Comment: Uses of Administrative Records: A Social Security Point of View

Abstract

Having a limited amount of space to comment on this issue-packed summary is bound to tempt one to be parochial. I have given in to the temptation to the extent that my remarks focus on the proposed goal of developing a Linked Administrative Statistical Sample (LASS; Kilss et al. 1984), based on the Social Security Administration's (SSA's) Continuous Work History Sample (CWHS), and on the related goal of improving data for epidemiological studies. Perhaps at some later date, I will have a chance to comment on SSA's interest in improving industrial codes and standardizing establishment reporting units. In fact, the entire study is of great interest and we are all indebted to Jabine and Scheuren; to the Internal Revenue Service (IRS) for sponsoring the two-volume report Statistical Uses of Administrative Records: Recent Research and Present Prospects (Internal Revenue Service 1984a); and to the Committee on National Statistics for the role that it has played in both of these efforts. The materials in this Administrative Records Handbook clearly document the warning in the report that the six goals set out in it require complex cooperative steps. Many of us have already had a series of frustrating experiences with efforts to develop the LASS. Although these have been complicated by the passage of the Tax Reform Act (Public Law 94-455, 1976), there are other basic difficulties that, quite aside from the Tax Reform Act, pose problems for implementation of the goals. Perhaps a brief review of the experience with efforts to enhance the CWHS will provide a simple, yet explicit, illustration of the nature of the problems. Let me anticipate my conclusion: the report's emphasis on the need for the willingness of government agencies to cooperate if these goals are to succeed is a necessary, but not sufficient, condition. In fact, even a skimming of the materials in volumes 1 and 2 of the report shows that a number of important agencies have demonstrated a willingness to cooperate on the LASS for nearly a decade. The basic problem is that the budget system does not provide a satisfactory vehicle for long-term development of statistical projects that cut across the