It’s almost job market time, so how about a post about what may have been the best job market paper from last year? Ben had the perfect combination in empirical work: a great dataset with a fantastic natural experiment, an important topic, and very careful modeling.
The question is as follows. Many consumers make terrible decisions when it comes to health insurance. They are seemingly unable to choose the premium/co-pay/deductible bundle (henceforth just “bundle”) that by all accounts is best for them among the set their employer offers. How important is this effect, and should we “nudge” people into better plans, perhaps by encouraging more careful reading of a plan prospectus, annual meetings with employees, or similar? The answers, it turns out, are that consumers do make massive mistakes, and that nudging them into making better choices may actually be welfare-decreasing for employees as a whole due to adverse selection.
Ben uses data on a large self-insuring firm. Three years into the dataset, the firm switched from offering one PPO to offering three plans, all of which had the same doctors and coverage, but which had varying deductibles and premia. The year of the switch, there was no default: all employees had to actively choose a plan. In future years, employees were reenrolled in the same plan unless they actively switched. The premia was set each year based on the previous year’s expenditures per person for that plan, such that the employer subsidy was the same for each plan. Because of this, and because of adverse selection into the high-premium plan, the price of the high-premium plan increased substantially in year four. Very few new employees signed up. Nonetheless, most employees who already had the plan stayed with it even though the cost was very high. Indeed, for many employees who stuck with the plan, the high-premium plan was strictly dominated by another plan, no matter what health outcomes were realized in year four. The implied “switching costs” are very high.
But might this be a good thing? We know that in markets with private information, adverse selection can destroy the market for a product. The truly sick employees would be substantially worse off if the high-premium plan disappeared. But the existence of high switching cost coworkers can overcome this adverse selection, and perhaps increase welfare overall. After estimating (from the data) the implies risk aversion distribution, Handel uses a model from Johns Hopkins that takes previous year health spending for each employee and computes the expected health needs of that employee in the future. He then considers the welfare impact of cutting switching costs. Clearly the workers who are induced to switch to their “actually best” insurance plan are better off. Ignoring dynamic effects, the increase in welfare (measured by certainty equivalent) is about 10%.
Consider the dynamic effect, however. As these (relatively healthy) workers switch to low-premium plans, the remaining workers in the high-premium plan are sicker on average. Therefore, the premium for their plan will increase in the following year. This will cause further shifts out of the high-premium plan the next year, and so on. In Ben’s simulation, after six years, 80% of high premium plan employees have switched out, and the premium for the remainder has nearly doubled. The welfare (measured by certainty equivalent, again) of non-switchers declines substantially while that of switchers increases slightly. The net effect is a 6% decline in welfare.
I suppose nothing is surprising here: when there are market frictions, efficiency is lost, so to the extent that consumers “make mistakes” and we avoid those frictions, everyone may be better off. But the magnitude of the result, and the quality of the modeling, is superlative in this paper.
http://www.econ.berkeley.edu/~bhandel/wp/switchingcosts.pdf (Latest version of WP)