This July, Donald Trump’s Council of Economic Advisers declared that poverty had, for all practical purposes, been eliminated. In a technical report, the economists wrote that the nation’s poverty rate was only 3 percent, rather than the Official Poverty Measure (OPM) of 12.3 percent for 2017, declared by the Census Bureau this September.

The claim is a stark about-face from Ronald Reagan’s famous declaration, in his 1988 State of the Union address, that “the federal government declared war on poverty, and poverty won.” It is also an admission that America’s anti-poverty social programs have had some good effect. Poverty has indeed been reduced, and until now, Republicans have been loath to admit it. But it has not fallen nearly enough.

Of course, the poverty rate is nowhere near as low as 3 percent. The reason for the CEA’s transparently false claim is that, if true, it might justify raising work requirements for the non-cash social programs that don’t already have them. People just don’t need the money, the CEA argues. And there are apparently many who want to believe the poverty rate is way down.

Cutting Medicaid outlays is a holy grail for the Republicans. If work requirements are added, they presume, fewer Americans will use the program. The same goes for the Supplemental Nutrition Assistance Program (SNAP), or food stamps, which already has moderate work requirements. New, more demanding requirements would cut their use. If they could, the Republicans would also attach work requirements to housing assistance.

A main component of the current American welfare program, Temporary Assistance for Needy Families (TANF), which was adopted in 1996 to replace the former welfare program, is just such work requirements. The budget for the original welfare program, Aid to Families With Dependent Children (AFDC), plus food stamps, came to only about 1 percent of GDP in the 1990s, when these programs were so vilified. Medicaid, which now accounts for some 3 percent of GDP, is the fat cat the Republicans would love to wound.

The current cynical strategy of Republicans is to argue, almost as one, that social programs (some lawmakers include Social Security and Medicare among them) must be cut substantially because the federal budget deficit is so large. The deficit is increasing rapidly, of course, but that’s in large part due to the recent massive Republican tax cuts for the well-off, a classic example of the starve-the-beast strategy propounded by Milton Friedman, Grover Norquist, and Ronald Reagan.

To further support the argument that able-bodied Americans are taking advantage of welfare programs like Medicaid and SNAP, the CEA presented several misleading charts that distort the straightforward facts. The council claims that millions of able-bodied citizens who can work are receiving Medicaid and SNAP (for more on Medicaid work requirements, see reports by Michelle Chen and Bryce Covert for The Nation).

Here are a few key facts: The Kaiser Family Foundation finds that 62 percent of Medicaid recipients have full- or part-time jobs. Regarding SNAP, four out of five recipients worked either the year before or after they received food stamps (the typical SNAP recipient from 2008 to 2012 received benefits for just one year). As for TANF, it is in fact mostly failing. In 1996, benefits were provided to 68 percent of poor families, compared to 23 percent today, under TANF. More to the point, the jobs that recipients had been required to take under TANF mostly turned out to be short-term. The expansion of the Earned Income Tax Credit, studies show, was more responsible for the increase in employment in the years following the introduction of TANF.

The whopper that the poverty rate is just 3 percent is just another product of the Trump dissimulation machine. Like the claim by the CEA last year that the GOP tax cuts would result in average wage increases of $4,000–$9,000 a year, it is nonsense. To accompany the wage-increase whopper, the CEA also issued a superficially sophisticated economics exercise to justify its claims. Almost all economic specialists found the methodology faulty. Of course, nothing like the promised wage increases has occurred, and virtually no economists are forecasting that they will.

Back to the current whopper: If the claim that poverty is now 3 percent rather than the reported 12.3 percent had any validity, it would mean that more than 30 million fewer Americans are poor than previously reported.

It might surprise readers to discover that the scholars who do the leading research supporting the CEA view, Bruce Meyer of the University of Chicago and James Sullivan of the University of Notre Dame, while defending the Trump CEA, concede that the 3 percent level is merely an “arbitrary” outcome. Their research has not been designed to find an actual poverty rate but to show the trend of poverty. To escape accusations of outright flimflammery, Meyer and Sullivan mention in a single sentence, without further explanation, that the threshold by which they calculate the 3 percent rate is arbitrary.

What the CEA does not tell you is that the methodology cannot determine an actual poverty rate. Meyer and Sullivan themselves admit, deeper into their research piece, that the poverty rate under their method could be as high as 12.7 percent, roughly the same as the official rate. But the authors bear responsibility for singling out the figure of 3 percent. The Wall Street Journal teaser for their article reads, “The White House tells the truth.” To bring the point home, in the annual poverty report published last year by the American Enterprise Institute, they wrote at the very top of their piece, “The consumption poverty rate fell from 3.4 percent in 2015 to 3.0 percent in 2016.” We have to look deeper into the piece to find that they don’t really mean this.

Meyer and Sullivan have long argued that poverty should be based on the consumption surveys of the federal government, which ask families how much they have spent. They say these are more trustworthy than the surveys of family income, which are used to compute the official poverty measure. If the consumption data were reliable, this would be true. Many economists agree that the income surveys are often inaccurate (though methods have been developed to adjust for this). But consumption data are also faulty. In particular, a family’s consumption can be supported by digging into savings or by borrowing in any given year, and the extent to which this occurs is not well analyzed. Consumption doesn’t measure one’s poverty in any given year.

To be sure, the OPM is by no means a model of statistical accuracy. In fact, I’d call it not merely archaic but also irresponsible. It is based on a survey of budgets for the poor from the mid-1950s, and it has only been adjusted for inflation. It does not include newer non-cash benefits like the Earned Income Tax Credit and food stamps. It has been adjusted for inflation over the years, but it has not been adjusted in other ways that would reflect changing patterns of consumption.

Using consumption rather than income surveys to measure poverty would include the non-cash sources of income, which on their own lower the poverty rate. But Meyer and Sullivan also use a “slower” price index, rather than the traditional consumer price index, which means the poverty line doesn’t rise as much in their calculation.

A relatively new, “experimental” measure of poverty, based on much research at the National Academy of Sciences and called the Supplemental Poverty Measure (SPM), does include non-cash payments, similar to the consumption poverty line. But, importantly, the SPM discards the old poverty line from the 1950s and uses a “relative” poverty threshold based on the most recent five-year consumer expenditures on key items.

Many have long recognized that a poverty threshold should rise over long periods of time to take into account a changing, more prosperous economy as well as new needs, such as computers and the Internet and a college degree, to lead a life of even minimal opportunity. Rose Friedman, Milton’s equally conservative economist spouse, argued that the poverty line should be raised every 10 years or so. Meyer and Sullivan now ignore this, thus leaving their poverty line low over time. Yet Meyer in 2008 proposed just such an idea in government testimony. As he said, “Periodically we may want to raise the cutoff for an absolute poverty measure…every decade or two.”

The SPM also deducts medical care and work-related and childcare expenses from people’s income. Meyer and Sullivan ignore such adjustments as well.

But this is still not the essence of how they derive the 3 percent poverty line. To make their calculation, the scholars set the poverty line for consumption at the 1980 level for the OPM (the consumption poverty rate and the OPM were thus both 13.0 percent at that time, according to Meyer and Sullivan). They then adjusted their poverty line over time by their preferred slower inflation rate. The result was a consumption poverty line of only $17,700 for a family of four in 2017, according to analyst Shawn Fremstad of the Center for Economic and Policy Research, compared to the OPM line of roughly $25,000 and an even higher threshold under the SPM. Given that their consumption figures include the non-cash benefits excluded from the OPM, this resulted in only 3 percent of the population below the low poverty line.

Here’s the kicker, however: As they actually note later in their piece, if they had set the poverty line at the OPM level for 2015, the same methodology would have produced a consumption poverty rate of roughly 12.7 percent. If they had gone back before 1980s, the poverty line would have been even lower than 3 percent (the consumption data did not exist then).

So what is the poverty rate, 3.0 percent or 12.7 percent? They can’t tell you. It could be higher or lower, or anywhere in between. This is what they mean by “arbitrary.” But if they can’t tell, as they admit to us, then why do they and their sponsors, the American Enterprise Institute, trumpet that 3 percent number? The Trump CEA trumpets it even louder.

Meyer and Sullivan are on more solid methodological ground when they cite the trend of poverty, rather than a specific rate. However, even this doesn’t bear up under serious scrutiny. Since 1980, they report, the consumption poverty line fell by more than nine percentage points, while the official poverty rate rose by half a percentage point. As The Wall Street Journal put it in the web headline for their piece, “Hardly Anyone Wants to Admit America Is Beating Poverty.”

But liberal economists have been making this argument for a long time. It is conservatives who have resisted, because to admit that there’s been a decline would serve as proof that the social policies adopted since the 1960s have indeed cut poverty, not increased it, as George Gilder, Charles Murray, and Reagan once vehemently argued.

The leading voices pointing out that poverty has been sharply reduced are scholars associated with the Columbia School of Social Work, which estimated the SPM back to the 1960s. In a widely read paper published in 2013, the Columbia group found that poverty rates measured by the SPM have fallen by roughly 40 percent since the 1960s. Still, to repeat, the latest SPM poverty rate, after the deduction, is 13.9 percent, still far too high.

Other economists on the center and broad left have been making the same argument. In 2015, for example, Christopher Jencks of Harvard presented an analysis in The New York Review of Books showing how far the official poverty rate had fallen when adding non-cash government benefits to the income of the poor and using a more slowly rising inflation measure.

But the best way to check the plausibility of Meyer and Sullivan’s falling consumption poverty rate is that it simply doesn’t make any sense. The Meyer and Sullivan consumption poverty rate fell substantially from its level in 2000, when the unemployment rate was only around 4 percent, to 2010, in the depths of the Great Recession, when the unemployment rate had risen sharply, to roughly 10 percent. As scholars Luke Shaefer and Joshua Rivera note, it is implausible that the consumption poverty line was considerably lower at the bottom of the steep recession than it was at the height of the Clinton boom.

Back in the real world, scholar Kathryn Edin, also with Luke Shaefer, published a book in 2016, $2.00 a Day: Living on Almost Nothing in America, which shows that 1.5 million families in America live on $2 of cash income a day per family member. They exclude the imputed income from SNAP because food stamps are not cash—you can’t use them to buy a warm coat for your child or to pay the electricity bill, they point out. Still, even if you include SNAP, 750,000 families live on $2 a day per family member, by their measure. That is roughly the same as the World Bank’s poverty line for poor nations.

Meyer and Sullivan, along with others in their camp and working from their consumption data, insist that the lower end of the spectrum is not doing that poorly. In a 2016 review of Edin and Shaefer’s book in The New York Review, Jencks reported on a rough calculation of income across the spectrum over the past five decades; since 1999, he finds, the income of Americans in the bottom 10 percent actually declined.

Meyer and Sullivan also claim that judging by surveys of material living standards, the poor are doing better than we think. So does the Trump CEA. Meyer and Sullivan note that the federal American Housing Survey reports that the bottom 20 percent of Americans live in conditions equal to the middle class a generation ago, in terms of the “square footage of their homes, the number of rooms per person, and the presence of air conditioning, dishwashers, and other amenities.”

Maybe, but of course changing technology has made many of these amenities much cheaper, relative to incomes. Moreover, must the poor live according to the same standards we used in the 1980s to be considered impoverished? And we haven’t even discussed educational quality in poor neighborhoods or rising homelessness. Homelessness, for example, tripled in the 1980s, under Reagan, and has remained high, though falling somewhat since 2007, according to recent data.

Shaefer and Rivera, looking at other measures of material deprivation, find that food insecurity, a Department of Agriculture measure of the threat of hunger, rose substantially for the poor throughout the 2000s—the same period when Meyer and Sullivan’s consumption poverty rate fell sharply. And material deprivations such as difficulty paying off mortgages and electricity and heating bills, and inability to pay for visits to the doctor, also rose in the 2000s.

Contrary to the assertions of the Trump CEA, poverty is seriously understated in America. Poverty measures can be complex, but here’s one clear way to look at it: When the OPM was first adopted in the 1960s, the official poverty line represented roughly 50 percent of typical, or median, family income in America. Today, it represents about 30 percent. If it were at about the level when first set compared to median income, we would be reporting roughly 10 million more poor in America than the 45 million we report today, perhaps more.