To mitigate the scarring effects of a criminal record, 35 states and over 150 cities and counties have adopted ban‐the‐box policies that delay questions about a job applicant’s arrest and conviction record. These policies are meant to increase hiring and employment among WCs by making it more difficult to screen applicants based on their criminal history, helping WCs get a foot in the door when seeking employment. However, ban‐the‐box policies do not address the underlying reasons that employers may conduct criminal background checks, such as the potential for lower productivity or higher downside risk. Employers may therefore still want to ask about an applicant’s criminal record later in the hiring process or make inaccurate judgments about an applicant’s criminal record based on his or her race or other demographic characteristics.
We use a field experiment involving decisions at nearly 1,000 U.S. businesses to test several alternative approaches to increasing the demand for WCs. The alternatives we consider are meant to address potential underlying reasons that businesses may conduct criminal background checks. We offer crime and safety insurance to address downside risk concerns, and we offer screening based on past performance reviews and the time elapsed since the most recent criminal record to address downside risk and productivity concerns. We also provide objective information on the average productivity of WCs compared with non‐WCs to address downside risk and productivity concerns through a different channel, including the possibility that manager beliefs about WC productivity may be inaccurate. We benchmark the effects of each of these alternatives against the effects of a wage subsidy, a natural but potentially costly approach to increasing the demand for WCs.
The partner for our study is a large nationwide staffing platform based in the United States, which third‐party businesses use to connect with available workers. Businesses submit job requests to the platform that include a job description (typically short term), the pay for the job, and qualifying criteria. The platform sends out the job offer to workers who meet the qualifying criteria, and workers can then accept the job on a first‐come‐first‐serve basis. Businesses rarely cancel jobs after they have been accepted due to clear procedures and fee structure. Cancellations of matches occur in less than 1 percent of cases. Presenting workers with the option to accept a job is therefore equivalent to that business extending a job offer to those workers.
The platform’s design allows us to ask hiring managers to make choices over potential hiring decisions, as opposed to callback or interview decisions that have generally been considered in past work. Hiring managers that use the platform were already familiar with submitting criteria for workers who can accept their jobs. We truthfully informed these managers that, in the same way, their responses during the experiment constituted high‐stakes decisions that could determine whether WCs would be allowed to accept their jobs in the future. For example, if a hiring manager indicated willingness to work with WCs under a certain insurance policy during the experiment, then WCs would be eligible to accept jobs posted by that manager after such an insurance policy is made available. The high‐stakes nature of these choices was not just theoretical; the choices that hiring managers made during the experiment actually affected whether WCs could accept their posted jobs.
In the experiment, the platform asked hiring managers at nearly 1,000 businesses whether they would allow WCs to accept their jobs given the availability and level of wage subsidies, crime and safety insurance, past performance reviews, and a more limited screening of criminal records. Starting with the baseline level of demand for WCs in our sample, we find that a sizable share of businesses, 39 percent, are willing to work with WCs without additional incentives or conditions. The level of demand, still without additional incentives or conditions, increases to about 45 percent for jobs that do not involve customer interactions and 51 percent for jobs that do not involve high‐value inventory, consistent with businesses perceiving risks related to customer safety or inventory theft. We also find that the share of businesses willing to work with WCs increases to 68 percent if businesses are having a hard time filling a job on the platform, consistent with businesses being more likely to consider nontraditional workers in tight labor markets.
We find that the share of businesses willing to work with WCs further increases by at least 10 percentage points when businesses are offered a modest level of crime and safety insurance, or when WCs must have received a single positive performance review to qualify, or when a more limited background check rules out a conviction or arrest in the past year. Wage subsidies can achieve similar increases but only at relatively high subsidy levels that may be cost‐prohibitive. For example, providing crime and safety insurance covering damages up to $5,000 increases the level of demand for WCs by about 12 percentage points, approximately equivalent to the effects of an 80 percent wage subsidy. Requiring WCs to have completed one prior job posted on the platform similarly increases the level of demand by about 11 percentage points, again roughly equivalent to the effect of an 80 percent wage subsidy. Limiting the pool of WCs to those who have not been convicted or arrested in the past year increases the level of demand by about 21 percentage points, greater than the effect of a 100 percent wage subsidy. We note, however, that limiting the pool of WCs reduces the level of demand to zero for the screened‐out individuals.
The final option we consider is providing hiring managers with objective information on the performance of WCs. We exploit the fact that some WCs inadvertently accessed the platform before their background screening results were known, allowing us to objectively compare the customer performance ratings of WCs and non‐WCs in their first jobs. We find that hiring managers underestimate the performance of WCs whether they have high or low performance ratings. Providing objective information on the true share of high performance ratings received by WCs leads to more‐accurate beliefs and increases WC hiring by about 6 percentage points on average, roughly equivalent to the effect of a 40 percent wage subsidy. Providing objective information on the share of low performance ratings, typically resulting from no‐shows, also leads to more‐accurate beliefs but does not have a significant impact on the willingness to work with WCs. The muted effect of low performance ratings compared with high performance ratings suggests that businesses may be less concerned with no‐shows when deciding whether to work with WCs and more concerned that WCs might not satisfactorily meet their performance standards while on the job.
Beyond demonstrating that a range of policies can increase the demand for WCs, our research provides new evidence on why businesses may conduct criminal background checks and hence what types of policies are likely to increase the demand for WCs. Several of our results, including the large demand response to crime and safety insurance and the even larger response to insurance among businesses whose jobs involve customer interaction or high‐value inventory, suggest that businesses are sensitive to the downside risk of hiring WCs. Other results, including the large response to objective information on the productivity of WCs, are consistent with the view that some businesses view WCs as less productive on average than otherwise similar non‐WCs (a view that is incorrect in our setting). The positive effects of the wage subsidies, performance screening, and screening of the most recent records are consistent with both explanations.
Our research also provides new evidence on why there may be such low take‐up of existing federal programs meant to increase the demand for WCs, such as the Work Opportunity Tax Credit or Federal Bonding Program. For example, the Work Opportunity Tax Credit provides a wage subsidy of up to 40 percent for hiring disadvantaged workers, including individuals convicted of a felony and released from prison in the past year. Our findings suggest that the reason the take‐up of this program is relatively low may be that employers are less willing to work with recently convicted or released WCs, particularly in the first year. The Federal Bonding Program instead provides insurance to employers hiring hard‐to‐place job candidates, which our findings suggest should increase the demand for WCs. The relatively low take‐up of this program is therefore more likely the result of employers being unaware of the program or having difficulty navigating the program requirements.
NOTE
This research brief is taken directly from Zoe B. Cullen, Will S. Dobbie, and Mitchell Hoffman, “Increasing the Demand for Workers with a Criminal Record,” NBER Working Paper no. 29947, April 2022.
