3. State responses to welfare reform
A key provision of PRWORA allows states to enact state-funded assistance programs
specifically targeted to their immigrant populations if they wished to attenuate the presumed
adverse impact of welfare reform on the foreign-born. Zimmermann and Tumlin
(1999) and Tumlin et al. (1999) summarize the various programs that states extended to
immigrants in the wake of welfare reform. Although there are many ways of describing
the states’ choices, one simple approach indicates if the states offered TANF, Medicaid,
food assistance, and SSI to pre-enactment and post-enactment immigrants during the initial
5-year bar. It turns out that almost every jurisdiction (50 out of 51) offered TANF and
Medicaid to pre-enactment immigrants. A few states went beyond this “minimal” level of
generosity and offered other programs to their pre-enactment immigrant populations and to
post-enactment immigrants during the 5-year bar. The first two columns of Table 2 summarize
these “beyond-the-minimum” state actions. It is worth noting that many of the states
with large concentrations of immigrants exceeded the minimal level of generosity.
To show how the “chilling effect” of welfare reform on Medicaid participation and
health insurance coverage depended on the decisions made by individual states, I pool
the 1994–1995 calendar years of the March CPS to provide a snapshot of the immigrant
and native population prior to welfare reform, and the 1998–2000 calendar years to provide
the respective snapshot after welfare reform.11 To easily summarize the evidence, I group states into two categories that signal their degree of generosity towards immigrants.
I initially use a definition of the state’s generosity based on the data summarized in the
first two columns of Table 2. A state is classified as “more generous” if it offered at least
one of the programs listed in these two columns; otherwise, the state is classified as “less
generous”. By this definition, 29 states are classified as more generous. Finally, I calculate
health insurance coverage rates in three mutually exclusive groups: natives, citizens, and
non-citizens.12
The first four columns of Table 3 summarize the evidence. The table clearly shows that
the decisions made by some states to offer a state-funded safety net to their immigrant
populations did not greatly alter the trend of Medicaid participation for native households.
For example, the probability that natives are enrolled in Medicaid declined by about 2–3
percentage points during the period, regardless of whether the state was generous to its
immigrant population. In contrast, the state decisions had a greater impact on Medicaid
enrollment rates among immigrants, both for naturalized citizens and non-citizens. For
example, the fraction of citizens enrolled in Medicaid declined by 1.5 percentage points
in the less-generous states, but rose in the more generous states. Similarly, the fraction
of non-citizens enrolled in Medicaid declined by 7.0 percentage points (from 18.1 to
11.1%) in the less generous states, but by 4.9 percentage points in the more generous
states (from 21.0 to 16.1%). It is clear that non-citizen households in the less generous
states experienced a much larger relative decline in Medicaid participation than native
households.
The differential trends for non-citizen households between the less generous and more
generous states are even sharper when the sample is restricted to the non-refugee population.
Although the CPS data do not report the type of visa used by a particular immigrant to enter
the country, one can approximate the refugee sample by using information on the national
origin of the foreign-born households. In particular, most refugees tend to originate in a small set of countries.13 I classified all persons residing in households where the household
head originated in the main refugee-sending countries as refugees, while all other persons
were classified as non-refugees. The non-citizen, non-refugees residing in the less generous
states experienced a 7.0 percentage point decline in their Medicaid participation rate, as
compared to the 3.1 percentage point decline for the non-citizen, non-refugees residing in
the more generous states.
The second panel of Table 3 replicates the analysis for health insurance coverage. The
probability that natives are covered by health insurance rose slightly in both the more
and less generous states. Moreover, the probability that immigrants are covered by health
insurance is also relatively stable over time: the probability fell by 0.3 percentage points
in the more generous states and by 1.1 percentage points in the less generous states. Most
strikingly, the health insurance coverage rate for non-citizens dropped by 1.7 percentage
points in the more generous states, but rose by 2.1 percentage points in the less generous
states. In short, the descriptive data reported in Table 3 do not reveal that the Medicaid
cutbacks experienced by non-citizens in the less generous states adversely affected their
overall rate of health insurance coverage.
The differential trends in non-citizen Medicaid participation and health insurance coverage
can be explained by a substantial increase in the probability that these immigrants were
covered by ESI. The bottom panel of Table 3 reports the trends in the rate of employerprovided
insurance for the various groups. The generosity of the state’s welfare program
towards immigrants does not affect the likelihood that natives are covered by ESI. The rate
of employer-sponsored insurance among natives rose by 2.6 percentage points in the more
generous states, and by 3.0 percentage points in the less generous states. In contrast, the rate
of ESI coverage for non-citizens rose by 2.7 percentage points in the more generous states,
and by an astounding 11.4 percentage points in the less generous states. The descriptive
evidence reported in Table 3, therefore, suggests a causal relationship between the Medicaid
cutbacks and the use of ESI coverage in the targeted population.The last four columns of the table report the trends in health insurance coverage in a
population that is of particular concern in the current context, namely children under the
age of 15 years.14 The differences in the trends among the various types of health insurance
coverage tend to be much sharper among children than in the general population. For example,
the fraction of non-citizen children covered by Medicaid fell by 4.5 percentage points
in the more generous states (from 35.4 to 30.9%), but it dropped by almost 9 percentage
points in the less generous states (from 31.1 to 22.2%).
Interestingly, the substantial decline in government-sponsored health insurance among
non-citizen children living in the less generous states did not materially affect the fraction
of those children who had some type of health insurance coverage. In particular, the rate of
health insurance coverage for non-citizen children in the more generous states fell by 1.3
percentage points (from 70.1 to 68.8%), but rose by 2.4 percentage points (from 63.3 to
65.7%) in the less generous states. The underlying reason for this differential trend was again
a sizable increase in the number of non-citizen children covered by employer-sponsoredinsurance. The rate of ESI coverage for non-citizen children living in the more generous
states rose from 35.5 to 37.1% during the period, as contrasted with a rise from 32.6 to 44.9%
for the children living in the less generous states. In short, the labor supply responses by the
parents of non-citizen children helped to completely offset the impact of the government
cutbacks in Medicaid assistance.
It is instructive to use a simple regression model to formalize and extend these descriptive
results. By controlling for various socioeconomic characteristics, the regression approach
helps us determine if the differential trends in health insurance coverage observed between
the more and less generous states arise because different types of immigrants tend to live
in different states, or if the variation can be attributed to state-specific trends in economic
activity or social conditions. To illustrate the basic methodology, pool the CPS data available
for the calendar years 1994, 1995, 1998–2000 and consider the triple-difference linear
probability specification:
3. State responses to welfare reformA key provision of PRWORA allows states to enact state-funded assistance programsspecifically targeted to their immigrant populations if they wished to attenuate the presumedadverse impact of welfare reform on the foreign-born. Zimmermann and Tumlin(1999) and Tumlin et al. (1999) summarize the various programs that states extended toimmigrants in the wake of welfare reform. Although there are many ways of describingthe states’ choices, one simple approach indicates if the states offered TANF, Medicaid,food assistance, and SSI to pre-enactment and post-enactment immigrants during the initial5-year bar. It turns out that almost every jurisdiction (50 out of 51) offered TANF andMedicaid to pre-enactment immigrants. A few states went beyond this “minimal” level ofgenerosity and offered other programs to their pre-enactment immigrant populations and topost-enactment immigrants during the 5-year bar. The first two columns of Table 2 summarizethese “beyond-the-minimum” state actions. It is worth noting that many of the stateswith large concentrations of immigrants exceeded the minimal level of generosity.To show how the “chilling effect” of welfare reform on Medicaid participation andhealth insurance coverage depended on the decisions made by individual states, I poolthe 1994–1995 calendar years of the March CPS to provide a snapshot of the immigrantand native population prior to welfare reform, and the 1998–2000 calendar years to providethe respective snapshot after welfare reform.11 To easily summarize the evidence, I group states into two categories that signal their degree of generosity towards immigrants.I initially use a definition of the state’s generosity based on the data summarized in thefirst two columns of Table 2. A state is classified as “more generous” if it offered at leastone of the programs listed in these two columns; otherwise, the state is classified as “lessgenerous”. By this definition, 29 states are classified as more generous. Finally, I calculatehealth insurance coverage rates in three mutually exclusive groups: natives, citizens, andnon-citizens.12The first four columns of Table 3 summarize the evidence. The table clearly shows thatthe decisions made by some states to offer a state-funded safety net to their immigrantpopulations did not greatly alter the trend of Medicaid participation for native households.For example, the probability that natives are enrolled in Medicaid declined by about 2–3percentage points during the period, regardless of whether the state was generous to itsimmigrant population. In contrast, the state decisions had a greater impact on Medicaidenrollment rates among immigrants, both for naturalized citizens and non-citizens. Forexample, the fraction of citizens enrolled in Medicaid declined by 1.5 percentage pointsin the less-generous states, but rose in the more generous states. Similarly, the fractionof non-citizens enrolled in Medicaid declined by 7.0 percentage points (from 18.1 to11.1%) in the less generous states, but by 4.9 percentage points in the more generousstates (from 21.0 to 16.1%). It is clear that non-citizen households in the less generousstates experienced a much larger relative decline in Medicaid participation than nativehouseholds.The differential trends for non-citizen households between the less generous and moregenerous states are even sharper when the sample is restricted to the non-refugee population.Although the CPS data do not report the type of visa used by a particular immigrant to enterthe country, one can approximate the refugee sample by using information on the nationalorigin of the foreign-born households. In particular, most refugees tend to originate in a small set of countries.13 I classified all persons residing in households where the householdhead originated in the main refugee-sending countries as refugees, while all other personswere classified as non-refugees. The non-citizen, non-refugees residing in the less generousstates experienced a 7.0 percentage point decline in their Medicaid participation rate, ascompared to the 3.1 percentage point decline for the non-citizen, non-refugees residing inthe more generous states.The second panel of Table 3 replicates the analysis for health insurance coverage. Theprobability that natives are covered by health insurance rose slightly in both the more
and less generous states. Moreover, the probability that immigrants are covered by health
insurance is also relatively stable over time: the probability fell by 0.3 percentage points
in the more generous states and by 1.1 percentage points in the less generous states. Most
strikingly, the health insurance coverage rate for non-citizens dropped by 1.7 percentage
points in the more generous states, but rose by 2.1 percentage points in the less generous
states. In short, the descriptive data reported in Table 3 do not reveal that the Medicaid
cutbacks experienced by non-citizens in the less generous states adversely affected their
overall rate of health insurance coverage.
The differential trends in non-citizen Medicaid participation and health insurance coverage
can be explained by a substantial increase in the probability that these immigrants were
covered by ESI. The bottom panel of Table 3 reports the trends in the rate of employerprovided
insurance for the various groups. The generosity of the state’s welfare program
towards immigrants does not affect the likelihood that natives are covered by ESI. The rate
of employer-sponsored insurance among natives rose by 2.6 percentage points in the more
generous states, and by 3.0 percentage points in the less generous states. In contrast, the rate
of ESI coverage for non-citizens rose by 2.7 percentage points in the more generous states,
and by an astounding 11.4 percentage points in the less generous states. The descriptive
evidence reported in Table 3, therefore, suggests a causal relationship between the Medicaid
cutbacks and the use of ESI coverage in the targeted population.The last four columns of the table report the trends in health insurance coverage in a
population that is of particular concern in the current context, namely children under the
age of 15 years.14 The differences in the trends among the various types of health insurance
coverage tend to be much sharper among children than in the general population. For example,
the fraction of non-citizen children covered by Medicaid fell by 4.5 percentage points
in the more generous states (from 35.4 to 30.9%), but it dropped by almost 9 percentage
points in the less generous states (from 31.1 to 22.2%).
Interestingly, the substantial decline in government-sponsored health insurance among
non-citizen children living in the less generous states did not materially affect the fraction
of those children who had some type of health insurance coverage. In particular, the rate of
health insurance coverage for non-citizen children in the more generous states fell by 1.3
percentage points (from 70.1 to 68.8%), but rose by 2.4 percentage points (from 63.3 to
65.7%) in the less generous states. The underlying reason for this differential trend was again
a sizable increase in the number of non-citizen children covered by employer-sponsoredinsurance. The rate of ESI coverage for non-citizen children living in the more generous
states rose from 35.5 to 37.1% during the period, as contrasted with a rise from 32.6 to 44.9%
for the children living in the less generous states. In short, the labor supply responses by the
parents of non-citizen children helped to completely offset the impact of the government
cutbacks in Medicaid assistance.
It is instructive to use a simple regression model to formalize and extend these descriptive
results. By controlling for various socioeconomic characteristics, the regression approach
helps us determine if the differential trends in health insurance coverage observed between
the more and less generous states arise because different types of immigrants tend to live
in different states, or if the variation can be attributed to state-specific trends in economic
activity or social conditions. To illustrate the basic methodology, pool the CPS data available
for the calendar years 1994, 1995, 1998–2000 and consider the triple-difference linear
probability specification:
การแปล กรุณารอสักครู่..