Subsidizing Remittances Through Welfare

The reconciliation bill signed by the president last week contains, among many other things, a “remittance tax”, which assesses a 1 percent fee on certain transfers of money abroad. Since earning money to send back home is one of the most common motivations for coming to the U.S., taxing remittances may discourage illegal immigration at the margins. A remittance tax could also help collect taxes on income that illegal immigrants would not otherwise report.

This post offers another justification for a remittance tax, which is to recover welfare money sent abroad. As discussed below, there appears to be significant overlap between households that send remittances and households that receive welfare. When U.S. taxpayers give means-tested benefits to residents who remit money outside our borders, the taxpayers are in effect subsidizing a portion of those remittances. A remittance tax could help recover U.S. taxpayer funds that were intended to raise living standards here but were instead diverted abroad.

Immigrant Welfare Use Is High

For context, the Center’s analysis of 2022 data shows that just over half of immigrant-headed households receive cash, food, medical care, or housing benefits from taxpayers. These benefits flow either to the immigrants themselves or to their U.S.-born children and other co-residents. The distinction makes little practical difference, however, since households are usually one economic unit, and any benefit to dependents is effectively a benefit to the household head who is responsible for them.

If there were no significant overlap between households that send remittances and households that receive welfare, remittances would have to come almost exclusively from the non-welfare receiving half of immigrant households. The next section investigates that proposition.

Evidence of Remittance-Welfare Overlap

Questions about remittances are hard to find in government surveys, and finding them alongside questions about welfare receipt is harder still. In fact, no dataset (to our knowledge) allows for an assessment of the remittance-welfare overlap on a large and representative sample of the U.S. population. However, the Census did add a set of supplemental “International Migration” questions to the August 2008 Current Population Survey (CPS). This one-time supplement contains the following question: “In the last 12 months, did anyone in this household give or send money to relatives or friends living outside the U.S.? Please include all gifts of money. Do not include loans.”

While that question directly measures remittances, the August CPS does not ask about means-tested benefits, which are instead tallied in the March supplement. Therefore, to measure remittances and welfare use in the same households, this analysis focuses on the small subset of responding households in August 2008 that were also interviewed in March 2009. This subset is not representative — the immigrants in it are all non-Hispanic, and they have a lower rate of welfare use than the overall sample. (See “Methodological Notes” for further discussion.) Nevertheless, the subset can still demonstrate the existence of a remittance-welfare overlap.

Table 1 shows the overlap between remittances and welfare in the CPS subset. Among immigrant-headed households that received welfare, 19 percent also sent remittances. This is a weighted calculation based on 41 welfare-using immigrant-headed households in the subset that sent remittances and 138 that did not. Conversely, among immigrant-headed households that sent remittances, 20 percent were also on welfare.

Interpretation

As noted above, the subset is unrepresentative, so the figures in Table 1 are not necessarily indicative of the general population. What Table 1 strongly suggests, however, is that a substantial remittance-welfare overlap does exist. Although this finding may seem counterintuitive, it should not be too surprising given the large proportion of immigrant-headed households that use welfare in the first place.

In fact, Table 1 probably understates the overlap. The reason is that the analyzed subset does not include Hispanics. Using separate analyses of the August 2008 CPS (for remittances) and the March 2009 CPS (for welfare), Table 2 demonstrates that Hispanic immigrants are more likely than non-Hispanics to use welfare and, in a separate sample, more likely to send remittances as well. Therefore, if Hispanics interviewed in both August and March could be identified in the data, they would probably show a higher remittance-welfare overlap than the non-Hispanics analyzed above.

Conclusion

Although available datasets do not allow for a precise estimate, they do show that a substantial overlap between households sending remittances and households receiving welfare almost certainly exists. The overlap indicates that U.S. taxpayer money has been converted into a form of foreign aid, contrary to its intended purpose of easing poverty in our own country. A remittance tax can help recover some of those lost funds.

Methodological Notes

A “welfare” program provides a means-tested, anti-poverty benefit. In this analysis, the cash category of welfare includes TANF, SSI, and state general assistance. (EITC is not counted as cash welfare here, but, as the Center’s report cited above shows, including it would raise the overall welfare use rate of immigrant-headed households from 50.4 percent to 53.5 percent.) Food includes SNAP, WIC, and free school lunch. Housing assistance includes both the subsidized and public types.

The CPS generally uses a 4-8-4 interview structure. Each household is interviewed in four straight months, followed by no interview for eight months, and then interviews again for four months. Because August is more than three months away from March, the 4-8-4 structure should not produce any common set of households interviewed in both months. However, the March supplement utilizes an expanded sample that brings back some households that would not ordinarily interview in March. The expansion includes both a Hispanic component and a non-Hispanic component. The non-Hispanic expansion is limited to households with minor children or a non-white (and non-Hispanic) member.

For unclear reasons, the Hispanic component of the expansion does not retain its household identification from previous months, meaning outside researchers can construct an August-March subset with only non-Hispanic households. The Hispanic exclusion has a theoretically ambiguous effect on measured welfare usage — Hispanics have above-average rates of welfare use, but so do the children and non-whites targeted in the non-Hispanic sample. Empirically, however, the March 2009 data shows that immigrant households in the subset (24 percent welfare use rate) are clearly less likely to be recipients than immigrant households overall (34 percent). The Center will pursue a more representative sample in future studies of the remittance-welfare overlap.

Another source of imprecision is the timing. Households occasionally add or subtract particular residents between months, which could affect remittances and welfare. Furthermore, the August 2008 CPS asks about remittances “in the last 12 months”, while the March 2009 CPS measures welfare usage during calendar year 2008. Parts of those time periods do not overlap, but they are close enough that they would seem to make little interpretive difference.