In its final workshop of the year, the MRDRC presented “Disability Research with the Health and Retirement Study [HRS]” on September 24, 2020. This was a Zoom event with about 50 participants. An online format removes, of course, the burdens of travel, so we were able to include researchers from across the United States and from Europe.
The HRS, which is collected at the University of Michigan, has been a cornerstone of research on retirement, pensions, and Social Security. It has, to date, played less of a role in studies related to Supplemental Security Income (SSI) and Social Security Disability Insurance (SSDI). On the one hand, the latter topics inherently pose challenges for sample size: Although the HRS has thousands of participants, only about 5% take up SSI and/or SSDI during their years in the survey. On the other hand, the HRS can offer an exceptionally rich array of covariates — on health, wealth, earnings, and family composition — as well as, starting at ages 51 to 61, panel data. It also has extremely useful links (available to researchers on a restricted-access basis) to administrative data sources with lifetime earnings, Medicare and Medicaid records, etc.
The workshop featured a panel of HRS users, who discussed the survey’s potential strengths and limitations. Panelists included Jody Schimmel Hyde (Mathematica), Philip Armour (Rand), and Lauren Nicholas (Johns Hopkins). A second panel sought a broad overview. It included John Bound (Michigan), Eric French (University College London), Kathleen Mullen (RAND), David Weir (principal investigator of the HRS), and John Phillips (National Institutes of Health/National Institute on Aging).
One of the points most frequently emphasized was the value of making the data set as user-friendly as possible. Elementary data processing steps can be very time consuming. Significant efficiency gains are possible if the steps need not be repeated for every research paper. In response, we have a new quick turnaround project now underway, headed by Susann Rohwedder (RAND), to expand the cleaned RAND HRS data files to encompass key SSDI and SSI variables. We will look for more opportunities along similar lines in the future.
Data development has been a topic of special interest to the MRDRC since its inception. We have, for example, used the expertise of our researcher team — which includes a number of HRS co-PIs — to present projects illustrating features of the HRS of special value for retirement-related policy analysis. With our mandate expanded in recent years to include disability studies, we look forward to highlighting potential roles for the HRS in SSI and SSDI research as well.