to the research.
And finally, my wife, Alison; for being a sounding board for ideas at all hours, a cheerleader
during times of despair, and most importantly for encouraging me to never stop learning. Thank
you.

iii

TABLE OF CONTENTS
Dedication ....................................................................................................................................... i
Acknowledgements ..........................................................................................................................ii
Table of Contents ............................................................................................................................ iii
List of Figures .................................................................................................................................. v
List of Abbreviations ...................................................................................................................... vi
Abstract ......................................................................................................................................... 1

Chapter One: Introduction .............................................................................................................. 2
1.1 Research Questions and Study Extent .............................................................................. 3

Chapter Two: Literature Review ..................................................................................................... 5
2.1 Financial Education Background ....................................................................................... 5
2.2 Current Financial Landscape ............................................................................................. 8
2.3 A Response to the Financial Crisis of 2008 ...................................................................... 10
2.4 Researching and Identifying Possible Factors ................................................................. 11
2.4.1 Financial Distress Factors ...................................................................................... 12
2.4.2 Potential for Financial Hardship Factors ............................................................... 15
2.5 Financial Vulnerability Research and Multi-Criteria Evaluation ...................................... 20

Chapter Three: Methods ............................................................................................................... 25
3.1 Outline of Data ................................................................................................................ 25
3.1.1 American Community Survey Data ....................................................................... 26
3.1.2 2010 Census Data ................................................................................................. 30
3.1.3 City Data Processing .............................................................................................. 31
3.2 Outline of Methods ......................................................................................................... 34
3.2.1 Normalization and Standardization of Factors ..................................................... 35
3.2.2 Kernel Density Estimation ..................................................................................... 37
3.2.3 Evenly Weighted Multi-Criteria Evaluation .......................................................... 37

iv

Chapter Four: Results .................................................................................................................... 39
4.1 Foreclosure and Tax Delinquency Results ....................................................................... 39
4.2 Census Data Results......................................................................................................... 45
4.3 Evenly Weighted Multi-Criteria Evaluation (MCE) Output Review ................................. 53

Chapter Five: Discussion ............................................................................................................... 57
5.1 Evenly Weighted MCE Analysis Discussion and Application ........................................... 57
5.1.1 Foreclosure and Tax Delinquency Significance ..................................................... 58
5.1.2 Census Factor and MCE Output Discussion .......................................................... 60
5.2 Implications and Limitations ........................................................................................... 63
5.3 Spatial Significance and Next Steps ................................................................................. 64
5.4 Summary .......................................................................................................................... 66
References .................................................................................................................................. 67

v

LIST OF FIGURES
Figure 1 - Regional View of the Study Area 4

Figure 2 - Methodological Flowchart Diagram 34

Figure 3 - KDE of Foreclosure Occurrence with Source Foreclosure Point Data 41

Figure 4 - KDE of Tax Delinquency Occurrence with Source Delinquency Point Data 42

Figure 5 - Results of the Foreclosure KDE Processing 43

Figure 6 - Results of the Tax Delinquency KDE Processing 44

Figure 7 - Households that Spend Greater than 35% of their Income on Rent 46

Figure 8 - Disabled Individual between the ages of 18 and 65 47

Figure 9 - Number of Individuals with Two Mortgages and a Home Equity Loan 48

Figure 10 - Ethnicity Factor- Number of African Americans and Latino Individuals 49

Figure 11 - Individuals that have not attained a High School Degree 50

Figure 12 - Households with Incomes Below the Poverty Level for the previous 12 months 51

Figure 13 - Individuals that have not worked in the past 12 months 52

Figure 14 - Number of Housing Units Currently Vacant 53

Figure 15 - Evenly Weighted MCE Overlay Output and High Educational Need Areas 56

Figure 16 - The City of Malden 2014 Zoning Map 60

vi

LIST OF ABBREVIATIONS
ACS American Community Survey

FLEC Financial Literacy and Education Commission

HEC Homeownership Education and Counseling

HUD Housing and Urban Development

KDE Kernel Density Estimation

IDA Individual Development Accounts

MCE Multi-Criteria Evaluation

1

ABSTRACT
The goal of this thesis is to complete a raster-based site suitability analysis that identifies
varying levels of need for financial education in the City of Malden Massachusetts. The research
product is intended to help decision makers evaluate where to site financial education training
events or invest in support programs for the communities or neighborhoods in need. The thesis
begins by reviewing the current state of financial education and arguments for its application.
Individual factors and supporting evidence that seek to identify individuals in need of financial
education is organized. The data preparation steps and multi-criteria evaluation (MCE) spatial
methodology used to create the final research output is detailed. The MCE results in the City of
Malden indentify multiple regions with a high level of need for financial education. A review of
these high need areas coincident with city zoning and a variety of geographic features highlights
additional spatial relationships of interest. The author concludes the research by outlining how the
final output can support deciding on locations for financial education events.
2

Chapter One: Introduction
The World Financial Crisis of 2008 affected the economic stability of the American financial
market, government, communities, and individuals. Median family income levels during the 2007
to 2010 period dropped an average of 7.7% (Bricker 2012). By 2009, the US housing market had
lost 28% of its value from its peak in 2006, and approximately four million families lost their homes
to foreclosure (Angelides and Thomas 2011). In 2009 testimony before the House Financial
Services Committee, Treasury Secretary Timothy F. Geithner, describing efforts to stabilize the US
economy post collapse said, "we now know that millions of Americans were ... unable to evaluate
the risks associated with borrowing to support the purchase of a home, a car, or an education.
(Treasury Secretary Timothy F. Geithner Testimony 2009). Financial education surveys have
attempted to quantify the American deficit in financial literacy concepts, and found only 18% of
1,700 American's were able to calculate basic compound interest and 56% could not properly
differentiate the performance differences between stocks, bonds and mutual funds over-time
(Lusardi and Mitchelli 2007).
As the prevalence of employer pensions continues to decrease, the responsibility for an
individual's financial well-being is shifting from financial experts to each individual (Wiatrowski
2012). The average level of American financial literacy in 2009 is likely well below that of the
average financial expert, which puts at risk the financial well-being many Americans who are
unprepared to make well informed financial decisions (Lusardi and Tufano 2009). The US Treasury
Department's Financial Literacy and Education Commission (FLEC), created in 2003, were tasked
with developing a national financial education strategy. Their 2011 Implementation Report seeks
to continue the development of a wide variety of financial education resources and to expand
3

relationships with organizations that are focused on improving the financial literacy Americans
(Commission 2011). Research in the delivery of financial education has indicated that many
attendees prefer formal training sessions in a communal setting over internet based financial
education resources (Corporation 2004). Since there is a place for seminar like financial education
opportunities, the locations of these events becomes an important factor in delivering financial
education. Current efforts by U.S. Bank to place financial education events are primarily
dependent on the location their training partner's access to an available facility (Bank 2013). This
partner facility approach to locating financial training events does not account for where those
individuals in greatest need are located and might exclude segments of these populations if
training locations are not easily accessible.
This research presents a spatial process that identifies the locations of populations with the
greatest relative need for financial education. This additional spatial information could help the
government, businesses, and non-profit agencies make informed decisions about how to locate
financial education events to reach those in most need of this kind of education.

1.1 Research Questions and Study Extent
There were two questions in this research. First, which factors are most appropriate for
identifying the need for financial education at a local scale? Second, what form of spatial analysis
is the most appropriate for identifying the need for financial education? The City of Malden in
Massachusetts is the spatial focus of this research and the areas in greatest need of financial
education are identified in this research.
4

The City Malden is located approximately six miles north of Boston (Figure 1). According to
the 2010 Census, Malden has an area of five square miles, a population of 60,374 and density of
11,780 people per square mile. According to 2010 census data, the City of Malden's non-white
ethnicities make up 47.5% of the City's population. All of Malden's neighboring cities and towns
have non-white ethnic populations below 31% and some as low as 10%. This makes Malden one of
the most ethnically diverse communities behind the City of Boston, which has a non-white
ethnicities making up 54% of the City. This City of Malden was interested in learning how they
could support their residents struggling after the Financial Crisis of 2008 and was supportive of this
research effort. Access to the local data, provided by the City GIS Department, City Assessors
Department, and City Treasurer allowed for a very granular perspective of the financial behaviors
and trends expressed by city residents to be examined.

Figure 1: Regional View of the Study Area
5

Chapter Two: Literature Review
The need for financial education is a complex phenomenon. In chapter two, research is presented
to both frame research efforts of the past and establish the current environment in which learning
will take place. The review begins with the presentation of financial education efficacy research
and the current state of financial education in America. Next, research outlining likely factors
indicative of the need for financial education is presented. Finally, spatial methodologies and
vulnerability studies are outlined to provide examples of research that could be adapted for this
research effort.

2.1 Financial Education Background
Financial education is not a new concept. In the 1950s and 1960s, states began mandating
the inclusion on personal finance, economics, and general consumer education into K-12
curriculum (Bernheim and Garrett 2003). Homeownership and mortgage counseling are also two
kinds of financial education that have existed for quite some time. These two kinds of educational
offerings are frequently provided by banks for interested new homebuyers. Retirement education
grew in earnest starting in 1990, when employers began preparing their employees for the
responsibility of investing for their own retirement (Martin 2007).
Even with these resources available today, making the "best" financial decision can still be
a difficult proposition. Financial experts use a number of different economic modeling techniques
and inputs specific to an individual's financial profile to make their "best" financial
recommendation. Is the average American today prepared to make the necessary financial
decisions, given the changing financial marketplace, without support? In an effort to assess the
6

financial literacy of Americans, many surveys have been conducted (Lusardi and Mitchelli 2007;
Lusardi 2008; Lusardi and Tufano 2009). The survey results clearly and repeatedly indicate a large
deficit in Americans' understanding of both basic and complex financial concepts across all
demographics.
Many experts believe that financial education is a viable approach to support individuals
that are unprepared to make informed financial decisions (Boshara et al. 2010; Lusardi and
Mitchelli 2007). Research into the efficacy of financial-education programs, primarily through
survey efforts, has documented a number of positive learning results. Homeownership education
and counseling (HEC) has been shown to lower default rates and limit other adverse financial
outcomes for HEC attendees (Martin 2007). Retirement education has been shown to increase the
amount of savings of attendees by 18%, and the impact was greatest on lower income attendees
(Lusardi 2004). Financial education has also been attributed to improvements in individual's
health, when participants are exposed to saving techniques like health savings accounts or flexible
spending accounts (Hastings, Madrian, and Skimmyhorn 2012). Financial education can teach
individuals to ask questions about basic daily financial decisions, increasing the prevalence of
comparison shopping habits. Finally, financial education allows individuals to increase their
financial stability and more financially stable individuals can make for more stable communities
(Harnish 2010).
To fill the gap in American's financial knowledge, there is a need for a large variety of
financial education options. Financial education today spans a wide set of topic areas, including
basic financial skills for young adults, homeownership and mortgages, basic saving and monetary
concepts, and retirement planning. National organizations have been collaborating to develop
financial education pedagogy and curriculum resources to aid in delivering impactful financial
7

education opportunities (Cordray 2013). These teaching methods have increasingly prescribed the
inclusion of survey techniques in training to better document financial impact. However, often
only anecdotal accounts of successes have been collected by researchers to date (Vitt 2005).
Recent financial education research efforts have been able to quantify the value of the just-in-time
paradigm for financial education (Fernandes, Lynch Jr, and Netemeyer 2014). Just-in-time
education is an approach that advocates for the delivery of education at a time when the student
has a current need for that particular educational subject. Their research attempted to quantified
the impact financial education has on the quality of financial decision making. Their research
results indicated financial education delivered more than 20 months prior to an individual’s actual
application the “learned” financial concepts, the final financial decision quality was no better
statistically than those who had received no financial education. Their research also indicated that
the quality of the financial decision-making was incrementally greater the closer the education
was delivered to the application of the newly acquired financial knowledge.
Since 2008, the availability of financial education resources focused on foreclosure and
mortgage refinancing has increased significantly. These resources are widely available online,
provided by a number of agencies including the Housing and Urban Development Department,
Federal Reserve Bank, The Home Ownership Foundation, Federal Housing Administration, Freddie
Mac's Mortgage Resource Center, and the National Foreclosure Mitigation Counseling Program.
These educational resources have been further adapted to meet specific state and local needs and
they are available through local agencies and groups.
However, the great number of institutions seeking to provide web guidance had an
unfortunate unintended consequence, fraudulent financial support websites. The increased
prevalence of quality homeowner support information gave rise to fallacious financial support
8

services seeking to take advantage of vulnerable people in search of financial guidance to more
confidently navigate the mortgage refinancing process (Leland 2009). Clear financial messaging
and guidance is essential for a steady recovery and for the health of the financial systems in the
future. Local community organizations and representatives are uniquely positioned to provide
targeted financial education to those community members in need, leveraging local relationships
and trust. By doing so, communities can financially strengthen and stabilize their populations.
Financial education is not an inoculation administered once to provide lifelong immunity
from poor financial decision-making. Financial education is a life-long learning process, requiring
personal commitment from individuals (Foundation 2013). Fortunately, there are tangible
economic and personal benefits to participants that decide to invest their time in financial
education.

2.2 Current Financial Landscape
The financial landscape has been changing for the average American and many of them are
unaware of the implications. The amount of revolving credit available to individuals has increased
steadily over-the past 30 years (Durkin 2000). This increased amount of readily available credit has
allowed consumers to finance a variety of purchases that might have been otherwise out of reach.
This economic freedom now makes it much easier for consumers to accumulate debts that are
beyond repayment or even servicing.
Individuals today are increasingly more responsible for managing their own retirement
investments. In 1985, 90% of Fortune 100 companies were offering defined benefit plans, which
paid a guaranteed amount during retirement. In 2013, only 30% of Fortune 100 companies are
9

offering defined benefit plans (McFarland 2013). This shift away from defined benefit plans has
been replaced by defined contribution plans, which do not provide a guaranteed benefits at
retirement and require individual to make the appropriate monthly contributions over-time (OECD
2006). In additional to selecting the correct level of monthly investment, the question of where
the money should be invested still remains. As more individual consumers enter the marketplace
looking to invest for their retirement, the variety and number of investment options individuals
have access to has also significantly increased. Consumers are required to evaluate investment
portfolios containing varying levels of diversification, risk, and return (Lusardi 2008). Survey's
targeted at assessing the readiness of Americans to make basic retirement investing decisions
have shown that 56% of respondents are unable to differentiate the performance of stocks,
mutual funds, and bonds overtime (Hilgert, Hogarth, and Beverly 2003).
The spending and saving habits of Americans over the past two decades have been
changing. Federal Reserve Economic Data reports show American personal saving rates have
dropped from 12.7% in November of 1980 to 2.9% in July of 2006 (Louis 2014). During this same
time period the ratio of debt to income over the same time period has risen from 3:5 to 1:1
(Dynan and Kohn 2007). This downward saving trend reversed dramatically after the financial crisis
of 2008, but Federal Reserve Bank data again shows the negative saving trend returned in 2011.
The major causes for this financial spending and saving shift were increasing housing prices and
financial innovation. Increasing housing prices required consumers to find more capital, which an
increasingly innovative financial industry was able to provide to consumers. With more money
invested in housing assets and consumers holding larger debts to service, consumer spending and
financial health is now more sensitive to market fluctuations (Dynan and Kohn 2007).
10

During financial fluctuations, the availability of credit can also help households weather
market downturns. However, this growing norm of indebtedness and necessity for credit puts
many Americans at risk of financial distress during market fluctuations. Households that lose their
ability to service their growing debt are at risk of great financial penalty and financial instability.
Between 2003 and 2007, many Americans entered into sub-prime mortgage contracts with rates
and conditions that were unsustainable. These sub-prime mortgage transactions were found to be
the core reason for the Financial Crisis of 2008 (Angelides and Thomas 2011).
The availability of American credit and the increasingly complex financial marketplace has
provided great opportunities to those knowledgeable enough to take advantage of these
resources and investment instruments. Unfortunately, there are also Americans who are
unprepared to navigate a more complex economic environment, but financial education can offer
these individuals guidance and knowledge, improving their relative ability engage in more complex
marketplaces.

2.3 A Response to the Financial Crisis of 2008
The Financial Crisis Inquiry Report of 2011 concluded that the financial collapse could have
been avoided but stewards of the financial system had ignored the warning signs of the impending
financial collapse (Angelides and Thomas 2011). Since the collapse, US government policymakers
have debated approaches to avoid another market collapse due to sub-prime mortgage lending.
One approach, developed by the Senate Banking Committee, has been proposed and was signed
into law on July 21, 2010. The Dodd-Frank Wall Street Reform and Consumer Protection Act
attempts to refine the way the mortgage sector of the banking industry functions, providing
11

greater consumer protections, and increasing the transparency of financial instrument for
consumers (Dodd-Frank Wall Street Reform and Consumer Protection Act 2010). Unfortunately,
legislation is only part of the solution; a fundamental problem remains largely unaddressed by the
Dodd-Frank legislation. The American population is lacking the basic financial literacy skills
required to make informed financial investment decisions (Commission 2012).
Surveys have shown that many Americans today are not equipped with the most basic
financial concepts, like compound interest and its implications in managing debt and making
mortgage decisions. This lack of skills severely impacts an individual's ability to make appropriate
financial decisions for their future well-being (Lusardi and Tufano 2009). Many financial experts,
educators, and researchers believe that financial education in its various forms can be a remedy
for this American deficit in financial literacy (Hilgert, Hogarth, and Beverly 2003; Greenspan 2005;
Morton 2005; Lusardi and Mitchelli 2007; Cordray 2013).

2.4 Researching and Identifying Possible Factors
This research effort seeks to identify those areas where individuals need financial
education. Financial education covers a variety of topics. Retirement planning education is focused
on providing targeted financial guidance in advance of a need. Basic financial literacy education is
more generic in focus seeking to serve a wide variety of attendees. Financial education can be
remedial, designed to support those individuals navigating forward from a difficult financial
situation, like foreclosure intervention efforts. Given the breadth of financial education topics, and
the audiences served, it is difficult to identify metrics that have a clear relationship with the need
for financial education. To address this problem, it was necessary to select additional logical proxy
12

criteria that would indicate a need for financial education. The proxy criteria included the
following two factors categories:
1. Factors that are indicative of current financial distress
2. Factors that are indicative of the potential for financial hardship
If individuals are currently in some form of financial distress, remedial forms of financial
education can improve financial decision-making abilities. By improving decision-making, these
groups in financial distress have a better chance at avoiding financial distress in the future. Each
financial decision made by individuals with the potential for financial hardship is also critical. Poor
financial decision making by these individuals can make satisfying existing financial obligations
impossible. When credit card, mortgage, loans or bills are left unsatisfied or paid, fees can be
levied, wages garnished, or property seized. All of these outcomes make returning to a financial
stability difficult. There is financial education designed to improve the basic financial decision
making ability of individuals. This education could reduce the likelihood of these groups
experiencing financial hardships. In the rest of this section, factors that fall into these two proxy
categories are reviewed that identify areas in need of financial education.
2.4.1 Financial Distress Factors
Financial distress is a state individuals experience due to a wide variety of decisions, within
an individual's control and many circumstances beyond the individual's control. This section will
review study factors indicative of financial distress brought about in part by decisions within an
individual's control and present supporting research for the categorization of such factors.
Assessing one's level of financial risk is a complex question, but there are events that can
indicate current distress. Foreclosure and tax delinquency are two factors that indicate financial
distress and can be used to identify those populations that would be well served by financial
13

education. There are situations where foreclosure and tax delinquency can occur for non-financial
reasons; however, the financial stability of an individual is often a factor in the resulting
foreclosure or tax delinquency event.
A housing foreclosure is a legal process by which a property may be sold and the proceeds
of the sale may be used to satisfy an existing mortgage debt (Apgar et al. 2005). The foreclosure
process begins after a series of missed payments on a mortgage. Foreclosure events significantly
affect the credit score of those undergoing the process and it will stay on a credit record for 7
years. A low credit score can make getting a loan, insurance, or even a job more difficult, since in
practice credit scores are used well beyond their intended purpose of evaluating the likelihood of
loan repayment (Kingsley 2009). Foreclosures have also been occurring disproportionately in areas
where there was a high density of sub-prime mortgage lending (Duda and Apgar 2004; Immergluck
and Smith 2005). A single foreclosed property can negatively impact the value of neighboring
properties within a half mile by 8.7% (Lin, Rosenblatt, and Yao 2009). Foreclosures negatively
affect both individual consumers and communities. Understanding where these events take place
most frequently can help communities target financial education and potentially reduce the
negative impacts of foreclosure.
The Federal Reserve Bank of Cleveland completed research that examined the factors that
contributed to housing price declines and found that foreclosure events occurred at the end of a
longer decline in community and individual financial health. One or more of the following can
often precede this downward financial trajectory, ending with foreclosure: tax delinquency,
property abandonment, or vacancy events (Whitaker and Fitzpatrick IV 2011). These pre-
foreclosure events were found to be just as important as foreclosure incidence when attempting
to indentify the factors at play in decreasing home values (Whitaker and Fitzpatrick IV 2011).
14

Tax delinquency is similar to foreclosure in the sense that homeowners have missed tax
payments over an extended period of time, which can result in municipality action to sell the
property to satisfy the outstanding tax balance and any accrued fees. Tax delinquency can provide
an earlier indication of financial distress because it often precedes foreclosure (Whitaker and
Fitzpatrick IV 2011). Since the effectiveness of financial education increases when delivered in
advance of financial hardship, tax delinquency is an especially valuable factor (Martin 2007).
When support is not available to owners of foreclosed, tax delinquent or vacant properties,
they often abandon their property management responsibilities. This deferred maintenance can
have a number of negative effects on the community as a whole. When properties are not cared
for frequently, they can begin to look unsightly. This makes a neighborhood less desirable to
prospective buyers. This series of events can ultimately reduce neighboring property values. If this
lack of homeowner care, stemming from financial distress, is not addressed by the community or
property owner, the downward trend can also lead to increased crime, vandalism and theft
(Whitaker and Fitzpatrick IV 2011). Research has estimated the cost to local government per
vacant or foreclosed property to be between $27,000 to $30,000 (Rogers and Winter 2009).
Both foreclosure and tax delinquency are factors that have significant impacts on the
financial health of individuals and communities. By actively organizing foreclosure and tax
delinquency data, communities will be better able to address property ownership issues and
support communities that are struggling with targeted financial education opportunities. Since
foreclosure and tax delinquency can indicate a state of financial distress stemming at least partly
from poor financial decisions, this study used these factors as indicators of a need for financial
education.
15

2.4.2 Potential for Financial Hardship Factors
There are a number of factors indicative of the potential for financial hardship; however,
with the growing availability and variety of financial resources provided by FLEC, financial
education can help an increasing variety of individuals with financial education needs become
more financially secure. It should also be made clear that individuals that regularly struggle with
finances are not necessarily bad financial decision makers. However, their even income to expense
ratio and potentially limited exposure to an increasingly complex financial market place can make
finding and maintaining financial stability difficult.
In 2010, America had 58.7 million individuals living with a disability. One in three of
disabled Americans live in poverty which is double the national average (Brault and Census 2012;
Institute 2012). The National Disability Institute has been working with 900 partner agencies to
increase the financial stability of American's with disabilities. Through financial education and
asset development assistance, this population has been able to become increasingly more
financially independent and economically self-sufficient (Institute 2012). The Federal Government
also recognizes the financial instability of this community and provides Supplemental Security
Income to support disabled Americans and certain elderly populations. Given the current income
disparity of disabled Americans and efforts to provide financial support to this group, it is clear
disabled Americans are often at risk of financial hardship and good candidates for financial
education.
Unemployment has also been a struggle for millions of American's after the Financial
Collapse of 2008. In January 2014, the Bureau of Labor Statistics reported the current rate of
unemployment to be 6.6%, which is a new 5 year low. However, this metric can mask the actual
health of the job market. There is another metric known generally as underemployment which
16

includes individuals that are underpaid, underutilized, overeducated, over skilled, or overqualified
for their current position (McKee-Ryan and Harvey 2011).
The rate of underemployment dropped from 17% in 2010 to 13% at the end of 2013.
However, the underemployment rate is more than double the unemployment rate (Insititue 2011;
Statitistics 2013). McKee-Ryan and Harvey (2011) have found that there is a positive relationship
between underemployment and the intention to quit a job. They also found a positive relationship
between the length of time an individual is unemployed and underemployment. Given large
numbers of individuals who would categorize themselves as underemployed, the relationship
between unemployment and underemployment, and the general desire of underemployed
workers to quit these kinds of jobs, the potential return of these individuals to the unemployed
segment is entirely possible. Currently, the federal government's Career One Stop Program
provides unemployed individuals to a variety of educational resources and training opportunities.
These opportunities, including basic financial skills training, and essential mathematics concepts,
are critical for more advanced financial education. Providing unemployed job seekers with the
skills they need to be successful is the goal of many unemployment-training programs. During
unemployed phases, individuals sometimes will have more limited incomes. This reduction in
income can increase the likelihood of financial distress. This qualifies unemployment as another
potential factor in this research.
Educational attainment is valuable metric for examining an individual's level of financial
knowledge. A review of common mortgage transactions with a broker has uncovered that less
educated mortgage buyers were charged $1,500 more on average than college graduates
(Campbell 2006). In another study, empirical evidence shows a negative relationship between the
ability to perform basic calculations and the propensity to default on a mortgage (Fernandes,
17

Lynch Jr, and Netemeyer 2014). Financial literacy surveys have shown that less educated
individuals struggled to grasp basic financial concepts, such as basic and compounding interest
concepts, more than more educated individuals (Lusardi and Tufano 2009; Lusardi and Mitchelli
2007). Individuals with low levels of educational attainment are more likely to have difficulty
evaluating financial products and managing debt obligations. The level of educational attainment
also directly impacts individuals’ life-time earning potential (Day and Newburger 2002). This lower
earning potential can make it much more difficult for these individuals to save money. This
reduced ability to amass financial reserves decreases the possibility of weathering negative
financial fluctuations without suffering negative impacts. Financial education can help increase the
savings rates of individuals with low educational attainment and reduce the likelihood these
individuals will experience financial hardships.
Individuals with incomes at or below the poverty level are closer to financial hardship than
most. With limited incomes, individuals living at the poverty level might struggle to save any
money in an emergency fund. Having available an emergency savings fund for times of economic
uncertainty is a common recommendation for all individuals and households (Brobeck 2013).
However, the American public's saving habits overtime have become increasingly more relaxed.
This trend increases the potential for greater financial hardship during economic downturns. The
personal savings rate of Americans has decreased every decade by 28% on average since the
1960's (Louis 2014). Following the most recent financial collapse in 2008, personal savings rates
increased greatly during 2009 and 2010. However, the 40-year-old trend of decreasing personal
savings rates resumed in January 2011 and the negative trend has continued to May 2014.
The United States Government has recognized the importance of providing financial
education and saving assistance to those in poverty (Clancy, Grinstein-Weiss, and Schreiner 2001).
18

In an effort to increase access to banking resources and incentivize saving habits, the US
government has developed Individual Development Accounts (IDA). IDA accounts are provided to
lower income individuals who often find it difficult to save using common banking resources
because of minimum balance requirement for opening saving accounts active. To remove these
barriers, the IDA program removed minimum account requirements set by commercial banks. To
incentivize consistent savings habits, the federal government matched the funds saved by IDA
account holders, providing immediate economic reinforcement of the value saving can provide. To
get access to an IDA and receive the matching funds offer, the participant needs to attend a series
of financial education seminars. The IDA program takes an applied approach to improving the
financial literacy of each participant. Teaching individuals the discipline to save while solidifying
basic financial literacy concepts connects the learning process to positive financial outcomes. The
IDA program has successfully helped struggling families improve their financial decision-making
skills, save for a down payment of a homes, and regularly save for college tuition.
In 2007, the ratio of US housing debt to disposable income hit an all time high of 130%
(Glick and Lansing 2011). Since the financial collapse of 2008, the debt to income ratio has
dropped from its peak, but the availability of credit remains at levels that supported the
development of debt obligations higher than available disposable income. The Housing and Urban
Development department provides small loans to homeowners and as of May 2013, the maximum
debt to net income ratio that an individual can have if they are going to underwrite a loan is 43%
(Galante 2013). Individuals that possess two mortgages and home equity loans are likely in a
position in which their debt to net income levels are at or above what is considered financially
healthy. Individuals with large amounts of debt to service are at risk of financial hardship if the
19

ability to service that debt is interrupted. The financial hardship in this case could come in the
form of increased rates, fees, or even property seizure and foreclosure.
Many of the studies that have explored the financial health of Americans have captured
demographic details about individuals. The ethnicity of the decision makers is one factor that has
been evaluated for trends. During the lead up to the financial crisis of 2008, the Housing and
Urban Development Department (HUD) made efforts to encourage lenders to provide lower
income and minority populations with mortgages to purchase homes (Angelides and Thomas
2011). These groups traditionally have had difficulty meeting the minimum mortgage
requirements in the prime mortgage market, but lenders developed the sub-prime mortgage
category to meet the "needs" of these borrowers. This HUD-encouraged mortgage practice was a
great success, and large number of sub-prime mortgages agreements were signed. Partly as a
result, between 2007 and 2009, two and one half million foreclosures were completed (Bocian, Li,
and Ernst 2010).
When looking at ethnicity, the rate of foreclosure for African Americans and Latino groups
was 42% greater than for Non-Hispanic white populations (Bocian, Li, and Ernst 2010). Within the
sub-prime mortgage market, these populations were also more likely to receive expensive loans
and terms including prepayment penalties (Bocian, Ernst, and Li 2008; Bocian and Zhai 2005).
Finally, survey's have documented a statistically significant relationship between ethnicity and low
levels of financial literacy (Lusardi and Tufano 2009).
One of the largest single expenses an individual has is their housing costs. Understanding
what an appropriate expenditure on housing is can help delimit individuals that are over extended
and potentially compromising their own financial health. The concept of affordable housing began
with the US National Housing Act of 1937, when income limits were established for access to
20

public housing programs (Schwartz and Wilson 2008). This housing affordability effort has worked
on defining the best income to housing expense threshold over the subsequent decades. Today
the conventional wisdom is that a maximum of 30% of an individuals income should be spent
housing costs (Schwartz and Wilson 2008). This factor is most important for lower income
populations, young people and the elderly to be aware of due to their more limited means.
However, the 30% rule still generally applies to all individuals. Individuals spending more than 30%
on housing costs are at greater risk of financial distress given their reduced ability to save and
invest.

2.5 Financial Vulnerability Research and Multi-Criteria Evaluation
In the previous section, a number of the factors that could indicate the need for financial
education were reviewed. To identify suitable locations, multiple factors are often combined to
calculate the scores that represent the magnitude of suitability (Eastman 1995). Several
researchers have attempted to combine multiple factors to measure financial risk.
The Local Initiative Support Corporation research used a number of public and private data
sources to develop their foreclosure risk score (Corporation 2013). The data sources include the US
Census Bureau, American Community Survey (ACS), Housing and Urban Development (HUD),
Mortgage Bankers Association, and the private loan analytics company LPS Applied Analytics. The
factors used to develop the foreclosure risk index include; housing counts, homeowner and tenant
occupancy data, mortgage foreclosure types, and loan performance metrics. The index
development process began by weighting, correcting, and normalizing the proprietary mortgage
data and then the loan data. The final foreclosure risk index by zip code area was an aggregation of
21

indices including the percentage of loan foreclosures weighted by foreclosure count, percentage of
subprime loans weighted by subprime mortgage count, and percentage of delinquent loans
weighted by delinquent loan count.
Italian researchers have developed a financial vulnerability indicator based up on financial
survey data from Italian participants. The survey focused on capturing and understanding
household debt trends, but also captured metrics on the inability to pay for basic monthly and
daily items (Anderloni, Bacchiocchi, and Vandone 2011). The research concluded that the level of
debt servicing is positively related to financial vulnerability and the relationship is stronger when
households are holding unsecured debts.
Outside of academic research, MBD Credit Solutions has developed their own consumer
financial vulnerability index, via quarterly consumer surveys, to evaluate the current state of
consumer credit in South Africa (Solutions 2013). MBD Credit Solutions is the largest account
receivables company in South Africa and has a stake in understanding the consumer's ability to
pay their bills. The index is composed of four indicators, including income, expenditure, savings,
and ability to service debt.
When it comes to the evaluation of multiple factors and the calculation of scores, one
national scale research project sought to develop an index measure to locate populations
vulnerable to environmental hazards by county (Cutter, Boruff, and Shirley 2003). 250 factors were
initially identified by the study. After being tested for multicollinearity and the data normalized,
the original 250 variables were reduced to 42 independent variables. Principal component analysis
further reduced the 42 independent variables to 11 significant factors. These 11 factors were
found to explain 76.4% of the variance in environmental hazard. The data source of all eleven
factors was the 1990 decennial census at the county level. The eleven factors were personal
22

wealth, age, density of the built environment, single sector economic dependence, housing stock
tenancy, race, ethnicity, occupation, and infrastructure dependence. Using an additive model, the
researchers calculated an environmental hazard risk index for each county. An additive model was
selected because no defensible method could be applied to weight the 11 factors.
The most common approach to combine factors is the weighted linear combination
(Eastman 1995). Weighted liner combination is an approach that multiplies each factor according
to a weighting scheme and then takes the sum of these results to produce a suitability index for a
particular location. Researchers have traditionally converted factor values to a numeric scale (e.g.,
0-100), which is commonly called standardization, before they multiply each factor (Eastman
1995). This process of standardizing and weighting a number of factors to develop a suitability
index is called Multi-Criteria Evaluation (MCE).
The MCE is a technique that is designed to cope with evaluating multi-criteria problems
and can be useful spatial analysis tool (Carver 1991). However, the selection and weighting of the
MCE criteria are critical to the outcome of the analysis. Since the selection and weighting decisions
for each MCE criteria are based upon the viewpoints and biases of researchers or decision-makers,
there is inherent uncertainty in the results that should be recognized. Clear criteria selection and
weighting methodology is necessary when designing a MCE if the results are to be valued. There
are a number of supplementary theoretical constructs to help optimize criteria selection and
weighting approaches including: Bayesian Probability Theory, Fuzzy Set Theory, and Dempster-
Shafer Theory (Stoms 1987). These approaches are beyond the scope of this research, but could be
a valued addition by future researchers.
Standardization is a critical first step to prepare the criteria for the MCE. The
standardization process re-scales the MCE criteria raw values to a zero to one scale. The re-scaling
23

process requires care by researchers to identify which criteria are indicative increased suitability
and which are indicative of decreased suitability. Depending on the criteria, either standardization
formula A or B is required, shown below. For criteria whose highest values indicate the highest
suitability, standardization formula A is required. Standardization formula A will assign the highest
value a score of one and the lowest value a score of zero. For criteria whose highest values
indicate the lowest suitability, standardization formula B is required. Standardization formula B
will assign the highest value a score of zero and the lowest value a score of one. With all the
criteria in the MCE standardized and weighted as appropriate, the highest output values of the
MCE are indicative to the greatest suitability given the criteria selected.

The MCE methodology has been used by a number of researchers with both spatial and
non-spatial research objectives (Store and Kangas 2001) (Pohekar and Ramachandran 2004)
(Raaijmakers, Krywkow, and van der Veen 2008). The research by Store (2001) used MCE to
explore habitat suitability for endangered species and applied the standardization and weighting
approach described by Carver (1991). The research by Pohekar (2004) used MCE to identify the
best sustainable energy practice by evaluating 90 different sustainable energy practices. The
Raaijmaker (2008) research sought to identify the perception of flooding risk using a variety of
qualitative and quantitative criteria. The breadth of research that has leveraged the MCE approach
to evaluate both spatial and non-spatial criteria extends well beyond what has been presented in
this research. This robust method seems well suited for supporting the thesis research objective of
identifying the areas in greatest need of financial education.
24

The financial landscape today is one of options and responsibility. Individuals seeking to
navigate financial environments successfully will need some form of financial education. There are
groups of individuals that have a greater need of financial education. Individuals currently in
financial distress and those that are living closely to a state of financial distress are often the
groups that have the greatest need for financial education. With a variety of criteria indentified to
help locate these groups in need of financial education, MCE can be implemented to present a
clear representation of the spatial need for financial education.

25

Chapter Three: Methods
This chapter presents the selected methods necessary to identify the areas in the City of Malden
that have the greatest need for financial education. Ten datasets indicative of a need for financial
education were processed, standardized, and combined to identify those areas in need of financial
education. These factors include educational attainment, foreclosure incidence, tax delinquency
incidence, poverty status, mortgage debt status, disability status, unemployment, housing vacancy
incidence, high rent expenditure, and ethnicity. A synopsis of each dataset used in the spatial
analysis is presented first. The data clean up and processing steps required to prepare each factor
for the multi-criteria evaluation (MCE) are outlined next. Lastly, the MCE processing steps to
combine each factor and produce scores representing the need for financial education are
reviewed.

3.1 Outline of Data
Ten datasets were identified for use in the spatial analysis designed to identify the need for
financial education. These ten datasets can be broken into three groups: ACS data, Census Data,
and City Data. Seven of the datasets were from the 2008 to 2013 American Community Survey
(ACS), one dataset was from the 2010 Census, and two of the datasets were provided by the City
of Malden, a list of the datasets can be found in Table 1. A more detailed review of these
processing steps for each of the three groups will follow.

26

Table 1 - Data Sources and Formats
Data Data Type Data Source
Low Educational Attainment Vector Polygon American Community Survey
Family Poverty Status Vector Polygon American Community Survey
Mortgage Debt Status Vector Polygon American Community Survey
Disabled Populations Vector Polygon American Community Survey
Unemployment Vector Polygon American Community Survey
Vacancy Status Vector Polygon American Community Survey
High Rent Expenditure Vector Polygon American Community Survey
Ethnicity Vector Polygon 2010 Census
Foreclosure Occurrence Vector Point City Assessing Dept.
Tax Delinquency Vector Point City Treasurer Dept.

3.1.1 American Community Survey Data
Seven ACS datasets were used in this research. These seven datasets were collected as
tabular data and then joined with census block group polygons. The census block group level was
the highest spatial resolution available for the seven datasets. There were 52 census block groups
within the City of Malden's boundaries. A review of the collection and processing of each of the
seven ACS factors found to be indicative of a need for financial education will follow.
The ACS educational attainment data captured counts of the highest level of educational
attainment achieved by individuals, ranging from no schooling attained to a graduate level
attainment. The educational attainment counts were captured at one-year increments of
schooling. The low educational attainment factor used in this research was the cumulative product
of educational attainment counts for individuals indicating educational attainment between no
schooling and a high school degree level of attainment. The final educational attainment factor
contained a single counts of individuals with educational attainment less than or equal to a high
school degree. The need for financial education using the educational attainment data was
indicated by a raw count of individuals in each census block group. Each census block group raw
27

count required normalization to account for the variation in size of each census block group. The
raw count data for each census block group was normalized by dividing the raw count by total
population for each census block group. The higher the counts of individuals with educational
attainment below the high school degree level the higher the need for financial education.
The family poverty data contained the count of families that have been below the poverty
level for the last 12 months. Poverty level is determined by comparing annual income of a family
to a set of dollar values called poverty thresholds that vary by family size, number of children and
age of householder. If a family's before tax income is less than the dollar value of their threshold,
then that family and every individual in it are considered to be in poverty. The ACS dataset field
that contained the counts information was B17010e2. The need for financial education using the
family poverty data was indicated by raw count of families in each census block group. Each census
block group raw count required normalization to account for the variation in size of each census
block group. The raw count data for each census block group was normalized by dividing the raw
count by total population for each census block group. The higher the counts of families living at
the poverty levels in a census block group, the higher the relative need for financial education.
The mortgage debt status data contained the count of housing units that had two
mortgages and a home equity loan. The ACS dataset field that contained the counts information
was B25081e6. The need for financial education using the mortgage debt status data was
indicated by raw count of housing units in each census block group. Each census block group raw
count required normalization to account for the variation in size of each census block group. The
raw count data for each census block group was normalized by dividing the raw count by total
population for each census block group. The higher the counts of housing units in a census block
group, the higher the relative need for financial education.
28

The disabled population data contains the count of disabled adults between the age of 18
and 65. The ACS defines disabled adults as having one or more sensory, physical, mental, self-care,
go-outside-home limitations. The ACS dataset field that contained the counts information was
C23023e3. The disabled population data indicated a need for financial education by raw count of
individuals in each census block group. Each census block group raw count required normalization
to account for the variation in size of each census block group. The raw count data for each census
block group was normalized by dividing the raw count by total population for each census block.
The higher the counts of disabled individuals in a census block group, the higher the relative need
for financial education.
The unemployment data contains the count of adult individuals between the age of 18 and
65 that have not worked in the past 12 months. This categorization was assigned to all
respondents who indicated they worked less than 2 weeks out of a 12-month period. The ACS
dataset field that contained the unemployment counts information was B23022e25. The need for
financial education using the unemployment data was indicated by raw count of individuals in
each census block group. Each census block group raw count required normalization to account for
the variation in size of each census block group. The raw count data for each census block group
was normalized by dividing the raw count by total population for each census block group. The
higher the counts of unemployed individuals in a census block group, the higher the relative need
for financial education.
The vacancy status data contains the count of housing units that are currently vacant
without rental or sale contracts pending. The ACS dataset column that contained the vacancy
counts information was B25002e3. The need for financial education using the vacancy data was
indicated by the raw count of housing units in each census block group. Each census block group
29

raw count required normalization to account for the variation in size of each census block group.
The raw count data for each census block group was normalized by dividing the raw count by total
population for each census block group. The higher the counts of vacant housing units in a census
block group, the higher the relative need for financial education.
The high rent expenditure data contained the count of renter occupied units in each census
block group spending greater than 35% of their gross income on rent. The ACS data captures rent
spending for a number of different spending ranges. This research extracted all renter occupied
unit counts for those categories that where renter are spending greater than 35%. The ACS dataset
columns that contained the renter occupied unit counts were B25070e8, B25070e9, and
B25070e10. The need for financial education was indicated by the raw count of renter occupied
units in each census block group. Each census block group raw count required normalization to
account for the variation in size of each census block group. The raw count data for each census
block group was normalized by dividing the raw count by total population for each census block
group. The higher the counts of renter occupied units in a census block group, the higher the
relative need for financial education.
The census block group boundaries are delimited based upon population. Densely
populated regions will have smaller area census block groups and more sparsely populated regions
will have larger area census block groups. Each of the seven ACS dataset has been aggregated into
the census block group regions. The respective size of each areal unit will affect the raw counts of
individuals, households, housing units and families and ultimately has the potential to influence
the outcome of this research. This issue is known as the modifiable areal unit problem and was
described in detail by Stan Openshaw (Openshaw 1983).
30

Working with the statewide source ACS data set required great effort to extract the data of
interest and organize appropriately for processing. Care should be taken to validate the data join
processing of the tabular ACS data and the census block group geometry. Another consideration
for researchers is to evaluate of the margin of error associated with each factor. The ACS makes
estimates of how the entire population of a census block would respond to a particular survey
question based upon a sample of respondents from that census block group. The ACS margin of
error is variable between data and care should be taken during ACS factor selection to evaluate
the margin of error for each factor.
3.1.2 2010 Census Data
Ethnicity census data from the 2010 decennial census was used in this research. The
Massachusetts census summary file, in a tabular format, was joined with census block polygons.
The census block aerial unit provided the highest spatial resolution available for the ethnicity
census data. There were 780 census block groups within the City of Malden's boundaries. A review
of the processing steps required to prepare the ethnicity data for use in this research will follow.
The 2010 census data contains a wide variety of ethnicity data that describes the density of
a number of different ethnicities. The two specific ethnic groups and their densities were of
interest to this study. The raw count of African Americans and Latinos populations by census block
group was collected. Since research has shown that these two ethnicities are more likely to have
low financial literacy skills, the counts of individuals by census block were summed together,
creating a single raw count. Each census block raw count required normalization to account for the
variation in size of each census block. The raw count data for each census block was normalized by
dividing the raw count by total population for each census block. The need for financial education
using the ethnicity data was indicated by the raw count of individuals in each census block. The
31

higher the counts of individuals meeting the ethnicity criteria in a census block group, the higher
the relative need for financial education.
The 2010 census block is another areal unit that has been defined based upon population.
This approach to areal aggregation based upon population creates a situation where the
modifiable areal unit problem (MAUP) could influence the interpretation of the ethnicity data.
Awareness of the potential MAUP is critical to the interpretation of the MCE results.
3.1.3 City Data Processing
The City of Malden's Assessing Department and City Treasurer provided tabular foreclosure
incidence and tax delinquency incidence data respectively. The GIS department also provided
parcel point centroid data for this research effort. The tabular foreclosure and tax delinquency
datasets were joined with the parcel centroid point vector data to provide a spatial representation
of the incidence of housing foreclosures and tax delinquent individuals. A detailed review of the
foreclosure and tax delinquency point data processing steps is presented in this section.
The City of Malden provided access to foreclosure and tax delinquency data for this study.
The City Assessing department archives yearly historic property assessment datasets. The archived
assessing data includes wide variety of property specific characteristics: the recent sale price,
current lot size, number of bathrooms and a unique identification code. The archived assessing
data was consistently captured on January 1st of each year and the data was maintained in a CSV
format. In the archived assessing data, the attribute of most interest was the non-arms length
property sale transaction information. The non-arms length attribute recorded a property sale or
land sale that occurred between a buyer and seller, who both had an existing relationship. A family
member to family member property transaction is a common example of a non-arms length
transaction.
32

Homeowners and banks are also an example of an existing buyer and seller relationship.
Foreclosure proceedings result in the transfer ownership from the homeowner to the bank, which
classifies as a non-arms length transaction. The archived assessing data for the 2008 and 2013
period was examined and the non-arms length transaction fields were extracted from each of the
yearly assessing dataset into a non-arms length transaction data table. The non-arms length data
was then examined in detail to identify records indicating a foreclosure transaction was the last
property sale transaction for that year. These foreclosure records were extracted with a unique
parcel identifier field.
Using the unique parcel identification, the yearly foreclosure data was joined with the
copies of the city's parcel centroid point dataset. The resulting 5 years of individual foreclosure
point records were merged into one single point dataset that ultimately contained 323 foreclosure
events.
The final foreclosure dataset contained many coincident points. In order to avoid an over-
estimation of the number of foreclosure events in a particular location, a data validation step was
required. The validation assessed if the previous year's foreclosure event had the same owner
information or not. If the same ownership information was found for a property exhibiting many
foreclosure transactions, it was assumed that the foreclosure events were more likely a single
foreclosure event initiated by the same owner. When multiple owners were identified in the same
parcel, it was assumed that two or more foreclosure events occurred. A single point in the center
of the parcel experiencing foreclosure would represent a single foreclosure event. When multiple
foreclosure events occur for a specific parcel these central points in a single parcel become
stacked. Those parcel centroids with stacked points and a single unique ownership record were
reduced to a single centroid point. Those properties that had two or more unique foreclosures
33

during the study period were manually unstacked and given non-coincident placement within the
source parcel. There was only one property in the 5 year data sample reviewed where one
property owner foreclosed on their home, the bank resold the property to a new owner and that
owner went into foreclosure.
The tax delinquency records were maintained by the City Treasurer. The tabular record
contained property owners that owed outstanding taxes on their property. This list contained tax
delinquent homeowners that had been paying down their delinquent taxes for many years and
those homeowners who as recently as the 2013 tax year became tax delinquent. Even though the
initial delinquency status could have started up to 30 years prior, this research will include all
outstanding tax delinquency records regardless of the delinquency start date because each tax
delinquency record indicates an currently tax delinquent property. The tax delinquency data also
contained the repayment status, property owner information by address, and the parcel ID for the
property in tax delinquency.
The tax delinquency property records were joined with the city parcel centroid dataset.
Similar to the foreclosure parcel centroid processing, any coincident tax delinquency points were
evaluated to determine if it was appropriate to count each tax delinquency parcel centroid point
as its own instance of tax delinquency. To assess this, the parcel centroids with stacked points
were evaluated to determine if more than one owner was in tax delinquency for a particular
property. There were no stacked tax delinquency points that had multiple unique occupants in a
delinquent tax state. The resulting total of tax delinquencies records used in the study was 332.

34

3.2 Outline of Methods
With the factor input data clearly defined, these data were processed by the following
steps: (1) Prepared and normalized the ACS, census data, parcel based data; (2) Rasterized the ACS
and census data; (3) Ran kernel density estimations for foreclosure and tax delinquency data; (4)
Standardized all factor data; (5) Combined all the data through an evenly weighted multi-criteria
evaluation (MCE) analysis.

Figure 2: Methodological Flowchart Diagram
Step one was important to the success of the research. All downstream processes relied
upon accurate data processing efforts. The specific tabular data of interest from the ACS and 2010
Census were organized, and then the repetitive joining and normalization were completed. The
city tabular data were also processed, and then each dataset was joined with the parcel point
centroid data.
Step two was required to prepare the ACS and census data for the MCE processing. The conversion
from polygon to raster was completed. The raster data cell size was set to 5 meters, to provide
enough spatial resolution to delimit census block groups and census block from adjacent regions.
Step three was required to create a surface that could represent the city point vector data in a
continuous way. Step four was necessary to rescale the normalized raw counts and density
measures onto a zero to one scale. This scaling method allowed an even comparison to be
1. Census Data
Processing
1. Parcel Data
Processing
3. Kernel
Density
Development
4. Factor
Standardization
5. Evenly
Weighted MCE
2. Feature to
Raster
Processing
35

completed between any of the factor. In step five, the evenly weighted MCE process was
completed to produce a single output representing areas in need of financial education.
3.2.1 Normalization and Standardization of Factors
Each of the ACS and 2010 Census data used in this study contained the raw counts of the
following factors; educational attainment, foreclosure incidence, tax delinquency incidence,
poverty status, mortgage debt status, disability status, unemployment, housing vacancy incidence,
high rent expenditure, and ethnicity. However, each of the raw counts needed to be normalized to
accomplish the evenly weighted MCE process. Given each census block group or block varies in
size by the population density of a given area, a population based normalization approach was
appropriate to ensure the raw counts for any given factor was relative to the population of the
census areal unit. The normalization process produced a ratio by dividing the raw count for each
enumeration unit by the total population of the enumeration unit. For example, if the raw count
for a particular census block group and the total population of the block group were 10 and 100
respectively, the normalized value was 0.1.
When comparing the normalization ratio values between ACS and Census factors, the
highest and lowest ratio values varied greatly between each factor. This variation in normalization
value was due to the wide variations in population and raw count values for each factor. In the
normalized state, combining all the factors ratios evenly would favor those factors with the higher
scores over those with lower scores. This variability in the normalized factor value ranges did not
support the evenly weighted analysis goals of this research. To compare all the factors evenly, the
ratio measures needed to be further standardized.
The standardization process was conducted to convert all normalized values onto a zero to
one scale. This allowed for an even comparison between factors to be completed without concern
36

over the scale and range of raw score values. The highest or lowest suitability values in any given
dataset will have the same standardization value, allowing an even comparison between all
factors.
The standardization formula selection was dependent on whether the factor data values
represent an increase or decrease in the need for financial education. The standardized value A
calculation ensured the highest score in a factor data range was equal to one and the lowest score
in the data range was equal to zero. The standardized value B calculation ensured the lowest score
in the data range was equal to one and the highest score is equal to zero. The standardized value A
formula was applied to all the ACS, 2010 Census, and city data factors in this research since the
highest raw count and density measures in each factor was indicative of a greater relative need for
financial education. The lower raw score and density measures were indicative of a lower relative
need for financial education.

37

3.2.2 Kernel Density Estimation
The foreclosure and tax delinquent parcel centroid datasets required a Kernel Density
Estimation (KDE) process, which was conducted to make inferences about the greater population
of incident points in areas where no actual incident point data existed. Each point represented one
incident in the datasets. If two incidents occurred in the same parcel, that parcel contained two
points. In this KDE process, a search radius was defined to determine the maximum extent any
given point had an influence on any nearby point. Accordingly, a distance for the foreclosure and a
distance for the tax delinquency were set at 425 meters and 433 meters respectively. Overlapping
radii that surrounding each point were then totaled, higher values indicating a greater number of
coincident radii. In addition, four points were placed outside the city boundary in both the
foreclosure and tax delinquency datasets. These four points do not correspond with actual
foreclosures or tax delinquency incidents. However, they were necessary to extend the bounding
box of the KDE processing output to cover the entire bounds of the city. The cell size of the output
raster data was 5 meters. The two surfaces produced provided an estimation of the distribution
and density of the foreclosures or tax delinquency incidents.
To include the two KDE raster surfaces in the MCE analysis, the standardization method
that was applied to each cell value in the KDE surfaces to allow for an even comparison between
all the factors in the analysis.
3.2.3 Evenly Weighted Multi-Criteria Evaluation
After all the factor values were rasterized and standardized, they were aggregated to
produce a single surface that represented areas in need of financial education. To combine all ten
factors, a weighted linear combination method was used. Given the lack of an informed method of
assigning weights to factors related to the necessity of financial education, an evenly weighted
38

approach was used in this study. Each of the ten factors was given a weighting factor of one. This
weighting strategy resulted in unaltered standardized values before the factors were combined.
The resulting raster surface with 5-meter cell size contained the sum of all the standardized factor
values. This output indicated where there highest relative need for financial education was located
in the City of Malden. Higher values were indicative of areas in greater relative need for financial
education and the lower values were indicative of areas in lower relative need for financial
education. The interpretation of these results will be covered in Chapter 4.

39

Chapter Four: Results
Identifying the spatial distribution of need for financial education was the goal of this research.
Research efforts identified ten factors that can indicate a need for financial education. The tabular
factor data required joining with the appropriate spatial features, the raw data values required
normalization, and the normalized values needed standardization in preparation for an evenly
weighted multi-criteria evaluation (MCE). In this chapter, the data preparation results for each
research factor are evaluated in detail. Afterward, the evenly weighted MCE result is presented
and the spatial distribution of high and low need areas for financial education is documented.

4.1 Foreclosure and Tax Delinquency Results
Foreclosure and tax delinquency data were prepared for the evenly weighted MCE using a
Kernel Density Estimation (KDE) method. The KDE processing effort produced two raster surfaces
found in Figures 3 and 4. The source data points for the KDE surfaces are present in Figures 3 and
4. The KDE results were thematically classified using a Jenks natural breaks optimization method.
This data classification technique optimizes the selection of classes to reduce the variance within a
particular class and maximize the variance between classes (Jenks 1967). This classification
method is appropriate given the left skewed distribution of the standardized values presented in
the histograms in Figures 5 and 6. The break values for the foreclosure and tax delinquency figures
vary since the break values are a product of the unique distribution of standardized value data in
each dataset.
To further compare and contrast the expression of the foreclosure and tax delinquency
point densities, a method used to quantify point feature clustering tendencies was applied.
40

Nearest neighbor analysis was used to evaluate how clustered or evenly dispersed each point
dataset was relative to a random distribution of points (O'Sullivan and Unwin 2003). The results of
the analysis produced z-scores that indicated how clustered or evenly distributed each point
dataset was. Negative nearest neighbor analysis z-scores less than -1.96 indicate a statistically
significant clustering trend. Positive nearest neighbor analysis z-scores greater than 1.96 indicate a
statistically significant even dispersion of points trend. The nearest neighbor analysis produced a z
score of -9.54 for foreclosure point dataset and a z-score result of -8.86 for the tax delinquency
dataset. Both the foreclosure and tax delinquency data sets are clearly exhibiting statistically
significant clustering trends. Relatively speaking, the tax delinquency data is more evenly
dispersed than the foreclosure data.
The statistically significant clustering of points identified by the nearest neighbor analysis
intuitively equates to high-density areas in both of the KDE results. Some of the largest high-
density KDE clusters in both the foreclosures and tax delinquency data sets are identified in
Figures 5 and 6 using black numbered boxes. The relative density of these point clusters is
graphically presented using a red to green scale. Red areas indicate the highest densities of points
in the KDE and green areas indicate much lower densities of points. The red areas, indicating a
high needs for financial education, cover much less area than green lower relative need areas. The
histograms in Figures 5 and 6, located below the map figures, provide a graphical depiction of the
distribution of need. The histogram vertical axis shows the count of raster cells in the KDE surface.
The horizontal axis presents each of the nine Jenks natural break classifications of standardized
values. The red, high need areas make up a very small percentage of the entire raster surface. The
tax delinquency KDE has ninety-two 5m x 5m raster cells identified in the highest need
classification bin, this equates to 0.56 acres. The high need areas numbered one through three in
41

the foreclosure Figure 5 and one through four in the tax delinquency Figure 6, will be referred to
as "cluster" during the result analysis and discussion. The largest foreclosure clustering occurs
primarily in center of the City and just south of the City center in a largely residential area,
identified as cluster Area 1 and cluster Area 3 in Figure 5. The large tax delinquency clusters in
Area 1 and Area 3 also fall in the same location as the foreclosure clusters. However, tax
delinquency cluster Area 2 and Area 4 are located in the eastern part of the city as well, north and
south of the industrial rail line.

Figure 3: KDE of Foreclosure Occurrence with Source Foreclosure Point Data Shown
42

Figure 4: KDE of Tax Delinquency Occurrence with Source Delinquency Point Data Shown
43

Figure 5: Results of the Foreclosure KDE Processing
2
3
1
44

Figure 6: Results of the Tax Delinquency KDE Processing
1
2
3
4
45

4.2 Census Data Results
Each of the ACS and Census factor datasets required normalization, rasterization, and
standardization to produce the results described in this section. The standardized factor values are
presented in each of the census figures. Each figure was thematically classified using an equal
interval data classification method with nine classes in each map and histogram.
The factor output for high rent expenditure indentifies two census block groups that have
standardized values in the highest equal interval class. Both census block groups are west of the
Orange line. The histogram in Figure 7 is a representation of the number of raster cells that fall
into each of the equal interval standardized value classes and provide insight into the distribution
of data. The Figure 7 histogram shows the majority of the high rent expenditure factor values
falling in the standardized value classes 0.11, 0.22, and 0.33 and very few census block groups
scored in the highest standardized value classes 0.78, 0.89, and 1. The histogram appears to be left
skewed. The mean standardized value of 0.31 aligns with this visual assessment of the histogram.
Those high need areas that have high-standardized values greater than 0.67 cover approximately 3
acres of the city and are located close to the Orange line.
This could be an indication that individuals are willing to pay more for rent when they live
in close proximity to convenient mass transportation into Boston. The eastern most third of the
City has low standardized values falling between classes 0.11 and 0.44. This lower rent
expenditure trend could be a function of the distance from the desirable mass transit options
further west.
46

Figure 7: Households that spend greater than 35% of their income on rent. (left) An equal interval color map
representing the standardized score for the factor by census block group. (right) A histogram of raster cell count
verses the standardized value distribution. The x-axis indicates the upper bound of each bin.
The disabled adult factor indentifies four census block groups with standardized scores in
the highest classification of scores (Figure 8). The histogram in Figure 8 is presenting a bi-modal
distribution with the highest number of raster cells falling in standardized classifications 0.22 and
0.89. This bi-modal representation in the 1st and 3rd quartile of standardized values is unlike any
of the other factors data distributions. The difference in cell counts between each equal interval
classifications is very small, with the largest difference equaling 2,000 5m x 5m cells or a 12-acre
difference in area. The high and low scores present across the city in no discernible pattern. The
data indicates that the disable population is spread throughout the city. Educational event planner
should consider accessibility issues when planning financial educational events in the City of
Malden.
47

Figure 8: Disabled Individual between the ages of 18 and 65. (left) An equal interval color map representing the
standardized scores for the factor by census block group. (right) A histogram of raster cell count verses the
standardized value distribution. The x-axis indicates the upper bound of each bin.
The mortgage debt status factor is unique among the census data when considering the
total number of respondents that met the survey criteria (Figure 9). The mortgage debt status
factor's raw ACS data contains the count of housing units that have two mortgages and a home
equity loan. Fifty-eight total housing units spread across four different census block groups during
the 2008 to 2013 time-period indicated they met the survey criteria. Due to very few housing units
meeting the criteria, only four census block groups have standardized scores above zero. Two of
these block groups are located in the northwestern corner of the city near the Orange line. The
other two block groups are located in the center of the city south of the industrial branch rail line.
Of the four groups, one block group falls in the highest standardized value class equal to one.
However, this is expected because the standardization formula ensures the highest raw score
receives a standardized score of one. The high-standardized value block group is located in a
residentially zoned part of the city south of the branch line and adjacent to the southern town
boundary (Figure 16).
48

Given the small number of housing units meeting the survey criteria and the fact that the
standardization formula must assign a value of one to the highest value in the dataset, researchers
should carefully evaluate if it is appropriate to include factors with limited results. This research
has determined the mortgage status factor to be a valuable indication of need for financial
education, but the impact of the mortgage status factor on the final MCE will be isolated in only a
few block groups but make a strong contribution to the final MCE score in those areas.

Figure 9: Number of Individuals with Two Mortgages and a Home Equity Loan (left). An equal interval color map
representing the standardized scores for the factor by census block group. (right) A histogram of raster cell count
verses the standardized value distribution. The x-axis indicates the upper bound of each bin.
The ethnicity factor is the only factor from the 2010 census that was aggregated on the
census block level (Figure 10). The ethnicity factor data is showing where there are large numbers
of African American and Latino individuals living in a single location. The histogram in Figure 10
shows a heavily skewed left distribution with 80% of the cell values falling in the lowest two
standardized value classes equal to 0 - 0.11 and 0.22. Focusing on the spatial distribution of the
remaining 20% of the cell values with standardized values above 0.22, there appears to a general
organization around the Orange line in the western side of the city. There are only two census
blocks in the highest standardization class equal to one. The highest need census blocks are
49

located near the city center, just north and south of the industrial branch rail line. The distribution
of values makes a strong statement that there are no large clusters of both African American and
Latino populations living in the same location. It is possible that African Americans and Latinos
prefer to inhabit separate neighborhoods, which would require additional research to validate. If
this is the case, separating the two groups might provide a better indication of the racial or ethnic
hotspots.

Figure 10: Number of African Americans and Hispanic Individuals (left). An equal interval color map representing
the standardized scores for the factor by census block group. (right) A histogram of raster cell count verses the
standardized value distribution. The x-axis indicates the upper bound of each bin.
The educational attainment factor has a standard score distribution most closely matching
a Gaussian distribution (Figure 11). This bell curve like distribution can be clearly seen in the Figure
11 histogram. The normal distribution observed in the histogram is also support by the fact that
the first quartile value, mean and third quartile values fall roughly at 0.25, 0.5, and 0.75
respectively. The standardized value class equal to 0.56 has double the amount of cell values of
the next highest class of 0.44, which makes for a very tall and narrow bell curve. The data
distribution has produced four census block groups that fall into the highest standardization value
class. Spatially, the educational attainment score values appear to be higher to the south and
50

west. There also seem to fewer individuals with low educational attainment in northeast of the
city. Two of the four highest need areas have schools within the same census block group
boundary. The other two high need areas located west of the Orange line are adjacent or in close
proximity to a school. This could be an indication that those with limited educational attainment
come to value formal education and situate themselves in close proximity to educational
resources.

Figure 11: Individuals that have not attained a High School Degree (left). An equal interval color map representing
the standardized scores for the factor by census block group. (right) A histogram of raster cell count verses the
standardized value distribution. The x-axis indicates the upper bound of each bin.
The factor identifying family poverty status has a distinct left skew shown in the Figure 12
histogram. This left skew distribution leaves very few census block groups with high standardized
values and has resulted in only one census block group being classified in the highest score
category. The largest census block group in the city, measuring 0.35 square miles, falls into the
standardized value class equal to 0.78 for households below the poverty level. This particular block
group contains the largest contiguous business and industrial zones (Figure 16). Those census
block groups that fall into the standardized value class equal to 0.56 ultimately have a higher
raster cell counts than the standardized value class equal to 0.78 and are coincident with
51

residential areas. The highest need area block group also happens to be bounded north and south
by block groups with the lowest level of need, an unusual juxtaposition of need. There could be an
opportunity to leverage local community knowledge to support neighbors in the highest need of
financial guidance or support.

Figure 12: Households with Incomes Below the Poverty Level for the previous 12 months (left). An equal interval
color map representing the standardized scores for the factor by census block group. (right) A histogram of raster
cell count verses the standardized value distribution. The x-axis indicates the upper bound of each bin.
The unemployment factor has strong left skew, with 63% of the raster cell values falling in
the lowest third of standardized value classes 0.11 , 0.22, and 0.33 (Figure 13). Similar to the high
rent expenditure and mortgage debt status factors, this left skew leaves only one census block
group classified in the highest need class category. This area with the highest need for financial
education is coincident the largest census block group, which is also the business and industrially
zoned area of the City of Malden (Figure 16). There are three census block groups with
standardized value in the class equal to 0.67. Two of these three block groups are adjacent or
coincident with industrially zoned areas of the city. Given the proximity these high need areas are
to industrial zones, could the individuals in these areas be experiencing unemployment because
the industrial areas are no longer offering the jobs that once attracted individuals to the location?
52

Figure 13: Individuals that have not worked in the past 12 months (left). An equal interval color map representing
the standardized scores for the factor by census block group. (right) A histogram of raster cell count verses the
standardized value distribution. The x-axis indicates the upper bound of each bin.
The final census factor to review is the housing vacancy factor, which like the
unemployment factors has left skewed standardized values (Figure 14). The spatial distribution of
census block groups with standardized values in lowest class equal to 0 - 0.11 are found in the
center of the city. This distribution is different from the other ACS factors that have often
expressed higher standardized values in the center of the city. This distribution could be due to the
value of real estate and rental space in the center of a city just outside of Boston. A wide
distribution of census block groups in standardized value classes greater than or equal to 0.56 are
coincident or adjacent to the industrial branch, Orange line, and the major roadway Route 60
(Figure 1). Properties adjacent to major transportation pathways with lots of traffic might be more
difficult to keep occupied with tenants or owners.
53

Figure 14: Number of Housing Units Currently Vacant (left). An equal interval color map representing the
standardized scores for the factor by census block group. (right) A histogram of raster cell count verses the
standardized value distribution. The x-axis indicates the upper bound of each bin.
The census factors collectively had a number of similarities. The data distribution was
often left skewed with the majority of raster cell values falling in the lowest three standardized
value equal interval classes. On average, the center of the city had a higher need for financial
education. The transportation corridors also have a role to play in many of the different factor
expression and should be an important feature to consider and review. The zoning of the city also
seems to have a relationship with the expression of need for financial education, but additional
review is required. Collectively, these ten factors will provide insight into the need for financial
education.

4.3 Evenly Weighted Multi-Criteria Evaluation (MCE) Output Review
Ten factors, each indicative of the need for financial education, were combined to produce
the evenly weighted MCE output shown in Figure 15. An equal interval data classification approach
was used to represent the MCE results. The areas with highest cumulative standardized values
indicate where the greatest relative need for financial education is located in the City of Malden.
54

The highest possible value that could be produced by the MCE is a standardized value of 10. This is
a function of the ten factors used in the MCE, each having the highest possible score of one. The
lowest possible score that could be produced by the MCE is a standardized value of 0. The highest
need areas are symbolized using orange and red thematic representations and a selection of some
of the "high need areas" are highlighted in Figure 15 using numbered black circles and ellipses. The
Figure 15 histogram for the MCE shows a slight left skew with a sharp drop as the cumulative
standardized values increase. This histogram shows that the highest need class contains only 12
5m x 5m cells or 300 m
2
of area. This area is just south of Salem Street, north of Playstead Road,
and between Hyde Street and Cross Street. This area contains 22 residential structures. This area
of highest need is identified in Figure 15 as Area 1. Area 1 is coincident with the highest scoring
census block from the Ethnicity data and the densest cluster of foreclosure data (Figures 5 and 10).
It was necessary to expand of the definition of great need for financial education beyond
the single highest need area with the highest MCE cumulative value, since it would not be effective
to target financial education to just 22 residences. A review of those areas that have cumulative
MCE standardized values greater than 3.24 and greater than 4.19 was conducted. High need areas
with scores greater than 3.24 covered 1.2 square miles of the city of Malden, which is 23.5% of the
total land area of the city. High need areas with scores greater than 4.19 covered 0.2 square mile
of the City of Malden, which is 3% of the total land area of the city.
Since the MCE analysis is an expression of all the individual factors identified in the
previous sections, similar spatial traits are present in the MCE output. The highest need areas are
coincident and adjacent to the orange line, branch industrial rail line, and Route 60. The city
center, business districts and industrial districts express a higher relative need for financial
education when compared to regions east and west of the center of Malden (Figure 16). An
55

isolated area in the east of the city, identified in Figure 15 as Area 3, has expressed higher needs
for financial education. This is primarily a product of the tax delinquency, educational attainment,
and disabled adults data. One of the largest areas in need for financial education, identified as
Area 2 in Figure 15, is also coincident with a single census block group. The high MCE value in Area
2 is primarily the product of the unemployment, family poverty status, disabled population, and
high rent expenditure factors. The majority of the factor data used in the MCE analysis is from the
ACS, which has well-defined census block group boundaries. However, there are a number of
scattered high need areas with less boundary uniformity. These areas of high need are a function
of the spatial distribution of the foreclosure and tax delinquency KDE data.
The incidence of low scores with values lower than 1.33 are found in the southeast
adjacent to the industrial rail, northwest adjacent to the western most boundary, and in the
northeast corner adjacent to the City of Saugus. More generally, these areas seem to be primarily
in the northern half of Malden. Low standardized scores areas should be interpreted as having a
lower immediate need for financial education.
56

Figure 15: Evenly Weighted MCE Overlay Output and High Financial Educational Need Areas
2
1
3
57

Chapter Five: Discussion
An individual's pursuit of financial education should never be complete. In response to the
increasingly complex financial systems, individuals are required to maintain a minimum level of
personal financial security. Being prepared for financial changes and turmoil requires self-study
throughout one's life, ideally with the study focused on answering the most pressing financial
questions. Efforts to increase the availability and diversity of financial education resources are
underway; however, the question of where there is the greatest need for financial education
remains. The focus of this research is to identify those areas in the City of Malden that have the
greatest need for financial education using a variety of socioeconomic and demographic factors.
These factors include educational attainment, foreclosure incidence, tax delinquency incidence,
poverty status, mortgage debt status, disability status, unemployment, housing vacancy incidence,
high rent expenditure, and ethnicity. By properly formatting, normalizing, standardizing, and
combining these factors using evenly weighted MCE, one can identify those regions that have a
high need for financial education.

5.1 Evenly Weighted MCE Analysis Discussion and Application
The final evenly weighted MCE result is a product of all the research factor variations in
need for financial education. At a high level, the city center, commercially intensive area in the
southwest, and isolated residential areas throughout Malden are expressing high need for
education. In aggregate, the MCE results are helpful to focus financial education support efforts.
However, evaluating the appropriateness of each high need MCE area requires researchers to
return the individual factors. Additionally, users should understand that the MCE results or
58

individual standardized factor values are not to be interpreted as a binary decree of need for
financial education. The analysis outputs should be interpreted as a continuous spectrum of need
for financial education, with some areas needing more support than others do.
The range of values produced by the MCE indicates the relative need for financial
education. The MCE results can help focus financial education support efforts in the areas of
greatest relative need. To do so, decision makers must first decide what standardized values best
define the greatest need for financial education in a given geographic study area. This decision is
best made by reviewing the results produced by the MCE closely.
The MCE is a decision making tool that can help direct financial education efforts.
Understanding which factors are impacting the MCE in a given area provides additional context to
the need for financial education. The sub sections following will provide additional discussion of
the individual factors and their impacts on the final MCE results.
5.1.1 Foreclosure and Tax Delinquency Significance
Beyond the clustering analysis and density of points, the expression of the foreclosure
points in the context of the City, provides another aid to interpret the KDE output. Figures 5 and 6
show high-density clusters of the foreclosure and tax delinquency data points respectively. These
clustered areas intersect often with a number of differently zoned areas, which can be used for a
number of different purposes and have different owners. High-density cluster Area 3, in both the
foreclosure (Figure 5) and tax delinquency (Figure 6) factors, occur in a residential zoned area
(Figures 16). The coincidence of residential homeowners with foreclosure and tax delinquency
occurrence aligns with the research goal of indentifying individual homeowners who are currently
in financial distress.
59

Reviewing the high-density cluster Area 1, for both the foreclosure (Figure 5) and tax
delinquency (Figure 6) factors, you see a more mixed presentation of zoning intersecting with the
clustering areas. Specifically, there are industrial zones in close proximity to the industrial branch
rail line and more mixed use buildings with both commercial and residential zoning along the main
east to west city thoroughfare, Route 60 (Figure 1). Given these variations in zoning and the KDE
factor clustering, additional effort was focused on understanding if these variations might
influence the need for financial education in these areas. This analysis focused on identifying
additional property and owner details for each of the individual points within high-density
foreclosure clusters.
A trend was identified that might explain the incidence of high-density clusters in
foreclosure Area 1, was located near the city center highlighted by a black bounding box (Figure 5).
These clusters were located in business zoning areas. The source foreclosure points in high-density
cluster Area 1 were primarily commercial property foreclosures. The differentiation of a
commercial foreclosure was identified by reviewing the source assessing data, which original
provided the non-arms length sale information.
The tax assessing information also contains the legal owner name and other billing
information that allowed for a differentiation between residential and commercial owners. The
reasonable assumption was made that when Limited Liability Corporations (LLC) and other
business name information was found in the assessing information, it was quite likely that these
properties in foreclosure were commercial entities not individuals. The original justification for
using the foreclosure factor in this research was that each foreclosure point was an individual
decision maker in financial distress, whose financial distress could be alleviated marginally with
financial education. Since the reasons for a commercial foreclosure are likely different from
60

residential foreclosures, additional research is required to evaluate if it is appropriate to keep
commercial foreclosures in the foreclosure factor dataset. Fully understanding the presentation of
factor data is critical if informed decisions are to be made about how to implement financial
education events.

Figure 16: The City of Malden 2014 Zoning Map. All the zones are mutually exclusive and Residential A zoning has a
hollow thematic representation which covering the remaining city area
5.1.2 Census Factor and MCE Output Discussion
The MCE high value Area 2 (See Figure 15), is the largest census block group in the City of
Malden. After reviewing all the factors in the Area 2, the unemployment factor (Figure 14),
household income below poverty level factor (Figure 13), and the high rent expenditure factor
(Figure 7), all presented with high scores in the same area. These particular scores contributed
greatly to the final MCE score in Figure 15. When reviewing more closely this particular census
block group to identify where the individuals in this large census block group were living, the
zoning map (Figure 16) quickly helped identify that this area is primarily zoned for commercial and
61

industrial activities. Area 2 has 19.2 acres zoned for residential zones B/C and 204 acres zoned for
industrial and business activities. Accordingly, only 9.4% of Area 2 has residential housing for City
of Malden residents, each of which is eligible to respond to the ACS. Given the ACS factors listed
above are primarily responsible for defining both the need for financial education and the bounds
of Area 2, this research concludes that spatial representation of need for financial education in
Area 2 to be misleading. Those individuals that are in need of financial education are most likely
located in the non-commercial and business properties and not over the entire extent of the
census block group. Since zoning policies vary from city to town, researchers should be careful to
understand if mixed residential / commercial zoning is present in order to assess where individuals
in need of financial education are actually located.
From a city level perspective, the proximity of Malden to Boston likely has an impact on the
demographic distribution of Malden residents and consequently the need for financial education.
The City of Malden is separate from Boston, but is well within the "Boston Metro Area". Malden
has many mass transportation options for commuting into Boston. With an average home price of
$278,000 and an average price per square foot of $215, Malden has some of the most reasonable
real estate when compared to its neighboring communities and given its proximity to Boston
(Trulia 2014). This mix of characteristics has made it very attractive to a number of different
populations, building what is now a very diverse community. To support some of the most needy
in this diverse community, City of Malden has 577 public housing units spread across 8 housing
projects (Layfield 2014). Four of the larger housing projects in Malden are in areas expressing a
high need for financial education. This coincidence of housing projects with the high need for
financial education areas is a potential method for validating the MCE analysis.
62

The relationship transportation corridors have with high MCE value areas was explored for
additional insight into the MCE results. There are two rail lines that transverse north to south and
east to west through the City of Malden. There is also a state highway, Route 60, which passes east
to west through the city as well (Figure 1). Adjacent to these transportation corridors you also find
zoning for industrial use and a highway businesses (Figure 16). These industrial and business zoned
areas are then adjacent to the residential areas. When placing differently zone areas in close
proximity, the abutting zones are common areas where negative externalities likely result in
decreased property values (Ohls, Weisberg, and White 1974). When reviewing each individual
factor's high standardized value areas and the MCE high value areas, a pattern of adjacency
between these high value areas and transportation corridors is clear. The explanation for exactly
why this spatial relationship exists requires additional research. However, the combination of
reduced property values, close proximity to transit corridors, and the need for financial education
is interesting and another potential metric for evaluating the MCE output.
The focus on high MCE value areas should also be shifted to the lower MCE value end of
the spectrum to help validate the model is running as expected. When investigating the three
areas of the MCE that had lower than average standardized values, the south western most area
adjacent to the industrial branch rail line, it intersects almost completely with the Holy Cross
Cemetery. Given the inhabitants of this area, it logically makes sense why this area is being
assigned lower scores. The other two areas with low MCE values are in close proximity to public
parks and the tax assessing records indicate the last home sale price to be above the average
Malden home price of $278,000 (Trulia 2014). One of these areas is adjacent to the northern town
boundary and in the Northwestern corner of the city. The other area is west the Orange line and
adjacent to the western town boundary. The greater level of relative affluence in these areas and
63

potentially increased level of education would equate to a reduction in the standardized scores for
these low MCE value areas.

5.2 Implications and Limitations
The large question that remains after evaluating the results of this research is if the evenly
weighted approach is the best methodology, given the factors selected. Identifying which factor(s)
best indicate the greatest need for financial education is an open question to be studied. An
argument could be made that the foreclosure and tax delinquency points should have a higher
weight because they more closely relate to actual financial decision-making events. However, how
to establish the degree to which any of the factors are weighted is a much more difficult question.
Another one of the study weaknesses is the relatively small number of factors identified
that indicate a need for financial education. The Cutter et al. (2003) study started with over 200
factors and reduced the number based upon factor analysis. The methodologies used to identify
factors in their study are viable, but additional effort to study financial decision-making and
financial education efficacy could help inform the factor selection criteria.
This study sorted through all the readily available free data that might indicate a need for
financial education. During this effort, the sources of the private data that could support research
on financial education were identified. The Mortgage and Bankers Association keeps detailed loan
payment information. This mortgage data could further assist in identifying financially stable or
unstable individuals. A membership with the National Realtors Association brings access to data
about the financial health of a particular community or neighborhood. Credit card companies
possess individual spending habits and payment history. This information could help identify how
64

fiscally responsibility an individual is by providing a window into spending and payment habits.
This data could also help identify the individuals that are overleveraged with credit card debt. If
more financial data with good spatial resolution was available, the MCE output could provide a
more robust expression of need for financial education.

5.3 Spatial Significance and Next Steps
The Consumer Financial Protection Bureau estimated that federal, state, and local
government, financial institutions, nonprofit organizations, charitable foundations, and others in
2013 spent $670 million providing financial education services. Even though this amount of
spending equates to only about two dollars per American, there is currently a lack of financial
literacy and a great need for financial education in America today. Given this level of investment, it
was surprising that this research effort was unable to identify any attempts by governments or
researchers to develop spatial methods to understand where there was the greatest need for
financial education. This research captured readily available local and national data and applied a
straightforward MCE method to see if any patterns of interest were present to assist in the
identification of need for financial education. With additional methodological refinements from
new research, additional factor data from private sources, and an expanded study area, this
research has the potential to increase the effectiveness of financial education spending.
In addition, the MCE output presented in this research would benefit from review by local
experts from the City of Malden. Ideally, representatives with expert community knowledge could
provide additional local insight beyond the review provided in this thesis. Unfortunately, this
research effort was unable to complete an extensive review with the City of Malden staff.
65

However, the City of Malden GIS experts reviewed the research result. They were confused by the
identification of high need for financial education in areas with large volumes of business and
commercial zoning. When a discussion about the inclusion of commercial foreclosure incidents in
the KDE analysis was raised, the City GIS experts thought that was a mistake. They thought that
commercial foreclosures should be removed from the foreclosure dataset because individuals and
business can act very differently.
The next step forward from this research is to use the resulting MCE surface indicating the
relative need for financial education to organize the delivery of targeted financial education
offerings. Successful educational training events require partners that can organize and
communicate with the intended audience for the financial education training opportunity.
Partners should be able to identify the critical factors that their potential attendees most require,
like convenient hours or days for training and best learning settings. These kinds of critical training
factors can contain additional spatial information, which could be leveraged to identify the most
appropriate training locations. To gauge the potential success of new partners, considering the
proximity of each partner or their training events to the areas in high need of financial education
could be valuable indicator of success. The identification of free training locations like, religious
institutions, community centers and other common meeting places would lend itself to a proximity
analysis to each high need area. The training partners should also consider the availability of mass
transportation and handicapped accessibility to each training event, especially in an urban context.

66

5.4 Summary
In this thesis, the factors and the method for identifying the relative need for financial
education was developed and applied for the City of Malden, Massachusetts. The method was
able to identify a number of high need areas spread throughout the City of Malden. Other
communities could use these factors and apply this methodology as a starting point for addressing
their community's need of financial education services. The preliminary research review by the
City of Malden was positive, but further evaluation is required.
Decision makers today at many levels understand that there is a place for financial
education; however, the best delivery methods and format are still areas under investigation.
Financial education research is focused on quantifying the efficacy of financial education, while
also trying to learn from already successful efforts. As more information is collected by researchers
and qualitative financial education lessons documented, the factor identification research used in
this research could be further refined.
This research has also exposed the fact that very limited spatial analysis has been
completed in the pursuit of understanding the spatial patterns associated with financial decisions
and financial knowledge. Additional efforts to provide spatial context to our financial systems and
the decisions being made by individual financial actors could aid many different groups seeking to
make more strategic business decisions. Even though there is room to improve this study, the
factor selection and the spatial analysis methodology are applicable for use by other cities and
towns.

67

REFERENCES
Anderloni, L., E. Bacchiocchi, and D. Vandone. 2011. Household Financial Vulnerability: An Empirical
Analysis. Available at SSRN 1959801.
Angelides, P., and B. Thomas. 2011. The Financial Crisis Inquiry Report: Final Report of the National
Commission on the Causes of the Financial and Economic Crisis in the United States (Revised
Corrected Copy): Government Printing Office.
Apgar, W. C., M. Duda, R. N. Gorey, and H. P. Council. 2005. The municipal cost of foreclosures: A Chicago
case study. Minneapolis, MN: Homeownership Preservation Foundation.
Bank, U. S. 2013. Office of Corporate Citizenship and Financial Education - Placement of Finanical Education
Events, October 8th 2013.
Bernheim, B. D., and D. M. Garrett. 2003. The effects of financial education in the workplace: evidence from
a survey of households. Journal of Public Economics 87 (7):1487-1519.
Bocian, D. G., K. S. Ernst, and W. Li. 2008. Race, ethnicity and subprime home loan pricing. Journal of
Economics and Business 60 (1):110-124.
Bocian, D. G., W. Li, and K. S. Ernst. 2010. Foreclosures by Race and Ethnicity.
Bocian, D. G., and R. Zhai. 2005. Borrowers in High Minority Areas More Likely to Receive Prepayment
Penalties on Subprime Loans. Center for Responsible Lending 1.
Boshara, R., J. Gannon, L. Mandell, J. W. Phillips, and S. Sass. 2010. Consumer trends in the public, private,
and nonprofit sector. National Endowment for Financial Education Quarter Century Project.
Brault, M. W., and U. S. B. o. t. Census. 2012. Americans with disabilities: 2010: US Department of
Commerce, Economics and Statistics Administration, US Census Bureau.
Bricker, J. K., Arthur B; Moore, Kevin B; Sabelhaus, John. 2012. Changes in US family finances from 2007 to
2010: Evidence from the Survey of Consumer Finances. Federal Reserve Bulletin (June):1-80.
Brobeck, S. 2013. Savings Accounts: Their Charactoristics and Usefulness: Consumer Federation of America.
Campbell, J. Y. 2006. Household finance. The Journal of Finance 61 (4):1553-1604.
Carver, S. J. 1991. Integrating multi-criteria evaluation with geographical information systems. International
Journal of Geographical Information System 5 (3):321-339.
Clancy, M., M. Grinstein-Weiss, and M. Schreiner. 2001. Financial education and savings outcomes in
individual development accounts: Working Paper 01-2. St. Louis, MO: Center for Social
Development, Washington University.
Commission, F. L. a. E. 2011. Implmentation Plan 2011 Promoting Financial Success in the United States. In
National Strategy for Financial Literacy.
Commission, S. o. t. U. S. S. a. E. 2012. Study Regarding Financial Literacy Among Investors. Washington
D.C.: U.S. Securities and Exchange Commission.
Cordray, R. 2013. Financial Literacy Annual Report: Consumer Financial Protection Bureau.
Corporation, F. D. I. 2014. Delivery Systems for Financial Education in Theory and Practice 2004 [cited
03/10/2014 2014]. Available from http://www.fdic.gov/bank/analytical/fyi/2004/092204fyi.html.
Corporation, L. I. S. 2013. Zip Code Foreclosure Risk Score Methodology 1-4.
Cutter, S. L., B. J. Boruff, and W. L. Shirley. 2003. Social vulnerability to environmental hazards*. Social
science quarterly 84 (2):242-261.
Day, J. C., and E. C. Newburger. 2002. The big payoff: Educational attainment and synthetic estimates of
work-life earnings: US Department of Commerce, Economics and Statistics Administration, US
Census Bureau Washington, DC.
Dodd-Frank Wall Street Reform and Consumer Protection Act. 111th, July 21th 2010.
Duda, M., and W. Apgar. 2004. Mortgage foreclosure trends in Los Angeles: Patterns and policy issues:
Report prepared for Los Angeles Neighborhood Housing Services.
Durkin, T. A. 2000. Credit cards: Use and consumer attitudes, 1970-2000. Fed. Res. Bull. 86:623.
68

Dynan, K. E., and D. L. Kohn. 2007. The rise in US household indebtedness: Causes and consequences:
Divisions of Research & Statistics and Monetary Affairs, Federal Reserve Board.
Eastman, R. J., Weigen; Kyem, Peter; Toledano, James. 1995. Raster Procedures for Multi-Criteria/Multi-
0biective Decisions. Photogrammetric Engineering & Remote Sensing 61 (5):539-547.
Fernandes, D., J. G. Lynch Jr, and R. G. Netemeyer. 2014. Financial Literacy, Financial Education, and
Downstream Financial Behaviors. Management Science.
Foundation, T. M. 2013. The Financial Literacy Challenge And its Impact in our Community, 1-18.
Galante, C. J. 2013. Mortgagee Letter 2013-05, ed. U. D. o. H. a. U. Development. Washington DC.
Glick, R., and K. J. Lansing. 2011. Consumers and the economy, part I: Household credit and personal saving.
FRBSF Economic Letter 1.
Greenspan, A. 2005. The importance of financial education today. Social Education 69 (2):64-66.
Harnish, T. L. 2010. Boosting Financial Literacy in America: A Role for State Colleges and Universities.
Perspectives, American Association of State Colleges and Universities, Fall.
Hastings, J. S., B. C. Madrian, and W. L. Skimmyhorn. 2012. Financial literacy, financial education and
economic outcomes: National Bureau of Economic Research.
Hilgert, M. A., J. M. Hogarth, and S. G. Beverly. 2003. Household financial management: The connection
between knowledge and behavior. Fed. Res. Bull. 89:309.
Immergluck, D., and G. Smith. 2005. Measuring the Effect of Subprime Lending on Neighborhood
Foreclosures Evidence from Chicago. Urban Affairs Review 40 (3):362-389.
Insititue, E. P. 2011. Underemployment Rates, ed. E. P. Insititue.
Institute, N. D. 2012. Building a Better Economic Future for Americans with Diabilities, 1-4. Washington DC.
Jenks, G. F. 1967. The data model concept in statistical mapping. International yearbook of cartography 7
(1):186-190.
Kingsley, T. G., Smith, Robin E., Price, David 2009. The Impacts of Foreclosures on Families and
Communities: A Primer, ed. M. H. a. C. Center: Urban Institute.
Layfield, D. A. c. 2014. Malden Housing Authority 2014 [cited 25 2014]. Available from
http://affordablehousingonline.com/housing-authority/Massachusetts/Malden/MA022/.
Leland, J. 2009. Swindlers find growing market in foreclosures. New York Times (January 15) A 1.
Lin, Z., E. Rosenblatt, and V. W. Yao. 2009. Spillover effects of foreclosures on neighborhood property
values. The Journal of Real Estate Finance and Economics 38 (4):387-407.
Louis, F. R. B. o. S. 2014. FRED Economic Database, ed. F. R. B. o. S. Louis.
Lusardi, A. 2004. Saving and the effectiveness of financial education. Pension design and structure: New
lessons from behavioral finance:157-184.
———. 2008. Financial literacy: an essential tool for informed consumer choice?: National Bureau of
Economic Research.
Lusardi, A., and O. Mitchelli. 2007. Financial literacy and retirement preparedness: Evidence and
implications for financial education. Business Economics 42 (1):35-44.
Lusardi, A., and P. Tufano. 2009. Debt literacy, financial experiences, and overindebtedness: National
Bureau of Economic Research.
Martin, M. 2007. A literature review on the effectiveness of financial education: Federal Reserve Bank of
Richmond Working paper series.
McFarland, B. 2013. Retirment Plans Offered by 2013. Insider, November 2013.
McKee-Ryan, F. M., and J. Harvey. 2011. “I have a job, but...”: A review of underemployment. Journal of
Management 37 (4):962-996.
Morton, H. 2005. Financial literacy: A primer for policymakers. Paper read at National Conference of State
Legislatures. April.
O'Sullivan, D., and D. J. Unwin. 2003. Geographic information analysis: John Wiley & Sons.
OECD. 2006. The Importance of Financial Education. Organization for Economic Co-Operation and
Development Policy Brief July.
69

Ohls, J. C., R. C. Weisberg, and M. J. White. 1974. The effect of zoning on land value. Journal of Urban
Economics 1 (4):428-444.
Openshaw, S. 1983. The modifiable areal unit problem: Geo Books Norwich.
Pohekar, S., and M. Ramachandran. 2004. Application of multi-criteria decision making to sustainable
energy planning—a review. Renewable and Sustainable Energy Reviews 8 (4):365-381.
Raaijmakers, R., J. Krywkow, and A. van der Veen. 2008. Flood risk perceptions and spatial multi-criteria
analysis: an exploratory research for hazard mitigation. Natural hazards 46 (3):307-322.
Rogers, W. H., and W. Winter. 2009. The impact of foreclosures on neighboring housing sales. Journal of
Real Estate Research 31 (4):455-479.
Schwartz, M., and E. Wilson. 2008. Who Can Afford To Live in a Home?: A look at data from the 2006
American Community Survey. US Census Bureau.
Solutions, M. C. 2013. Consumer Financial Vulnerability Index South Africa.
Statitistics, B. o. L. 2013. Local Area Unemployment Statistitics In Alternative Measure of Labor
Underutilization for States, 2013 Annual Averages, ed. U. B. o. L. Statistics. Washington DC.
Stoms, D. 1987. Reasoning with uncertainty in intelligent geographic information systems. GIS 87:693-699.
Store, R., and J. Kangas. 2001. Integrating spatial multi-criteria evaluation and expert knowledge for GIS-
based habitat suitability modelling. Landscape and urban planning 55 (2):79-93.
2009. House Financial Services Treasury Secretary Timothy F. Geithner Testimony. 7/24/2009.
Trulia. 2014. Malden Market Trends 2014 [cited April 27th 2014]. Available from
http://www.trulia.com/real_estate/Malden-Massachusetts/market-trends/.
Vitt, L. A. 2005. Goodbye to complacency: Financial literacy education in the US 2000-2005: AARP.
Whitaker, S., and T. J. Fitzpatrick IV. 2011. The Impact of Vacant, Tax-Delinquent and Foreclosed Property
on Sales Prices of Neighboring Homes. Federal Reserve Bank of Cleveland Working Paper 11023.
Wiatrowski, W. J. 2012. Last Private Industry Pension Plans: A Visual Essay, The. Monthly Lab. Rev. 135:3.

Abstract (if available)

Abstract The goal of this thesis is to complete a raster‐based site suitability analysis that identifies varying levels of need for financial education. The research product is intended to help decision makers evaluate where to site financial education training events or invest in support programs for the communities or neighborhoods in need. The thesis begins by reviewing the current state of financial education and arguments for its application. Individual factors and supporting evidence that seek to identify individuals in need of financial education is organized. The data preparation steps and multi‐criteria evaluation (MCE) spatial methodology used to create the final research output is detailed. The MCE results identify multiple regions with a high level of need for financial education. A review of these high need areas coincident with city zoning and a variety of geographic features highlights additional spatial relationships of interest. The author concludes the research by outlining how the final output can support deciding on locations for financial education events.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Site suitability analysis: small-scale fixed axis ground mounted photovoltaic power plants in Fresno, CA

PDF

Site location suitability analysis for a smart grid network

PDF

A site suitability analysis for an inland port to service the ports of Los Angeles and Long Beach

PDF

Community gardens for social capital: a site suitability analysis in Akron, Ohio

PDF

Relocation bay: identifying a suitable site for the Tampa Bay Rays

PDF

Population disaggregation for trade area delineation in retail real estate site analysis

PDF

Integrating spatial visualization to improve public health understanding and communication

PDF

Installing public electric vehicle charging stations: a site suitability analysis in Los Angeles County, California

PDF

Predicting archaeological site locations in northeastern California’s High Desert using the Maxent model

PDF

Onshore wind power systems (ONSWPS): a GIS-based tool for preliminary site-suitability analysis

PDF

A fire insurance map geocoder for pre-earthquake San Francisco

PDF

Surface representations of rainfall at small extents: a study of rainfall mapping based on volunteered geographic information in Kona, Hawaii

PDF

Estimating populations at risk in data-poor environments: a geographically disaggregated analysis of Boko Haram terrorism 2009-2014

PDF

Selection of bridge location over the Merrimack River in southern New Hampshire: a comparison of site suitability assessments

PDF

A threat-based least-cost path decision support model for national security resource allocation along the US-Mexico border

PDF

Institutional inscription in the Minorcan Quarter of Saint Augustine, Florida

PDF

Harnessing GIST-enabled resources in the classroom: developing a Story Map for use with secondary students

PDF

Eye.Earth Pro (Beta v1.0): application development and spatial financial analysis utilizing the PESTELM framework

PDF

Developing and implementing a GIS-based framework to identify optimal locations for clean water wells in sub-Saharan Africa

PDF

Providing a new low-cost primary care facility for under-served communities: a site suitability analysis for Service Planning Area 6 in Los Angeles County, California

Asset Metadata

Creator Breeding, David James (author)

Core Title Locating the need for financial education

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Geographic Information Science and Technology

Publication Date 07/01/2014

Defense Date 05/09/2014

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

Tag financial education,MCE,multi-criteria evaluation,OAI-PMH Harvest,site suitability

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Oda, Katsuhiko (Kirk) (committee chair)

Creator Email david.breeding84@gmail.com,daville7@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-428401

Unique identifier UC11286763

Identifier etd-BreedingDa-2602.pdf (filename),usctheses-c3-428401 (legacy record id)

Legacy Identifier etd-BreedingDa-2602.pdf

Dmrecord 428401

Document Type Thesis

Format application/pdf (imt)

Rights Breeding, David James

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