Scott Winship, Yuval Levin, and William Voegeli offer insightful critiques of my Liberty Forum essay on the 20th anniversary of welfare reform. Actually, I use the term “critique” advisedly, because there may not actually be that much daylight between our positions. Still, there are quibbles, if not major differences.

Scott Winship clearly feels the 1996 reform has been a bigger success than I do. He correctly notes that using alternative measures of poverty, rather than the official Census Bureau measurement, shows some reductions in child poverty in the mid-1990s, some of which is attributable to the 1996 Act and the reorientation of welfare toward work.

Still it is worth noting that, whether by the Census Bureau measure or by the alternative one, working age poverty has not seen significant reductions, nor has there been much reduction in the child poverty rate in the 21st century. This, despite welfare spending’s going up by hundreds of billions of dollars over that same period, to the point that federal and state governments spend roughly a trillion dollars each year on assistance to low-income people in one form or another.

Clearly, welfare reform did not reduce spending. But the more relevant question might be whether incremental increases in spending produced commensurate incremental gains in poverty-reduction.

One also can look at changes in the recipiency and dependency rates highlighted in a recent report from the Health and Human Services Department’s Office of the Assistant Secretary for Planning and Evaluation. The rate of families receiving any benefits from TANF, SSI, or SNAP throughout a year was 23.6 percent in 2012, an all-time high, and significantly higher than the 16 percent it was in 1996. Part of this is due to the expansion of SNAP, and families receive lower benefit amounts even when working. The dependency rate, which measures the proportion of families receiving more than 50 percent of annual income from one of these three programs, is pretty much the same in 2012 (5.1 percent) as it was in 1996 (5.2 percent).

I do want to highlight one area where I think Winship gets it exactly right, though. He correctly calls for “reforming disability programs—SSI and the Social Security Disability Insurance (SSDI) program—to promote work and independence among beneficiaries with relatively minor physical or mental ailments should be another goal of antipoverty policy.” Yes, and we should go further, by making it easier for people to work to the extent they can (and in most cases, want to, and would, if not for existing policy barriers). I agree with Scott that the 1996 reform and implementation of the EITC were certainly a step in the right direction from this perspective, but we can do much more in this regard. Making work both easier and more attractive will go a long way toward increasing the economic mobility aspects of the welfare system, aspects that I think he and I agree are relatively weak today.

Yuval Levin makes a strong case for subsidiarity, arguing in favor of letting “state social service agencies to local civic groups to churches and nonprofits—try different ways of meeting this daunting challenge in different circumstances” because this is “the logical response to the sheer complexity of the problem.” I agree wholeheartedly. As Levin notes, not only do federal bureaucrats lack knowledge of how to intervene in the lives of the poor, “such a system is essentially impossible in principle at the federal level because such knowledge is impossible to achieve and sustain in a centralized way in a society as vast as ours.”

I strongly agree. In my book The Poverty of Welfare (2003), I make the case for civil society as the most effective way to reduce poverty. And, if we are going to have public welfare programs, they will almost certainly be more effective at the state and local level. However, I have concerns about Levin’s call for allowing churches and nonprofits to utilize “public resources” for their programs. This sounds suspiciously like President Bush’s “faith-based initiative.” Public resources inevitably come with strings attached that undermine both the effectiveness and the essential character of those institutions. When, for example, Catholic Charities receives the majority of its revenue from the government, it ceases to be either Catholic or a charity.

I also agree with William Voegeli that Americans are deeply conflicted about the welfare state. But I think people’s confusion applies more to programs that benefit them, or those like them, than to programs that benefit the poor (especially if the poor are perceived as non-white). Social Security and Medicare are hugely popular. But welfare as applied to the poor seems considerably less so. A 2015 Gallup poll found that only 14 percent of Republicans and Republican leaners were satisfied with the “work the federal government is doing on poverty,” compared to 18 percent among Democrats and Democratic leaners. While the source of dissatisfaction might differ, there is widespread disillusion with the current system. Other polls corroborate that the public places a much higher priority on preserving programs like Social Security and Medicare than on funding the welfare system targeted toward Americans living in poverty.

I also agree with him that our goal should not be solely to make the welfare state more efficient, but to reduce its overall size, cost, and intrusiveness. One way to accomplish this, I still maintain, is to enable poor people to move out of poverty, thereby making it less necessary for them to turn to the government for support. Most of what is required to accomplish this—from school choice to job creation to criminal justice reform—will take place outside the welfare system. But to the degree that we can reform welfare to make it more likely to help people ascend the economic ladder, we can hasten the day when they leave these programs. This could actually be the most effective way to reduce the scope and scale of the welfare system.

We also need to recognize that in many cases the poor are poor because of government action, from failed government schools to housing policies to our past history of racism and sexism. Obviously our first priority should be ending government actions that are making the problem worse. But there also would seem to be an obligation to make restitution for past wrongs committed by the government.

Finally, where Winship, Levin, and Voegeli seem to agree with each other, to a greater or lesser degree, but differ with me is on the question of providing cash benefits and then expecting recipients to budget and make decisions for themselves.

Admittedly, there have been few studies of how American welfare recipients would respond to receiving cash instead of traditional benefits. However, the ones we have seen have shown that even when recipients are given totally unrestricted cash, they show no more inclination to spend their money on vices than the rest of us. For example, state requirements for drug-testing welfare recipients have failed to uncover evidence of drug use higher than overall use in society. Likewise, a recent working paper for the World Bank reviewed 30 studies examining the relationship between cash aid and “temptation goods” such as alcohol and tobacco, with quantitative data and survey responses. Their comprehensive review found that “almost without exception, studies find either no significant impact or a significant negative impact of [cash] transfers on temptation goods.”

Voegeli notes the contradiction between my arguments in favor of letting poor people manage their own lives and my acknowledgment that there is a behavioral or cultural component to poverty. That is, people are more likely to be poor if they drop out of school, don’t have a job, or become pregnant outside of marriage. But while acknowledging the role that poor choices play in pushing people into poverty and trapping them there, we should also understand that choices take place in a wider context. People relegated to neighborhoods with concentrated poverty, poor schools, few role models, fewer jobs, police misconduct, and a history of racial discrimination will find it much harder to do the things necessary to avoid poverty.

Staying in school, getting a job, and avoiding pregnancy outside of marriage might seem basic to someone who grew up in a two parent household in a respectable suburb, but the framework for many poor communities is much different. If we want poorer communities and poor people to behave differently, it is hard to see how denying them agency or responsibility for their lives will get them there.

In the same regard, Levin worries that “federal cash payments to individuals tend to isolate individuals from their communities and from responsibilities to others. They encourage an understanding of society as consisting of isolated individuals and a distant national government.”   Here I disagree. Not that poor communities are not socially isolated, reinforcing the culture of poverty—Raj Chetty’s work, among others, makes it clear that they are—but that this dislocation would be made worse by cash payments if these replaced in-kind benefits.

Because only certain providers are both qualified and willing to accept payment through the many in-kind social welfare programs, the poor are currently forced to live in areas where poverty is concentrated. Cash payments, on the other hand, could reduce the tendency to segregate the poor geographically, although obviously they would not eliminate it. (For example, the poor would still naturally gravitate to areas with lower rents.) But to the degree that we are better able to integrate the poor into the larger community, we would reinforce behaviors that could help get people out of poverty. Similarly, if the poor were able to purchase goods and services with cash from a broader range of providers, they would be exposed to more opportunities to change their situation.

In the end, as we look back over 20 years of welfare reform, I think we can all agree that we can and should do better.