Gary Huang on Improper Payment Studies

Hi, I’m Gary Huang, a Technical Director and Fellow at ICF Macro, Inc. in Calverton, Maryland. My colleagues, Sophia Zanakos, Erika Gordon, Gary McQuown, Rich Mantovani, and I are presenting at AEA’s upcoming conference on improper payment (IP) studies. We conduct research and evaluation relating to benefit eligibility and payment errors under rubric of IP. This kind of research, required by law (IPERA 2010, formerly IPIA 2002), is becoming increasingly important for improving government accountability and financial integrity.

Lessons Learned: To define benefit eligibility error and to make decisions on data sources and methods to use to generate IP estimates, we must prioritize stakeholders’ different interests. This includes meeting the technical and statistical rigor required by the Office of Management and Budget (OMB), understanding the intricacies in program concerns by federal agencies, dealing with the reluctance to cooperate among local agencies, and facing the logistic challenges for surveying program participants. Two types of data sources are used in IP studies: program administrative records and survey data.

Hot Tip: A comprehensive IP study of the assisted-housing programs at HUD involves a stratified sample survey and administrative data collection to generate nationally representative estimates of 1) the extent of erroneous rental determinations, 2) the extent of billing error associated with the owner-administered program, and 3) the extent of error associated with tenant underreporting of income. The extensive data collection effort requires coordination and data quality control to ensure data accuracy in tenant file abstraction, in-person CAPI interviewing, third party information, and data matching with Social Security and National Directory of New Hires databases.

Hot Tip: Some agencies conduct national representative surveys of individuals served and entities paid for providing services. In some cases, these surveys bear close similarities to audits and are overt or covert with the data collector posing as a customer. The Food and Nutrition Service (FNS) is increasingly emphasizing the use of administrative data to update estimates obtained from surveys. However, the administrative data are usually biased, and therefore must be modified. Statistical modeling for updating improper payment estimates seems a possible and efficient alternative in IP studies.

Hot Tip: For the Center for Medicare Medicaid Services (CMS) to identity probable fraudulent claims and the resulting improper payments to health care providers, computer programs were developed to examine four years of Medicaid administrative claims data for all US states and territories, applying a variety of algorithms and statistical processes. Both individual health care providers and related institutions were reviewed. For such large administrative data analyses, evaluators struggle to understand various issues from technical, managerial and political perspectives.

Rad Resources: Check OMB’s implementing guidance to all federal agencies ( on IP measurement and policy and technical requirements for IP studies.

This contribution is from the aea365 Tip-a-Day Alerts, by and for evaluators, from the American Evaluation Association. Please consider contributing – send a note of interest to Want to learn more from Gary? Gary and his colleagues will be presenting as part of the Evaluation 2011 Conference Program, November 2-5 in Anaheim, California.

Leave a Comment

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.