Temporary help service firms' use of employer tax credits: implications for disadvantaged workers' labor market outcomes.

Workers in Means Gender N = 681

Female 0.695 0.121 *** Race N =681

White 0.305 -0.009

Black 0.432 0.024

Hispanic 0.057 -0.007

Other race (nonwhite) 0.206 0.007 Education N = 562

Less than high school 0.413 0.097 ***

High school or equivalent 0.489 -0.075 ***

Some college 0.093 -0.021

College degree or higher 0.005 -0.002 Children N = 575

No. of children < 18 in

household 1.97 0.267 ***

No. of children < 6 in

household 1.09 0.156 *** Age N = 669

Age in quarter of job start 27.0 -1.76 *** Residence N = 573

Lives in Milwaukee 0.621 0.010 Sample includes sample means for all WOTC certifications in Wisconsin for workers with WOTC jobs starting in July 1999-December 2001 who had some welfare or food stamp receipt in 1998-2001. These data are a subset of those used in Table 2A. That data set had 17,018 observations. We limit the sample to those starting a job in the period July 1999-December 2001 and then keep only those workers who received welfare or food stamps sometime in 1998-2001 since we have demographic characteristics for this subset. These restrictions result in a sample size of 6374 person-jobs, of which 681 are THS jobs. The total number of workers in the sample is 5928. Mean differences are labeled: * if significant at 10%, ** if significant at 5%, and *** if significant at 1%. Table 3. Employment Outcomes in WOTC and THS Employment Job Characteristics THS Firm WOTC certified Quarters at job:

2.03 (1.51)

Total earnings at job:

$2868 (4863)

Average earnings per quarter:

$1079 (963) WOTC eligible but Quarters at job: not certified 1.80 (l.40)

Total earnings at job:

$1656 (4016)

Average earnings per quarter:

$643 (822) Job Characteristics End-User Firm WOTC certified Quarters at job:

3.07 (3.33)

Total earnings at job:

$5972 (13,636)

Average earnings per quarter:

$1174 (1168) WOTC eligible but not certified Sample contains jobs obtained by THS and/or WOTC workers, as used in Tables 1 and 2B. This table contains sample means and standard deviations, by subgroup, using information from each record in Tables 1 and 213, based on the UI data for the relevant job. (The upper left cell uses the data on THS WOTC workers from Table 213, which is a slightly larger sample than that in Table 1.) Table 4. Propensity Score Estimation for Selection into WOTC Certification Variable Odds Ratio Z-Statistic Age 1.120 *** 1.73 Age squared 0.999 -1.35 Female 1.108 0.57 High school diploma 1.448 *** 2.94 Some college 1.510 * 1.77 College degree 1.372 0.30 Black 0.423 *** -5.11 Hispanic 0.520 ** -2.65 Other race (nonwhite) 0.723 * -1.77 Number of children under 6 years of age 0.765 *** -4.46 Number of children ages 6-17 0.934 -1.41 Milwaukee resident 0.544 *** -4.13 Firm headquarters in Wisconsin 0.813 * -1.73 Size of disadvantaged workforce

at firm (in 1000s) 0.997 -0.89 Last year's average quarterly

earnings/worker at firm 1.0007 *** 6.97 This sample is based on the sample of THS workers in Table 1. We drop workers with more than one WOTC certification in the period from 1996 to early 2003 and workers with missing (and not easily imputed) data on the independent variables or with UI matches that have job start dates inconsistent with the WOTC records. The sample size is 10,962 (of which 320 are WOTC workers). The last variable was constructed by assembling all records for the relevant firm that were in our data in the last year; all these workers' earnings were added together and divided by the number of workers to get a sense of the usual earnings of workers at this firm. Quarter indicators and an intercept are also included in the estimation. Omitted indicators are Education = less than high school and Race = white. Estimated via logit; pseudo-[R.sup.2] = 0.101. * significant at 10%; ** significant at 5%; *** significant at 1%. Table 5. Effects of WOTC Certification on Wages and Tenure of THS Workers

Effect of WOTC Outcome of Interest Certification Total wages (all jobs) in first

year following THS job start $310 (425) Total wages (all jobs) in second

year following THS job start $333 (461) Total quarters employed in first

year following THS job start 0.071 (0.099) Total quarters employed in second

year following THS job start 0.141 (0.114) Sample size: [N.sub.treatment] = 314, [N.sub.comparison] = 10,642. Standard errors are estimated via 100 bootstrap replications. We match on log-odds ratios of the propensity score to account for choice-based sampling (see Smith and Todd 2005). We use a bandwidth of 2, but point estimates with bandwidths of 0.5, 1, 3, and 4 are all quite similar. We also use 2% trimming of the treated individuals to address common support concerns, as described in Todd (1999). The measures with "year following" do not include the quarter in which the job started. Sample sizes are smaller than in Table 1 because of additional restrictions regarding missing data and multiple records. Panel estimates use a slightly smaller data set because we drop those who were under the age of 16 two years before the relevant job start. Table 6. Propensity Score Estimation for Selection into THS Employment Variable Odds Ratio Z-Statistic Age 1.082 * 1.75 Age squared 0.999 -1.56 Female 0.738 ** -2.12 High school diploma 1.307 ** 2.38 Some college 1.404 * 1.73 College degree 1.643 0.69 Black 0.932 -0.47 Hispanic 1.224 0.83 Other race (nonwhite) 1.186 1.04 Number of children under 6 years of age 0.937 -1.21 Number of children ages 6-17 0.986 -0.32 Milwaukee resident 1.089 0.64 Target group =poor veteran 1.797 * 1.82 Target group = ex-felon 3.026 *** 4.14 Target group = high-risk youth 1.635 ** 2.03 Target group = vocational rehabilitation 0.646 -1.23 Target group = food stamp youth

(ages 18-24) 1.295 1.35 Target group = supplemental security

income 0.816 -0.72 Target group = WtW 1.333 * 1.61 Size of disadvantaged workforce at firm

(in 1000s) 0.991 *** -3.97 Occupation: clerical/sales 2.053 ** 2.19 Occupation: service 6.413 *** 5.81 Occupation: other 7.412 *** 6.21 Sample: All Wisconsin workers who were WOTC/WtW certified for a job starting in July 1999-December 2001 and who had welfare or food stamp receipt sometime in 1998-2001. We drop workers with more than one WOTC certification in the period from 1996 to early 2003 and workers with missing (and not easily imputed) data on the independent variables or with UI matches that have job start dates inconsistent with the WOTC records. Sample size is 3978 (of which 453 are THS workers). Quarter indicators and an intercept are also included in the estimation. Omitted indicators are Education = less than high school, Race -white, Target group = welfare, and Occupation = professional/technical/managerial. Estimated via logit; pseudo-[R.sup.2] = 0.102. * significant at 10%; ** significant at 5%: *** significant at 1%. Table 7. Effects of THS Employment on Wages and Tenure of WOTC Workers Outcome of Interest Effect of THS Job Total wages (all jobs) in first year

following WOTC job start -$247 (417) Total wages (all jobs) in second year