Selection by health status and medical care use in a social health insurance with an equalization fund: The Colombian case

Under a social health insurance with an equalization fund (SHIEF), revenues from a mandatory payroll tax are pooled into an equalization fund that reimburses for-profit insurance companies according to a capitated formula based on sex and age. Managed competition with capitated payment should induce private insurers to control costs without reducing the quality of service necessary to attract consumers. Although this incentive may force private insurers to produce efficiently, it also introduces an incentive for them to select against bad risks. Using 1997 Colombia household survey data, I studied whether risk selection occurs in a SHIEF by comparing the health status of participants to that of non-participants. I also analyzed how social health insurance coverage affects the individual probability of using medical care (i.e., physician visits, hospital use and outpatient medical care use). I estimated a fully specified model that captures the interactions among medical care use, social health insurance, and private insurance.; To account for the endogeneity problem between an individual's health insurance coverage and medical utilization, I jointly estimated a three-equation model using a full information maximum likelihood method (FIML) with regional instrumental variables. In addition, I employed a discrete factor method (DFM) to relax the joint normality distributional assumptions of the unobserved variables in the model.; The FIML results suggest a lack of evidence of selection of good risks (i.e., positive selection). According to the self-reported health status, healthier individuals are more likely to participate in SHIEF; however, other measures of individual health status indicate selection of bad risks (i.e., negative selection).; Controlling for selection problems in health insurance is important in order to obtain accurate estimates of the relationship between health insurance and medical care use according to the FIML and DFM. Once I removed the bias from the health insurance parameters, the new estimates had a positive but smaller effect on outpatient medical care use according to the FIML estimation. Using FIML instead of a simple (uncorrelated) model to control for the selection problem due to health insurance, the effect remained positive but became larger for measures of preventive care and hospital. These results could be due either to the selection of good risks in both types of health insurance programs, or to the differences in individuals' preferences toward quality.; According to the simulation results, enforcing universal coverage in the social health insurance would slightly affect the consumption of preventive services. The results reflect a larger effect on the use of hospital services and outpatient services. Since truly mandated social health insurance would not cause a significant change in the use of medical services, its impact on the costs of medical care may be lower than expected.