Immigration & welfare

George J. Borjas

Copyright National Review; New York; Jun 16, 1997


Solving the welfare problem will solve the welfare problem-not the immigration problem.

I began to investigate the link between immigration and welfare in 1987. Since that time, I have given countless presentations at universities and think tanks. Invariably, someone will come up after the seminar and say something like: "I believe the data and I agree with you that there is a problem. But the problem is not immigration; the problem is welfare." This response has recently been enshrined by Cato Institute analysts as their rallying cry against the accumulating evidence on immigrant use of welfare: "Immigration Yes, Welfare No!"

When I first heard this response a decade ago, it struck me as odd. Yes, there are things that are wrong with the welfare system. But if the social experimentation of the past few decades has taught us anything, it is that we have no clue about how to tackle poverty. In view of our resounding defeat in the War on Poverty, why would we consciously pursue policies-such as our current immigration policy-that exacerbate poverty?

Today's immigrants are more likely to receive welfare than natives and than earlier immigrants. In 1970, the Census indicated that immigrant households were less likely to receive cash benefits (such as AFDC, SSI, or General Assistance) than native households. By 1990, the Census showed that immigrants were more likely to receive cash benefits than native households. In fact, if one adds non-cash programs (such as Medicaid, Food Stamps, and housing assistance), it turns out that 21 per cent of immigrant households receive some type of aid, as compared to 14 per cent of native households and 10 per cent of white non-Hispanic native households. In short, the "welfare gap" between immigrants and natives has reversed direction and grown substantially in a very short time.

Why did immigrant use of welfare rise so rapidly? It's elementary: today's immigrants are relatively less skilled than those who came two or three decades ago. Since the enactment of the 1965 Amendments, the United States has been granting entry visas to persons who have relatives in the United States, with no regard to their skills or economic potential. Immigrants who arrived in the mid to late 1960s entered the U.S. labor market with a wage disadvantage of about 17 per cent; today's immigrants enter with a wage disadvantage of about 32 per cent. In the 1960s, the economic adaptation experienced by immigrants as they found out about job opportunities and learned to speak English guaranteed that the initial 17 per cent wage gap would disappear within a couple of decades. If today's immigrants have the same rate of adaptation as earlier immigrants, we can expect that they will have a wage disadvantage of about 15 per cent throughout much of their working lives. The inherently unstable combination of unskilled immigration and a generous welfare state has brought us to the current situation.

The geographic clustering of immigrants in the United States is remarkable. In 1990, three-quarters of the immigrants lived in only six states. This clustering has created a mosaic of ethnic neighborhoods or enclaves in some American cities, and has fostered the creation and growth of ethnic networks that transmit information about life in these United States to potential migrants in the source countries.

Do these ethnic networks provide information about welfare programs to new immigrants? Case studies of the Russian and Chinese communities leave little doubt that they do. Russian- and Chinese-language newspapers print detailed reports about the application process and eligibility requirements for particular programs. There are "Dear Abby"-style columns in newspapers to help readers with welfare problems. And bookstores in Taiwan, Hong Kong, and the United States sell a Chinese-language book that contains a 36-page guide to SSI and other benefits.

Network effects probably grew in importance as the networks became more established. It is well known that "takeup" rates (i.e., the fraction of eligible persons who actually receive benefits from a particular welfare program) are well below 100 per cent. As the ethnic networks expanded, the take-up rates of immigrants probably increased.

The welfare state can also have a magnetic effect on immigrants. Welfare programs in the United States, though not generous by Western European standards, stack up pretty well when compared to the standard of living available in most of the world's less-developed countries. While it is true that many immigrants come to the United States for job opportunities, decades of economic research into the determinants of migration decisions have demonstrated that it is potential income that is the significant factor. And the welfare state provides a lot of income opportunities, especially for persons with few skills. As a result, the question is not whether magnetic effects exist-they do. Rather, the question is whether these magnetic effects are numerically important.

Three different types of magnetic effects influence immigrant behavior. It is possible that welfare programs attract persons who otherwise would not have migrated to the United States. This is the magnetic effect most people have in mind, but is also the one about which there is least empirical evidence.

Second, the safety net might also discourage immigrants who "fail" in the United States from returning to their home countries. A recent study (by economists Randall Olsen and Patricia Regan of Ohio State University) provides the first hint of such a magnetic effect by showing that the probability of out-migration is greatly reduced if the household receives public assistance in the United States.

Finally, magnetic effects arise from the huge disparities between different states' welfare benefits. In 1970, California's AFDC benefit level was only 68 per cent that of the median state; twenty years later it was 2.5 times that of the median state. By 1990, California's benefit package was the second most generous in the nation (surpassed only by Alaska's). It turns out that the fraction of new immigrants not on welfare who chose to live in California dropped between 1980 and 1990, from 30.1 to 28.9 per cent. But the fraction of new immigrants on welfare who chose to live in California rose sharply, from 36.9 to 45.4 per cent. The evidence, therefore, suggests a clustering effect upon immigrant welfare recipients as California's benefit level rose above that of other states.

The existence of ethnic networks and magnetic effects implies that immigrants respond to variations in welfare benefits. As a result, it should not be too surprising that the welfare problem in the immigrant population has grown considerably in the past two decades.

In 1996, after years of concern over the link between welfare and immigration, Congress included a number of immigrant-related provisions in the welfare-reform bill. According to the Congressional Budget Office, almost half of the $54-billion savings in this legislation can be traced directly to the restrictions on immigrant use of welfare. The welfare-reform bill banned most types of assistance for immigrants who would enter the country after August 22, 1996 (with the ban being lifted when the immigrants become citizens), and it mandated that most non-citizens present in the country on August 22, 1996, be kicked off the SSI and Food Stamp rolls within a year. The welfare-reform legislation was a capitulation by Congress to the idea that the problem was indeed welfare, not immigration.

Congress was wrong, however. Even before the ink dried the rumblings had begun that the restrictions on immigrant use of welfare were onerous, inhumane, and unjust. The immigrant provisions brought together a number of powerful interest groups all of which lobbied hard for their repeal. Governors and mayors could read between the lines of the welfare-reform bill: Hundreds of thousands of current (and future) immigrants on the federal payroll would, more likely than not, become charges of state and local governments. After all, someone would have to bear the costs of supporting the disabled, many of whom are elderly and not employable. Immigrant-rights organizations stressed the unfairness of a piece of legislation that made no provision for legal immigrants who had paid their taxes, had been eligible to be drafted, and had contributed in many other ways to U.S. economic life. Finally, as the shutting-off date approaches, the everalert media have begun airing case studies of very ill immigrants who received letters from the Social Security Administration notifying them that their benefits would be cut off within months.

Not surprisingly, it seems as if these interest groups will carry the day. Congress has already voted to extend the deadline for cutting off SSI benefits to the disabled. And many of the restrictions on immigrant use of public assistance will probably be repealed. An important component of the budget agreement reached recently between the Republicans in Congress and President Clinton allocates funds for the provision of aid to disabled non-citizens.

The unraveling of this major piece of social legislation teaches us many lessons. No one much disputes the fact that immigrant use of social benefits grew rapidly in the past three decades. So it is hard to argue that the restrictions on immigrants in the welfare-reform bill were based on faulty analysis or data. Congress saw an actual problem-rising welfare use by immigrants-and tried to do something about it.

Our political system finds it difficult, if not impossible, to dismantle any part of the safety net protecting the economically disadvantaged, no matter how serious may be the intended consequences of that safety net. It is all but unimaginable that the day will come when TV cameras will record for posterity the instant an immigrant Alzheimer patient loses his SSI benefits, is evicted from his apartment or nursing home, and ends up wandering aimlessly around the streets of his neighborhood.

It is irresponsible to dismiss the problems associated with welfare and immigration by passing the blame off on the welfare state-and by asserting that we could solve the problem by repealing all the programs that make up the welfare state. It is far from clear that the removal of immigrants from the welfare rolls today would generate long-run budgetary savings-particularly for those programs where the welfare state remains active, such as the provision of emergency medical care to the needy. Pre-natal care, for instance, is far cheaper than the cost we would have to bear when an immigrant woman who had not seen an obstetrician arrived at a hospital emergency room to deliver her baby. Put differently, the cost of providing last-minute services to immigrants could easily exceed the cost of putting them on a "retainer" through welfare benefits.

My point is simple: It is unclear that we want to bear the political, social, and economic costs of removing immigrants already in the United States from the welfare rolls. That does not mean that we give up on reforming the immigration system. But we must adapt our immigration policy to this political reality. In the end, it will be much easier and cheaper to reduce the welfare costs of immigration not by "ending welfare as we know it," but by reforming immigration policy.

If we truly want to reduce the number of immigrants on the welfare rolls, we will have to restrict immigrant access to the welfare system at the point where the restrictions are most likely to be effective and stick-at the point of entry. We must accept the fact that the consequences of the misguided immigration policies that we pursued in the past three decades are here to stay. Sunk costs, as economists say, are sunk.

What types of point-of-entry restrictions might we want to consider? It is obvious, it has been said before, but it is worth saying again. Immigrant use of welfare programs would be greatly reduced if we improved the skill composition of the immigrant flow. It makes a great deal of economic sense to restrict the entry of persons who will probably become public charges. We could set up a point systems la Canada-where the success of a visa application would depend not only on whether the applicant had relatives in the United States, but also on the applicant's education, age, occupation, and job prospects. The point system would also nullify the influence of ethnic networks or magnets in generating excess use of welfare programs by immigrants.

In the absence of such point-of-entry restrictions, some states have begun to take unilateral actions that weaken the link between immigration and welfare within their borders. The welfare-reform bill gives states much more leeway in setting benefit levels. States will now compete when setting welfare benefits. Immigrant-receiving states, such as California, have a huge incentive to race to the bottom in benefits as they attempt to reduce the fiscal burden imposed by the immigration of less-skilled workers.

It is remarkable that while the heated debate over the relationship between welfare and immigration dominates the public arena, a consensus has evolved that the gains from immigration to the U.S. economy as a whole are small. The National Academy of Sciences reports that immigration generates a net gain for U.S. natives of at most $10 billion annually, or about $120 per native household in the United States. At the same time, however, the National Academy reports that immigration imposes a $1,174 annual fiscal burden on the typical native household in California, and a $229 annual fiscal burden on the typical native household in New Jersey.

We still have not really addressed the problems introduced by rising use of welfare benefits in the immigrant population, and the debate over the welfarereform bill indicates that we will not solve these problems until we tackle the much larger question of what type of immigration policy the United States should pursue.


Mr. Borjas is a Professor of Public Policy at the John F. Kennedy School of Government, Harvard University.