Lessons from NYC’s Randomized Conditional Cash Transfer Program

The research organization MDRC recently released its comprehensive evaluation of Opportunity NYC- Family Rewards, a conditional cash transfer (CCT) pilot program with the goal of helping families break free of the cycle of poverty. This program is particularly notable because it is the first comprehensive CCT program in a developed country, and it was a large-scale randomized control trial. CCTs offer cash assistance, but only if certain conditions are met, in this case these conditions are concentrated in the three spheres of children’s education, preventive health care utilization and parents’ employment. There was no case management component to the program, the cash-based incentives were the only mechanism in place. The initiative had the twin goals of reducing current poverty (material hardship) and incentivizing these low-income families to invest in developing their human capital, which is important for their ability to attain a level of self-sufficient prosperity. After the conclusion of the three-year pilot program, while the program’s cash transfers produced some results in the first goal of reducing material hardship, as you might expect from a significant cash transfer to families with otherwise limited incomes, it failed to have much of an impact in its second goal related to the development of human capital. Family Rewards failed to produce any meaningful effects. There are caveats to what these findings mean in a broader sense, but they convey some of the limitations of transfers, and of antipoverty policies in general, in addressing the more complex and difficult aspects of poverty.

Six community-based organizations ran the Family Rewards pilot in six of the city’s communities with the highest levels of poverty, with the program running for three years concluding in August 2010. MDRC split roughly 4,800 families with 11,000 children into treatment and control groups, and analyzed the effects of the CCT program on a range of different metrics by comparing the groups at two and six years after the program began. Family Rewards included 22 different rewards tide to specific activities like taking the PSATs, attending parent-teacher conferences, and sustaining full-time work.

Throughout the three years of the program, participating families received an average of over $8,700 with a majority of those families receiving at least $7,000 and the top quintile receiving more than $13,000 in cash transfers. These substantial cash transfers reduced the share of families in poverty by 12 percentage points (from a baseline of 68 percent in poverty). There were associated reductions in measures of material hardship like the proportion of families dealing with food insufficiency or inability to pay rent, relative to the control group. These gains were concentrated among families living in severe poverty, while the reductions in hardships were “small and statistically insignificant among those whose poverty was not as severe at the time they began the program.”

The program fared much worse in its second goal of enticing people to invest in their human capital and develop the ability to support themselves long-term. With the exception of increases in high school achievement and graduation rates for students who were already more prepared, Family Rewards failed to have a substantial impact on the longer-term goals related to avoiding future poverty.  The program had no substantia impact on school outcomes for students in elementary and middle school, and no positive effects for less proficient high school students.  Despite cash-incentives to maintain full-time work, parents did not significantly increase their earnings in jobs covered by unemployment insurance, although it is possible that parents increased informal earnings that were not covered. In fact, Family Rewards reduced work effort for parents with limited educational attainment: parents in the program lacking a high school diploma had an average quarterly employment rate three percentage points lower than the control group, with average earnings roughly $2,900 lower. For the parents with the most limited job prospects, the program had the opposite of its intended effect, as the cash transfer amounting to a substantial portion of what they would be able to earn led them to reduce work effort. The minimal results for both parents and their children highlight some of the limitations of this strategy for trying to help them find a sustainable path out of poverty.

The results of this large randomized control CCT program show the challenges and limitations of government efforts to address the complicated problem of poverty.  While there were some reductions in material hardship due to the program’s substantial cash transfers, after the program ended the there were not significant differences between the program group and control group in terms of average monthly income and annual poverty rates. This is perhaps the starkest sign that the program largely fell short in its second goal of incentivizing people develop the capacity self-sufficient prosperity in the longer term, as the effects faded out when the transfers did. Instead of the intended goal of encouraging parents to maintain full-time work, Family Rewards actually had an adverse impact on the work effort of some of the most disadvantaged parents, which could have led them to become even more disconnected and isolated, which potentially comes with a whole range of negative spillover effects. The groups behind Family Rewards incorporated these findings and updated the design of the program in a second iteration that ran from 2011 to 2014, with results and more lessons learned from that pilot available this year. The current welfare system is a tangled mess and in dire need of reform. Pilot programs and evaluations like this one are valuable because they help us see what works, or in this case, what doesn’t.