Table 7 reports the relevant regression coefficients from various specifications of the
model in Eq. (2). The estimated coefficients consistently show that the labor supply of
non-citizen men declined substantially in those states that were most generous with their
immigrant populations in the aftermath of PRWORA, even after controlling for differences
in a vast array of socioeconomic characteristics and state-specific factors. For example, the triple-difference coefficient measuring the impact of the welfare cutbacks on non-citizen
male labor force participation is 0.060 (with a standard error of 0.016); the coefficient
measuring the impact on log annual hours worked is 0.041 (0.058); and the coefficient
measuring the impact on the probability that the worker works a full-time week is 0.040
(0.022). In contrast, the triple-difference coefficient measuring the relative impact of the state
programs on the labor supply of citizen men is numerically closer to zero and statistically
insignificant. Finally, although the labor supply effects are (statistically) weaker for women,
the estimated coefficients suggest that hours of work for non-citizen women increased more
if they lived in states that were not generous to their immigrant populations in the aftermath
of PRWORA.