The child support incentive program: constraints and the capacity for organizational cheating by street-level bureaucrats

Many have argued that government "should be run like a business." This argument has evolved throughout the years, but implies that implementing performance measures, benchmarks and incentives will provide inducements for improving organizational performance. While the literature is replete with such arguments, there has been little attention paid to the "wrong-side" of performance management and whether incentive structures result in instances where individuals and organizations engage in efforts to cheat, or "game the system." The nation's child support program is a good example of a government program that has adopted an incentive-based approach. Four state child support agencies, representing varied degrees of performance, allowed their staff to participate in a survey that explored issues of cheating and knowledge of the child support incentive program. This study constructed cheating as the capability of artificially inflating performance levels. The underlying research hypothesis, based on principal-agency theory, suggested that higher levels of cheating would be found among higher incentive earning states. The findings of the study suggested otherwise. The study found that while front-line child support professionals identified performance as being important, few were aware that a child support incentive system existed. This finding is important as it suggests that the incentive structure does little to influence front-line worker behavior and does not appear to have much impact on workers' performance levels. Furthermore, there appeared to be no difference in knowledge levels among workers between high and low performing states. Using multivariate analysis, the key finding of the study was that respondents reporting high levels of role conflict, and lower levels of job satisfaction were more likely to report feeling pressure to cheat. The finding related to role conflict is consistent with the constructs of principal-agency theory suggesting that the more clearly roles are defined, the less likely agents are to act against the wishes of principals. The link between job satisfaction and role conflict on feeling pressure to cheat provides an intriguing starting point from which to further explore the issue of cheating and the applicability of principal-agent theory to explain cheating behavior.

THE CHILD SUPPORT INCENTIVE PROGRAM: CONSTRAINTS AND THE CAPACITY FOR ORGANIZATIONAL CHEATING BY STREET-LEVEL BUREAUCRATS by Tracy R. Graham A dissertation submitted to the faculty of The University of Utah in partial fulfillment of the requirements for the degree of Doctor of Philosophy Department of Political Science The University of Utah December 2010 Copyright © Tracy R. Graham 2010 All Rights Reserved The University of Utah Graduate School STATEMENT OF DISSERTATION APPROVAL The dissertation of has been approved by the following supervisory committee members: , Chair Date Approved , Member Date Approved , Member Date Approved , Member Date Approved , Member Date Approved and by , Chair of the Department of and by Charles A. Wight, Dean of The Graduate School. Tracy R. Graham Richard T. Green 12/4/2009 Matthew J. Burbank 12/4/2009 James J. Gosling 12/4/2009 J. Steven Ott 12/4/2009 Leslie Francis 12/4/2009 Matthew J. Burbank Political Science ABSTRACT Many have argued that government "should be run like a business." This argument has evolved throughout the years, but implies that implementing performance measures, benchmarks and incentives will provide inducements for improving organizational performance. While the literature is replete with such arguments, there has been little attention paid to the "wrong-side" of performance management and whether incentive structures result in instances where individuals and organizations engage in efforts to cheat, or "game the system." The nation's child support program is a good example of a government program that has adopted an incentive-based approach. Four state child support agencies, representing varied degrees of performance, allowed their staff to participate in a survey that explored issues of cheating and knowledge of the child support incentive program. This study constructed cheating as the capability of artificially inflating performance levels. The underlying research hypothesis, based on principal-agency theory, suggested that higher levels of cheating would be found among higher incentive earning states. The findings of the study suggested otherwise. The study found that while front-line child support professionals identified performance as being important, few were aware that a child support incentive system existed. This finding is important as it suggests that the incentive structure does little to influence front-line worker behavior and does not appear to have much impact on iv workers' performance levels. Furthermore, there appeared to be no difference in knowledge levels among workers between high and low performing states. Using multivariate analysis, the key finding of the study was that respondents reporting high levels of role conflict, and lower levels of job satisfaction were more likely to report feeling pressure to cheat. The finding related to role conflict is consistent with the constructs of principal-agency theory suggesting that the more clearly roles are defined, the less likely agents are to act against the wishes of principals. The link between job satisfaction and role conflict on feeling pressure to cheat provides an intriguing starting point from which to further explore the issue of cheating and the applicability of principal-agent theory to explain cheating behavior. CONTENTS ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x Chapters INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1 PERFORMANCE MANAGEMENT IN THE PUBLIC SECTOR: AN OVERVIEW . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Weberian Organizational Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 PPBS Budgeting and Management Systems . . . . . . . . . . . . . . . . . . . . . . . 8 Management by Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Zero Base Budgeting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Open System Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Total Quality Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 The Reinvention Argument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 The New Managerialism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2 DEFICENCIES WITH PERFORMANCE MANAGEMENT AND INCENTIVE SYSTEMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Principal-Agency Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Weaknesses of Principal-Agency Theory . . . . . . . . . . . . . . . . . . . . . . . . . . 20 3 THE CHILD SUPPORT INCENTIVE PROGRAM . . . . . . . . . . . . . . . . . . . . 27 The Social Security Amendments (1975) . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 The Child Support Enforcement Amendments (1984) . . . . . . . . . . . . . . . . . 29 The Family Support Act (1988) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 The Personal Responsibility and Work Opportunity Reconciliation Act (1996) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 The Child Support Performance and Incentive Act (1998) . . . . . . . . . . . . . . 35 The Deficit Reduction Act of 2005 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 vi The American Recovery and Reinvestment Act of 2009 . . . . . . . . . . . . . . . . . 40 4 RESEARCH DESIGN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Research Questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 State Programs Participating in the Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 State "A" . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 State "B" . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 State "C" . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 State "D" . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Survey Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 5 MEASURING THE KNOWLEDGE OF CHILD SUPPORT INCENTIVE PROGRAM BY FRONT-LINE WORKERS . . . . . . . . . . . . . . . . 56 6 ASSESSING ORGANIZATIONAL CHEATING . . . . . . . . . . . . . . . . . . . . . . . 65 7 PART ONE: INFLUENCING VARIABLES ON CHEATING-ANALYSIS FOR ALL SURVEY RESPONDENTS . . . . . . . . . . . . . . . . . . . . . 75 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Dependent Variable #1: Possibility of Cheating . . . . . . . . . . . . . . . . . . . . . . . . 79 Role Conflict and the Possibility of Cheating . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Job Stress and the Possibility of Cheating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 Dependent Variable #2: Awareness of Others Cheating . . . . . . . . . . . . . . . . . . 84 Role Conflict and Awareness of Others Cheating . . . . . . . . . . . . . . . . . . . . . . . 85 Job Satisfaction and Awareness of Others Cheating . . . . . . . . . . . . . . . . . . . . . 87 8 PART TWO: INFLUENCING VARIABLES ON CHEATING-ANALYSIS OF STATE LEVEL RESPONDENTS . . . . . . . . . . . . . . . . . . . . . . 90 Findings for Cheating- State A . . . . . .. . . .. . . . . . . . . . . . . . . . . . . . . . . . . .. . 91 State A- Job Satisfaction and Possibility of Cheating . . . . . . . . … . . . . . .. . . 92 State A- Strategizing and Possibility of Cheating . .. . . . . . . . . . . . . . . . . . .. . . 93 Findings for Cheating- State B . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 State B- Awareness of Others Cheating and Job Satisfaction . . . . . . . . . . . . . 96 Findings for State C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 State C- Role Conflict and the Possibility of Cheating . . . . . . . . . . . . . . . . . . 98 State C- Supervision and the Possibility of Cheating . . . . . . . . . . . . . . . . . . . . 99 State C- Role Conflict and the Awareness of Others Cheating . . . . . . . . . . . . 101 State C- Job Satisfaction and Awareness of Others Cheating . . . . . . . . . . . . . 103 Findings for State D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 State D- Role Ambiguity and Awareness of Others Cheating . . . . . . . . . . . . . 106 vii State D- Knowledge of Performance Appraisal and Awareness of Others Cheating . . . . . .. …. … .. . . . . . . . … . . . . . . .. . . . . 107 9 FEELING PRESSURE TO CHEAT: A PREDICTIVE MODEL . . . . . . . . . 112 Dependent Variable: Pressure to Cheat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Independent Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 Findings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115 Job Satisfaction and Feeling Pressure to Cheat . . . . . . . . . . . . . . . . . . . . . . . . 117 Role Conflict and Feeling Pressure to Cheat . . . . . . . . . . . . . . . . . . . . . . . . . . 118 Possible to Inflate Performance and Feeling Pressure to Cheat . . . . . . . . . . . . 119 10 DISCUSSION AND CONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 Finding #1- Performance Matters for Child Support Professionals . . . . . . . . . 124 Finding #2- Unawareness of Incentive System . . . . . . . . . . . . . . . . . . . . . . . 125 Finding #3- Understanding Cheating Within the Child Support Context . . . . 127 Finding #4- The Importance of Role Conflict and Job Satisfaction with Cheating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 Finding #5- Respondents' Ideas About the Possibility of Cheating Are Related to Feeling Pressure to Cheat . . . . . . . . . . . . . . . . 132 Finding #6- Limitations of the Principal-Agent Model . . . . . . . . . . . . . . . . . . 132 Future Research . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 APPENDICES A- STATE "A" PROFILE DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 B- STATE "B" PROFILE DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139 C- STATE "C" PROFILE DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 D- STATE "D" PROFILE DATA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 E- CHILD SUPPORT SURVEY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 F- SCALE QUESTIONS AND CONSTRUCTION . . . . . . . . . . . . . . . . . 153 WORKS CITED . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 LIST OF TABLES Table Page 4.1 State incentive rankings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 5.1 Responses of state's ability to receive incentive funding . . . . . . . . . . . . . . . 58 5.2 Responses for possibility of penalties for poor performance . . . . . . . . . . . . . 59 5.3 Responses of whether organizations have been penalized . . . . . . . . . . . . . . . 60 5.4 Responses comparing incentive and nonincentive tasks . . . . . . . . . . . . . . . . . 61 5.5 Standardized measures for incentive and nonincentive tasks . . . . . . . . . . . . . 62 5.6 Measures for incentive and nonincentive tasks (excludes caseload size variable) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 6.1 Responses for impact of personal performance on organizational performance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 6.2 Responses for strategizing with co-workers to improve performance . . . . . . . 67 6.3 Responses for strategizing with supervisors to improve performance . . . . . . . 68 6.4 Responses for organizational strategizing to improve performance . . . . . . . . 69 6.5 Responses for possibility of cheating (Inflating performance) . . . . . . . . . . . . 71 6.6 Responses for knowledge of others cheating (Inflating performance) . . . . . . . 72 6.7 Responses for workers feeling pressure to cheat (Inflate performance) . . . . . 73 7.1 Survey questions pertaining to cheating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 7.2 Summary of association of independent variables on the possibility of cheating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 ix Table Page 7.3 Summary of association of independent variables on others cheating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 7.4 Survey questions pertaining to role conflict . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 7.5 Effect of role conflict on possibility of cheating for all states . . . . . . . . . . . . . 82 7.6 Survey questions pertaining to job stress scale . . . . . . . . . . . . . . . . . . . . . . . . 83 7.7 Effect of job stress on possibility of cheating- All states . . . . . . . . . . . . . . . . . 84 7.8 Effect of role conflict on awareness of others cheating- All states . . . . . . . . 85 7.9 Survey questions pertaining to job satisfaction scale . . . . . . . . . . . . . . . . . . 87 7.10 Effect of job satisfaction on awareness of others cheating- All states . . . . . . 88 8.1 Summary table of significant variables for the possibility of cheating . . . . . . 90 8.2 Summary table of significant variables for awareness of others cheating . . . . . 91 8.3 Effect of job satisfaction on possibility of cheating- State A . . . . . . . . . . . . . . 92 8.4 Effect of strategizing and possibility of cheating- State A . . . . . . . . . . . . . . . . 94 8.5 Effect of job satisfaction and awareness of others cheating- State B . . . . . . . . 96 8.6 Effect of role conflict on the possibility of cheating- State C . . . . . . . . . . . . . . 99 8.7 Effect of supervision on possibility of cheating- State C . . . . . . . . . . . . . . . . . 100 8.8 Effect of role conflict on awareness of others cheating- State C . . . . . . . . . . . 102 8.9 Effect of job satisfaction on awareness of others cheating- State C . . . . . . . . . 104 8.10 Effect of role ambiguity on awareness of others cheating- State D . . . . . . . . . 106 8.11 Effect of knowledge of performance appraisal on awareness of others cheating- State D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 9.1 Logistic regression predicting pressure to cheat from: Gender, race, state, job satisfaction, role conflict, possible to inflate performance, and awareness of incentives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 ACKNOWLEDGMENTS The culmination of this dissertation has not been an isolated endeavor and a number of individuals and organizations deserve special thanks from the author. First, Rick Green and Matt Burbank have been invaluable sources of support and knowledge as this project has unfolded over the past four years. My employer, the Office of Recovery Services, has been supportive throughout the years as I have worked to complete my dissertation and Ph.D. Special thanks go to Mike Tazelaar, Emma Chacon, and Mark Brasher for allowing me the flexibility and support as I have worked through this lengthy process. And last, but certainly not least, my wife Micah, and three children (Mallaree, Jarron, and Nikkee) deserve special recognition for putting up with me as I have tried (sometimes not too successfully) to balance the demands of work, school, and family. Once this is done, I will focus my efforts to be a better husband and father to you all! INTRODUCTION Public administration in the United States has a lengthy obsession with the values and practices associated with democracy on the one hand, and with values and practices of the business world on the other. The discipline of public administration has struggled throughout its history with reconciling these two distinct philosophical approaches to public management. In recent years, the field has leaned heavily toward the business model, with members of Congress, presidents, scholars, and the media asserting that government should be run more like a business. This call to action has been bolstered by the implementation of incentive programs, performance management, and privatization for public government services. While the rhetoric espouses the intrinsic worth of these private-sector influenced methods, there has been limited, though insightful, discussion in the public administration literature pertaining to the possibility that these types of programs may introduce questionable practices within public organizations. Personal experience has caused this author to inquire of the inherently problematic aspects associated with incentive structures and performance measurement systems in the public sector. While the literature suggests the possibility of employees manipulating performance, there has been little empirical work conducted on the subject. The questions remain: do incentive systems offer perverse incentives to organizations and their employees to perform well in incentive areas? Is it possible that performance 2 management systems can be manipulated by organizations and their employees? If this manipulation occurs, under what conditions does it take place? Do certain organizational characteristics curtail or enhance the likelihood of what Bohte and Meier (2000) call organizational cheating? These fundamental questions represent the underpinnings of this study. Bohte and Meier (2000) studied the occurrences of organizational cheating within Texas public school systems prior to the national passage of the No Child Left Behind (NCLB) initiative. Their findings suggested that schools lacking resources and those that felt pressure from "extreme task demands" were more likely to cheat than other schools. Bohte and Meier concluded their article by calling on the discipline of public administration to pay more attention to the conditions under which organizations may feel pressure to cheat and to remedy those situations accordingly. This study attempts to answer Bohte and Meier's call to action by focusing on the impact the nation's child support incentive system has on front-line child support workers. The nation's child support program presents a valuable case study in relation to the problematic side of incentive systems and performance management. Since so many aspects of the child support program are related to collections, the process of quantifying inputs, outputs, and outcomes has been relatively simple compared with other social welfare programs. In addition, the child support program is one based on a shared partnership between the federal and state governments. Under this partnership, states have been given a great deal of latitude in establishing the ways in which they choose to organize themselves. Some states have chosen to centralize service delivery, while others rely on programs administered at the local level. In addition, some states rely on private 3 service contracts to provide services to the public. This organizational diversity allows for a comparison between states and the subsequent methods each adopts to perform effectively. More important, however, is each state's ability to participate in an active incentive program which seeks to reward states for performing well on a predetermined performance objective system. The incentive-based program identifies key performance areas for the child support program in five substantive areas. Ultimately states can "maximize" their incentive dollars by scoring well on preselected performance standards. Earned incentive dollars are intended to be reinvested in the child support program to improve the program's effectiveness. It is also important to note that penalties exist for poor performance in several of the performance measures. It is, therefore, in each state's best interest to perform well in the program. This study examines the degree to which the federal child support incentive program acts as a constraint upon street-level child support workers and their daily work activities. If the incentive system acts as a mechanism to focus worker's tasks, a subsequent research question surrounds the likelihood of child support agencies' willingness to engage in manipulative behaviors by attempting to artificially increase performance in the key incentive areas. Specifically, this study will explore the experiences of individual child support workers and the pressures they may feel to engage in these questionable practices. Chapter 1 of this dissertation provides an underlying foundation regarding the historical development of performance management systems in American public administration. As mentioned previously, the concept of performance management is not 4 a new one for the discipline. In fact, this analysis demonstrates that the issue of measuring performance is one that cycles through the course of our nation's administrative history. Chapter 2 provides a more detailed analysis of the nature of principal-agency theory in relation to the use of governmental incentive programs. While some suggest this mode of thinking results in effective methods for instituting control and enhancing efficiency, others counter this idea by pointing out some of the deficiencies associated with principal-agent type relationships. As this chapter demonstrates, there are some potentially negative consequences related to incentive systems and the use of performance management. As a result, incentive programs have the potential to introduce perverse incentives resulting from implementation of incentive structures. Chapter 3 presents a historical analysis of the development of the child support enforcement program in the United States. The chapter places specific focus on the development of the child support incentive system in the wake of significant policy changes that resulted from the welfare reform initiatives of the early 1990s. In addition, the chapter provides additional details concerning each of the performance measures including how the measures are calculated and their underlying reasons for existing. Chapter 4 addresses the methodological constraints used by this study in attempting to address the research questions at hand. The chapter explains that data were gathered from four state administered child support agencies using an online survey. The chapter also provides the rationale used to categorize the four state programs in relation to the incentive levels each earned. Of particular importance is identifying the constraining impact the incentive has at the individual worker level. Do overwhelming 5 pressures exist within child support agencies to perform well in incentive areas? Do workers feel compelled to manipulate computer systems to artificially increase performance outcomes? These, and other, important questions were addressed through the questions posed by the survey. The remaining chapters present the findings of the study. Chapter 5 focuses on the results of the study concerning the knowledge street-level child support workers have regarding the child support incentive systems. The results of this analysis presented some surprising results regarding how well workers understood the concepts and tasks related to the incentive system. Chapter 6 focuses specifically on assessing the level of organizational cheating in each of the agencies studied, while Chapters 7 and 8 focus on eight independent variables and their subsequent impact on cheating. Finally, Chapter 9 implements a multivariate logistic model that tests the effects of several key variables on the pressure to cheat. Performance management systems and incentive programs have been implemented in a wide-range of programs to enhance their effectiveness and their value to the public at large. While this study does not adopt the perspective that performance management systems are inherently bad, it does highlight deficiencies and unintended consequences that emerge when implementing these types of programs. The primary justification for writing this dissertation is to shed some light on attempts to develop performance management systems in a way that does not compromise the integrity of their disciplines and practices. CHAPTER 1 PERFORMANCE MANAGEMENT IN THE PUBLIC SECTOR: AN OVERVIEW One of the methods used to improve oversight and accountability in government has been to implement performance measurement systems in public agencies. Proponents of these systems argue that they have two purposes: to enhance efficiency and to promote accountability of government agencies (Broom et al. 2002; Coggburn and Schneider 2003; De Lancer and Holzer 2001; Garvey 1995). Heinrich (2003) relied heavily upon Thompson's (1967) descriptions of organizations in developing a useful framework which described the development of performance measurement systems in the United States. Thompson's approach is used in this proposal as a means to facilitate a discussion of the history of performance measurement in the United States. James Thompson suggested that organizations can best be understood by looking at two broad categories of organizational theory that attempt to describe how organizations function. The first group, labeled as "closed systems," relies heavily upon notions of rationality. From a closed perspective, organizations are viewed as rational entities seeking to maximize efficiency, organizational performance, and outputs (Thompson 1967). Closed system approaches tend to emphasize the organization's need to establish certainty in their environment through the establishment of hierarchies, 7 formal rules, and repetitive processes (Thompson 1967). In contrast, Thompson found that open system theories were not as dependent upon hierarchy and rules as closed system models. Instead, open systems are viewed as "natural systems" comprised of interdependent parts that cooperate with one another in order to cope with uncertain and unstable environments (Thompson 1967). These two approaches are used as a guiding theme in analyzing the historical development of performance management in American government. Weberian Organizational Theory Closed system approaches rely upon the Weberian notion of bureaucracy (Thompson 1967). While Weber's writings were not translated into English until the mid- twentieth century, ideas closely resembling Weber's theory of bureaucracy were expressed by many of the early theorists in the field of public administration. For Weber, bureaucratic processes and structures were constructed by individuals in an attempt to accomplish a given set of goals (Coser 1977; De Lancer and Holzer 2001). The underlying theme of rationality lies at the heart of Weber's theory of bureaucracy. Weber viewed organizations as embodying inherent interests, goals, and purposes (Thompson 1967). The establishment of rules, hierarchy, and authority help create a sense of uniformity within bureaucracies to assist in accomplishing organizational purposes (Burke 1986; Perrow 1972). These structures also serve to buffer the organization from external influences in an intentional effort to increase certainty within the organizational environment. According to Perrow (1972, 5), the elimination of all unwanted "extraorganizational influences" proved to be a rational approach for maximizing 8 efficiency. As such, the bureaucratic model exhibited the potential to remove the imperfections associated with human behavior from organizational settings. "The more bureaucracy is ‘dehumanized', the more completely it succeeds in eliminating from official business love, hatred, and all purely personal irrational and emotional elements which escape calculation. This is the specific nature of bureaucracy and it is appraised as its special virtue" (Weber 1946, 216). Other developments in the closed system approach include Taylor's (1912) scientific management movement, Luther Gulick's (1937) POSDCORB principles, and Chester Barnard's (1938) influential book The Functions of the Executive. Each of these approaches emphasized the role of a relatively small number of variables in environments where goals and tasks were known, and where a prediction of outcomes was generally considered to be reliable (Heinrich 2003). PPBS Budgeting and Management Systems Under the constructs imposed by closed system approaches, early attempts at developing modern performance management systems began to emerge by the 1950s. Many of these approaches were imported to public sector organizations from the business world. One of the first attempts at developing a more holistic approach to systematic management was Planning Programming Budgeting Systems (PPBS) (Gosling 2002; Nelson et al. 1998). Schick (1966) identified PPBS as moving beyond control and managerial budgeting through the use of extensive planning and goal setting. "One of the major aims of PPB[S] is to convert the annual process of preparing a budget into a conscious appraisal and formulation of future goals and policies" (Schick 1966, 244). Schick 9 claimed that the three functions of budgeting (control, management, and planning) would come into balance under PPBS. Most importantly, PPBS required top policy makers to establish broad policy objectives (Schick 1966). Once these objectives were identified, they acted as a constraint for lower ranking officials in the bureaucracy and served as a guide for subsequent actions. While PPBS promised to improve government, it eventually failed as a budgeting system in the United States. The reasons for its failure included the difficulty of identifying measurable objectives, inadequate training in implementing PPBS for agency budgeting officials, and the fractured federal governmental structure which hindered coordination efforts between the executive and legislative branches (Gosling 2002). Management by Objectives Another private sector management approach that was applied to the public sector was Management by Objectives (MBO). This approach, developed by Peter Drucker (1954), placed a heavy emphasis on the role of management in accomplishing organizational objectives. According to Drucker, management has three distinct purposes: ensuring economic performance, enhancing efficiency/productivity, and managing people effectively to accomplish the first two managerial functions. With these goals in mind, Drucker argued that the best way to improve management was through the systematic identification of objectives deemed critical to the success of an organization. "Objectives are needed in every area where performance and results directly and vitally affect the survival and prosperity of the business" (Drucker 1954, 63). By identifying and measuring these objectives, management would 10 improve its position to meet the needs of customers and improve organizational performance. The importance of Drucker's MBO model was the emphasis it placed on quantitative (and qualitative) measurement of individual and organizational performance in order to improve efficiency and productivity. Versions of MBO were implemented in both the private and public sectors (Brady 1978; De Woolfson 1975) and proved to be influential in the development of subsequent performance management systems in American management. Zero Base Budgeting A final example of a closed system model was the implementation of zero-base budgeting (ZBB) during the 1970s and 1980s. After being elected president in 1976, Jimmy Carter brought the zero-base budgeting concept with him after using it while serving as governor of the state of Georgia. ZBB required government agencies to construct four separate budgets on the basis of four different funding scenarios (Gosling 2002). This process required budgeting officials to examine their entire budget each year and to identify priorities within their respective programs. While ZBB was ultimately abandoned by the federal government as a formal budgeting system with the election of Ronald Reagan, key principles of ZBB continued to be used by budget managers at both the federal and state levels of government (Gosling 2002). It is also important to recognize that key characteristics of ZBB dealt with performance management and monitoring (Brown 1981; Lauth 1985). In citing Georgia's experience with ZBB, Lauth (1985) found that ZBB offered decision makers the opportunity to direct resources to programs that had achieved key program objectives. 11 Brown's (1981) article suggested that decision makers using ZBB systems would make their final budgetary decisions based upon which programs demonstrated high cost-effectiveness. In sum, ZBB presented a budgetary system that continued funding for programs which demonstrated effective performance in key areas. The examples of MBO, PPBS, and ZBB all represented closed system approaches as defined by Thompson (1967) and Heinrich (2003). Each of these perspectives attempted to describe organizations and how they prescribe remedies to help them improve performance internally. Interestingly, each of these viewpoints describes organizations as endogenous entities where all variables are known and where individuals respond willingly to authority and rules (Heinrich 2002). As organizations became more intricate over time, a new perspective would develop to account for the emerging complexities facing organizations. Open System Models Unlike the closed system approaches, open system models viewed organizations as "natural systems" (Thompson 1967) where organizational structures were more fluid and open to outside influences. While rational-legal models (i.e., closed system approaches) placed emphasis upon the bureaucratic response of buffering the organization from external forces, open system approaches embraced the challenges associated with organizational interdependence (Heinrich 2003; Thompson 1967). The origin of open-systems thinking was grounded in the work of Herbert Simon who criticized rational-decision making processes (Thompson 1967). Simon's (1945) pointed critique found that rational decision making assumes that knowledge is complete and that all factors can be identified through systematic analysis. Simon suggested that 12 this is not the case. Instead, Simon proposed the notion of "bounded rationality" to help describe decision making in an environment of uncertainty and unknowns. "In making administrative decisions it is continually necessary to choose factual premises where truth or falsehood is not definitely known …" (Simon 1945, 60). Simon's contributions to the field of organizational theory go well beyond his theory of decision making. Simon's influence helped scholars to begin questioning the rationally-based closed system approaches of organizational theory and move toward a more open, systemic conception of how organizations operate (Steiss and Daneke 1980). New managerial theories and concepts emerged in the wake of challenging the dogmas related to traditional organizational theory. The main emphasis in many of these approaches embraced an external (or open) perspective in finding ways to increase organizational productivity in an ever increasingly interconnected world. Total Quality Management Total Quality Management (TQM) emerged in the 1980s as dissatisfaction with previous attempts to improve management had failed (Heinrich 2003). The primary emphasis of TQM addressed the role "customers" play in the business world. "The consumer is the most important part of the production line. Quality should be aimed at the needs of the consumer, present and future" (Deming 1986, 5). By emphasizing the customer and improving quality at all levels of an organization, TQM promised to "transform" management systems in the American business world. In developing TQM, Deming identified fourteen points that would help his efforts to transform American management systems. Included in these fourteen points was a need for management to create a "constancy of purpose" (23) for employees, 13 emphasizing quality at all levels, promoting employee involvement, and focusing on constant improvement of products and processes to benefit the customer. TQM took the business world by storm and was subsequently adopted by public sector agencies across all levels of government. The way in which TQM was implemented varied, but the one common theme that emerged was the reliance upon performance measures by management to improve productivity within organizations. While it can be argued that the way TQM was actually implemented differed from Deming's ideas, the TQM movement reinforced the ideals associated with measuring performance in the psyche of American management. The Reinvention Argument Calls to reinvent government grew in fervor with the Reagan Era, which brought with it taxpayer revolts, demands for downsizing government, and policies promoting devolution. This resulted in a call from academics, politicians, and the general public to take a serious look at how government conducted its business. Reinventing government emerged as a dominant theme, even a new paradigm, for improving government performance, while at the same time, reducing government programs and expenditures (Poister 2003). Osbourne and Gaebler's (1993) Reinventing Government had a tremendous impact on government reform in the early 1990s. They argued that in order for government to improve, it must adopt an entrepreneurial approach that would enable government workers to cut through "bureaucratic red tape" and to perform more effectively. 14 It is important to note that Osbourne and Gaebler's reinvention approach viewed government as being inherently different from business. While supportive of some private sector practices, Osbourne and Gaebler suggested that reinvention attempts must refocus on shifting government to an outcome-based perspective. In order to accomplish this goal, government should use privatization, contracting for services, decentralizing decision making authority, and adopting performance measures to monitor performance (Osbourne and Gaebler 1993). Osbourne and Gaebler's work heavily influenced Vice President Al Gore's National Performance Review (NPR). Commissioned by President Bill Clinton, Gore oversaw the reinvention efforts for the federal government. The mission given to Gore was two fold: first, make government work better, and second, make it cost less (Gore 1993). Reinvention relied upon methods of performance management as its primary tool (Gore 1993). "Performance management programs should have a single goal: to improve the performance of individuals and organizations" (Gore 1993, 26). The NPR went on to say that "agencies will gradually build performance information into their own budget guidance and review procedures, into their strategic and operational plans, and into revised position descriptions for their budget, management and program analysts" (Gore 1993, 111). With the publication of the NPR, the Clinton administration embarked on a mission to reinvent the federal government using performance management as a key method to accomplish their goals. During the proceeding years, the Clinton administration and Congress passed welfare reform legislation and adopted the 15 Government Performance and Review Act (GPRA) which required all federal agencies to identify specific performance measures for their organizations by September 30, 1997 (Kravchuck and Schack 1996). Proponents of GPRA argued that it would provide an effective tool to assist Congress in obtaining additional oversight of the bureaucracy. Supporters claimed that GPRA would ensure accountability to citizens and elected officials alike, in addition to helping instill a results oriented perspective for government (Mihm 2001). Mihm contended that once results oriented government was implemented, employees in this type of system would "understand the importance and the connection between their individual performance and the organization's success" (2002, 42-43). The New Managerialism The call for reinventing government in the 1990s, combined with a reliance upon rationally-based economic theory, resulted in the development of a new managerial philosophy in the public sector known as both the New Public Management (NPM) and managerialism (Hughes 1994). These terms will be used interchangeably in this study. Managerialism represented an ideology which praised the values associated with management itself (Pollitt 1993). NPM assumed that individuals in government organizations were rational and would respond accordingly through proper performance monitoring by supervisors (Hughes 1994). Essentially, NPM suggested that effective management, by itself, was capable of improving government programs and services (Pollitt 1993). Hughes (1994) contended that NPM placed an emphasis on governmental results. In order to improve government results, managers adopting the NPM perspective rely 16 heavily upon developing flexible (i.e., nonbureaucratic) organizations, promoting performance monitoring systems, and bringing market forces to bear in government through the practices of contracting for services and privatization (Hughes 1994). Proponents of the NPM claimed that these approaches resulted in an improved and more economical government for citizens. While Pollitt (1993) has pointed out some inherent weaknesses with managerialism (including the lack of clearly established goals in public programs, and the model's inability to deal with the conflicting values of fairness and equality in comparison with that of efficiency), this new managerial philosophy has developed into what Rosenbloom (2002) has called the dominant managerial approach in public administration today. It appears as if the field's love affair with performance monitoring is here to stay. In conjunction with the growing popularity of performance monitoring and incentive plans, it is important to note that the child support program at the national level has enamored itself with these practices. Chapter 3 provides a detailed history of the development of the child support program during the past thirty years. Special emphasis is placed on how the program has used a combination of performance management techniques and the use of incentives, to dramatically emphasize the importance of performance at the state and local levels of government. In turn, this emphasis on performance allows states to draw additional federal funding for good performance as well as the possibility of penalties for poor performance levels. How successful the reinvention effort is remains to be seen. While some are critical of the reinvention developments in the field of public administration (Moe 1994; 17 Nathan 2001; Radin 2002), it seems as if the proposals initiated with PPBS have come full circle. Contemporary public administration relies heavily upon quantitative measures as the basis for improving results, performance, and accountability. Performance management is presented as a uniquely qualified means to solve many of government's most serious problems. CHAPTER 2 DEFICENCIES WITH PERFORMANCE MANAGEMENT AND INCENTIVE SYSTEMS The reasons given for implementing performance management systems fall into two broad categories: first, the values associated with efficiency, and, second, to exert more control and accountability over public programs. This chapter focuses on the latter. It explores the literature associated with bureaucratic control through performance management and how the use of incentives attempts to "reign in" the bureaucracy. Brehm and Gates (1997) suggested that one body of literature provides an underlying foundation for explaining how the use of performance management and incentives affect subordinates and superiors within contemporary organizations. This literature employed an economic perspective that suggested that individuals are ultimately self-interested and act in a rational manner to fulfill their own needs (Downs 1967). Assuming that subordinates are ultimately self-interested, one can rationally infer that they will only carry out the wishes of others if properly induced into acting according to their superior's wishes. Principal-agent theory has built upon this notion and has attempted to develop a model which explains how superiors (i.e., principals) can exert control over subordinates (agents). 19 Principal-Agency Theory Rational actor models hinge upon the idea that individuals will almost always act in a manner that benefits (or in the verbiage of economists, "maximizes") their own self-interest (Brehm and Gates 1997; Downs 1967; Miller 1990). Because it is assumed that individuals will act rationally, developers of rational-actor models make the broad claim that they offer a robust approach that can be used to predict numerous types of behavior. Economists (in addition to other social scientists) developed principal-agent models to describe the relationship that exists between superiors and subordinates in a number of fields, including public management (Arrow 1985; Lynn 1996; Pratt and Zeckhauser 1985). Specifically, principal-agent models rely heavily upon the notion of incentives, inducements, and punishments as motivating factors in controlling subordinate behavior. The result of such systems is a model which creates "linear, hierarchical, top-down" relationships that act as a constraint upon subordinate behaviors and actions (Feldman and Khademian 2002, 543-44). Superiors in principal-agent relationships act as the principal. Principals, in general, have a given set of preferences regarding some type of activity that they would like to see implemented. Since principals typically do not have the time, resources, or even knowledge to complete these tasks, they must rely on subordinates (agents) to carry out these assignments on their behalf. Under such relationships, the principal's primary duty is to establish a system and design "incentives such that the agents find it in their own interests to take the best possible set of actions (from the principal's perspective)" (Miller 1990, 325). 20 While principals may establish these systems, they are inherently reliant upon agents to carry out their wishes. Under such relationships, agents are viewed as having their own distinct preferences and other self-interests that may differ from the preferences of principals (Arrow 1985; Brehm and Gates 1997; Hammond and Miller 1985; Lynn 1996; Miller 1990). As such, lower-level subordinates have the potential to wield a great deal of power and influence in principal-agent relationships (Mechanic 1962). Because these differences among preferences persist, the primary duty of the principal is to create a situation where the preferences of agents come into line with those of the principal. According to principal-agent theory, the best way to do this is through the development of a contractual type relationship which relies heavily upon the notion of inducements to control and direct subordinate behavior and actions (Wood and Waterman 1994). In terms of this study, the child support program in the United States relies heavily upon performance measures created by Congress and the Office of Child Support Enforcement (OCSE) (the principals) which act as a constraint on state and local programs' (the agents) collective behavior. For nearly two decades economists have claimed that their models are robust and predictive in nature given the assumptions of rationality and utility-maximization. While their assertions may be true (again, given their limited characterization of human nature), a growing body of literature has emerged which criticizes the applicability of principal-agency theory in a number of substantive areas. Weaknesses of Principal-Agency Theory The first of these criticisms relates to issues of hidden or unobservable actions (Lynn 1994). Lynn found that a significant problem in principal-agent relationships is 21 "securing reliable task performance from agents" (1994, 115). Lynn goes on to suggest that agents can engage in behaviors or activities that principals fail to observe. While principals, in their efforts to control agent behavior, seek to minimize these situations, they continue to persist. The result is that principals cannot possibly observe all activities of their agents due to cost restraints (Brehm and Gates 1997), the difficulty associated with monitoring certain activities (Pratt and Zeckhauser 1985), and the general limitations in rewards that principals can offer agents (Arrow 1985). A second problem with principal-agency relationships according to Lynn is that information asymmetries exist where agents, by the nature of their work, are placed in positions to know more about their work than do their principals (Lynn 1994). These information asymmetries result in potential disruptions to the power structure implicit between principals and agents. Where principals have been viewed as being in control, agents are in the position of having more information available to them and may use it to their advantage (Brehm and Gates 1997). A consequential third problem is that issues pertaining to trust between principals and agents are frequently problematic. Miller (1990) claimed that principal-agency relationships tend to focus on agent activities that are easy to observe. In establishing an incentive system based on observable actions, agents are left in the potentially precarious position of being held accountable for things that are outside of their control. "When this occurs, the principal and the agent can only contract for outcomes which may be jointly determined by the agent's actions and some external random variable[s]" (Miller 1990, 329). 22 The result of these types of situations is an agent blaming her performance on factors falling outside of her direct control. The principal, who is placed in a position of not knowing for sure, may not believe the agent and suggest she is shirking her responsibilities. In the end, neither the principal nor agent trust each other leaving both parties in a worse position (Miller 1990). A fourth problem with principal-agency models is that they present situations where agents are placed in the position where they may manipulate outcomes to benefit themselves (Hammond and Miller 1985). It is important to recognize that the incentive to manipulate outcomes is likely the result of the three previous criticisms (hidden actions, information asymmetries, lack of trust) associated with principal-agent relationships. While principals can attempt to create situations where manipulation is less likely, agents frequently will find themselves in a position where they can inherently improve their situations vis a vis their principals. This criticism poses serious questions for the validity associated with contemporary incentive systems. Stone provided some additional insight into why agents may be willing to manipulate outcomes. While numerical analysis represents the predominant mode of implementing control in principal-agent relationships, Stone has noted that measuring only represents "one of many ways to describe" social phenomena (2002, 163). Stone's perspective of measurement in contemporary policy analysis provides a useful construct from which to view performance management systems in the public sector. According to Stone, the use of numbers as a measurement device is not as unbiased as many would like to think. Stone argued that numbers are used strategically by individuals or organizations in an effort to promote specific perspectives. As such, 23 numbers are used symbolically as a storytelling device concerning problems and possible solutions. "Numbers … are measures of human activities, made by human beings, and intended to influence human behavior" (Stone 2002, 177). Using numbers as a foundation for measuring schemes results in four identifiable outcomes that fundamentally question the objectivity frequently attributed to the use of quantitative methods. First, whenever numbers are used, a conscious decision must be made about what counts and what does not (Stone 2002). Those who make these decisions must construct various classification schemas in deciding where variables fall and how they will ultimately be counted. The result is that boundaries must be created so decisions can be made about what is included and excluded. In other words, "counting always involves deliberate decisions about counting as" (Stone 2002, 164). As a result, the decision of what counts and what does not entails a second problem-ambiguity. While many might make the assumption that the use of numbers results in less ambiguity, Stone contended, "Ambiguity-the range of choices in what to measure or how to classify-always lies just beneath the surface of any counting scheme" (2002, 165). Stone's suggestion, that numbers may be ambiguous, flies in the face of the intuitive notion that numbers imply precision and accuracy. The result is that numbers and measurements result from a value-laden, heavily subjective, and inherently political process that defines how the numbers are constructed. A careful analysis of the counting process reveals a less than objective perspective on the use of numbers and measurement systems. 24 A contemporary example of some problems with measuring is that public schools are now required to report progress on student performance under the guidelines of the No Child Left Behind policy initiative. While at first glance this seems fairly easy to accomplish, difficult questions begin to emerge. For example, should the scores of mentally challenged students be included in the school totals? What about the scores of students who speak English as a second language? Decisions on what counts and what does not has a dramatic impact on final scores depending upon how different schools include these groups of students in their respective counting schemes. Third, by measuring something, one implies a need for action (Stone 2002). Why would something be measured if it did not warrant being monitored in the first place? The simple act of finding a way to measure something suggests it is important enough to be measured. In addition, measurement also implies that a problem (or potential problem) may exist and that a solution is available to remedy the problem. The success of any given solution requires adequate measurement to be able to demonstrate that the solution was an effective antidote to the problem. Once a measurement has been introduced, it implies that individual or organizational behavior is being monitored. The fourth impact of measurement is that people, both individually and organizationally, respond to being measured (Stone 2002). "Measurement, like a mirror, triggers the natural desire to look good. People want to conform to their own ideals as well as to general social values. Measurement provokes people to ‘play the role' and to present themselves as they want to be seen" (Stone 2002, 178). Stone identified this phenomenon of responding to measurement as reactivity. 25 Reactivity is an important component of the measuring process. The entire philosophy of performance management is to improve performance through measuring activities and setting goals to increase individual and organizational achievement. Designers of performance management systems fully expect people to react to being measured and to improve their own performance. "People change their behavior in response to being measured, and the changed behavior leads to results that are actually different from what they would have been without the measure" (Stone 2002, 180). While people may want to look good when being measured, it is important to consider that self-image is not the sole factor influencing behavior in measurement systems. A fifth result of measurement is the role inducements play in the measurement process. According to Stone, inducements rely heavily on the carrot-stick approach. Inducements can be viewed as either a reward (positive inducements) or as punishment (negative inducements). "Every reward contains an implicit or potential penalty of withdrawal, and every penalty short of death contains an implicit reward of cessation" (Stone 2002, 265). When an inducement is offered to someone, their reactivity level to the measurement can be greatly influenced beyond that of just simply looking good. "When people are measured … something more than their self image and their desire to please is on the line; their fate is at stake" (Stone 2002, 179-80). Many performance management systems rely on inducements as an enforcement mechanism to alter behavior. As a result, performance measures can be viewed in both positive and negative perspectives. When so much is at stake, Stone has suggested that people "have a strong incentive to manipulate measures" to improve their own position and standing (2002, 180). 26 If Stone's theory is accurate, a serious dilemma is raised concerning the objectivity of performance measurement systems. Understanding reactivity is so important … because, unlike deliberate falsification of numbers, it is an inextricable feature of social measurement. Moreover, reactivity violates the canons of good scientific practice, on which all statistical reasoning is based. Scientific method assumes a strict separation between the observer and the observed. The subject of measurement should never measure himself or herself. To do so would be the essence of subjectivity, and scientific measurement pursues the ideal of objectivity, where neither the subject nor the observer has incentive or opportunity to manipulate the way a measure appears. (Stone 2000, 180) Stone's framework provides a useful perspective on the impact of measurement and performance management systems. In addition to Stone's critique of the use of numbers, others have noted that performance management can contribute to dysfunctional organizations (Bouckaert 1995), "gaming the system" (Poister 2003), and even organizational cheating (Bohte and Meier 2000). These issues lie at the heart of understanding the development of the child support incentive program in the United States. Chapter 3 details the history of the program relative to its emphasis of performance management and the development of a comprehensive incentives system for state and local programs. CHAPTER 3 THE CHILD SUPPORT INCENTIVE PROGRAM Child support, until the mid 1970s, was largely considered a private matter left to be resolved between parents and the state or local court systems (Sorenson and Hill 2004). As previously mentioned, this "private" child support system was one that hinged upon judges making determinations regarding child and spousal support for both divorced and never-married parents and their children. A primary problem inherent with this private system was its requirement of hiring attorneys in an attempt to obtain court-ordered child support (Garfinkel et al., 1998). This system resulted in a haphazard and inequitable allocation of child support for families throughout the country. The problem of children not receiving support from parents became more apparent over time. By 1970, one child out of eight was living in a family headed by a single mother (Huang et al. 2004). In a related trend, the amount of public dollars being spent on welfare tripled between the years of 1955 and 1975 (Garfinkel et al., 1998). Policy experts found that single-parent families were severely at risk for living in poverty and depending entirely upon the nation's welfare system for daily sustenance. The early emphasis of the child support program focused on using child support to help offset the costs associated with the nation's public assistance program. The emergence of the nation's "public" child support system culminated in 1975 when Congress narrowly 28 passed Title IV-D of the Social Security Act which established the nation's first comprehensive child support system (The history of child support and its partners, 2003). This chapter provides an overview of the nation's child support program and its subsequent development during the past thirty years. While its beginnings were modest, it has emerged as a major policy component in the nation's welfare system. Congress expanded the program in 1984, 1988, and most significantly in 1996 with the passage of the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA). Each of these acts added programmatic elements in an effort to make the child support program more effective and efficient (The history of child support and its partners, 2003). Most importantly, for this study, are the culminating changes brought with the passage of the Child Support Performance and Incentive Act of 1998 which established the system's current incentive program. The following pages will provide background on this important legislative initiative. The Social Security Amendments (1975) Until the passage of the Social Security Amendments in 1975, the nation's child support program was essentially a private program administered by parents and their interactions with the state's various court systems (Sorenson and Hill 2004). As the number of children residing in single-parent households increased, pressure mounted by women's groups for government to do something about the growing problem of children not being supported by absent parents, usually fathers (Garfinkel et al., 1998). The culmination of this pressure, in addition to the realization that child support could be used to offset some of the costs associated with welfare programs, led to the narrow passage of the Social Security Amendments in 1975 which created the modern child support system. 29 Specifically, Title IV-D of the Social Security Act required all states to establish an independent organization to oversee child support collections for families receiving Aid for Families with Dependent Children (AFDC) (The history of child support and its partners, 2003). This landmark legislation created the federal Office of Child Support Enforcement (OCSE) to oversee the implementation of the child support program by the states and authorized the federal government to pay three-fourths of the bill associated with start-up costs for the program (Garfinkel et al. 1998; The history of child support and its partners, 2003). The states were given these funds to help "locate absent parents, establish paternity, establish child support orders, and obtain child support payments" (Huang et al. 2004, 109). While the creation of OCSE and the IV-D program was initially controversial, Congress soon realized that in order for the child support program to move forward, it needed additional powers and authority to become more effective. By 1984 when President Ronald Reagan signed The Child Support Enforcement Amendments, both the House of Representatives and the U.S. Senate unanimously passed the legislation demonstrating a desire for the child support program to become more aggressive than it had been in the past (U.S. Department of Health and Human Services, 1984). The Child Support Enforcement Amendments (1984) The Child Support Amendments made four significant changes to the child support program. First, and foremost, the 1984 legislation expanded the child support program and its services to both AFDC recipients as well as nonwelfare families (The history of the child support program and its partners, 2003). Upon passage of the 30 amendments, the nation's child support program and its services became available to anyone in need, regardless of their involvement with the welfare system. The second important emphasis of the 1984 law was the change in the requirement for states to find ways to expedite the establishment of child support and paternity (The history of child support program and its partners, 2003; Garfinkel et al. 1998). In its annual report to Congress, OCSE reported that in 1981, more than 8.1 million women were raising children alone. Of those, only 4 million had established a child support obligation (United States Department of Health and Human Services 1984). Child support experts began to place emphasis on the importance of establishing paternity and child support orders quickly so that financial support could begin as soon as possible for parents and children in need. Third, the new legislation required states to adopt standardized child support guidelines in an attempt to reduce the inconsistent way in which child support orders were established. These inconsistencies in ordered child support amounts were typically the result of differences in judicial philosophy being applied in different judicial jurisdictions (Huang et al., 2004). For the first time in the program's history, states would be required to develop a standardized methodology in the way its judges determined what a child support obligation would be. This would prove to be an important first step in standardizing the process used for establishing child support orders throughout the country and eliminating biases that had previously existed in the way these support amounts were determined. The fourth change that the 1984 legislation brought about was a number of new enforcement techniques to be utilized by the states. The most important of these new 31 enforcement methods was the mandate that states must begin implementing immediate income withholding for all child support delinquencies in excess of thirty days (Sorensen and Hill 2004). Under this new provision, employers were required to withhold child support directly from the noncustodial parent's wages. Second, states were required to establish processes to withhold state tax returns from noncustodial parents who were delinquent in their child support obligations (The history of child support and its partners, 2003). This change provided states with an additional tool to collect monies owed by delinquent noncustodial parents by applying it to past-due child support arrearages owed to either the State or the family. Finally, states were required to adopt "long-arm" provisions to make it easier to collect child support across state lines (Garfinkel et al. 1998). The changes emerging from the 1984 legislation represented an important step in increasing the necessary tools to effectively collect child support by the nation's IV-D agencies. These changes represented the foundation of the IV-D program. However, future amendments to the nation's child support program would be required to further enhance collection efforts on the part of state agencies and the OCSE. The Family Support Act (1988) Four years elapsed before Congress and the president were willing to enhance the child support program again. An important aspect of the 1988 legislation involved the mandate for states to begin developing fully automated computer information systems (The history of child support and its partners 2003). The new legislation built upon standards that were first introduced with the Child Support Amendments in 1984. 32 One of the most significant changes of the Family Support Act was the mandate for states to implement mandatory income withholding, regardless of whether a delinquency existed, by the year 1994 (Huang et al. 2004). Income withholding, as Sorenson and Hill (2004) noted, has been child support enforcement's most effective collection tool to date. The ability for states to use income withholding has been a critical addition in helping IV-D agencies increase child support collections throughout the nation and has been used increasingly since 1988. In addition to the income withholding provisions, the Family Support Act also enhanced the 1984 Child Support Amendments by requiring states to adopt provisions making their child support guidelines presumptive (United States Department of Health and Human Services 1989). Under this new requirement, noncustodial parents would need to make significant arguments as to why their child support amounts should deviate from the guideline amount. If a judge allowed a deviated amount to be entered in a court order, the new legislation required the judge to enter specific findings of fact explaining why the deviation was warranted. With the ascent to power of Newt Gingrich and the "Contract with America" Republicans in 1995 came new and fervent promises to reform the nation's welfare system. Congress and the president would once again look to the child support program as a central player in moving people from welfare to work. The Personal Responsibility and Work Reconciliation Act of 1996 (PRWORA) ended the AFDC program and replaced it with a broad block grant entitled Temporary Assistance for Needy Families (TANF). The child support program also underwent serious enhancements as it played 33 an important role in providing monetary support for families moving from welfare toward self-sufficiency. The Personal Responsibility and Work Opportunity Reconciliation Act (1996) President Bill Clinton signed the Personal Responsibility and Work Opportunity Reconciliation Act (PRWORA) into law on August 22, 1996. While much of the attention related to PRWORA focused on reforming the nation's welfare program, significant changes were also made to the child support system. Under PRWORA, child support enforcement was viewed as an integral component of welfare reform that could provide financial stability for families thus keeping at-risk families from using government assistance. PRWORA enhanced the child support system in a number of important ways. First, employers were required to report all newly hired employees to IV-D agencies who would maintain the data as part of a State Directory of New Hires (SDNH) and transmit this information to the National Directory of New Hires (NDNH). This data base would be made available to all of the nation's IV-D agencies for the purposes of collecting child support (Sorenson and Hill 2004). In a related matter, states were also required to pass the Uniform Interstate Family Support Act (UIFSA) which made the enforcement of child support obligations uniform across state boundaries (The history of child support and its partners 2003). UIFSA was viewed as a critical component to solving a persistent problem of enforcing child support orders when the noncustodial parent resided in another state. 34 Third, PRWORA placed an increased emphasis on establishing paternities for out-of-wedlock children born in hospitals. States were required to change their laws to allow for paternity acknowledgement forms to be filled out and signed by parents while they were in the hospital and then forward the signed documentation to the states' offices of vital records. Fourth, a number of new enforcement remedies were added to the child support arsenal. These additional tools included provisions to deny passports to individuals who were more than $5,000 delinquent in child support, increased emphasis of income withholding (United States Department of Health and Human Services 1996; The history of child and its partners 2003), and mandated the requirement for states to establish centralized State Disbursement Units (SDUs) from which all child support payments would be sent (United States Department of Health and Human Services 1996). Finally, PRWORA paved the way for the development of a contemporary incentive program. Like much of the rhetoric surrounding the "reinventing government" movement of the 1990s, PRWORA adopted an outcomes-based approach in many of its programs, including child support. While the full-blown incentive program would not be officially sanctioned until 1998, PRWORA implemented audit criteria which would focus on child support outcomes. These incentives provided an additional revenue stream to support the IV-D program and would reward child support programs for good performance (United States Department of Health and Human Services 1996). PRWORA fundamentally changed the way in which human service and welfare agencies conducted their business. While debates continue regarding the success of the welfare reform initiative, researchers have found that the nation's child support program 35 played an important part in reducing the number of individuals receiving TANF benefits under the PRWORA initiatives (Huang et al. 2004; Mead 2005). The Child Support Performance and Incentive Act (1998) In an attempt to bolster the contributions of state child support agencies, policymakers in Congress were quick to adopt the Child Support and Performance Incentive Act (CSPIA) of 1998 to offer state child support programs the ability to earn additional funds for good performance. Undoubtedly influenced by the Government Performance and Review Act (GPRA), the newly proposed child support incentive program moved beyond the scope of simple cost-effective measures and boldly embraced the contemporary movement associated with performance management and carrot-stick incentive structures (Cassetty and Hutson 2005). The CSPIA directly resulted from the PRWORA changes in 1996 which required the Secretary of the Department of Health and Human Services (DHHS), in conjunction with state IV-D directors, to develop a new performance-based incentive system (Gardiner et al. 2003). The CSPIA constructed a new child support incentive system for state IV-D agencies which would be based on performance in four key performance areas in addition to the previously used cost-effectiveness measure. In order to ensure that states did not cheat in reporting their progress on these standards, a comprehensive data reliability audit would be conducted of state systems to ensure that the data maintained and entered into state IV-D computer systems was accurate and complete. In establishing the CSPIA, Congress guaranteed a capped pool of money would be available beginning in federal fiscal year 2000 and continue through 2008. The amount of incentive money each state would receive was granted on the basis of how 36 each state did on the performance measures and was ultimately calculated on a weighted basis in terms of the total amount collected by the IV-D agency (Gardiner et al. 2003). Each state IV-D agency is required to submit annual reports (the OCSE 157, and the OCSE 396) to the federal office which is used as the basis for granting the states' incentive awards. As mentioned previously, the CSPIA created five key performance standards that would be used as the basis for rewarding state IV-D agencies. The standards chosen were a reflection of OCSE's strategic plan and emphasized key areas of the child support program. Additionally, the CSPIA adopted a mechanism through which OCSE could impose penalties on child support agencies for failing to meet minimum standards in these areas of performance. The penalties associated with poor performance resulted in a penalty against the states' grant under Temporary Assistance for Needy Families (TANF). The first year generally resulted in a penalty of one percent of the state's TANF grant and continued up to 5 percent unless the state demonstrated improvement in the problem area. The first performance measure was the Paternity Establishment Percentage (PEP). A fundamental aspect of the child support program revolves around establishing paternity for children who are born to unmarried parents. In developing this standard, OCSE and the nation's IV-D directors agreed to offer each state the opportunity of choosing one of two methods for measuring the PEP score: the IV-D PEP measure or the statewide PEP measure. 37 A state choosing the IV-D paternity standard (45 CFR 305.2(a)(i)) is measured on the number of paternities that are established within cases assigned to the IV-D agency. Under this performance standard, the IV-D agency is only responsible for establishing paternities over which it directly has control. The IV-D PEP measure is determined by the following formula: The Total Number of Children in the IV-D caseload in the Fiscal Year, or at the option of the State, as of the end of the Fiscal Year, who were Born Out-of-Wedlock with Paternity Established or Acknowledged The Total Number of Children in the IV-D Caseload as of the end of the preceding Fiscal Year who were Born Out-of-Wedlock Conversely, if a state chooses to employ the statewide paternity standard (45 CFR 305.2(a)(ii)), the IV-D agency is held responsible for establishing paternities statewide- even for situations where the agency may not have a case. Under this scenario, child support agencies are required to work with birthing facilities in the state to ensure that paternity establishment processes are completed and reported to the states' vital record agencies. The statewide PEP measure is determined by the following formula: The Total Number of Children Born Out-of-Wedlock and for Whom Paternity has not Been Established or Acknowledged During the Fiscal Year The Total Number of Children Born Out-of-Wedlock During the Preceding Fiscal Year The minimum standard that states must maintain, without imposition of a penalty, for both of the PEP standards is a 90 percent score. Gardiner et al. (2003) reported that there appears to be no difference in state performance between the two PEP options; however, they have noted that the IV-D measure seems to be improving more rapidly than is the statewide measure. The second incentive measure monitors the number of cases within the IV-D program that have an actual child support order in place (45 CFR 305.2(a)(2)). Again, 38 this standard is directly related to OCSE's strategic plan and is a necessary element if child support is to be paid to custodial parents on behalf of their children. The minimum performance standard is that 40 percent of cases in the IV-D caseload must have a child support order. This standard is determined by the following formula: The Number of IV-D Cases with Support Orders During the Fiscal Year Total Number of IV-D Cases During the Fiscal Year The percentage of current child support paid (45 CFR 305.2(a)(3)) is the third incentive measure created by the CSPIA. This measurement directly concerns itself with the collection of support that is owed on a monthly basis. Again, this measurement is related to the OCSE's strategic plan to provide regular (i.e., monthly) financial support to families. States are expected to maintain a minimum percentage of 40 percent to avoid penalty. This standard is computed by the following formula: The Number of Dollars Collected for Current Support in IV-D Cases Total Dollars Owed for Current Support in IV-D Cases The fourth measurement relates to the number of past-due collections (45 CFR 305.2(a)(4)) brought in by IV-D agencies. This measurement attempts to capture the amount of collections child support programs are successful in collecting that are overdue or late. This measurement is computed by the following formula: Total Number of IV-D Cases Paying Toward Arrears Total Number of IV-D Cases with Arrears Due 39 Finally, the CSPIA continues the tradition of monitoring IV-D agencies' cost effectiveness ratios (45 CFR 305.2(a)(5)). Simply put, this measure provides a return-on-investment approach that attempts to demonstrate the ratio that comes about by looking at the IV-D agency's expenditures and the amount of collections that are made by that IV-D agency. The measurement is determined by the following formula: Total IV-D Dollars Collected Total IV-D Dollars Expended The federal government requires state IV-D agencies to submit to regular data reliability audits. Using information contained on the OCSE 157 report, auditors monitor state systems for completeness and accuracy. In order to conduct these audits, Gardiner et al. (2003) reported that OCSE requires at least 120 cases from each state to conduct their analysis. Once completed, states must pass their data reliability tests at a 95 percent confidence interval to avoid penalty. Should a state be unable to meet the 95 percent threshold, it is considered exempt from participating in the incentive program for the area in which it failed. In sum, the CSPIA adopted a "carrot and stick" approach to the newly formed child support incentive system. It is important to note that states that perform well on these measures can bring in additional revenues to support their respective programs. The opposite also holds true. States that perform poorly, or that fail to pass the data reliability component, are penalized for their "inadequate" performance. In theory, the child support incentive program provides a mechanism which acts as a constraint upon IV-D agency behavior at the state level and guides state efforts in key areas important to the federal government. 40 The Deficit Reduction Act of 2005 An additional factor relevant to this analysis pertains to the passage of the Deficit Reduction Act (DRA) of 2005. The legislation, which passed by the narrowest of margins, brought serious budgetary cuts to nearly all of the nation's domestic programs. The child support program was not exempt from these budgetary shortfalls. Specifically, the cuts associated with the child support program focused on the child support incentive plan. While the incentive program was left intact, one critical component was removed. When the CSPIA was initially adopted, it was accepted that states would be able to draw down their incentive award, which counts as program revenue for the state. The state was then authorized to draw down federal match dollars (66 percent), on the amount of incentive that was earned by the state. The Deficit Reduction Act ended the practice of drawing down the additional 66 percent match on incentive payments beginning October 1, 2007. The cuts were justified by some on the basis that state practices of receiving federal funds from the incentive and then matching those with the IV-D match rate of 66 percent (also federal funds) constituted another example of states engaging in "double dipping." The American Recovery and Reinvestment Act of 2009 Since the passage of the Deficit Reduction Act, concerted efforts within the child support community have tried to get Congress to reinstate the federal match on incentives earned. The recent passage of the American Recovery and Reinvestment Act (ARRA), or commonly referred to as the "stimulus package" temporarily addressed this issue. The ARRA allows, on a temporary basis, states to match their incentive dollars with the federal match. States can match their earned incentive with federal match funds (66 41 percent) for a two-year timeframe that ends in 2010. After that date, the prohibition of matching funds on incentives will be reinstated. The economic crises of the past several years have resulted in a direct impact on child support programs within the states. While the full impact of these cuts remain to be seen, there is no doubt that state programs who have worked toward improving their performance in the incentive program are the most likely to be hurt under the new funding limitations imposed by the DRA. The temporary reinstatement of federal matching dollars will help, but the cuts that result from this drop in federal funding pose significant challenges for cash-strapped state programs. This chapter has provided details pertaining to the development and implementation of an incentive program for the federally mandated child support program. Thus far, the analysis has focused on the factual and theoretical aspects of this program. The question remains as to how effective the incentive program has been in acting as a constraint on child support agencies and how it has affected street-level child support workers and their day-to-day activities. The following chapters will present empirical results that will shed some additional light on these important questions. CHAPTER 4 RESEARCH DESIGN The previous chapter provided a detailed history of the development of the child support incentive program in the United States. While states have used this incentive program as a means to increase funding for their respective child support programs, little research has examined the influence this program has on street-level child support workers. This dissertation explores the impact of the child support incentive program for front-line child support workers and to further entertain the conditions under which these workers provide child support services to the public. Principal-agent theory provides the theoretical framework from which this analysis is conducted. Research Questions The first research question inquires into the influence the child support incentive program has on child support agencies. Does the incentive program shape individual behavior for street-level child support workers? Do the daily activities child support workers engage in support the key tenets of the incentive system? Do workers consciously complete daily tasks that serve to increase individual and organizational performance in the child support incentive system? If the incentive system acts as a constraint upon individual activities, the second research question delves into the dark side of performance management. Do child 43 support workers feel undue pressure to increase performance in the areas covered by the incentive system? Principal-agency theory suggests that agents (in this case front-line staff) may have some inherent interests to engage in manipulation or to "game the system" in an attempt to improve their standing within their organization. In some instances, these activities are justified as they may benefit the organization in comparison with other child support programs from around the country. In order to answer these questions, child support workers from four different states were surveyed about their daily work activities and their general attitudes about the pressures associated with working in a child support agency. A copy of the survey instrument is found in Appendix B. Because some of the information obtained from the surveys was somewhat delicate, the four selected states remain anonymous and are referred to as: "State A," "State B," "State C," and "State D." Appendices A-1 through A-4 contain state profile information, including performance data and other important characteristics for each state. A brief summation of each of the states' characteristics and performance under the child support incentive program is conducted in the following pages. State Programs Participating in the Study The participating states were selected based upon each state's receipt of federal incentive funds. In order to determine the amount of incentives each state received, an examination of all state programs was conducted using the federal fiscal year 2005 incentive statistics. Each state program was ranked by three factors: 1) incentive dollars per case, 2) incentive dollars per full-time employee (FTE), and 3) incentive dollars as a percent of total collections. Once each of the three rankings was figured, an average 44 score was computed for each state. After reviewing the list of states, four state programs agreed to allow their employees to participate in the survey. Table 4.1 provides the ranking data used to select the states. Based on the criteria used, State "A" is considered a "high incentive state," State "B," a "medium-high" incentive state, State "C," a "medium-low" incentive state, and State "D," a "low" incentive state. All four of the state programs are state administered. No county-based programs agreed to participate in this study. The four states varied in size. For instance, the number of FTEs ranged from 247 to over 1,600 for one of the states. Not surprisingly, there was a wide range in terms of the number of cases each state was assigned (ranging from a low of 45,000 to a high of 377,000) and the amount of dollars collected (ranging from a low of $94 million to a high of $613 million) for FFY 2005. Each state program participating in this study agreed to allow their employees to participate. The survey was administered between the months of October and December 2007 to over 1,300 child support workers in the four programs. While response rates varied from state to state, the total response was 54.5 percent. The following pages provide a brief description of the four state programs that participated in this study. Table 4.1 State incentive rankings State Incentives by Incentive Incentive by Average Caseload by FTE Collections Rank A 7 12 11 10.000 B 17 14 4 11.667 C 26 13 21 20.000 D 25 39 22 28.667 45 State "A" State A was the smallest of the child support programs that participated in this study. State A had a relatively small number of employees (247) and a modest caseload consisting of about 45,000 cases. Based on the report data, it appeared that the number of FTEs and the number of cases remained relatively constant throughout the years. With that said, total collections for State A dropped in FFY 2006 for the first time since the inception of the child support incentive system in 1999. The performance measures for State A demonstrated good and improving performance in the five incentive areas. The only potential blemish in State A's performance history pertained to the state's Paternity Establishment Percentage (PEP) score. State A dropped below the 90 percent threshold in 2001 but has shown constant improvement since that time. It appears that the improvement was enough to avoid federal financial penalties to the state. State A represented the lowest amount of participation in the survey of the states participating in the study. Only 27 members of State A's staff responded to the survey. While it was nearly impossible to determine the response rate to the surveys (repeated efforts were made to contact the state about the number of individuals given the survey, but no response could be obtained), the response rate was estimated to be somewhere around 20 percent. State "B" State B was the most unique of the four states participating in this study as it was the only one that had a couple of counties which elected to manage their own child 46 support program. Those surveyed for this study worked for the state program and not the counties'. Of 185 total child support workers, 100 responded to the survey (resulting in a 54 percent response rate). State B's total number of child support employees (including state and county) has remained constant, albeit with a slight upward trend. The state's reports also demonstrated a consistent drop in the number of cases since FFY 1999. The state's performance has been consistent as well. Like State A, State B has experienced problems attaining its PEP score. Unlike all of the other states, however, State B used the IV-D measurement instead of the statewide measure. The state's woes in this area began in FFY 2000 when the state earned a PEP score of 61.5 percent, well below the 90 percent threshold. In addition to the poor number, the state failed its data reliability audits in the paternity establishment area. While the state improved its score enough in years 2000-2001 to avoid any further financial penalties, it did not achieve the 90 percent level until 2003. The data show constant improvement in each subsequent year. In addition to the data reliability failure in the PEP score, State B also failed a data reliability audit for payments toward arrears in 2001. Failing the data reliability review results in the state's inability to participate in incentive funding for that particular area. In sum, it appeared that State B received some penalties toward its incentive, although it is important to recognize that these losses occurred several years ago. State "C" In State C, 530 employees received the survey and 369 responded to it. The response rate of 69.7 percent was the highest among the states participating in this study. 47 State C reported 650 FTEs in FFY 2006, which represents a significant increase over past years. In addition, State C has experienced a significant increase in its caseload since 1999. Likewise, State C has seen increases in its total collections. In terms of performance, State C has shown constant improvement in its five incentive areas. Like States A and B, State C has also struggled achieving its PEP score. In FFY 2000, State C's PEP score was 81.4 percent. The state improved this number to 86.3 percent the following year, but it dropped again (80.7 percent) in FFY 2002. It is unclear at this time if State C received a financial penalty for this failure to meet the 90 percent performance level. Again, the incentive problems experienced by State C are beginning to fade due to the length of time in which the infractions occurred. State "D" State D also reported high response rates for the survey. Four hundred and thirty two potential respondents were given the survey and 263 responded for a response rate of 60.9 percent State D appeared to have experienced some dramatic decreases in the number of FTE's in FFY 2006 (930 down from 1, 041 in FFY 2005). The caseload in State D has experienced a steady downward trend, decreasing from a high of 390,000 cases in FFY 2002 to nearly 368,000 in FFY 2006. While caseloads and staff have dropped, State D has experienced increases in collections culminating in an eight-year high of almost $550 million in FFY 2006. State D follows a pattern similar to that of the other states in this study. Its performance has steadily increased, but once again, the state has experienced some problems with the PEP score. Initially, State D reported a PEP score of over 91 percent 48 in 2000. A year later, however, the state reported a score of 86.70 percent followed the next year by a 79.7 score in 2002. The 2002 score should have resulted in a financial penalty to the state. FFYs 2003 and 2004 showed modest increases in State D's PEP score culminating in the state achieving 92.5 percent in FFY 2005. The state has remained above the 90 percent threshold ever since. In summary, each of the states participating in this dissertation exhibits some very clear similarities. Each has experienced some historic problems in achieving the minimum standards for the PEP score, although these are probably starting to fade in the institutional memory of each of the programs and its staff. In all of the other areas, the states have all maintained or improved upon each of the five incentive measurements. Along with those similarities, there also appeared to be some dramatic departures between the states. States B and C have both experienced significant staffing increases, while State A has remained steady and State D actually dropped. In addition, State C was the only state to see significant increases in its caseload. All of the other states experienced decreases in the number of cases assigned to the child support program. Finally, States B, C, and D all experienced increases in collections. State A experienced a drop in collections in FFY 2006, which dramatically departs from collections experienced in the other states observed in this study. Hypotheses Based upon the construction of the dependency on incentives (see Table 4.1), hypotheses can be constructed to test the underlying research questions posed by this study. The first of these hypotheses dealt directly with the issues of the federal incentive amounts each state receives. It was reasonable to expect that workers from high 49 incentive states, who theoretically are more dependent upon incentive dollars, would identify tasks pertaining to the achievement of performance goals at higher rates than lower incentive states. Therefore, I hypothesize that: H1: A>B>C>D where > indicated state workers having more knowledge of the federal child support incentive and its correspondingly acted as a constraint on worker tasks: A second hypothesis, more central to the overall goal of this research was also constructed regarding the specific issue of organizational cheating by child support workers. I hypothesized that the more dependent a state was on incentive dollars, the more likely incidents of organizational cheating would be. Therefore, it was reasonable to hypothesize: H2: A>B>C>D where > indicated states having more incidents of organizational cheating. Survey Construction In order to gather the data necessary for this study, permission was obtained from each selected state's IV-D director to interview front-line staff regarding their knowledge of the federal incentive system. Specific questions were tailored to measure the extent that day-to-day work was constrained by the federal incentive program. The survey, found in Appendix B, explored child support workers' opinions related to the pressures they may or may not feel to improve performance in the areas identified in the incentive 50 program. The survey was administered to all child support workers in each state program that agreed to participate in the study. The introduction of the survey explained that the purpose of the study was to "research the attitudes, beliefs, and practices" of child support workers from around the country. In addition, the introduction explained that workers' responses would remain anonymous, and they were free not to participate in the study. It also explained that their response to the survey served as their consent to participate in the study. The introductory statement emphasized that workers' responses were important to improving the nation's child support program and would bolster survey response rates. The questionnaire was delivered to recipients using an online survey source. This method of data gathering posed a number of distinct advantages over traditional (paper-based) survey administration. First, the online survey provided a simple way to deliver the survey to a relatively large number of potential survey respondents. More importantly, all of the child support workers who participated in the survey had access to the Internet to accomplish their job duties. As a result, delivering the survey through online means made sense since all of the potential survey respondents had the means to participate, if they wished, and none were excluded because they were unable to access the survey instrument. Finally, the online survey was a reasonably inexpensive method to administer the survey to a large and geographically diverse number of individuals. The online survey was designed to allow respondents to skip questions and each question provided respondents the opportunity to respond with a "don't know/choose not to answer" type of response. If a respondent chose not to respond to a question, they 51 were allowed to continue with the survey and to complete as many or as few of the questions as they wished. The survey began by asking respondents some generally simple questions (Questions 1-4) regarding their work history and experience within the child support program. These questions were asked early in the survey to assure the respondents that the questions discussed pertained to their jobs and were relatively simple to answer. The next series of questions (Questions 5-9) inquired into the level of knowledge front-line employees have about the child support incentive system. The responses to these questions allowed for further analysis of worker compliance within the overall incentive structure and acted as an independent variable concerning the second research question. Incidents of organizational cheating, because of their nature, are not readily observable. With that said, one of the primary purposes of this study was to gauge the degree of organizational cheating that occurs within child support agencies. The measures established for organizational cheating in this study (Questions 13-15) acted as the dependent variable for the purposes of analysis. These questions, by design, attempted to measure the degree of organizational cheating from a couple of different perspectives. For example, Question #13 attempted to measure the degree to which organizations developed specific strategies to improve performance in key areas. Respondents were asked to indicate their opinion to the following question: "My organization spends a lot of time developing strategies to increase performance in key areas." This question was quite subtle, but it attempted to identify how workers viewed 52 organizational attempts to improve performance. It is important to note that the responses to this question did not directly ask about the issue of cheating. Rather, it was an attempt to measure the level of organizational influence from a worker's perspective. Question #14 was more pointed. It asked respondents to identify if "It is possible to take actions in my caseload that will artificially inflate my individual performance levels." Again, this question, by design, did not imply worker culpability. Rather, it simply inquired into the possibilities of cheating within their workplace. Question #15 asked respondents to answer: "I am aware of others in my organization who take actions to artificially inflate their individual performance levels." Again, I did not ask respondents whether they specifically engaged in such activities, but the question attempted to focus workers on what was actually going on in terms of cheating within the organization while not specifically implicating themselves. Finally, Question #16 bluntly asked workers if "In the past year, I have felt pressure from someone in my agency to manipulate my caseload to improve performance." Again, workers did not indicate if they succumbed to these supervisory demands, only if they felt this type of pressure from their managers. The responses to these four questions provided the foundation for this research. They are important elements in determining the degree of organizational cheating that may, or may not, exist within the child support agencies participating in this study. A primary goal of this study was to find a remedy to offset the likelihood of organizational cheating. In order to accomplish this goal, survey questions were designed to measure a number of additional variables that may help identify factors that led to increases in organizational cheating. 53 Bohte and Meier (2000) identified a number of variables that potentially acted as motivation for organizations to engage in organizational cheating. One of these variables was poor organizational performance. Poor performance may provide justification for some to engage in questionable activities to improve performance. Poor performance may be determined by an agency's inability to meet minimum performance standards or in a decline of performance over time. The performance data gathered from each of the states determined if performance problems existed in the past, and if these performance woes acted as a precursor for propensities to engage in cheating. Additionally, Bohte and Meier suggested the likelihood of organizational cheating increased when task demands at the individual, team, and organizational levels were overwhelming. Peterson et al. (1995) developed a validated scale that measured a number of role-related items. These scales were used in the survey instrument for this research. The first of these scales measured the worker's sense of work overload and stress (Questions 21-25 of the survey). In theory, if workers felt extreme pressure because of overwhelming workloads, they might be more likely to engage in questionable behaviors to overstate their individual performance. The survey posed several questions about the degree to which workers felt capable of accomplishing the necessary daily tasks that would result in successful job performance. In a similar vein, if task demands were such that workers felt pressure to manipulate outcomes, a common strategy was to engage in "creaming" or selecting easy cases to work (Blau 1955; Wilson 1989). Working cases that are more likely to result in payments may be a strategic device workers employ in order to meet the demands managers and the organization place on them to perform. Working "easier" cases in lieu 54 of the more difficult ones was a mild form of cheating but was one that may be relatively common among caseworkers. Survey questions (Questions 11-13) inquire about the emphasis workers place on finding effective strategies to improve performance. The survey also asked child support workers if they felt others monitored their work product within their respective organizations. The survey instrument adopted a validated scale developed by Oldham and Cummings (1996) which measured worker's perceptions of organizational supervision. This scale was created by combining questions 17, 34, 35, 36, and 37 in the survey. These questions asked a battery of questions focused on the individual worker (or agent) in relation to their supervisors (or principals). As principal-agency theory suggests, it is imperative for principals to implement processes to monitor the behavior of agents. As demonstrated in Chapter 3, where monitoring is considered to be inadequate, principal-agent theory suggests that agents will pursue their own self-interests. While all forms of cheating are probably viewed in negative terms by organizations, the degree to which performance is monitored likely varies between and among child support agencies. Workers may be more willing to engage in forms of cheating if they felt that monitoring was done haphazardly. Therefore, the survey attempted to assess employee attitudes concerning how closely their work product was monitored. A number of additional independent variables were measured by the survey. First, the instrument measured the degree of employee job satisfaction (Questions 38-39), by using a scale developed by Cammann et al. (1983). Notions of role ambiguity (Questions 25-29) and role conflict (Questions 30-32) were investigated by using scaled questions developed by Peterson et al. (1995). Each of these variables was important, 55 because they provided insights into the clarity child support workers had in regard to their job. The concept of role conflict was one that could be potentially important as the possibility of receiving competing task demands might have impeded individual decision-making for child support workers. Finally, the survey explored how well employees understood how their employer measures performance. Williams and Levy (1992) developed a series of questions that measured employee understanding of performance appraisal systems (Questions 41-44). The survey concluded by asking some additional general demographic questions (Questions 45-50). Survey respondents were also given an opportunity at the end of the questionnaire to answer an open-ended question that asked, "Is there anything else you would like to share regarding your work performance and experience?" The survey tool described in this chapter was the basis for the analysis that follows. The questions were designed for specific purposes that would allow further exploration of the research questions which lie at the heart of this study. The following chapters focus on the degree to which child support workers understand the incentive system structure, and whether they were willing to engage in "cheating" behaviors to enhance their individual and organizational performance. CHAPTER 5 MEASURING THE KNOWLEDGE OF THE CHILD SUPPORT INCENTIVE PROGRAM BY FRONT-LINE WORKERS This research investigates the impact that incentive systems have on incidents of organizational cheating in the substantive area of child support enforcement. In order to gather relevant data, I interviewed child support professionals from four state level programs using a battery of questions pertaining to their workplace. Some of the questions posed to workers related to their individual knowledge of the child support incentive system. This chapter seeks to arrive at a conclusion concerning the individual knowledge and understanding of the child support incentive system by front-line child support workers. State child support programs can earn additional funding for performing well in five specific categories (see Chapter 3 for more detail). It is reasonable to hypothesize that state workers coming from "high" incentive states may have more knowledge of the incentive program than those working in "low" incentive states. This hypothesis is: H1: A>B>C>D A couple of different methods could be used to arrive at an answer to the question of how knowledgeable staff is about the incentive plan. The simplest way was to ask 57 them about it. With that said, it may not have been reasonable for front-line staff to know specifically whether an incentive program existed. Given that line of reasoning, the survey instrument measured specific job-related tasks that correlated with four of the five incentive measures. In order to establish front-line worker knowledge of the incentive program, child support workers were asked three specific questions pertaining to their organization's involvement in an incentive program. The first question asked respondents whether their organization received additional funding for "outstanding organizational performance." The overwhelming response from all workers, as presented in Table 5.1, was "I don't know." The responses to this question demonstrated an overall low level of worker knowledge that their programs could earn additional funding for performance. While responses from State A (the high incentive state) suggested that a third of respondents felt their program could earn additional funding, two-thirds said either their state did not earn additional funds, or they did not know if additional funds could be earned. Conversely, State D (the low incentive state) had the second highest correct response (21.07%) to the question. Given the responses to this question, it appeared there is quite a bit of uncertainty among respondent state workers about their organization's ability to earn incentive dollars. Over half of all respondents indicated that they did not know if additional funds were available and almost 28 percent incorrectly stated that their program could not earn incentive dollars for performance. A second question asked survey respondents about the possibility of penalties for 58 Table 5.1. Responses of state's ability to receive incentive funding State Yes No Don't Know Total A 9 33.33% 5 18.52% 13 48.15% 27 100% B 10 10.99% 45 49.45% 36 39.56% 91 100% C 69 19.60% 93 26.42% 190 53.98% 352 100% D 51 21.07% 56 23.14% 135 55.79% 242 100% Total 139 19.52% 199 27.95% 374 52.53% 712 100% poor performance. The question inquired about the possibility of penalties in the event their organization failed "to perform in certain areas." When couched within these "negative terms," the overwhelming response from respondents was that their agency could be penalized. The survey results for this question are presented in Table 5.2. Over three-fourths of all of those surveyed (75.89%) indicated that it was possible for their organization to be penalized for poor performance. Only 2 percent said that their organizations were not susceptible to financial penalties, while a significant portion (21.84%) said that they did not know if their organization was subject to penalties. This pattern holds true when the data are examined at the state level. The only minor exception to these findings was in the response-set from State B participants. Fewer respondents indicated that their state could be penalized (66.30%), while more said their state could not be penalized (6.52%). In addition, a higher proportion of State B respondents indicated that they did not know if their agency could be penalized (27.17%) than in the other participating states. Substantively, more respondents in all states recognized that their agencies could be punished financially for poor performance than knowing that their agency could earn additional incentive funding for meeting, or exceeding goals. The responses suggested that the majority of respondents from all states seemed confident that negative incentives 59 Table 5.2. Responses for possibility of penalties for poor performance. State Yes No Don't Know Total A 21 77.78% 0 0.00% 6 22.22% 27 100% B 61 66.30% 6 6.52% 25 27.17% 92 100% C 265 76.59% 7 2.02% 74 21.39% 346 100% D 188 78.33% 3 1.25% 49 20.42% 240 100% Total 535 75.89% 16 2.27% 154 21.84% 705 100% could result for poor performance, but were unaware of the reality that IV-D child support programs were capable of earning extra monies for good performance. A third question asked respondents if their state had ever experienced financial penalties for poor performance. As Chapter 3 detailed, the IV-D child support program provides several mechanisms through which states can be penalized for not meeting performance standards in the five incentive areas. Once again, the data demonstrated that most respondents did not know if their state has ever been penalized. The results for this question are displayed in Table 5.3. The responses to this question suggested a lot of ambiguity existed among workers concerning their understanding of the possibility of penalties being levied against state child support programs. Three-fourths of the respondents indicated that they did not know if their program received any financial penalties. These results were remarkably consistent among the respondents from each state. In order to assess the accuracy of this question, the responses need to be compared to any instances where the participating state may have experienced actual financial penalties for failing to meet incentive standards. While "official" data could not be obtained from the federal Office of Child Support Enforcement regarding this matter, the data presented in the Appendices regarding each state's performance levels suggested 60 Table 5.3. Responses of whether organizations have been penalized. State Yes No Don't Know Total A 2 7.41% 5 18.52% 20 74.07% 27 100% B 13 14.13% 12 13.04% 67 72.83% 92 100% C 56 16.09% 27 7.76% 265 76.15% 348 100% D 18 7.47% 44 18.26% 179 74.27% 241 100% Total 89 12.57% 88 12.43% 531 75.00% 708 100% that in only one case, State D, was there ever a likelihood of financial penalties being levied against the state. Once again, the survey results indicated a general finding of low worker knowledge about the incentive system. One possibility for this low-level understanding of the incentive system by workers may have been that workers did not identify with an "incentive system" but would identify work tasks that directly pertain to that system. As a result, a series of questions in the survey asked respondents to rank eleven task-oriented items, five of which were the actual incentive items, on a scale of one to ten, with one representing a low priority for the agency and ten being high. The purpose of these questions was to measure the degree of importance front-line workers assigned to incentive and nonincentive tasks. The results of the analysis suggested that front-line workers hardly distinguished between incentive and nonincentive tasks. This finding provides further evidence that front-line workers do not understand the federal incentive system. These data are found in Table 5.4. In order to determine the overall understanding of front-line workers of the performance incentive items, an additive scale was created to compare the knowledge of incentive and nonincentive work tasks. These scales were created by adding the sum of all worker responses to the five incentive and six nonincentive tasks. The range for 61 Table 5.4. Responses comparing incentive and nonincentive tasks. Incentive Items Nonincentive Items Paternity Establishment IV-A Collections State Mean Median Standard Deviation State Mean Median Standard Deviation A 10.11 11 2.19 A 10.56 11 1.121 B 9.88 11 2.524 B 8.75 10 3.156 C 8.85 11 3.332 C 8.53 10 3.392 D 9.99 11 2.2 D 8.86 10 2.971 Order Establishment Customer Responsiveness State Mean Median Standard Deviation State Mean Median Standard Deviation A 9.93 11 2.218 A 10.2 11 1.581 B 10.05 11 2.279 B 9.82 11 1.986 C 8.97 11 3.266 C 9.68 11 2.157 D 10.19 11 1.862 D 10.03 11 1.859 Current Support Cost Controls State Mean Median Standard Deviation State Mean Median Standard Deviation A 10.37 11 1.944 A 7.24 9 3.876 B 10.23 11 1.955 B 6.49 8 4.166 C 9.68 11 2.368 C 6.87 8 4.032 D 10.07 11 1.866 D 7.55 9 3.614 Past Due Collections Caseload Size State Mean Median Standard Deviation State Mean Median Standard Deviation A 10.15 11 2.013 A 6.64 7 3.451 B 9.62 11 2.340 B 5.7 5 3.652 C 9.05 10 2.610 C 5.6 5 3.631 D 9.55 10 2.211 D 6.35 6 3.653 Cost Effectiveness Total Collections State Mean Median Standard Deviation State Mean Median Standard Deviation A 7.15 8 3.613 A 10.44 11 2.022 B 6.43 7.5 4.000 B 10.16 11 2.089 C 6.57 7 3.825 C 9.33 11 2.777 D 7.78 9 3.536 D 9.91 11 2.145 Federal Timeframes State Mean Median Standard Deviation A 10.44 11 0.961 B 9.62 11 2.557 C 9.19 11 2.868 D 9 10 3.122 62 incentive items was 5-55 and nonincentive items ranged from 6-66. Once the additive value was constructed, a standardized score was figured by dividing the incentive scale score by 5 (the number of incentive tasks) and the nonincentive scale by 6 (the number of nonincentive tasks). The results of this computation are presented in Table 5.5. At first glance, the construction of these scale items demonstrated a slight difference in the way front-line workers ranked incentive and nonincentive work tasks in importance. The results clarified that most workers slightly identified incentive tasks as being more important than the nonincentive tasks. These findings tended to support the general finding that most workers demonstrated little knowledge of the incentive system. It is also important to note that there was also very little difference between how the high incentive and low incentive state workers ranked these items. In looking at the individual items that made up the scale, it became apparent that caseworkers did not feel it was important for their organizations to keep caseload size to a manageable level. This was the lowest of all of the ranked items and was not surprising given some of the comments survey respondents attached to their surveys. In an effort to create a more balanced measure, a second incentive scale was created to remove the caseload question from the analysis. These results, shown in Table 5.6, demonstrated an Table 5.5. Standardized measures for incentive and nonincentive tasks. Incentive Items (Range 1-11) Nonincentive Items (Range 1-11) State Mean Median Standard Deviation State Mean Median Standard Deviation A 9.541 10 1.678 A 9.253 9.667 1.503 B 9.236 9.7 2.030 B 8.424 8.667 1.993 C 8.619 9.1 2.337 C 8.199 8.500 2.076 D 9.516 10.2 1.831 D 8.617 9.000 2.004 63 Table 5.6. Measures for incentive and nonincentive tasks (excludes caseload size variable). Incentive Items (Range 5-55) Adapted Non Incentive Scale (Range: 5-55) State Mean Median Standard Deviation State Mean Median Standard Deviation A 47.7037 50 8.389 A 48.88 51 6.49564 B 46.1778 48.5 1.015 B 44.8409 46 1.01 C 43.0936 45.5 1.169 C 43.5917 45 1.05 D 47.5809 51 9.153 D 45.3872 48 9.451 almost indecipherable difference between the incentive and nonincentive task scales. In other words, child support workers from all four state programs generally identified nearly all of the tasks as being "highly" important for their organizations. In sum, the results from this analysis do not support the hypothesis that workers from high incentive states would rank incentive tasks higher than nonincentive tasks. As the data show, workers, regardless of their state performance on incentives, ranked both categories nearly equally. Couple this finding with the results from the questions that inquired into the ability for state programs to earn additional funding for high performance; it was readily apparent that most child support workers do not have a general, let alone a detailed, understanding of the child support incentive program. This finding was surprising given how state programs that perform at high levels have become dependent upon these additional incentive dollars to bolster their programs. Principal-agency theory suggests it is rational for organizations to emphasize key performance areas, particularly where good performance in those areas can result in positive (in this case financial) outcomes for the organization. In the four participating states, survey respondents did not rank incentive tasks in the way principal-agency theory would suggest. While organizations might rely heavily upon these dollars, it does not 64 appear that the states surveyed for this project communicated those needs to the street-level worker. CHAPTER 6 ASSSESSING ORGANIZATIONAL CHEATING One of the key purposes of this study was to identify whether incidents of organizational cheating were higher among states drawing more incentive dollars. The underlying theory suggests that state programs which draw more performance-based incentive dollars might be more inclined to engage in cheating-types of activities in order to facilitate the continued increase of incentive dollars to their programs. According to principal-agency theory, agents might be willing, in some situations, to bend the rules in order to maximize their own profitability. In order to assess this important question, survey respondents were asked a battery of questions pertaining to their behaviors concerning cheating. Careful attention was paid to construct questions in a way that would answer the question without posing any risk to the survey participants. The survey questionnaire and research methods were consistent with the policies of the University of Utah's Institutional Review Board (IRB) and received IRB approval prior to gathering any data. Before addressing the questions related to cheating, it was important to assess other activities that were perceived to be related to the issue of organizational cheating. First, survey respondents were asked how they felt their own individual performance impacted overall organizational performance. It seemed reasonable to expect higher 66 levels of self-reported incidents of individual contributions in high incentive states compared with low incentive states. This expectation was not met in this study. Table 6.1 demonstrates that the vast majority of survey respondents (regardless of their state's level of incentive dollars) felt that their performance had an impact on their organization's performance. Ninety percent of respondent's either "strongly agreed" or "agreed" that their individual performance mattered in terms of organizational performance. There appeared to be wide consensus among respondents that individuals play an important part of organizational performance. A second series of questions inquired into the role strategy played in improving performance. In terms of organizational cheating, it seemed reasonable to expect that state programs that spent more time strategizing might find themselves with higher instances of organizational cheating. Two questions were asked about the degree to which workers strategize to improve performance with co-workers and supervisors. An additional question asked survey respondents to rate whether their organization spent a lot of time trying to develop strategies to improve performance. Using these questions as a foundation, an additive scale was constructed to measure the total amount of strategizing going on within the organizations. Again, I hypothesized that high incentive states would strategize at higher levels than low incentive states. Table 6.1. Responses for impact of personal performance on organizational performance. State Strongly Agree Agree Neither Agree or Disagree Disagree Strongly Disagree Total A 13 48.15% 13 48.15% 1 3.70% 0 0.00% 0 0.00% 27 B 46 50.55% 38 41.76% 6 6.59% 1 1.10% 0 0.00% 91 C 116 33.72% 179 52.03% 29 8.43% 12 3.49% 8 2.33% 344 D 103 44.02% 120 51.28% 9 3.85% 1 0.43% 1 0.43% 234 Total 278 39.94% 350 50.29% 45 6.47% 14 2.01% 9 1.29% 696 67 Overall, survey respondents indicated moderate levels of strategizing with co-workers, but higher levels with their supervisors. These findings were consistent at the state level as well. As Tables 6.2 and 6.3 indicate, workers said they either strongly agreed or agr