The Urban Institute's Microsimulation Model for Reinsurance: Model Construction and State-Specific Application

The Urban Institute’s Microsimulation Model for Reinsurance: Model Construction and State-Specific ApplicationA. Bowen Garrett, Lisa Clemans-Cope, Paul Masi, and Randall R. Bovbjerg May 2008State Coverage Initiatives is a national program of the Robert Wood Johnson Foundation administered by AcademyHealth.State Coverage InitiativesThe nonpartisan Urban Institute publishes studies, reports and books on timely top-ics worthy of public consideration. This report was funded by a grant from the State Coverage Initiatives program. Some information on Washington comes from comple-mentary work funded by that state’s Office of Financial Management. The views ex-pressed are those of the authors and should not be attributed to the Urban Institute, its trustees, or its funders. 2Microsimulation Model for ReinsuranceTable of ContentsFour Steps in Modeling the Effects of Reinsurance 1Step 1. Creating a Baseline Database for Each State 1Baseline Step A: Medical Expenditure Panel Survey-Household Component (MEPS-HC) 1Baseline Step B: “Stretching” the distribution of MEPS-HC health care expenditures 2Baseline Step C: Adjusting for expenditures under-reported in MEPS 2Baseline Step D: Making other adjustments to MEPS-HC spending data 2Baseline Steps E and F: Reweighting the MEPS to resemble a particular state 3Baseline Step H: Imputing premiums for coverages 3Baseline Step G: Comparing initially simulated premiums with known premium benchmarks 6Baseline Step I: Arriving at the final state-specific baselines 6Step 2. Modeling reinsurance reform options 6Step 3. Simulating changes in employer offer and individual take-up of coverage 7Step 4. Computing overall costs of reinsurance policy and tabulating the model’s results 9Endnotes 9 1This report describes the Urban Institute’s microsimulation model of reinsurance, which was built to simulate the impacts of various state-specific ways of using state-funded reinsur-ance to subsidize primary insurance premiums. The type of reinsurance modeled reimburses primary carriers at the end of a year for the insured health care spending that falls within a specified “corridor” of aggregate annual indi-vidual medical claims expense—e.g., reinsurance of 90 percent of expenses between $30,000 and $90,000 per person per year.This work was done as part of the Reinsurance Institute, a project of the Robert Wood Johnson Foundation’s State Coverage Initiatives program, administered by AcademyHealth, and the Urban Institute, that provided technical assistance to competitively selected states interested in using reinsurance as an aspect of health financing reform. A companion report describes the overall project, including the qualitative elements of our work that complemented the quantitative modeling described here.1 Results of the modeling are reported in deliverables to the participating states.2Four Steps in Modeling the Effects of ReinsurancePredicting the effects of reinsurance reform options within our three states of Rhode Island, Washington, and Wisconsin involved four main tasks:Creating a baseline database that reflects each state’s distribution of health insurance coverage, demographic characteristics, individual health expenditures, employer characteristics, and health insurance premiums paid by employer group and non-group purchasers of coverage;Modeling the reinsurance policy and estimating the changes in premiums that would result from each particular reinsurance reform option considered;Estimating the effects of those premium changes on employers’ health insurance offer behavior and on individuals’ take-up of group and non-group coverage; andComputing expected changes in health care expen-ditures and costs to the state for each reinsurance option simulated, as well as summarizing the changes in health insurance coverage expected to result for different populations of interest.Step 1. Creating a Baseline Database for Each StateComprehensive modeling of the effects of rein-surance policies in a state requires individual-level data that are population-based, so that we can simulate the costs and behavior of the uninsured. Data on insurance claims alone are insufficient. A useful model also requires data that are representative of state health expendi-tures and coverage, demographic characteristics, and employer characteristics. Additionally, since reinsurance policies affect the composition of the risk pools for each insurance product, simu-lation of reinsurance requires that the baseline data also characterize the health expenditure profile and number of individuals enrolling in coverage together in each risk pool.Finally, modeling reinsurance requires having an accurate distribution of health care expenses in the upper tail of the distribution across risk groups, because some reinsurance options can be expected to target such very high expendi-tures. The implication of this final requirement is that the data need to have a sufficiently large number of observations to obtain an ade