States evaluating whether to use their own data or SIPP data should consider both the level of detail available in their state data and the effort required to organize and use their state data, as compared to the SIPP. Each data source has its own advantages and disadvantages, both because the availability of specific data elements within state data stores varies and because state policies (and therefore, the relative importance of certain data elements) vary.
As described in the ASPE companion brief, Data Sources for Modified Adjusted Gross Income (MAGI) Conversions, the Department of Health and Human Services (HHS) has developed a national model to simulate Medicaid eligibility for use in the recommended Standardized MAGI Conversion Methodology using SIPP data. A key feature of SIPP data is that it contains information on enrolled Medicaid beneficiaries as well as individuals who are likely eligible and therefore is better for estimating the effects of a converted threshold on the net number of people eligible. This model, however, does not capture state-specific, eligibility group specific household composition, income counting or disregard rules, given limitations in the data the SIPP collects and the feasibility of programming for all states. The SIPP model uses one household composition configuration for all states and all eligibility groups, which includes parents, children and siblings. State data could capture these data points using multiple household composition configurations varying for family groups, such as section 1931, and for other AFDC-related eligibility groups [e.g., employing “prohibited deeming” financial responsibility rules under Section 1902(a)(17) of the Social Security Act (the Act)]. Similarly, state data would capture the rules employed by states that test multiple configurations of household compositions to give an applicant the greatest possible benefit before finding the individual ineligible.
The SIPP model has selected most frequently used disregards that are captured in the SIPP, which CMS believes will be sufficient for states that use typical disregards. State data has the potential to capture actual disregards used under the state rules, including less frequently used ones and ones not be captured in the SIPP. Using state data gives states the opportunity to use a full year of data, if they choose, which could be important in states that experience seasonal variations in eligibility and/or disregards; whereas the SIPP model uses one month of data (the April 2010 cross section of the 2008 SIPP panel). State data will by definition capture the demographics of the enrolled population in the state. The SIPP model is approximating the demographics of each state using a re-weighting strategy whereby certain characteristics relevant for the income conversion process are given more or less weight for each state. Finally, for eligibility groups that have an asset test, the state data should capture this information; whereas the SIPP model will not be using an asset test when selecting cases to use for the calculations.
However, in performing tests with sample state data in our research, we found a number of challenges that states may also encounter. The Standardized MAGI Conversion Methodology requires individual-level data that includes information on eligibility category, income, and total disregard amount. There may be a substantial effort involved in extracting, transforming and loading the data. Some data may not be collected consistently; even if it is collected, it may not be maintained in history. Furthermore, some data elements present in SIPP may not be present in state data sources, such as stepparent income and parent income for young adults living at home.
"rb.pdf" (pdf, 1.32Mb)