It is the Federal Reserve’s mandate to promote full employment, but policy makers in Washington have not done a great job of achieving this goal. Over the the past 35 years, there has been just one stretch during the late 1990s in which the U.S. economy was operating at full employment.

The Fed’s failure in this regard isn’t just about a lack of will, however. Unlike the central bank’s inflation goal of 2%, what constitutes full employment changes over time. Take, for instance, the phenomenon of the falling labor force participation rate. Many economists have argued that this trend is due to the aging of the population, as an older workforce means more retirees. But with a smaller share of the population looking for work, the rate of full employment tends to go down as well.

Trends like these have led the Federal Reserve to consistently lower its estimate of full employment throughout the economic recovery, from 6% to as low as 5% today. In a new working paper published on Monday by the National Bureau of Economic Research, economists Regis Barnichon and Andrew Figura argue that other forces are at play too, which could drastically lower what full employment really is.

They argue that since the 1990s there has been a marked decline in the number of people who are both out of the workforce (because they haven’t actively looked for a job for more than a month), but still say they want a job. And this is happening independently of the aging of the workforce. In fact, we’re seeing the steepest declines in the unemployed who say they want a job among women who are in their prime working years.

Barnichon and Figura hypothesize that much of the decline in this group can be blamed on the 1990s welfare reform, both the creation of the earned income tax credit and restrictions on receipts of payments to mothers who don’t work. The authors argue that while these reforms encouraged more mothers to work, it also pushed others further away from employment and instead pushed them into the Social Security disability insurance program, which requires that beneficiaries be unable to work to receive benefits. The authors point out that after the Aid to Families with Dependent Children was eliminated, not all former welfare recipients found work once their aid dried up. They write:

The decline in the number of caseloads was substantially larger than the corresponding gains in employment: for instance, while employment among single mothers rose by approximately 820,000 between 1995 and 2001, welfare caseloads fell by approximately twice as much, suggesting that a number of traditional welfare recipients ended up neither employed nor on welfare.

Meanwhile, use of the Social Security disability program has nearly doubled since 2000, with one expert, Richard Burkhauser, saying, “The disability program is increasingly becoming a long-term unemployment program.” The program is in serious financial trouble; it is expected to run out of funds and be forced to cut benefits after next year.

The rising cost of childcare may be contributing to the declining labor participation among working-age mothers, a possibility that the authors of the study do not address. The authors point out that low income women account for most of the increase in the number of people out of the labor force who don’t want a job. These are likely the same people who are choosing between work and paying for childcare. And with the cost of childcare rising quickly and wages for average workers remaining stagnant, it’s no surprise that a “greater share of mothers are not working outside the home than at any time in the past two decades,” according to the Pew Research Center.

The welfare reform of the 1990s is often hailed as a success, and it did lead to rising pay and increased employment for some mothers formerly on welfare. But the data clearly show that many others have been forced into programs like Social Security disability that might not have the wherewithal to support them in the coming years.