A working paper released this week by Notre Dame economists William Evans and Daniel Fitzgerald makes the head-scratching claim that refugees, despite below-average incomes and high rates of welfare use, pay $21,000 more in taxes than they receive in benefits during their first 20 years in the United States. Immigration-boosting wonks such as Matt Yglesias and Dylan Matthews immediately trumpeted the findings, and the Washington Post and FiveThirtyEight added favorable write-ups.
They should have been more skeptical. The claim that refugees contribute more in taxes than they receive in benefits is simply implausible. The public sector struggles with budget deficits every year, and it relies on the taxes paid by the highest-earning Americans to stay even close to fiscal balance. If disproportionately low-earning and high-welfare-consuming Americans were somehow net fiscal contributors, then governments at all levels would be deep in the black, and politicians would be sparring over what to do with the massive surpluses.
So how does the Evans-Fitzgerald paper come to such an implausible result? First, the authors count all (or nearly all) taxes paid by refugees but reduce the services they receive to six social programs — cash welfare, SSI, Social Security, food stamps, Medicare, and Medicaid. All other costs that governments might incur from immigration — housing, infrastructure, education, law enforcement, and so on — do not count.
Second, they fail to adjust for the underreporting of those social programs in the American Community Survey (ACS). The difference between reported costs in survey data and actual costs can be large. For example, we know from administrative data that about 22 million households were on food stamps in 2012, but the 2012 ACS counts fewer than 16 million of them. (In fact, these numbers understate the reporting gap. The administrative data give us the number of households on food stamps in a typical month, while the ACS asks each household whether it has received food stamps at any time in the past 12 months. The ACS count should actually be higher than the administrative count.) On the other side of the ledger, the ACS captures 83 to 87 percent of earnings by survey respondents. Adjusting survey-based earnings and program costs upward to reflect reality is essential for meaningful results.
Third, the paper excludes refugees' minor children. When refugees cannot afford to provide food, housing, or medical care to their children, taxpayers foot the bill. Most of those costs are omitted.
Fourth, the authors restrict the refugee age range to 18-65, cutting off the analysis just before the age where most people stop working and begin participating in the nation's costly retirement programs.
The problems listed above are not ad hoc concerns. Experts in fiscal analysis have long known to avoid these pitfalls. The best example is a book-length report on immigration published last fall by the National Academies of Sciences (NAS). When the NAS performed its own fiscal impact analysis, it tallied all government services, adjusted for undercount in survey data, included minor children, and examined immigrants over their whole age range. The results were clear: The least-educated immigrants pay less in taxes than they receive in services, while the highest-educated pay more.
Refugee resettlement is a complex issue, and the fiscal impact is only one part of the debate. Policymakers should have no illusions, however, about the budgetary cost of welcoming people who have lower earnings and higher welfare consumption.