treated population is therefore the wives of married male veterans ages 55 to 64,
and the control group is the wives of married male non-veterans in the same age
group. We define 1992-1995 as the pre-policy period and 1998-2002 as the post-
policy period because changes in the VA health care were rolled out during 1996
and 1997.9
The CPS allows us to study labor market outcomes on the extensive
margin such as labor force or employment exit, and on the intensive margin, such
as hours worked, or movement into part-time work. It further allows us to
examine earnings, although the universe for which we can study current earnings
outcomes is limited during this time period, and type of labor force participation,
such as self-employment. We are also able to examine the effects for different
demographic groups, such as by education. 10
C. Main Specification
64, 50-64, and 55-64. In general these results are qualitatively the same as our main results.
Quantitatively, Table 3 results on work outcomes are slightly larger in magnitude and significance
for these subsets (Table 7, Panel I provides some results) and Table 4 results are quantitatively
similar.
9 We end our study period in 2002 because VA revised the rules for obtaining health care January
2003. We have also estimated our regressions restricting our post-period to 1998-2001 because of
a concern that particular Vietnam Era veterans are affected by a 2002 change that categorized
diabetes as a war-related injury for veterans who may have been exposed to Agent Orange
(Duggan et al. 2010, Autor and Duggan 2007). Results are qualitatively almost identical and
significance increases in some regressions when we remove 2002 from our sample. Table 7, Panel
III provides these results.
10 Results by VA means-tested status are very similar to those by education level and are therefore
not included.
12