«By JONATHAN D. KETCHAM, NICOLAI V. KUMINOFF AND CHRISTOPHER A. POWERS* We develop a structural model for estimating the welfare effects of poli- cies ...»
In the LE scenario, only 2% of consumers are made better off. For most people, the utility loss from being forced to switch plans more than offsets the cost savings, risk reduction, and improvements in plan quality experienced by switchers. The small fraction of winners all belong to the suspect group. Hence, if we think that inertia primarily reflects hassle costs and consumer preferences, then the menu restriction makes the vast majority of consumers worse off in order to produce small benefits for a small share of people in the suspect group because they are less able to choose inferior plans.
The first two columns of Table 6 summarize how the policy affects insurer revenue per enrollee, government spending per enrollee via the premium subsidy, and the average change in expected welfare per enrollee. The ME scenario predicts a slight increase in 34 average consumer surplus ($12), driven by large gains for a small fraction of consumers.
The LE scenario predicts a larger reduction (-$108). Both scenarios predict increases in government spending ($4 to $17) and insurer revenue ($6 to $36). Hence, one of the main effects of the policy is to transfer revenue from consumers and taxpayers to insurers.
The last six columns show comparable results for three other hypothetical rules for how CMS could determine which plans to keep on the menu: the plans that are on the efficiency frontier for the greatest number of people; the plans with the minimum average cost to the enrollee; and the plans with the highest net revenue per enrollee.34 Our results on consumer welfare are qualitatively robust across these scenarios. The most striking difference is the order of magnitude increases in insurer revenue and government spending that occur when insurers are allowed to retain their highest profit plans. The more profitable plans tend to be the higher-premium ones that provide more risk reduction and have higher quality ratings. Hence, insurers would have strong incentives to persuade regulators to allow them to retain their more comprehensive plans at the expense of consumers and taxpayers. With approximately 7.7 million people currently participating in the standalone Medicare prescription drug markets the change in annual government transfers would range from a reduction of $0.07 to $0.14 billion under the minimum expenditure rule to an increase of $1.6 to $1.8 billion under the maximum profit rule.
C. Distributional Effects of Personalized Decision Support
Our second policy experiment evaluates the welfare effects of a hypothetical information campaign modeled on a randomized field experiment conducted by Kling, Mullainathan, Shafir, Vermeulen, and Wrobel (2012) [henceforth KMSVW]. Their analysis is motivated by the observation that while Medicare enrollees can learn about their personal PDP options and potential savings by calling 1-800-Medicare or using various cost calculators available online, a minority of enrollees report doing so, as seen in Table 1.
KMSVW attribute this to “comparison friction” which they define as the wedge between 34 For profitability, we assume that there is sufficiently little variation in within the set of plans offered by each insurer that it does not affect the ranking of plans by revenue per enrollee. Under this assumption the ranking of plans within each brand is defined by ∑.
35 available information and consumers’ use of it. KMSVW tested an intervention in which several hundred treatment group enrollees who agreed to participate in the experiment were sent a decision support letter containing personalized information about their potential personal cost savings from switching to their lowest cost available plan. The letter also identified the name of the low cost insurer and contact information to initiate a switch.
KMSVW observed a 7 percentage point increase in the rate at which the treatment group switched to their lowest cost plan relative to a control group that received a general letter with no personalized decision support, and an 11.5 percentage point increase in the overall switching rate for the treatment group.
In this experiment we estimate the heterogeneous welfare implications of a national rollout of the decision support tool in which the government mails letters to all enrollees that would be similarly worded to the one sent to the treatment group in KMSVW’s study. Such a policy may affect welfare via several pathways. First, as the authors suggest, providing enrollees with personalized information may mitigate psychological biases and/or reduce information costs, making them better off. In the context of our model, this would be realized as increases in the switch rate and cost savings. Because the decision support tool does not embed risk protection and quality, however, the net effect on welfare is ambiguous. Second, an important feature of the information campaign—if it were implemented by the government—is that it would necessarily be based on incomplete information about enrollees’ drug needs. While CMS has full information about an individual’s claims over their prior years in the PDP market, the individual may have private information about their own drug needs over the upcoming year. If enrollees with private information about changes in their drug needs choose to switch plans based on outdated information provided by CMS then these misinformed individuals could experience welfare losses.35 Finally, increased switching initiated by a national rollout could induce feedback effects on premiums that would further affect welfare.
We use KMSVW’s estimated treatment effects as moments that we can use to calibrate and. Specifically, in the ME scenario we multiply the estimated inertia paIn principle such a phenomenon could exist if the free but imperfect information from CMS reduces efforts for people to acquire private information about their own future drug needs. Carlin, Gervais, and Manso (2013) explore these ideas more generally.
In the LE scenario, there is assumed to be no change in the behavior of the suspect group so we use (16.a) and (16.c), in which case and will have to be larger than in the ME scenario in order to induce sufficient switching among the non-suspect group to replicate the treatment effects estimated by KSMVW.
Figure 3 summarizes the distributional effects of the personalized information treatment. In both the ME and LE scenarios the policy is welfare enhancing for more than two thirds of consumers. The winners are between 3 to 6 percentage points less likely to be diagnosed with dementia or depression and 1 to 4 percentage points less likely to have a college degree. There is virtually no difference between winners and losers in terms of gross drug expenditures.
FIGURE 3: DISTRIBUTION OF WELFARE EFFECTS FROM PERSONALIZED DECISION SUPPORT37 $1,500 Most effective nudge: inertia is not latent preferences; policy changes behavior $1,000 Least effective nudge: inertia is latent preferences; policy does not change behavior $500 $0 0 0.5 1
5 ‐5 ‐$1,000 ‐15
Note: The figure shows CDFs of the expected change in welfare from a personalized decision support tool that is based on the field experiments of Kling et al. (2012). The model is calibrated to reproduce their estimated treatment effects on the rates at which people switch plans. The bar charts report demographics for the average enrollee with welfare gains and losses under alternative assumptions about the efficacy of the nudge.
In the ME scenario the winners are more likely to be in the suspect group. The information treatment induces them to place more emphasis on cost savings and many of them switch plans as a result. Figure 4 shows that the winners enjoy an average reduction in premiums and OOP expenditures of $41 along with risk reduction and quality improvements worth another $12. The losers have substantially higher OOP expenditures. This is because the low cost plan that is featured by the information treatment is the one that minimizes their expenditures based on their prior year of drug use. A small share of people who experience significant health shocks would spend substantially more in the recommended plan than in the plan that they actually chose for themselves. These individuals are concentrated in the non-suspect group. This illustrates the potential welfare losses that can arise from a nudge based on incomplete information. More broadly, this suggests a tradeoff between the potential benefits of simplifying the presentation of information and the potential costs of deemphasizing important details about the assumptions underlying that information.
In the LE scenario the winners are 4 percentage points more likely to have incomes over $25k and 17 percentage points less likely to be in the suspect group. This is because the information treatment is assumed to not change the behavior of individuals in the suspect group. The individuals with the largest welfare gains tend to be people in the nonsuspect group who benefit because the information treatment lowers their cost of collecting information about alternative plans and induces them to actively switch plans. Overall, 67% of enrollees are better off as a result of the policy.
By inducing consumers to switch to lower cost plans, the information treatment reduces insurer revenue and government expenditures per enrollee. These figures are shown in the first two columns of Table 7. Recall that our model of decision making assumes that consumers have unbiased expectations of their actual drug use in the upcoming year. This could cause us to understate the policy’s benefits. If consumers are myopic in the sense that they expect their drug use to be the same as the prior year then there is less scope for the information treatment to reduce welfare. The last two columns of Table 7 demonstrate this and show that when we repeat the estimation and simulation based on the assumption that consumers are myopic, then between 76% and 91% of consumers benefit from the policy and the average change in welfare is an increase of between $75 and $193. Adverse selection still makes some people worse off by raising plan premiums.
Our final policy experiment evaluates the welfare effects of replacing CMS’s current revealed preference approach to defining each person’s reenrollment default plan with an alternative policy that would set the default to be the plan that would minimize each enrollee’s costs. We envision the policy being implemented as a stronger version of the decision support nudge. Not only would enrollees be informed of their minimum cost options, they would be automatically assigned to those options unless they chose to opt out by overriding the reassignment and choosing a different plan. As before, we assume CMS would predict each enrollee’s minimum cost plan using their drug claims from the prior year. Consistent with CMS’s current approach, first-time enrollees would still be required to make active decisions.
In the ME scenario, the policy completely erases inertia for enrollment in the new low-cost default. Nevertheless, some consumers may still prefer their original plans if those plans provide greater quality or variance reduction. Assuming it is costless for enrollees to opt out and continue in their old plans, the policy could reduce consumer welfare for only two reasons: (1) endogenous increases in premiums, or (2) (mis)assignment to plans requiring higher expenditures due to changes in drug needs. Figure 5 shows that across the distribution of enrollees the aggregate effect of these two mechanisms is almost always dominated by the aggregate effect of lower expenditures and the elimination of inertia. Overall, 90% of consumers have gains in expected welfare. Figure 6 shows that approximately half of them choose to stay in the low cost default.36 The others choose to opt out and remain in their current plans. The average winner experiences a substantial reduction in their premium ($56) and OOP expenditures ($85) and a smaller reduction in the monetary value of plan quality ($38). The policy tends to benefit people who have more drug claims, more drug expenditures, higher rates of cognitive illness, 36 While the aggregate probability of remaining in the low cost default is 48% the probabilistic nature of the logit model requires that there is a positive probability that every enrollee chooses to remain in their low cost default. The probability weighted consumer surplus from this potential outcome feeds into each individual’s expected welfare gain.