Public Housing Assistance, Public Transportation, and the Welfare-to-Work Transition

Abstract

Using data for welfare recipients who left the Temporary Assistance for Needy Families program during 1996 in Cuyahoga County (Cleveland), Ohio, the authors compare the determinants of labor market outcomes across three classes of housing assistance: those who receive a certificate or voucher, those who reside in a traditional public housing project, and those who reside in a Section 8 housing project. The statistical model includes spatially based measures of job opportunities for welfare recipients as well as measures of access to those opportunities. As might be expected, the authors’ analysis reveals that welfare exiters living in housing projects (either public or Section 8) are more spatially concentrated in Cleveland than those not receiving housing assistance, whereas those receiving certificates and vouchers are more spatially dispersed. Even so, welfare exiters receiving certificates and vouchers are employed closer to their homes, spend less time commuting to work, have superior public transit connections to their jobs, and generally have greater access to job openings relative to those who live in housing projects or receive no assistance. This evidence is consistent with the notion that certificate and voucher recipients have the spatial and economic flexibility in residential choice and that they exercise this choice to reduce commuting distance and time. In statistical models explaining various labor market outcomes, the authors do not find much difference among recipients of the three types of housing assistance. Without statistical controls for general neighborhood conditions, all three types of housing assistance are negatively related to the level of earnings. However, this finding disappears when statistical controls for general neighborhood conditions (poverty rate) are included in the model. This suggests that the lower earnings often attributed to housing assistance are actually a neighborhood-based effect.