Michael Clemens, the leading proponent of a dramatic increase in less-skilled immigration to the U.S., writes in praise of a new OECD report which analyzes the fiscal impact of immigrants across 28 countries:
People who support more immigration from poorer countries think immigrants make a net positive fiscal contribution, and people who have the opposite opinion believe the opposite fact. Who is right?
The OECD analysis proceeds to calculate the true “net direct fiscal position” impact of immigrants, using the world’s best data harmonized across most OECD countries. The net direct fiscal position equals the taxes and social security contributions of immigrants minus the social transfers they receive. In the figure below, the vertical axis shows this net direct fiscal position for an average household. A positive number means they pay more in taxes than they use in benefits. The blue bars are immigrant-headed households, the white diamonds are native-headed households.
First, note that Clemens explicitly refers to support for immigration from poorer countries. The OECD’s analysis is not limited to immigrants from poorer countries. Rather, it includes immigrants from poor countries as well as affluent countries (“the OECD has taken the view that the foreign-born are the appropriate target group”). It is also important to keep in mind that immigration from poorer countries is heterogeneous, as some immigrants from poorer countries are highly educated while others are not. The OECD calculation of “net direct fiscal position” Clemens cites does not differentiate between skilled migrants and less-skilled migrants. It also does not differentiate between the old and the young. Keep in mind that one of the central issues in the immigration debate in the U.S. and throughout the OECD is whether skilled migrants should be given a preference over less-skilled migrants.
To its great credit, the OECD makes clear that its analysis is very sensitive to its assumptions:
The fiscal impact of immigration cannot be pinned down to a single and undisputable figure, as its measurement depends on a number of key assumptions, including the degree to which the cost for the public purse of certain public services and the public capital stock (such as for infrastructure and public administration) and non-personal taxes (such as the corporate income tax) is attributed to the immigrant population. Inclusion or exclusion of these items often changes the sign of the impact.
Clemens continues:
In the United States, the OECD finds that the fiscal impact of the average native household and the average immigrant household are similar and positive. The average immigrant household makes a positive contribution equivalent to more than 8,000 Euros per year (around US$11,000) more in taxes than they take out in benefits.
This impact is “direct” because it only counts current cashflows into and out of the public purse. Such an estimate is conservatively low because, the OECD notes, it “neglects the indirect implications resulting from migrants’ broader impact on the economy”. Other analyses show that including these broader economic stimulus effects raises the fiscal impact of US immigration, such as a recent peer-reviewed academic study by Chojnicki, Docquier, and Ragot. It is furthermore conservatively low because it omits the fiscal impact of US immigrants’ children, which is known to be positive and large.
Let’s turn to the abstract of the Chojnicki, Docquier, and Ragot study (“Should the US have locked heaven’s door?,” published in the Journal of Population Economics in January 2011):
This paper examines the economic impact of the second great immigration wave (1945–2000) on the US economy. Our analysis relies on a computable general equilibrium model combining the major interactions between immigrants and natives (labor market impact, fiscal impact, capital deepening, endogenous education, endogenous inequality). Contrary to recent studies, we show that immigration induced important net gains and small redistributive effects among natives. According to our simulations, the postwar US immigration is beneficial for all natives cohorts and all skill groups. Nevertheless, the gains would have been larger if the US had conducted a more selective immigration policy.
What exactly did Chojnicki et al. mean by “a more selective immigration policy”? The authors construct a selected immigration (SI) scenario, which is designed to illustrate the impact of the relative immigrant skill decrease since 1965.
Our analysis suggests that the US postwar immigration is beneficial for all the parties concerned. Would the gains from immigration be larger if the US had pursued a selective immigration policy? The selected variant relies on the same stocks of immigration than the baseline. It assumes that the skill distribution of immigrants converges towards the distribution of natives. In such a scenario, the average education level of immigrants increases but the average experience decreases (immigrants have less experience but more schooling). Quantitatively, such a scenario induces similar effects on the skill premium, wage inequality and taxes than the no immigration variant (Table 3). The magnitude of the effects is usually lower.
As appearing on figure 3b and table 3, all natives would have gained from a stronger selection: all determinants of the utility level are positively affected. Contrary to the no immigration variant, the distributive impact are more important. A stronger selection would obviously be more profitable to low-skill workers than to medium and high-skill workers. For recent cohorts, gains for the low skilled are twice as large as for the highly skilled.
The authors conclude that “all generations would have benefited from a stronger selection of immigrants,” and that selection “would have particularly benefited to low-skill workers.” This strikes me as an important finding, which policymakers ought to take seriously.
One of the more peculiar aspects of Clemens’ argument is that he claims that while “taxpayers like me do not finance schooling for other families’ children because we enjoy giving cash for other people to blow on frivolous lifestyle consumption,” he implicitly suggests that taxpayers like him finance means-tested transfers programs because they do enjoy giving cash for other people to blow on frivlous lifestyle consumption. This is a definition of means-tested transfer programs that some conservatives and libertarians might be inclined to embrace, but it doesn’t strike me as very sound:
The impacts above exclude the cost of schooling for young children. That is a correct omission. Taxpayers like me do not finance schooling for other families’ children because we enjoy giving cash for other people to blow on frivolous lifestyle consumption. We finance public schooling because in a society where today’s adults got education when they were children, everything is better—including the economy. Educating the next generation is not financing the consumption of others, it’s an investment in the nation, an investment with positive returns.
One assumes that advocates of foodstamps would characterize SNAP as a measure that makes “everything better—including the economy,” e.g., because transfers to low-income households have broader economic stimulus effects and because they help prevent children raised in low-income households from being underfed and underweight. Indeed, the stimulus effects of transfers to low-income households have been a central part of the economic policy debate in the post-crisis era. The same logic applies to Medicaid and a wide range of other anti-poverty programs. The distinction between consumption and investment is not as clear as Clemens suggests. Since we can characterize foodstamps and Medicaid as “an investment in the nation, an investment with positive returns,” surely we should also acknowledge that the financing of K-12 education doesn’t simply yield benefits for “the nation”—it also yields benefits for its direct beneficiaries as well, by raising their future incomes, etc.
Beyond this, it would be nonsensical to compare schooling usage by non-migrants and migrants. Non-migrants tend to be much older on average than typically young migrant families. Across the entire United States, elderly Americans’ property taxes finance public schooling for other American families’ young children, even though elderly Americans’ kids left school long ago. I’ve never heard a nativist politician calling for an end to those subsidies on the grounds that young American families are “consuming” the money of the elderly. Public money that is spent educating the next generation of Americans—whether those Americans’ parents were born here or elsewhere, has positive returns for society. Education is an investment.
Clemens’ seems to believe that support for K-12 spending is rooted in an “investment in the nation” hypothesis, in which voters are largely indifferent to whether or not they benefit from it. Why, asks Clemens, would elderly people favor spending money on K-12 schools if their own children left school many years ago? One problem is that elderly people in some jurisdictions actually do oppose high levels of K-12 spending while others favor it. Much seems to depend on whether or not voters benefit from school quality via the channel of local property values.
Dartmouth economist William Fischel’s work on the homevoter hypothesis and the history of K-12 school districts strongly suggests that support for K-12 financing at the local level has been greatly influenced by the fact that the quality of local schools is capitalized in property values. See the abstract of Fischel’s 1998 paper on school finance legislation and property tax revolts:
The use of local property taxes to fund public schools in the United States has been under attack since the 1970s as a result of reform-minded lawsuits. Court-ordered reforms typically involve a greater proportion of state funding, more equal expenditures, and less local fiscal control. I explain in nontechnical language why this movement has reduced educational quality. The more extreme cases, such as Serrano v. Priest in California, have contributed to tax revolts that have starved education. The advantage of local fiscal control is that home values rise when schools get better, provided that the additional property-tax bite is not excessive. All homeowners, not just those with school-age children, have an interest in efficiently-run schools when education is financed locally. This fiscal feedback is lost when school funds are provided from statewide tax revenues.
What I find most peculiar about Clemens’ post is that he characterizes it as “the last nail in the coffin for nativists’ cooked-up narratives about benefits use by immigrants,” yet the OECD analysis itself includes many caveats and cautionary notes regarding the limitations of its data sources. Clemens quotes lead investigators behind the report, Jean-Christophe Dumont:
“Immigrants are thus neither a burden to the public purse nor are they a panacea for addressing fiscal challenges. They have a positive net direct fiscal position in most countries, except in those with a large share of older migrants. … Contrary to public belief, low-educated immigrants have a better fiscal position than their native-born peers. And where immigrants have a less favorable fiscal position, this is not driven by a greater dependence on social benefits, but rather by the fact that they often have lower wages and thus tend to contribute less.
“Most immigrants do not come for social benefits, they come to find work and to improve their lives and those of their families. Indeed, employment is the single most important determinant of migrants’ net fiscal contribution and raising immigrants’ employment rate to that of the native born would result in substantial fiscal gains, notably in European OECD countries. Efforts to better integrating immigrants should thus be seen as an investment rather than a cost.”
Here is the a slightly longer excerpt from Dumont’s post:
In the 2013 edition of the OECD flagship publication on migration, the International Migration Outlook, we present new evidence on this issue for all European OECD countries, plus Australia, Canada and the United States. We show that although the fiscal impact of immigration cannot be pinned down to a single and undisputable figure – as its measurement depends on many assumptions – the impact of the cumulative waves of migration that arrived over the past fifty years in OECD countries is on average close to zero, rarely exceeding 0.5% of GDP in either positive or negative terms.
Immigrants are thus neither a burden to the public purse nor are they a panacea for addressing fiscal challenges. They have a positive net direct fiscal position in most countries, except in those with a large share of older migrants. This means that they contribute to the financing of public infrastructures although admittedly to a lesser extent than the native-born. Contrary to public belief, low–educated immigrants have a better fiscal position than their native-born peers. And where immigrants have a less favourable fiscal position, this is not driven by a greater dependence on social benefits, but rather by the fact that they often have lower wages and thus tend to contribute less.
Most immigrants do not come for social benefits, they come to find work and to improve their lives and those of their families. Indeed, employment is the single most important determinant of migrants’ net fiscal contribution and raising immigrants’ employment rate to that of the native born would result in substantial fiscal gains, notably in European OECD countries. Efforts to better integrating immigrants should thus be seen as an investment rather than a cost.
Clemens might object to the “on average close to zero” because it discounts the economic stimulus effects of immigration, etc. Of course, less-skilled immigration might exacerbate linguistic isolation or cause other problems that can’t be rigorously, but let’s accept what I take to be one of Clemens’ central premises—than any reference to cultural considerations is racist or ethnocentrist. There are nevertheless a few issues we might raise with Dumont’s conclusions:
1. He doesn’t allow for a distinction between skilled migrants and less-skilled migrants—it is not at all clear why it would be wrong for the United States to craft an immigration policy that welcomes skilled migrants under a certain age while limiting the influx of less-skilled migrants and older migrants, the latter of whom will have fewer peak earning years than younger migrants.
2. He suggests that the relevant public belief is that “low-educated immigrants have a better fiscal position than their native-born peers.” I’m not actually sure that the public belief in the United States is that less-skilled immigrants from Latin America contribute less in taxes and consume more in benefits than native-born individuals with similarly low levels of educational attainment in rural poor in Appalachia or in inner-city neighborhoods. Even if this were true, however, it doesn’t seem as though the fact that less-skilled immigrants have a better fiscal position than less-skilled natives ought to have meaningful implications for U.S. immigration policy given that the U.S. has the option of recruiting immigrants with a much better fiscal position than less-skilled natives.
3. In arguing that OECD countries ought to see the expenditures associated with educating the children of less-skilled immigrants from countries in which entrenched intergenerational poverty is pervasive as an investment is an interesting one. But it doesn’t strike me as obvious that OECD countries should not, as a general rule, prefer skilled migrants, as the children of skilled migrants might on average have a number of advantages over the children of less-skilled migrants. The OECD countries in question would still be investing in the children of immigrants.
Michael Clemens and his allies have made mincemeat of the arguments that all immigration should immediately cease and that immigrants with an 8th grade education who are either banned from receiving various means-tested transfers or face various challenges in accessing them (including limited English language proficiency) make less of a net direct fiscal contribution than native-born Americans with an 8th grade education who are much less likely to work and who are eligible for a full complement of means-tested transfers. Though I haven’t really encountered these arguments in the mainstream immigration debate (zero immigration? less-skilled natives are productivity dynamos who make much bigger contributions to the public fisc than their immigrants peers?), I’m willing to concede that influential, non-fringe people are making them. What advocates of a sharp increase in less-skilled haven’t done is seriously dent the argument that if the U.S. cares about net direct fiscal contribution, it ought to craft an immigration policy that favors the skilled over the less-skilled and young adults over old adults.