Re‐Estimating Female Domestic Work: Based On The British Survey Evidence From 1986–7

Jenkins and O'Leary (SJPE, 1997) estimated unpaid work using some of the variables collected in the 1986–7 SCEL Initiative. However, the discovery of low filial effects should be attributed to the specification of their model, especially to the inclusion of proxies for contemporaneous paid work aiming to capture population heterogeneity. It seems that the two researchers could have utilised other data from people's past work histories (which were also solicited in the survey) to construct alternative paid‐work proxies and avoid the simultaneity problem altogether. The introduction of such proxies preserves the structure of the Jenkins/O'Leary model, and yields both lower full‐time and part‐time employment effects on unpaid work, and higher children‐effects. In fact the children‐effects remain pronounced even if one drops all paid work regressors or introduces a non‐selection hazard from the probit on paid work participation, instead.