T HE United States is on the verge of another great debate over immigration. Thus far the focus of this still-inchoate debate has been on illegal immigration or welfare benefits to legal immigrants, not on the larger issue of the character and consequences of the current high levels of legal immigration. Economic factors by themselves should not and will not decide the outcome of this debate. But they will play an important role. Economics helps us to frame answerable questions about immigration: Who gains by it? Who loses? And in light of the answers to these questions, what should U.S. immigration policy be?
There have been two major shifts in immigration policy in this century. In the twenties the United States began to limit the number of immigrants admitted and established the national-origins quota system, an allocation scheme that awarded entry visas mainly on the basis of national origin and that favored Germany and the United Kingdom. This system was repealed in 1965, and family reunification became the central goal of immigration policy, with entry visas being awarded mainly to applicants who had relatives already residing in the United States.
The social, demographic, and economic changes initiated by the 1965 legislation have been truly historic. The number of immigrants began to rise rapidly. As recently as the 1950s only about 250,000 immigrants entered the country annually; by the 1990s the United States was admitting more than 800,000 legal immigrants a year, and some 300,000 aliens entered and stayed in the country illegally. The 1965 legislation also led to a momentous shift in the ethnic composition of the population. Although people of European origin dominated the immigrant flow from the country's founding until the 1950s, only about 10 percent of those admitted in the 1980s were of European origin. It is now estimated that non-Hispanic whites may form a minority of the population soon after 2050. More troubling is that immigration has been linked to the increase in income inequality observed since the 1980s, and to an increase in the costs of maintaining the programs that make up the welfare state.
These economic and demographic changes have fueled the incipient debate over immigration policy. For the most part, the weapons of choice in this debate are statistics produced by economic research, with all sides marshaling facts and evidence that support particular policy goals. In this essay I ask a simple question: What does economic research imply about the kind of immigration policy that the United States should pursue?
EVERY immigration policy must resolve two distinct issues: how many immigrants the country should admit, and what kinds of people they should be.
It is useful to view immigration policy as a formula that gives points to visa applicants on the basis of various characteristics and then sets a passing grade. The variables in the formula determine what kinds of people will be let into the country, and the passing grade determines how many will be let into the country. Current policy uses a formula that has one overriding variable: whether the visa applicant has a family member already residing in the United States. An applicant who has a relative in the country gets 100 points, passes the test, and is admitted. An applicant who does not gets 0 points, fails the test, and cannot immigrate legally.
Of course, this is a simplistic summary of current policy. There are a lot of bells and whistles in the immigration statutes (which are said to be only slightly less complex than the tax code). In fact the number of points a person gets may depend on whether the sponsor is a U.S. citizen or a permanent resident, and whether the family connection is a close one (such as a parent, a spouse, or a child) or a more distant one (a sibling). Such nuances help to determine the speed with which the visa is granted. A limited number of visas are given to refugees. Some are also distributed on the basis of skill characteristics, but these go to only seven percent of immigrants.
Although the United States does not officially admit to using a point system in awarding entry visas, other countries proudly display their formulas on the Internet. A comparison of these point systems reveals that the United States is exceptional in using essentially one variable. Canada, Australia, and New Zealand have more-complex formulas that include an applicant's educational background, occupation, English-language proficiency, and age along with family connections.
Sometimes a host country awards points to people who are willing to pay the visa's stated price. Canada, for example, has granted entry to virtually anyone who would invest at least $250,000 in a Canadian business. Although this "visas-for-sale" policy is a favorite proposal of economists (if we have a market for butter, why not also a market for visas?), it is not taken very seriously in the political debate, perhaps because policymakers feel a repugnance against what may be perceived as a market for human beings. I will therefore discuss the implications of economic research only for policies in which points are awarded on the basis of socioeconomic characteristics, not exchanged for dollars.
THE academic literature investigating the economic impact of immigration on the United States has grown rapidly in the past decade. The assumptions that long dominated discussion of the costs and benefits of immigration were replaced during the 1980s by a number of new questions, issues, and perceptions.
Consider the received wisdom of the early 1980s. The studies available suggested that even though immigrants arrived at an economic disadvantage, their opportunities improved rapidly over time. Within a decade or two of immigrants' arrival their earnings would overtake the earnings of natives of comparable socioeconomic background. The evidence also suggested that immigrants did no harm to native employment opportunities, and were less likely to receive welfare assistance than natives. Finally, the children of immigrants were even more successful than their parents. The empirical evidence, therefore, painted a very optimistic picture of the contribution that immigrants made to the American economy.
In the past ten years this picture has altered radically. New research has established a number of points.
The relative skills of successive immigrant waves have declined over much of the postwar period. In 1970, for example, the latest immigrant arrivals on average had 0.4 fewer years of schooling and earned 17 percent less than natives. By 1990 the most recently arrived immigrants had 1.3 fewer years of schooling and earned 32 percent less than natives.
Because the newest immigrant waves start out at such an economic disadvantage, and because the rate of economic assimilation is not very rapid, the earnings of the newest arrivals may never reach parity with the earnings of natives. Recent arrivals will probably earn 20 percent less than natives throughout much of their working lives.
The large-scale migration of less-skilled workers has done harm to the economic opportunities of less-skilled natives. Immigration may account for perhaps a third of the recent decline in the relative wages of less-educated native workers.
The new immigrants are more likely to receive welfare assistance than earlier immigrants, and also more likely to do so than natives: 21 percent of immigrant households participate in some means-tested social-assistance program (such as cash benefits, Medicaid, or food stamps), as compared with 14 percent of native households.
The increasing welfare dependency in the immigrant population suggests that immigration may create a substantial fiscal burden on the most-affected localities and states.
There are economic benefits to be gained from immigration. These arise because certain skills that immigrants bring into the country complement those of the native population. However, these economic benefits are small -- perhaps on the order of $7 billion annually.
There exists a strong correlation between the skills of immigrants and the skills of their American-born children, so that the huge skill differentials observed among today's foreign-born groups will almost certainly become tomorrow's differences among American-born ethnic groups. In effect, immigration has set the stage for sizable ethnic differences in skills and socioeconomic outcomes, which are sure to be the focus of intense attention in the next century.
The United States is only beginning to observe the economic consequences of the historic changes in the numbers, national origins, and skills of immigrants admitted over the past three decades. Regardless of how immigration policy changes in the near future, we have already set in motion circumstances that will surely alter the economic prospects of native workers and the costs of social-insurance programs not only in our generation but for our children and grandchildren as well.
IF economic research is to play a productive role in the immigration debate, research findings should help us to devise the formula that determines admission into the United States. We need to decide what variables are to be used to award points to applicants, and what is to be the passing grade. Before we can resolve these issues, however, we have to address a difficult philosophical question: What should the United States try to accomplish with its immigration policy?
The answer to this question is far from obvious, even when the question is posed in purely economic terms. We can think of the world as composed of three distinct groups: people born in the United States (natives), immigrants, and people who remain in other countries. Whose economic welfare should the United States try to improve when setting policy -- that of natives, of immigrants, of the rest of the world, or of some combination of the three? The formula implied by economic research depends on whose interests the United States cares most about.
Different political, economic, and moral arguments can be made in favor of each of the three groups. I think that most participants in the U.S. policy debate attach the greatest (and perhaps the only) weight to the well-being of natives. This is not surprising. Natives dominate the market for political ideas in the United States, and most proposals for immigration reform will unavoidably reflect the self-interest and concerns of native voters.
Immigration almost always improves the well-being of the immigrants. If they don't find themselves better off after they enter the United States, they are free to go back or to try their luck elsewhere -- and, indeed, some do. A few observers attach great weight to the fact that many of the "huddled masses" now live in relative comfort.
As for the vast populations that remain in the source countries, they are affected by U.S. immigration policy in a number of ways. Most directly, the policy choices made by the United States may drain particular skills and abilities from the labor markets of source countries. A brain drain slows economic growth in the source countries, as the entrepreneurs and skilled workers who are most likely to spur growth move to greener pastures. Similarly, the principles of free trade suggest that world output would be largest if there were no national borders to interfere with the free movement of people. A policy that restricts workers from moving across borders unavoidably leads to a smaller world economy, to the detriment of many source countries.
The three groups may therefore have conflicting interests, and economics cannot tell us whose interests matter most. The weight that we attach to each of the three groups depends on our values and ideology. For the sake of argument I will assume a political consensus that the objective of immigration policy is to improve the economic well-being of the native population.
Beyond that, we have to specify which dimension of native economic well-being we care most about: per capita income or distribution of income. As we shall see, immigration raises per capita income in the native population, but this does not mean that all natives gain equally. In fact some natives are likely to see their incomes greatly reduced. We must therefore be able to judge an immigration policy in terms of its impact on two different economic dimensions: the size of the economic pie (which economists call "efficiency") and how the pie is sliced ("distribution"). The relative weights that we attach to efficiency and distribution again depend on our values and ideology, and economics provides no guidance on how to rank the two.
For the most part, economists take a very narrow approach: policies that increase the size of the pie are typically considered to be better policies, regardless of their impact on the distribution of wealth in society. We shall begin our construction of an immigration policy by taking this narrow approach. In other words, let's assume that immigration policy has a single and well-defined purpose: to maximize the size of the economic pie available to the native population of the United States. We shall return to the distributional issues raised by immigration policy later on.
TO see how natives gain from immigration, let's first think about how the United States gains from foreign trade. When we import toys made by cheap Chinese labor, workers in the American toy industry undoubtedly suffer wage cuts and perhaps even lose their jobs. These losses, however, are more than offset by the benefits accruing to consumers, who enjoy the lower prices induced by additional competition. An important lesson from this exercise, worth remembering when we look at the gains from immigration, is that for there to be gains from foreign trade for the economy as a whole, some sectors of the economy must lose.
Consider the analogous argument for immigration. Immigrants increase the number of workers in the economy. Because they create additional competition in the labor market, the wages of native workers fall. At the same time, however, native-owned firms gain, because they can hire workers at lower wages; and many native consumers gain because lower labor costs lead to cheaper goods and services. The gains accruing to those who consume immigrants' services exceed the losses suffered by native workers, and hence society as a whole is better off.
Immigration therefore has two distinct consequences. The size of the economic pie increases. And a redistribution of income is induced, from native workers who compete with immigrant labor to those who use immigrants' services.
The standard economic model of the labor market suggests that the net gain from immigration is small. The United States now has more than 20 million foreign-born residents, making up slightly less than 10 percent of the population. I have estimated that native workers lose about $133 billion a year as a result of this immigration (or 1.9 percent of the gross domestic product in a $7 trillion economy), mainly because immigrants drive down wages. However, employers -- from the owners of large agricultural enterprises to people who hire household help -- gain on the order of $140 billion (or 2.0 percent of GDP). The net gain, which I call the immigration surplus, is only about $7 billion. Thus the increase in the per capita income of natives is small -- less than $30 a year. But the small size of this increase masks a substantial redistribution of wealth.
My calculation used the textbook model of a competitive labor market: wages and employment are determined in a free market that balances the desires of people looking for work with the needs of firms looking for workers. In this framework an increase in the number of workers reduces wages in the economy -- immigrants join natives in the competition for jobs and bid down wages in the process. There is a lot of disagreement over how much native wages fall when immigrants enter the labor market. Nevertheless, a great deal of empirical research in economics, often unrelated to the question of immigration, concludes that a 10 percent increase in the number of workers lowers wages by about three percent.
If we accept this finding, we can argue as follows: We know that about 70 percent of GDP accrues to workers (with the rest going to the owners of companies), and that natives make up slightly more than 90 percent of the population. Therefore, native workers take home about 63 percent of GDP in the form of wages and salaries. If the 10 percent increase in the number of workers due to immigration has lowered wages by three percent, the share of GDP accruing to native workers has fallen by 1.9 percentage points (or 0.63 x 0.03). Thus my conclusion that in a $7 trillion economy native earnings drop by $133 billion.
Those lost earnings do not vanish into thin air. They represent an income transfer from workers to users of immigrants' services -- the employers of immigrants and the consumers who buy the goods and services produced by immigrants. These winners get to pocket the $133 billion -- and then some, because the goods produced by immigrant workers generate additional profits for employers. Under the assumption that a 10 percent increase in the number of workers reduces wages by three percent, it turns out that the winners get a windfall totaling $140 billion. Hence the $7 billion immigration surplus.
We can quibble about assumptions, but the rigor of economic theory suggests that this nitpicking may not alter our conclusions much. For example, one could argue -- and many do -- that immigrants do not reduce the earnings of native workers. If we wished to believe this, however, we would also be forced to conclude that natives do not benefit from immigration at all. If wages do not fall, there are no savings in employers' payrolls and no cost savings to be passed on to native consumers. Remember the lesson from the foreign-trade example: no pain, no gain.
One could also argue that immigration has reduced the earnings of natives very substantially -- by, say, 10 percent. The immigration surplus would then be about $25 billion annually. The net gain from immigration, therefore, remains small even with an unrealistically high estimate of the impact of immigration on native earnings. Imagine what U.S. policy would look like today if our earnings had fallen by 10 percent as a result of past immigration.
The immigration surplus has to be balanced against the cost of providing services to the immigrant population. Immigrants have high rates of welfare recipiency. Estimates of the fiscal impact of immigration (that is, of the difference between the taxes paid by immigrants and the cost of services provided to them) vary widely. Some studies claim that immigrants pay $25-$30 billion more in taxes than they take out of the system, while other studies blame them for a fiscal burden of more than $40 billion on natives.
It is doubtful that either of these statistics accurately reflects the gap between taxes paid and the cost of services provided. Studies that claim a beneficial fiscal impact tend to assume that immigrants do not increase the cost of most government programs other than education and welfare. Even though we do not know by how much immigrants increase the cost of police protection, maintaining roads and national parks, and so forth, we do know that it costs more to provide these services to an ever larger population. However, studies that claim a large fiscal burden often overstate the costs of immigration and understate the taxes paid. As a result, estimates of the fiscal impact of immigration should be viewed with suspicion. Nevertheless, because the immigration surplus is around $7 billion, the net benefit from immigration after accounting for the fiscal impact is very small, and could conceivably be a net loss.
IN principle, we should admit immigrants whenever their economic contribution (to native well-being) will exceed the costs of providing social services to them. We are not, though, in a position to make this calculation with any reasonable degree of confidence. In fact, no mainstream study has ever attempted to suggest, purely on the basis of the empirical evidence, how many immigrants should be admitted.
This unfortunate lack of guidance from economic research has, I believe, led to sudden and remarkable swings in policy proposals. As recently as 1990 Congress legislated an increase in the number of legal immigrants of about 175,000 people annually. Last year the Commission on Immigration Reform, headed by Barbara Jordan, recommended that legal immigration be cut by about 240,000 people a year -- a proposal that was immediately supported by President Clinton. (The Clinton Administration, however, successfully resisted congressional efforts to follow up on the commission's recommendations.)
Although we do not know how many immigrants to admit, simple economics and common sense suggest that the magic number should not be an immutable constant regardless of economic conditions in the United States. A good case can be made for linking immigration to the business cycle: admit more immigrants when the economy is strong and the unemployment rate is low, and cut back on immigration when the economy is weak and the unemployment rate is high.
Economic research also suggests that the United States may be better off if its policy of awarding entry visas favors skilled workers. Skilled immigrants earn more than less-skilled immigrants, and hence pay more in taxes, and they are less likely to use welfare and other social services.
Depending on how the skills of immigrants compare with the skills of natives, immigrants also affect the productivity of the native work force and of native-owned companies. Skilled native workers, for example, have much to gain when less-skilled workers enter the United States: they can devote all their efforts to jobs that use their skills effectively while immigrants provide cheap labor for service jobs. These gains, however, come at a cost. The jobs of less-skilled natives are now at risk, and these natives will suffer a reduction in their earnings. Nonetheless, it does not seem far-fetched to assume that the American work force, particularly in comparison with the work forces of many source countries, is composed primarily of skilled workers. Thus the typical American worker would seem to gain from unskilled immigration.
How does immigration affect companies' profits? Companies that use less-skilled workers on the production line gain from the immigration of the less-skilled, who reduce the earnings of less-skilled workers in favor of increasing profits. However, other companies -- perhaps even most -- might be better off with skilled immigrants. Many studies in economics suggest that skilled labor is better suited to the machines that are now used widely in the production process. Most companies would therefore gain more if the immigrant flow were composed of skilled workers.
Most workers prefer unskilled immigrants, whereas most companies prefer skilled immigrants. This conflict can be resolved only by measuring how much native workers gain from unskilled immigration and how much companies gain from skilled immigration, and comparing the two. Although there is a lot of uncertainty in the academic literature, we do know that the productivity of capital is very responsive to an influx of skilled workers. The large increase in the profits of the typical company, and the corresponding reduction in the cost of goods produced by skilled workers, suggest that the United States might be better off with a policy favoring skilled immigrants.
The gains from skilled immigration will be even larger if immigrants have "external effects" on the productivity of natives. One could argue, for example, that immigrants may bring knowledge, skills, and abilities that natives lack, and that natives might somehow pick up this know-how by interacting with immigrants. It seems reasonable to suspect that the value of these external effects would be greater if natives interact with highly skilled immigrants. This increase in the human capital of natives might offset -- and perhaps even reverse -- the harm that immigration does to the wages of competing workers.
Although such effects now play a popular role in economic theory, there is little empirical evidence supporting their existence, let alone measuring their magnitude. I find it difficult to imagine that interaction with immigrants entering an economy as large as that of the United States could have a measurable effect. Nevertheless, if external effects exist, they reinforce the argument that the United States would gain most from skilled immigrants.
PARTICIPANTS in the immigration debate routinely use the results of economic research to frame the discussion and to suggest policy solutions. Perhaps the most important contributions of this research are the insights that immigration entails both gains and losses for the native population, that the winners and the losers are typically different groups, and that policy parameters can be set in ways that attempt to maximize gains and minimize losses. If the objective of immigration policy is to increase the per capita income of the native population, the evidence suggests that immigration policy should encourage the entry of skilled workers. It is important to remember, however, that even though the immigration of skilled workers would be beneficial for the United States as a whole, the gains and losses would be concentrated in particular subgroups of the population.
As we have seen, the net gains from current immigration are small, so it is unlikely that these gains can play a crucial role in the policy debate. Economic research teaches a very valuable lesson: the economic impact of immigration is essentially distributional. Current immigration redistributes wealth from unskilled workers, whose wages are lowered by immigrants, to skilled workers and owners of companies that buy immigrants' services, and from taxpayers who bear the burden of paying for the social services used by immigrants to consumers who use the goods and services produced by immigrants.
Distributional issues drive the political debate over many social policies, and immigration policy is no exception. The debate over immigration policy is not a debate over whether the entire country is made better off by immigration -- the gains from immigration seem much too small, and could even be outweighed by the costs of providing increased social services. Immigration changes how the economic pie is sliced up -- and this fact goes a long way toward explaining why the debate over how many and what kinds of immigrants to admit is best viewed as a tug-of-war between those who gain from immigration and those who lose from it.
History has taught us that immigration policy changes rarely, but when it does, it changes drastically. Can economic research play a role in finding a better policy? I believe it can, but there are dangers ahead. Although the pendulum seems to be swinging to the restrictionist side (with ever louder calls for a complete closing of our borders), a greater danger to the national interest may be the few economic groups that gain much from immigration. They seem indifferent to the costs that immigration imposes on other segments of society, and they have considerable financial incentives to keep the current policy in place. The harmful effects of immigration will not go away simply because some people do not wish to see them. In the short run these groups may simply delay the day of reckoning. Their potential long-run impact, however, is much more perilous: the longer the delay, the greater the chances that when immigration policy finally changes, it will undergo a seismic shift -- one that, as in the twenties, may come close to shutting down the border and preventing Americans from enjoying the benefits that a well-designed immigration policy can bestow on the United States.
The Atlantic Monthly; November 1996; The New Economics of Immigration; Volume 278, No. 5; pages 72-80.