WEST VIRGINIA WORKS:
Welfare Reform Pilot Program

Special Report

Program Outcomes


OFFICE OF LEGISLATIVE AUDITOR
Performance Evaluation and Research Division
Building 1, Room 314W
State Capitol Complex
CHARLESTON, WEST VIRGINIA 25305
(304) 347-4890


Section 1


Introduction


Congress made sweeping changes to the nation's welfare system through passage of the Personal Responsibility and Work Opportunity Reconciliation Act of 1996 (P.L. 104-193). The Act was signed into law by the President in August of 1996. Title I of the Act eliminated the old welfare system, Aid to Families with Dependent Children (AFDC), and replaced it with a block grant system known as Temporary Assistance for Needy Families (TANF). TANF was designed to place a greater emphasis on work, and thereby, reduce the level of dependency on government assistance. One of the more important changes of the Act was a five-year lifetime limit for receiving TANF benefits.

The new law also eliminated the Job Opportunities and Basic Skills Training Program (JOBS) which was an AFDC program. The JOBS program was intended to use work, education and training programs to develop work motivation and reduce dependency on welfare, thus reducing welfare caseloads. The 1988 legislation that established JOBS intended the program to focus on education and training more so than on placing individuals in actual work positions. Consequently, most States had few recipients in work programs such as the Community Work Experience Program (CWEP) in which recipients worked a certain number of hours for their AFDC payment to gain work experience. In fact, West Virginia was one of only four States that had a significant portion of its JOBS participants in CWEP. Under TANF, greater emphasis is placed on getting individuals employed, with a lesser emphasis on education and training.

The federal legislation permits the operation of more than one type of TANF system. This allows States to experiment with different approaches to welfare reform before one approach is used Statewide. TANF provides States discretion and guidelines with respect to eligibility determination. However, certain requirements of TANF must be complied within all counties including pilot counties. Two important requirements that are in effect Statewide are the five-year lifetime limit of receiving benefits, and all families must work after 24 months on assistance in order to continue receiving assistance. Consequently, West Virginia has two TANF systems in place: 1) Forty-six counties operate under TANF, and 2) nine pilot counties operate under the WORKS program which conforms with TANF but has different eligibility requirements than the non-WORKS counties.

Statewide operation of TANF and the West Virginia WORKS pilot program began on January 1, 1997. Nine counties were chosen to begin implementation of WORKS. They include Greenbrier, Mercer, Monroe, Nicholas, Pocahontas, Tyler, Wetzel, Wayne, and Wood. The 46 non-WORKS counties are divided into four groups, each of which will be phased into the WORKS program by January 1998. The counties with the largest caseloads are scheduled to be included first. The gradual phasing in of the program allows for costs to be absorbed and organizational changes to take place more easily.

One important difference under WORKS is eligibility determination. Table 1 illustrates some of the common and different provisions between non-WORKS and WORKS counties. The asset limits are higher under WORKS. This can result in some individuals to be eligible for WORKS that would not have been eligible for AFDC. However, more income is included in determining eligibility, particularly SSI income. The inclusion of SSI has caused some who were formerly eligible to become ineligible under WORKS. The Personal Responsibility and Work Opportunity Reconciliation Act of 1996 redefines children's disability under SSI, which will take some off of SSI and make them eligible for cash assistance under WORKS.

Another difference under WORKS is caseload management. WORKS staff caseload will be established at a ratio of no more than 100 cases per staff position. This is drastically reduced from pre-reform levels. Staff will have the opportunity to work more closely with families. The agency indicated that 18 new people were hired to implement WORKS in the pilot counties. Nine were strictly for WORKS and nine were hired to replace people who transferred from within the agency. The total cost was $422,499. As WORKS is implemented Statewide, additional workers will be needed. If the same ratio of workers and expenditures holds for the remaining 46 counties, then 92 new staff will be needed at a cost of $2.2 million.

Table 1
Eligibility Requirements

Provision

TANF 46 COUNTIES

WV WORKS 9 COUNTIES
Assets$1,000 assets limit per case$2,000 asset limit per case

Vehicle

$1,500 equity value
1 vehicle excluded regardless of value

Income

Income counted in accordance with AFDC State Plan

All income counted except the Earned Income Tax Credit, Tax refunds, reimbursements and income excluded by federal law. SSI payments are counted
Personal Responsibility Contract
No Provision

Required
Benefit Time Limit60 months lifetime limit60 months lifetime limit
Work RequirementsWork required once individual has received assistance for 24 monthsWork required once individual has received assistance for 24 months
Marriage IncentiveNo Provision10% increase on monthly cash assistance for married couples
Prohibition of payments to individuals convicted of drug felonies
Provision Applies

Provision Applies

Section 2


WORKS Pilot Counties Are a Good Representation of the State

Senate Bill 430 required the selection of the pilot counties to fairly represent both rural and urban areas. The program was also to include a minimum of 15% of the state population that qualified for AFDC. These requirements were met in the selection process. The AFDC caseload for the pilot counties was 15.4% of the Statewide total at the start of 1996, as indicated in Figure 1.

The rural and urban composition of WORKS county population was very similar to the Statewide averages and the non-WORKS counties. Table 2 reports economic variables for the pilot counties, and a comparison of pilot county averages with State averages and non-WORKS averages. The unemployment rate in the pilot counties was less than the State average and non-WORKS counties by about one percentage point in 1996. The average county population was very close for the State, WORKS counties and non-WORKS counties.

A comparison was also made of caseload statistics as of January 1996. For example, the percent of AFDC cases with single female head of household were virtually identical. These statistics and others along with a detailed discussion are contained in Appendix D. According to this analysis, the WORKS pilot counties provide a good economic and demographic cross-section of the rest of the State.



Table 2
Economic and Demographic Comparison

County
Unemployment Rate: 1996 AverageTotal Population: 1990 CensusUrban % of Population: 1990 CensusRural % of Population: 1990 CensusPer Capita Income: 1990 Census
Greenbrier8.6%34,69318.4%81.6%$16,747
Mercer5.2%64,98030.5%69.5%$17,607
Monroe6.0%12,4060.0%100.0%$13,466
Nicholas10.9%26,77520.7%79.3%$13,685
Pocahontas13.0%9,0080.0%100.0%$15,873
Tyler7.4%9,7969.3%90.7%$13,466
Wayne7.2%41,63628.9%71.1%$13,611
Wetzel10.0%19,25844.7%55.3%$16,500
Wood6.0%86,91565.9%34.1%$19,573
State Average9.1%32,60936.1%63.9%$15,416
WORKS County Average
8.3%

33,941

36.2%

63.8%

$15,614
Non-WORK S County Average
9.2%

32,348

36.1%

63.9%

$15,377

Section 3


Caveats to the WORKS Program Evaluation

There are important caveats to the analysis of the West Virginia WORKS project which must be discussed in order to properly interpret the results of the project. One caveat is the time required to implement WORKS. The new program was first implemented in nine pilot counties in January of 1997. In order to implement the program, the pilot counties had to convert each of their existing AFDC cases to WORKS, along with any new cases they received. The actual conversion of existing cases involved first determining if individuals still qualified for cash assistance under the new rules. Recipients who previously were receiving AFDC prior to the beginning of WORKS had to visit the county office to be reevaluated. Persons who qualified for the WORKS program went through an orientation session explaining the new requirements. Counties accomplished this through individual or group sessions. The conversion process resulted in a gradual phase in of the WORKS program of each county's total caseload. Overall, the conversion process took five to six months to convert a county's total caseload to WORKS. Consequently, the evaluation of WORKS cannot measure the complete impact of the program because only a portion of the cases operated under WORKS for the first eight months of the program.

Table 3 below documents the number of existing cases in January 1997. During the months of January through July 1997, 493 new cases entered the system adding to the number of cases to be converted.

Table 3
Existing AFDC Cases Prior to Conversion
County Existing Cases, Jan. 1997
Greenbrier387
Mercer1266
Monroe107
Nicholas539
Pocahontas78
Tyler170
Wayne772
Wetzel336
Wood1215

The conversion of one case according to one community service manager could take up to 2 _ hours, allowing an individual worker to convert only three cases per day. In addition to the completion of the orientation session, applicants must also sign a Personal Responsibility Contract, and immediately register with the Employment Security Division of the Bureau of Employment Programs. The Personal Responsibility Contract basically states, that the individual signing it understands that becoming employed is the overall goal of this program and that the individual will move to become employed and self-sufficient. In addition, the client promises to take care of any dependent children and keep them in school. Qualifying residents have two years to participate in a work activity or face the loss of eligibility for cash assistance. During this time they may be enrolled in a Community Work Experience Program (CWEP) or subsidized employment program if they are unable to find unsubsidized work. There is a 60 month lifetime limit to receive cash benefits for all persons physically able to work.

As indicated by Table 4, the percentage of AFDC cases converted to WORKS was low during the first month. The percentage of cases converted by each of the 9 counties ranged from 0.3% to a high of 17.8%. During February and March, the rates leveled off somewhat. However, at the end of March, most counties had more than half their cases converted to WORKS. Wood County continually lagged behind in the conversion with only 36% of the cases being converted at the end of June 1997 while the other eight counties had converted 95% or more.


Table 4
Rate of Conversion of AFDC Cases to WV WORKS
CountyJan 1997Feb 1997Mar 1997April 1997May 1997June 1997
Greenbrier 16.5%42.6%67.1%95.4%97.9%100.0%
Mercer8.0%33.2%68.9%93.2%98.8%99.4%
Monroe17.8%39.7%64.7%94.8%96.5%95.7%
Nicholas9.8%31.1%64.2%92.3%99.3%98.8%
Pocahontas15.2%35.0%64.7%97.7%100.0%100.0%
Tyler6.7%29.1%52.9%72.7%92.9%100.0%
Wayne3.2%16.0%45.1%65.9%91.4%96.7%
Wetzel11.3%32.1%50.8%85.1%93.0%100.0%
Wood0.3%5.0%11.6%18.8%24.8%36.5%

Interviews with personnel from each of the districts involved in the pilot program indicated that several factors affected the conversion process. One of these factors is the computer systems currently in use by DHHR staff. Entries may have to be made in up to three systems. Data maintained in one system called the C-219 allows the department to write checks and automatically write correspondence to clients. Data is also entered into the Work Information System (WIS) which tracks persons registered to work or train. During the conversion to WORKS, some of the offices in these districts were undergoing conversion to the RAPIDS computer system for their non-AFDC cases. The RAPIDS system is used under WORKS for data related to Food Stamps and Medical Card recipients. Workers must also maintain manual records for each individual client. WORKS records are maintained separately from other social services records.

Many of the workers now assigned to the WORKS program were either experienced in Income Maintenance or Work and Training. Under WORKS, a worker had to be knowledgeable in both areas of the welfare system. One Community Service Manager indicated that workers required two months of intense training to understand the amount of policy each worker must know. However, only one week or less of training was provided prior to the beginning of the conversion process. Some offices were also understaffed according to staff interviewed. The problems of multiple computer systems, inexperienced personnel, new rules and regulations, a manual application process and understaffing slowed the process. All of these factors affected Wood County according to personnel interviewed, especially the understaffing resulting in only 36.5% of the cases being converted by June 1997.

Because of the task of converting cases to WORKS, the complete impact of the program cannot be measured. Persons whose cases had not been converted to WORKS during this time period were subject to the Federal TANF rules but not the WORKS changes. Those cases which were converted had only a few months under WORKS. This may not provide enough time to effect measurable change. Furthermore, any measurable impact may not be complete because of the small number of months to measure. Although the study shows some positive results under the pilot project, the improvements could increase over time, although the cases will become more difficult as well. Additional months of operation will be needed to allow trends to develop which will provide a better picture of how well the program is operating. Moreover, it was not within the scope of this study to measure the specific behavior of caseworkers to determine the exact impact on recipients obtaining employment. Although the Personal Responsibility Contract implies initiative on the recipients, some caseworkers may play more of a role in placements than others.

Finally, it was beyond the scope of the study to calculate the federal participation rate which requires DHHR to have 30% of all families receiving temporary assistance to be working 20 hours a week, and 75% of two-parent families working 35 hours a week. There are other work components, training and education components that can count in calculating the participation rate. However, data limitation prevented the total participation rate from being calculated. At the time of the study, DHHR did not have available all of the components.

Section 4


Employment Placements Under the WORKS Program

The passage of Senate Bill 430 of the 1997 Legislative session required the Performance Evaluation and Research Division, within the Legislative Auditor's Office, to undertake a statistical study of the West Virginia WORKS program as implemented under a pilot project by the Department of Health and Human Resources. The primary interest of the legislative study is to evaluate the rate at which participants in the WORKS pilot program move to each of the following three work categories:

Unsubsidized Employment Employment provided by an employer who does not receive a subsidy.

Subsidized Employment Employment in which a portion of the recipient's earnings are subsidized by the state to the employer.

Work Experience Work experience involves recipients working a certain number of hours for their welfare benefits to gain work experience.

Table 5 presents the total number of unsubsidized employment placements over the first eight months of 1997 under WORKS, and the first eight months of 1996 prior to WORKS. As Table 5 indicates, unsubsidized employment placements increased substantially, but results varied widely. Five counties, Mercer, Nicholas, Pocahontas, Wayne and Wetzel showed sizeable gains over the same period of the previous year. Three counties, Greenbrier, Monroe and Tyler showed little or no change, while Wood County had a large drop in the number of employment placements. The low job placements for Wood County were statistically significant, indicating that they were unusually low compared to the pre-WORKS period. However, this does not mean that the WORKS program caused the drop in employment placements in Wood County, only that the drop occurred during the WORKS period. DHHR indicated that Wood County experienced personnel problems that significantly hindered its performance. Therefore, Wood County presents a distortion in measuring the impact from WORKS since it is not clear what would have occurred in Wood County if there were no staffing problems. It is possible that employment placements could have increased or decreased, or showed no change. As a result, including Wood County would understate the impact of the WORKS program, but excluding it could be overstating the WORKS impact depending on what would have occurred under normal circumstances. With this caveat, the study reports employment placements with and without Wood County. Table 5 reports unsubsidized employment placements. When Wood County is included, the pilot counties experienced an 11% increase in job placements, while excluding Wood County shows a 42% job placement increase. Because of relatively small caseloads, counties such as Monroe, Pocahontas, and Tyler will show large percentage changes even though the absolute change is relatively small.

The types of part-time or full-time jobs obtained by recipients vary, but most are in clerical, sales, and service occupations. The wage scale range from minimum wage to nine dollars an hour. However, most of the jobs are in the five dollar an hour range.


Table 5
Unsubsidized Employment Placements
PRE-WORKS
January - August
1996
WORKS
January - August
1997

Difference
1997 minus 1996

Percentage
Change
Greenbrier959500.0%
Mercer2072544722.7%
Monroe1615-1-6.3%
Nicholas901617178.8%
Pocahontas62317283.3%
Tyler1820211.1%
Wayne631165384.1%
Wetzel33653297.0%
Wood22587-138-61.3%
Totals753836+8311.0%
Totals Excluding Wood County
528

749

+221

41.9%
Source:Department of Health and Human Resources, Work Information System (WIS) Reports for selected months.

Table 6 compares employment over the last five years using the first eight months of each year. The data shows a downward trend during 1995 and 1996, with 1997 representing a sharp upturn. Although the data includes only the first eight months of each year, the downward trend in 1995 and 1996 occurred for a full calendar year. A similar table in Appendix B shows employment for the entire calendar year from 1993 to 1996. Employment was up by 11.7% in 1994, then dropped by 15.7% in 1995 and by 10.6% in 1996. When Wood County is excluded from the calculations, job placement growth was less in 1994 and 1995 and improved somewhat in 1996. However, the eight counties had job placement growth from 1.9% to 41.9% from 1996 to 1997.


Table 6
Unsubsidized Employment Placements
Total Placements & Percentage Change Over
the First Eight Months of Each Year
Jan. - Aug.
1993
Jan. - Aug.
1994
Jan. - Aug.
1995
Jan. - Aug.
1996
Jan. - Aug.
1997
Greenbrier1291031179595
Mercer173223172207254
Monroe3628201615
Nicholas9513212490161
Pocahontas181719623
Tyler1122101820
Wayne88674263116
Wetzel4952143365
Wood24128927622587
Total840933794753836
Percentage
Change

na

11.1%

-14.9%

-5.2%

11.0%
Percentage
Change Excluding Wood County

na

7.5%

-19.6%

1.9%

41.9%
Source:Department of Health and Human Resources, Work Information System (WIS) Reports for selected months.

Are Employment Gains Caused by WORKS or by the Economy?

The obvious question is "Would the employment gains still have occurred without the WORKS program?" In order to answer this question, regression analysis was performed on monthly unsubsidized employment placements from January 1993 through August 1997. The unemployment rate for each county was included to control for economic influences on employment placements. Another variable was included to determine if there was a statistically significant difference between job placements during the months under WORKS (January 1997 to August 1997) and the months under the old welfare system (January 1993 to December 1996).





The results of the regression analysis are reported in Table 7. For Mercer, Nicholas, Pocahontas, Wayne, and Wetzel counties, gains in employment placements can be attributed to the WORKS program after controlling for economic influences. No statistical difference in job placements was found in Greenbrier, Monroe or Tyler counties, while Wood County had on average 136 fewer jobs for the WORKS period compared to pre-WORKS.

Table 7
Impact on Unsubsidized Employment
From WORKS Program

Counties

Monthly Average Difference in Job Placements Under WORKS vs.
Pre-WORKS
Mercer64 more job placements
Nicholas48 more job placements
Pocahontas8 more job placements
Wayne56 more job placements
Wetzel24 more job placements
GreenbrierNo Change
MonroeNo Change
TylerNo Change
Wood136 Less job placements
The impact from WORKS was statistically significant at the 95% confidence interval for those counties with a positive or negative impact. The unemployment rate was not significant in any of the regressions at the 95% confidence interval, and in only one county (Nicholas) the unemployment rate was significant at the 90% confidence interval. The lack of correlation is also revealed in the fact that despite a declining unemployment rate in every county since 1994, there was a two year decline in unsubsidized employment placements. This suggests that employment placements may be determined more by recipient and agency efforts than by economic conditions. This contention is supported in that the JOBS program by legislative intent was more an education and training program than an employment program. Furthermore, economic conditions are not significantly different in 1997 than in 1995 and 1996 as measured by the unemployment rates. In most counties, unemployment rates have been steady over the last two or three years. However, the greater work emphasis under WORKS has resulted in sharp increases in job placements in some pilot counties. [The regression models for the analysis above are illustrated in Appendix C.]






Work experience placements, or the Community Work Experience Program (CWEP), increased by 70% under WORKS (96% excluding Wood County). Under CWEP, individuals work at non-profit organizations or local government offices. The employer does not pay the recipient, instead recipients work for their cash assistance to gain work experience. CWEP was utilized more frequently in most pilot counties during the first eight months of 1997 than during the same period of 1996. Placements in the program had been declining in 1995 and 1996. Although large gains occurred in CWEP placements, the structure of the program was not in compliance with the minimum wage law through September of 1997. The U.S. Department of Labor has stated that programs like CWEP must meet the minimum wage requirement and other Fair Labor Standards Act requirements. Prior to WORKS, the number of hours worked in CWEP was determined by dividing the minimum wage rate into the AFDC grant amount. This calculation determined the number of hours a recipient could work and still be in compliance with federal minimum wage laws. In most cases, the number of hours worked would not fulfill the federally required weekly hours of work under welfare reform. Consequently, the agency changed the structure of CWEP at the time WORKS was implemented in the pilot counties to require recipients to work 20 hours a week for single parent families and 35 hours a week for two-parent families. This had the effect of causing many recipients to work below the minimum wage rate. The agency indicated that an emergency manual revision was released in October conforming the work experience program with the Fair Labor Standards Act. The effect of having CWEP conform with the Fair Labor Standards Act could make it difficult for many CWEP placements to count towards the federal participation rate because the weekly number of hours worked will be insufficient. The federal participation rate requires 30% of all family recipients to work at least 20 hours a week, and 75% of two-parent recipients to work 35 hours a week. States stand to lose federal funding if participation rates are not met.



Table 8
Work Experience (CWEP) Placements
First Eight Months of 1997 Compared to First Eights Months of 1996
PRE-WORKS
January - August
1996
WORKS
January - August
1997

Difference
1997 minus 1996

Percentage
Change
Greenbrier7372-1-1.4%
Mercer5310653100.0%
Monroe1217544.6%
Nicholas62872540.3%
Pocahontas671 16.7%
Tyler156752346.7%
Wayne4711770148.9%
Wetzel2610579303.8%
Wood6635-31-47.0%
Totals360613+25370.3%
Totals Excluding Wood County
294

578

+284

96.6%
Source: Department of Health and Human Resources, Work Information System (WIS) Reports for selected months.

Table 9 shows that 1997 CWEP placements were in sharp contrast with previous years. As in the case of unsubsidized employment, CWEP placements were also in decline in 1995 and 1996. However, the drop in 1996 was more pronounced for CWEP placements, declining by 37%. This resulted in a large rebound in 1997 from a relatively low level in 1996.

Regression analysis shows that CWEP placements are not correlated with the unemployment rate. Furthermore, despite the large increase in CWEP placements in 1997, only two counties had statistically significant improvements compared to past use of CWEP positions, Tyler and Wetzel counties. Tyler County had six more CWEP placements per month than in previous years, and Wetzel County had eight more per month. Although Mercer, Nicholas, and Wayne counties had large gains over 1996, their CWEP placements in 1997 were typical to past performance. When Wood County is excluded, the placement growth rates do not change significantly from 1994 through 1996. However, there is a larger difference for 1997.


Table 9
Work Experience (CWEP) Placements from 1993 - 1997
First Eight Months of Each Year
Jan. - Aug.
1993
Jan. - Aug.
1994
Jan. - Aug.
1995
Jan. - Aug.
1996
Jan. - Aug.
1997
Greenbrier46104787372
Mercer1321268053106
Monroe1523151217
Nicholas7486836287
Pocahontas9141367
Tyler1416271567
Wayne11214011447117
Wetzel48415326105
Wood1401261086635
Total590676571360613
Percentage
Change

na

14.6%

-15.5%

-37.0%

70.3%
Percentage
Change Excluding Wood County


na


22.2%


- 1 5 .8%


-36.5%


96.6%
Source:Department of Health and Human Resources, Work Information System (WIS) Reports for selected months.

Subsidized employment is represented by the Employment Incentive Program (EIP) and its precursor, the Jobs Training Partnership Act/On the Job Training (JTPA/OJT) program. EIP is structured so that the state pays half of a welfare recipient's salary paid by participating non-profit organizations, and private or public sector employers. The employer is required to commit to retaining the employee after the completion of the contract. An EIP contract can last for a period of time between 200 and 600 hours, with a beginning hourly wage rate of $5.99 or more. The goal of subsidized employment is for placements to become permanent, unsubsidized positions.

This form of employment has not been greatly utilized by WORKS pilot counties, as indicated by Table 10. DHHR statistics indicate that, as of August, only three WORKS counties have placed recipients in EIP during 1997. Prior to 1997, only one county, Nicholas County, appears to have made more than one EIP placement in any given month. The Nicholas County DHHR Office made ten EIP placements in 1995, five in 1996, and one in 1997.


Table 10
Yearly Totals for Subsidized Employment Placements
19931994199519961997
Greenbrier10201
Mercer10102
Monroe01000
Nicholas121051
Pocahontas00000
Tyler35200
Wayne31100
Wetzel1610120
Wood00000
Source:Department of Health and Human Resources, Work Information System (WIS) Reports for selected months.
*Note: Subsidized employment placements prior to November 1994 were made under the JTPA/OJT program. At that time the program was replaced by EIP. Totals for 1997 are for the period covering January through August.

When examining the total number of subsidized employment placements at the end of each month, another interesting trend is apparent. Wetzel County had as many as seven EIP contracts in place at one point during 1993, with a total of 16 placements for the year. This same county has not completed any EIP contracts during 1997 and only two placements were made in 1996.

The WORKS program has had no positive impact on the number of subsidized employment placements. The use of subsidized employment has been sporadic statewide for as far back as this study goes. Three pilot counties (Greenbrier, Mercer, and Nicholas) have made one or two placements in 1997, since the beginning of WORKS, but as of August, the other six counties have not made an EIP placement this year.

The process of adapting to the WORKS program is a partial explanation for the infrequent use of subsidized employment. Many WORKS caseworkers who transferred from other Department of Health and Human Resources units, such as Income Maintenance, lack experience in Work and Training and were initially unfamiliar with the use of subsidized employment programs. Conversion of cases to WORKS has required much of the county offices staff time.

Also, federal participation requirements make many employers hesitant to hire a welfare recipient under EIP. Since a recipient must work at least 20 to 35 hours per week, an employer who wants someone to work only 20 hours per week may not want to commit to additional hours of work. This problem also affects the ability to expand use of the CWEP program.

An EIP placement takes considerable time and effort for a caseworker since each placement must be individually negotiated with an employer. Furthermore, unsubsidized placements are preferable from the standpoint of the Department of Health and Human Resources. Since JTPA funding for subsidized employment is now gone, TANF funds must be used. Therefore, for budgetary reasons, an unsubsidized placement is more cost-effective and it is the ultimate goal of WORKS. Employer incentives are available for unsubsidized placements as well as subsidized placements. Even if an unsubsidized placement is made, an employer is still eligible for the Work Opportunity Tax Credit which amounts to as much as $2,100 per year. WORKS caseworkers do not have to make a more costly subsidized employment placement in order to provide an incentive to hire welfare recipients.

EIP is utilized far less frequently than work experience programs. This may be attributable to the nature of a subsidized employment placement. The goal of a subsidized position is to obtain a permanent job. Once the subsidized position becomes permanent no other recipients can be placed in that position. This is unlike CWEP, in which any number of recipients can eventually be placed into the same position. Public agencies may also require a number of CWEP placements at the same time. In many cases, caseworkers can place several recipients with the same CWEP employer. Multiple unsubsidized employment placements can sometimes be made with a single employer as well. The time-consuming process of arranging a single subsidized placement is, therefore, frequently less practical than the alternatives.

What Percent of Recipients are Participating in Employment?

Federal law requires States to meet certain participation rates or they stand to lose a portion of federal funding. Under TANF, 30% of all families receiving AFDC in FY 1998 are required to participate in a work activity, and 75% of two-parent families are required to be in work activities. A list of work activities is located in Appendix C. Unsubsidized employment, subsidized employment and work experience positions are among some of the work activities.

This analysis focuses only on the three work activities mentioned previously. It is not the intent to calculate the participation rate including all work activities. However, this analysis does examine if WORKS achieves a higher number of employment placements as a percent of all cases that are required to be in work activities. Under WORKS and under the former JOBS program, certain recipients are exempt from participating in work activities. The participation rate calculation involves dividing the number of individuals in work activities by the number of individuals required to be in work activities. This analysis makes this calculation only with respect to unsubsidized employment, subsidized employment and work experience. The calculations are shown below in Table 11.

Table 11
Unsubsidized Employment, Subsidized Employment and CWEP Positions
as A Percent of Cases with Work Requirements
(Monthly Average for Each Calendar Year)
19931994199519961997*
Greenbrier10.2%11.8%12.1%12.5%14.1%
Mercer4.7%5.5%4.9%6.0%10.7%
Monroe8.9%9.3%7.7%11.6%10.8%
Nicholas4.5%5.6%6.8%5.6%16.2%
Pocahontas9.7%11.4%13.3%9.8%15.4%
Tyler3.3%4.5%4.3%3.5%14.7%
Wayne6.8%7.6%6.8%6.1%10.9%
Wetzel5.4%4.9%3.7%3.6%15.6%
Wood6.0%6.3%6.7%4.9%2.6%
Source:Department of Health and Human Resources, Work Information System (WIS) Reports for selected months, and C-219 reports.
*Note: Figures for 1997 are for the period covering January through August.

Section 5


Welfare Cases Dropped by Nearly 50%

Since the beginning of the West Virginia WORKS program in January 1997, the total number of cash assistance cases has been roughly cut in half in each of the pilot counties. As Table 12 illustrates, there was a 46.8% reduction in cash assistance cases in the pilot counties when comparing totals for January 1997 and August 1997.

Table 12
Cash Assistance Cases for January & August of 1997

Pilot
Counties

Total Cases: January 1997

Total Cases: August 1997

Change in Number of Cases

Percent Change in Caseload
Greenbrier387166-221-57.1%
Mercer1,266646-620-49.0%
Monroe10746-61-57.0%
Nicholas539237-302-56.0%
Pocahontas7834-44-56.4%
Tyler17092-83-45.9%
Wayne820430-390-47.6%
Wetzel336155-181-53.9%
Wood1,215858-357-29.4%
Total4,9182,616-2,302-46.8%


Caseload Reductions have Stabilized at Current Levels

It appears that the decline in caseload has slowed as of July and August. Wood County has not completed converting all of its cases over to WORKS. Therefore, Wood County will likely see its caseload drop further as it converts its cases to WORKS completely. Figure 2 shows total and net case closures. During the period from January 1997 to April 1997, at the beginning of the program's implementation, a larger number of cash assistance cases were closed than in the following months. The peak number of closures occurred in February when 802 cases were closed. By May 1997, most existing cases had been converted to WORKS and, therefore, the number of case closures in the pilot counties decreased dramatically. In Mercer and Wayne counties, total cases increased slightly in July and August, while Monroe and Nicholas counties had cases increase slightly in August. When new cases are subtracted from closed cases, the results are net case closures. By August, new cases exceeded closed cases by 11. Overall, the case reduction appears to be stabilizing in the pilot counties with the exception of Wood County.


Were Case Closures Related to WORKS or the Economy?

Regression analysis shows that there was a downward trend in AFDC cases from 1993 through 1997 for all pilot counties. This is shown in Figure 3. The monthly average caseload increased slightly in 1994, however, it began declining in 1995. By the end of 1996, the average AFDC caseload dropped by nearly 800 cases over the three year period. The range of case reduction was as low as five cases a year in Pocahontas County to seven cases a month in Mercer County. However, under WORKS, the average caseload dropped by 1,855 in eight months. The unemployment rate was shown to be influential in AFDC case levels in all counties except Pocahontas. In Mercer County, 55 AFDC cases were associated with every 1% in the unemployment rate. With economic conditions represented by the unemployment rate, regression analysis shows that the average caseload under the first eight months of WORKS was substantially lower than the average caseload for the 1993-1996 period. The difference was statistically significant for every county at the 95% confidence interval, indicating that WORKS was a primary factor in the large caseload reduction during 1997.

Why Have Cash Assistance Cases Declined?

The dramatic decrease in the number of cases was caused by a number of factors. The major factors include the following.

Employment Placements were an estimated 25% of the total case reduction.
Recipients chose not to participate in 25% of the case closures. Recipients either requested to be taken off the welfare rolls, they did not appear in the office to complete paperwork, or they would not comply with the new requirements. Discussions with WORKS staff indicated that recipients chose to have their case closed for a variety of reasons, some of which included wanting to save as much of their five-year benefit limit for a more serious need, they were working secretly while collecting welfare, or they did not want to report an absent parent for child support.
New eligibility requirements made many recipients ineligible. The inclusion of Supplemental Security Income (SSI) and step-parent's income were major changes in eligibility determination. West Virginia is one of five States (Alabama, Arizona, Florida and Wisconsin) that includes SSI income in determining eligibility. Including this income has caused some West Virginia residents to apply for welfare in Virginia. Forty-eight new cases from West Virginia were opened in Virginia's western border counties due to the inclusion of SSI income under WORKS. The exact number of case closures resulting from new eligibility requirements is not known because of limitations in the data. Closures due to the inclusion of SSI income are estimated to be between 10% and 20% based on DHHR hand-counts for five of the first eight months of 1997.

The majority of case closures were cases that did not have a work requirement under the old welfare system. This is indicated in Figure 4. These types of cases dropped by 70% and represented 60% of the total number of cases closed. The bulk of these cases were closed because of SSI

income, customers requesting to have their case closed, or lack of compliance. As previously stated, many cases have been closed because of ineligibility, some did not want to comply, or some could do without welfare at the time. With these types of cases eliminated from the rolls, the Department of Health and Human Resources is left with cases that will prove more difficult to reduce through encouraging self-sufficiency.
Figure 5 shows major categories for case closures. The C-219 data system has numerous case closure categories that a caseworker can use to indicate the reason for a case being closed.


Given that there were 836 employment placements during the first eight months of 1997, and there were 3,111 cases closed, employment was about 27% of case closures. This is an approximation, the percent is likely lower because not every employment placement would have lead to a case being closed, especially if the job was part-time work. The exact number cannot be determined from the C-219 data system because there is no category for cases closed due to employment. It is likely that employment closures are recorded in the Excess Income category. However, this category could include other types of case closures.

Other major categories include Customer Request, Failure to Comply, and Failure to Appear. These reasons accounted for 25% of case closures. Customer request simply means that a recipient expressed the desire to have his or her case closed. Possible reasons for requesting case closure were previously cited as they were already working while receiving AFDC, they did not want to report an absent parent for child support, or they simply did not wish to participate in the WORKS program. The actual reason for the request is not known. Failure to comply or failure to appear could mean, for example, that a recipient failed to provide necessary information to approve their case or failed to come into the office when required.

The Composition of the Cases Has Changed Under WORKS

As the number of cases has fallen, the composition of the caseload has changed. The most important change has been the increase in the proportion of cases that require work activities. Prior to WORKS only 46% of the cases had work requirements,' under WORKS the percentage is 70 % (see graph on following page). A majority (60%) of case closures were cases which did not have a work requirement. Single parents (primarily mothers) without a work requirement under the old welfare system dropped by 70% (from 1,483 cases to 447) under WORKS. These individuals either accepted the new work requirements, found work, or refused to participate in the program. Proportionally, the greatest decline was 72% (from 1,059 to 294) of cases involving cases where children were determined eligible without including a parents income. Under the old guidelines if a parent was receiving SSI income, it would not be included in determining if a child was eligible for AFDC. Also, if a step-parent was involved, his or her income was not included. However, under WORKS, SSI income is included and step-parents are considered a part of the family unit, therefore, their income is included. The inclusion of SSI income and step-parent income resulted in a large number of cases being ineligible for cash assistance.

Single mothers required to work were only one-third of the total caseload under the former welfare system. Under WORKS the percentage is over half. Figures 6 and 7 on the following page illustrate the large increase in the percentage of cases that require work and those cases involving single mothers who are required to work. One contributing factor is that under the old welfare system, a single mother was exempt from work requirements if she was caring for a child three years of age or younger. Under WORKS, the work exemption applies when the child is one year old or younger. With the greater emphasis of getting these individuals to work, the agency may have to provide greater resources in the area of day care or transportation expenditures.




Expenditures on Welfare and Food Stamps Could Drop 33% to 55%
The total amount expended on AFDC and AFDC-U in fiscal year 1996 is reported in Table 13. Also, recipients who receive AFDC may be eligible for food stamps. The total amount expended on food stamps to those on AFDC is also shown in Table 13. However, food stamps allotments are 100% federally funded, therefore, there should be no fiscal impact to the State. These expenditures are expected to decline during fiscal year 1998 as the remaining counties implement WORKS. To provide an estimate as to how much of a decline in public assistance expenditures will occur, the assumption is made that case reductions will be similar to what was experienced in the pilot counties.

Regression analysis was conducted on monthly AFDC caseload and AFDC expenditures to determine the relationship between the two variables for each pilot county. The regression was done in logarithms to show the percentage change in expenditures due to a percentage change in AFDC caseload. The results show that the elasticities range between 0.94% and 1.14%. This means that for some counties a 1% change in AFDC cases will result in a change in AFDC expenditures between 0.9% and 1.14%. Seven of the eight counties had elasticity values greater than one. In general, for every 1% drop in AFDC cases, AFDC expenditures are expected to drop by close to 1.1%.

Therefore, if the pattern of case reduction occurs for the remaining counties which will be phased into WORKS, AFDC caseload could drop by 30% to 50% in fiscal year 1998 depending on the rate in which cases are converted to WORKS and on economic conditions. The total amount expended on AFDC and AFDC-U (for the unemployed) in Fiscal 1996 was $99,870,761. The Federal/State match was approximately 75% federal and 25% State. Expenditures for 1997 could not be used because they exclude RAPIDS data. Without knowing the amount of cash assistance for 1997, the figures cannot be estimated properly. However, if cash assistance is close to $100 million as it was in 1996, the cost reductions could be $33 million to $55 million, with the State's share being $8.3 million to $13.8 million.




Table 13
Expenditure Amounts for AFDC-U & Public Assistance Food Stamps
State Fiscal Year 1996

AFDC & AFDC-U
Public Assistance Food Stamps
July 19958,433,8329,293,916
August8,544,4859,347,216
September8,539,7129,413,568
October8,494,6849,715,920
November8,369,0769,590,357
December8,294,2969,276,663
January 19968,303,2239,524,742
February8,401,4179,568,931
March8,281,1829,482,253
April8,268,8569,425,921
May8,060,3679,215,052
June7,879,6319,024,185
Totals99,870,761112,878,724
Source:Office of Audit, Research and Analysis, Department of Health and Human Resource, C-219 data.


Although it is expected that the State will experience cost savings from case reductions, there are additional costs the State will incur that must be considered. For example, as WORKS is implemented Statewide, additional workers will be needed. The agency indicated that 18 new people were hired to implement WORKS in the pilot counties. Nine were strictly for WORKS and nine were hired to replace people who transferred from within the agency. The total cost was $422,499. As WORKS is implemented Statewide, additional workers will be needed. If the same ratio of workers and expenditures holds for the remaining 46 counties, then 92 new staff will be needed at a cost of $2.2 million. Also, it is expected that to promote self-sufficiency to the more difficult cases that remain, day-care expenditures and transportation subsidies will increase. See Appendix E for day-care expenditures and transportation costs.

Table 14 shows the change in welfare expenditures under WORKS for the first eight months of 1997 compared to the first eight months of 1996. As the figures indicate, total expenditures dropped by more than 35%.



Table 14
Total AFDC Expenditure & Public Assistance Food Stamps Expenditure

First Eight Months of 1997 Compared to First Eight Months of 1996

AFDC
1996

AFDC
1997
PA-Food Stamps
1996
PA-Food Stamps
1997
Greenbrier694,166377,957768,762428,812
Mercer2,277,0161,310,5742,541,4861,571,311
Monroe186,301105,903199,358122,382
Nicholas1,155,958590,8191,363,053701,084
Pocahontas138,05376,018148,91875,204
Tyler316,823213,856368,773268,620
Wayne1,543,964870,4241,768,227992,650
Wetzel706,848399,705812,377491,013
Wood2,264,5691,864,3372,572,6042,176,976
Totals9,283,6985,809,59310,543,5586,828,052
Percent
Change

-37.4%

-35.2%
Source:Office of Audit, Research and Analysis, Department of Health and Human Resource, C-219 data.