Spousal Labor Market Effects from Government Health Insurance: Evidence from a Veterans Affairs Expansion

estimating equation (1) on outcomes for married men only using the men’s
characteristics as controls.¹³ Veterans are more likely than the control group to be
not working after receiving VA health insurance, with a significant coefficient of
0.008 points once a full set of controls is added, a 2.3% increase relative to the
pre-period veteran average of 0.35. Veterans also increase part-time work with
significant coefficients of 0.016 points in both specifications, an increase of
15.8% relative to the pre-period. Table 2, Panel II, provides a robustness check
for these results, demonstrating that there is not a pre-trend by cutting the universe
to only include pre-period data and creating a “fake post” variable that is 1 for
1994-1995 and 0 for 1992-1993. As would be expected if there was not a pre-
trend, results are not significant, and indeed, are opposite-signed for the not
working outcome.

Having demonstrated that the VA expansion decreased married men’s
labor supply, we turn to the spillover effects of this coverage on their wives who
are not eligible to use VA health care. First, we test the effects of public
insurance on wives’ labor supply on the extensive margin. As shown in Table 3,

¹³ These results differ slightly from those in Boyle and Lahey (2010) because the universe for that
exercise included single men (who are shown to be more likely to leave the labor market than
married men after receiving health insurance) and, in order to be consistent with the previous job-
lock literature, limited to men who were working in the previous year. We do not condition on
previous employment to examine the spillover effects on spouses. Additionally, in Boyle and
Lahey (2010) we included industry and occupation controls in the set of full controls, but in our
regressions with wife outcomes, the small size of some of these cells causes observations to drop
out in the probit specifications, potentially resulting in selection biases. Results are nearly
identical for the men and qualitatively similar for the women results with these controls included.

17

<<   15   16   17   18   19   20   21   >>

More intriguing information