Lone Star Image System Evaluation
Final Report

Executive Summary


 
 

Deanna T. Schexnayder

Jerome A. Olson
Daniel P. O’Shea
Davy Norris
Daniel Schroeder
Christopher T. King

 
 

August 1997
 
 
 
 

Center for the Study of Human Resources
Lyndon B. Johnson School of Public Affairs

The University of Texas at Austin


This report was prepared with funds provided through Interagency Agreement No. UTA97-0019 from the Texas Department of Human Services to the Center for the Study of Human Resources at the University of Texas at Austin.The views expressed here are those of the authors and do not represent the positions of the agency or of The University.

 
 

Executive Summary

The Texas Department of Human Services (DHS) has been conducting a welfare reform demonstration project entitled “Lone Star Image System” in Bexar and Guadalupe counties, as required under the provisions of Texas House Bill 1863, enacted by the Texas Legislature in May 1995.  This project uses electronic finger-imaging to detect and deter duplicate applications for Aid to Families with Dependent Children (AFDC) and Food Stamp benefits.  DHS contracted with the Center for the Study of Human Resources of the LBJ School of Public Affairs at The University of Texas at Austin (CHR) to evaluate the impact of this demonstration on Food Stamp and AFDC caseloads and to estimate the costs and/or savings of this initiative.
 

 

Overview of the LSIS Demonstration

DHS began operating the LSIS demonstration in Bexar and Guadalupe counties in late October 1996, with pilot operations scheduled to continue through August 1997. According to the project design, all non-exempt adults, heads of households and minor parents with a dependent child who request Food Stamp or AFDC benefits must comply with an electronic finger-imaging and photographing procedure at the time of application or recertification.

DHS has contracted with North American MORPHO, Inc. to install and operate the imaging equipment in the demonstration offices. The refusal or failure of a non-exempt individual to be imaged results in denial of the application or case. If a match occurs and fraud is suspected, a referral is made to the Office of Inspector General at DHS for investigation. All ten DHS local offices that serve residents of these counties are participating in the demonstration.

 

Overview of the Evaluation

The evaluation of the LSIS demonstration consists of both impact and cost analyses over the first seven months of its operation. The purpose of the impact analysis is to measure the effect of the LSIS demonstration on Food Stamp and AFDC caseloads and to explain changes in the rates at which persons enter and leave the two programs.  The cost analysis examines the cost features of the LSIS demonstration and the potential benefit savings resulting from the program. Findings are based on an analysis of administrative caseload and cost data from October 1995 through May 1997, as well as interviews with Food Stamp and AFDC recipients who have recently exited from at least one of the programs.

Impact Analysis

Impact Research Questions

The objective of the LSIS pilot is to reduce duplicate receipt of Food Stamp and AFDC benefits.  The research questions developed to estimate the effect of duplicate receipt of benefits on caseloads are:

1. What effect has the demonstration had on factors influencing the Food Stamp and AFDC caseload flow, especially the disposition of initial applications for benefits and recertifications?

2. What effect has the finger-imaging demonstration had on the size of the active Food Stamp and AFDC caseloads?

These questions were addressed both through the statistical analysis of Food Stamp and AFDC caseloads over time and interviews with former Food Stamp and AFDC recipients.

The statistical analysis utilized a pre-post/pilot-comparison site research design. Comparison DHS offices were selected by identifying the ‘nearest neighbor’ for each of the ten pilot offices, based on a number of variables that have been shown to affect Food Stamp and AFDC program dynamics. Both unadjusted and adjusted net effects were calculated to measure the effect of LSIS on Food Stamp and AFDC caseload flows. A dynamic simulation model was developed to estimate the effect of LSIS on the size of the caseloads.[1]

CHR also interviewed randomly-selected heads of Food Stamp and AFDC cases who failed to recertify for benefits for two consecutive months prior to the interview. Face-to-face interviews were conducted with nearly 200 individuals over a four-month period from four demonstration offices and four comparison offices. Questions were asked to gather information about former recipients’ exits from public assistance and to discern any effects that electronic imaging may have had on these exits. Researchers also ascertained recipients’ receptivity to imaging and their awareness of fraudulent receipt of public assistance benefits.
 

Key Impact Analysis Research Results

Statistical Results

The Food Stamp and AFDC caseloads in both the pilot and comparison sites were declining steadily from the beginning of the baseline period until May 1997, the last month for which data were available. A dynamic simulation model was developed to estimate how much of this decline was attributable to the LSIS demonstration. Implementation of this statistical model produced a net increase for the Food Stamp caseload of five cases out of 55,000 (0.01 percent), and a net decrease for the AFDC caseload of seven cases out of 18,486 (0.04 percent). Neither of these results was statistically significant. Thus, there is no statistical evidence that any of the observed caseload decline in Bexar and Guadalupe counties occurred because of LSIS.

The effect of LSIS was calculated for nine Food Stamp client flows and 18 AFDC flows affecting the caseloads, including changes in approval and denial status, placement in temporary hold status, or movement to and from other parts of Texas.  For Food Stamps, the demonstration caused a statistically significant increase in exits from the active caseload of 1.3 percent. However, this was offset by an increase in caseload entries, many of whom were the same individuals who had exited. Thus, one effect of the demonstration was to induce temporary exits for a significant number of Food Stamp cases. The pilot also increased the flows into and out of hold status, suggesting that Food Stamp cases took longer to process as a result of the imaging requirement. This may have occurred because many Food Stamp cases included more than one adult, each of whom had to visit the DHS office to be imaged.

Of the 18 AFDC caseload flows, only one minor flow changed significantly as a result of the demonstration. The statistical analysis produced no evidence that LSIS had any effect on the major flows in and out of the AFDC caseload.

Recipient Perceptions

Clients reported that most of their exits from welfare were associated with increased earnings and income, client choice (such as missed appointments or deciding that benefits were not worth the effort) or problems with service delivery.  Since considerable shares of all exits occur due to client choice or service delivery problems, additional tasks, such as requiring all adults in a household to visit the welfare office to fulfill the imaging requirement, could contribute to dips in the patterns of welfare receipt.

Biometric imaging of public assistance recipients may have three effects: imaging may deter individuals from attempting fraud by caseload duplication, detect caseload duplication by providing an identification match, or influence individuals to defect from public assistance by creating an unacceptable barrier to their continued receipt of benefits. While the interviews did not reveal any detection or deterrence due to duplicate benefits, a small subset of the interview population found imaging unacceptable or burdensome, resulting in a few of them giving up their Food Stamp benefits.

Food Stamp and AFDC recipients generally share the widespread public perception that fraud and abuse is a major problem in our welfare system. Nearly three-quarters of the respondents supported electronic imaging as a method to reduce fraud and abuse. However, only small shares of the respondents indicated first-hand knowledge of Food Stamp or AFDC fraud. At most, only one of the 36 examples of fraud of which respondents were aware may have involved duplicate benefits.

The types of fraud most commonly mentioned by respondents were unreported income, misrepresented household composition, and selling Food Stamp benefits. These are the same types of fraud regularly subject to field investigations by the Office of Inspector General at DHS and for which several automated checks are already in place. Electronic imaging would not be effective in detecting the vast majority of these types of fraud.

 Cost Analysis

Cost Research Questions

The cost analysis estimated the net savings/loss from operating the LSIS demonstration, as well as the net flow of federal funds into the state of Texas and the flow of federal and state funds into the demonstration counties as a result of LSIS. The three research questions addressed by this analysis are:

1. To what extent are the added or incremental costs of the LSIS pilot demonstration offset by savings due to reduced participation in the AFDC, Food Stamp and Medicaid programs, computed separately by program and overall?

2. To what extent are federal funds to Texas reduced as a result of LSIS pilot implementation?

3. To what extent are federal and state funds to Bexar and Guadalupe counties (the demonstration sites) reduced as a result of LSIS pilot implementation?

The cost analysis describes the costs resulting from LSIS development, implementation and operations from July 1996 through May 1997. The estimate of benefit savings is derived from the impact analysis dynamic simulation model and encompasses the first seven months of the demonstration program operation from November 1996 through May 1997.[2]

In addition to these three questions, the original evaluation plan called for an assessment of the costs of implementing this demonstration on a statewide basis. That research question was dropped from the analysis after the Texas Legislature authorized the implementation of biometric imaging on a statewide basis and appropriated a fixed amount of funds to accomplish this. The requisite cost data needed to estimate statewide costs were unavailable at the time this report was prepared.

Key Cost Research Results

The LSIS demonstration cost the state of Texas $1.7 million for the first seven months of operation, and, excluding development and implementation costs, yielded a net operating loss of $892,798.  Furthermore, since LSIS yielded no net impacts on benefit payments, there was no indication, based on demonstration results, that the initiative would recoup the development and implementation costs of $805,093.

The demonstration did increase the availability of funds at the state and regional levels. The LSIS pilot resulted in a net inflow of $270,000 in federal funds to the state of Texas. There was also a net inflow of $305,734 in state and federal funds to the demonstration counties, resulting in a total positive long-term impact—including economic multiplier effects—of $859,357 for the economies of these counties. Thus, the demonstration provided some benefit at the regional level by increasing temporary employment opportunities.

 

Conclusions

In the LSIS demonstration, electronic imaging has failed to produce the expected effects. The demonstration has not reduced caseloads significantly by detecting or deterring duplicate benefits.  Instead, it appears to have induced some temporary exits among Food Stamp recipients for whom the process of getting Food Stamps has become more difficult due to the need to have all adults on a case report to a DHS office for imaging. The LSIS demonstration cost the state of Texas $1.7 million for the first seven months of operation and yielded no savings in benefit payments.

 

[1] While researchers used the most powerful available statistical techniques to conduct this analysis, these techniques would not account for caseload changes due to duplicate benefits occurring in other parts of the state.
[2] The cost analysis has several limitations, foremost among which are problems relating to data availability. A detailed discussion of limitations is included in the full report.