Welfare Reform Reduced Poverty and No One Can Contest It
This post originally appeared on Forbes
[Update, 1/15: Added a comment near the end about the importance of SNAP during the recovery, based on info provided me about under-reporting of SNAP benefits in surveys. I thank Tim Smeeding for the pointer.]
There’s a big give-away that liberals are making claims about the purported dangers of welfare reform they can’t support.
Their primary claims are twofold. The left asserts that welfare reform in 1996 was harmful—or would have been without expansions elsewhere in the safety net. Therefore, it follows that future reforms of food stamps (the “Supplemental Nutrition Assistance Program”) and other safety-net programs along these lines would be even more harmful, because those were the programs that expanded to mitigate the effects of the harmful welfare reforms.
Here’s the give-away: none of these liberals actually tell you what happened to poverty among children or single-parent families after welfare reform. And they don’t do it because the fact of the matter is that poverty among these groups declined. Even without taking the safety net into account, poverty is lower than it was at its pre-welfare-reform peak in 1993. Any other evidence about what happened during the recession or about trends in this or that part of safety-net spending is simply noise.
It’s worth repeating facts I highlighted in a column last week and challenging welfare reform’s detractors to counter them.
Poverty among children was the same in 1993 as in 1967
Some on the right occasionally claim that poverty is no lower today than 50 years ago. That’s not true—it’s just that the official poverty measure doesn’t count income from the biggest safety-net programs. It is certainly possible to argue about what the poverty trend would have looked like in the absence of these programs. It may well be that more people would have worked and avoided poverty that way instead. But it does no good to cite bad statistics to make that case.
Here’s the thing though: it really is true that child poverty was basically the same in 1993 as in 1967. The most comprehensive poverty measure available today, produced by Columbia University researchers, includes a variety of cash and noncash safety-net benefits and accounts for income and payroll taxes. Figure 3 of their paper shows the child poverty trend using the official Census Bureau measure and using their better “anchored supplemental poverty measure.” The improved measure indicates 29% of children were poor in 1993. That was slightly below the 1983 peak, but it was well above any year in the late 1960s or anytime during the 1970s. If you cover the post-1993 period with your hand, it’s clear you’d never have predicted in 1993 that child poverty was due to fall. Try it, I’ll wait.
The reason that poverty fell over time but the child poverty rate did not is well-understood, though welfare reform’s detractors often fail to note it. We managed to reduce poverty among the elderly dramatically. Again, it is entirely possible that poverty among the elderly would have fallen quite a bit without the safety net—people would have privately saved more, for instance—but the point here is that what happened to the elderly has no bearing on debates over welfare reform.
Receipt of cash welfare benefits plummeted for the first time ever, starting in 1995
Cash welfare used to mean the Aid to Families with Dependent Children program (AFDC). After 1996 it was replaced with Temporary Assistance for Needy Families (TANF). As a fraction of families headed by a single mother, 1972 was the peak year for welfare receipt. That year 71% of single mothers received AFDC benefits (Figure 1).* Over the next 17 years, there was a slow but steady decline in welfare receipt, to the point where just 51% of single mothers were on welfare in 1989. The decline reversed between 1989 and 1992 as the economy turned down, jumping to 58%. But though the unemployment rate dropped by nearly 1.5 points from 1992 to 1994, welfare receipt persisted at the same level.
Then in 1995, welfare receipt began what would become an unprecedented decline. In four years it fell 24 percentage points to a new historic low (going back at least to 1962). TANF receipt continued to fall thereafter, reaching 17 to 18% in 2008.** The 1996 legislation did indeed end welfare as we knew it.
It defies reason to believe that this drop would have happened without welfare reform. True, the expansion of the Earned Income Tax Credit in 1993 and the fall in unemployment might have accounted for the decline in welfare receipt to 50% in 1996. But that would have put it at a level not seen since 1969. The subsequent drop has no precedent, and if you cover with your hand the post-1996 years in Figure 1 of this study of mine, you will see that no one would never have predicted what happened.
And even the 1994-to-1996 decline may have reflected the influence of welfare reform. By then, a majority of states were operating their welfare programs under federal waivers designed to encourage work and independence. And all signs at the federal level pointed to welfare becoming a much worse deal. President Clinton highlighted his reform plans in the State of the Union in early 1994. He unveiled his welfare reform proposal in June, which strengthened work requirements substantially and introduced time limits. What would eventually become the final welfare reform bill was introduced in the Gingrich Congress at the start of 1995. This initial bill—later vetoed by President Clinton—passed in September. The writing was on the wall well before 1996.
Work among single mothers, after rising slowly for 15 years, jumped in 1997 and 1998, then stayed elevated
From 1980 to 1996, for instance, employment among never-married mothers rose by less than ten percentage points (See Figure 2 from this congressional testimony by the great Ron Haskins). From 1996 to 1999, it jumped by around 15 points. This was also an unprecedented change, and welfare reform is clearly behind it.
Employment among single mothers rose a bit more through 2000 and then fell modestly thereafter. In 2013, however, the employment rate among never-married mothers was over ten points higher than it had been in 1996.
Child poverty, and poverty among households headed by a single mother, fell a lot between 1993 and 2000
The Columbia estimates show child poverty dropping sharply from 1993 to 1995—before welfare reform, before welfare receipt started to decline, and before employment rates among single mothers jumped. This likely reflects the improved economy, combined with the expansion of the EITC in 1993. That is, the drop would not have been unusually large if not for the EITC. This is apparent by comparing poverty trends among single mothers before and after accounting for taxes (Figure 1, from Haskins, citing the Congressional Research Service). Poverty fell before taking taxes into account, but not as much. Similarly, the Columbia figures show a smaller drop using the official poverty measure—which does not account for taxes—than using their improved measure.
After 1996 or 1997, however, poverty among children and among single-mother households continued to drop, reaching historic lows and falling until 2000.
Child poverty stayed relatively low after 2000
This is the part where welfare-reform detractors have to start obfuscating and changing the subject. To believe that the decline in poverty through 2000 simply reflects the 1990s economic boom, one has to argue that the extraordinary fall in the welfare rolls after 1994 would have happened anyway. There is nothing in the historical record to suggest that. And while the EITC surely increased incentives to work, no scholar believes its effects are strong enough to produce anything like the jump in work between 1996 and 1998.
But more to the point, if the poverty declines through 2000 were all about the economy, then we would have expected to see poverty rise back up to its previous levels once the economy went into recession. Especially, since rates of poverty and welfare receipt were at historic lows and since employment rates among single mothers were at historic highs.
To be clear: that did not happen. Child poverty bottomed out in 2009, and in 2012–the most recent year in the Columbia study, it was at the same level as in 2000. That is, child poverty in 2012 was lower than at any point before 2000 and much lower than at any point before 1996. The same is true of households headed by a single mother (see the bottom two lines of Haskins’s Figure 1).
Poverty is down relative to pre-welfare-reform levels even before taking into account the safety net programs that expanded after 1996, and even before taking into account the programs that conservatives propose to block grant
This is the coup de grace. Faced with the evidence that poverty has not gone up, welfare-reform-detractors attempt to argue that the only reason that happened was because other programs–typically SNAP is invoked–stepped up. But the evidence doesn’t support this claim.
The worst that can be said is that according to the official poverty measure, child poverty was about the same in 2012 as its 1993 peak, slightly higher than in 1996. But note that (1) the unemployment rate was 8.1% in 2012 but just 6.9% in 1993, so we would expect poverty to be higher than back then, (2) the Columbia estimates adjust the poverty line upwards too much every year by using the “CPI-U-RS” rather than the “PCE deflator” to account for the rise in the cost of living, so the 1993 poverty rate was actually higher properly taking inflation into account, (3) under-reporting of transfer income has increased over time, (4) the official poverty measure excludes income not only from SNAP but from the EITC, which conservative reformers are mostly proposing to expand, and (5) those reformers are not talking about doing away with SNAP and other programs, so it is a little hysterical to cite this trend as indicating what would happen with further block-granting.
But that’s the worst that can be said, and it’s clearly not right to think things have been this bad. The Congressional Research Service estimates Haskins cites for households headed by a single mother show that before taking any safety-net benefits into account, poverty was lower in 2013 than before welfare reform (48% compared with 50 to 56% before 1997, see Table C-11). That is also reflected in the official poverty measure for this group.
SNAP had a relatively small role in reducing poverty during the Great Recession
Now I’m just trying to provoke people. But it’s true! Look at the Congressional Research Service estimates again (Figure 1). Compare the second line from the top to the third line from the top. In 2007, the markers for the two lines almost touch each other. After 2008 they are wider apart, but adding SNAP benefits reduces the growth in poverty among households headed by a single mother by 2 percentage points between 2007 and 2012 (Table C-11). That translates into an important reduction in poverty, to be sure, but it pales in comparison with the importance of the combined effect of unemployment insurance, the Supplemental Security Income program (SSI), and TANF. It’s less important than refundable tax credits.*** None of the conservative welfare reform proposals has suggested changes to unemployment insurance or SSI. All have proposed expanding the EITC.
[UPDATE: Tim Smeeding shared with me a paper of his finding that the poverty-reducing impact of SNAP in any year is greater than these estimates suggest--perhaps twice as great--because of under-reporting of SNAP receipt in survey data. Because under-reporting has risen over time, the impact of SNAP on mitigating the growth in poverty during the recession is also understated by the CRS report. It is not clear that that changes the conclusion that refundable tax credits were more important than SNAP in mitigating the rise in poverty; the under-reporting of SNAP means that CRS poverty estimates that include the tax credits also are too high (because they too include SNAP amounts that are too low).]
These facts are uncomfortable to people like Jared Bernstein, who want to ignore actually existing poverty trends and to instead talk about how SNAP’s counter-cyclical rise in spending–in contrast to the supposed failure of TANF to respond–saved the day during the Great Recession.
Jared’s column–including the chart he uses to contrast the responsiveness of SNAP and TANF–deserve another post. Until then, I’ll wait to hear from the pro-science left about where I’ve gone wrong here. Prediction: the subject will be changed to “extreme poverty” trends. That’s a third post…
* Technically, since a small number of welfare recipients are single fathers or two-parent families, it is not quite right to say that the ratio of families on welfare to single mothers is the “percent of single mothers receiving welfare,” but the difference is negligible, especially in terms of trends.
** For estimates after 2002, see https://www.acf.hhs.gov/programs/ofa/programs/tanf/data-reports and https://www.census.gov/hhes/families/data/families.html, Table FM-2.
*** Table C-11 of the Congressional Research Service report indicates that the net impact of federal income and payroll taxes is to reduce poverty, sometimes by nearly as much as SNAP does. But since taxes paid increase poverty, that means that the refundable credits more than compensate and therefore had a bigger effect on the increase in poverty than SNAP.
Scott Winship is the Walter B. Wriston Fellow at the Manhattan Institute for Policy Research. You can follow him on Twitter here.
Interested in real economic insights? Want to stay ahead of the competition? Each weekday morning, e21 delivers a short email that includes e21 exclusive commentaries and the latest market news and updates from Washington. Sign up for the e21 Morning eBrief.