The Trump administration appears to be considering whether lawful immigrants should be deported after receiving a certain amount of public assistance, according to a draft of an executive order obtained by The Washington Post.

Such a move would represent a departure from current practice but would be consistent with the goals of Trump advisers Stephen K. Bannon and Stephen Miller, who, in the words of the Los Angeles Times, “see themselves as launching a radical experiment to fundamentally transform how the U.S. decides who is allowed into the country.”

The draft order appears to be premised on the notion that immigrants place an untenable burden on the economy through their use of public assistance. It asserts that immigrant households are “much more likely than those headed by citizens to use Federal means-tested public benefits,” which include Medicaid, SNAP, housing assistance, supplemental security income, TANF, and general assistance. However, it offers no evidence for that claim.

On the contrary, an extensive 2011 Harvard Business School review of the economic literature on the effects of immigration found that “on average, immigrants appear to have a minor positive net fiscal effect for host countries.” Economists generally agree that the typical American citizen would be better off if more foreign workers were let into the country each year, regardless of skill level.

Moreover, the assertion that immigrant households are more likely to use federal benefits appears inaccurate. A 2013 analysis by the Cato Institute, a libertarian think tank, found that poor, noncitizen immigrants actually use public assistance at a lower rate than low-income native-born citizens across a wide array of federal programs.

Strict eligibility guidelines that prevent many immigrants from receiving federal aid are one of the primary drivers of these differences, according to the Cato Institute report. Undocumented immigrants and workers or students on temporary visas are generally not eligible at all. Lawful permanent residents are eligible only after five years.

Moreover, “the average value of benefits per recipient is almost always lower than for the native-born,” the report finds. “For Medicaid, if there are 100 native-born adults, the annual cost of benefits would be about $98,400, while for the same number of noncitizen adults the annual cost would be approximately $57,200.”

Among children of noncitizens, the discrepancy is even larger: “For children, a comparable calculation for 100 non-citizens yields $22,700 in costs, while 100 citizen children of citizen parents cost $67,000 in benefits.”

Not all researchers agree on these numbers. The Center for Immigration Studies, a group that favors more restrictions on immigration, finds that overall, immigrant households are more likely to rely on public assistance than nonimmigrant households.

But as the Cato authors point out, this is simply because immigrant households are more likely to be low-income households. When you adjust for income, as in the Cato study, immigrant use of public assistance is actually lower.

Defenders of the policy might argue that income-adjustment is irrelevant here, since the overall burden derives from the sheer number of low-income immigrants, rather than their rates of program use. But the same standard could be applied to any demographic group with higher-than-average (14.8 percent) poverty rates: children (21.1 percent), the black population (26.2), non-high school graduates (28.9), people living in rural areas (16.5) or people living in the South (16.5), for instance.

In reality, Trump's draft order, were it to come into effect, would reduce public assistance use for people who are less likely to rely on it than the typical American citizen, once you adjust for income.