that are closer to ex-offender populations (or that receive a greater number of applications from
ex-offenders) are more likely to run background checks (Holzer et. al., forthcoming).
Table 3 also shows the averages of these establishment characteristics stratified by
whether employers were legally required to check. Here, we find somewhat similar patterns to
those we found for the frequency of checking. Establishments that are larger, not-for-profit, in
service industries, have no unskilled jobs, as well as those that have collective bargaining
agreements and higher percentages of black applicants are overrepresented among those that are
legally required to check. On the other hand, manufacturing, construction, smaller, and
minority-owned firms, as well as firms with a large fraction of unskilled jobs are
underrepresented amongst firms that are legally required to check.
Taken together, these results suggest that the greater propensity of firms that always
check (e.g., large and service firms) is in large part prompted by the legal requirement to do so.
Given the very widespread legal barriers to employment in many occupations that occur in most
states, it is perhaps not surprising that legal requirements drive a great deal of employer behavior
in this regard. But these findings also suggest that the laws that prevent employers from hiring
offenders might need to be reviewed, in light of the strong negative effects they appear to have
on the ability of ex-offenders to gain employment.
The substantial increase in the proportion of establishments that always check the
criminal histories of applicants over the 1990s, as shown in Figure 5, also suggests that the
availability of low-cost checking services in the private market may be in part driving this
increase. Although we do not have data in both 1992 to 94 on the method by which employers
check backgrounds to fully explore these factors, our 2001 survey does ask this question. Figure
7 shows the method by which employers check criminal histories stratified by their responses to