X hits on this document





15 / 42

percentage of job that are unskilled,11 the total hires over the past year, establishment location,

whether the establishment checks criminal background, is union, is non-profit, and whether the

establishment is minority-owned. We also analyze the relationship with the racial composition

of applicants.

There are several clear patterns in Table 1.12

First, the distribution of industries among

those most willing to accept ex-offenders is skewed towards manufacturing, construction, and

transportation, or those industries that likely have fewer jobs that require customer contact.

Moreover, the distribution of establishments that actually hire ex-offenders is similarly skewed,

indicating a strong correlation between willingness to hire and actual hiring of ex-offenders in

these industries. We also find that establishments willing to hire ex-offenders are

disproportionately those with a large fraction of unskilled jobs (>0.200) and those that hired over

20 workers over the past year (reflecting size and/or turnover). On the other hand, we find the

service industries more represented among those unwilling to accept and that have not hired ex-


We also find some cases in which the characteristics of those willing to accept ex-

offenders do not correlate with those of establishments that actually hire them. For instance,

establishments that always check criminal backgrounds are disproportionately represented

among those that say they are not willing to accept applicants with criminal records. But these

establishments are equally represented among those firms that have hired ex-offenders v. those

that have not. This evidence suggests that, with respect to hiring ex-offenders, the prospective

11 Unskilled jobs refer to those jobs that do not require any particular skills, education, previous training or experience when filled.

12 We also estimated probit regressions in which willing to accept applicants with criminal histories and have hired an ex-offender over the past year are binary dependent variables, with the list of variables identified in Table 1 included as independent variables. The results of these regressions were largely consistent with and qualitatively similar to the descriptive results we present in Table 1. A multinomial logit of employer willingness to hire ex-


Document info
Document views65
Page views65
Page last viewedWed Oct 26 13:51:59 UTC 2016