Abstract
Applying propensity score matching to the National Child Development Study, we found that conviction reduces the employment probability of middle-aged British females by about three times more than that of comparable males. Moreover, while males recover a part of the disadvantage by increasing self-employment, conviction results in a strong labour market marginalization for females, as unemployment and, overall, inactivity increase. Robustness checks tend to confirm these findings. This suggests both a stronger discouragement effect for females and/or a different attitude towards self-employment and/or excluding factors (e.g. access to borrowing).
Notes
1 See Holzer et al. (Citation2006) for an application of the statistical discrimination thesis, Rasmusen (Citation1996) and Sciulli (Citation2010) for an economic explanation of stigma associated to conviction, and Entorf (Citation2009) for negative signals of worker's productivity.
2 Conviction, because of stigma or negative signals, may be also associated with discouragement favouring inactivity, or preferences for self-employment to avoid stigmatization or negative effects from screening.
3 We exclude cohort members living in jail in 2000.
4 Dehejia and Wahba (Citation2002) argued that estimation bias in matching methods may be reduced by using simple PSM and controlling for past outcomes.
5 Because the analysis is justified only if performed over the common support region (Heckman et al., Citation1998), our estimations rely on 4551 males and 4976 females.
6 The gap between selection into crime rather than into conviction depends on the probability of offenders being punished.
7 The estimation results of the propensity score equation and the propensity balancing tests are available upon request.
8 The employment probabilities are expressed in terms of percentage points.