Healthcare Plans and Patient Outcomes: Evidence from Bankruptcy-Induced Random Assignment in Colombia (2023). With Leonardo Bonilla, Nick Papageorge, Christian Posso, and Matthew Zahn. [Working Paper].
We examine the effect of random assignment to health insurance plans on mortality and healthcare utilization. Our data are from Colombia, which provides a unique context to study how health plans affect outcomes. First, the entire public healthcare system is a form of “managed competition” where profit-seeking firms receive government subsidies to provide healthcare to patients, with the exact same financial coverage. Second, when health insurance firms go bankrupt, which happens with some frequency, patients are ran- domly assigned to remaining firms by a government scheme that mimics a “shift-share” instrumental variable. We exploit this random assignment to assess how different insur- ances companies affect patients. We first provide evidence that random assignment works as intended: actual assignment is nearly identical to what the government’s randomization scheme would suggest. Two, we show that bankruptcies in general lead to lower average mortality, suggesting that firms going bankrupt do a relatively poor job of keeping patients alive and that “managed competition” helps remove particularly bad firms from the market. Third, we show healthcare utilization declines following a bankruptcy, which we attribute to the reallocation of patients to more efficient health insurers.