State welfare-reform impacts: content and enforcement effects.
- U.S. Congressman Clay Shaw, September 1996
The Personal Responsibility and Work Opportunity Act (PRWOA) of 1996 has spawned an array of competing claims about the efficacy of welfare reform. Politicians at every level of government have credited reform legislation for the precipitous drop in welfare caseloads that began in 1994. The Council of Economic Advisers has even claimed that the caseload declines that occurred as much as a year before the PRWOA was signed were an anticipatory response to pending program changes (Council of Economic Advisers, 1997). Skeptics (e.g., Martini and Wiseman, 1997) have questioned both the theoretical and empirical foundation for such claims.
The resolution of competing claims has been impeded by the misspecification of program parameters in econometric models. Specifically, virtually all economic models of program impact assume an interjurisdictional homogeneity in program content, timing, and enforcement that simply does not exist. In doing so, economists are ignoring a storehouse of public administration and public policy literature that documents the variation in interjurisdictional program content created in the implementation process at subnational levels that extend all the way to "street-level bureaucracies" (Lipsky, 1980). The model misspecifications that result generate inefficient and inconsistent impact estimates - the fodder of the continuing controversy.
Neither the debate over program efficacy nor the underlying specification problem is unique to the 1996 reform legislation. The same controversy has arisen over the efficacy of earlier state-level reforms. From 1989 through 1995, individual states obtained a spate of "waivers" from the U.S. Department of Health and Human Services (HHS) to modify the parameters of the Aid for Families with Dependent Children (AFDC) program. Analysts have yet to reach a consensus on whether or how these state-initiated "demonstrations" affected welfare caseloads. That uncertainty is also due in part to the incomplete specification of program parameters.
This paper attempts to refine estimates of program impact in two ways: first, by incorporating more of the explicit interjurisdictional variability in program parameters into the evaluation model, and second, by recognizing the implicit variation across jurisdictions caused by local discretion in program implementation. These empirical innovations are more demonstrative than definitive. However, they may help pave the way toward more comprehensive evaluation models and ultimately more precise estimates of program impacts.
II. EXPLICIT VARIATION IN STATE WELFARE DEMONSTRATIONS
The Family Support Act (FSA) of 1988 mandated structural reforms in the AFDC program and a timetable for their implementation. The most important elements of that reform legislation were (1) the extension of AFDC eligibility to two-parent families (AFDC-U) in all states, (2) the establishment of mandatory self-help activity for an increasing portion of the AFDC caseload, and (3) the creation of transitional eligibility for in-kind benefits for welfare recipients who begin paid employment. The AFDC-U expansion fostered a potential short-run rise in AFDC caseloads, while the second and third features were expected to reduce caseloads over time. This was particularly true for the self-help mandate, which set rising participation thresholds in all states during the implementation period (1990 - 1995 for AFDC, 1990-1997 for AFDC-U).
The FSA also established the authority of the HHS to grant waivers to states seeking to modify the provisions of that act. This strategy was based on similar provisions introduced in the Omnibus Budget Reconciliation Act of 1981, which also both set new national welfare regulations and encouraged state-level experimentation. Once again, the provision for state experimentation spawned a surge of state welfare-reform initiatives. By the middle of 1996, 42 states had been granted administrative waivers, some of them more than once.
The waiver process spawned an exceptionally diverse set of welfare reforms. Table 1 displays the incidence of 19 of the most common waiver provisions. Some of these provisions (e.g., the relaxation of AFDC-U eligibility rules, the liberalization of income disregards) broadened welfare eligibility, while other provisions (e.g., narrowed criteria for Job Opportunities in the Business Sector [JOBS] participation exemptions, stricter sanction rules, and time limits on benefit eligibility) were designed to limit welfare demand or supply. This maze of reform provisions is further complicated by the tendency of states to introduce multiple provisions simultaneously and/or sequentially (e.g., the much publicized Wisconsin reforms encompassed nine separate waivers and at least 18 different program features; a listing by state is available from the author).
The enormous variability in state-level program features obviously precludes any generalization about the effects of a standardized program model. There are more statutory combinations than states, thus precluding the statistical isolation of parameter-specific effects. Some aggregation of these provisions is required for statistical analysis. For this purpose, state reform initiatives are here classified by their presumed "toughness," i.e., the degree to which they limit welfare access. Salient access-limiting provisions include (1) time limits on benefit eligibility, (2) family caps, and (3) narrowed exemption criteria for JOBS participation. Although none of these provisions directly reduces AFDC caseloads, they all diminish the expected utility of welfare recipiency. They may also signal a change in public attitudes toward welfare.1
A dichotomous characterization is used to distinguish "tough" and "soft" state-level reforms. "Tough" reforms include at least two of the access-limiting features noted above; all other reform initiatives are characterized as "soft." The tough/soft dichotomy is useful for distinguishing reforms that emphasized limits on welfare access. Even states that adopted such provisions, however, often introduced access-widening [TABULAR DATA FOR TABLE 1 OMITTED] provisions at the same time, often as part of a political compromise. Hence, the tough/soft dichotomy still obscures significant variation in program features, therefore limiting the power of the statistical tests.(2)
As noted above, several states had multiple waivers during the FSA implementation period. Accordingly, the characterization of a state's welfare initiative can change from one year to another. During the observation period, sequential waivers consistently narrowed welfare access; no states converted from "tough" to "soft" reforms.
During the observation period, states were also differentially affected by FSA's extension of AFDC-U eligibility. Prior to FSA, 29 states already had an AFDC-U program and thus were unaffected by this new FSA stricture. The remaining states had the option of limiting AFDC-U eligibility to 6 months per year during the implementation phase. Thirteen of the non-AFDC-U states elected this option.
The first objective of the present paper is to determine whether these myriad state welfare initiatives had any discernible impact on AFDC caseloads. The central hypothesis is that state-initiated reforms - particularly tough reforms - reduce caseload growth. To test this hypothesis, a pooled time-series, cross-section regression is conducted to identify significant correlations between annual caseload growth and specific features of state reform initiatives. The general model underlying this test postulates that annual variations in AFDC caseloads (W) are the outcome of state-specific changes in economic circumstances (E), program parameters (P), and reform initiatives (R):
(1) [Mathematical Expression Omitted]
In this formulation, R refers to waiver-specific reform elements. Changes in state program parameters not subject to federal approval (e.g., benefit levels) are expressed in the vector P. To the extent that economic and programmatic influences are controlled, this model should identify the caseload change uniquely attributable to the specific features of state reform initiatives.
III. IMPLICIT VARIATION IN STATE PROGRAMS
The statutory features of state welfare reform legislation are not the sole determents of welfare reality. A substantial body of political science, public administration, and public choice research has demonstrated how formal program features are altered in the implementation process. Pioneering studies by Berman (1978), Brodkin (1986, 1990), Elmore (1979), Handler (1986, 1992), and Lipsky (1980) all emphasize how bureaucratic discretion in the implementation of social programs substantially alters their content and ultimately their effectiveness. Lipsky goes so far as to suggest that the "street-level bureaucrats" who interact with clients are the ultimate arbiters of public policy. Although policy analysts may find it expeditious to assume that the implementation process is "logical, rational, well-ordered, and hierarchical," such an assumption is empirically unsupportable (Elmore, 1979). As a result, analysts often end up trying to evaluate the efficacy of program designs that no one has implemented.
There are numerous studies documenting the scope, nature, and motivations for local discretion in the implementation of welfare reforms (e.g., Brodkin, 1986, 1997; Brock, 1992; Hasenfeld and Steinmetz, 1981; Kane and Bane, 1994; Meyers et al., 1998; Palumbo and Calista, 1990; Petersen, 1996; Prottas, 1979, 1981; Ricco and Hasenfeld, 1996; Schiller, 1996; Snyder, 1992). Several of these authors emphasize the critical role of local caseworker discretion in the determination of welfare eligibility and the imposition of sanctions.
Public choice theory also plays a role in local responses to state mandates. Specifically, local administrative discretion may alter not only the timing but also the substance of welfare reform (Nathan, 1996). Schiller (1996) documented that such local discretion had an independent effect on the caseload effects of federal (FSA) reforms. The impact of state welfare reform initiatives are likely to be constrained by the same institutional features (Brock, 1992; Snyder, 1992). Because local welfare administrators tend to be more liberal than state bureaucrats (Piskulich, 1993; Payne, 1996), they are likely to oppose state-initiated reforms that limit welfare access or increase the utilitarian cost of recipiency. That opposition may manifest in such passive resistance as delayed or desultory enforcement of legislated rules. The ultimate power of local caseworkers to approve welfare applications, to exempt recipients from work requirements, to verify continuing eligibility, and to impose sanctions gives them substantial discretion to alter caseload flows (Dattallo, 1994; Mead, 1996). Local welfare departments are also responsible for generating baseline data on program activities and outcomes and are thereby in a position to manipulate perceptions of program effectiveness (Snyder, 1992).
The demonstrable influence of local administrative discretion on program content and outcomes requires a respecification of the conventional evaluation model. As Schiller (1996) demonstrated, the "naive" model (equation  above) can be augmented with a vector of"discretionary" variables that reflects the degree of symmetry between state-legislated designs and locally implemented programs. Such a "discretionary" vector effectively recalibrates the major elements of welfare reform initiatives to their institutional reality. This adjustment permits more precise estimation of the effectiveness of fully implemented reform features.
Among the potential elements of the discretionary vector [A.sub.i,t] are (1) the local approval rate for welfare applicants, (2) the exemption rate (from mandatory self-help activities) for AFDC adults, (3) the participation rate in self-help activities (especially a "full-time" equivalence rate), and (4) the assignment rate across program activity options. The institutional structure of the AFDC program imbues local caseworkers and administrators with substantial discretion to alter these various indices of program implementation. This being the case, higher approval and exemption rates could be signals of local empathy with the poor and/or local hostility to tougher work-oriented reforms. Lower (hours-adjusted) participation rates would send the same signal, as would low assignment rates to work-oriented program options. In this expanded, "discretionary" model, caseload changes are determined by
(2) [Mathematical Expression Omitted],
where the discretionary elements of [A.sub.i,t] influence caseload trends independently or interactively with the reform elements [R.sub.i,t].
The omission of discretionary implementation elements (the vector A) in standard evaluation models introduces not only measurement error but potential bias as well. As suggested above, caseworker decisions on rule enforcement may be conditioned on the features of program reform (R). Caseworker implementation decisions may also be correlated with perceived labor-market opportunities (E). In these circumstances, the naive model will generate weak and inconsistent estimates of caseload effects.
IV. THE DATA
To test the proposition expressed in equation (2), state AFDC caseload data for the period 1991-1996 have been assembled. Annual application and termination rates have also been computed so that the dynamics of net caseload changes can be examined.
A. Control Variables
Variables used to control for economic influences (E) on the AFDC caseloads include annual changes in state population, per capita income, unemployment, and poverty. Population growth enlarges the potential demand for welfare, especially when the population expansion is fueled by foreign immigration. Rising per capita income may reduce the demand for welfare but increase the willingness of taxpayers to supply it. Cyclical forces tend to have lagged and indirect effects on the economic status of the single mothers who account for 90% of the AFDC caseload (Schiller, 1998; see also Gabe, 1992; Blank and Ruggles, 1996).
AFDC program features not subject to federal waiver include the need standard used to determine eligibility for AFDC and related welfare benefits (e.g., Food Stamps, Medicaid) and the maximum benefit actually payable. States have consistently raised need standards even while holding the line on benefit ceilings in order to qualify needy residents for non-AFDC benefits that may entail higher federal cost shares.
Caseload characteristics that might affect the responsiveness of welfare demand to reform initiatives include the proportion of no-adult cases and the percentage of cases with children under age 3. No-adult cases often originate when an undocumented (illegal) immigrant gives birth on U.S. soil. As U.S. citizens, these "border babies" are eligible for AFDC, even though their parents are not. The parents are also immunized against the work or other self-help requirements that are often the centerpiece of welfare reform. In California, one out of four AFDC cases is a no-adult case. Nationally, one out of six cases is so classified as a result of border babies, temporary family placements, and nonmaternal guardianships. The higher the proportion of such cases, the less responsive to reform will be the demand for welfare.
AFDC households with very young children also tend to be less susceptible to welfare-reform strictures. The FSA exempted from mandatory self-help activity all mothers of children under age 3 and extended that exemption to age 6 for mothers with unmet childcare needs. Most state regulations offer similar exemptions. As a result, AFDC cases with very young children are also unlikely to be responsive to work-oriented welfare reforms unless the availability of childcare services is increased at the same time.
The availability of AFDC-U (two-parent) benefits prior to 1990 also affected the composition of state caseloads and their potential response to welfare reform (Riggin and Ward-Zukerman, 1995). As noted earlier, of the 23 states without an AFDC-U program prior to 1990, 13 opted for a restricted (6 month per year eligibility) program and the rest elected an unrestricted program. This latter subgroup had the greatest potential for a short-run jump in AFDC caseloads.
B. Local Discretion Variables
One of the central hypotheses of this exposition is that local discretion in program implementation significantly affects the observed effects of welfare-reform initiatives. Four levers of local control are particularly interesting in this regard: the approval rate, the exemption rate, the participation rate, and the assignment rate to different program activities. Eligibility determination is not a clerical function, but instead a complex interpersonal outcome affected by the skills and attitudes of local caseworkers (Lindsey, 1993, and earlier citations). In this context, higher approval rates are likely to extend welfare recipiency to households with greater labor-market experience and potential (Bernstein, 1994). By increasing access to welfare in this way, local administrators may increase the cyclical sensitivity of AFDC caseloads.
Higher exemption rates may dull the thrust of work-oriented welfare reforms by limiting the pool of "mandatory" participants. As others have shown (e.g., Bane and Ellwood, 1994) the greatest net impacts for welfare-to-work programs are likely to come from the long-term recipients with little work experience who are unlikely to volunteer for self-help activities. The FSA and many state waiver initiatives target these hard-to-employ recipients. However, local welfare administrators have the authority to exempt these and other recipients for a variety of reasons (e.g., health, transportation, or childcare problems). Like beauty, the justification for program exemption may lie in the eye of the beholder. What appears to be an insurmountable participation barrier to one sympathetic caseworker may appear to be a mere inconvenience to a less sympathetic caseworker. This inherent subjectivity may help explain why annual statewide exemption rates varied from a low of 10% to a high of 97% during the early 1990s.(3) Because this extreme variance far exceeds interstate differences in family structure (Schiller, 1996), it may best be regarded as a barometer of local administrative discordance with legislated welfare rules. As such, the exemption rate may signal both the quantity and the quality of implemented reforms.
Participation rates are computed against the pool of mandatory (nonexempt) adult recipients. In principle, variations in participation rates should be another barometer of local program implementation (Provencher, 1990). In practice, reported participation rates incorporate substantial administrative discretion concerning who is counted as a participant and for how many hours. This discretionary factor helps explain the near perfect record of states in satisfying FSA's escalating participation quotas.
The distribution of reported JOBS participants across program activities may be a better signal of local dispositions to legislated welfare reforms. The FSA mandated four employment-oriented program options to which states could assign mandatory participants: job search, on-the-job training, wage subsidies, or community work experience programs (CWEPs). States could also count school enrollment as a JOBS activity or pursue other options to satisfy gross participation quotas. The assignment rate to specific activities like CWEPs may reflect local commitment to welfare-to-work program designs. By contrast, high participant concentrations in Assessment and Development (essentially a holding stage) may be a signal of low commitment to state or national reform agendas.
Table 2 displays the range and means for these data elements. All of the data are drawn from statistics compiled by the U.S. HHS from official state reports. The dependent variable and the first six control variables are expressed as annual growth ratios. All of the remaining independent variables except AFDC-U and REFORM are expressed as year-specific percentages. The AFDC-U and REFORM variables are categorical. The AFDC-U classification is constant across years (1990-1996); a state's REFORM categorization may change from one year to the next. Although 37 of the states introduced one or more welfare reform initiatives during the observation period, only one-sixth of the state-year observation units were so characterized. Most of these reform years occurred late in the observation period.
The characterization of an entire state's welfare-reform initiative implies an intrastate uniformity that is inconsistent with the central hypothesis of local diversity. In fact, many states implemented their reforms gradually, often beginning with only a few counties. For this reason, any reforms that officially started in November or December of a given year were deemed to be initiated in the following calendar year. The use of both contemporaneous [TABULAR DATA FOR TABLE 2 OMITTED] and lagged values for REFORM also accommodates implementation delays. Finally, the local discretion variables are designed to introduce an element of heterogeneity into state reform initiatives. Ideally, both control and outcome variables would be measured at the county level (see Schiller and Brasher, 1993 for such an application). Only state-level aggregates are available for these key variables, however. In effect, the central hypothesis of local public choice must be tested with interstate variances. The implied measurement error tends to reduce the statistical significance of both reform [Mathematical Expression Omitted] and discretion (A).
There are really two hypotheses at the core of this model. The first is that the content of state welfare-reform initiatives affects the rate of caseload growth. The second and corollary hypothesis is that the observed effectiveness of state reforms is conditioned by local implementation practices. These corollary hypotheses are tested at three levels of aggregation, in concert with the following questions:
(1) Do the explicit features of welfare reform affect caseload dynamics?
(2) Are caseload effects significantly affected by local implementation factors, including (a) exemption rates or (b) other discretionary variables?
These questions are addressed by different specifications of the state REFORM variable and a varying set of discretion variables. For the first question a dichotomous specification is used initially to indicate the presence of a state reform in a particular year. Then, the REFORM variable is categorized into three classes: no reform, soft reform, tough reform (as discussed above). In the final iteration, REFORM (in both dichotomous and trichotomous specifications) is interacted with XMPT to approximate the caseload coverage of state reforms.
The dependent variable in all three models is the annual percentage change in a state's AFDC caseload. The estimates reported here are based on ordinary least squares regressions of data pooled across states and years. The pooling is justified by the fact that all states were permitted to pursue waivers during this period. Logarithmic transformations were also estimated, with similar results (available from the author).
A. Generic Reforms
Table 3 offers detailed estimates for the first set of specifications. The first reduced-form model tries to explain caseload change on the mere existence of a welfare-reform initiative (R). This is dubbed the "naive" model because it takes all reform pronouncements at face value, without regard to either the substance of reforms or their actual implementation. In this base model, any state-initiated reform is presumed to have some identifiable caseload effect.
Two discretionary adaptations of the core (dichotomous) model are also depicted in Table 3. The first adaptation adds only the exemption rate as a proxy for caseload discretion. As noted above, the exemption rate is a proxy for the actual coverage of nominal reforms. The second adaptation includes a more complete set of discretionary variables: the approval, participation, and CWEP assignment rates. If local discretion is a material factor in the efficiency of state reforms (and varies systematically across states), these added variables should have statistical significance. The inclusion of these discretionary elements in the model should also change the "naive" estimate of reform impact.(4)
The results depicted in Table 3 are consistent with the core hypotheses. To begin with, population growth, changes in the poverty rate, and cyclical changes in unemployment all have significant effects on annual AFDC caseloads, in the expected directions. More important, the estimates also indicate that state reform initiatives are effective in reducing caseload growth. Caseload growth drops nearly 4 percentage points below trend in the year state reforms are implemented, and by a comparable amount in the following year.
The "exemption" model (second column in Table 3) reveals that this basic barometer of local discretion is also a statistically significant determinant of caseload growth, in the predicted direction. The absolute size of the exemption effect (.00048) is very small. Given [TABULAR DATA FOR TABLE 3 OMITTED] the variance in exemption rates (Table 2), however, it appears that this dimension of local implementation can have a nontrivial impact on caseload outcomes. A swing of one standard deviation around the mean of the exemption rate could cut in half the estimated effects of a state reform initiative (R).
The more complete specification of local discretion (column 3) confirms that participation rates (PART) are also a significant determinant of caseload growth. The contemporaneous effect is negative, while the lagged effect is positive. This suggests that program activities delay exits but accelerate them after program completion. Given the high incidence of high school enrollment in JOBS programs, this result seems eminently reasonable. This result is also consistent with the observed magnet effect of improved welfare services (Johnson et al., 1994).
A comparison of the three models in Table 3 reveals some interesting patterns. The apparent influence of economic forces on caseload growth diminishes significantly when local discretionary factors are included in the model. This is seen in the sequential diminution of the coefficients on POP, POV, and UNEM as the naive model is expanded. This is consistent with the notion that reform initiatives empower welfare administrators to gain greater control of welfare caseloads.
B. "Soft" versus "Tough" Reforms
Table 4 takes account of the explicit content of state reforms in assessing their caseload impact. The dichotomous REFORM variable used in Table 3 is first replaced by a trichotomous specification denoting the "toughness" of a state's reforms. Then the trichotomous specification is multiplied by the nonexemption rate in an effort to capture the elements of both explicit content and local discretion. For convenience, only the results for these key variables and related test statistics are shown (the results for the full model are available from the author).
The top rows of Table 4 display the estimates for two alternative "reform" specifications within the context of the naive impact model. These results strongly support the central hypothesis of the study. The trichotomous specification reveals that "tough" welfare-reform provisions account for most of the caseload reduction attributed to the aggregated (dichotomous) REFORM variable. These results confirm that reform provisions that limit accessibility to welfare do have their intended caseload effect. In view of the fact that "tough" provisions were often accompanied by liberalized ("soft") reform features, even these estimates may understate the true caseload effects of such reforms (long-term effects may grow larger still if tougher eligibility provisions alter expectations and tastes for welfare; see Goertzel and Young, 1996).
The second set of rows in Table 4 summarize results from the "exemption" model for three specifications of the REFORM variable. The gap between "soft" and "tough" reform provisions widens in this case, suggesting the importance of enforcing tougher reform provisions with reduced use of personal (case-specific) exemptions.
Column (3) refers to an interactive REFORM specification that combines elements of reform content and enforcement. The REFORM variable used here is a multiple of the state nonexemption rate (0-100) and the trichotomous specification (0,1,2) of reform provisions. This interactive program variable substitutes for both the independent XMPT and REFORM variables in the model. The estimated effects are all in the expected direction and significant at the 1% level for contemporaneous "tough" provisions and lagged "soft" provisions.
The bottom rows in Table 4 summarize the results from the "full discretion" model that includes measures of approval, participation, and assignment rates. The explained variance increases substantially in this more complete specification. Notice also how the coefficients for all specifications of the REFORM variable drop significantly in value. This is consistent with the central notion that the explicit features of reform are only partially responsible for any observed program effects. Even when the explicit content of reform is controlled, variation in implementation will condition observed program impacts.
The multidimensional complexity of welfare reform has frustrated efforts to isolate the effects of reform. The thicket of reform provisions created by multidimensional and overlapping [TABULAR DATA FOR TABLE 4 OMITTED] state and federal initiatives destroys any pretense of "controlled" experimentation. Local discretion in program implementation further impairs our ability to describe or evaluate the actual parameters of reform initiatives.
This paper has tried to confront rather than ignore these institutional realities by incorporating measures of programmatic and administrative heterogeneity into the standard evaluation model. The effort has been modestly successful in confirming (1) the caseload-reducing effects of reform provisions that diminish the attractiveness or accessibility of welfare benefits and (2) the importance of local administrative enforcement in shaping aggregate caseload effects.
These initial findings lengthen the agenda for evaluation research. A salient implication of these findings is that unidimensional evaluation models (with a single program [reform] variable) are seriously misspecified. The complexity of a fully specified evaluation model is inconsistent with such "reduced form" pretensions. The research challenge is to develop more compelling measures of content and enforcement than these introduced here and to incorporate them in evaluation models. By progressing in these directions, it may be possible to discern more precisely whether and how welfare reform alters welfare dynamics.
AFDC: Aid for Families with Dependent Children
AFDC-U: Two-parent families
CWEP: Community work experience program
FSA: The Family Support Act
HHS: Health and Human Services
JOBS: Job Opportunities in the Business Sector
PRWOA: Personal Responsibility and Work Opportunity Act
1. If access-limiting features are a response to rising caseloads, they are endogenous rather than exogenous interventions. There is no significant correlation, however, between prereform caseload trends and reform features in the observation period.
2. The data set is not large enough to search for the statistically most significant combination of program features. In any case, such statistical searches are atheoretical. The hypothesis tested here is whether access-limiting program features actually have their intended impact. A model based on the presence of any one access-limiting feature proved to be inadequately discriminatory.
3. Much of the variation in local approval and exemption rates may be due to applicant characteristics and labor market conditions. To the extent that these factors are statistically controlled, the residual variance reflects local discretion. The same is true for other measures of discretion.
4. Some data items were not available for all states in all years, thus reducing the effective sample size, as shown in Table 3.
Bane, Mary Jo, and David Ellwood, Welfare Realities, Harvard Press, Cambridge, Mass., 1994.
Berman, Paul, "The Study of Macro- and Micro-implementation," Public Policy, 26:2, 1978, 157-184.
Bernstein, Jared, The Determinants of the AFDC Caseload: A Time-Series, Cross-Sectional Approach, PhD dissertation, Columbia University, 1994.
Blank, Roberta, and Patricia Ruggles, "When Do Women Use Aid to Families with Dependent Children?" Journal of Human Resources, Winter 1996, 56-89.
Brock, Thomas W., The Implementation of Welfare Reform: Policy and Organizational Effects on Service Delivery in Los Angeles County, PhD dissertation, UCLA, 1992.
Brodkin, Evelyn Z., The False Promise of Administrative Reform: Implementing Quality Control in Welfare, Temple University Press, Philadelphia, 1986.
-----, "Implementation as Policy Politics," in Implementation and the Policy Process: Opening Up the Black Box, Dennis J. Palumbo and Donald J. Calista, eds., Greenwood Press, New York, 1990, 107-118.
-----, "Inside the Welfare Contract: Discretion and Accountability in State Welfare Administration," Social Service Review, 71, 1997, 1-33.
Council of Economic Advisers, "Explaining the Decline in Welfare Receipt, 1993-1996," White House, Washington, D.C., May 9, 1997.
Dattalo, Patrick, "Agency and Community Characteristics Associated with Level of AFCD Client Use of Social Services," Administration in Social Work, 18:4, 1994, 25-49.
Elmore, Richard F., "Backward Mapping: Implementation Research and Policy Decisions," Political Science Quarterly, 94, 1979, 601-616.
Gabe, Thomas, Demographic Trends Affecting Aid to Families with Dependent Children (AFDC) Caseload Growth, Congressional Research Service, Washington, D.C., December 9, 1992.
Goertzel, Ted G., and Gary Young, "New Jersey's Experiment in Welfare Reform," The Public Interest, Fall 1996, 72-80.
Handler, Joel F., The Conditions of Discretion: Autonomy, Community, Bureaucracy, Russell Sage Foundation, New York, 1986.
-----, "Discretion: Power, Quiescence, and Trust," in The Uses of Discretion, Keith Hawkins, ed., Clarendon Press, Oxford, 1992, 331-360.
Hasenfeld, Yeheskel, and Daniel Steinmetz, "Client-Official Encounters in Social Service Agencies," in The Public Encounter: Where State and Citizen Meet, Charles T. Goodsell, ed., Indiana University Press, Bloomington, Ind., 1981, 83-101.
Johnson, Terry, D. Klepinger, and F. Dong, "Caseload Impacts of Welfare Reform," Contemporary Economic Policy, January 1994, 89-104.
Kane, Thomas J., and Mary Jo Bane, "The Context for Welfare Reform," in Welfare Realities: From Rhetoric to Reform, by Mary Jo Bane and David T. Ellwood, Harvard University Press, Cambridge, Mass., 1994, 1-27.
Lindsey, Elizabeth W., "Training Georgia's Eligibility Workers," Public Welfare, Spring 1993, 35-41.
Lipsky, Michael, Street Level Bureaucracy: Dilemmas of the Individual in Public Services, Russell Sage Foundation, New York, 1980.
Martini, Alberto, and Michael Wiseman, "Explaining the Recent Decline in Welfare Caseloads," Challenge, 40:6, November/December 1997, 6-20.
Mead, Lawrence. "Welfare Policy: The Administrative Frontier," Journal of Policy Analysis and Management, 15:4, 1996, 587-600.
Meyers, Marcia K., Bonnie Glaser, and Karin Mac Donald, "On the Front Lines of Welfare Delivery: Are Workers Implementing Policy Reforms?" Journal of Policy Analysis and Management, 17:1, 1998, 1-22.
Nathan, Richard, Turning Promises into Performance: The Management Challenge of Implementing Workfare, Columbia University Press, New York, 1996.
Palumbo, Dennis J., and Donald Calista, "Opening Up the Black Box: Implementation and the Policy Process," in Implementation and the Policy Process: Opening Up the Black Box, Dennis J. Palumbo and Donald J. Calista eds., Greenwood Press, New York, 1990, 317.
Payne, James L., "Absence of Judgement," Policy Review, November/December 1996, 50-54.
Petersen, Carol Dawn, Intra-state Inequalities in the Financing and Delivery of Noneducational Social Services to Children and Youth, PhD dissertation, University of Texas, Austin, 1996.
Piskulich, C. Michelle, "Toward a Comprehensive Model of Welfare Exits: The Case of AFDC," American Journal of Political Science, 37:1, 1993, 165-185.
Prottas, Jeffrey Manditch, People Processing: The Street-Level Bureaucrat in Public Service Bureaucracies, Lexington Books, Lexington, Mass., 1979.
-----, "The Cost of Free Services: Organizational Impediments to Access to Public Services," Public Administration Review, 41:5, 1981, 526-534.
Provencher, Paul J., Welfare Recipients and Employment: The Influence of the Attitudes of Case Managers and Other Factors on Program Performance in Local Welfare Offices, PhD dissertation, Brandeis University, 1990.
Riccio, James, and Yeheskel Hasenfeld, "Enforcing a Participation Mandate in a Welfare-to-Work Program," Social Service Review, 70, 1996, 516-542.
Riggin, Leslie J., and Betty Ward-Zukerman, "Effects on the AFDC-Basic Caseload of Providing Welfare to Two-Parent Families," The Social Science Journal, 32:3, 1995, 265-278.
Schiller, Bradley R., "Assessing the impact of Welfare Reform: The Role of Local Discretion," Unpublished manuscript, American University, 1996, mimeo.
-----, The Economics of Poverty and Discrimination, 7th ed., Prentice Hall, Upper Saddle, N.J., 1998.
Schiller, Bradley R., and C. Nielsen Brasher, "Effects of Workforce Saturation on AFDC Caseloads," Contemporary Policy Issues, April 1993, 39-49.
Snyder, Nancy M., "Managerial and Organizational Impediments to Effective Welfare Reform," Journal of Health and Human Resources Administration, 14:4, 1992, 445-464.
|Printer friendly Cite/link Email Feedback|
|Author:||Schiller, Bradley R.|
|Publication:||Contemporary Economic Policy|
|Date:||Apr 1, 1999|
|Previous Article:||Time limits on welfare receipt.|
|Next Article:||The demand for cigarettes in Taiwan: domestic versus imported cigarettes.|