One of the most important policy discussions going on right now concerns how we maintain the accuracy of government reported statistics, especially those concerning the economic condition of our fellow Americans. Empirical and credible data are essential for policymakers and program administrators to measure properly program performance and to understand how these programs affect the lives of participants. In fact, the Commission on Evidence-Based Policymaking, which was jointly sponsored by Speaker Paul Ryan and Senator Patty Murray last year, is currently developing new strategies to strengthen evidence-building and improve how government programs gather and interpret data.
The commission’s work is especially relevant to me. Throughout my 20 years of working in the social services agencies of New York State and New York City, I was constantly aware that I had a clearer picture of what was going on in low-income households than what was being reported in the Census Bureau’s annual reports on the economic condition of Americans, including those living in my state. Using data systems common in every state, I could see who was receiving food stamps or other welfare benefits, in what neighborhoods, and in what types of families. I knew the education levels of recipients, their family structures, and employment statuses. In short, I knew how much assistance New Yorkers were receiving from various government programs.
Despite having this valuable on-the-ground data, I and other state and local officials did not share this information with federal statistical agencies in any organized, comprehensive or effective way. Without this precious information, Census Bureau reports and research based on Census Bureau datasets were becoming increasingly inaccurate and were not reflecting the true condition of America’s low-income populations. On top of that, these flawed reports — based on surveys of a sample of Americans — were having far more influence on the public debate than the reliable data I had at my fingertips. Especially troubling was the fact that the Census Bureau was indicating greater economic distress than what my colleagues and I knew was really the case.
My suspicions were confirmed by the work of Bruce Meyer, Wallace Mok, and James Sullivan. Their collective research found that the quality of household surveys, from which official rates of unemployment, poverty, and other policy guiding statistics are derived, has been in severe decline. Households have simply become less inclined to respond to surveys; and when they do, they are less likely to answer certain questions and provide accurate information, particularly when they are being asked about receiving various forms of public assistance.
By comparing survey results with more reliable administrative data from federal and state welfare agencies, Meyer and his coauthors revealed significant underreporting of receipt rates and the value of benefit received for most poverty-reducing programs. For example, surveys failed to capture almost half of the dollars given out by the Temporary Aid to Needy Families program. This underreporting impairs our perception of the poverty-reducing effects generated by public assistance programs.
The solution to this problem lies with the administrative data residing within state departments of social services. In addition to having in-depth knowledge on program participants, the datasets generated by these departments offer many advantages, such as large sample sizes, low measurement error, low attrition rates, and longitudinal measurement opportunities. However, there is no legal authority or established incentives for state agencies to share their datasets with the Census Bureau. Right now, agencies often balk at data sharing opportunities simply because it has not been done before; they perceive costs to be too great; or they do not want to expose themselves to additional oversight.
The Evidence-Based Policymaking Commission needs to take a strong stand on mandating that state social service departments share their administrative datasets with the Census Bureau. If this is applied to food stamp benefits, housing assistance, Medicaid, TANF, and child support enforcement, it would dramatically improve the ability of the Census Bureau to describe the real economic condition of Americans. Data sharing arrangements should be done in a cost effective and confidential way so as to not jeopardize the privacy of program participants. Such a measure would allow for a much-needed correction in how we understand poverty and perceive government programs in the country, ultimately contributing to more targeted and more effective policy decisions.
A key to making this happen is the recognition of the Census Bureau as an honest, reliable, and exclusive holder of this data. The data shared on these programs should not be used by the federal oversite agencies to play “gotcha” with state providers of services, as that would undermine the federal-state partnership inherent in our federalist system. The mandate to share with the Census, however, cannot be in the form of a request; it must be a requirement for accepting the federal money that is used to finance the various federal programs. I know state social services commissioners — I was one myself — and we respond well to direction. But options, which require all the bureaucratic hassle that comes with any decision, often end up going in the circular file. Moreover, the provision of this data must be nationwide — every state must participate in order to get an accurate picture of the nation as a whole. That will only happen with a clear mandate. Unlike other mandates, this new data sharing requirement would not interfere with the states’ appropriate flexibility in determining how to administer their programs. The point is simply to make available data showing how much assistance is going to how many people.
Such data sharing could have far reaching impact on federal policy by allowing us to know the actual value of the benefits we provide, and how effective these programs are in moving families toward self-sufficiency. There is no more important fact to know in the war on poverty.
Robert Doar is the Morgridge Fellow in Poverty Studies at the American Enterprise Institute. From 2003–2006 and from 2007–2014, he was respectively commissioner of social services for the state of New York and commissioner of the New York City Human Resources Administration.