The first column of Table 8 reports the estimated structural coefficient when the dependent
variable in the second-stage equation is the probability that the typical person in cell k
is covered by employer-sponsored insurance. As the first row shows, there is a sizable
negative correlation between the probability of receiving Medicaid and the probability of being covered by ESI. The coefficient is numerically large and marginally significant from
zero (−1.79, with a standard error of 0.94). Moreover, this coefficient is not significantly
different from one. The IV estimate of δ, therefore, confirms the key insight from the
descriptive statistics presented in the previous sections: the availability of Medicaid seems
to completely crowdout the presence of employer-sponsored insurance in the sample of
disadvantaged immigrants most affected by welfare reform