Sanford Journal of Public Policy

Sanford Journal of Public Policy
Vol. 4 | Spring 2013

Vol..4 Spring.2013
Mission The Sanford Journal of Public Policy was created in 2009 as a forum for public policy students and professionals to contribute to the current policy discourse through insightful analysis and innovative solutions. About The Sanford Journal is run by graduate students from the Sanford School of Public Policy and across the Duke University community. The Sanford Journal is published annually and is available in print, ebook, and pdf formats. The Sanford Journal solicits articles from across the spectrum of public policy in a variety of formats, including policy research and position papers, issue briefs, and opinion pieces. The Sanford Journal website features blog posts from our staff contributors and select faculty addressing contemporary policy issues. It is a forum where public policy students and practitioners can stay connected and express their own views on todays policy challenges.
Visit us online at http://sites.duke.edu/sjpp/
The Sanford Journal of Public Policy is a graduate student-run publication. The views and opinions expressed in the Journal are the authors own and do not necessarily represent the views of the Sanford School of Public Policy or Duke University.
Sanford Journal of Public Policy 201 Science Drive Box 90243 Durham, NC 27708-0239 sjpp@duke.edu
Editors-in-Chief
Jeff Bartelli Rachel Leven
Managing Editor
Drew Monger
Senior Print Editors

Mariel Beasley Sondra Appelson
Submissions Coordinator
Daniel Jasper
Multimedia Editor
Zarak Khan
Business Manager
Chris Marsicano
Assistant Webmaster
Hal Beresford
Layout Editor
Ankur Verma
Staff Editors
Adam Lutterloh Andrew Olson Colin Walsh Greg McDonald Hal Beresford Jarrett Lane Lucas Westmaas Mau Pantoja Michael Chiulli Natalie Trebes Nina Brooks Rachel Goldstein Rachel Pittenger Robert Mason Sarah Grantham Smriti Sharma Sonia Sekhar Tameeka Norton Tayeba Maktabi Wil Morrison
iii
Acknowledgements
This year, the Sanford Journal has benefited from the commitment and assistance of many people at the Sanford School and Duke University. Many faculty members offered their time and experience to help improve the Sanford Journal and we are extremely grateful to each of them for their help. We wish to thank our faculty advisor, Elizabeth Frankenberg, for her guidance and supervision in realizing the goals of the Sanford Journal. Dr. Frankeberg has provided critical support for the Sanford Journal since its inception. Her continued leadership helped to guide the growth of the Sanford Journal. This year several professors were involved in reviewing the excellent submissions we received. We would like to thank Mac McCorkle (Esq.), Dr. Joel Rosch, Dr. Bruce Jentleson, Dr. David Schanzer, Dr. Charlie Clotfelter, and Dr. Jay Hamilton for their advice and flexibility. Their participation was a critical component to elevating this years content. We look forward to standardizing this review process in the future. We are also grateful to Nick Carnes for contributing to the Sanford Journal blog and for encouraging his students to submit their work to the Sanford Journal. Professor Carnes dedication to the students of the Sanford School of Public Policy was a source of inspiration for the Sanford Journal staff. Several members of the Sanford School administration offered us key support throughout the year. Assistant Dean Michael Case offered us valuable support and resources to promote our mission and scope. Helene McAdams graciously provided us with the assistance we needed to accomplish our goals. We also wish to thank the Sanford Student Council and the Duke University Graduate and Professional School Student Council (GPSC) for their continued funding and support for the mission of the Sanford Journal. We also wish to thank Ankur Verma for his assistance in the layout and design of this volume of the Sanford Journal. We thank all the authors who submitted to the Sanford Journal. Many examples of high quality research and policy analysis came to the Sanford Journal and we struggled to select only six papers for inclusion in Volume IV. Finally, the Sanford Journal would not have succeeded without the tireless efforts of its volunteer staff. The student leaders and team members of the Sanford Journal guaranteed the production of a high quality and relevant policy publication. Through their tireless efforts the Journal blossomed from an annual project into a Sanford institution.
v
Table of Contents
Foreword Dean Bruce Kuniholm.......................................................................................... ix Articles F is for Foreclosed: The Effects of the Housing Crash on NC Student Achievement....................................................................................... 3 Walker Swain Incentivizing Responsible Small-Dollar Lending in Low-Income Communities .................................................................................. 25 Megha Bansal Negotiating Online Privacy Boundaries: Self-Revelation in the Facebook Generation ................................................................................. 47 Katherine Bies Recommendations for Migrant Labor Reform in Qatar: Fulfilling a U.S. Foreign Policy Objective of Greater Respect for Human Rights In Gulf Cooperation Council Countries ................................. 69 Kharmika Tillery Gideons Bullhorn: Sounding a Louder, Clearer Call for a Civil Right to Counsel .......................................................................................87 Jonathan D. Kelley The Lens of Law Enforcement: A Geospatial Statistical Program Evaluation of Denvers HALO Camera Surveillance System .......................... 109 John Papazian
vii
Foreword
Most real-world problems have a policy dimension. It has been my privilege, as Dean of the Sanford School of Public Policy, to watch our students grapple with national and global issues and develop new policy prescriptions for addressing some of the most pressing problems of our time. I was especially pleased to see graduate students launch the Sanford Journal of Public Policy in the spring of 2010. It provides a needed showcase for research and ideas from young leaders around the nation. The Sanford Journal is run by graduate students who donate their time to thinking hard about complex policy problems. Every year they work with academics, students, and professionals to produce a truly impressive collection of articles. The staff collaborate with faculty (something we value highly at Sanford) and contribute timely pieces to a biweekly blogsomething few other policy journal staff do. Last fall, the Sanford Journal reached out across Duke University and added staff from the law and divinity schools. The staff also worked with Sanford professors to review print articles for accuracy and to assess their contributions to policy analysis. The result is a collection of articles that are thought-provoking and policy relevant. In F is for Foreclosed and Incentivizing Responsible Small-Dollar Lending in Low-Income Communities, two Duke alumni address key financial challenges for the nations poorest populations. In the former article, Walker Swain, a 2012 Sanford MPP alumnus and a PhD candidate at Vanderbilt University uses county-level data to link student failure in the classroom with the stressful experience of home foreclosure. Meanwhile, Megha Bansal, a fellow 2012 alumna and Deloitte consultant, seeks to identify better financial service programs for low-income communities. Jonathan Kelley and John Papazian take on sensitive questions in civil law and law enforcement. In Gideons Bullhorn, Kelley, a Duke J.D. candidate, advocates for the extension of the right-to-counsel in civil courts. In The Lens of Law Enforcement, Papazian, MEM/MPP 12, evaluates the effectiveness of camera surveillance systems in law enforcement, finding evidence of crime reduction for certain offenses. As the United States experiences major shifts in its global influence, new approaches to foreign policy are imperative. Foreign Service Officer Kharmika Tillery, MPP 12, explains how the United States can fulfill its foreign policy objective of greater respect for human rights in her article, Recommendations for Migrant Labor Reform in Qatar.
ix
The Sanford Journal strives to be a forum for policy discussion across disciplines and institutions, which is why I am particularly excited to see the work of Katherine Bies, a star alumna from the University of Michigan who is completing her tenure with Teach for America. Bies article Negotiating Online Privacy Boundaries takes a fresh look at privacy laws and perceptions of personal boundaries for the Net Generation. Together, these articles present an academically rigorous approach to questions of real and practical consequence. They show a strong commitment to addressing the countrys problems. Along with the Sanford Journal editors, I invite you to hear the voices of a new generation of leaders in policy analysis.
Bruce R. Kuniholm, Dean
Articles
F is for Foreclosed: The Effects of the Housing Crash on North Carolina Student Achievement1
Walker Swain
Walker Swain is a doctoral student of Leadership and Policy Studies at Peabody College of Education and Human Development, Vanderbilt University. His research focuses on the impacts of social policy on schools, teacher labor markets, school integration, and choice. He previously worked as an education policy consultant and as a middle school teacher. He also holds a Master of Public Policy from the Sanford School of Public Policy at Duke University.
Abstract Since the passage of No Child Left Behind, elevating student achievement on standardized tests has become the most prominent goal of our education system. And since the onset of the Great Recession, the housing crash and subsequent foreclosure crisis have been at the forefront of political and economic discourse. Yet, social scientists and policymakers alike have strikingly little empirical information about how the latter has affected the former. In this paper I examine the effects of the housing crash on North Carolina public schools. Specifically, I analyze whether rises in county level home foreclosure rates hurt student academic achievement and whether they left more students economically vulnerable. Controlling for county characteristics that are constant over time and adjusting for statewide trends in test scores, increases in county level foreclosures are significantly associated with drops in both math and reading achievement. Coefficients from the basic ordinary least squares (OLS) regression models indicate that an increase of one home foreclosure per 100 county residents is associated with decrease of approximately .3 standard deviations in the percent proficient for math and reading. The negative relationship remains significant after controlling for county level unemployment and is strongest for black students. The specific effects of home foreclosures
1 I am grateful for the guidance of Helen Ladd in helping shape this project. I would also like to thank Anna Gassman-Pines and Jacob Vigdor for early conversations that helped me formulate my research question and methods.Thanks to Nick Carnes and Julia Howley for the frequent feedback and support.
Swain: F is for Foreclosed
cannot be separated from other statewide trends during the recession years. However, the data suggest that districts hit harder by the recession performed quantifiably worse in both math and reading. These findings have implications for the fairness and efficiency of policies that base teacher compensation or employment decisions on simple calculations of student value-added scores. They also support policies that would provide additional aid to high foreclosure districts to help students in foreclosed homes stay at their schools (i.e. covering transportation costs) or ease their transitions to new schools (i.e. counseling and records management). And so the question is whether all of usas citizens, and as parentsare willing to do whats necessary to give every child a chance to succeed. That responsibility begins not in our classrooms, but in our homes and communities. President Barack Obama, State of the Union Address 2011 Introduction In this paper I examine the effects of the Great Recession on North Carolina public schools. Specifically, I examine how the rapid increase in housing foreclosures has hurt North Carolina students performance on standardized tests. I also explore the relationship between county rates of foreclosure and the percentage of students classified as economically disadvantaged, highlighting the disparate effects of the economic downturn. These questions are of interest for two reasons. First, before the housing market had remotely stabilized, the President of the United States cited international test scores as our generations Sputnik moment in the State of the Union address and argued that the fate of our nation depends on elevating student achievement. Yet few scholars have sought to explain how the recent surge in household and community instability has damaged students chances at academic success. If foreclosures hurt student achievement, school policy and housing policy should be better coordinated. Second, this study adds to a growing literature that questions the use of high stakes test scores as accountability measures in a time of widespread economic instability. While many researchers who criticize test score accountability have shed light on the relationship between poverty
4
and academic success, this study focuses primarily on a group of students who are not traditionally considered economically disadvantaged - the children of homeowners.1 In so doing, it highlights the non-fixed nature of students socioeconomic status and the subsequent limitations of standard value-added test score assessments of teacher effectiveness. I use North Carolina county level foreclosure data from the Office of the Administrator of the Courts and district level End of Grade Test (EOG) Scores aggregated to the county level from 1999 to 2010 to examine the relationship between achievement and foreclosure rates, while controlling for county level characteristics (fixed effects) and test score trends over time. I also examine the link between the housing crash and students economic wellbeing using data from the Department of Public Instruction on students qualifying for free and reduced price lunch. This indicator helps illuminate one of the mechanisms by which the housing crash has affected students lives and potentially their learning. By combining these two basic metrics, I develop a rough sketch of how the housing crash has affected students across the state. In the sections that follow, I provide a brief background on the foreclosure crisis in North Carolina. Next, I provide a review of relevant literature that informed my hypotheses about how foreclosures could impact student achievement. I then describe the data and methods used to reach my preliminary conclusions. Finally, I present the findings and discuss the policy implications of achievement being injured by high rates of foreclosures in a community. Background Since 2006, housing markets in the United States have been in a state of turmoil. Between 2006 and 2009, equity in U.S. real estate decreased by $6.8 trillion (Dastrup & Betts 2012). Across the country, banks foreclosed on record numbers of American homes. The Center for Responsible Lending estimates lenders completed 2.5 million foreclosures between January 2007 and the end of 2009, the vast majority of which were on owner-occupied properties (Bocian et al. 2010). Approximately 56 percent of foreclosed families were white and non-Hispanic. However, African-American and Latino families were substantially overrepresented relative to their share of mortgage holdings (Bocian et al. 2010).
5
Foreclosure rates have risen dramatically in most NC counties; in 2000, the average North Carolina county saw homes foreclosed at a rate of approximately 20 per 10,000 residents. By 2010, that rate had more than tripled, with the average county in North Carolina experiencing approximately 62 housing foreclosures per 10,000 residents. There has also been considerable variation in the severity of the foreclosure epidemic, with county per capita foreclosure rates in 2010 ranging from less than 30 to more than 250 per 10,000 residents. In 2008 the Brookings Institute estimated that nearly 40,000 children in North Carolina experienced home foreclosures in the previous year (Lovell & Isaacs 2008). To put that figure in perspective, if those children were a school district, it would be the sixth largest district in the state. Since 2008, each year set a new record for foreclosure filings, including nearly 70,000 across the state in 2010 (see figure 1). That same year, banks foreclosed more than 10,000 homes in Mecklenburg County alone. County foreclosure rates were positively correlated with higher levels of median family income, indicating they were not isolated to poor counties. As of July 2011, the trend toward rapidly-increasing rates of foreclosure had yet to subside.
Figure 1: North Carolina foreclosures per 12 months from 1998 to 2011
Source: Filings with North Carolina Administrative Office of the Courts (2011) 6
As a proxy for the impact of foreclosures on students economic wellbeing, I examine the changes in the percentage of students who participate in the National School Lunch Program. Students from families with income at or below 130 percent of the federal poverty line (FPL) are eligible for free lunch and families with incomes between 130-185 percent qualify for reduced price (USDA 2012). The North Carolina Department of Public Instruction classifies these students collectively as economically disadvantaged for reporting purposes. No Child Left Behind (NCLB) required districts to report disaggregated test score data that include the number of students that qualify for free and reduced lunch and how well these students collectively perform on End of Grade exams (EOGs). With these annual reports, I can track approximate changes in the number of students experiencing economic stress. In this study I restrict my calculation rates for grades three through eight, where all students participate in EOGs. In 2011, nearly 55 percent of tested students were designated economically disadvantaged, marking a nine-percentage point increase in less than a decade. In the past three years alone the number of students participating in the program statewide increased by over 97,000. This spike was more than quadruple the increase from the previous three years (an increase of 23,121 students). As Figure 2 indicates, the Great Recession left a substantially higher portion of North Carolina students in difficult financial situations.
Figure 2: NC public school students participation in the National School Lunch Program
Economically Disadvantaged Students, Grades 3-8
Propotion of Students
Year
Source: Authors calculations based on data from North Carolina Department of Public Instruction (2011).2 7
Hypotheses & Research It is impossible to observe the effects of individual foreclosures on specific students using county level data. However, I can estimate the effect of community level housing foreclosures on district level economic wellbeing, overall student achievement, and the achievement of specific subgroups. Previous research suggests a few basic hypotheses. First, I expect that increased foreclosure rates adversely affect overall student achievement on End of Grade exams at the county level. Foreclosures often force students to leave their homes.This process can radically disrupt a students social life and academic performance, particularly when families are forced to temporarily live with friends or family, or seek refuge in homeless shelters (Goux & Maurin 2005; Maxwell 2003). In cases where foreclosures force families to leave neighborhoods, the process disturbs the social networks of both the children who leave and those left behind (Gruman et al. 2008). The process of changing schools can also pose a host of challenges for students as they adjust to the new set of peers, classroom culture, and curriculum (Alexander et al. 1996; Lash & Kirkpatrick 1994; Mehana & Reynolds 2004; Schwartz et al. 2007). Even when students do not move out of their house or neighborhood, the housing instability associated with foreclosures can impose a substantial amount of stress on students and their families. Qualitative research has documented that this stress can cause depression, affect students ability to focus, increase absences, and hurt overall academic performance (Kingsley et al. 2009). Even the anticipation of housing instability can have dramatic effects on the mental wellbeing of students. Previous research has demonstrated that county level effects of economic hardship, in the form of mass job loss, can negatively affect aggregate student achievement (Ananat et al.2011). Several studies have documented a strong correlation between foreclosure rates and student test scores using residence maps linked to specific schools. The largest was an analysis of New York public schools that showed elementary school math and reading proficiency levels were ten percent lower for schools whose students reside in the neighborhoods in the top quartile of foreclosures and those who live in neighborhoods in the bottom quartile of foreclosures (Been et al. 2010). In a recent study, researchers in San Diego were able to match student address records from the school district to public records of mortgage defaults, and then analyze the effects of default on test scores.
8
The sample size of over 190,000 allowed the researchers to control for a wide array of observable and unobservable student, family, school, and teacher characteristics, using district and county records as well as fixed effects. Controlling for a range of potential confounding variables, the authors found statistically significant decreases in math scores and increases in absenteeism for students whose homes were foreclosed. They did not, however, find any effects on reading scores or effects on students whose parents were renters of foreclosed homes (Dastrup& Betts 2012). Second, counties with higher foreclosure rates should experience increases in the percentage of students designated as economically disadvantaged. Foreclosure rates are likely an indicator of economic stresses like loss of wealth and borrowing power, or increased unemployment and loss of earnings. Thus, increased foreclosure rates and the economic hardship they represent leave more families dependent on the National School Lunch Program. Finally, because students whose parents experienced foreclosures would not typically qualify for free or reduced lunch (Wachter& Smith 2010), increases in housing foreclosure rates will have worse effects on the test scores of students who are not traditionally economically disadvantaged. Black students may also experience greater losses, because their parents homes were foreclosed at disproportionately high rates (Bocian et al. 2010; Been et al. 2010). Students whose families are not directly affected by foreclosures may suffer from a loss of resources because of lowered property values (a phenomenon referred to as the contagion effect) or from having their instructional time disrupted by adding new students who were forced to change schools because their families were affected (Hanushek et al. 2004; Mehana& Reynolds 2004). Finally, students (especially adolescents) living in high foreclosure communities may also be affected by changes in the behavior of peers whose homes are foreclosed, which can disrupt academic performance (Calv-Armengol et al. 2009; Weinberg 2006). Data & Methodology Data used in this study come from five distinct sources. My primary independent variable, county level foreclosure numbers, comes from the Administrative Office of the Courts by way of the North Carolina Justice
9
Data
Center. This dataset is composed of the number of housing foreclosures filed in each month in each county from January 1998 to July 2011. To normalize these numbers for population size, I divided the county foreclosure numbers by the annual county population totals. I used this unconventional rate (foreclosures per resident), rather than foreclosures per homeowner or per mortgage, to better capture the extent to which the community is affected.3 Annual county population estimates were attained from the North Carolina Office of State Budget and Management, for every county between the years 2000 and 2010. I used demographic data (race, ethnicity, median income, renter occupied housing, household vacancies, and male unemployment rates) on the counties from the 2010 US Census to describe the areas affected by housing foreclosures (US Census Bureau). I incorporated data on the same variables from the 2000 Census to compare rates of change for the various demographic characteristics and to identify variables that change dramatically and differentially over time, such as the percent of the population that is Hispanic or unemployed. To measure student achievement, I used district level EOG Test Scores, which I accessed from the North Carolina Department of Instruction. I specifically examined the third through eighth grade composite math and reading comprehension assessments. I used the proficiency percentage for each of the 115 districts from 2002 to 2011 for third through eighth grade students in math and reading, broken down by gender, race, and free and reduced lunch status, to assess the effects on various subgroups. This data should be largely representative given all students at grade levels three and eight are required to take the math and reading comprehension assessments and all school districts are required to report scores to the department.4 I used eighth grade scores, rather than high school End of Course (EOC) scores, because the latter are likely to be affected by selection biases caused by dropouts or students taking different math subject tests. To account for variation in county level unemployment, I used annual county level employment (2001 through 2010) from the Bureau of Labor Statistics. I merged this dataset with the foreclosures, demographics and test score data to complete the panel set. My unit of analysis is the county. Most of the 100 counties in North Carolina contain only one school district. For the counties that contain multiple school districts, I use a pupil-weighted average
10
of district level test scores and percentages of economically disadvantaged students. Methodology To measure the effects of housing foreclosures on student test scores, I estimated two empirical models for both reading and math test scores. The estimated equation for Model 1 takes the following form: Scorect = 1 foreclosureratect-1 + county fixed effects+ ect
(1)
In this model, the coefficient 1 estimates the difference in test scores for counties with increases in foreclosure rates, controlling for county characteristics and trends in test scores over time.5 In the equation, t refers to the year (2002-2011) and the c refers to counties. The outcome of interest, Scorect, represents the weighted county average percent proficiencies for each year. In the regressions, the percent proficiency is replaced with a yearly mean-referenced Z-score. By representing the districts percent proficiency in math or reading as a score relative to the yearly average, I account for statewide variation in test scores based on the difficulty of the test or any other statewide trend. The foreclosure rate referenced in this model represents the county foreclosures per capita in the previous year t-1.6 County fixed effects control for all observed and unobservable characteristics of the county that do not change over time. These account for the possibility that counties that have higher foreclosure rates may also have lower test scores on average. The second regression model follows the same basic form as the first, except it incorporates time varying annual county level unemployment rates. The recent rise in unemployment could affect both housing stability and student achievement. When a person loses his or her job it makes it difficult to keep up mortgage or rent payments. It is only logical that instability of employment could, in turn, make housing less stable, and recent studies have shown county level unemployment rates adversely affect student performance on standardized tests (Ananat et al. 2011). This model accounts for the possibility that counties that experience increases in foreclosure rates may also have experienced increases in unemployment that are actually the cause of drops in student test scores. The estimated equation for Model 2 takes the following form: Scorect = 1 foreclosureratect-1 + 2 percent unemployedct-1 + county fixed effects + ect
(2)
11
Effects of Foreclosures on Student Economic Status I also examined the effect of the foreclosures that counties experienced in the previous year on levels of the percent of students that are classified as economically disadvantaged (qualifying for free or reduced lunch) by estimating the following equation:
(3) Percent FRLct = 1 foreclosurect-1 + 2 unemployedct-1 + 3 percent
Hispanicct + county fixed effects + year fixed effects + ect
This equation mirrors the second model for student achievement but changes the outcome of interest and adds the time varying county characteristic of percent Hispanic. Here, the Percent FRLct variable represents the proportion of third through eighth grade students in a county that participate in the National School Lunch program. The percent Hispanic variable controls for the percent of students in a county of Hispanic ethnicity a time-varying characteristic that has changed dramatically over the last ten years and differentially across counties. The Hispanic population in North Carolina grew extremely rapidly over the past decade and may have influenced both housing stability and student economic status. Recent immigrants, with fewer resources at their disposal, face unique housing difficulties and generally have low incomes. For the sake of clarity, I also present the results of economic status regressions with less restrictive forms (no control for unemployment and no year fixed effects) mirroring those of Model 1 and 2 for the student achievement data. Results Controlling for county fixed effects and annual trends in test scores finds increases in foreclosure rates both decreased math and reading proficiency levels and increased the number of students classified as economically disadvantaged. The magnitude of these associations was greatest among black students and equally strong among male and female students. Tables 1 and 2 report the basic results for math and reading respectively. The results for both follow the same general pattern. The first column represents the results of the basic model in which I control for county fixed characteristics and standardize proficiency percentages to adjust for
12
yearly trends in student achievement. The coefficient on foreclosure rate in the first model indicates that an increase of one home foreclosure per 100 county residents is associated with decrease of approximately .3 standard deviations in the percent proficient for math and reading. The large R-Squared value (over .85) for regression Model 1 indicates that the model has a strong fit with the data in both math and reading test scores. The coefficients reported for Model 1 indicate that in counties with increases in foreclosure rates, the percent of students scoring proficient in either subject dropped relative to the mean on the following years EOG tests. The magnitude of the effect doubles when I run the regression only including black students proficiency (Table 3). The effect size also appears to be larger when only girls are included, particularly with regards to reading proficiency. The second column of each table reports the results for regression Model 2 where I add a control for county level unemployment rates. The decrease in the coefficient for foreclosure rates indicates that, not surprisingly, there is some overlap in these two variables that measure different aspects of hard times in the Great Recession. It is notable, however, that the effect
Table 1: Foreclosures Previous Year on 3-8 Composite Math Score (Z score)
Model 1 (SE) Foreclosure Rate Percent Unemployed Constant Model Characteristics County Fixed Effects Controls for Unemployment Year Fixed Effects Observations R-squared -.343*** (.089) .948*** (.026) (1) yes no no 880 .861 Model 2 (SE) -.283*** (.010) -.007 (.005) .380*** (.048) (2) yes yes no 880 .862
Notes. Robust standard errors (clustered by county) are in parentheses. Sample size for each model is based on 88 counties over ten years. Each value represents a coefficient from an ordinary least squares (OLS) regression model. The values in the row labeled foreclosure rate represent the estimated effect of an increase in foreclosures per resident on student achievement (county average percent proficient). 2002 used as reference year. *** p<.01, ** p<.05, * p<.1 Source: NC Department of Public Instruction, NC Office of the Courts, Bureau of labor Statistics.
13
size of the foreclosures alone remains significant. This suggests that home foreclosures have a significant relationship with student achievement distinct from that of unemployment. The unemployment rates appear to have a larger impact on reading scores while the effect of foreclosures on math scores is relatively unchanged by the added control. The larger effect on math scores is consistent with prior findings (Dastrup & Betts 2012). Subgroup Analysis Table 3 presents the disaggregated results of foreclosure rates on student math achievement. Black students are the only racial subgroup with drops in test scores associated with increased rates of foreclosures controlling for county unemployment rates and fixed effects. All subgroups exhibit a significant negative association between county unemployment rates and reading scores with the largest effects on black students and girls.7 Female students also appear to have stronger associations between increased foreclosure rates and tests scores, particularly in math.8
Table 2: Foreclosures Previous Year on 3-8 Composite Reading Score (Z score)
Model 1 (SE) Foreclosure Rate Percent Unemployed Constant Model Characteristics County Fixed Effects Controls for Unemployment Year Fixed Effects Observations R-squared -.307*** (.0679) .791*** (.019) (1) yes no no 880 .866 Model 2 (SE) -.146** (.070) -.020*** (.003) .254*** (.038) (2) yes yes no 880 .874
Notes. Robust standard errors (clustered by county) are in parentheses. Sample size for each model is based on 88 counties over ten years. Each value represents a coefficient from an ordinary least squares (OLS) regression model. The values in the row labeled foreclosure rate represent the estimated effect of an increase in foreclosures per resident on student achievement (county average percent proficient). 2002 used as reference year. *** p<.01, ** p<.05, * p<.1 Source: NC Department of Public Instruction, NC Office of the Courts, Bureau of labor Statistics.
14
Table 3: Disaggregated Effects of Foreclosure Rates on Student Math Achievement

Race & Ethnicity Subgroup Foreclosure Rate All Students -.283*** (.101) Percent Unemployed -.007 (.005) Constant .380*** (.048) Observations R-squared 880 .862 Black -.739*** (.185) .003 (.010) .304*** (.074) 850 .539 White -.109 (.088) -.013*** (.004) 1.051*** (.043) 880 .728 Hispanic .383 (.228) .019** (.009) .370*** (.077) 873 .524 Gender Female -.349*** (.106) -.003 (.005) .445*** (.050) 880 .816 Male -.215** (.099) -.012** (.005) .318*** (.050) 880 .877 FRL -.166 (.134) .007 (.006) -.765*** (.047) 879 .802
Notes. Robust standard errors (clustered by county) are in parentheses. Sample size for each model is based on 88 counties over ten years. Each value represents a coefficient from an ordinary least squares (OLS) regression model. The values in the row labeled foreclosure rate represent the estimated effect of an increase in foreclosures per resident on student achievement (normalized county average percent proficient). The regression model controls for county fixed effects. *** p<.01, ** p<.05, * p<.1. Source: NC Department of Public Instruction, NC Office of the Courts, Bureau of labor Statistics
Effects on the number of students qualifying for Free or Reduced price Lunch Increases in county level foreclosure rates are associated with increases in the percentage of students that are economically disadvantaged, while controlling for any fixed characteristics of the counties, any statewide changes that occur year to year, and the time variant influences of county unemployment rates. Table 4 reports the results of the regression analysis on the effects of county level housing foreclosures on students economic status. All three columns report results for the same basic panel regression model, which controls for county fixed effects as well as changes in the percentage of Hispanic students in the county. However, the second column adds a control for unemployment rate and the third adds controls for year fixed effects. All coefficients represent the increase in percent of students qualifying for free and reduced priced lunch associated with the increase in the variables listed on the left. For example, the coefficient for foreclosure rate in the third column indicates that every per capita increase in foreclosure rates is associated with a seven percent increase in students classified as economically disadvantaged the following year.
15
Swain: Swain: F F is is for for Foreclosed Foreclosed
The purpose of presenting the data in this manner is to isolate the effects of the foreclosures from the broader range of hardships or demographic trends. The only difference between Model 1 and Model 2 is the incorporation of unemployment data. The drop in the magnitude of the coefficient represents the amount of the effect that is actually explained by unemployment correlating with foreclosures. Model 3 controls for the statewide recession trends using year fixed effects, and although the magnitude of the effect of foreclosures is substantially smaller, it remains statistically significant. This finding indicates that students socioeconomic status and the aggregate socioeconomic statuses of districts are not fixed characteristics and appear to be hindered by the home foreclosures in a district. Discussion & Conclusions The housing crash and subsequent Great Recession expelled many students from foreclosed homes and left numerous parents searching for
Table 4: Effect of Foreclosures on Student Economic Status (Percent Free and Reduced Lunch) Model 1 Foreclosure Rate Percent Unemployed Percent Hispanic Students Constant Model Characteristics County Fixed Effects Percent Hispanic Normalized Test Scores Controls for Unemployment Year Fixed Effects Observations 14.574*** (2.151) 1.057*** (.115) .220*** (.008) (1) yes yes yes no no 872 Model 2 11.090*** (2.279) .006*** (.001) .886*** (.102) .201*** (.008) (2) yes yes yes yes no 872 Model 3 7.089*** (2.210) .008*** (.003) .396** (.180) .166*** (.013) (3) yes yes yes yes yes 872
R-squared .928 .937 .943 Notes. Robust standard errors (clustered by county) are in parentheses. Sample size for each model is based on 88 counties over ten years. Each value represents a coefficient from an ordinary least squares (OLS) regression model. The values in the row labeled foreclosure rate represent the estimated effect of an increase in foreclosures per resident on student achievement (county average percent proficient). 2002 used as reference year. *** p<.01, ** p<.05, * p<.1 Source: NC Department of Public Instruction, NC Office of the Courts, Bureau of labor Statistics. 16
scarce employment opportunities. It also created major revenue shortfalls for state governments in places like North Carolina, forcing lawmakers to make serious cuts to education funding. More students were economically vulnerable than at any time since the Great Depression, and the teachers they looked to for stability and continuity were in danger of mass layoffs. My results suggest that increased rates of home foreclosures in North Carolina counties are significantly associated with drops in percentage of students proficient in math and reading as well as increases in percentages of students who are designated economically disadvantaged. They also indicate that the harm was greatest among black students, a group that was already struggling to keep pace before the economy turned south. My analysis indicates that the differences in foreclosure rates across counties serve as indicators of a broader range of economic hardships that some counties experienced as a result of the Great Recession. Thus, with the data I have, I cannot conclude that interventions that target stabilization of housing for mortgage holders would improve overall student test scores. However, the data do suggest that a significant portion of the variation in districts relative progress towards meeting NCLB accountability goals is driven by the severity of the districts experience of the Great Recession. This preliminary finding raises questions about the fairness of holding the teachers and administration of schools in different districts accountable to the same standard without regard for the circumstances in which they operate. It highlights the problems raised by unpredictable influences on student achievement even when comparing a students gains to his or her average progress from previous years. It is good politics to say that there is no excuse for not maximizing all students academic achievement. Yet, it may not be wise policy to fire or cut the pay of teachers and administrators who fail to adequately progress students in the midst of an economic hurricane. In the wake of the housing crash, the federal government provided stopgap funding to avoid mass teacher layoffs, but it gave no additional funds or requirements for schools to implement transition counseling for the increased number of students transferring to new schools. Its guidelines for the uses of the funding also limited any potential attempts at reapportioning of resources to adjust for the differential impacts of the recession on
17
counties within states. Furthermore, the funds that the federal government did provide as budget stopgaps to states were inadequate to close recessionfueled deficits. My preliminary findings suggest that students academic achievement would benefit from policymakers considering a broader range of policy interventions addressing hardships they face outside of school. They support calls for additional aid to high foreclosure districts in order to help students in foreclosed homes stay at their schools (i.e. covering transportation costs) or ease their transitions to new schools (i.e. counseling and records management). They also imply the need for greater contextualization in teachers and schools evaluations. Though federal education funds undoubtedly saved teacher jobs, the restrictions tied to the stabilization funds and Race to the Top grants limited states capacities to deal with the recessions broad range of impacts and forced cuts to social services that might have cushioned the blow to students and their families. In light of these shortcomings, it is difficult to characterize the federal response to the Great Recessions effects on schools as anything but wholly inadequate. If the students of North Carolina are to overcome the new obstacles to academic success brought on by the great recession, our policymakers should consider a bolder, broader range of interventions to address their needs. Endnotes
1
The incomes of the average homeowner (including those who hold mortgages) are nearly twice those of renters (Wachter & Smith 2010). Shaded area designates the official recession period. Using foreclosures per mortgage would have exaggerated the percentage of the community affected in counties that have low home ownership rates. Students with certain disabilities and new students with limited English proficiency are excluded from tests in North Carolina. All models are estimated using ordinary least squares (OLS) and robust standard errors that are clustered at the county level. I also calculated effects with different time intervals with respect to the EOG test date to determine whether the strength of the effect is related to the timing
2 3
18
of foreclosures (Appendix A,B) The coefficients are smaller in magnitude but remain statistically significant.
7 8
See appendix C for subgroup analysis on reading proficiency. Disaggregated reading scores are displayed in Appendix C. They follow a similar pattern with a smaller effect size. The Hispanic population of North Carolina more than doubled between 2000 and 2010 with considerable variation across counties (US Census). I also conducted the same regression analysis using the foreclosure rates from the same year the tests were given, and found significant though smaller associations (Appendix B).
10
References Cited Alexander, K.L., Entwisle, D.R., and Dauber, S. L. (1996). Children in motion: School transfers and elementary school performance. The Journal of Education Research, 90, 3-12. Ananat, E. O., A. Gassman-Pines, D. V. Francis, and C. Gibson-Davis (2011). Children Left Behind: The Effects of Statewide Job Loss on Student Achievement.NBER Working Paper No. 17104. Been, Vicki, Ingrid Gould Ellen, Amy Ellen Schwartz, Leanna Stiefel and Meryle Weinstein. (2010). Kids and Foreclosures: New York City. New York, NY: New York University. Bocian, D.G., Wei, L., & Ernst, K. (2010) Foreclosures by Race and Ethnicity: The Demographics of a Crisis. Center for Responsible Lending. 3-25. Retrieved from http://www.responsiblelending.org/ mortgage-lending/research-analysis/foreclosures-by-race-andethnicity.pdf. Calv-Armengol, A., Patacchini, E., & Zenou, Y. (2009). Peer Effects and Social Networks in Education. Review of Economics Studies, 76(4), 1239-1267. Dastrup, S. R., & Betts, J. R. (2012). Elementary Education Outcomes and Stress at Home: Evidence from Mortgage Default in San Diego. Retrieved from http://econ.ucsd.edu/~sdastrup/pdfs/ DastrupJobMarketPaper.pdf.
19
Ellinwood, M. (2011, January 4). Proposed Education Cuts would be a Huge Setback for North Carolina. NC Policy Watch. Retrieved from http://www.ncpolicywatch.com/2011/01/04/proposed-educationcuts-would-be-a-huge-setback-for-north-carolina/. Garrett, N. (2010, August 24). North Carolina Wins Race to the Top Money. WRAL. Retrieved from http://wraltechwire.com/business/ tech_wire/biotech/blogpost/8181783/. Goux, D., &Maurin, E. (2005).The effect of overcrowded housing on childrens performance at school. Journal of Public Economics, 89, 797-819. Gruman, D.H., Harachi, T.W., Abbott, R.D., Catalano, R.F., and Fleming, C.B. (2008).Longitudinal effects of student mobility on three dimensions of elementary school engagement. Child Development, 79 (6), 1833-1852. Retrieved from http://onlinelibrary.wiley.com/ doi/10.1111/j.1467-8624.2008.01229.x/full. Hanushek, E.A., Kain, J.F., and Rivkin, S.G. (2004).Disruption Versus Tiebout Improvement: The Costs and Benefits of Switching Schools. Journal of Public Economics, 88, 1721-1746. Harding, John P., Eric Rosenblatt, and Vincent Yao. (2009). The Contagion Effect of Foreclosed Properties. Journal of Urban Economics, 65(3): 237-251. Retreived from http://www.sciencedirect.com/science/ article/pii/S009411900900045X. Kingsley, G.T., Smith, R., and Price, D. (2009).The Impacts of Foreclosures on Families and Communities: A Primer. Washington, D.C.: Urban Institute. Retrieved from http://www.urban.org/ UploadedPDF/411910_impact_of_foreclosures_primer.pdf. Lash, A.A. and Kirkpatrick, S.L. (1994). Interrupted lessons: Teacher views of transfer student education. American Educational Research Journal, 31 (4), 813-843. Lovell, P., Isaacs J. (2008). The impact of the mortgage crisis on children and their education. First Focus, Brookings Institution. Retrieved from http://www.brookings.edu/~/media/Files/rc/papers/2008/04_ mortgage_crisis_isaacs/04_mortgage_crisis_isaacs.pdf.
20
Maxwell, L.E. (2003). Home and school density effects on elementary school children: The role of spatial density. Environment and Behavior, 35, 566-578. Mehana, M. and Reynolds, A.J. (2004).School Mobility and Achievement: A Meta-Analysis. Children and Youth Services Review, 26, 93 -119. Retrieved from http://www.sciencedirect.com/science/ article/pii/S0190740903001191. Sanchez, C. (2011, July 31). Education Cuts Squeeze N.C. Teachers. National Public Radio. Retrieved from http://www.npr. org/2011/07/31/138862695/north-carolina-cuts-squeeze-educationprograms. Schwartz, A.E., Stiefel, L., and Chalico, L. (2007). The multiple dimensions of student mobility and implications for academic performance: Evidence from New York City elementary and middle school students. A Condition Report for the New York Education Finance Research Consortium. New York, NY: Institute for Education and Social Policy. Retrieved from http://www.albany.edu/ edfin/documents/schwartzsteifelmobilitypaper.pdf. U.S. Department of Education. (2010, August). Projected FY 2010 Education Job Fund Allocations. Retrieved from http://www.ed.gov/ sites/default/files/edjobsfun-allocations.pdf. U.S. House of Representatives. (2009, February 17). H.R.1: American Recovery and Reinvestment Act of 2009. Retrieved from http://thomas.loc.gov/cgi-bin/ bdquery/?d111:HR00001:@@@D&summ2=m&. Wachter, S. & Smith, M. (2010). The American Mortgage System: Crisis and Reform. 62-66. Weinberg, B. (2006). Social interactions and endogenous association (OSU Working Paper 13038). Columbus, OH: Ohio State University. Retrieved from http://economics.sbs.ohio-state.edu/weinberg/socint1. pdf. Winning the Future. (2011). Congressional Digest, 90(3), 66-69.
21
Appendix A: Math scores using foreclosures from the same year.

Regression model Subgroup (N) All Students 1036 Only FRL 1035 Only Non-FRL 1036 Only Black 988 Only White 1036 Only Hispanic 1012 Only Female 1036 Only Male 1036 *Significant at 5% significance level **Significant at 1% significance level Source: NCDPI, Office of the Courts, US Census Change in EOG Math Model I (Controlled for County Fixed) Coefficent (Standard Error) -.1484 (5.177508)** -.084 6.41005 .015 3.68715 -.383 (15.73362)** -.081 4.43803 .118 19.00615 -.180 (5.113026)** -.115 (5.734264)* Change in EOG Math Model II (County & Time Fixed) Coefficent (Standard Error) .102 6.66063 .100 8.95158 .120 (6.000408)* .012 19.58827 .104 5.88040 .052 19.23581 .105 6.59374 .103 7.29081
Appendix B: Comparison for Year of data for FRL regression

Effect of Foreclosures on Student Economic Status (% Free and Reduced Lunch)
Regression model Same Year All Counties Coefficent (Standard Error) Foreclosure Rate 3.44705 (1.022315)** 1,036 *Significant at 5% significance level **Significant at 1% significance level Source: NCDPI, Office of the Courts, US Census Previous Year All Counties Coefficent (Standard Error) 5.36199 (1.092267)** 1,036 Previous Year Without Vacation Counties Coefficent (Standard Error) 8.12165 (2.169216)**
22
Appendix C: Disaggregated Effects of Foreclosure Rates on Student Reading Achievement

Race & Ethnicity Subgroup Foreclosure Rate All Students -.146** (7.016) Percent Unemployed -.020*** (.003) Constant .254*** (.038) Observations R-squared 880 .874 Black -.244* (12.696) -.033*** (.005) .933*** (.050) 850 .601 White -.072 (6.541) -.008*** (.003) .882*** (.033) 880 .782 Hispanic .081 (20.986) -.015** (.007) 1.112*** (.077) 873 .624 Gender Female -.164* (8.284) -.033*** (.004) .603*** (.043) 880 .799 Male -.126* (6.803) -.008** (.003) -.084** (.036) 880 .904 FRL .036 (9.328) -.015*** (.004) -.837*** (.035) 879 .806
Notes. Robust standard errors (clustered by county) are in parentheses. Sample size for each model is based on 88 counties over ten years. Each value represents a coefficient from an ordinary least squares (OLS) regression model. The values in the row labeled foreclosure rate represent the estimated effect of an increase in foreclosures per resident on student achievement (normalized county average percent proficient). The regression model controls for county fixed effects. *** p<.01, ** p<.05, * p<.1. Source: NC Department of Public Instruction, NC Office of the Courts, Bureau of labor Statistics.
23
Incentivizing Responsible Small-Dollar Lending in Low-Income Communities

Megha Bansal
Originally from the suburbs of Washington DC, Megha Bansal completed a Master of Public Policy degree from Duke University in 2012, and received dual Bachelor of Arts degrees in 2009 from the University of Maryland in Economics and Psychology, as well as a minor in International Development and Conflict Management. Prior to graduate school, Megha worked as a Paralegal at the Federal Trade Commissions Bureau of Consumer Protection, helping to prosecute telemarketing, financial, and advertising fraud. Her work experience at the Federal Trade Commission exposed her to the many issues that low-income individuals face, particularly in terms of access to the financial services sector. In graduate school, Megha focused her studies on social policy as it affects low-income communities, and spent a summer with the Brookings Institution Metropolitan Policy Program exploring anti-poverty policies and low-income tax credits. Meghas policy interests intersect many social and economic issues, including poverty, health, justice, and disaster response. Currently, Megha is a Federal Strategy & Operations Consultant at Deloitte Consulting in Washington DC, working with federal government clients to address their most complex problems while improving performance, efficiency, and innovation in the public sector.
Abstract Recent economic turbulence has contributed to an increased need for short-term small-dollar loans, particularly for low-income individuals. In a world with no payday loan programs, low-income, high-risk consumers might be unable to obtain loans and would have little to no access to shortterm cash options. Thus, the need for a small-dollar lending program is apparent. However, the current structure of the payday lending market is rife with information asymmetries, leaving lenders uncertain about risk levels associated with different consumers, and borrowers unaware of the consequences associated with how short-term loans are designed. Furthermore, proper incentives do not currently exist in the system to correct these information asymmetries and market failures. Based on the
25
Bansal: Incentivizing Responsible Small-Dollar Lending in Low-Income Communities
problems identified with the small-dollar lending industry, the concept of a borrowers card would address problems in term limits, borrowing amounts, and repayment plans. It would also increase data collection on the borrowing patterns of consumers, and create the most incentive for lenders to extend loans to consumers seeking small-dollar loans, while providing incentive for consumers to improve their own repayment behavior. Background Recent economic turbulence has contributed to an increased need for short-term small-dollar loans, particularly for low-income individuals. In many states, the payday lending market operates to meet this demand (Plunkett and Hurtado 2011:31-32; Saunders et al. 2010:4). Typically, individuals write a post-dated personal check to a lender or authorize electronic access to their bank accounts for the amount of the desired loan, generally less than $500, plus a transaction fee, typically a $15-$20 fee for a two-week $100 loan (Bair 2005:6-7). The lender holds the check until an agreed-upon date, usually the next payday, at which point the borrowers have the option to repay the loan and fee or allow the lender to deposit the check (Bair 2005:4). If borrowers are unable to repay the loan in full, they either face potential insufficient funds or overdraft fees, roll over the loan for an additional fee, or take out additional loans to pay back the first loan (Logan and Weller 2009:2). Lenders charge a higher interest rate to compensate for the risk of non-repayment (Caskey 2002:15) and the postdated check serves as a low-cost form of collateral for lenders (Elliehausen 2009:2). Consumers Needing Payday Loans Display Varying Levels of Credit Risk Payday loans are typically taken out for convenience, to cover an emergency, or to pay for consumption needs. Consumers perceive payday lenders as an optimal way to meet immediate cash flow needs (Logan and Weller 2009: 2; Caskey 2002: 2). Generally, low-income individuals who use these alternative financial products may lack financial savings, have bad credit and/or high debt-to-income ratios, cite privacy concerns, have a lack of comfort with formal financial services, [and] lack a basic consumer finance education (Birkenmaier and Tyuse 2005: 72). While borrowers do
26
have checking accounts, some do not have credit cards, have poor credit scores, or have exceeded their credit limits (Barr 2004:2). Payday lenders generally do not stringently underwrite their loans to assess credit risk or worthiness even though payday loan borrowers typically have a higher risk profile than prime borrowers (Center for Financial Services Innovation 2005:1; Lawrence and Elliehausen 2008:300). The underwriting for a loan consists simply of assessing whether the individual has a source of income to repay the loan (Lawrence and Elliehausen 2008:300). The focus is thus on an ability to collect, versus a consumers past payment behavior or capability to pay (Herrmann and Tescher 2008:15; Lawrence and Elliehausen 2008: 300). Payday Lending Market Contains Information Asymmetries Minimal barriers to entry make the payday loan industry competitive in nature (Schneider and Koide 2010:19). However, scholars suggest that the payday loan sector is a textbook case of market failure, because information asymmetries between borrowers and lenders lead to problems of moral hazard and adverse selection (Campbell et al. 2011:3). Moral hazard is perpetuated by models of payday lending in which lenders fail to properly verify repayment capability prior to issuing a loan, and instead, permit consumers to take out loans that they may not be able to afford. Adverse selection exists because there is no clear way for lenders to discern high-risk consumers from moderate- or low-risk borrowers. Representative data about varying consumers use of payday loans is limited (Lawrence and Elliehausen 2008: 304). Instead, it appears that all consumers are pooled and offered similar terms without consideration of variation in borrowing behavior and ability to repay. Furthermore, borrowers are often unaware of the terms of their loans. While lenders are required to disclose necessary information pertaining to loans under the Truth in Lending Act (TILA), consumers may not easily understand the financial language (Mann and Hawkins 2007:882). For example, some consumers do not have an accurate understanding of just how expensive payday loans are and either report unrealistically low annual percentage rates (APR) or are unaware of their APR (Caskey 2002:3; Lawrence and Elliehausen 2008:308). This lack of understanding may lead consumers to inaccurately compare their financial options (Mann and Hawkins 2007: 882). The calculation stipulated by TILA for examining
27
annual percentage rates also excludes loan fees and charges, which could lead some borrowers to underestimate the true cost of payday loans (Plunkett and Hurtado 2011:50). Additionally, consumers may lose trust or confidence in the markets ability to function, which could lead them to abstain from the traditional financial market altogether (Campbell et al. 2011:3). Payday Loans Are Poorly Designed Many borrowers who take out payday loans are unable to repay within a single pay period and thus must accrue additional fees to roll over their loans (Herrmann and Tescher 2008:6). The Center for Responsible Lending estimated that in 2008, the typical consumer who took out a payday loan ultimately paid $800 for a $300 loan, due to the accrual of additional fees and interest (Logan and Weller 2009:3).The organization also estimated that, of approximately 19 million payday loan borrowers, 12 million were trapped in a cycle of at least five payday loan transactions per year (Logan and Weller 2009:3). Lenders claim that when the interest rates and fees are stated in terms of an annual percentage rate, it distorts the purpose of the loan, which is intended to be short-term in nature (Caskey 2002:3). However, what starts off as short-term can turn into long-term because of rollover and chronic borrowing patterns (Caskey 2002:32). These rollover options may be helpful in the short-run; however, they are particularly risky over time, because they allow these high interest rates to compound and drive already financially vulnerable individuals further into debt (Birkenmaier and Tyuse 2005:73; Saunders et al. 2010:14). Additionally, consumers often borrow from multiple lenders and thus accrue varying levels of debt based on prior loan obligations (Stegman 2007:176-177). Some estimates suggest that borrowers take out eight to nine payday loans annually (Saunders et al. 2010:4). According to the Center for Responsible Lending, 76 percent of payday loans are made to consumers who need to pay off previous payday loans (Plunkett and Hurtado 2011:34). Data on Borrowing and Repayment History Is Not Readily Available Only a few states have standardized databases with information on the borrowing and repayment history of payday loan consumers,which could help enforce restrictions on rollovers and aggregate borrowing (Elliehausen2009:7). The lack of a centralized system makes it difficult to keep track of where consumers are originating their loans, how often
28
they are borrowing, and their true ability to repay (Stegman 2007:177). In the mainstream financial sector, consumers are linked to purchases and spending patterns through their credit score. However, data on short-term loans is generally entered on a 30-day cycle into credit bureau systems, despite the typical two-week repayment period (Herrmann and Tescher 2008:15). This lag time may affect the accuracy of data being used to underwrite loans. Furthermore, while some payday lenders collect and share data among themselves, the data is not widely available, particularly because the lenders often operate as private entities (Flannery and Samolyk 2005:4). Since payday lenders do not typically report borrower behavior to credit bureaus in a timely manner, if at all, those consumers that are able to repay cannot experience the credit rating benefits associated with timely repayment (Fair Community Credit 2011: 5; Herrmann and Tescher 2008:15). Without a proper link to their borrowing history, consumers that are unable to reap the benefit of good credit ratings may have less incentive to repay their loans on time, in full, or at all. In the long term, these strategies are detrimental to establishing financial security. Ultimately, this situation creates a fringe financial class without the capacity to build credit (Fair Community Credit 2011:5). A More Responsible Small-Dollar Lending Program Is Needed In a world with no payday loan programs, low-income, high-risk consumers might be unable to obtain loans and would have little to no access to short-term cash options. Thus, the need for a small-dollar lending program is apparent. However, the current structure of the payday lending market is rife with information asymmetries, leaving lenders uncertain about risk levels associated with different consumers, and borrowers unaware of the consequences associated with how short-term loans are designed. Furthermore, proper incentives do not currently exist in the system to correct these information asymmetries and market failures. Scholars agree that if equivalent programs are designed more responsibly, they can benefit the population of consumers that needs the loans and has the capacity to repay (Campbell 2011:101). More responsible loan programs have a variety of characteristics that ensure access to credit without trapping borrowers in additional debt. Ideal loans would comply with state and federal consumer protection and fair
29
lending laws, be underwritten based upon a borrowers ability to repay, be long enough to allow for affordable repayment, and come with additional assistance (e.g., financial counseling or mandatory savings components) to assist consumers (New Yorkers for Responsible Lending [NYRL] n.d.:1). Consumers should have options in pricing, borrowing, and repayment plans, dependent on their specific needs, socioeconomic status, and discipline (Herrmann and Tescher 2008:13-14). Any program developed as an alternative to a payday lender must consider borrower need and behavior, as well as what would happen if access to loans was eliminated (Campbell 2011: 0). Criteria A responsible payday loan alternative should seek to: Minimize risk associated with consumers ability to repay loans: This criterion aims to reduce risk associated with consumer behavior and/or ability to repay by either better assessing consumer riskiness or ensuring that consumer risk does not prohibit or limit the potential for loan repayment. Any alternative must minimize this risk in order to ensure that lenders continue to provide access to short-term smalldollar loans. Provide incentive for lenders to extend loans to consumers seeking small-dollar loans: In order to operate, lenders must be able to accrue a positive expected return. Since the demographic that tends to need small-dollar lending may be high-risk, incentives must be offered for lenders to extend a loan without using discriminatory criteria. A viable alternative will provide incentives by reducing the level of consumer risk, either by allowing lenders to better examine consumer risk, or by guaranteeing compensation in spite of the risk levels. Provide incentive for consumers to improve repayment behavior: Consumers are not necessarily linked to their borrowing and repayment histories and thus are provided with little incentive to improve their borrowing behavior. A viable alternative should provide a mechanism by which consumers choose to improve their own repayment behavior, in order to mitigate their levels of risk in the eyes of a lender.
30
Maximize sustainability of program implementation: Program design should consider factors of sustainability, such as operational cost, simplicity of implementation, political pushback, or scalability. The tension that exists with designing an alternative program is in balancing consumer need, firm incentives, and market failures resulting from information asymmetries,in order to meet the demand for these loans without incentivizing unscrupulous or predatory behavior. Alternatives The following alternatives provide plausible program designs for a responsible small-dollar lending program. Each alternative will be discussed in detail and weighted against the specific criteria identified above. Each criterion is weighted equally in the analysis. 1. Use a referral process to provide loans to approved low-income and/or high-risk consumers. 2. Design a borrowers card system to collect information about consumers borrowing and repayment behaviors. 3. Fund a loan loss reserve pool to back loans made to low-income and/or high-risk consumers. This analysis is based on the distinct design components, not the specific city or state programs discussed, and seeks to analyze the design components independent of one another in implementation. Alternative 1: Referral Process to Provide Loans This alternative is modeled after the Fair Community Credit program in Kansas City, which utilizes trained program referral partners to make loans to low-income individuals or subprime borrowers. The program also includes a loan loss reserve funded by foundations and individuals, a minimum income level, capped interest rates, limited borrowing amounts and term limits, and financial education. In its first year of operation, the program expected to make only 500 collateralized loans (Associated Press 2012).
31
The referral system incorporates and leverages borrowers social networks into the lending process (Schneider and Koide 2010:78). Community, peer-based international microfinance programs (Herrmann and Tescher 2008:12) have shown that loans made to individuals within an existing social network are more likely to be repaid, because individuals have been identified as responsible or trustworthy (Fair Community Credit 2011:11). The Kansas City group cites Grameen Banks 97 percent repayment rate, attributing this success to monitoring within social networks that exerts social pressure to repay and provides a safety net if borrowers experience financial distress (Fair Community Credit 2011:11). Fair Community Credit uses a referral system of trained community and neighborhood institutions. These institutions may be nonprofit organizations, congregations, or social service providers that are specifically targeted as referral partners. Training is provided by professional community organizers. According to Fair Community Credit, good borrowers tend to exhibit four characteristics: they have an acute and specific financial need, maintain a pre-existing relationship with a referral partner, demonstrate a willingness to improve their financial situation, and meet the minimum criteria for underwriting. After receiving a letter of endorsement, information about the clients loan is entered into an online tracking system to which only the referral partner, the partner bank representative, and Fair Community Credit have access. The online tracking system does not collect sensitive information like social security numbers or credit scores. Fair Community Credit uses the system to track referrals made by each partner to ensure that quality referrals are being made. After receiving a referral, all borrowers must also undergo an underwriting process with the partnering bank before qualifying to receive the loan. Analysis of Alternative 1 The risk associated with repayment stems from uncertainty about the individuals ability or propensity to repay the loan. A referral system provides some mitigation of risk compared to a walk-in customer for whom there is no basis for assessing ability to repay. The referral system uses trusted community partners that are able to speak to consumers character, morals, or other related characteristics beyond financial means that might affect their tendency to repay.
32
While a standardized referral procedure would minimize subjectivity concerns, it is uncertain whether referral systems provide enough incentive for lenders to extend loans to borrowers perceived as particularly risky. Referral and community partners may be biased in favor of borrowers drawn from their population of interest. Banks and lenders are more concerned with repayment, and without a guarantee of collateral or profitability, the referral may not be enough to incentivize them to extend loans. Furthermore, it is uncertain whether a referral system would provide a concrete incentive for consumers to improve repayment behavior, particularly if the referral is based on a pre-established relationship. Indirectly, borrowers may try to utilize their relationships with community partners more in order to gain favor prior to and during the referral process. Because the referral is more of an assessment of the consumers situation and need, it may not directly provide the incentive to improve behavior. A referral system keeps the pool of borrowers relatively small, since community partners would have to base referrals on established relationships, extensive counseling, and referral meetings. Maintaining a smaller pool of approved borrowers helps with financial sustainability. Moreover, community partners could always be added to the program if the program expanded or if partners withdrew. Additionally, low default rates of repayment would help with sustainability, since over time, lenders might observe that the referred consumers were reliable and be more willing to develop a relationship with the referral program. However, more evaluation is needed to determine default history and repayment with the referral process. Alternative 2: Borrowers Card to Collect Information Alternative 2 proposes a program similar to grocery store discounts. Cards would be used to track consumer purchases and spending histories. In the case of grocery store discount cards, the information a store gains about consumers is likely more valuable than the discounts enabled by the card. Individuals who desire a small-dollar loan could opt into this voluntary card program through an existing credit union. Participants would receive
33
a personalized borrowers card and be required to submit a fingerprint scan, to minimize the chances of borrowers obtaining multiple cards. To ensure access for low-income consumers in need, such as undocumented immigrants or those with nontraditional jobs, the card would not be linked to an individuals social security number. Each time a consumer receives a small-dollar loan, details about the loan amount, borrowing date, interest rates, and repayment schedule could be collected electronically. The loan would not necessarily have to accrue to the borrowers card, though there are program models (such as Progreso Financiero) that provide their loans through a pre-paid card (Baddour et al. 2011:48). An incentive system could be worked into the program design, in which consumers would receive a rating similar to a credit score. This shadow credit rating would be calculated based on information from borrowers cards, including how quickly consumers repaid their loans, whether consumers received financial counseling, whether consumers participated in financial education or homeownership classes, and whether consumers opened savings accounts or maintained a minimum balance, among other factors. These behaviors would provide bonus points for the shadow credit score. Progreso Financiero developed its own credit assessment underwriting model to assess repayment ability that uses an indepth online questionnaire and more than 1,300 attributes of nontraditional data, such as rental history, utility payments, and home or car ownership (Del Bosque 2011). Progreso Financiero developed this model because only three percent of its loan applicants have a sufficient credit history or high enough credit score (Baddour et al. 2011:47). Lenders could offer incentives such as lower interest rates, longer repayment plans for those with positive payment histories, or higher loan amounts to encourage consumers to build shadow credit scores. Under the Progreso Financiero model, individuals who repay their loans on time are able to receive better terms on their next loan (Moran 2011). Progreso Financiero also reports borrowers payments to credit bureaus and found that almost 90 percent of their borrowers increased their credit score from their first loan (Saunders et al. 2010:23). Some discussion of innovative small-dollar lending programs focuses on whether consumers can transition to the mainstream financial system after a certain amount of time (McKernan and Compton 2010:6). With the borrowers card, if consumers reach a shadow credit score of a certain
34
threshold amount based on their behavior and loan repayment, they may reach interest rate levels with the small-dollar lending institution comparable to the prime interest rates obtainable at a bank. Slowly and with counseling, consumers may then be able to transition over to the traditional banking sector. Piloting this program at an existing community financial institution or credit union may help to link shadow credit ratings to real credit scores. Furthermore, an existing credit union may experience lower operating costs due to preexisting infrastructure, staff, physical space, and processes (Bair 2005:28). Analysis of Alternative 2 The borrowers card increases data collection on a consumers loan history, borrowing behavior, and repayment ability, and makes it available to participating lenders (e.g., credit unions). This system fully meets the criterion of minimizing risk associated with a consumers ability to repay loans. First, the card would hold consumers responsible for their lending and borrowing patterns by linking their behavior to their potential to obtain loans. The card provides an incentive for consumers who frequently borrow and repay their loans to engage in the program because of the potential to receive reduced interest rates, larger loan amounts, and varied payment plans. The data collection thus helps to separate very risky individuals, who likely would not qualify for a loan at all, from those that are less risky. This provides an incentive for high-risk consumers to improve their behavior. However, to the extent that these consumers do not improve behavior, it is possible that their financial outcomes will worsen because moderate- or lowrisk individuals formerly pooled with them will no longer be subsidizing their loans. Amassing any large database of information may pose privacy concerns, especially if a biometric was used to ensure that borrowers could only receive one card. However, there are ways to minimize privacy concerns. The data collected could exclude personally identifiable information, such as social security numbers and credit scores. The database should be nonpublic and only shared with participating lending institutions and credit unions that agree to the terms of the borrowers card loan program. This provision would exclude any payday lending operation from access to the
35
database, since their loan terms and interest rates would not comply with program design. Furthermore, limiting access may actually encourage more credit unions to participate,in order to increase the amount of information they receive about potential borrowers. In order for lenders to extend loans to an individual, there must be some assurance of profitability. When repayment is uncertain, lenders are less likely to extend a loan to an individual. The borrowers card system would create a viewable database of information on a consumers borrowing and repayment history, providing lenders with incentive to extend loans for two reasons. First, lenders are able to use the consumers financial behavior to assess their ability and history of repayment. Second, consumers who borrow frequently and consistently repay are more likely to use the card because of the potential for a reduced interest rate. Thus, lenders have some indication that the borrower is able and willing to repay. This mechanism is somewhat comparable to the way in which the credit card industry has used market segmentation to offer varying terms and interest rates to consumers based on their respective financial characteristics. The information technology advances in the credit card industry may have some applicability to the small-dollar loan or payday lending industry as well (Mann and Hawkins 207:912). Very high-risk consumers who frequently take out loans they are unable to repay may be shut out of the borrowing market, because the card system exposes their behavior and inability to repay. However, a responsible lending program need not ensure that every individual pursuing a loan receives one. The National Consumer Law Center states that the criteria for responsible small loans must be considered on their own merits, without excessive concern for whether or not they will permit the wide availability of easy credit that gets payday borrowers in trouble today (Saunders et al. 2010:8). Thus, it is critical that those who need loans and can repay them are not denied access because they are clustered with high-risk borrowers. Alternative 2 creates an incentive for consumers to participate in this program, particularly those that frequently take out these loans or genuinely need these loans to supplement their income. By taking ownership of their borrowing behavior and building a history of good repayment, consumers are able to receive needed financial assistance at a potentially lower interest rate. Additionally, through their behavior, individuals can demonstrate their
36
ability to repay and subsequently, lower their perceived risk as a borrower. This feature helps address issues of adverse selection in the population. A few states have created centralized databases for information regarding consumers payday loans, in order to prevent rollover or chronic borrowing (Elliehausen 2009:7). However, these databases simply collect information for lenders to use and do not seem to provide consumers any incentive to improve their own financial behavior. This is in contrast to the borrowers card design, which would use the possibility of lower interest rates to incentivize higher repayment rates and better loan terms. However, this assumes that interest rates are a strong enough incentive mechanism. Other incentives might include longer repayment plans or higher loan limits. Furthermore, the potential for using good repayment behavior in the construction of a shadow credit score as the basis for transitioning to the traditional banking sector would provide individuals who are typically unable to access prime interest rates incentive to work toward that goal. This alternative helps create a long-term relationship between consumers and their borrowing behavior, by pegging an individuals borrowing history to future interest rates and loan terms. Individuals are linked to their borrowing behavior over time, creating a more sustainable lending model than systems that allow individuals to take out many loans from a variety of lenders. In the case of a state where it might be difficult to block political legislation that allows payday lending to take shape, this alternative might provide a way of tackling the issue from the other end, by reducing the need and demand for payday loan shops and encouraging more reliance on credit unions or other community lenders. Alternative 3: Loan Loss Reserve Pool to Back Loans In the small-dollar lending industry, borrowers tend to be risky and lenders are often hesitant to provide loans without some sort of collateral or guarantee of repayment. This alternative would create a loan loss reserve pool of funds in order to back the loans made to borrowers. Lenders could draw from the fund when borrowers are unable to repay their loans, in order to ensure that the lender is able to break even. The loan loss reserve fund would help lenders obtain more financial capital to help expand their business and increase the number of loans that could be made affordable (Schneider and Koide 2010:19).
37
Depending on the design, loan loss reserve pools could back the loans in entirety, or could partially back the loans up to a certain percentage, leaving the lender to potentially face some risk of loss. Loan loss reserve pools could also be funded in different ways. The Fair Community Credit programs $200,000 loan loss reserve pool is funded by individual and community donations, and is held in a Certificate of Deposit at the partner bank. The fund earns interest and is used to back 100 percent of loans made to individuals who are referred to the partner bank. In contrast, the Pennsylvania Credit Union Better Choice program funded their loan loss reserve pool from monies earned above the market rate-of-return on a $20 million dollar investment from the Pennsylvania Department of Treasury into the credit union system (FDIC 2010:75). The loan loss reserve pool was then used to reimburse up to 50 percent of any lender losses (Schneider and Koide 2010:11). Analysis of Alternative 3 Lenders want to ensure that whatever loans they extend will be repaid, typically so their businesses can break even or make a profit. In general, lenders are most concerned about protecting their businesses from loss, and the loan loss reserve pool provides that protection. Depending on the design of the loan loss reserve and the share of a loan a fund can guarantee, it either partially or fully meets the criterion of minimizing the risk of not being repaid. However, it is important to note that the loan loss reserve pool is specifically set up with the lenders interests and profitability concerns in mind. The loan loss reserve alternative partially meets the criterion of providing lenders incentives to extend small-dollar loans since they have a positive expected return. However, it is possible that lenders are provided with too much incentive to make loans to any individual, regardless of risk, simply to ensure a profit. With the assumption that all loans are backed, potentially risky consumers may be offered loans they cannot afford. A blanket guarantee against any loan losses makes riskier loans seem more attractive, contributing to the moral hazard consumers and lenders both experience. A partial reimbursement may provide the correct amount of incentive, while not overcompensating the risk pool. This alternative does not meet the criterion of providing consumers with incentive to improve their own repayment behavior. The loan loss
38
reserve pool is used to back all loans made to consumers in the hope that lenders will find more incentive to extend loans. Consumers are aware that even if they are unable to repay, the lender will still receive a return on the loan. With a loan loss reserve pool, consumers are not necessarily held accountable for the loans they receive. Furthermore, the existence of the loan loss reserve suggests that as long as a loan is repaid, no matter by whom, the loan is appropriate to make. This situation reduces the obligation borrowers feel to repay. Even if penalties are imposed on consumers in order to induce them to repay, the loan loss pool itself is not incentivizing the improvement in their behavior. This could be a particular problem for more high-risk consumers, who may seek out these small-dollar loans more frequently since they know that lenders may be more willing to extend a loan to them. A critical concern with sustaining a loan loss reserve pool is where funds will come from and how the pool can be replenished over time. Within the program design, there may be ways to include additional fees to help sustain the loan loss reserve. For example, every consumer that takes out a loan could pay a per-loan fee, a one-time fee, or an annual fee toward the pool. However, this tactic poses additional trade-offs with both access and overuse. If fees were assessed with each loan, high-frequency consumers would be required to pay more. But annual or one-time fees may actually encourage consumers to take out more loans, since they would incur no further cost. Finally, the concept of charging additional fees may be inconsistent with the idea of offering an alternative to payday lending, which hinges its own profitability model on additional fees to consumers. Thus, while there are ways to incorporate additional mechanisms to fund a loan loss reserve, they come with significant trade-offs. Recommendation Based on the problems identified with the small-dollar lending industry, Alternative 2 provides the most flexibility to create a responsible smalldollar lending program. The concept of a borrowers card would address problems interm limits, borrowing amounts, and repayment plans. It would also increase data collection on the borrowing patterns of consumers and could provide further insight into creating other financial programs that could more appropriately target individuals based on their level of risk.
39
While Alternative 1 and Alternative 3 offer benefits that the borrowers card perhaps does not, Alternative 2 best incentivizes responsible smalldollar lending by both encouraging lenders to extend loans to low-income individuals and encouraging consumers to take ownership of their own financial behavior. References Cited Associated Press. (2012). Kansas City Group Creates Alternative to Payday Loans. Kansas City, MO: Columbia Missourian. Accessed on April 18, 2012 at http://www.columbiamissourian.com/ stories/2012/01/23/kansas-city-group-creates-alternative-paydayloans/ Baddour, A., Lightsey, R, Monger, J., & Coleman, M. (2011). Reshaping the Future of Small-Dollar Lending in Texas: Alternatives to HighCost Payday and Auto Title Loans. Austin, TX: Texas Appleseed. Accessed on April 18, 2012 at http://www.texasappleseed.net/index. php?option=com_docman&task=doc_download&gid=698&Itemid= Bair, S. (2005). Low-Cost Payday Loan: Opportunities and Obstacles. Amherst, MA: A Report by the Isenberg School of Management University of Massachusetts at Amherst, Prepared for the Annie E. Casey Foundation. Accessed on April 18, 2012 at https://folio.iupui. edu/bitstream/handle/10244/101/fes3622h334.pdf?sequence=1 Barr, M.S. (2004). Banking the Poor: Policies to Bring Low-Income Americans Into the Financial Mainstream. Washington, DC: The Brookings Institution. Accessed on April 18, 2012 at http://www. frbsf.org/community/issues/assets/bankonsf/resources/banking_the_ poor.pdf Bianchi, N. (2012). Profiting from Poverty: How Payday Lenders Strip Wealth from the Working-Poor for Record Profits. Chicago, IL: National Peoples Action. Accessed on April 18, 2012 at http://www. npa-us.org/files/images/profiting_from_poverty_npa_payday_loan_ report_jan_2012_0.pdf Birkenmaier, J., &Tyuse, S.W. (2005). Affordable Financial Services and Credit for the Poor: The Foundation of Asset Building. Journal of Community Practice 13(1): 69-85.
40
Blank, R.M. (2008). Public Policies to Alter the Use of Alternative Financial Services Among Low-Income Households. Washington, DC: The Brookings Institution. Accessed on April 18, 2012 at http://www.brookings.edu/~/media/Files/rc/papers/2008/0416_low_ income_blank/0416_low_income_blank.pdf Campbell, J.Y., Jackson, H.E., Madrian, B.C., and Tufano, P. (2011). Consumer Financial Protection. Journal of Economic Perspectives, 25(1): 91114. Caskey, J.P. (2002). The Economics of Payday Lending. Prepared for the Center for Credit Union Research. Madison WI: Filene Research Institute. Accessed on April 18, 2012 at https://www.cuany.org/ access_files/outreach/Filene_-_The_Economics_of_Pay_Day_ Lending.pdf Center for Financial Services Innovation. (2005). Balancing Convenience and Risk: Developing Alternative Payday Loans. Chicago, IL: Center for Financial Services Innovation. Accessed on April 18, 2012 at http://cfsinnovation.com/?q=node/440682/ lightbox2&url=system/files/imported/managed_documents/ alternative_payday_loans.pdf Del Bosque, M. (2011). Taking an Interest: A New Business Model Challenges Predatory Payday Lenders. Austin, TX: Texas Observer. Accessed on April 18, 2012 at http://www.texasobserver. org/cover-story/taking-an-interest Elliehausen, G. (2009). An Analysis of Consumers Use of Payday Loans. Washington, DC: George Washington University School of Business Financial Services Research Program and Board of Governors of the Federal Reserve System Division of Research and Statistics. Accessed on April 18, 2012 at http://www.cfsaa.com/portals/0/ RelatedContent/Attachments/GWUAnalysis_01-2009.pdf Ernst, K., Farris, J., & King, U. (2004). Quantifying the Economic Cost of Predatory Payday Lending. Durham, NC: Center for Responsible Lending. Accessed on April 18, 2012 at http://cfsinnovation. com/?q=node/440682/lightbox2&url=system/files/imported/ managed_documents/crlpaydaylendingstudy.pdf
41
Fair Community Credit. (2011). Business Plan: Product Design, Structure, and Governance. Kansas City, MO: Communities Creating Opportunity. Federal Deposit Insurance Commission (FDIC). (2010). A Template for Success: The FDICs Small-Dollar Loan Program. The FDIC Quarterly 4(2): 28-41. Flannery, M., and Samolyk, K. (2005). Payday Lending: Do The Costs Justify The Price? Working Paper No. 2005/09. Washington DC: Federal Deposit Insurance Corporation Center for Financial Research. Accessed on April 18, 2012 at SSRN: http://ssrn.com/ abstract=771624 or http://dx.doi.org/10.2139/ssrn.771624 Hayes, P.L. (2009). A Noose Around the Neck: preventing Abusive Payday Lending Practices and Promoting Lower Cost Alternatives. St. Paul, MN: William Mitchell Law Review. Accessed on April 18, 2012 at www.wmitchell.edu/lawreview/documents/10.Hayes.pdf Herrmann, M.J., &Tescher, J 2008). A Fundamental Need: SmallDollar, Short-Term Credit. Chicago: Center for Financial Services Innovation, ShoreBank Corporation. http://cfsinnovation.com/ node/330391 H.R. 4173 (111th Congress). (2009). Dodd-Frank Wall Street Reform and Consumer Protection Act. Accessed on April 18, 2012 at GovTrack. us (database of federal legislation) at http://www.govtrack.us/ congress/bills/111/hr4173 Illinois Asset Building Group. (2009). Alternative Small-Dollar Loans in Illinois: Creating Sound Financial Products through Regulation and Innovation. Chicago: Illinois Asset Building Group. Accessed on April 18, 2012 at www.illinoisassetbuilding.org/sites/default/files/ asdlbrieffinal.pdf Kent, M. (2011). Re: Follow-Up Regarding Short-Term Financial Services Loan Act, S3841/A7047.New York: Department of Consumer Affairs. Accessed cached version on April 18, 2012 at http://www. nedap.org/programs/campaigns/documents/Memo-DCAToInterested Parties-201105-1_000.pdf.
42
King, U. and Parrish, L. (2011). Payday Loans, Inc.: Short on Credit, Long on Debt. Durham, NC: Center for Responsible Lending. Accessed on April 18, 2012 at http://www.responsiblelending.org/ payday-lending/research-analysis/payday-loan-inc.pdf Kovach, A. (2010). The Changing Landscape for Alternative Loans. Chicago, IL: Sargent Shriver National Center on Poverty Law. Accessed on April 18, 2012 at http://www.theshriverbrief. org/2010/07/articles/asset-opportunity/asset-building/the-changinglandscape-for-alternative-smalldollar-loans/ Lasher, M.C. (2011). Memorandum in Opposition (Draft). New York: City of New York Office of the Mayor. Accessed cached version on April 18, 2012 at http://www.nedap.org/programs/campaigns/documents/ NYCMemo_S3841-A7047-11smalldollarlending_MCLApproved. pdf. Lawrence, E.C., & Elliehausen, G. (2008). A Comparative Analysis of Payday Loan Customers. Contemporary Economic Policy 26(2): 299-316. Logan, A., and Weller, C.E. (2009). Who Borrows From Payday Lenders? An Analysis of Newly Available Data. Washington, DC: Center for American Progress. Accessed on April 18, 2012 at http://www. americanprogress.org/issues/2009/03/pdf/payday_lending.pdf Mann, R.J., and Hawkins. J. (2007). Just Until Payday. Los Angeles, CA: University of California Los Angeles Law Review (Vol. 54). Accessed on April 18, 2012 at SSRN: http://ssrn.com/ abstract=921844 Martinez, M. (2012). Kansas City: Fair Community Credit (Program Coordinator, fccfaircommunitycredit@gmail.com) McKernan, S.M., and Compton, J.F. (2010). Developing A Research Agenda on Small-Dollar Credit and Financial Empowerment. Washington, DC: Urban Institute. Accessed on April 18, 2012 at http://www.urban.org/UploadedPDF/412307-Developing-aResearchAgenda.pdf
43
Moran, S. (2011).Some Ways to Get Started as a Social Entrepreneur. New York: New York Times. Accessed on April 18, 2012 at http:// www.nytimes.com/2011/06/23/business/smallbusiness/23sbiz. html?pagewanted=all Mountain Association for Community Economic Development. Designing an Effective and Viable Alternative to Payday Lending: Lessons from the Save It! Loan. Berea, KY. Accessed on April 18, 2012 at http://cfsinnovation.com/node/440548 New Yorkers for Responsible Lending.Principles of Sound SmallDollar Lending. New York. Accessed on April 18, 2012 at http://www.nedap.org/programs/campaigns/documents/ FINALPrinciplesofSoundSmallDollarLending.pdf The NYC Legislature Should Reject the Short-Term Financial Services Loan Act S.3841-A/A.7047-A. New York. Accessed on April 18, 2012 at http://www.nedap.org/programs/campaigns/documents/ NYRLFactSheet-6Reasons6-13-11.pdf New York Times. (2012). Hurting Poor Borrowers. New York: New York Times. Accessed on April 18, 2012 at http://www.nytimes. com/2012/02/01/opinion/hurting-poor-borrowers.html?_r=1 Pennsylvania Treasury. (2011). PA Treasurer McCord: Innovative Payday Lending Alternative Saves PA Consumers $15 Million. Harrisburg, PA. Accessed on April 18, 2012 at http://www.patreasury.gov/ PressReleases-2011-02-24%20Better%20Choice%20Year%20 end%202010.html Plunkett, L.A., & Saunders L.K. (2010). Put Your Money Where the Sharks Mouth Isnt: Loan Options for Mississippi Borrowers That Are Better Than Payday Loans. Boston: National Consumer Law Center. Accessed on April 18, 2012 at http://www.nclc.org/images/ pdf/high_cost_small_loans/payday_loans/rpt-loan-options-ms.pdf Plunkett, L.A., &Hurtado A.L. (2011). Small-Dollar Loans, Big Problems: How States Protect Consumers from Abuses and How the Federal Government Can Help. Suffolk University Law Review, Vol 44. Rev. 31. Accessed cached version on April 18, 2012 at http://www. law.suffolk.edu/highlights/stuorgs/lawreview/Volume_44/upload/ Plunkett_Lead_Macro_Final.pdf
44
Reckard, E.S. (2010). Lender Embraces Latino Immigrants. Los Angeles: LA Times. Accessed on April 18, 2012 at www.progressfin.com/ Lender_Embraces_Latino_Immigrants.pdf Saunders, L.K., Plunkett, L.A., & Carter, C. (2010). Stopping the Payday Loan Trap: Alternatives That Work, Ones That Dont. Boston: National Consumer Law Center. Accessed on April 18, 2012 at http://www.nclc.org/images/pdf/high_cost_small_loans/payday_ loans/report-stopping-payday-trap.pdf Schneider, R. and Koide, M. (2010). How Should We Serve the ShortTerm Credit Needs of Low-Income Consumers? Chicago, IL: Center for Financial Services Innovation. Accessed on April 18, 2012 at http://cfsinnovation.com/node/440501 Smale, P. (2005). Payday Loans: Federal Regulatory Initiatives. Cornell University ILR School. Accessed on April 18, 2012 at www. policyarchive.org/handle/10207/bitstreams/3850.pdf Stegman, M.A. (2007). Payday Lending. Journal of Economic Perspectives 21(1): 169-90. Washington, E. (2006). The Impact of Banking and Fringe Banking Regulation on the Number of Unbanked Americans. Journal of Human Resources, Winter 2006, 41 (1), 106-137. Accessed on April 18, 2012 at www.econ.yale.edu/faculty1/washington/impactbank.pdf
45
Negotiating Online Privacy Boundaries: Self-Revelation in the Facebook Generation

Katherine Bies
Katherine Bies graduated from the University of Michigan (Ann Arbor) in 2012 with a Bachelor of Arts in Communication Studies. This policy research paper is an adaption from her senior honors thesis, which was awarded the Virginia L. Voss Memorial Award for excellence in writing. She is currently a Teach for America corps member at Brownsville Ascend Charter School in Brooklyn, New York.
Abstract Social media sites, while facilitating the collection and sharing of personal information online, have had negative consequences for personal privacy. Policy makers have yet to define which aspects of online communication on social media sites, if any, can be legally considered private, leaving many users with a false sense of security online. This study compares the offline behavior of 381 survey respondents to their behavior on social media sites in order to determine if offline privacy law can be applied to cyberspace. An analysis of these responses finds that: 1) social media users have negotiated a privacy boundary online; 2) online privacy protection is similar to privacy protection used offline; 3) the use of online privacy protection varies by generational cohort. These results suggest that action should be taken to legally afford social media users the same legal protections that are guaranteed offline. Introduction The development of interpersonal communication technology has bridged the physical distance between individuals in modern society by facilitating the ability to gather and share information. However, these innovative technologies such as the telephone, recording devices, and cellphones come at a potential cost to personal privacy (Agre and Rotenberg 2007; Moore 1984; Caloyannides 2003; Ware 1986; Wolinksy and Sylvester 1992). The Fourth Amendment,which guards against unreasonable searches and seizures, has traditionally served as the constitutional safeguard of personal privacy. As technological innovations have made it easier to
47
Bies: Negotiating Online Privacy Boundaries
obtain and share personal information, the Supreme Court has defined and redefined the boundary between public and private, reconciling the capabilities of new technologies with a constitutional right developed before these technologies were even imaginable. The Internet has facilitated the collection and sharing of personal information on an even greater scale (Caloyannides 2003). Social media sites, such as Facebook, LinkedIn, and Twitter, are communication platforms used by hundreds of millions of people to share personal information online. Many users limit access to personal information with privacy controls developed by these sites; however, the question remains: does this action guarantee an expectation of privacy on social media? This study explores this question by comparing privacy protection on social media sites with the offline expectation of privacy established by the Supreme Court. Background An Overview of Facebook Facebook, launched in 2004, allows a registered user to create an online profile and to add other users as friends. Facebook invites users to disclose private information by posting information on their profile, sharing photos, exchanging wall posts, creating status updates, and sending Facebook messages. Facebook was chosen as the focus of this study for a variety of reasons. First, Facebook is the worlds largest social media site with over 800 million users worldwide (Facebook Statistics 2011). While 48 percent of its users are 18 to 34 year olds, the 35 and older cohort, currently 30 percent of the user base, is growing steadily (Facebook Statistics 2011). Also, Facebook provides both opportunities for sharing private information and privacy controls that are similar to other social media sites like Instagram, Twitter, and LinkedIn. The average Facebook user spends more than 55 minutes on Facebook per day (Facebook Facts 2010). Facebooks ubiquity may have substantial implications for the current social privacy norms upon which developments in online privacy law will be based. Privacy Risks on Social Media: The Facebook Case Facebook has obvious social benefits, as well as risks, to personal privacy. The site has played an important role in developing and maintaining
48
relationships with friends by allowing users to interact with a large social network (Debatin et al. 2009; Ellison et al. 2007). The facilitation of friendship networks is positively correlated with the maintenance and creation of social capital (Ellison et al. 2007). However, Facebook users do not always have control over, or even knowledge of, the many individuals with access to their personal information. Privacy breaches may bring unintended negative consequences. Facebook users have faced scrutiny from potential employers and law enforcement based on the material they post on social media- even when protected by privacy settings (Smith and Kidder 2010; Stone 2006; Kornblum and Marklein 2006). Employers admit using Facebook for recruiting and assessing applicants (Smith and Kidder 2010; Zeidner 2007). Police officers receive training about Facebook as a way to investigate and reduce crime (The Fuzz 2006). The FBI recently released an advertisement looking for companies with the capability to build a data-mining application for social media sites, including Facebook (Giles 2012). However, Facebooks continued popularity proves that its benefits override the potential loss and even invasion of privacy (Debatin et al. 2009). The Development of Privacy Law In constitutional law, a right to privacy has been based on the Fourth Amendment: The right of the people to be secure in their persons, houses, papers, and effects, against unreasonable searches and seizures, shall not be violated, and no Warrants shall issue, but upon probable cause, supported by oath or affirmation, and particularly describing the place to be searched, and the persons or things to be seized (U.S. Const. amend. IV). The Supreme Courts interpretation of the Fourth Amendment has expanded from solely protecting a citizens right against government intrusion of physical property to encompass other aspects of personal privacy. The contemporary concept of personal privacy, defined as the right to control the access and use of personal information, was established in the late 19th century by future Supreme Court Justices Warren and Brandeis. Concerned that modern media technologies such as instant photographs and tabloid newspapers were intruding into the private lives of American citizens, Warren and Brandeis (1890) argued that individuals have a right to be let alone that extends beyond a persons right to physical property to include his thoughts, sentiments, and emotions.
49
For the first half of the 20th century, privacy was limited to the physical space of the home. The 1928 case Olmstead v. United States, 277 U.S. 438, allowed law enforcement to wire-tap phone conversations if they did not physically enter the suspects house. Justice Brandeis, dissenting in the Olmstead case, argued that the court must meet modern conditions by expanding the interpretation of the Fourth Amendment to include the capabilities of new technologies that the Founders could not have predicted when writing the constitution, 277 U. S. 465, (1928). The Supreme Court expanded its interpretation again with the development of the public telephone booth. In Katz v. United States, the court ruled that regardless of physical location, a person with a reasonable expectation of privacy is protected from unreasonable search and seizure under the Fourth Amendment, 389 U.S. 360 (1967). Justice Stewart, writing for the majority, argued that the Fourth Amendment protects against intrusion in a public setting if the individual takes steps to create a private space such as shut[ting] the door to a telephone booth, 389 U.S. 352, (1967). Justice Harlan, writing a concurring opinion, further refined the concept of a reasonable expectation of privacy in a two-part rule: first, an individual must exhibit an expectation of privacy and secondly, society must acknowledge this expectation of privacy as reasonable. Legal scholars have recommended that policy makers further expand the concept of personal privacy to safeguard against privacy intrusions on social media sites (Beckstrom 2008; Hodge 2006). Junichi Semitsu (2011) analogizes conversations on Facebook to the telephone booth conversation recognized by the Supreme Court in Katz to warrant an expectation of privacy. While he acknowledges that Facebook communications can be broadcast to a multitude of Facebook friends rather than one listener, Semitsu argues that the Supreme Court never suggested that additional message recipients necessarily defeats an expectation of privacy (Semitsu 2011). Applying the Katz decision to Facebook, Semitsu (2011) suggests that courts today should consider Facebook as the twenty-first century equivalent of a phone booth. An individual who uses Facebook privacy controls makes the equivalent action of shutting the telephone booth door and therefore has a reasonable expectation of privacy. These recommendations are based solely on normative argument, however. While important in developing the legal theory behind court
50
decisions, empirical claims measuring the current cultural conception of privacy would add weight to any decision regarding a concept traditionally rooted in social practices (Solove 2002). Empirical Research Regarding Facebook Privacy Benson (2009) conducted open interviews with college students in order to determine the extent to which they share personal information both online and offline. Expanding upon Bensons approach, I compare privacy practices on Facebook and offline. By comparing the offline and online behaviors of Facebook users, I hope to add empirical clarity to Semitsus application of Katz to Facebook, and most importantly, to the over-arching legal debate and question: do Facebook users have a reasonable expectation of privacy on Facebook? Hypotheses Self-Revelation In Bensons study, self-revelation is the act of disclosing private information. Benson identified several categories of Facebook content considered inappropriate to share in a public setting, including: excessive drug and alcohol use, name calling, and political comments (Benson 2009). Benson found that self-revelation did not vary significantly from online to offline for college students (Benson 2009). However, older users may behave differently than younger users. Tapscott (2009) predicts that generational cohort is an important determining factor in an individuals attitude towards and experience with online technologies. He defines the Net Generation as individuals who are currently aged 29 and younger and have grown up with the Internet (Tapscott 2009). Because of their online experiences, Palfry and Gasser (2008) suggest that these individuals relate information and experience friendship differently from their parents. To test these theories, I conduct a comparison of self-revelation across three generational cohorts, including the Net Generation, a middle cohort (26-35), and an older cohort (36 and older). Based on Bensons findings, I predict that: H1: The Net Generation will engage in more consistent selfrevelation on Facebook and offline than older generations.
51
Correlates of Privacy Protection Before I compare privacy practices on Facebook and offline, I first determine to what extent privacy protection is used on Facebook. Privacy protection refers to actions that limit access to private information. Facebook privacy controls allow a user to limit his or her content to specified groups of users. A variety of factors may influence the extent to which Facebook privacy controls are used. The Net Generation. While generational cohort has an important influence on online behavioral norms, discussion on this topic has mostly focused on self-revelation (Tapscott 2009; Palfry and Gasser 2008). Early studies suggest the Net Generation may differ from older generations in its online privacy protection. Palfry and Gasser (2008) argue that the Net Generation is more likely than older generations to feel that sharing information online is the norm. The Net Generation may also feel less compelled to protect information. Expanding on Palfry & Gassers theory, I predict that: H2: The Net Generation employs less stringent privacy controls on Facebook than older generations. Privacy Controls Knowledge. A Facebook users familiarity with privacy controls may also influence the stringency of his or her privacy protection. In the last six months, Facebook has updated its privacy controls three times. If users are not current with the most recent privacy controls, their content may not be protected as intended. Butler et al. (2011) compared Facebook users perceived privacy settings to what their actual settings were and found that only 23 percent correctly stated their settings. Based on Butler et al.s findings, I predict that: H3:Users with greater privacy controls knowledge will employ more stringent privacy protection on Facebook. Demographic Factors. Previous psychological research has found gender and personality attributes to influence online privacy concerns. Lewis et al.(2009) discovered that females employ more stringent privacy settings than males on Facebook. Junglas et al. (2008) examined the relationship between the Big Five personality variables and a concern for online privacy and found only three of the five attributes correlate significantly: agreeableness, conscientiousness,
52
and openness. Highly agreeable individuals had lower privacy concerns than non-agreeable individuals. Both conscientious individuals and individuals open to new experiences had a higher concern for privacy than non-conscientious or non-open people. A Comparison of Offline and Online Privacy Protection Legal scholars have argued that the use of privacy controls is the online equivalent to various forms of legally recognized offline privacy protection (Semitsu 2011; Beckstrom 2008; Hodge 2006). I test this argument by statistically comparing the offline and online privacy protection of Facebook users. Do Facebook users take similar steps to protect their Facebook content as they do to protect private information offline? If Facebook privacy protection and offline privacy protection are similar, legal scholars are more supported in their argument that Facebook users exhibit an expectation of privacy. Therefore, I predict: H4: Facebook users who engage in more stringent privacy protection offline will also engage in more stringent privacy protection on Facebook. The second part of the Katz test is whether or not society will recognize this expectation of privacy as reasonable. If only certain groups of Facebook users act in a way that affords an expectation of privacy online, then this expectation may not be reasonable by societys standards. Moderating Effect of Generational Cohort. Based on discussions above, I predict that generational cohort influences the relationship between online and offline privacy protection. Benson (2009) found that the Net Generation consistently reveals private information online and offline. Expanding on this finding, I predict that the Net Generation will also be consistent in its online and offline privacy protection. Older generations, growing up in a different technological environment, may be cautious to accept the norms that the Net Generation has already established on Facebook, especially if these behaviors are not entirely consistent with traditional privacy norms (Tapscott 2009; Palfry and Gasser 2008). Therefore, I predict: H5: The Net Generation will engage in more consistent privacy protection on Facebook and offline than older generations.
53
Data and Methods Sample I created and administered a survey to measure the five hypotheses. The non-representative sample was composed of 381 Facebook users recruited online using Amazon Mechanical Turk which allows registered workers to browse for and complete computer-based tasks for compensation. All respondents were required to be United States residents older than 17 years of age and to have a Facebook account in order to complete the survey. Of the Mechanical Turk respondents, 194 (51 percent) were male and 187 (49 percent) were female. Forty-nine percent of participants were nonHispanic Caucasian, 25 percent were Asian, and 16 percent were Hispanic. Fourteen percent of respondents had received a high school diploma as their highest level of education, 21 percent were currently enrolled in college, 38 percent had received a college degree and 26 percent were currently pursuing or had completed a higher degree. Sixty-four percent of respondents had a Facebook profile for two years or more. Half of the respondents check Facebook every day. This sample is composed of three pre-determined age groups (18-25, 26-35, and 36 and older) based on Tapscotts generational cohort definitions and Facebooks demographic make up (Tapscott 2009; Facebook Statistics 2011). In order to best compare the three generational cohorts, each one needed a relatively equal percentage of survey respondents: the Net Generation cohort had 155 respondents (41 percent), the middle (2635) cohort had 110 respondents (29 percent), and the oldest (36 and older) cohort had 116 respondents (30 percent). The mean age of the sample was 32 years (Standard deviation = 11.55; Range = 18-77). Variables The survey instrument consisted of 66 questions to measure the central constructs developed in the hypotheses. The research focuses on two comparisons: (1) online self-revelation to offline self-revelation and (2) online privacy protection to offline privacy protection. Variables shown in previous studies to be related to privacy concerns were also measured as controls: privacy controls knowledge, personality traits,and demographic variables such as gender and generational cohort.
54
Online and Offline Comparisons. Online self-revelation measures the extent to which the respondent reveals private information on Facebook. A thirteen-item index was created based on Bensons findings about information that Facebook users found socially inappropriate (Benson 2009). Respondents rated statements on a four-point scale with 1 as never, 2 as sometimes, 3 as occasionally, and 4 as often. A higher score signifies greater self-revelation (Cronbachs = .84). Offline self-revelation was measured by analogizing interactions with Facebook friends to offline interactions with friends. To keep this measure consistent with the online self-revelation index, selected online items were converted into a six-item offline self-revelation index. Raw scores were summed, with higher values indicating greater self-revelation (Cronbachs = .82). Online privacy protection measures the extent to which a Facebook user applies privacy settings to limit the audience who is able to see a users content online. This index is made up of three questions about the use of Facebooks privacy settings: public, friends only, and custom. The custom privacy control goes a step further in privatizing content because it not only limits content from the public view, but also from the view of specific Facebook friends. Responses are scored from 1 as public to 3 as custom with higher values representing more stringent privacy controls (Cronbachs =.87). Offline privacy protection measures behaviors that limit to whom a respondent reveals personal information. The items were built in two parts: (1) private information revealed, based on the self-revelation index, and (2) limiting who this information is being revealed to, based on the Katz analogy. Respondents rated statements on a four-point scale from 1 as never to 4 as often. The eight-item index was moderately reliable (Cronbachs =.71). The internal consistency of this measure is not as high as the previous measures, but it shows acceptable reliability. The reliability was not improved by removing any of the items. Control Variables. Privacy controls knowledge is a five-item index comprised of questions about respondents knowledge of and comfort with Facebooks privacy controls based on a similar study by Butler et al. (2011). Four of the five items were aggregated with higher scores demonstrating
55
greater knowledge of privacy controls (Cronbachs =.68). One item was removed in order to improve internal consistency. The Big Five personality traits were assessed using a ten-item scale developed by Gosling et al. (2003) and validated by Muck et al. (2007). Responses for each item were scored with a seven-point Likert scale, ranging from 1 as strongly disagree to 7 as strongly agree. Demographic variables of gender, age, race, income, and current education level were also measured. Results Self-Revelation H1 predicted that the Net Generation would have more comparable offline and online self-revelation than older generations. An interaction was calculated to determine the distinct relationship between online and offline self-revelation by generational cohort (See Figure 1). The Net Generation exhibited a strong, positive, and significant correlation between offline and online privacy protection. I fail to reject H1. Facebook users in the Net Generation who are more self-revelatory offline are also more selfrevelatory online. Older generations exhibited a much smaller, and even slightly negative, association between self-revelation offline and online, which means that older generations who share more information offline actually share less information online. Correlates of Privacy Protection The average Facebook user was found to use privacy settings to limit content to Facebook friends. To determine if this finding was consistent for the generational distribution as a whole, a regression was performed to control for demographic characteristics associated with predicting privacy protection (see Table 1). The Net Generation and high privacy controls knowledge were found to have a significant, positive association with privacy protection. Individuals with greater conscientiousness, a higher education level, and females were also found to use more stringent privacy settings. The other measured variables were not found to be significant predictors of privacy protection.
56
Figure 1:The Relationship of Offline and Online Self-Revelation by Generational Cohort1
Amount of Online Self Revelation
Amount of Offline Self-Revelation
Generational Cohort. H2 predicted the Net Generation to have less stringent privacy controls than older generations. However, the older cohorts both had small, negative associations with privacy protection (see Table 1; the oldest cohorts association is marginally significant at p < .10). This demonstrates that the Net Generation, the excluded group in this regression, exhibited the most stringent use of privacy controls of any generation as a whole. I reject H2. The Net Generation was more likely than older generations to use the more stringent custom Facebook privacy control. Possible explanations of this phenomenon will be explained in the Discussion. Privacy Controls Knowledge. Privacy controls knowledge was predicted to positively influence the use of Facebook privacy controls in H3. The Net Generation was found to not only use the most stringent privacy settings, but also to be most knowledgeable about privacy controls (see Table 1 and Figure 2). This difference between the privacy controls
1
Scale of 0-1, with 1 being highest amount of self-revelation. High self-revelation respondents fall in the top third and low self-revelation respondents fall in the bottom third.
57
Table 1: A Regression Analysis of the Correlates of Online Privacy Protection

Predicting Variables White Asian Black Female Income Education Agreeableness Conscientiousness Emotional Stability Extraversion Openness 26-35 cohort 36 and older cohort Privacy Controls Knowledge N R2 Online Privacy Protection (B) 0.04 0.01 -0.02 0.04^ 0.04 0.10^ 0.04 0.14* -0.11 -0.04 0.09 -0.02 -0.05^ 0.21** 348 0.109 Standard Error 0.04 0.04 0.07 0.02 0.04 0.06 0.07 0.07 0.06 0.05 0.07 0.03 0.03 0.08
*significant at p < .05, **significant at p< .01, ^significant at p<.10
Figure 2: A Comparison of Privacy Controls Knowledge Means by Generational Cohort
58
knowledge of the Net Generation and the oldest generation was found to be statistically significant at p<.05 after performing an Independent Samples T-Test. I fail to reject H3. A Comparison of Offline and Online Privacy Protection To gauge the similarity of offline and online behaviors, a regression was performed to determine if a relationship still exists after controlling for a large number of demographic predictors. Offline privacy protection had a strong, positive, and significant correlation (.23 at p<.01) with Facebook privacy protection (see Table 3). I fail to reject H4. These results suggest that the use of Facebook controls is similar to steps taken to protect privacy offline. Moderating Effect of Generational Cohort. H5 predicted the Net Generation to engage in more consistent privacy protection online and offline than older generations. Table 3 shows generational cohort is a moderating variable for the online and offline privacy protection comparisons even when controlled for alternative predicting variables. An interaction was calculated to determine the distinct relationship between privacy protection and generational cohort (see Figure 2). The oldest cohort had a large, significant, and positive correlation (at p <. 05) between offline and online privacy protection. The Net Generation did not have a significant correlation. I reject H5. The Net Generation engages in more stringent privacy protection online regardless of its offline privacy protection. Offline and online privacy protection is actually more consistent for older generations. This suggests that the Net Generation is more concerned about online privacy than older generations. Possible explanations will be discussed later. Discussion Do Facebook users exhibit an expectation to privacy? The overall pattern of results from my survey of 381 adults around the country suggests that the first Katz criterion has been met: Facebook users exhibit an expectation of privacy. First, the median Facebook user implements the Friend privacy control to restrict his or her content from public view. Secondly, Facebook users, regardless of generational cohort,
59
Table 2: A Regression Analysis of the Relationship between Online and Offline Privacy Protection
Moderating Variables Offline Privacy Protection White Asian Black Female Agreeableness Conscientiousness Emotional Stability Extraversion Openness Income Current Education 26-35 cohort 36+ cohort N R2 Privacy Controls (B) 0.23** 0.06 0.02 -0.02 0.04* -0.01 0.14* -0.1 -0.04 0.13^ 0.03 0.06 -0.02 -0.06* 339 0.118 Standard Error 0.08 0.04 0.04 0.07 0.02 0.07 0.07 0.06 0.05 0.07 0.04 0.06 0.02 0.03
*significant at p<.05, **significant at p<.01, ^significant at p<.10
Figure 3: The Relationship of Offline and Online Privacy Protection by Generational Cohort2
Stringency of Online Privacy Protection
Stringency of Offline Privacy Protection Scale of 0-1, with 1 being highest stringency of privacy protection. High stringency respondents are in the top third and low stringency respondents are in the bottom third.
60
were found to engage in online privacy protection that is similar to offline privacy protection. These results support Semitsus suggestion that hiding Facebook content behind a protective cyberwall is analogous to the Katz example of concealing a private telephone conversation behind a telephone booth wall (Semitsu 2011). In follow-up interviews with a sub-sample of survey respondents, Facebook users discussed their expectation that privacy controls on Facebook should be respected. My content is more private than a persons who has their settings set to public, said Lucy,age 19. When asked how he would react if someone were to find a way to get past his privacy settings without his knowledge or consent, George, age 22 said, I would feel violated. You are creating a barrier between the people you want to see your content and the people you dont. By using privacy controls, it appears Facebook users are defining a privacy boundary on Facebook. Is this expectation reasonable? Results suggest that generational cohort has an important influence on both the use of self-revelation and privacy protection on Facebook. The Net Generation is more likely than older generations to use the Custom setting. Survey results and follow-up interviews reveal why the Net Generation is more likely than older generations to take this extra step to protect their privacy. Since the Net Generation shares more private information on Facebook, these individuals may have a greater desire to protect their content. Survey data found that the Net Generation reveals more personal information and is more knowledgeable about Facebook privacy settings than older generations. The Net Generations greater concern with privacy may be the reason these individuals are savvier users of Custom controls. For example, Christina, age 20, explained, I use the custom privacy setting a lot because some of my Facebook friends are family members If Im using profanity or something I dont want my daddy to see, I make sure to customize him out. On the other hand, older generations may not take the time to use more stringent settings, because they do not need to. Joyce, age 42 explained, I dont take the time to learn about privacy controls because I dont share information that Im concerned about protecting. While this study limited its research to whether or not Facebook users implement Custom privacy settings, further research could focus on why Custom
61
settings are used and which Facebook friends are customized in or out by Facebook users. Another explanation is that the Net Generation, having grown up with the Internet and social media sites, has established online behavioral norms and an expectation of which behaviors on Facebook guarantee a sense of privacy (Tapscott 2009; Palfry and Gasser 2008). Older generations, growing up in a different technological environment, may be cautious to accept the norms that the Net Generation has already established on Facebook, especially if these behaviors are not entirely consistent with traditional privacy norms (Tapscott 2009; Palfry and Gasser 2008). Since results show the online and offline privacy protection of older generations to be similar, older generations may be applying offline privacy norms as they navigate privacy boundaries in cyberspace rather than taking cues from the Net Generation. Because the Net Generation varies significantly from older generations in its online behavior, it is difficult to say that all generations will currently agree on a reasonable expectation of privacy on social media. While it appears that a privacy boundary has been negotiated online, society may not have reached a definitive consensus on whether or not this affords a reasonable expectation of privacy. Summary of Limitations While these results suggest that a general consensus about privacy expectations online has yet to coalesce, several limitations in the current study would need to be addressed before recommending legal action. A large sample was used, but it was not a representative one. Since survey respondents were recruited online, these individuals are more tech-savvy than the average user. The high correlation between knowledge of privacy controls and the use of stringent controls may not actually be reflective of the larger public. The fact that respondents self-selected to complete the survey for payment could also bias results. Another limitation is that a causal direction between online and offline privacy protection cannot be determined from the results. Further research on this question with an experimental design could attempt to definitively determine the directionality of this relationship. It should be noted, however, that in some Supreme Court cases, such as Katz, it has been enough for legal argument to analogize between two actions in order
62
to guarantee an expectation of privacy; for example, a traditional behavior and a behavior made possible by a new technology. Therefore, it may be sufficient to say that these offline and online privacy protections are similar in order to determine that an expectation of privacy exists on Facebook. Another limitation is that new measures were used. Although reliable, these measures may not be valid. I attempt to analogize interactions and behaviors on cyberspace with offline interactions and behaviors. Since the development of the Internet, justices and legal scholars have attempted to create these analogies in order to determine how law should apply to cyberspace. However, there are inherent differences between cyberspace and offline that make it very difficult to develop these analogies. Policy Considerations and Further Research This study suggests that social media users have negotiated a privacy boundary on Facebook by using privacy controls. However, this sense of privacy, if not legally recognized, may not be realized. Even though results suggest that Facebook users take similar steps to protect their online and offline privacy, regardless of generational cohort, current privacy law does not afford these individuals with an expectation of privacy in cyberspace. I hope that this research will contribute to the development of a new Katz style standard that protects privacy both online and offline. Justice Alito, in a case concerning the use of GPS tracking devices to monitor vehicles, wrote that the Katz test should be applied to a society that has a well-developed and stable set of privacy expectationsUnited States v. Jones, 565 U.S., (2012:10). While it is important to determine a Facebook users expectation of privacy, it is also important to acknowledge the groups on Facebook, such as older generational cohorts, who may disagree with this determination. Because there are significant differences in online behaviors across generations, it is difficult to currently label this reasonable expectation of privacy as well-developed and stable until it is better understood why these groups differ. There are many other questions that should be considered before policy is enacted. Further research could better determine why and in what situations Custom controls are used on Facebook. Also, privacy expectations may differ between Facebook and other social media sites such as Twitter and Instagram. Lastly, an important consideration in this
63
discussion is the fact that social media sites create privacy policies to which users must agree and adhere. Does agreeing to a privacy policy inherently alter a users expectation of privacy? Recent developments in privacy policy regarding social media sites offer clues to the direction which privacy law is headed. The Federal Trade Commission recently worked with Facebook to update its privacy policy to limit Facebooks ability to override its users privacy settings (Sengupta 2011). Facebook has also recently released statements warning employers against accessing a job applicants private content as a privacy violation (Duncan 2012). In order to better protect privacy on social media sites, District Court Judge Morrow ruled that social media messages and privatized wall posts were subject to the Stored Communications Act (SCA), Crispin v. Christian Audigier Inc, 717 F. Supp. 2d 991 (2010). The SCA, 18 U.S.C. 2701, is a legislative initiative that was passed to limit the ability of third-party Internet service providers to reveal information to government and non-government entities. This ruling follows a recent Supreme Court decision, Warshak v. United States, 532 F.3d 521 (2008), which limited the ability of Internet service providers to turn over the content of a users emails without the users knowledge. In general, these policy changes have favored greater privacy protections on social media sites. With these policy changes, social media sites are increasingly being guaranteed the same legal protections offered on other communication platforms. While some may argue that social media users should simply reveal less information online, this argument does not take into account changing behavioral norms. The Net Generation varies significantly from older generations in both self-revelation and privacy practices on social media sites. Instead of attempting to change these established behavioral norms on the Internet, policymakers should continue to protect and more concretely define the already established privacy boundary on social media sites. References Cited Barrington Moore, Jr., Privacy: Studies in Social Cultural History (Armonk, NY: M. E. Sharpe, 1984).
64
Bernhard Debatin, Jennette P. Lovejoy, Ann-Kathrin Horn, and Brittany N. Hughes, Facebook and Online Privacy: Attitudes, Behaviors, and Unintended Consequences, Journal of Computer Mediated Communication, 15 (2009): 83-108. Brad Stone, Web of Risks: Students adore social-networking sites like Facebook, but indiscreet postings can mean really big trouble, Newsweek, August 21, 2006, 76. Carol Wolinsky and James Sylvester, Privacy in the Telecommunications Age, Communications of the ACM, 35(2) (1992): 23-25. Crispin v. Christian Audigier, Inc. 717 F. Supp. 2d 965 (C.D. Cal. 2010) Daniel Solove, Conceptualizing Privacy, California Law Review, 90, (2002): 1087-1156. Darryn Beckstrom, Whos Looking at Your Facebook Profile- The Use of Student Codes to Censor College Students Online Speech, Willamette Law Review, 45, (2008-2009): 261-312. D.C. Beckstrom. Whos Looking at Your Facebook Profile- The Use of Student Codes to Censor College Students Online Speech, Willamette Law Review, 45 (2008-2009): 261-312. Donald Tapscott, Grown Up Digital: How the Net Generation is Changing Your World (NewYork: McGraw-Hill, 2009), 4. Elizabeth Butler, Elizabeth McCann, and Joseph Thomas. Privacy Setting Awareness on Facebook and Its Effect on User-Posted Content, Human Communication, 14(1), (2011): 52. Facebook Facts and Figures for 2010, DigitalBuzz, March 22, 2010, http://www.digitalbuzzblog.com/facebook-statistics-facts-figuresfor-2010. Facebook Statistics, Stats & Facts for 2011, DigitalBuzz, January 18, 2011, http://www.digitalbuzzblog.com/facebook-statistics-statsfacts-2011/. The Fuzz wants to add you as a friend, ZDNet, August 15, 2006, http:// www.zdnet.com/blog/education/the-fuzz-wants-to-add-you-as-afriend/411.
65
Geoff Duncan, Facebook warns employers: dont ask for passwords, Yahoo! News, March 23, 2012, http://news.yahoo.com/facebookwarns-employers-dont-ask-passwords-163806953.html. Iris Junglas, Norman Johnson, and Christiane Spitzmuller. Personality traits and concern for privacy: an empirical study in the context of location-based services, European Journal of Information Systems, 17 (2008): 387-402. Janet Kornblum and Mary Beth Marklein, What you say online could haunt you; Schools, employers scrutinize social websites such as MySpace and Facebook, USA Today, March 9, 2006, http://www. usatoday.com/educate/college/firstyear/articles/20060319.html. Jenny Benson, Saving Face: The Offline Implications of Behavior on Facebook, Dalhousie University: Unpublished doctoral dissertation, (2009): 1-140. Jim Giles, FBI releases plans to monitor social networks, New Scientist, January 25, 2012, http://www.newscientist.com/blogs/ onepercent/2012/01/fbi-releases-plans-to-monitor.html. John Palfry and Urs Gasser, Born Digital: Understanding the First Generation of Digital Natives (New York: Basic Books, 2008), 3-7. Junichi Semitsu, From Facebook to Mug Shot: How the Dearth of Social Networking Privacy Rights Revolutionized Online Government Surveillance, Pace Law Review, 31, (2011): 291-381. Katz v. United States, 389 U.S. 347 (1967). Kevin Lewis, Jason Kaufman, and Nicholas Christakis. The Taste for Privacy: An Analysis of College Student Privacy Settings in an Online Social Network, Journal of Computer Mediated Communication, 14 (2008): 79-100. Matthew Hodge, Fourth Amendment and Privacy Issues on the New Internet: Facebook.com and Myspace.com, Southern Illinois University Law Journal, 31(1), (2006): 95-122. Michael Caloyannides, Privacy vs. Information Technology, IEEE Security & Privacy Magazine, 1(1) (2003): 100-103.
66
Muck, P.M., Hell, B. and Gosling, S.D. Construct validation of a short five-factor model instrument: a self-peer study on German adaptation of the ten-item personality inventory (tipi-g). European Journal of Psychological Assessment, 23(3) (2007): 166175. Nicole B. Ellison, Charles Steinfield, and Cliff Lampe. The benefits of Facebook friends: Social capital and college students use of online social network sites. Journal of Computer-Mediated Communication, 12(4) (2007). Olmstead vs. United States, 277 U.S. 438 (1928). Philip E. Agre and Marc Rotenberg, Technology and Privacy: The New Landscape (Cambridge: MIT Press, 2007). Rita Zeidner, How deep can you probe? HR Magazine, October (2007):57-60. Samuel Gosling, Peter Rentfrow, and William Swann. A very brief measure of the big-five personality domains, Journal of Research in Personality 37 (2003): 504528. Samuel Warren and Louis Brandeis, The Right to Privacy, Harvard Law Review 4(5) (1890). Somini Sengupta, F.T.C. Settles Privacy Issue at Facebook, The New York Times Online, November 29, 2011, http://www.nytimes.com/2011/11/30/technology/facebook-agreesto-ftc-settlement-on-privacy.html. The Stored Communications Act 18 U.S.C. 2701 to 2712 (1986). United States .v Jones, 565 U. S. ____ (2012), No. 101259, slip op. (U.S. January 23, 2012), http://www.supremecourt.gov/ opinions/11pdf/10-1259.pdf. United States v. Warshak.532 F.3d 521 (2010). William Smith and Deborah Kidder, Youve been tagged! (Then again, maybe not): Employers and Facebook, Business Horizons, 53 (2010): 491-499. Willis Ware, Emerging Privacy Issues, Computers & Security, 5(2) (1986): 101-113.
67
Recommendations for Migrant Labor Reform in Qatar: Fulfilling a U.S. Foreign Policy Objective of Greater Respect for Human Rights In Gulf Cooperation Council Countries
Kharmika Tillery
Kharmika Tillery was born in Chicago and raised in Greensboro, North Carolina. She holds a Master of Public Policy Degree from the Sanford School at Duke University with a concentration on foreign policy and diplomacy. She received her Bachelor of Arts in International Relations and a minor in Spanish Language from the University of North Carolina at Chapel Hill. Kharmika has interned at the John F. Watson Public Policy Institute in Trenton, New Jersey; served on former Governor of New Jersey Jon Corzines re-election campaign; and interned on Capitol Hill in the office of Representative Barbara Lee (D-CA). Kharmikas research was inspired by her time as an intern at the U.S. Embassy in Doha, Qatar. She has also spent significant time volunteering and studying in Guatemala, Singapore, Malaysia, Thailand, and China. Kharmika is now serving as a U.S. Diplomat to Mexico at the U.S. Consulate General in Nuevo Laredo where she is in charge of political and economic affairs.
Abstract The heavy reliance on migrant labor in Qatar presents significant problems for these workers. Qatar benefits from imported labor, and migrants are able to earn more money than they would in their home countries. However, migrant workers from South and Southeast Asia are often deceived and exploited by sponsors and recruiters in the receiving and sending countries through physical and emotional abuse, inadequate wages, poor living conditions, and denial of freedom to travel or change jobs. The practices of recruitment agencies in the sending countries and the current labor laws of the Gulf Cooperation Council (GCC) do not favor the workers and have facilitated the continuation of abusive practices. The sponsor system should be reformed through liberalization of the current
69
Tillery: Recommendations for Migrant Labor Reform in Qatar
system. Simultaneously, the recruitment system should be reformed through bilateral engagements between Qatar and other nations with large numbers of citizens working in Qatar. Background Qatar has relied on significant levels of foreign labor since the 1930s. The oil boom of 1973 allowed it to accumulate vast revenues from oil exports (Winckler1997:480-493). As a result, the nations of the Persian Gulf were able to invest heavily in infrastructure, energy stations, governmental ministries and services, social services, healthcare, and education. Although Gulf nations possessed a significant amount of capital to invest in projects, they lacked a capable workforce (Ibid). Today, approximately 87 percent of Qatars workforce is composed of migrant laborers. At the beginning of the oil era, the majority of migrant laborers came from surrounding Arab nations. Qatar and other Gulf nations were also destinations for Palestinians following the Arab-Israeli war in 1948 and the occupation of Palestine (Kapiszewski 2006). The religious, cultural, and linguistic similarities initially made Arab labor more attractive than immigrants from other areas. However, the preference for Arab workers has changed significantly in recent years (Kapiszewski 2006). Leaders of the Gulf countries became particularly fearful of the radical social and political beliefs that non-Gulf Cooperation Council (GCC) Arabs brought with them, such as secularist and pro-Soviet ideologies. Also, Arab-led labor strikes threatened Qatars internal stability (Ibid). These prevailing attitudes have caused the employment of Arab laborers to decrease over time (Ibid). As a result, migrant labor from South and Southeast Asia to Qatar has surged. Asian migrant workers typically present lower costs to employers because they are viewed as more efficient, obedient, less demanding, and easier to fire (Kapiszewski 2007). Asian laborers are also more willing to migrate and work without their families, whereas many Arab laborers expect to bring their families and settle permanently. The fact that many South and Southeast Asians practice Islam adds to their favored status in Qatar, as GCC countries greatly value religion and tend to be very conservative (Ibid). Because of the major economic contributions, principally remittances, migrant laborers have made to Asian countries, Asian governments have
70
been instrumental in encouraging their citizens to work in Qatar and other GCC countries (Gardner 2011). Foreigners now make up a significant part of Qatars population. While migrant laborers have satisfied the labor needs of the Gulf countries, many of their citizens have begun to feel excluded from labor markets.As a result, Qatar has attempted to nationalize, or promote higher participation of the indigenous population in the national workforce (Dito 2010:70-86). A nationalization program called Qatarization has attempted to limit the amount of foreign labor by increasing citizen participation of the local workforce through training, skill acquisition, and pay incentives (Winckler 2010:9-13). Qatar was able to achieve higher participation of the citizenry in the public sector by lowering knowledge and experience requirements for recruits, and offering higher salaries, job security, and less strenuous work schedules (Sharp 2004). The Qatarization initiative has been less successful in private sector jobs because of the irregular work hours and a greater emphasis on employee performance (Naithani 2010). Qatari citizens often lack the technical and professional qualifications needed to realize the necessary level of productivity in the private sector (Attiya 2010). Many South and Southeast Asian workers also are willing to work in sectors that members of the national population are not, such as labor-intensive jobs in construction, household upkeep, and janitorial services.
Figure 1: Top Source Countries of Immigrants in GCC Countries in 2010
Number of GCC Countries
Country
Chart Adapted from Data Source: The Migration and Remittances Factbook 2011, World Bank.
71
Since the nationalization program has yet to be effective at replacing migrant laborers with Qatari nationals of the same qualifications, Qatar is likely to continue receiving large amounts of migrant labor to maintain the nations productivity. Most of the labor will continue to come from South and Southeast Asia. The table below illustrates how the migrant labor population continues to grow in Qatar, Oman, and Bahrain and slightly decreases or remains steady in the remaining GCC countries.
Table 1: International Migrants as a Percentage of the Population from 1990-2010
U.S.-Qatar Relations The United States believes that respecting human rights helps to maintain the peace, discourage aggression, fight against crime and corruption, promote the rule of law, strengthen democracies, and prevent humanitarian disasters. Due to the United States amiable relationship with Qatar, ideas about migrant labor reform have a platform in Qatar. The ruler of Qatar, Shaikh Hamad bin Khalifa Al-Thani, has initiated efforts to politically liberalize Qatari society. The countrys effort to align itself with the United States can facilitate openness regarding migrant labor reform. Qatar has served as an ally to the UnitedStates in the war on terrorism, and the United States and Qatar have favorable business relations through economic ties in the oil and gas industry. Problem The heavy reliance on migrant labor in Qatar presents significant policy problems for the migrant workers. These problems include a restrictive sponsorship system, physical and emotional abuse, inadequate wages, and poor living conditions. The practices of sending countries recruitment
72
agencies and current labor laws in Qatar disadvantageworkers and have facilitated the continuation of problems. Restrictive Sponsorship System The kafala system is the legal basis for residency and employment of foreigners employed by GCC nations (Winckler 1997). More than 85 percent of Qatars workforce is made up of third-country nationals who are entirely dependent on their sponsors. Under Qatars sponsorship system, migrant workers receive an entry visa and a residence permit only if a Qatari citizen or institution sponsors them. Employers are required to assume full economic and legal responsibility for the employees while they are contracted to those sponsors. The laborers can only work for the employers that are sponsoring them. Thus, the workers ability to remain in the country depends on maintaining his or her contract with that sponsor (Colton 2010). The sponsorship system is the central institution that defines identities, rights, and obligations for employers and migrant laborers. Although they can be renewed, entry visas are typically short-term and can be terminated by the employer at any time (Ibid). The government of Qatar does not grant citizenship to migrant workers. As a result, the kafala system creates a dynamic of uncertainty and job insecurity (Auwal 2010). On August 1, 2009, Bahrain introduced and later passed the Labor Market Regulatory Law known as Decision No. 79 (Sajjad 2009). The objective of the new law was to abolish the sponsor system; allow foreign workers to easily move in and out of new jobs; improve working conditions for migrant laborers; increase the ability for Bahrainis to obtain jobs, by prohibiting employers from forcing workers to remain employed indefinitely; end the practice of bringing in workers and dumping them in the market for monthly fees; and increase salaries and wages (Ibid). The law also requires companies to apply for a visa through the Labor Market Regulation Authority (LMRA) whenever they want to hire a foreign employee, whichwas not previously the case. The LMRA would then become the workers official sponsor. Under this type of sponsorship, the worker has the capacity to switch jobs after issuing an end-of-contract
73
notice delivered by mail three months in advance (Ibid). The law applies to both the public and private sectors, but it does not apply to domestic workers (Scrapping Sponsorship 2009). Mistreatment and Abuse Mistreatment and abuse mostly involves nationals of South Asia, Indonesia, and the Philippines. Accounts of mistreatment include nonpayment or late payment of wages, sexual harassment, contract switching, physical torture or torment, and overwork. Embassies in Qatar are known to provide shelter and relief to their citizens who have fled as a result of altercations and abuses. In 2007, the Embassies of India, Nepal, and Sri Lanka received a combined total of more than 15,000 complaints from both male and female workers reporting mistreatment by their employers (Department of State 2010). However, many workers did not press charges for fear of losing their jobs (Ibid). Deceptive Recruiting Practices Recruiting agencies serving as labor brokers in the sending countries factor into the exploitation of many South and Southeast Asian migrants seeking work in the Gulf. Labor brokers charge a large fee to potential laborers in exchange for the opportunity to seek better employment abroad. Thus, it is in the agencies interest to avoid discussing the common challenges and difficulties low-skill migrants face in the GCC nations. Labor brokers often portray an image of higher wages and better working conditions, but fail to inform workers of the limited ability to return home (Ibid). Inadequate Living Conditions The living conditions of foreign workers are a serious concern in Qatar. Diplomatic representatives who visited migrant residences reported that the majority of workers are living in poor conditions. Living quarters are overcrowded and dirty, and often lack running water, electricity, and sufficient food (Ibid). An investigation by the Labor Inspection Section concluded that many of the labor camps investigated through random assignment were below minimum benchmarks as specified by international labor standards (Ibid).
74
Lack of Government Enforcement The Ministry of Energy and Industry, the Ministry of Health, and the Labor Department are responsible for the enforcement of the labor regulations created by the government (Department of State 2011). Though reported to be improving, these agencies suffer from inadequate training and low levels of personnel levels (Ibid). Qatar does not have an official minimum wage. In most cases, the pay foreign workers receive is not sufficient to provide a decent standard of living for a worker and his or her family. Under Qatari law, the workweek is not supposed to exceed 48 hours and should include a 24-hour rest period (Ibid). During the month of Ramadan, workers are not allowed to work more than 36 hours a week (Ibid). However, the law was often disregarded with respect to unskilled laborers and domestic and personal employees (Ibid). Many of these workers worked seven days a week and more than 12 hours a day. In addition, they received few holidays and no overtime pay (Ibid). Weak Enforcement of Labor Laws The cycle of exploitation continues because of the weak labor laws and the failure to prioritize migrant labor rights in both sending and receiving countries (Auwal 2010). Labor laws in GCC countries favor employers, because current laws make it very difficult for migrants to seek legal recourse when they are mistreated. Xenophobia among GCC country officials has led governments to implement laws and policies that violate the fundamental human rights of migrant workers (Ibid). Dr. Susan Martin, an associate professor of international migration at Georgetown University, believes the biggest impediment to guaranteeing migrant laborers more rights can be summed up as a combination of problems created by unfair sponsorship systems, deceptive recruiters, and unresponsive governments. She identifieda need for a more liberal sponsorship system, more government enforcement, and more transparency and accountability on behalf of those recruiting migrant laborers, in order to alleviate and ultimately eliminate mistreatment and hardship foreign workers experience.
75
Criteria The following set of criteria is used to weigh the validity of the proposed policy options based on U.S. foreign policy interests. Ensure policy options are effective in eliminating human rights abuses against migrants. Ensure that the selected policy option is politically feasible given the political economy and culture of Qatar. Protect and promote interests central to U.S. foreign policy, such as greater respect for human rights and the promotion of strong and stable democracies. Policy Options 1. Liberalize the sponsorship system. 2. Reform the recruitment system through bilateral relations with large sending countries. 3. Create a government agency exclusively for overseeing migrant laborers and enforcing migrant labor laws. 4. Modify the nationalization process currently in place, based on an analysis of the mistakes and successes from past nationalization efforts. Analysis In the analysis, I weigh the above policy options against the criteria to evaluate which policy options are the most viable. 1. Liberalize the Sponsorship System. Maximize Effectiveness. Liberal policies include being able to transition from one job to another without permission and being able to move freely within the country and in and out of the country. If Qatar were to adopt liberal measures as a means of improving the sponsorship system, the nation could succeed in improving migrant labor conditions. While Bahrain did modify its sponsorship system, it has not appeared to give migrants more rights. Qatar has the opportunity to implement a type of migrant labor
76
reform already tested in a neighboring GCC country and modify the law to obtain a more favorable outcome. The results from Bahrains new legislation suggest abolition has not been entirely effective. Observers have stated that the passage of the law appeared to be merely symbolic,as many of the measures taken were applauded by the media but had no real impact for migrant workers. Reports state that while the law now requires new workers to be sponsored by the government, this law is not significantly different from the previous system, under which the government issued the two-year work visas. Workers are still required to have a company or individual sponsor them in order to remain in the country (Scrapping Sponsorship 2009). The law has strengthened migrant laborers ability to move from one sponsor to another with more ease. The capacity to leave has helped workers escape abusive employers or demand better working conditions and wages. However, the new law is wrought with technicalities that favor employers (Ibid). In the event that a worker does want to leave his or her employer, vengeful sponsors can punish their workers by canceling their visas before they have a chance to switch sponsors. The law has proven ineffective in combating deceptive recruiting practices (Ibid). The lack of accountability and enforcement allow these practices to still occur at high rates (Ibid). Although reforming the sponsorship system by abolishing it seems to be ineffective in Bahrain, Qatar can modify the policies used in Bahrain to achieve the desired results, or something equivalent. While Qatar may be unwilling to abolish the sponsorship system, changing certain aspects of the system to give more autonomy to migrants can be effective. Ensure Political Feasibility. Although legislation to abolish the sponsorship system passed in Bahrain, the measure was not without opposition. While Bahraini Labor Minister Dr. Majeed Al Alawi likened the sponsorship system to modern slavery, the decision was met with fierce resistance from the business community, which called for Al Alawis resignation (Toumi 2009). Similar discontent exists in other GCC nations. Emirati business leaders in the UAE also strongly oppose abolishing the sponsorship system, as they believe it will affect the interests of the private sector and alter the demographic structure (Ibid). While there is strong opposition from some like the Bahraini business community, there also exists the belief that an alternative to the sponsorship
77
system is needed. Leaders from the six GCC countries have proposed alternatives to the sponsorship system, such as imposing residence caps on foreigners (Ibid). Lieutenant General Dhani Khalfan Tamim, Dubais Police Chief, supports abolishing the sponsorship system because, Emiratis have become responsible for each and every detail of their employees lives. This system has overburdened UAE nationals and made them accountable for their employees problems, and therefore it should be scrapped (Ibid). Al Alawi has been at the forefront of this movement in the GCC. In spite of the opposition in other countries, influential decision-makers in Qatar desire to change the sponsorship system. In 2010, Qatar Prime Minister ShaikhHamad Bin Jasem Bin Jabor Al Thani agreed that changes to the sponsorship system were necessary. He stated, The government is now seriously looking into the sponsorship system, and in light of the latest developments, the issue requires some legal and other measures (Toumi 2010). Even if Qatari officials do not want to eliminate the system, they are open to making changes to the system, which could include increased liberalization. Protect and Promote Interests Central to U.S. Foreign Policy. Reforming the sponsorship system should improve human rights and bring about stable political conditions, both of which are conducive to positive business connections. Removing the sponsorship system in Qatar would improve the status of most migrant workers and reduce their risk of exploitation. Reforming the sponsorship system would also increase workers ability to select where they work, ultimately forcing employers to provide better conditions and better wages (Scrapping Sponsorship 2009). Qatar can liberalize the sponsorship system by creating laws that favor migrant workers and ultimately uphold the United States foreign policy interests in human rights and political stability. 2. Reform the Recruitment System through Bilateral Relations with Large Sending Countries. Maximize Effectiveness. Reforming the recruitment system through bilateral engagement with nations that have large numbers of their citizens working in the Gulf would be effective in eliminating abuses of migrant workers. Some sending countries,in partnership with the GCC nation of Kuwait, have created domestic policies to eliminate the number of deceptive
78
recruiters and to provide more accountability for low-skilled and unskilled migrant workers in leaving and entering their respective countries. Dr. Andrew Gardner, a professor of anthropology at the University of Puget Sound in Tacoma and a researcher of the unskilled labor force in Qatar, states, Many of these labor migrants have little or no contact with their official sponsor anyway, but instead face a phalanx of managers and middlemen who are often the direct culprits in creating the problems they face. Collaborative efforts between Qatar and sending countries increase scrutiny of deceptive recruiters looking to exploit uneducated migrant laborers and eliminate the inefficiency caused by confusing networks that lack transparency and accountability. Requiring labor agents to register and receive a permit is one way to decrease the number of exploitative labor recruiters. Qatari officials can collaborate with officials in the sending countries embassies to monitor the treatment, payment, and living conditions of foreign workers. Combining efforts through bilateral collaboration can achieve positive results in migrant labor reform. Accountability measures can work to decrease the number of deceptive recruiters that exploit migrants by taking advantage of the lack of documentation in sending and receiving countries. Dr. Garner also highlights that the nature of the policy plays a role in its effectiveness, noting that a policy that has less emphasis on shaming Qatar and has more emphasis on collaboration and cooperation will be more successful. Ensure Political Feasibility. Reforming the recruitment system through bilateral relations with large sending countries would be politically feasible in Qatar. No part of Qatari society has spoken out against bilateral collaboration with sending nations. According to Dr. Mary Breeding, a consultant with the World Banks Independent Evaluation Group, this type of engagement is already taking place on some levels. The Ministry of Overseas Indian Affairs has worked with Qatari officials in some capacity and has encouraged them to be more open with their recordkeeping and documentation. Dr. David Mednicoff, an assistant professor of public policy at the University of Massachusetts-Amherst and a specialist in Middle Eastern politics, explains another reason why Qatari officials would be open to bilateral engagement. He states, Qataris will feel like the finger isnt just
79
being pointed at them; both countries will be regulating the policies. Qatar really cares about diplomacy; foreign relations and treaties make it a more complex policy game that could work very well. Protect and Promote Interests Central to U.S. Foreign Policy. Reforming the recruitment system through bilateral engagement will help protect and promote human rights interests central to U.S. foreign policy. Accountability from both Qatar and the sending country will have a double impact on decreasing the deception that occurs when labor recruiters take advantage of unknowing workers. Increased accountability will also reduce the number of abuses that happen without legal recourse. Workers will be more empowered to take action against the wrongs they suffer regarding nonpayment, delayed payment, poor working conditions,and unsatisfactory living conditions. Fewer human rights abuses will occur if both the sending country and Qatar are accountable for the migrant worker. 3. Create a Government Agency Exclusively for Overseeing Migrant Laborers and Enforcing Migrant Labor Laws. Maximize Effectiveness. Creating an agency exclusively for overseeing migrant laborer affairs risks being ineffective in ending abuses of migrant laborers. The agencies that collectively oversee the migrant labor affairs in Qatar are the Ministry of Energy and Industry, the Ministry of Health, and the Labor Department. These agencies already lack staff members and proper training; the creation of a separate agency risks experiencing the same problems. Zahra Babar elaborated on this point when she said, In Qatar, one of the issues is also just lack of state capacity in terms of monitoring the conditions for workers. They have set up agencies such as the National Human Rights Council and other bodies to protect workers rights, but the capacity is still lacking. Creating a separate agency may be well intentioned, but ultimately ineffective if Qatari officials lack the expertise and training to oversee agency functions. Ensure Political Feasibility. Dr. Gardner has observed changes in Qatari labor officials views on enforcing labor laws. Dr. Gardener stated, For the first time in a decade, I was hearing labor migrants in the camps talking about labor courts and positive, or at least non-devastating, verdicts on those courts. The existence of labor courts and the increased enforcement of existing labor laws indicate that government officials in
80
Qatar are attempting to use institutions to limit mistreatment among foreign workers. A government agency created exclusively for overseeing migrant labor rights is therefore politically feasible given the observed changing atmosphere regarding migrant labor. Babar, also agreed that the creation of a new agency is politically feasible. Qatar has recently established an agency designed to support migrants in accessing the legal system when a worker has a dispute with his employer over wages or working conditions. The point of the agency is to provide assistance to migrants when they are dealing with the courts (Babar 2012). If Qatar is already working to establish institutions of accountability for migrant workers, it may be feasible for the government of Qatar to create an institution that monitors migrants. Protect and Promote Interests Central to U.S. Foreign Policy. Establishing an effective oversight agency to handle migrant labor affairs will promote U.S. interests in human rights by demonstrating accountability and transparency. Documenting migrant workers, monitoring the way they are treated, and enforcing existing migrant labor laws when they are broken will increase the protection of migrants in Qatar, which is consistent with U.S. foreign policy objectives in the region. Creating an agency exclusively for migrant rights can also promote political and economic stability. An agency monitoring migrant labor affairs and enforcing migrant labor laws may promote domestic stability by addressing issues appropriately and fairly, thereby quelling the possibility of unrest among migrant workers. 4. Modify the Nationalization Process Currently in Place, Based on an Analysis of the Mistakes and Successes from Past Nationalization Efforts. Maximize Effectiveness. Revamping the nationalization processes will have little to no effect on improving the human rights situation for migrant workers because,as a cultural norm, Qataris do not occupy low or unskilled labor and domestic positions. Nationalization efforts in Qatar, or Qatarization, are very important in helping nationals feel secure about their cultural identity in a country where almost 90 percent of the inhabitants are third-country nationals. However, nationalization efforts have had little impact on employment sectors. Dr.
81
Gardener explained, The GCC states have been talking about localization [i.e., nationalization] for decades and its had almost no impact. Zahra Babar believes revamped nationalization efforts will most likely not succeed. She stated, Considering that there are only about 100,000 Qataris in the workplace, or less, and about a million foreign workers, Qatarization is never going to result in Qatari citizens taking all the jobs. The problem with Qatarization is that it virtually guarantees Qataris job placement either in the public or private sector, and as a result impedes motivation, competition, and proper skill development for advancement amongst Qataris. Based on Babars assessment, Qataris are getting placed in high-skilled professional jobs simply to keep non-citizens out of them. However, the abuses against migrants mostly affects South and Southeast Asian workers in domestic, low-skilled, or unskilled jobs. Consequently, attempts to nationalize the workforce will not have a meaningful impact on eliminating abuses against migrants. Ensure Political Feasibility. Restructing the nationalization efforts is politically feasible in Qatar. Nationalization efforts are an enduring movement in many GCC countries, just in different forms. For example, a sub-regional proposal presented in 2004 at the Gulf rulers summit aimed to limit the period a foreigner can work in all Gulf States to six years (Kapiszewski 2007). In 2008, Qatari Minister of Labor and Social Affairs Hassan Al Dhabit Al Dosri announced that all private sector and public sector companies had to reserve 20 percent of their workforce for Qatari citizens (Qatar Announces). The government also wants Qatar to be able to compete in the private sector and has taken action outside of mandating quotas to ensure this. In 2003, the country announced plans for a restructuring of the kindergarten through 12th grade education system, with help from the Rand Corporation to educate a better skilled labor force for the future (Glasser 2003). Protect and Promote Interests Central to U.S. Foreign Policy. Modifying the nationalization efforts in Qatar will have a minor effect on protecting and promoting interests central to U.S. foreign policy. Limiting foreign workers can have a negative impact on political and economic stability in Qatar by causing unrest among those foreigners that will be replaced if the nationalization process is successful. Revamped nationalization can
82
also impact the sending countries by decreasing remittance flows to their economies and prompting their citizens to work in other parts of the world. Recommendation It is recommended that Qatar pursue are form of the recruitment system through bilateral engagement between Qatar and other nations that have large numbers of citizens working in Qatar, and through liberalization of the current system. Liberal policies include being able to transition from one job to another without permission and being able to move freely within, into, and out of the country. Of all the policy options, liberalization and reform through bilateral engagement are the most viable after being weighed against the three criteria.
References Cited Ahmed, Attiya. Migrant DomesticWorkers in Kuwait: The Role of State Institutions. Viewpoints: Migration andtheGulf(2010). http://www. voltairenet.org/IMG/pdf/Migration_and_the_Gulf.pdf Auwal, Muhammad A. Ending the Exploitation of Migrant Workers in the Gulf. The Fletcher Forum of World Affairs Journal (2010): 87108 Babar, Zahra. Skype correspondence, February 29, 2012. Breeding, Mary. telephone conversation, February 29, 2012. Colton, Nora Ann Colton. The International Political Economy of Gulf Migration. Middle East Institute. http://www.mei.edu/content/internationalpolitical economy-gulf-migration Department of State. Country Reports on Human Rights Practices. April 8, 2011. http://www.state.gov/j/drl/rls/hrrpt/2010/nea/154471.htm ---. Qatar: Background and U.S. Relations. Congressional Research Service Report. Order Code RL31718, 2006. http://fpc.state.gov/ documents/organization/33741.pdf
83
Dito, Muhammad E. Labor Migration in the GCC Countries: SomeReflections on a Chronic Dilemma. Viewpoints: Migration and the Gulf (2010). http://www.voltairenet.org/IMG/pdf/Migration_and_ the_Gulf.pdf Gardener, Andrew. Labor Camps in theGulf States. Viewpoints: Migration and the Gulf (2010). http://www.voltairenet.org/IMG/pdf/Migration_ and_the_Gulf.pdf ---. Skype correspondence, February 29, 2012. ---. Why do theykeepcoming? Labor Migrants in the Gulf. Migrant Labor in theGulf Summary Report, no. 2 (2011). http://www12.georgetown.edu/ sfs/qatar/cirs/ MigrantLaborSummaryReport.pdf Glasser, Susan B. Qatars Educational Reform Triggers Unease Among Muslims http://www.guardian.co.uk/education/2003/feb/13/ internationaleducationnews.hig International Organizationfor Migration. 2010 World Migration Report: The Future of Migration: Building Capacity for Change. http://publications. iom.int/bookstore/index.php?main_page=index&cPath=37 Kapiszewski, Andrzej Arab Versus Asian Migrant Workers in the GCC Countries. United Nations Expert Group Meeting on International Migration and Development in the Arab Region. Beirut, Lebanon, May 15-17, 2006. ---. De-Arabization in the Gulf: Foreign Labor and the Struggle for Local Culture. Georgetown Journal of International Affairs (2007): 81-88. Hein Online. Keane, D. and McGeehan, N. Enforcing Migrant Workers Rights in the UAE. International Journal on Minority and Group Rights (2005) 81115. Khadria, Binod. Paradigm Shifts in Indias Migration Policy toward the Gulf. Viewpoints: Migration and the Gulf (2010). http://www. voltairenet.org/IMG/pdf/Migration_and_the_Gulf.pdf KhatriSridhar K. Labour Migration, Employment and Poverty Alleviation in South Asia. Proceedings of the Regional Seminar jointly organized by South Asia Centre for Policy Studies (SACEPS) and Friedrich Ebert Stiftung (FES). Hotel Himalay, Kathmandu, Nepal, August 9-10, 2007.
84
Martin, Susan R. e-mail message to author, February 2, 2012 PranavNaithani. Challenges Faced by Expatriate Workers in Gulf Cooperation Council Countries. International Journal of Business and Management 5 (2010): 98-103. Accessed September 12, 2011. http://www.ccsenet.org/ijbm Qatar Announces Nationalization Job Drive. Gulf News. http:// gulfnews.com/news/gulf/qatar/qatar-announces-job-nationalizationdrive-1.92899 Roper, Steven D. Labor Migration to the Gulf: Understanding Variations in the Kafala System. Accessed September 9, 2011 at http://www. stevendroper.com/Kafala.pdf. Sajjad, Valiya S. How Government Sponsorship Works in Bahrain. Arab Times. http://www.arabtimesonline.com/NewsDetails/tabid/96/ smid/414/ArticleID/160050/reftab/96/Default.aspx Scrapping Sponsorship will Reduce the Risks Gulf Daily News. May 15, 2009. LexisNexis. AccessedJanuary 18, 2012. Toumi, Habib. GCC tostudy alternatives tosponsorship Gulf News. July 20, 2009. General OneFile. Accessed January 18, 2012. http:// gulfnews.com/news/gulf/bahrain/gcc-to-study alternatives-tosponsorship-1.501890 ---. Qatar moves towards scraping controversial sponsorship system. Gulf News. November 10, 2010. General OneFile. AccessedJanuary 18, 2012. Viewpoints: Migration and the Gulf. 2010. Middle East Institute. www. mei.edu Winckler,Onn. The Immigration Policy of the Gulf Cooperation Council Countries (GCC) States. Middle Eastern Studies 33.3 (1997) 48093. Doi:10.1080/00263209708701165 ---. Labor Migration tothe GCC States: Patterns, Scale, andPolicies. Viewpoints: Migration and the Gulf(2010). http://www.voltairenet. org/IMG/pdf/Migration_and_the_Gulf.pdf
85
Gideons Bullhorn: Sounding a Louder, Clearer Call for a Civil Right to Counsel
Jonathan D. Kelley
Jonathan D. Kelley is a Juris Doctor candidate at the Duke University School of Law. He is a Fulbright Fellow, and earned his Bachelors degree from Arizona State University in 2007. After graduating from law school in May 2013, he will work in private practice in New York City.
Abstract While the right to counsel in criminal proceedings was established by the Supreme Court in Gideon v. Wainwright, the Supreme Court has chosen not to extend that right to the civil context. Since inception, the Civil Gideon movement has sought to establish a civil right to counsel, but has been plagued by setbacks. This article examines those setbacks and advocates for the implementation of a realistic and pragmatically constructed civil equivalent to the right to counsel established in Gideon. By reconciling the heart of Gideon with modern conceptions of morality and offering both criticisms and suggestions, this article offers a renewed vision for the Civil Gideon movement, one that is rooted in the moral/formal tensions of legal liberalism, and that attempts to strike a proper balance between utopian visions of legal paradise and realistic understandings grounded in the pessimism of temporal and human limitation.
Introduction In August of 2006, the American Bar Association (ABA) House of Delegates passed Resolution 112A, which specifically encouraged legislatures to provide legal counsel to low-income persons in civil proceedings involving basic human needs (American Bar Association 2006). This resolution in many ways embodies the heart of the Civil Gideon movement, a movement designed to provide civil defendants the same right to counsel given to criminal defendants. The bedrock, and namesake, of the
87
Kelley: Gideons Bullhorn
Civil Gideon movement is the Supreme Courts 1963 decision in Gideon v. Wainwright (see Table1) ,a decision as unlikely as it was radical. In it, the Court held that the Sixth Amendment of the U.S. Constitution, which provides that in all criminal prosecutions the accused should enjoy the assistance of counsel, is obligatory upon states through the Fourteenth Amendments Due Process Clause (Gideon v. Wainwright 1963).1 Though Gideon was progressive in 1963, today, most Americans take the right to counsel in criminal proceedings for granted, as many have grown up knowing nothing but access to this right. As time passes, and as the United States continues to weather the worst recession since the Great Depression (Domitrovic 2013), it is the right time to take an honest look at what a modern equation of civil justice, couched within a rigorously moralistic framework, should look like. With the current historically high unemployment rate and overborne prisons and judicial systems, our responsibility to provide recourse for those charged with a civil wrong ought to be re-examined from both a legal and public policy perspective (Houseman 1998:369). Because determining the correct legal outcome is rendered almost impossible without effective counsel, (Sweet 1998: 503) we, as a just society, ought to ask whether the judiciary is providing this. In the words of Civil Gideon advocate Alan Houseman, [n]ow is the time to reassess the mission, purposes, objectives, and structure of the national civil legal assistance system and determine how it should be changed to achieve equal justice for all (Houseman 1998:369). As propitious a beginning as Gideon v. Wainwright signaled for indigent litigants, the decision itself completed only half of the equation. Because Gideon applies only to criminal defendants, indigent defendants in civil suits still lack right to counsel, and are therefore left to finance the high price of counsel themselves, or, alternatively, represent themselves pro se.2 Consequently, shortly after and in continuity with the Courts decision in Gideon, a Civil Gideon movement began with the goal of providing a constitutional right to counsel in some or all civil cases . . . [thereby ensuring] increased funding and provid[ing] access to civil legal assistance in some cases (Houseman 1998:369). In light of this goal, it was hoped that a case similar to Gideon might allow the Supreme Court to establish a right to counsel in the civil context. In 1981, Civil Gideon proponents were hopeful that Lassiter v. Department of Social Services (see Table 1) would be the case to establish a
88
right to civil counsel. Lassiter concerned the Durham County Department of Social Services attempt to terminate the parental rights of Abby Gail Lassiter. Lassiter lost custody of her infant child to the Durham County Department of Social Services for failing to provide adequate medical care. A year later, Lassiter was convicted of second-degree murder and incarcerated for two years. When the Department of Social Services brought suit and terminated her parental rights, Lassiter was unable to secure a lawyer and was denied court appointed counsel. Lassiter appealed, arguing that the court should provide legal counsel because she could not afford such assistance herself. Ultimately, the Supreme Court held that the Fourteenth Amendments Due Process Clause did not entitle Lassiter to legal counsel, and that her parental rights could be terminated without representation (Lassiter v. Department of Social Services 1981:24, 33). Consequently, the Court held that a case-by-case decision was appropriate indetermining when due process required representation, and that such analysis should be based on three factors previously identified in Matthews v. Eldridge (1976).3 The Court concluded that, an indigent litigant has a right to appointed counsel only when, if he loses, he may be deprived of his physical liberty (Lassiter v. Department of Social Services 1981:26-27). This ruling was not ideal for Civil Gideon proponents hoping for legal recognition of a civil right to counsel. By setting the standard for required counsel at the deprivation of physical liberty, while denying extension of Gideons holding to the civil context, the Court both instituted similar case-by-case judgment previously repudiated by the Court in Gideon, and ensured that very few litigants could satisfy the constitutional standard for required appointment of counsel (Brown 2011:901). Despite the fact that the Supreme Court decided not to extend Gideon to the civil context in Lassiter, the time is ripe for a re-examination of the values espoused in both Lassiter and Gideon. In addition to its legal holding, the Courts decision in Gideon stands beside the American conception of due process guaranteed by the Constitution. The Courts reasoning represents both a moral and a legal statement about what fundamental fairness the U.S. judiciary should provide citizens as mandated by the Constitution. In light of this clarion call, and in furtherance of the vision established by the sounding of Gideons trumpet,4 this article argues for the implementation of a pragmatically constructed civil equivalent to the right to counsel established in Gideon. Such a right must simultaneously
89
convince the ambivalent and appease the zealot, and in so doing strike the appropriate balance between utopian visions of legal paradise (Lee 2004:367) and realistic understandings grounded in the pessimism of temporal and human limitation. If middle ground between the law-policy dichotomy can be found, it is plausible to ensure for all the principle enshrined on the Supreme Courts marble portico: Equal Justice Under Law. Part I of this article discusses case law established after Gideon and offers a resolution to the law-policy dichotomy by reconciling the Gideon decision with modern conceptions of morality. Part II provides an overview of alternatives to the civil right to counsel and discusses recent advances in the Civil Gideon movement. Part III argues for the refashioning of key elements of Civil Gideon, and articulates a new, streamlined version of the movement that might enable it to achieve recognition of a constitutional right to counsel in civil cases (see Table 1 for a summary of cases discussed in this article).
Table 1: Summary of Pertinent Supreme Court Cases and Respective Holdings

Case Powell v. Alabama Gideon v. Wainwright Lassiter v. Department of Social Services Strickland v. Washington Year 1932 1963 1981 1984 Supreme Courts Holding Court appointed counsel must be capable of providing effective and substantial aid Every criminal defendant is entitled to counsel Civil defendants are not entitled to counsel Two-pronged inquiry for establishing ineffective assistance of counsel
Part I: Gideons Flaws and Reconciling the Inherent Law-Policy Dichotomy Gideons Foundational Fissures, Wrought By Antithetical Precedent The historically unprecedented advances made by the Gideon decision for indigent defendants permanently changed the face of legal representation in the United States. However, despite the evolution of constitutionally-required representation, counsel provided to Americas
90
poorest citizens (Matthew 25, Bible)5 is often woefully inadequate, and attempts to challenge the inefficacy of such representation have been onerous (Cole 2006:101). Despite the fact that criminal defendants enjoy the right to an attorney, there are few quality control mechanisms in place for appointed counsel. In his book Gideon v. Wainwright and Strickland v. Washington: Broken Promises in Criminal Procedure Stories, Georgetown Law Professor David Cole exposes numerous examples of inadequate court-appointed counsel, including instances in which appointed counsel slept during portions of the trial, used drugs throughout the trial, admitted to having insufficient knowledge of the facts of the case, or was unable to name a single Supreme Court case on the death penalty (Cole 2006:101). As illustrated by these examples, appointing counsel who provide subpar representation strips Gideon of its radical dexterity, and significantly undermines its core objective. Despite the frequent inefficacy of appointed counsel, the threshold for successfully suing an attorney for legal malpractice remains extremely high, and as a result, such suits are rarely successful. None of the aforementioned offenses related by David Cole resulted in a successful malpractice suit. Gideons Questionable Efficacy: Examination of Two Pertinent Cases While the fault lines riddling post-Gideon jurisprudence may be selfevident, the cause of Gideons relative failures remains speculative. Many argue that the overburdened public defender system is the obvious culprit (Brown 2011:896). Others believe the problem is rooted in a chronic lack of funding to legal aid organizations that might otherwise provide service to the underrepresented (Rhode 2004:369). Whatever the reason, the problems posed by the inadequacy of court-appointed counsel are further compounded by the Supreme Courts 1984 decision in Strickland v. Washington (see Table 1), which established the standard by which a client may sue his counsel for inadequate representation. Though both involve the right to counsel in the criminal, not civil, context, Gideon and Strickland together offer a potential conceptual framework within which the Civil Gideon movement might eventually operate. Should the Civil Gideon movement become successful in its mission to extend a constitutional right to counsel to civil defendants, the movement ought to carefully consider the evolution, or devolution, of the right to counsel in the criminal context, and adjust its strategy accordingly.
91
Despite the failures of Gideon and the setback posed by Strickland, it is important to note that the Supreme Court has long recognized the right to effective assistance of counsel when counsel is provided by the state. In 1932, in Powell v. Alabama (see Table 1), nine young black men were charged with the rape of two young white women while traveling by freight train through Alabama. The trial court attempted to appoint the defendants counsel by appoint[ing] the whole bar of Scottsboro (six men) to represent them. Only one attorney agreed to do so only on the morning of trial (Cole 2006:106). The defendants, who had previously escaped a lynch mob when first arrested, were sentenced to death, and eventually sought appeal to the U.S. Supreme Court. The Supreme Court reversed the trial court, holding that the defendants were entitled to court-appointed counsel under the Due Process Clause of the Fourteenth Amendment but had been deprived of that right. The Court reasoned that the designation of counsel as was attempted was either so indefinite or so close upon the trial as to amount to a denial of effective and substantial aid (Powell v. Alabama 1932:287). The Supreme Courts holding in Powell demonstrates that prior to the Courts definitive standard for measuring efficacy of counsel in a criminal case, the Court was cognizant of two things: 1) the Fourteenth Amendment guaranteed the right to court-appointed counsel in some cases; and 2) court-appointed counsel must be capable of providing effective and substantial aid (Ibid.). In Strickland, the Court established a two-part test for establishing a claim of ineffective assistance of counsel that today functions as thede facto standard in criminal cases. Under this test, a defendant cannot obtain relief against his former counsel unless it is established that the counsels performance fell below an objective standard of reasonableness, and thatthere is a reasonable probability that the result of the proceeding would have been different if counsels performance had been adequate (Strickland v. Washington 1984:694). Ironically, just as Lassiter posed a series of relatively unsympathetic facts, which in turn yielded an unfavorable result for the defendant and for the Civil Gideon movement, so the facts in Strickland made it what Professor Cole called probably the worst possible case to set forth the parameters for effective assistance of counsel (Cole 2006:108).6 Unsurprisingly, the standard set by the Court in Strickland is widely criticized as unrealistic, unsympathetic, and prohibitively high (Cole 2006:547). Beyond its limiting nature, critics argue that Strickland
92
violates the Sixth Amendment right to effective assistance of counsel for many criminal defendants . . . [and] provides less incentive for states to fund indigent defense services adequately (Brown 2011:547).7 Strickland offers another reason to lament the fact that Gideon has not fulfilled its promise of providing competent counsel to indigent criminal defendants (Brown 2011:895). An examination of Strickland makes clear that the low quality of court appointed counsel is not the only factor to blame for the lack of substantial change in defendants rights. However, it is undeniable that since Gideon, the change originally intended by the decision has not been actualized. Therefore, fashioning Gideons civil equivalent on the basis of a decision that has experienced trouble realizing its potential is tantamount to continuing investment in a company already on the verge of bankruptcy. But that is not to say that Gideon does not still have some validity in the marketplace of ideas. Rather, Gideon should be used as a bare-bones example of a rightsimproving law from which effective civil solutions can be fashioned. Reconciling Gideon with Modern Conceptions of Morality The Supreme Courts decision in Gideon derived as much from the gusto of American scruples as from constitutional requirements of the Due Process Clause. In words that would become an indispensible element of the Civil Gideon mantra, Justice Black concluded, in our adversary system of criminal justice, any person hauled into court, who is too poor to hire a lawyer, cannot be assured a fair trial unless counsel is provided for him. This seems . . . to be an obvious truth (Gideon v. Wainwright 1963:344). This statement embodies both moral and legal judgment, in that one cannot say that a person ought to have access to legal counsel based on the promise of a fair trial without simultaneously inferring that a fair trial is a morally actionable goal. Thus, the access to justice envisaged by the Due Process Clause and mandated by the Court in Gideon concerns a fundamental moral fairness, a fairness regarded by Justice Black and others as an obvious truth (Ibid.). The publicly held conception of justice and morality derive from the moral/formal tensions of liberal legalism, a tension produced by the dual commitments of law and moral politics (Alfieri 2005:1459). The efficacy of liberal legalistic ideology requires balancing the tension between law and advocacy, and is both relevant and necessary to any lawyers practice. Law
93
and advocacy are often perceived as incompatible because [l]aw heralds process values, and its practice entails formal commitments to principles of neutrality, objectivity, and reason. . . . [while] Moral politics, in contrast, honors intrinsic norms and extrinsic results, and its performance involves instrumental commitments to principles of partisanship, subjectivity, and passion (Ibid). The tension between the two is both an asset and a liability with regard to Civil Gideon, in that establishing a civil right to counsel depends upon constructing an argument that, like Gideon, appeals as much to fundamental fairness as it does to the law. A successful balanced argument depends upon achieving a delicate equilibrium between pathos and ethos, between moral fundamentalism and legal realism. An imbalance of this delicate tension significantly contributed to the disastrous decision in Lassiter, which greatly offset theadvancement of a civil right to counsel. While Lassiters legal argument may have been convincing, the relatively unsympathetic facts of the case proved detrimental. The heart of Gideon lies within the reconciliation of the moral/formal tensions of legal liberalism. Forward movement in either the judicial or legislative context can only be achieved by appealing to both elements of this tension. Similarly, lawyers advocating for Civil Gideon must balance a commitment to the legal process with realistic political commitments to democratic access and racial equality norms (Alfieri 2005:1459). Because modern conceptions of morality are significantly more inclusive than they were over 30 years ago when Gideon was decided, current conditions may be opportune for a judicial challenge, should the appropriate case surface.8 Part II: Progress and Finding Effective Alternatives Legislative Strategies, Task Forces, and Progress for Civil Gideon While those charged with a criminal offense are now provided with court-appointed counsel, the indigent defense system exists in a state of crisis (Chen 1996; Drecksel 1991; Dripps 1997; Schulhofer and Friedman 1993). That is not to say that the Civil Gideon movement has not gained considerable ground, even in the wake of decisions like Lassiter. After the 2006 ABA passage of Resolution 112A, numerous state and local bar associations passed similar resolutions and established task forces with the goal of expanding the right to counsel in certain civil matters.9 In this same spirit, the California legislature enacted the Sargent Shriver Civil Counsel
94
Act, which funds pilot programs focused on representing clients in civil matters affecting basic human needs (Cal. Govt Code 68651a 2010). Additionally, just after California passed the Sargent Shriver Act, the ABA passed the Model Access Act, designed to assist interested legislators to establish a statutory right to counsel for indigent defendants in those basic areas of human need identified in the 2006 Resolution (Am. Bar Assn, Report To The House Of Delegates, Resolution 104 2010:6). Despite the Supreme Courts reluctance to establish a civil right to counsel, efforts to recognize such a right at the state level enjoy greater acceptance (Brown 2011).10 These efforts have also begun to resolve the law-policy dichotomy through an attempt to change the policy surrounding the law solidified in Lassiter. As organizations like the ABA attempt to frame a right to civil counsel as a both an actionable goal and moral issue, the legal process will ideally evolve in tandem. Legislative strategies employed at the state level have seen greater success in a movement toward Civil Gideon, in large part due to the sheer number of state statutes providing appointed counsel in a variety of civil cases. While ambitious, such statutes are nothing novel in themselves. In fact, when the Court decided Lassiter, 33 states and the District of Columbia already had statutes in place requiring counsel in parental rights termination cases (Young 1997). Thus, with the considerable momentum that Civil Gideon has enjoyed at the state level, efforts to put in place mechanisms providing for the representation of indigent litigants must continue to be encouraged at every possible opportunity, until such provision enjoys uniformity of recognition across the country. Initially, progress at the state level may not even take the form of legislation, but instead surface as recommendations to implement legislation uniquely tailored to each states needs. For example, task forces established to determine effective ways to expand the right to counsel in specific jurisdictions offer realistic alternatives to immediate and broad legislative action. The Boston Bar Association Task Force on Expanding the Civil Right to Counsel, for instance, operates with the specific mandate to recommend that the legislature fund programs to establish a right to counsel in civil cases such as immigration, family, and housing cases (Boston Bar Assn 2008). Thus, with the opportunity to utilize such task forces, legislatures can move at whatever pace they desire, whether they prefer to unilaterally implement holistic legislation aimed at civil representation, or to move forward with
95
piecemeal legislation aimed at solving specific problems identified by task forces (Abel and Vignola 2011). Alternatives in the Interim: Legal Services Organizations as a Pathway to Effective Civil Representation While effective legislative changes are a major part of the solution to achieving equal opportunity under law, until a right is constitutionally recognized, Civil Gideon can be advanced through the utilization of alternatives to court-appointed representation. Such alternatives are currently provided by various legal services organizations dedicated to representing those who cannot afford counsel. The provision of these alternatives to court appointed counsel effectively advances resolution of the law-policy dichotomy by creating policy aimed at providing what the law, alone, cannot. As the policy surrounding civil provision of counsel changes, the law is encouraged to evolve, and the legal process becomes closer to the policy ideal. Despite being relatively disorganized in their infancy decades ago, today legal aid organizations provide meaningful legal assistance to indigent clients across the country. Civil legal assistance for individuals unable to pay for representation began in New York City in 1876 with the founding of the Legal Aid Society of New York, a private and charitable program created primarily by lawyers (Houseman 1995; MaGuire 1928; Tweed 1954). In the years following the Legal Aid Societys establishment, the world of civil legal aid evolved (Houseman 1995).11 Each diverse form of legal aid enjoyed one common and unfortunate characteristic: impossibly inadequate resources (Houseman 1995).12 Because of the endemic lack of resources, legal aid programs found it difficult to develop any sense of continuity or purpose within the general social fabric (Houseman 1995). By 1970, the basic structure of the legal services program was in place, which included federal funding. From its inauspicious beginnings, modern civil legal assistance in the United States has, over the last four decades, evolved from a relatively insignificant and disorganized program that provided limited services in only a few areas of the country, with little financial support and political recognition, into a system that provides a broad panoply of legal services to the low-income community nationwide (Houseman 2007). In furtherance of this vision, legal services organizations have won victories in several
96
individual cases that forced major governmental and economic institutions to respect the rights of poor people. Foremost among these victories was a successful challenge to a $200 million cut in Medicaid in California, in which legal services went head-to-head with then-Governor, Ronald Reagan (Morris v. Williams 1967). Such victories proved potent reminders of the importance and efficacy of legal services organizations on a national scale. The ripple effect of early legal services victories promoted the overarching goal of creating an integrated and comprehensive civil legal assistance system, while simultaneously working toward the achievement of a Civil Gideon (Houseman 1998).13 The method by which a sophisticated legal aid system ought to be implemented has been the subject of much discussion. Throughout the debate, Legal Aids goal has remained to effect change at both the micro and macro levels. With this mission in mind, and with both arms of Legal Aid working in tandem, Civil Gideon will continue to make advances in the halls of legislatures across the country, while simultaneously providing counsel to those who could not otherwise afford it. Part III: Sounding Gideons Bullhorn Fashioning a Newer, Stronger Gideon Preempting a Civil Strickland Without a doubt, myriad reforms must take place prior to Civil Gideon reaching its desired goal and before fundamental fairness is achieved for all criminal and civil defendants. Before beginning at the obvious starting point(Gideon),one must not overlook the debilitating nature of Strickland in the criminal context, and the often insurmountable barrier that it presents to any criminal defendant hoping to find some recourse against the ineffective assistance of counsel. Before seeking to establish a civil right to counsel, the Civil Gideon movement must be cognizant of avoiding a Strickland type decision in the civil context, so as to ensure that Civil Gideon does not become a hollow right (Brown 2011:916). In the event that Civil Gideon is eventually achieved at the federal level, the right to effective assistance of counsel will likely follow, as the Court has long recognized that the right to counsel is the right to the effective assistance of counsel (McMann v. Richardson 1970:771 n.14). Currently, most jurisdictions providing counsel in parental termination cases either expressly adopt the rigorous
97
Strickland standard or impliedly follow it, inevitably leading those courts to hold they decline to find ineffectiveness (Calkins 2004:214-15). Thus, to preempt the gradual erosion of a right to counsel that Strickland begat in the criminal context, the Civil Gideon movement ought to ensure, if possible, that legislation granting a statutory right to counsel also provides that such counsel must be effective (Brown 2011:919). Such a standard, which would ensure that civil defendants have a realistic standard to enforce against misbehaving counsel, should be thought of as a Civil Gideon with teeth.14 While potentially onerous, providing a framework for the formulation of an alternative Strickland standard in civil cases may not be as difficult as one might imagine. Several civil-side standards, developed by the American Bar Association, the National Legal Aid and Defender Association, and other standard-setting bodies, exist that can serve as useful guides (Abel 2006). In 2006, the American Bar Association published its Standards for the Provision of Civil Legal Aid, which contained standards for practitioners through which courts might be able to fashion an alternative rubric to that posed in Strickland. These guidelines and strategies might effectively enable the Civil Gideon movement to push toward a more just standard in judging the efficacy of court-appointed counsel in civil cases, if and when a civil right to counsel is achieved. Morality is as Morality Does: Applying Gideons Moral Framework to the New Civil Context Consistent with the trumpet call of equality sounded in Gideon, the Civil Gideon movement must seek a fusion of liberal democracy and legal equality. In doing this, the fundamental moral fairness at once espoused and enacted in Gideon will be transposed from one analogous context to the next, and absolve the moral/formal tensions of liberal legalism. A critic may note that criminal and civil situations do not seem to be analogous with regard to the gravity of their respective outcomes. While it may be true that a defendant risks losing his physical freedom in a criminal context, it is not true, nor would it be logical to assert, that a defendant in that instance has more to lose than in a civil action. In Lassiter, for example, the defendant lost her parental rights, one of the most fundamental rights that a citizen possesses, without any court-appointed representation. The legal severing of a parent-child relationship involves one of the weightiest decisions that
98
a court can make. The moral decision asserted by the court in this instance is the same from one context to the next, from criminal to civil, because in either case an incarceration is being threatened through forced isolation. While civil and criminal cases both involve society (represented by the judge) making analogous moral decisions, the right case must come along to overrule Lassiter, one with sympathetic facts and a clear legal application. When this moment comes, individuals schooled in liberal legalism, including those embracing the Civil Gideon movement, ought to jump at the opportunity to embrace both rigorous legal process and moral judgment through advocating and advancing a civil equivalent to Gideon. Dont Try Harder, Try Smarter: Igniting and Uniting Each Element of the Civil Gideon Movement With each element of the Civil Gideon movement, from legislative advocacy and judicial representation to the precedent-setting efforts of legal services organizations such as Legal Aid, the movement gains both momentum and legitimacy, moving closer toward eventually establishing a constitutional right to counsel in civil cases. The arguments for a civil right to counsel have been established through existing scholarship; all that is needed is the right case, the right combination of law and fact, to overrule Lassiter. As it was put by the Honorable Judge Robert Sweet, [as] every trial judge knows, the task of determining the correct legal outcome is rendered almost impossible without effective counsel. . . . To put it simply, denial of representation constitutes denial of access to real justice (Sweet 1998:505). The Supreme Court in Gideon recognized this, and reached its historic decision by extending the right of representation to all criminal defendants. In this same spirit, to expand the judgment made by the Supreme Court in Gideon to the civil context would be to take the next logical step toward ensuring that equal justice under law is more than empty rhetoric. To extend such a right would be more than simply fair or equitable, it would constitute the logical outworking of the fundamental moral fairness expressed in Gideon by Justice Black, a fairness fundamental enough to be recognized as an obvious truth (Gideon v. Wainwright 1963). Until this right is recognized constitutionally, the sounding of Gideons trumpet will continue in the interim through myriad strategic avenues at both state and local levels. Legislative strategies for the furtherance of Civil Gideon have gained traction and likely stand the best chance of establishing
99
something close to a uniform right to counsel in civil cases in the absence of a Supreme Court decision (Brown 2011). Legal services organizations such as Legal Aid have been vigilant in their attempts at a holistic representation of indigent clients, a representation that at once takes the cause of the dispossessed to the county courthouse and the legislative floor. Perhaps the greatest asset that legal services organizations possess in furtherance of their mission is their commitment to a holistic, dualistic approach, which provides aid to those in need until a right to that aid is established in civil cases. Significant progress has been made toward a Civil Gideon through the ABAs Model Access Act (2010) and the numerous state-level statutory provisions requiring appointed counsel in civil contexts. Organizations such as the ABA and the Boston Bar Association are uniquely poised to shape the ongoing debate bypassing resolutions and forming task forces aimed at guiding interested legislators toward mandatory civil representation. The ABAs passage of Resolution 112A in 2006, for example, provided a framework for states to fashion such aright. Still other organizations like Legal Aid are specially equipped and experienced to reinforce the rights of civil litigants by providing a holistic representation aimed at treating both the client and the underlying problem. In short, the time may never have been better to reassess, recapitulate, and reengage the Civil Gideon movement through any of its various members. Through the perpetuation of liberal legalism, the best of both morality and legality are brought together and function in harmony as legal advocates work together for the movements advancement. Though the scars of Lassiter are still keenly felt, and the overall efficacy of Gideon is routinely called into question, just as the good that Gideon did has not been outdone entirely by Strickland, so the civil right to counsel has not been precluded by Lassiter. Gideons trumpet will sound again, but this time for the rights of civil defendants, and this time with all the support and advocacy that Gideon did not enjoy in its day. When that happens, advocates will have the benefit of hindsight, and will be wary of the fact that adoption of this type of law is not a panacea for indigent civil litigants. Rather, hindsight teaches that advocates should avoid a civil Strickland, so as to ensure the fundamental efficacy of counsel provided by that right. Taken together, these ongoing efforts prove the solvency of the idea at the heart of Gideon, that fundamentally equal justice under law cannot be provided without
100
fundamentally equal access to counsel. In the end, Gideons trumpet will sound again, but this time, it will be a bullhorn. Endnotes
1
The Fourteenth Amendments Due Process Clause states: [N]or shall any State deprive any person of life, liberty, or property, without due process of law[.] U.S. CONST. amend. XIV, 1. In Gideon, the Court worried that a criminal defendant, without counsel and facing a possible deprivation of life or liberty, would not receive Constitutionally guaranteed process because: Left without the aid of counsel [the defendant] may be put on trial without a proper charge, and convicted upon incompetent evidence, or evidence irrelevant to the issue or otherwise inadmissible. ... He lacks both the skill and knowledge adequately to prepare his defense, even though he have a perfect one. He requires the guiding hand of counsel at every step in the proceedings against him. Without it, though he be not guilty, he faces the danger of conviction because he does not know how to establish his innocence. (Gideon v. Wainwright 1963:345).
Pro se is defined as [a]ppearing for oneself, as in the case of one who does not retain a lawyer and appears for himself in court. BLACKS LAW DICTIONARY 1099 (5th ed. 1979). Adverse outcomes are significantly more likely for unrepresented litigants. (Engler2010, 4666) (surveying studies finding that lack of counsel is a significant variable affecting outcomes in housing, family, small-claims, and administrative agency actions). The three factors identified in Eldridgewere the private interests at stake, the governments interest, and the risk that the procedures used will lead to erroneous decisions. Lassiter, 452 U.S. at 27. Gideons Trumpet refers to a book of the same name, which drew its title from the book of Judges, in the Bible, which reads: But the Spirit of the Lord came upon Gideon, and he blew a trumpet. BIBLE, Judges 6, 34. See also ANTHONY LEWIS, GIDEONS TRUMPET 1 (Random House, 1964). Or the least of these as it was put in the Gospel of Matthew. (BIBLE, Matthew 25). (COLE 2006:108); (Thomas III, 2010:549) (describing Strickland as The Case from Hell). Professor David Cole writes about the history behind the
101
case that would eventually establish the limits upon the Sixth Amendment right to appointed counsel in criminal cases: On a two-week crime spree in September 1976, David Leroy Washington robbed and killed a minister robbed and killed a woman in her home and then shot all three of the sisters-inlaw in the head; and kidnapped a man, stabbed him repeatedly, and eventually killed him. (COLE 2006).
7
Another reason that the Strickland standard is frequently criticized is the fact that courts do not generally attach Stricklands strong presumption in favor of reasonable performance when reviewing the performance of other professionals, such as physicians, surgeons, accountants, and architects (Klein 1986:640). By the right case, I simply mean a case that balances both legal and moral arguments, a case that simultaneously advances a convincing legal argument and a sympathetic set of facts, just as was accomplished in Gideon. Such a balance is achievable in either legislative or judicial settings, and should be pursued in both. See Paul Marvy & Laura Klein Abel, Current Developments in Advocacy to Expand the Civil Right to Counsel, 25 TOURO L. REV. 131, 132-44 (2009) (identifying bar association efforts in Alaska, California, Hawaii, Maine, Massachusetts, Minnesota, New Hampshire, New York, Ohio, Pennsylvania, and Washington). In fact, commentators have found that legislative efforts may be far better poised to achieve success for the Civil Gideon movement than judicial efforts. Judicial efforts toward a Civil Gideon are fought primarily on three different fronts: due process claims, federal equal protection claims, and state constitutional claims. Those claims have found most of their success in cases involving the termination of parental rights (Schwinn 2006:606).
10
Many legal aid programs were private corporations; others were part of local bar associations, relying primarily on the donated time of lawyers. Some were part of governmental units (usually municipalities) or other social agencies. Still others were run by law schools or clinics. 12 Court appearances were rare. Appeals were nonexistent. Administrative representation, lobbying, and community legal education were not even contemplated. Legal aid had little effect on the individuals it served and no effect on the client population as a whole.
11 13
Such a system should have the capacity to: (1) educate and inform low-income persons of their legal rights and responsibilities and the options and services
102
available to solve their legal problems; and (2) ensure that all low-income personshave meaningful access to a full range of high-quality legal assistance providers when they have chosen options that require legal aid and assistance.
14
This term is the authors own.
References Cited Abel, Laura K. A Right to Counsel in Civil Cases: Lessons From Gideon v. Wainwright, 15 Temp. Pol. & Civ. Rts. L. Rev. 527, 549 (2006). Abel Laura K. and Susan Vignola, Economic and Other Benefits Associated with the Provision of Civil Legal Aid, 9 Seattle J. for Soc. Just. 139 (2011). Alfieri, Anthony V. Gideon in White/Gideon in Black: Race and Identity in Lawyering, 114 Yale L.J. 1459 (2005). American Bar Association, Report to the House of Delegates, Resolution 104 (2010), available at http://www.abanet.org/legalservices/sclaid/ downloads/104_Revised_FINAL_Aug_2010.pdf ---. Report to the House of Delegates, Resolution 112A (2006). Bible, Matthew 25. Blacks Law Dictionary 1099 (5th ed. 1979). Boston Bar Association Task Force on Expanding the Civil Right to Counsel, Gideons New Trumpet: Expanding the Civil Right to Counsel in Massachusetts 2 (2008). Brown, Mark. Establishing Rights Without Remedies? Achieving an Effective Civil Gideon by Avoiding a Civil Strickland, 159 U. Pa. L. Rev. 893, 900 (2011). California Government Code 68651(a) (West Supp. 2010). Calkins, Susan. Ineffective Assistance of Counsel in Parental-Rights Termination Cases: The Challenge for Appellate Courts, 6 J. App. Prac. & Process 179 (2004). Chen,Harold H. Malpractice Immunity: An Illegitimate and Ineffective Response to the Indigent-Defense Crisis, 45 Duke L.J. 783 (1996).
103
Cole, David. Gideon v. Wainwright and Strickland v. Washington: Broken Promises in Criminal Procedure Stories, 101, 104 (Carol S. Steiker ed., 2006). Domitrovic, Brian. The Worst Economic Crisis Since When?, Forbes, Feb. 5, 2013, available at http://www.forbes.com/sites/ briandomitrovic/2013/02/05/the-worst-economic-crisis-since-when/ Drecksel, Paul Calvin. The Crisis in Indigent Criminal Defense, 44 Ark. L. Rev. 363 (1991). Dripps, Donald A.Ineffective Assistance of Counsel: The Case for an Ex Ante Parity Standard, 88 J. Crim. L. & Criminology 242 (1997). Engler, Russell. Connecting Self-Representation to Civil Gideon: What Existing Data Reveal About When Counsel Is Most Needed, 37 Fordham Urb. L.J. 37 (2010). Gagnon v. Scarpelli, 411 U.S. 778, 790 (1973). Gideon v. Wainwright, 372 U.S. 335 (1963). Houseman, Alan W. and Linda E. Perle, A Brief History of Civil Legal Assistance in the United States, Center for Law and Social Policy 49 (2007) available at http://www.clasp.org/admin/site/publications/ files/0158.pdf Houseman, Alan W. Civil Legal Assistance for the Twenty-First Century: Achieving Equal Justice for All, 17 Yale L. &Poly Rev. 369 (1998). ---. Political Lessons: Legal Services for the Poor A Commentary, 83 Geo. L. J. 1669 (1995). Klein, Richard. The Emperor Gideon Has No Clothes: The Empty Promise of the Constitutional Right to Effective Assistance of Counsel, 13 Hastings Const. L.Q. 625 (1986). Klein, Richard and Robert Spangenberg, The Indigent Defense Crisis 25 (1993). Lassiter v. Department of Social Services., 452 U.S. 18 (1981). Lee, Kyung. Reinventing Gideon v. Wainwright: Holistic Defenders, Indigent Defendants, and the Right to Counsel, 31 Am. J. Crim. L. 367 (2004).
104
Lewis, Anthony, Gideons Trumpet (Random House 1964). MaGuire, John M. The Lance of Justice 3 (1928). Marvy, Paul and Laura Klein Abel, Current Developments in Advocacy to Expand the Civil Right to Counsel, 25 Touro L. Rev. 131 (2009). Matthews v. Eldridge, 424 U.S. 319 (1976). McMann v. Richardson, 397 U.S. 759 (1970). Morris v. Williams, 433 P.2d 697 (Cal. 1967). Powell v. Alabama, 287 U.S. 45 (1932). Rhode, Deborah. Access to Justice: Connecting Principles to Practice, 17 Geo. J. Legal Ethics 369 (2004). Scott v. Illinois, 440 U.S. 367 (1979). Standing Comm. on Legal Aid & Indigent Defendants, Am. Bar Assn, Standards for the Provision of Civil Legal Aid (2006), available at http://www.abanet.org/legalservices/sclaid/downloads/ civillegalaidstds2007.pdf Schulhofer, and Stephen J. and David D. Friedman, Rethinking Indigent Defense: Promoting Effective Representation Through Consumer Sovereignty and Freedom of Choice for All Criminal Defendants, 31 Am. Crim. L. Rev. 73 (1993). Schwinn, Steven D. The Right to Counsel on Appeal: Civil Douglas, 15 Temp. Pol. & Civ. Rts. L. Rev. 603 (2006). Steinberg, Robin and David Feige, Cultural Revolution: Transforming the Public Defenders Office (Aug. 2002), available at http://www.ksg. harvard.edu/criminaljustice/publications/cultural_rev.pdf Strickland v. Washington, 466 U.S. 668 (1984). Sweet, Robert W. Civil Gideon and Confidence in A Just Society, 17 Yale L. & Poly Rev. 503 (1998). Thomas III, George C. Historys Lesson for the Right to Counsel, 2004 U. Ill. L. Rev. 43 (2004). Tweed, Harrison The Legal Aid Society New York City 1876-1951 (1954).
105
Young, Rosalie R. The Right to Appointed Counsel in Termination of Parental Rights Proceedings: The States Response to Lassiter, 14 Touro L. Rev. 247 (1997).
106
The Lens of Law Enforcement: A Geospatial Statistical Program Evaluation of Denvers HALO Camera Surveillance System
John Papazian
John Papazian is a denizen of Denver and a proud alumnus of the MPP program at Duke University. He is currently studying statistics at the Institute for Advanced Analytics in Raleigh. He is grateful for the assistance that he received from Professor Philip Cook, Professor Elizabeth Frankenberg, and the staff of the Sanford Journal while conducting this research.
Abstract The Denver Police Department has recently implemented a new hightech surveillance program to prevent crime throughout the city. The High Activity Location Observation (HALO) cameras can transmit video to police headquarters in real time through an Internet-based wireless network. The department has installed more than 100 HALO cameras at various high crime areas in Denver as of 2012. This investigation attempts a program evaluation of the surveillance system through a geospatial statistical analysis of property crime. Although cameras have been installed across the city, this investigation focuses on cameras installed in Police District #6, which encompasses the central business district. This investigation establishes a statistically significant relationship between the installation of the HALO cameras and a reduction of thefts from motor vehicles in the viewshed of the cameras in downtown Denver. The difference-indifference econometric approach suggests that the relationship is causal. Other categories of crime also may have been reduced due to the HALO cameras, but the statistical evidence is not strong enough to make a causal claim. An alternative approach based upon kernel density estimation hot spot maps is also explored. Policy recommendations are developed based upon the empirical results of the program evaluation.
109
Papazian: The Lens of Law Enforcement
Introduction In 2009, the Denver Police Department unveiled a new video surveillance program, the High Activity Location Observation (HALO) system, to help prevent crime. These HALO cameras were an improvement over traditional closed-circuit television cameras because they incorporate night vision with zoom and pivot capabilities that can read a license plate from a block away and transmit video through a wireless network (Osher 2010). However, the cameras cost $20,000 per unit. As of 2012, the Police Departmenthad installed more than 100 HALO cameras at various high crime areas in Denver. The American Civil Liberties Union has criticized these types of cameras as ineffective in reducing crime (Biale 2008). However, no formal statistical analysis of the Denver HALO cameras has yet been carried out by the police department, the ACLU, or external researchers (Maher 2009). This paper evaluates the surveillance system in downtown Denver through a geospatial statistical analysis of property crime. The camera system was also designed to help prevent violent crime such as robberies. However, the data available on these types of crime is insufficient to draw conclusions, and therefore the focus is exclusively on property crime. The geographic focus is on the central area of the city encompassed by Denver Police District #6, where most of the HALO cameras were installed. I employ a difference-in-differences econometric method to analyze the data derived from a quasi-experiment. With this approach, I compare property crime incidents in treated sites (areas within the HALO camera viewsheds) to control sites (areas with similar characteristics but without HALO cameras) both before and after the intervention. This investigation establishes a statistically significant relationship between the installation of the HALO cameras and a reduction of thefts from motor vehicles in the viewshed of the police cameras in downtown Denver. Background The city of Denver began experimenting with cameras in 2006 when the former Chief of Police, Gerald Whitman, assigned Lieutenant Ernie Martinez to launcha pilot video surveillance project, which later grew into the HALO crime prevention program (Maher 2009). The police department wanted to keep up with technological advances pursued by larger cities
110
such as Chicago. The departments Operations Manual states that the top four principle objectives of the HALO program were (1) enhancing public safety, (2) preventing and deterring crime, (3) reducing the fear of crime, and (4) identifying criminal activity (Denver Police Department 2011). The program had immediate success; the first installed camera helped the police to capture gang members committing arson at the Holly Square Shopping Center, a strip mall in northeast Denver. Following this initial success, the Denver Police expanded the program. In 2008, the federal government granted the Denver Police $1 million to install 50 additional cameras to help with security while the city hosted the Democratic National Convention. Those cameras remained in place after the convention concluded (Maher 2009). By January 2010, Lieutenant Martinez had deployed the 81 wireless cameras to high crime areas as part of Phase II of the HALO program. In November 2010, the Colfax Business Improvement District helped finance the purchase of additional cameras added as part of Phase III to help prevent crime on Colfax Avenue, a thoroughfare notorious for prostitution and organized crime (Martinez interview August 8, 2011). All HALO cameras are currently monitored and controlled from police headquarters by uniformed officers. As of January 2012, the Denver Police Department has released the exact locations of all cameras installed (HALO Street 2012). Welsh and Farringtons (2009) meta-analysis suggests that cameras installed in city centers led to small, non-significant reductions in crime. They compute an odds ratio as a comparable metric of relative effect size across all studies included in their meta-analysis. However, their results may not apply to the more powerful wireless Avrio cameras because their study only analyzes the effects of traditional close circuit television (CCTV) cameras. Additionally, their results may not apply to locations in the United States. Most of their studies use data from metropolitan areas in the United Kingdom, which were early adopters of surveillance technology. Finally, many of the studies included in their meta-analysis lack a rigorous quasiexperimental program evaluation and only analyze crime data before and after the camera intervention without developing any type of control. One recent study suggests that cameras may have helped prevent crime in certain areas of Los Angeles. Cook and MacDonald (2010) analyze crime data in Los Angeles before and after the formation of a number of
111
Business Improvement Districts (BIDs). These BIDs raise money from local businesses to pay local services such as private security guards, trash collection, commerce promotion, and CCTV camera installation. The Hollywood Entertainment BID has installed eight CCTV cameras at intersections in the district. Cook and MacDonald aggregate neighborhood time series crime data, and then assignthe values to the corresponding business improvement district. Next, they use the panel data to examine the effects of BIDs through a difference-in-difference econometric model with fixed effects for neighborhood and year. Their results showa statistically significant effect of BIDs on crimes and arrests per year. Applying Cook and MacDonalds approach, this investigation evaluates the original 44 HALO cameras installed in Denver Police District #6 to determine whetherthe surveillance system helped to prevent property crime in downtown Denver. Data Preparation The city of Denver provides geospatial files on their website for depicting neighborhoods, police districts, streets, zoning restrictions, and census information. Many of the maps produced in this report are derived from those geospatial files (Denver Maps 2012). The Denver Police Department provides data on crime incidents from 2006 to the present on their websiteas part ofthe National Incident Based Reporting System (NIBRS). NIBRS lists each reported crime as a separate incident, recording its type, location, date, and time when the incident was first reported by civilians or by officers (Crime Data 2012). For this investigation of the HALO cameras, I analyze Federal Bureau of Investigation Uniform Crime Reporting (UCR) Part 1 property crimes in central Denver. UCR Part 1 crimes are serious crimes that occur regularly throughout the country and include burglary, larceny, theft from motor vehicle, auto theft, and arson. These types of crimes are likely to be reported to authorities. Chris Wyckoff, the Director of the Data Analysis Unit for the Denver Police, reports that the online database provides useful information for analyzing UCR Part 1 property crime rates in different areas of Denver (Wyckoff interview August 16, 2011). Crime incident data downloaded from the Denver Police Department must be geocoded before it can be imported into a geographic information
112
system. Geocoding involves assigning a specific longitude and latitude to the address location of a criminal incident. The web application BatchGeo was used to perform this task. As noted, the Denver Police Departments website provides data on crime incidents since 2006, but the address data are not properly cleaned for geocoding (Crime Data 2012). Therefore, I cleaned the raw data before geocoding. Cleaning procedures included spelling out common abbreviations, standardizing the entry format, and removing miscellaneous information. In order to determine consistency across the 250,000 criminal incidents in the database, I programmed Visual Basic scripts to parse the address information. Once I assigned a specific longitude and latitude to each criminal incident, I imported the crime incident datainto the geographic information system ArcGIS and wrote scripts in Python to perform the actual spatial analysis. Figure 1 displays a choropleth map of property crimes across all neighborhoods throughout the entire cityof Denver. Figure 2 displays the original 44 HALO cameras in central Denver with a push pin dot for every incident of property crime in the time frame of 2006-2007. Although cameras were installed across the city, this investigation focuses only on cameras installed in Police District #6, which encompasses the central downtown area of the city. Methodology The differences-in-differences econometric approach compares the treatment sites against control sites before and after the 2008 camera expansion. Under a randomized controlled policy experiment, sites would be randomly assigned to receive treatment or not. The HALO intervention did not use such a method. The camera sites were chosen primarily because they were at intersections near high crime areas of the city. Moreover, it is likely that some businesses lobbied to have cameras installed near their stores. In this regard, the Denver Police surveillance program is a quasi-experiment. Since control sites were not determined prior to the intervention, the control sites must be determined in the present and then applied retroactively. I identified control sites through a Monte Carlo process, which selects numbers at random similar to a roulette wheel at a casino. In the
113
Figure 1: Cloropleth Map of Property Crimes across the Neighborhoods of Denver, Colorado, 2006-2007
first stage, the treatment sites are examined to determine what geographic characteristics they have in common. Street corners with characteristics similar to street corners containing HALO cameras are labeled as candidate control sites. These characteristics include similar zoning restrictions, low housing utilization, and proximity to alcohol sales. In the second stage, some of the candidate control sites are selected as control sites. I assigned
114
Figure 2: Map of the Original 44 HALO Cameras (Treatment Sites) and Locations of Property Crime Incidents, 2006-2007
every candidate site a random number between zero and one and then selected the 44 sites with the highest numbers to serve as control sites. A stochastic process removes human bias and ensures that control sites are scattered at random, similar to darts thrown on a board. Figure 3 illustrates the candidate control sites that have similar geographic characteristics as the original 44 HALO cameras sites in Police District #6. Figure 4 plots the
115
44 control sites selected at random to match the original 44 HALO cameras sites in order to balance the treatment and the control. Control sites were selected based on three characteristics: (1) zoning restrictions, (2) low housing utilization, and (3) proximity to alcohol sales. The zoning restrictions in central Denver are intricate with many different ordinances regulating the development of land. According to the citys geographic records, nearly all of the original 44 HALO camera sites are located in the Downtown Zone. This is the area of the city where skyscrapers are allowed to be built. Therefore, candidate control sites are selected only from this zone. Furthermore, the original 44 HALO camera sites are all located on city blocks that are relatively unpopulated. Most of the buildings in this area are commercial rather than residential, and they have low housing utilization. Therefore, the candidate control sites are restricted to city blocks that house less than 400 people. Finally, the original 44 HALO camera sites are located near bars, nightclubs, and stores selling alcohol. Past scholarship has demonstrated a linkage between the accessibility of alcohol and crime (Block and Block 1995). Therefore, the candidate control sites are restricted to locations that are within a half mile of a business with a liquor license in order to be similar to the HALO cameras sites. Researchers frequently model crime through a Poisson distribution. Discrete count data is bounded below by zero but not bounded above by any integer. Count data does not follow a Normal distribution because of the lower bound. The Denver crime data displays signs of over dispersion because the variance is much larger than the mean. In fact, the mean is 14.7 criminal incidents per observation, while the variance is above 232. In other words, there are some street corners with very high levels of crime, while there are other street corners with very low levels of crime. Therefore, I model crime count as a negative binomial process rather than a Poisson process. Ordinary Least Squares requires the residuals to be normally distributed, which is not true in this case. Thus, I develop a generalized linear model to test the impact of the HALO intervention using Maximum Likelihood Estimation with Stata statistical software. As noted, I use a differences-in-differences regression model to measure the effect of the treatment. In this model, the unit of the observation, Y, is the count of crime in each individual viewshed (whether real or hypothetical) for a period of two years. A viewshed is defined as a circular area around a site with a radius of 50 yards as seen in Figure 6. The HALO cameras are
116
Figure 3: Map of the Candidate Control Sites to Match Original 44 HALO Cameras
visible from roughly half a block away, and therefore their maximum crime deterrence is set to a distance of 50 yards. For the treatment group, the count of crime in the real viewshed after a camera was installed is compared to the count of crime in the same viewshed before any camera was installed. For the control group, the count of crime in a hypothetical viewshed in the postperiod is compared to the count of crime in the same hypothetical viewshed in the pre-period.
117
Figure 4: Map of the 44 Control Sites to Match Original 44 HALO Camera Sites
118
To estimate a causal impact of HALO cameras on property crime in downtown Denver, I use the following equations: log(count of crime) = (logYit) = 0 + 1*T1 + 2*At + 3*(Tt*At) count of crime = Yit = exp(0 + 1*T1 + 2*At + 3*(Tt*At))
(1) (2)
(3) count of crime = Yit = exp(0)*exp(1*T1)*exp(2*At)*exp(3*(Tt*At))....
There are three key variables in this approach. The treatment dummy variable Ait = 1 for sites that actually received a HALO camera and Ait = 0 for the control sites. The time period dummy variable Tit = 1 for observations occurring after the cameras were installed and Tit = 0 in the period before installation. The interaction term of Tit*Ait = 1 for observations corresponding to treatment areas after the intervention. In a negative binomial regression, the log of the outcome variable (the count of crime) is modeled as a linear combination of the predictor variables. The incident rates ratio for a predictor variable can be calculated by taking the exponent of its coefficient. In this sense, the incident rates ratio has a multiplicative effect in the y-scale of crime count (UCLA 2012). But, when a dummy variable is zero, the exponentiation of the coefficient equals one, and the predictor variable has no multiplicative effect.
Figure 5 - Histogram of Crime Incidents across all Observations
119
To review, the unit of observation is the count of crime in each individual viewshed (whether real or hypothetical) for a period of two years. There are 44 sites located in Denver Police District #6 that were eventually treated by installing HALO cameras in 2008. These sites are listed inthe Appendix. Therefore, a count of crime was computed in each real viewshed after installation and in each hypothetical viewshed before installation for a total of 88 observations in the treatment group. There are 44 sites located in Denver Police District #6 that have been selected to serve as the control. These sites were never treated with a HALO camera but were selected through a Monte Carlo process. A count of crime is computed in each hypothetical viewshed in the post-period and in each hypothetical viewshed in the pre-period for a total of 88 observations in the comparison group. Thus, a sum total of 176 observations are used in this analysis.
Table 1: Derivation of the Difference-in-Difference Estimate (3)
Dummy Variable A Area 1: Area 0: HALO Viewshed non-HALO Sites (Treatment) (Control)
Dummy Variable T
Time Period 0: Before Intervention Jan. 2006 to Dec. 2007 Time Period 1: After Intervention Jan. 2010 to Dec. 2011
Coefficient 0 1 2 3
Calculation a c-a b-a (d - b) - (c - a)
Results To determine if the HALO intervention had an effect on reducing crime, I examined the coefficients for the Difference-in-Difference estimates (3) for each specific category of crime. I developed a separate econometric model for each specific category of property crime: auto theft, burglary, larceny, and theft from motor vehicles. All of the coefficients for the Difference-in-Difference estimates (3) are negative. The coefficient for theft from motor vehicles is negative and statistically significant even with robust standard errors. The percent change in the incident rate of theft from motor vehicles is roughly a 50% decrease in the treated area between the two time periods relative to the change in the control area.
120
Figure 6: Map of Crime Incidents Overlaid on Top of the Viewsheds of a Subset of the HALO Cameras and Control Sites
However, none of the other coefficients is statistically significant, which could be due to sample size. The statistically significant coefficient for theft from motor vehicles gives causal evidence that the HALO cameras did reduce crime for that category in downtown Denver. Figure 7 displays time series data of all categories of property crime in the downtown area,
121
while Figure 8 displays time series data for just thefts from motor vehicle in the downtown area. Figure 8 shows a clear divergence of the treatment versus the control in the period after the HALO intervention.
Table 2: The Difference-in-Difference Estimates (3) for Categories of Property Crime
Categories of Property Crime Auto Theft Burglary Larceny Theft From Motor Vehicle Coefficient -0.272 (0.350) -0.245 (0.658) -0.242 (0.390) -0.689 (0.331) P>|z| 0.438 0.710 0.536 0.038 95% Conf. Interval -0.958 -1.536 -1.006 -1.339 0.415 1.045 0.523 -0.040
Note: Robust Standard Errors are listed in parentheses
Table 3: Tabulation of Property Crime Counts for All Sites Grouped Together
Categories of Property Crime Auto Theft Burglary Larceny Theft From Motor Vehicle
Treatment Before 87 21 264 174
Treatment After| 43 14 353 106
% Change -50.6% -33.3% +33.7% -39.1%
Control Before 74 27 158 178
Control After 48 23 269 216
% Change -35.1% -14.8% +70.3% +21.3%
in % Change -15.5% -18.5% -36.6% -60.4%
Arson is also a category of property crime. However, arson is excluded because there are very few incidents of arson in downtown Denver.
122
Figure 7: All Property Crime in Viewsheds of HALO Cameras vs. Control Sites in Downtown Denver over Time
Offense_Ca Count of OBJECTID
Theft from Motor Vehicles Crime in Viewsheds of HALO Cameras vs. Control Sites Figure 8: Theft from Motor Vehicles Crime in Viewsheds of HALO Cameras vs. Control Sites in Downtown Denver over Time
140
Count of Crime Incidents per Year
120
100 80 60 40 20 0 Control Treatment
2006
2007
2008
2009
2010
2011
123
Alternative Approach Diffusion of the treatment effect to control intersections would undermine the above economic approach. Diffusion likely did not occur here because each control intersection is at least one block removed (~100 yards) from its nearest treatment intersection, while HALO cameras are only visible from a distance of about 50 yards. However, criminals might be deterred from committing transgressions in central Denver irrespective if a camera is watching them. It is possible that the high concentration of cameras in the city center creates a halo effect, where the treatment diffuses throughout the entire downtown area rather than only impacts certain intersections. In the case of a virtuous diffusion, the result would be to reduce the estimated effect sizes because the treatment would spill over into the control. Since the econometric approach can only detect a localized effect of the treatment, crime density hot spot maps are created to complement the regressions and analyze the effect of the intervention on a larger scale. This approach draws insight from the environmental criminology research of Brantingham and Brantingham (1981) in their use of ecological principles to model crime patterns. One important element in this approach is to collect data at the lowest geographic units of analysis. According to Weisburd et al. (2009), crime maps based upon high units of analysis could be misleading because of an ecological fallacy. For example, the choropleth map [Figure 1] of property crime across neighborhoods in Denver has a deceptive quality. There are some neighborhoods colored light gray because of relatively low level of property crime, which are adjacent to neighborhoods colored dark gray because of relatively high level of property crime. It appears that if one were to walk across the border that one would encounter more crime. However, there are likely some places in dangerous neighborhoods that are quite safe, while there are places in safe neighborhoods that are quite dangerous. Aggregating crime statistics to the neighborhood level distorts the truthbe cause there are places inside each neighborhood that are not representative of the whole. Unfortunately, choropleth maps at all geographic levels (counties, neighborhoods, census blocks) are susceptible to the ecological fallacy because administrative boundaries are not based upon crime patterns. The
124
Figure 9: Map of Kernel Density Estimate Hot Spots of Property Crime in Denver Before the Instalation of HALO Cameras (2006-2007)
125
Figure 10: Map of Kernel Density Estimate for Property Crimes in Central Denver Between the Two Time Periods
126
solution is to disaggregate to the lowest unit possible, which is the geo coded incident of a criminal activity. However, a map [Figure 2] displaying a plethora of push pin crime dots can make it difficult to identify highcrime areas because multiple crimes at one location will only show up as a single dot. One solution is to create crime density hot spot maps. Crime density maps can be used as an alternative evaluation technique to determine if surveillance cameras have dampened the level of crime. Waples, Gill, and Fisher (2009) use the concept of a kernel density estimate in their criminology research on surveillance cameras. Crime in Denver is concentrated at specifichotspots because a lot of crime occurs repeatedly at the same locations over time. Kernel density estimation (KDE) is a nonparametric procedure that can be used by ArcGIS to estimate the probability density function of a random variable (e.g. crime) over a geographic region (ESRI 2010). Figure 9 illustrates a kernel density estimate of property crime in downtown Denver before the installation of the HALO cameras (20062007). Waples et al (2009) use kernel density estimation to create crime maps before and after the intervention of police cameras to detect if the hotspots have moved once the cameras are installed. Then, they create a change detection map to capture the kernel density differences between the prior and the posterior maps. I replicated their technique to create maps [Figure 10] of the property crime in Denver before and after the installation of the HALO police cameras. Similar to the regression approach, the time period before the intervention (2006-2007) and the time period after the intervention (2010-2011) both constitute two years and contain the same calendar months. The colored areas on the change detection map [Figure 10] illustrate locations where there is a statistically significant change in the kernel density estimate of crime between the two periods. The map depicts a statistically significant decrease in the kernel density estimate of property crime in the northwestern quadrant of downtown Denver. This is near the location where five HALO cameras were installed prior to 2010. This section of the city, known as the LoDo district (abbreviation for Lower Downtown) is a popular nightlife area between two sports stadiums that historically has been plagued by a high level of crime. However, it would be imprudent to assign the drop in crime in the LoDo district solely to the HALO camera intervention. It is unwise to infer a causal relationship from the KDE map because there could be other
127
explanations for the drop in property crime. For example, the LoDo district has recently undergone significant gentrification. Unlike the differencein-difference econometric approach, the kernel density estimation change detection does not employ a rigorous control. Nonetheless, the map does delineate an interesting phenomenon that should be studied in more detail. Conclusion In retrospect, this investigation has established a statistically significant relationship between the installation of the HALO cameras and a reduction of thefts from motor vehicles in the viewshed of the cameras in Denver Police District #6. The difference-in-difference econometric approach suggests that the relationship is causal. HALO cameras may have reduced other categories of property crime, but the statistical evidence is not strong enough to make a causal claim. There are three potential weaknesses in my methodology. First, important variables were likely omitted when choosing control sites. Control sites were selected based on geographic characteristics they had in common with treatment sites: zoning restrictions, low housing utilization, and proximity to alcohol sales. These geographic characteristics are somewhat arbitrary, and they are based upon my personal understanding of downtown Denver. There are other characteristics that may have been more appropriate. For example, there has been considerable construction in downtown Denver, which alters travel patterns around the city. Ultimately, there is no way to select control sites that perfectly match the treatment sites. Although the control sites are not ideal, I believe they are sufficiently similar to the treatment sites to test the counter factual in the context of a quasi-experiment. Second, my methodology assumes parallel trends. The treatment and control sites do not have to be identical for a differences-in-differences approach. But, in the absence of cameras, crime should otherwise increase or decrease at the same rate in the treatment sites as in the control sites. Unfortunately, I cannot prove that this parallel trends assumption holds true. HALO cameras are clustered around the 16th street mall, the Denver Convention Center, and the Civic Center Park. Those three locations have no perfect substitutes in the fabric of downtown Denver. However, the control sites are randomly scattered around the treatment sites as seen in Figure 4. The control sites are not placed in a suburban strip mall far away
128
from downtown. Rather, the control sites are placed as near as possible to the treatment sites to help ensure the assumption of parallel trends. Third, there may have been errors in geocoding. Since there were over 250,000 criminal incidents in the database, I could not check every address individually to ensure that it was correctly plotted on the map. Instead, I devoted considerable resources to data cleaning and then used external software to assign a specific longitude and latitude to each address location of a criminal incident. There are possible errors in my data, where criminal incidents are plotted at wrong locations on the map. Any large scale error in geocoding would cascade into the econometric analysis and corrupt the statistical results. Unfortunately, there is no metric to gauge the level of accuracy in the geocoding. However, I used the web application BatchGeo to complete this process. BatchGeo is a well-respected geocoding tool that relies on geo spatial data from Google to plot addresses correctly onto a map. While it is possible that a minimal number of criminal incidents are plotted incorrectly, it is unlikely that the errors are large scale enough to dramatically alter my findings. Bearing in mind these potential weaknesses in my methodology, I recommend three strategies based upon the empirical results: 1. An expansion of new HALO cameras into other areas of Denver experiencing high levels of theft from motor vehicles. Ideally, the number of police cameras should be increased until the marginal benefit of crime prevention equals the marginal cost of camera installation. Future research could be conducted to estimate the full economic benefits and costs of the HALO cameras. 2. An upgrade of the information system to cross-reference the NIBRS crime incident data to actual arrests and convictions. Publicly available crime data on the Denver Police website does not link arrests and convictions to usage of the HALO cameras. It is important to learn if the number of arrests has increased in the viewshed of the cameras in order to learn their true effectiveness. 3. An implementation of a randomized controlled experiment in the next phase of the HALO program. Determining the control sites before the intervention is superior than determining the control sites after the intervention. This would enable researchers to be more confident in ascribing causality.
129
References Cited Biale, Noam. (2008, June 25). What Criminologists and Others Studying Cameras Have Found. American Civil Liberties Union. Retrieved from http://www.aclu.org/technology-and-liberty/expert-findingssurveillance-cameras. Block, Richard and Carolyn Rebecca Block. (1995). Space, place and crime: hot spot areas and hot places of liquor-related crime. In John E. Eck and David Weisburd (Eds.), Crime and Place: Crime Prevention Studies, Volume 4, (pg. 145-183). Monsey, NY: Criminal Justice Press. Brantingham, Patricia and Paul Brantingham. (1981). Environmental Criminology. Beverly Hills, CA: Sage Publications. Cook, Philip, and John MacDonald. (2010, April). Public Safety through Private Action: An Economics Assessment of BIDs, Locks, and Citizen Cooperation. National Bureau of Economic Research. Retrieved from http://www.nber.org/papers/w15877. Crime Data. (2012). Denver Police Department Data Analysis Unit. Retrieved from http://data.denvergov.org/dataset/city-and-county-ofdenver-crime. Denver Maps. (2012). City and County of Denver. Retrieved from http:// www.denvergov.org/maps. Environmental Systems Research Institute. (2011). ArcGIS Desktop Help: How Kernel Density works.Retrieved from http://webhelp.esri.com/ arcgiSDEsktop/9.3/index.cfm?TopicName=How%20Kernel%20 Density%20works. HALO Policy in Police Operations Manual. (2011). Denver Police Department. Retrieved from http://www.denvergov.org/Portals/720/ documents/OperationsManual/119.pdf. HALO Street Cameras. (2012). Denver Police Department. Retrieved from http://www.denvergov.org/police/PoliceDepartment/ SafetyPrevention/StreetCameras/tabid/442831/Default.aspx.
130
Maher, Jared Jacang. (2009, June 18). Smile! You could be on the Denver Police Departments candid camera. Westword. Retrieved from http://www.westword.com/2009-06-18/news/smile-you-could-beon-the-denver-police-s-candid-camera. Martinez, Lieutenant Ernie. (2011, August 8). Personal Interview. Negative Binomial Regression. (2012). University of California at Los Angeles Statistical Consulting Group. Retrieved from http://statistics. ats.ucla.edu/stat/stata/dae/nbreg.htm. Osher, Christopher. (2010, June 6). Denvers surveillance system draws praise, concerns. Denver Post. Retrieved from http://www. denverpost.com/ci_15236766. Waples, Sam,Martin Gill, and Peter Fisher. (2009, May). Does CCTV displace crime? Criminology and Criminal Justice, 9(2), 207-224. Weisburd, David, Wim Bernasco, and Gerben Bruinsma. (2009). Putting Crime in its Place.New York, NY: Springer Publications. Welsh, Brandon, and David Farrington. (2009). Public Area CCTV and Crime Prevention: An Updated Systematic Review and MetaAnalysis. Justice Quarterly, 26(4), 716-745. Wyckoff, Chris. (2011, August 16). Personal Interview.
131
Appendix A: List of the Original 44 HALO Cameras

Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 132 Location 120 W 14th Ave 12th & Chopper Circle 12th & Welton 1351 Cherokee 13th & Champa 13th & Cherokee 13th & Delaware 13th & Welton 1450 Bannock 1450 Broadway 14th & Bannock 14th & Broadway 14th & California 14th & Champa 14th & Cherokee 14th & Curtis 14th & Delaware 14th & Larimer 14th & Stout 14th & Welton 15th & California 15th & Curtis 15th & Market 15th & Wazee 15th & Welton 16th & California 16th & Cleveland 16th & Curtis 16th & Lawrence 16th & Market 16th & Wazee 16th & Welton 170 W 14th Ave 17th & Arapahoe 18th & Arapahoe 19th & Market 20th & Blake 27th & Welton 9th & Chopper Circle Colfax & Broadway 15th Street & Cleveland 60 W 14th Ave 15th Street and Colfax Avenue Park Avenue West & Lawrence Longitude -104.9888 -105.0058 -104.9961 -104.9917 -104.9976 -104.9916 -104.9929 -104.9949 -104.9906 -104.9874 -104.9907 -104.9874 -104.9946 -104.9963 -104.9916 -104.9972 -104.9929 -104.9998 -104.9955 -104.9937 -104.9934 -104.9960 -104.9995 -105.0013 -104.9925 -104.9922 -104.9878 -104.9948 -104.9965 -104.9983 -105.0000 -104.9913 -104.9898 -104.9944 -104.9933 -104.9947 -104.9943 -104.9780 -105.0069 104.9874 -104.9888 -104.9886 -104.9890 -104.9881 Latitude 39.7375 39.7495 39.7408 39.7377 39.7437 39.7369 39.7401 39.7417 39.7388 39.7393 39.7397 39.7383 39.7433 39.7447 39.7385 39.7453 39.7385 39.7473 39.7440 39.7427 39.7443 39.7463 39.7489 39.7503 39.7436 39.7452 39.7418 39.7472 39.7485 39.7499 39.7512 39.7445 39.7381 39.7488 39.7497 39.7527 39.7543 39.7548 39.7479 39.7401 39.7410 39.7385 39.7401 39.7551
133

Sanford Journal of Public Policy - Volume 4 No. 1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sanford Journal of Public Policy - Volume 4 No. 1

Uploaded by

Copyright:

Available Formats

Vol. 4 | Spring 2013