Gender wage gap when women are highly inactive: Evidence from repeated imputations with Macedonian data

The objective of this research is to understand if large gender employment and participation gaps in Macedonia can shed some light on the gender wage gap. A large contingent of inactive women in Macedonia including long-term unemployed due to the transition process, female remittance receivers from the male migrant, unpaid family workers in agriculture and so on, is outside employment, but is not necessarily having the worst labour-market characteristics. In addition, both gender wage gap and participation gap enlarge as education decreases, revealing the importance of non-random selection of women into employment. Though, the standard Heckman-type correction of the selectivity bias suggests that non-random selection exists, but the resulting wage gap remains at the same level even when selection has been considered. Instead, we perform repeated wage imputations for those not in work, by simply making assumptions on the position of the imputed wage observation with respect to the median. Then, we assess the impact of selection into employment by comparing estimated wage gaps on the base sample versus on an imputed sample. The main result is that selection explains most of the gender wage gap in the primary-education group (75%), followed by the secondary-education group (55%). In the tertiary group, the small initial gap vanishes once selection considered. This suggests that indeed non-working women are not those with the worst labour-market characteristics. Results suggest that gender wage discrimination in Macedonia is actually between 5.4% and 9.8% and does not exist for the highly-educated women. The inability of the Heckman-type correction to document a role for selection in explaining the gender wage gap may be due to the criticisms to the exclusion restrictions and the large amount of missing wages.