Work Opportunity Tax Credit: Employers Do Not Appear to Dismiss  
Employees to Increase Tax Credits (13-MAR-01, GAO-01-329).       
                                                                 
In 1997, 4,369 corporations earned a total of $135 million in    
Work Opportunity Tax Credits (WOTC). The employers who earned    
most of the credit were large companies with gross receipts      
exceeding $1 billion and engaged in nonfinancial services and    
retail trade. GAO's analysis of state agency data for California 
and Texas from 1997 through 1999 showed that three percent of    
participating employers accounted for 82 percent of all hires of 
WOTC-certified workers. Many employers who participated in the   
tax credit program in those two states in 1999 say that, besides 
the opportunity to obtain the credit, their participation in the 
program was also greatly influenced by such factors as the need  
to address a labor shortage and the opportunity to be a good     
corporate citizen. The results of GAO's two state analysis       
indicate a low probability of replacing employees who were not   
eligible for the tax credit.
                                     
-------------------------Indexing Terms------------------------- 
REPORTNUM: GAO-01-329                                                 
ACCNO: 577854                                                     
TITLE:   Work Opportunity Tax Credit: Employers Do Not Appear to  
Dismiss Employees to Increase Tax Credits                        
DATE: 03/13/2001 
SUBJECT                                         
                                                                 
             Disadvantaged persons                               
             Employee dismissal                                  
             Personnel recruiting                                
             Surveys                                             
             Tax credit                                          
             Welfare recipients                                  
             California                                          
             Supplemental Security Income Program                
             Targeted Jobs Tax Credit Program                    
             Temporary Assistance for Needy Families             
             Program                                             
             Texas                                               
             Work Opportunity Tax Credit                         

******************************************************************
** This file contains an ASCII representation of the text of a  **
** GAO Testimony.                                               **
**                                                              **
** No attempt has been made to display graphic images, although **
** figure captions are reproduced.  Tables are included, but    **
** may not resemble those in the printed version.               **
**                                                              **
** Please see the PDF (Portable Document Format) file, when     **
** available, for a complete electronic file of the printed     **
** document's contents.                                         **
**                                                              **
******************************************************************
GAO-01-329
     
Report to the Chairman, Subcommittee on Oversight, Committee on Ways and
Means, House of Representatives

United States General Accounting Office

GAO

March 2001 WORK OPPORTUNITY TAX CREDIT

Employers Do Not Appear to Dismiss Employees to Increase Tax Credits

Page i GAO- 01- 329 Work Opportunity Tax Credit Letter 1

Appendix I Objectives, Scope, and Methodology 23

Appendix II Survey Instruments 27

Appendix III Statistical Analysis of State Databases 39

Appendix IV GAO Contacts and Staff Acknowledgments 42

Tables

Table 1: Distribution of Businesses Earning WOTC and the Amount of Credit
Earned by Size of Gross Receipts, Nationwide in 1997 9 Table 2: Distribution
of Businesses Earning WOTC and the Amount

of Credit Earned, by Industry, Nationwide in 1997 10 Table 3: Distribution
in Texas and California of Number of

Employers, Number of WOTC Hires, and Average Time in the Program, by Size of
Employer, 1997 -99 11 Table 4: Number of Population and Sample Cases in
Strata for

Employers Hiring WOTC- Eligible Employees 24 Table 5: Logistic Regression
Analysis of the Likelihood of

Separation of Employees in California and Texas 40

Figures

Figure 1: Factors Greatly Influencing Employer Participation in California
and Texas 12 Figure 2: Reported Reasons for Vacancies Filled by WOTC-
Eligible

Hires in California and Texas 15 Figure 3: Percentages of Participating
Employers in California and

Texas Who Changed Recruitment Practices to Secure the Credit 16 Figure 4:
Percentages of Participating Employers in California and

Texas Who Changed Hiring Practices to Secure the Credit 17 Contents

Page ii GAO- 01- 329 Work Opportunity Tax Credit

Figure 5: Percentages of Participating Employers in California and Texas Who
Changed Training Practices to Secure the Credit 18 Figure 6: Employment
Rates of Targeted and Nontargeted Welfare

Recipients, 1995- 99 19 Figure 7: Separation Rates for Certified WOTC
Employees in

California and Texas, by Earnings, 1997- 99 20 Abbreviations AFDC Aid to
Families With Dependent Children CPS Current Population Survey EIN Employer
identification number IRS Internal Revenue Service TANF Temporary Assistance
for Needy Families WOTC Work Opportunity Tax Credit

Page 1 GAO- 01- 329 Work Opportunity Tax Credit

March 13, 2001 The Honorable Amo Houghton Chairman, Subcommittee on
Oversight Committee on Ways and Means House of Representatives

Dear Mr. Chairman: The Work Opportunity Tax Credit (WOTC) is intended to
increase the employment and earnings of workers belonging to certain
disadvantaged groups by providing employers with an incentive to hire and
retain these workers. However, as government and academic analysts have
pointed out, the credit may also create incentives for employers to dismiss
workers who do not make the employer eligible for the credit. Specifically,
employers can increase their tax credit by dismissing workers for whom the
employer has never received a tax credit in order to hire other workers who
make the employer eligible for a credit; this practice is called
displacement. Employers can also increase their tax credit by dismissing
workers whose eligibility for the credit has ended and hiring other workers
who make the employer again eligible for a credit; this practice is called
churning.

Little is known about the employers who participate in the WOTC program or
about participating employers? motivations for hiring and retaining workers
for whom the employers can receive a tax credit. As agreed with your office,
our objectives in this study were to determine (1) the characteristics,
including motivations, of the employers who have participated in the credit
program and (2) the extent, if any, to which employers have practiced
displacement or churning.

To determine the characteristics of employers, we analyzed nationwide data
from IRS on the use of the WOTC and data from agencies in California and
Texas on the number of WOTC- certified employees hired by each employer. 1
To obtain information relating to the extent of churning and displacement,
we (1) surveyed a probability sample of employers who

1 State agencies are responsible for determining the eligibility of
individuals as members of targeted groups and issuing certifications of
eligibility to the employers of those individuals.

United States General Accounting Office Washington, DC 20548

Page 2 GAO- 01- 329 Work Opportunity Tax Credit

participated in the credit program in California and Texas and (2) analyzed
data from agencies in these two states on the actual earnings and employment
histories of certified employees. We chose California and Texas primarily
because they are among the states that certified the largest number of
employees to participate in the WOTC program in fiscal year 1999 and have
electronic databases of their WOTC program data. We did not evaluate the
effectiveness or efficiency of the WOTC in promoting the hiring of
disadvantaged individuals. For example, we did not determine the extent to
which employers may have received ?windfall?

credits for employees whom they would have hired anyway. Our methodology is
described further in the scope and methodology section of this letter and in
appendix I. Our survey instruments are reproduced in appendix II.

Large employers earned most of the credit and hired most of the employees
under the Work Opportunity Tax Credit program. In 1997, 4,465 corporations
earned a total of $135 million in tax credits. The employers who earned most
of the credit were large companies, with gross receipts of $1 billion or
more, and engaged in nonfinancial services and retail trade. Our analysis of
state agency data for California and Texas from 1997 through 1999 showed
that 3 percent of participating employers accounted for about 83 percent of
all hires of WOTC- certified employees. According to our survey, many
employers who participated in the credit program in those two states in 1999
said that, besides the opportunity to obtain the credit, their participation
in the program was also greatly influenced by such factors as the need to
address a labor shortage (an estimated 36 percent) and the opportunity to be
a good corporate citizen (an estimated 41 percent).

Although we were unable to definitively determine the extent to which
displacement and churning occur, our review suggests that churning is likely
to be very limited, if it occurs at all. Our survey of employers in two
states indicates that 93 percent of participating employers said that
displacement and churning have little or no cost- effectiveness. According
to our survey, those employers estimated that the WOTC offset, on average,
47 percent of employers? costs of recruiting, hiring, and training certified
workers. If employers were practicing churning, it would make the most sense
for them to dismiss their WOTC- certified employees near the earnings level
($ 6,000) that would yield the maximum credit. Data from agencies in
California and Texas on the employment of WOTCcertified employees showed
that their employment rarely ends near that earnings level. These data also
showed that certified workers with Results in Brief

Page 3 GAO- 01- 329 Work Opportunity Tax Credit

earnings within plus or minus $1, 000 of the $6,000 credit- maximizing level
were no more likely to separate from their employers than other certified
workers, whose earnings fell short of or exceeded this range. The agency
data did not allow us to perform similar tests for the occurrence of
displacement, but we would not expect employers to undertake a practice that
they said was not cost- effective. Because the results of both our survey
and our analyses of employment data were similar for California and Texas,
and because many of the firms employing the bulk of WOTC employees in these
two states operate in multiple states, we believe the results of our two-
state analysis indicate a low probability of displacement and churning in
other states as well.

The WOTC is intended to encourage employers to hire individuals from eight
targeted groups that have consistently high unemployment rates. The targeted
groups are

 individuals in families currently or previously receiving welfare benefits
under the Temporary Assistance for Needy Families (TANF) program or its
precursor, the Aid to Families With Dependent Children (AFDC) program;

 veterans in families currently or previously receiving assistance under a
food stamp program;

 food stamp recipients- aged 18 through 24 years- in families currently or
previously receiving assistance under a food stamp program;

 youth- aged 18 through 24 years- who live within an empowerment zone or
enterprise community; 2

 youth- aged 16 and 17 years- who live within an empowerment zone or
enterprise community and are hired for summer employment only;

 ex- felons in low- income families;

 individuals currently or previously receiving Supplemental Security
Income; and

 individuals currently or previously receiving vocational rehabilitation
services.

2 Empowerment zone and enterprise community refer to an area or combination
of areas designated by the U. S. Department of Housing and Urban Development
or the U. S. Department of Agriculture that meet certain population, size,
and poverty criteria. Effective for the period beginning January 1, 2002,
the Community Renewal Tax Relief Act of 2000 (P. L. 106- 554) expands this
target group and the target group of youth employed during the summer to
include qualified individuals who live in a renewal community that is
designated by the Secretary of Housing and Urban Development. Background

Page 4 GAO- 01- 329 Work Opportunity Tax Credit

Additional eligibility criteria apply to these groups. For example, welfare
recipients must have received AFDC or TANF benefits for any 9 months during
the 18- month period ending on the hiring date in order to be eligible for
the program.

The amount of tax credit that employers can claim under this program depends
upon how long they retain credit- eligible employees and the amount of wages
they pay to WOTC- certified employees. Employers who retain certified
employees for at least 120 but less than 400 hours qualify for a credit of
25 percent of up to $6,000 in wages, for a maximum credit of $1,500.
Employers who retain certified employees for 400 hours or more qualify for a
credit equal to 40 percent of up to $6,000 in wages, for a maximum credit of
$2,400. 3 The credit is calculated using the actual first year wages paid or
incurred. Employers must reduce their tax deductions for wages and salaries
by the amount of the credit. In addition, as part of the general business
credit, the WOTC is subject to a yearly cap. 4 However, excess WOTC can be
used to offset tax liabilities in the preceding year or in any of 20
succeeding years.

The WOTC was first authorized in the Small Business Job Protection Act of
1996 to improve upon and replace a similar, expired program- the Targeted
Jobs Tax Credit program. 5 The WOTC was designed to mitigate some
shortcomings that had been identified in the previous credit program-
specifically, that it gave employers windfalls for hiring employees that
they would have hired anyway and that too many crediteligible employees left
their jobs before they received much work experience. Some target groups
were reformulated with the intention of focusing narrowly on those who truly
need a credit for firms to risk hiring them. In addition, the minimum
employment period for receiving the higher rate of credit was lengthened.
The WOTC became effective

3 The tax credit for youths in summer jobs has a lower wage limit than that
for the other target groups. See appendix III for a detailed description of
the limit on credit- eligible wages for youths in summer jobs.

4 The general business credit combines several tax credits, including the
WOTC, for the purpose of computing an overall dollar limitation on the
reduction of tax liability. The general business credit may not exceed net
income tax minus the greater of (1) the tentative minimum tax or (2) 25
percent of the net regular tax liability above $25, 000.

5 Small Business Job Protection Act (P. L. No. 104- 188).

Page 5 GAO- 01- 329 Work Opportunity Tax Credit

beginning in October 1996 and has since been reauthorized. It is due to
expire in December 2001.

In fiscal year 1999, 335,707 individuals were certified as members of the
targeted groups, making their employers eligible for the credit if the
workers remained on the job for at least 120 hours. Individuals in the
welfare target group made up 54 percent of the individuals certified. Youth
in the food stamp target group made up another 20 percent of the individuals
certified. The other six target groups each accounted for 1 to 8 percent of
the remaining certifications.

Federal and state agencies share responsibility for administering the WOTC
program. The Department of the Treasury, through the Internal Revenue
Service (IRS), is responsible for the tax provisions of the credit. The
Department of Labor, through the Employment and Training Administration, is
responsible for developing policy and program guidance and providing
oversight of the WOTC program. In addition, the Department of Labor awards
grants to states for administering the eligibility determination and
certification provisions of the program. State agencies verify and report to
the Department of Labor on state certification activities. All 50 states,
the District of Columbia, Puerto Rico, and the U. S. Virgin Islands
participate in the program. Neither the Department of the Treasury nor the
Department of Labor regulations require these agencies to take any actions
regarding displacement or churning.

The State of New York and the Department of Labor have undertaken studies
that may have findings relevant to whether employers engage in displacement
or churning practices. The New York study, which was issued in 1998,
concluded, among other things, that employer windfalls from churning
employees are minimal. This conclusion was based on analysis of state WOTC
and Wage Reporting databases with records on 12,609 individuals in New York
covering the fourth quarter of 1996 through the first quarter of 1998. The
study did not address displacement. The Department of Labor study is
ongoing, so its results are not yet available. The study is using in- depth
interviews with 16 employers who hire a large number of employees under the
WOTC program to examine the hiring, retention, and career advancement
experiences of WOTC employers and employees.

Page 6 GAO- 01- 329 Work Opportunity Tax Credit

To obtain information on the characteristics of employers, we analyzed
national tax data from the IRS? Statistics of Income Division for 1997, the
most recent year that data were available, and state WOTC data from agencies
in California and Texas for 1997 through 1999. To obtain information
relating to the extent of displacement and churning, we surveyed a
stratified probability sample of employers who have participated in the WOTC
program in California and Texas. The participating employers that we
surveyed are those with repeated and recent experience in the program in
that they hired at least one WOTC employee in 1999 and hired at least one
WOTC employee in another year. Our sample is projectible to the entire
population of 1,838 employers in California and Texas who met these hiring
criteria.

For information relating to churning, we also analyzed WOTC and unemployment
insurance data for these states. With these data, we determined the total
earnings and length of employment of WOTCcertified employees and examined
this information for evidence concerning the extent and likelihood of
churning.

For additional information relating to displacement, we analyzed national
employment data in the Commerce Department?s Current Population Survey (CPS)
for 1995 through 1999. We used the CPS data to estimate employment rates for
members of groups targeted by the credit and members of groups not targeted
by the credit but who may substitute in employment for target group members.
The absence of a centralized database containing the necessary detailed
information precluded a nationwide survey of employers and analysis of
employment practices.

We chose California and Texas because they are among the states that

 certified the largest number of employees to participate in the WOTC
program in fiscal year 1999,

 have electronic databases of their WOTC program data, and

 provided a somewhat geographically diverse population. Together,
California and Texas certified about 12 percent of WOTCeligible individuals
in fiscal year 1999, ranking them second and fifth, respectively, in WOTC
certifications for that fiscal year. When reporting our estimates derived
from the sample and our analysis of program and unemployment insurance data,
we combined data from both states because the results in the two states were
similar. Furthermore, the confidence intervals for all point estimates in
the letter of this report are no more than 10 percentage points on either
side of the estimate. Scope and

Methodology

Page 7 GAO- 01- 329 Work Opportunity Tax Credit

Our survey and state agency data pertain only to participating employers in
California and Texas. However, to assure ourselves that our findings are
likely to apply to WOTC employers in the rest of the nation, we examined the
federal laws and regulations related to the credit, surveyed state
administrators responsible for the credit, and analyzed the data on
participating employers. The federal tax benefits offered by the WOTC are
the same across all states. Therefore, we have no reason to believe that
employers in California and Texas respond differently to these incentives
than employers in other states. We spoke to the officials who were
responsible for administering the WOTC program in all 50 states, and they
all confirmed that their states made no effort to either encourage or
discourage displacement or churning. From the participating employer data,
we determined that employers who operate in multiple states account for most
of the WOTC hires in California and Texas. Moreover, we found no differences
relevant to churning and displacement between employers in California and
Texas in the results of our survey and agency data analyses, suggesting that
our conclusions would be generalizable to employers in other states as well.

We did not evaluate how effective or efficient the WOTC has been in
increasing the employment and earnings of target group members. To do this,
we would have had to determine the extent to which

 the credit caused employers to hire workers that they would not otherwise
have hired,

 employees? experience with WOTC employers increases their current and
future earnings, and

 employers received ?windfall? credits for employees whom they would have
hired anyway.

We did not address any of those issues in this report. We did not verify the
state and federal databases we used. However, agreements between the
Department of Labor and state WOTC offices require the states to conduct
audits of the accuracy of their WOTC records. A review of studies of the
accuracy of unemployment insurance data, which was conducted for the
National Research Council, concluded that the data appear to be accurate.
The study notes that employers are required by law to report the data and
that intentional inaccuracies are subject to penalties. This same review of
studies found that the CPS data are a valuable source of information on the
national low- income population, with broad and fairly accurate measures of
income. However, the study noted that sample sizes may be small for some
subpopulations

Page 8 GAO- 01- 329 Work Opportunity Tax Credit

(e. g., welfare recipients in particular states), and the percentage of some
subpopulations covered by the survey appears to have declined modestly in
recent years. The tax data from IRS? Statistics of Income Division undergo
numerous quality checks but do not include information from amended tax
returns.

We conducted our review from January 2000 through December 2000 in
accordance with generally accepted government auditing standards. Our scope,
methodology, and the sources of the data we used are discussed further in
appendix I. We requested comments on a draft of this report from the
Department of Labor and asked cognizant agencies in California and Texas to
review the draft?s discussion of their WOTC efforts. The comments are
discussed near the end of this letter.

Employers who were large in terms of gross receipts earned most of the
credit reported in 1997, the latest year for which data were available. Data
from the agencies that certify WOTC employees in California and Texas showed
that a relatively small number of employers did most of the hiring in the
WOTC program from 1997 through 1999. Employers? participation in the program
was greatly influenced by such factors as the opportunity to obtain the
credit, address a labor shortage, and be a good corporate citizen.

In 1997, nationwide, an estimated 4,465 corporations earned an estimated
total of $134.6 million in tax credits. 6 Approximately 66 percent of the
credit was earned by corporations with gross receipts of $1 billion or more.
Table 1 shows the amount of credit that businesses earned by amount of gross
receipts.

6 The credit data we present include the amount earned by subchapter S
corporations. A subchapter S corporation is treated similarly to
partnerships for federal income tax purposes. Shareholders claim the credit
on their individual tax returns. Aside from these shareholders, a negligible
amount of credit was earned by individual taxpayers. Large Employers

Made the Most Use of the Credit

Large Employers in a Few Industries Earned Most of the Credit

Page 9 GAO- 01- 329 Work Opportunity Tax Credit

Table 1: Distribution of Businesses Earning WOTC and the Amount of Credit
Earned by Size of Gross Receipts, Nationwide in 1997

Dollars in millions

Businesses Credit earned Gross receipts Number Percent Amount Percent

$0 - less than $1 1,577 35.9 $6.5 4. 8 $1 - less than $10 1, 494 33.5 7. 0
5.2 $10 - less than $50 643 14.4 7. 0 5.2 $50 - less than $100 133 3.0 2. 3
1.7 $100 - less than $250 160 3.6 5. 5 4.1 $259 - less than $500 118 2.7 8.
4 6.2 $500 - less than $1, 000 106 2.4 9. 4 7.0 $1,000 and greater 232 5.2
88.5 66.3

Total 4,465 100.0 $134.6 100.0

Note: Columns may not sum to totals because of rounding. Source: GAO
analysis of IRS data.

Most of the credit was reported by businesses engaged in nonfinancial
services, such as hotel, motel, and other personal services, and retail
trade. These industries accounted for 81 percent of the credit reported.
Table 2 shows the credit amounts earned by businesses in each industry in
1997. The aggregate amount of WOTC earned by taxpayers is likely to have
grown significantly between 1997 and 1999 because the number of WOTC
certifications grew significantly nationwide over that period- from 126,113
to 335,707. However, based on the certification data we have from California
and Texas, we believe that the percentage distribution of the credit by size
of employer and by industry has not changed dramatically. The size
distribution of employers measured by number of WOTC hires did not change
significantly in either California or Texas during that period. The
distribution of certifications by industry also changed little in Texas; we
do not have industry information for California.

Page 10 GAO- 01- 329 Work Opportunity Tax Credit

Table 2: Distribution of Businesses Earning WOTC and the Amount of Credit
Earned, by Industry, Nationwide in 1997

Dollars in millions

Businesses Credit earned Industry Number Percent Amount Percent

Construction 760 17.0 $2.7 2. 1 Manufacturing 649 14.5 15.8 11.8
Transportation and public utilities 25 0.6 3. 6 2.7 Wholesale trade 126 2.8
1. 3 1.0 Retail trade 1,057 24.1 75.0 55.8 Finance, Insurance and Real
Estate 342 7.7 2. 0 1.5 Nonfinancial Services 1,480 33.2 33.7 25.0 Other 9
0. 2 0.3 0. 2

Total 4,465 100.0 $134.6 100.0

Note: Columns may not sum to totals because of rounding. Source: GAO
analysis of IRS data.

Our analysis of WOTC certification data in California and Texas for 1997
through 1999 showed that a few employers did most of the hiring in the WOTC
program. Employers who hired more than 100 WOTC- certified employees
represented about 3 percent of all employers in the program but accounted
for about 83 percent of all hires. About 65 percent of employers in the
program made only one WOTC hire.

The larger WOTC employers spent more time in the program. Employers who
hired more than 100 WOTC- certified employees were in the program for an
average of 10 or more quarters, while those hiring 5 or fewer employees were
in the program for an average of less than 3 quarters. The larger WOTC
employers also hired more frequently. Employers who hired in every year
accounted for about 83 percent of total hires while representing about 8
percent of all employers. Table 3 shows the distribution of the number of
employers, the number of WOTC- certified employees, and time in the program,
by size of employers (in terms of WOTC- certified hires) for 1997 through
1999. A Relatively Small Number

of Employers Accounted for Most WOTC Hires in California and Texas

Page 11 GAO- 01- 329 Work Opportunity Tax Credit

Table 3: Distribution in Texas and California of Number of Employers, Number
of WOTC Hires, and Average Time in the Program, by Size of Employer, 1997-
99

Employers WOTC hires Size in terms of WOTC hires Number Percent Number
Percent Average

time a

1 4,220 64.5 4, 220 2.6 0. 2

2 - 5 1,265 19.3 3, 573 2.2 2. 8

6 - 10 279 4.3 2, 210 1.4 5. 8

11 - 15 133 2.0 1, 708 1.1 6. 7

16 - 20 97 1.5 1, 734 1.1 7. 4

21 - 25 53 0.8 1, 207 0.7 7. 0

26 - 50 159 2.4 5, 661 3.5 8. 6

51 - 100 116 1.8 8, 113 5.0 9. 3

101 - 200 101 1.5 14,150 8. 7 10.4

201 - 500 63 1.0 20,189 12.4 10.8

Greater than 500 59 0.9 99,823 61.4 12.1 Total 6,545 100.0 162,588 100.0 1.
7

Note: Totals may not sum because of rounding. a Time in quarters.

Source: GAO analysis of California and Texas WOTC databases.

The employers that we surveyed in two states reported that the opportunity
to obtain a tax credit was by far the factor that most influenced their
decisions to participate in the WOTC program, followed by the opportunity to
address labor shortages and be a good corporate citizen. According to our
survey, the opportunity to obtain the credit was the largest influence, with
an estimated 85 percent of participating employers in California and Texas
saying they were greatly influenced by this opportunity. Figure 1 shows the
extent to which employers in the states we reviewed said that specific
factors greatly influenced their participation in the program. Several
Factors Influenced

Participation in the WOTC Program

Page 12 GAO- 01- 329 Work Opportunity Tax Credit

Figure 1: Factors Greatly Influencing Employer Participation in California
and Texas

Source: GAO analysis of survey data.

Participation in the program appears often to have had support from high
levels within the companies. For example, for an estimated 57 percent of
California and Texas employers, the possibility of participating in the
program was raised by someone inside the company rather than by an outside
organization. In those situations, high- level management was responsible
for raising the idea of participating in the WOTC program about three-
quarters of the time, according to our survey- based projections.

84.6 41

36.4 27

11.8 0 10

20 30

40 50

60 70

80 90

100 Obtain tax credit

Be good corporatecitizen Address labor shortage

Positive past experience with the target groups

Influence of communityleaders Estimated percentage

Reason for participation

Page 13 GAO- 01- 329 Work Opportunity Tax Credit

Displacement and churning are likely to be limited, if they occur at all,
because, as our survey of employers in California and Texas indicates, most
employers view these practices as having little or no costeffectiveness.
This view is consistent with the employers? estimate that the credit offsets
less than half the costs of recruiting, hiring, and training credit-
eligible employees. Our employer survey also indicates that most vacancies
filled by credit- eligible employees occur for reasons unrelated to
displacement and churning, such as voluntary separations. Furthermore, our
survey indicates that most employers change at least one recruitment,
hiring, or training practice, which, studies suggest, may make these
employers more likely to retain new hires. Our analysis of program and
employment data from state agencies supports what we learned from the survey
regarding the low probability of churning. These data show that employment
rarely ends near the earnings level that yields the maximum credit, and
employees earning the maximum are no more likely to separate than are other
WOTC- certified employees. The agency data do not allow us to perform
similar tests for the occurrence of displacement. However, displacement is
less likely to occur when employers are increasing their workforce- as has
been the case since the introduction of the credit- because they have less
need to dismiss non- WOTC workers in order to hire WOTC workers.

Most employers do not consider displacement and churning to be costeffective
employment practices. Based on our survey, we estimate that 93 percent of
participating employers in California and Texas would agree that
displacement is cost- effective to little or no extent. An estimated 93
percent of employers also hold that view regarding churning. Displacement
and churning are not cost- effective if the cost of recruiting, hiring, and
training a new employee exceeds the amount of WOTC that an employer expects
to earn from that employee. Under those circumstances, the WOTC provides no
incentive for that employer to dismiss an existing employee to hire a WOTC-
certified one. According to our employer survey, on average, the tax credit
offsets less than one- half (47 percent) of this If Displacement and

Churning Occur, They Are Likely to Be Very Limited

California and Texas Employers View Displacement and Churning as Not
CostEffective

Page 14 GAO- 01- 329 Work Opportunity Tax Credit

cost. 7 Furthermore, employers told us that it is important to reduce the
turnover of WOTC- certified employees. Based on our survey, we estimate that
for 71 percent of participating employers in those two states, retaining
employees after the maximum tax credit has been secured is very important.
An additional 20 percent would view retention of employees after the maximum
tax credit is secured as somewhat important.

For those employers who could tell us the reasons for the vacancies that
were filled by WOTC- certified employees, an estimated average of 61 percent
of such vacancies arose because the previous employees quit. On average, the
next most frequent reasons for the vacancies were that the previous
employees were terminated for cause and that the positions were newly
created. Figure 2 shows the distribution by California and Texas employers?
responses regarding the reasons for vacancies. None of the reasons given
were related to displacement or churning.

7 Although the WOTC does not appear to provide an incentive for employers to
purposely create vacancies for credit- eligible employees by dismissing
existing employees, the credit may, nevertheless, provide an incentive for
employers who already have vacancies to fill them with credit- eligible
employees. According to our survey, the difference between the average cost
of recruiting, hiring, and training a credit- eligible employee ($ 3,799)
and the average cost of the same activities for a noneligible employee ($ 3,
265) is not statistically significant. Therefore the credit may more than
offset any additional cost associated with choosing a credit- eligible
individual over a noneligible individual to fill an existing vacancy. (The
maximum credit is $2, 400, but most WOTC- certified employees leave their
jobs well before the maximum credit is reached. See the data presented later
in this report). We did not attempt to determine whether the WOTC is
effective in encouraging employers to hire individuals from the target
populations (see the scope and methodology section of this report). However,
we did determine that, for an estimated 70 percent of participating
employers in the two states, the tax credit?s insufficiency to offset
recruitment, hiring, and training costs has little or no deterrent effect on
employer willingness to fill vacancies with credit- eligible individuals.
Participating Employers

Reported, on Average, That Most Vacancies Result From Voluntary Separations

Page 15 GAO- 01- 329 Work Opportunity Tax Credit

Figure 2: Reported Reasons for Vacancies Filled by WOTC- Eligible Hires in
California and Texas

Note 1: We asked each employer what percentage of these vacancies was due to
each of the reasons listed in the figure. We then averaged those percentages
across responding employers.

Note 2: The ?other? category included responses such as ?seasonal hiring
buildup? and ?additional

workforce needed.? Source: GAO analysis of survey data.

About 85 percent of employers in California and Texas have changed a
recruiting, hiring, or training practice to secure the WOTC and better
prepare credit- eligible new hires, according to estimates that are based on
employer- reported information from our survey. Furthermore, an estimated 43
percent of employers in these two states have changed their practices in all
three of these areas. A 1999 study conducted by Jobs for the Future found
that employers who successfully employed welfare recipients- which is the
largest targeted group in the WOTC program- developed strategies to improve
access, retention, and advancement of Most Participating

California and Texas Employers Changed Recruiting, Hiring, or Training
Practices

Page 16 GAO- 01- 329 Work Opportunity Tax Credit

those individuals. 8 The strategies used by employers in our survey included
targeted recruitment; outreach and screening assistance from organizations
that know and understand the targeted group; preemployment training, such as
training in communication skills; and mentors, among other strategies. These
strategies are consistent with ones these researchers identified in other
studies.

Based on the results of our survey, we estimate that about two- thirds of
participating employers in the two states changed at least one recruitment
practice to secure the tax credit. The most frequent change in recruitment
practice was that employers listed job openings with a public agency or
partnership. An estimated 49 percent of participating employers in the two
states took such an action. Figure 3 shows the extent to which participating
employers changed recruitment practices to secure the credit.

Figure 3: Percentages of Participating Employers in California and Texas Who
Changed Recruitment Practices to Secure the Credit

Source: GAO analysis of survey data.

8 Business Participation in Welfare- to- Work: Lessons from the United
States, (Boston: Jobs for the Future, 1999). Jobs for the Future is a
national nonprofit organization that focuses on workforce development
issues.

5 12.1

11.9 12.9

29.1 33.8

42.6 48.8

020 4060 80100 Requested referrals of applicants

Partnered with agency to identify applicants Partnered with agency to screen
applicants Partnered with agency to prepare applicants

for employment Changed recruitment literature

Set up separate WOTC recruitment process Changed wording in job listings

Estimated percentage Type of change Listed job openings with public agency

or partnership

Page 17 GAO- 01- 329 Work Opportunity Tax Credit

An estimated three- quarters of participating employers in the two states
changed at least one hiring practice to secure the tax credit. Our survey
indicated that the most frequent change in hiring practices was that
employers began training their managers about the tax credit, with an
estimated 66 percent of employers making that change. Figure 4 shows the
extent to which participating employers changed hiring practices to secure
the credit.

Figure 4: Percentages of Participating Employers in California and Texas Who
Changed Hiring Practices to Secure the Credit

Source: GAO analysis of survey data.

Based on our survey, we estimate that about one- half of participating
California and Texas employers changed at least one training practice to
better prepare WOTC new hires. For example, an estimated 40 percent began
providing mentors to their new hires. Figure 5 shows the extent to which
participating employers changed training practices to secure the credit.

4.6 20.7

20 21.6

9.8 40.3

48.7 65.8

0 20 40 60 80 100 Estimated percentage Trained managers about WOTC

Used consultants to process WOTC paperwork

Changed or modified application packet Identified specific positions for
WOTC hiring

Offered hiring incentives to managers Set up separate WOTC hiring process
Changed or modified hiring standards

Used agency to conduct hiring process Type of change

Page 18 GAO- 01- 329 Work Opportunity Tax Credit

Figure 5: Percentages of Participating Employers in California and Texas Who
Changed Training Practices to Secure the Credit

Source: GAO analysis of survey data.

Displacement is less likely to occur when employers are increasing their
workforce because they have less need to dismiss non- WOTC workers in order
to hire WOTC workers. Since the introduction of the credit in the last
quarter of 1996, employment in the U. S. economy has grown robustly, even
for low- skilled workers. Using the CPS data, we found that employment rates
grew over the period for certain target group members and closely related
nontarget group members that may substitute in employment for the target
groups. For example, we estimated employment rates for welfare recipients in
the CPS (those on welfare for 9 or more months in the previous year) who
would be members of the group targeted by the credit. 9 We also estimated
employment rates for welfare recipients who would not be target group
members (those on welfare less than 9 months of the previous year). The
employment rate of the target group welfare recipients grew by 47 percent
and nontarget welfare recipients by 12 percent from 1995 through 1999.
Figure 6 shows employment rates over the period for members of the targeted
and nontargeted welfare groups.

9 The welfare target group is members of families receiving benefits under
TANF or AFDC for any 9 of the last 18 months, ending on the hiring date. The
data elements in the CPS permit us to identify most, but not all, of the
welfare target group members in the CPS database. Robust Employment

Growth Makes Displacement Less Likely

5 6.7

20.9 33.2

38 39.9

020 4060 80100 Estimated percentage Provided mentors

Provided basic work readiness Lengthened training times Changed training
materials Used an agency to provide additional

training or mentors Set up separate WOTC training process

Type of change

Page 19 GAO- 01- 329 Work Opportunity Tax Credit

Figure 6: Employment Rates of Targeted and Nontargeted Welfare Recipients,
1995- 99

Source: GAO analysis of CPS data.

Our analysis of the WOTC and unemployment insurance data in California and
Texas showed that most certified employees do not earn enough income while
working for WOTC employers for churning to make sense for those employers.
Sixty- seven percent of certified employees separated from their employers
after earning less than $3,000. Furthermore, only a relatively small number
of certified employees earned incomes in the range where churning may be
most likely to occur. Employers wishing to maximize their credit would
retain WOTC employees until they had earned a total of $6,000, the maximum
earnings eligible for the credit. Only about 7 percent of certified
employees separated after earning incomes between $5,000 and $7,000 (a range
of earnings within $1,000 of the credit maximizing level). If employers did
not churn when employees reached this level of earnings, it seems less
likely that they would churn at other levels of earnings. Figure 7 shows the
percentage of employees separating after earning a given amount of income.
Relatively Few WOTC

Hires Are Still Employed When Churning Is Most Likely

Page 20 GAO- 01- 329 Work Opportunity Tax Credit

Figure 7: Separation Rates for Certified WOTC Employees in California and
Texas, by Earnings, 1997- 99

Note: The figure does not include the 9 percent of employees who earned
$10,000 or more while working for the WOTC employer. Including these
employees would require extending the figure with numerous additional $1,000
total earnings categories. However, the shape of the figure would be
unchanged because the certified employees in each of these additional
categories would represent less than 2 percent of total certified employees.

Source: GAO analysis of Texas and California WOTC and Unemployment Insurance
databases.

In addition to determining the percentage of WOTC- certified employees who
separated near the maximum earnings level, we also analyzed the effect of
reaching the maximum earnings level on the likelihood of separation. We used
a statistical technique to measure the likelihood of separation of WOTC-
certified employees who reach the maximum earnings level in a given quarter
relative to the likelihood of separation of WOTC- certified employees who do
not reach the maximum. The technique that we used allows us to measure the
effect on the likelihood of separation, while controlling for the effects of
other employee characteristics, such as membership in a particular target
group. The measured effect is, therefore, the net effect on the likelihood
of separation (i. e., net of the effects of the other characteristics).

Using this technique, our analysis showed that WOTC- certified employees who
reach the maximum earnings in a given quarter (i. e., those whose Employees
Who Reached

the Maximum Earnings in California and Texas Were No More Likely to Separate
From WOTC Employers Than Those Who Do Not

Page 21 GAO- 01- 329 Work Opportunity Tax Credit

cumulative earnings are between $5,000 and $7,000) are no more likely to
separate from their WOTC employers than those employees who do not reach the
maximum. In addition, the analysis showed that reaching the maximum has no
effect on the likelihood of separation across most target groups. For
example, members of the welfare target group are no more likely to separate
in the quarter in which they reach the maximum than are members of other
target groups who reach the maximum. Besides differences in target group
membership, this analysis also controlled for differences in the occupation
of employees, size of employers in terms of total employment, and other
factors. This analysis is described in more detail in appendix III.

The fact that an overwhelming majority of WOTC employers whom we surveyed in
California and Texas considered displacement and churning to have little or
no cost- effectiveness leads us to conclude that few of them would engage in
these practices. Our analyses of WOTC employment data compiled by the two
states provides further support for this conclusion with respect to
churning. Further, although our survey and state agency data pertain only to
participating employers in California and Texas, we believe that our
conclusions regarding the occurrence of displacement and churning are likely
to hold true in the remainder of the nation. The federal tax benefits
offered by the WOTC are the same across all states. Therefore, we have no
reason to believe that employers in California and Texas would be less
responsive to those incentives than employers in other states. Moreover,
employers that operate in multiple states account for most of the WOTC hires
in California and Texas. We spoke to the officials who were responsible for
administering the WOTC program in all 50 states and they all confirmed that
their states made no efforts to either discourage or encourage displacement
or churning. The fact that there were no differences relevant to
displacement and churning between the results of our survey and agency data
analyses for California and those for Texas also gives credence to the
generalizability of our conclusions.

The Department of Labor sent e- mail comments on a draft of this report to
us on March 1, 2001. The Department of Labor made suggestions for clarifying
information in the report. We modified the report where appropriate. The
Department of Labor also stated that, given the wealth of evidence in our
report indicating that displacement and churning are limited, our
conclusions regarding the use of these practices could be stronger. We did
not strengthen our characterization of the extent to which displacement and
churning may be occurring because we believe Conclusions

Agency Comments

Page 22 GAO- 01- 329 Work Opportunity Tax Credit

that our conclusion appropriately reflects the strength of our methodology
and resulting data.

Agencies in California and Texas responsible for the WOTC program also
reviewed our draft report regarding our description of the credit program in
their state and our analysis of state data. The agencies stated that they
had no suggestions for changes in our report.

As agreed with your office, unless you publicly announce its contents
earlier, we plan no further distribution of this report until 30 days from
the date of this letter. At that time, we are sending copies of this report
to Representative William J. Coyne, Ranking Minority Member, Subcommittee on
Oversight, House Committee on Ways and Means; the Honorable Elaine L. Chao,
Secretary of Labor; the Honorable Charles O. Rossotti, Commissioner of
Internal Revenue; Mark Heilman, Chief, Job Services Division, California
Employment Development Department; and John Carlson, WOTC Coordinator, Texas
Workforce Commission. Copies of this report will be made available to others
upon request.

If you have any questions regarding this report, please contact me or James
Wozny at (202) 512- 9110. Key contributors to this report are acknowledged
in appendix IV.

Sincerely yours, Michael Brostek Director, Tax Issues

Appendix I: Objectives, Scope, and Methodology

Page 23 GAO- 01- 329 Work Opportunity Tax Credit

The objectives of this report were to determine (1) the characteristics of
employers who have participated in the WOTC program and (2) the extent, if
any, to which employers have practiced displacement and churning. To obtain
information on the characteristics of employers, we analyzed national tax
data from the Statistics of Income Division of the Internal Revenue Service
for 1997, the most recent year that data were available, and state WOTC data
from agencies in California and Texas for 1997 through 1999. To obtain
information relating to the extent of displacement and churning, we surveyed
a stratified probability sample of employers who have participated in the
WOTC program in California and Texas. Our survey of employers is discussed
in more detail below.

For information relating to churning, we also analyzed WOTC and unemployment
insurance data for California and Texas. With these data, we determined the
total earnings and length of employment of WOTCcertified employees and
analyzed this information for evidence concerning the extent and likelihood
of churning. Our methodology for this analysis is discussed in detail in
appendix III. For additional information relating to displacement, we
analyzed national employment data in the Commerce Department?s Current
Population Survey (CPS) for 1995 through 1999. We used the CPS to estimate
employment rates for members of groups targeted by the credit and members of
groups not targeted by the credit but who may substitute in employment for
target group members.

To obtain information relating to the extent of displacement and churning,
we identified participating employers from databases of employees who had
applied for certification under the WOTC program. These databases are
maintained by the state agencies in California and Texas that are
responsible for determining the eligibility of employees as members of
targeted groups and issuing certifications of eligibility to employers.

Our desired survey population initially was managers who were hiring WOTC
program employees nationwide. However, since this information is kept by
each state office in various forms, it was not feasible to assemble a
national sampling frame. Therefore, we used data from two of the five states
with the largest numbers of WOTC employee participants in 1999. California
and Texas were the two states of the five largest with manageable electronic
databases of WOTC employees in 1999. We identified employers from these
lists by their unique employer identification numbers (EIN), which are used
by IRS. In order to have a population of employers with repeated and recent
experience with the program, we included only those who had hired at least
one certified Appendix I: Objectives, Scope, and

Methodology Survey of Employers

Appendix I: Objectives, Scope, and Methodology

Page 24 GAO- 01- 329 Work Opportunity Tax Credit

employee, hired at least once in 1999, and hired at least once in 1997 or
1998.

To identify employers from the databases of WOTC- eligible employees, we
aggregated the employees according to their employer?s EIN. For the purposes
of our sample, we defined ?employer? as a unique EIN and selected a
stratified random sample of 157 employers from the 975 total employers in
California and 148 employers from the 863 total employers in Texas. 1 The
strata were defined by how many WOTC employees the employer hired. Because
employers who had more than 100 WOTC hires accounted for 80 percent of the
total WOTC hires, those employers hiring more than 100 employees were a
separate stratum from those hiring between 2 and 100 WOTC employees. In this
way, we were able to sample more employers with larger numbers of WOTC
hires. Table 4 shows the breakdown by state and stratum of the number of
employers in the population, the number selected into the sample, and the
number who responded to the survey. In total, we sampled 305 employers and
received responses from 225, for an overall response rate of 74 percent.

Table 4: Number of Population and Sample Cases in Strata for Employers
Hiring WOTC Employees

Stratum 2 to 100 hires Over 100 hires Total

California Population 775 200 975 Selected 92 65 157 Completed survey 66 50
116 Percent completed 72% 77% 74% Texas

Population 610 253 863 Selected 86 62 148 Completed survey 71 38 109 Percent
completed 83% 61% 74%

Source: GAO analysis of WOTC and survey data in California and Texas

1 In California, there were 6,658 employers who filed requesting eligibility
for their employees under the WOTC program. Of these, 975 met the criteria
for inclusion in the population of hiring at least one employee who was
certified as eligible, hiring in 1999, and hiring in at least one other
year. In Texas, there were 4,240 employers who hired under the program, and
863 of these who met the criteria for inclusion in the population.

Appendix I: Objectives, Scope, and Methodology

Page 25 GAO- 01- 329 Work Opportunity Tax Credit

In addition to the EINs for the employers associated with WOTC- eligible
employees, the databases included limited information for a contact person.
To try to ensure that our surveys reached the correct person at the employer
site, we contacted every sampled employer by phone first. In this initial
phone call, we explained the purpose of the survey, the kinds of questions
we would be asking, and the location for which we were interested in
obtaining information, and we asked for the name of the most appropriate
respondent. Most initial contacts indicated that they were the most
appropriate respondent or that they would receive the survey and forward it
as necessary. Approximately 4 weeks after the initial mailout, we conducted
a second mailout to those who had not yet responded. Approximately 4 weeks
after that, we followed up with all remaining nonrespondents by telephone,
reminding them that they had not responded and asking them to complete a
shorter version of the questionnaire over the telephone.

Because the survey results come from a sample, all results are estimates
that are subject to sampling errors. These sampling errors measure the
extent to which samples of these sizes and structure are likely to differ
from the populations they represent. Each of the sample estimates is
surrounded by a 95- percent confidence interval, indicating that we can be
95- percent confident that the interval contains the actual population
value. Unless otherwise noted, the 95- percent confidence intervals for all
percent estimates in the letter of the report do not exceed plus or minus 10
percentage points around the estimate.

In addition to the reported sampling errors, the practical difficulties of
conducting any survey may introduce other types of error, commonly referred
to as nonsampling errors. For example, differences in how a particular
question is interpreted may introduce variability into our survey results
that is difficult to measure. We conducted pretests of the survey to
evaluate the wording of the questions. One particular source of nonsampling
error unique to this survey involves the location to which that respondent?s
answers refer. In some cases, the employer or EIN that we selected
corresponded to a very large corporation, and our contact was in a hiring
division located outside the state or local office of interest. In the
initial phone calls, the location of interest was specified; however, the
respondent may have responded with a different location in mind or may have
been unable to take into account variation in hiring practices across
several local offices. Careful pretesting of the survey did not uncover such
issues, but this possibility may lead to additional variation in our survey
results.

Appendix I: Objectives, Scope, and Methodology

Page 26 GAO- 01- 329 Work Opportunity Tax Credit

Our survey and state agency data pertain only to participating employers in
California and Texas. However, to assure ourselves that our findings are
likely to apply to WOTC employers in the rest of the nation, we examined
federal laws and regulations related to the credit, surveyed state
administrators responsible for the credit program, and analyzed data on the
participating employers. The federal tax benefits offered by the WOTC are
the same across all states. Therefore, we have no reason to believe that
employers in California and Texas respond differently to these incentives
than employers in other states. We spoke to the officials who were
responsible for administering the WOTC program in all 50 states, and they
all confirmed that their states made no effort to either encourage or
discourage displacement or churning. Moreover, employers that operate in
multiple states account for most of the WOTC hires in California and Texas.
We found no significant differences between employers in California and
Texas in the results of our survey and agency data analyses, suggesting that
our conclusions will be generalizable to employers in other states as well.

We did not verify the state and federal databases we used. However,
agreements between the Department of Labor and state WOTC offices require
the states to conduct audits of the accuracy of state WOTC records. A review
of studies of the accuracy of unemployment insurance data conducted for the
National Research Council concluded that the data appear to be accurate. The
review noted that employers are required by law to report the data, and
intentional inaccuracies are subject to penalties. This same review of
studies found that the CPS data are a valuable source of information on the
national low- income population, with broad and fairly accurate measures of
income. However, the study noted that sample sizes might be small for some
subpopulations (e. g., welfare recipients in particular states) and the
percentage of some subpopulations covered by the survey appears to have
declined modestly in recent years. The sample size for the targeted and
nontargeted groups in our analysis was sufficiently large that the
confidence intervals for the estimated employment rates were no more than 6
percentage points on either side of the estimate. We concluded that the
slight decline in coverage of welfare recipients is unlikely to affect our
analysis of trends in employment rates over the period.

As noted, we analyzed the tax data from IRS? Statistics of Income Division.
These data undergo numerous quality checks but do not include information
from amended tax returns (i. e., revisions made by taxpayers themselves
after their initial filings). Limitations of Our

Analysis

Appendix II: Survey Instruments Page 27 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments

Appendix II: Survey Instruments Page 28 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 29 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 30 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 31 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 32 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 33 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 34 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 35 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 36 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 37 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix II: Survey Instruments Page 38 GAO- 01- 329 Work Opportunity Tax
Credit

Appendix III: Statistical Analysis of State Databases

Page 39 GAO- 01- 329 Work Opportunity Tax Credit

To investigate whether reaching the maximum earnings in a given quarter
affects the likelihood that employees will separate from their WOTC
employers, we used state WOTC and unemployment insurance data on total
earnings and duration of employment. We also used data from these sources on
other employee characteristics, such as target group and occupation, and
employer characteristics, such as total employment and the industry of the
employer. The data were collected for 108,935 WOTCcertified employees and 5,
347 employers in California and Texas for the years 1997 through 1999. 1

We used the logistic regression model to quantify the effect of reaching the
maximum earnings on the probability that the employee separates from the
employer. We also used the model to estimate the effect of other employee
characteristics, such as current wages (total earnings in a given quarter)
and membership in a target group, on the probability of separation. The
results of this analysis are presented as odds ratios in table 5. An odds
ratio is a measure of relative risk of the occurrence of an event- in this
case, the separation from an employer. The reported odds ratios indicate the
effect of a particular characteristic (e. g., reaching the maximum earnings)
on the probability of separation, controlling for the effects of other
characteristics included in the analysis. The estimate of the effect,
represented by the odds ratio, is the net effect of the characteristic (i.
e., net of the effects of all other characteristics).

If the characteristic increases the probability of separation, the odds
ratio will be greater than 1, and if it decreases the probability of
separation, the odds ratio will be less than 1. This interpretation is
slightly different when the characteristics are different categories. An
example of such a

?categorical? characteristic is membership in a target group where the
categories are welfare recipients, veterans, food stamp youth, and so on. In
such cases, the analysis omits one of the categories (called the

?reference group?) and tests whether the included categories have greater 1
Using the WOTC databases provided to us by the states, we identified a total
of 154, 708 employees who were certified from 1997 through 1999 in
California and Texas. We requested that the states match the social security
numbers of these certified employees with the quarterly wage records in the
states? unemployment insurance databases. For both states, we found that
nearly all the certified employees had wage records with some employer
during the 12- quarter period. However, not all the employees had wage
records with the employer identified in the WOTC database as the employer
requesting the certification. A total of 108,935 certified employees were
identified as working for the WOTC employer in the unemployment insurance
database, indicating that 70 percent of certified workers had wage records
that could be used for our analysis. Appendix III: Statistical Analysis of
State

Databases

Appendix III: Statistical Analysis of State Databases

Page 40 GAO- 01- 329 Work Opportunity Tax Credit

or less chance of separation relative to the omitted category. An odds ratio
of greater than1 indicates greater probability of separation, while an odds
ratio of less than 1 indicates less probability of separation.

Table 5: Logistic Regression Analysis of the Likelihood of Separation of
Employees in California and Texas

Variable name Odds ratio

Current wages 0.92 a

State

California 0. 61 a Texas Reference group Maximum earnings 1. 01

Target group

Welfare recipients 1. 06 a Veterans 1.29 a Food stamp youth Reference group
Employment zone/ economic community youth 1. 07 a Ex- felons 1.54 a
Vocational rehabilitation 0. 68 a SSI recipients 0.84 a

Interactions: target group and maximum earnings

Welfare recipients and maximum earnings 0. 99 Veterans and maximum earnings
1. 36 Employment zone/ economic community youth and maximum earnings 1.17
Vocational rehabilitation and maximum earnings 1.19 Ex- felons and maximum
earnings 1. 56 a SSI recipients and maximum 1. 17 Food stamp youth and
maximum Reference group a Indicates significance at the 5- percent level.
Employees belonging to the summer youth target group were not included in
this analysis. This target group has a different wage cap ($ 3,000 of
eligible first year wages), and members of the group are seasonal workers.
The circumstances of their separations are therefore not comparable to those
of members of other target groups.

Source: GAO analysis of California and Texas WOTC and Unemployment Insurance
databases.

Table 5 shows that reaching the maximum earnings has no statistically
significant effect on the odds that employees will separate from their
employers. The variable called ?maximum earnings? indicates the quarter in
which an employee?s cumulative earnings are between $5,000 and $7,000. This
interval includes $6,000 as its midpoint and indicates that reaching the
maximum occurs in the quarter when the employee is within $1,000, more or
less, of the maximum earnings eligible for the credit. The odds ratio for
this variable is not significantly different from 1, meaning that employees
whose earnings are within $1,000 of the maximum in a

Appendix III: Statistical Analysis of State Databases

Page 41 GAO- 01- 329 Work Opportunity Tax Credit

quarter are no more likely to separate than employees whose earning are
outside this range. Table 5 shows that reaching the maximum has no effect on
the likelihood of separation across most target groups as well. For example,
members of the welfare target group are no more likely to separate in the
quarter in which they reach the maximum than are members of other target
groups who reach the maximum. 2

We also used the logistic regression model to analyze the effect of reaching
the maximum earnings separately for each state. The separate analysis
permitted more characteristics of the employees and employers to be included
because data on characteristics were not always available for both states.
We analyzed the likelihood of separation in each state using only the
characteristics in table 5, and then expanded the analysis to include the
additional variable characteristics available in each state. This analysis
shows that the conclusion about the effect of reaching the maximum on
separation does not change when additional characteristics are added to the
model. When variables indicating the occupation of the employee are added to
the analysis in California, reaching the maximum earnings continues to have
no effect on separation. When variables indicating the employer?s industry
and size in terms of total employment are added to the analysis in Texas,
reaching maximum earnings is significant, but employees reaching the maximum
are still slightly less likely to separate. Specifically, they are 9 percent
less likely to separate than are employees who do not reach maximum
earnings.

2 As table 5 shows, only members of the ex- felon target group had a
statistically significant greater chance of separation when they reach the
maximum earnings than members of other target groups when they reach the
maximum. However, ex- felons represent only about 3 percent of all certified
employees that we analyzed, and the results of the analysis for this group
may not be as reliable as the results for other target groups.

Appendix IV: GAO Contacts and Staff Acknowledgments

Page 42 GAO- 01- 329 Work Opportunity Tax Credit

Jim Wozny, (202) 512- 9110 Kevin Daly, (202) 512- 9110

In addition to those named above, Kerry Dunn, Tre Forlano, Wendy Ahmed, Sam
Scrutchins, Stuart Kaufman, Barry Seltser, and Cheryl Peterson made key
contributions to this report. Appendix IV: GAO Contacts and Staff
*** End of document. ***