labor supply changes within couples may vary upon the husbands’ public
insurance receipt. Some husbands with employer-provided insurance may
continue full-time work in spite of the availability of VA insurance, with the
expectation that they will reduce their labor supply once their wives are able to
find an alternate source of insurance (i.e the wife might change her labor supply
before or at the same time as the husband). Nevertheless, we may expect to see
stronger results for couples in which the husband is either not working full-time
or is not working at all, as these are couples for whom the offer is more likely to
have had a direct effect (i.e. for whom the VA coverage potentially caused a
decrease in the husband’s labor supply). Panel II does demonstrate results with
larger magnitudes for these two groups. Wives of husbands who are not working
full time are more likely to work and work about 1.37 more hours per week after
public health insurance is offered to their husbands. Similarly, wives of husbands
who are not working at all are even more likely to work (decrease of not working
of 4.3 ppt) and increase their hours worked by 1.76 hours per week.
Panel III provides additional robustness checks. Columns (1) and (2)
demonstrate that fully interacting the independent variables in the model with
veteran provides results similar to the base regression. Columns (3) and (4)
demonstrate the same for removing the year 2002 (when diabetic Vietnam
veterans potentially exposed to Agent Orange were re-categorized as having
work-related injuries).
23