We investigate how environmental regulation under the U.S. Superfund program and Clean Air Act affected exposures to fine particulate air pollution and hazardous waste for Americans over age 65 during the 2000s. Our research design uses quasi-random features of how the two programs enforce regulations and provide information to estimate their causal effects on migration and pollution exposure. We show that senior Americans' average pollution exposures declined substantially. We also show that spatially heterogeneous improvements in environmental quality had little-to-no effect on residential sorting. This led to relatively large reductions in pollution exposure for seniors living in the dirtiest areas.

We find that long-term exposure to fine-particulate air pollution (PM2.5) degrades health and human capital among older adults by increasing their risk of developing Alzheimer’s disease and related dementias. We track U.S. Medicare beneficiaries’ cumulative residential exposures to PM2.5 and their health from 2004 through 2013, leveraging within- and between-county quasi-random variation in PM2.5 resulting from the expansion of Clean Air Act regulations. We find that a 1 ig/m3 increase in decadal PM2.5 increases the probability of a dementia diagnosis by 1.68 percentage points. The effects are as large or larger when we adjust for mortality-based sample selection and additional Tiebout-sorting dynamics. We do not find relationships between decadal PM2.5 and placebo outcomes. Our estimates suggest that the federal regulation led to nearly 182,000 fewer people with dementia in 2013, yielding $214 billion in benefits. Further, PM2.5’s effect on dementia persists below the current regulatory thresholds.

We study whether teams' productivity improves as they gain experience working together. We leverage unique clinical data to observe team experience and individual physician and staff experience in coronary catheterization laboratories. Teams are composed of cardiologists, nurses, and technicians who work together synchronistically. We observe teams and individuals at hospitals across the United States from 2001 to 2009, including the rate at which they gain experience using drug-eluting stents (DES) from their introduction in the US in 2003 onward. We estimate models of productivity and clinical outcomes that account for team experience, physician experience, and staff experience conditional on each other and on time-invariant physician and staff characteristics, hospital-specific monthly effects, and an extensive set of patient-level clinical factors. Greater experience performing DES cases together improves teams' productivity, lowering total case time, procedure time, and non-physician labor costs while leaving clinical outcomes unchanged. In contrast, physicians' and staffs' individual experience with DES does not improve productivity conditional on other factors. The effects of team experience with DES appears generalized, with gains from experience with competitor brands of DES about as large as those from the specific brand of DES being used for a given case.

We study whether teams’ productivity improves as they gain experience working together. We leverage unique clinical data to observe team experience and individual physician and staff experience in coronary catheterization laboratories. Teams are composed of cardiologists, nurses, and technicians who work together synchronistically. We observe teams and individuals at hospitals across the United States from 2001 to 2009, including the rate at which they gain experience using drug-eluting stents (DES) from their introduction in the US in 2003 onward. We estimate models of productivity and clinical outcomes that account for team experience, physician experience, and staff experience conditional on each other and on time-invariant physician and staff characteristics, hospital-specific monthly effects, and an extensive set of patient-level clinical factors. Greater experience performing DES cases together improves teams’ productivity, lowering total case time, procedure time, and non-physician labor costs while leaving clinical outcomes unchanged. In contrast, physicians’ and staffs’ individual experience with DES does not improve productivity conditional on other factors. The effects of team experience with DES appears generalized, with gains from experience with competitor brands of DES about as large as those from the specific brand of DES being used for a given case.

We hypothesize that analyzing individual-level secondary data with instrumental variable (IV) methods can advance knowledge of the long-term effects of air pollution on dementia. We discuss issues in measurement using secondary data and how IV estimation can overcome biases due to measurement error and unmeasured variables. We link air-quality data from the Environmental Protection Agency's monitors with Medicare claims data to illustrate the use of secondary data to document associations. Additionally, we describe results from a previous study that uses an IV for pollution and finds that PM2.5's effects on dementia are larger than non-causal associations.

This study provides the first revealed preference evidence on the value of statistical life (VSL) for US seniors aged 67–97 from the rates at which they choose to consume medical care relative to other private goods, and by the effects of their choices on their survival probabilities. These effects are estimated from individuals’ survey responses linked with their Medicare records. Instrumental variables estimators provide robust evidence that the mean VSL is below $1 million and that it decreases with age, and, given age, increases with income, education, and health and is higher for women and people who never smoked. JEL classification: D90, J14, J17, Q51

: We propose new methods to model choice behavior and conduct welfare analysis in complex environments where it is untenable to assume that choices fully reveal preferences. In particular, we investigate how Medicare beneficiaries choose prescription drug plans (PDPs) under the Medicare Part D program. Our approach is novel in that we estimate a multinomial logit model for PDP choice that allows for heterogeneity in both preferences and the behavioral choice process. We find the data can be well characterized by a mixture of three behavioral types: The “rational” type constructs expected out-of-pocket costs E(OOP) rationally, and, ceteris paribus , seeks to minimize premiums plus E(OOP) as theory suggests. The second type constructs expected out-of-pocket (OOP) costs rationally, but puts too much weight on premiums relative to E(OOP) in choosing plans. A third type, who we label “confused,” places weight on irrelevant financial aspects of drug plans, implying they fail to construct E(OOP) rationally. A consumer is more likely to be the “confused” type if they suffer from Alzheimer’s disease and/or depression. We use the model to quantify the monetary and welfare losses that arise from suboptimal decision making for the population, for the behavioral types, and for people with cognitive limitations. We also evaluate policies to simplify the choice set to reduce these losses.

We develop a method that embeds signals about consumers' knowledge to evaluate prospective choice architecture policies. We analyze three proposals for U.S. Medicare prescription drug insurance markets: (i) menu restrictions, (ii) personalized information, and (iii) defaulting consumers to cheap plans. We link administrative and survey data to identify informed enrollment decisions that proxy for preferences of observationally similar misinformed consumers. Results suggest that each policy yields winners and losers, with the menu restrictions harmful to most but personalized information beneficial to most. These results are robust across signals of consumers' knowledge but differ from the benchmark that excludes such signals.

We propose new methods to model behavior and conduct welfare analysis in complex environments where some choices are unlikely to reveal preferences. We develop a mixture-of-experts model that incorporates heterogeneity in consumers' preferences and in their choice processes. We also develop a method to decompose logit errors into latent preferences versus optimization errors. Applying these methods to Medicare beneficiaries' prescription drug insurance choices suggests that: (1) average welfare losses from suboptimal choices are small, (2) beneficiaries with dementia and depression have larger losses, and (3) policies that simplify choice sets offer small average benefits, helping some people but harming others.

We develop a structural model for bounding welfare effects of policies that alter the design of differentiated product markets when some consumers may be misinformed about product characteristics and inertia in consumer behavior reflects a mixture of latent preferences, information costs, switching costs and psychological biases.We use the model to analyze three proposals to redesign markets for Medicare prescription drug insurance: (1) reducing the number of plans, (2) providing personalized information, and (3) defaulting consumers to cheap plans.First we combine administrative and survey data to determine which consumers make informed enrollment decisions.Then we analyze the welfare effects of each proposal, using revealed preferences of informed consumers to proxy for concealed preferences of misinformed consumers.Results suggest that each policy produces large gains and losses for some consumers, but the menu reduction would unambiguously harm most consumers whereas personalized information would unambiguously benefit most consumers.

The purpose of this rejoinder is to clarify key areas of agreement and disagreement with Abaluck and Gruber and address aspects of their reply to our comment, both of which appear in the December 2016 issue of the American Economic Review. Readers of our exchange may wonder how we can reach such divergent conclusions from analyzing the same data. In this rejoinder we show how. We demonstrate that Abaluck and Gruber’s criticism of our analysis is based on their mistaken claims about theory and empirics, their omission of key facts, and their emphasis on results that obscure our many areas of agreement.

Neoclassical and psychological models of consumer behavior often make divergent predictions for the welfare effects of paternalistic policies, leaving wide scope for researchers' choice of a model to influence their policy conclusions.We develop a framework to reduce this model uncertainty and apply it to administrative data on consumer decision making in Medicare Part D. Consumers' choices for prescription drug insurance plans can be explained by Abaluck and Gruber's (AER 2011) model of utility maximization with psychological biases or by a neoclassical version of their model that precludes such biases.We evaluate these competing hypotheses using nonparametric tests of utility maximization and a trio of model validation tests.We find that 79% of enrollment decisions in Medicare Part D from 2006-2010 satisfied basic axioms of consumer preference theory under the assumptions of full information, zero transaction cost, and no measurement error.The validation tests provide evidence against widespread psychological biases.In particular, we find that precluding psychological biases improves the structural model's out-of-sample predictions for consumer behavior.

We study whether people became less likely to switch Medicare prescription drug plans (PDPs) due to more options and more time in Part D. Panel data for a random 20 percent sample of enrollees from 2006--2010 show that 50 percent were not in their original PDPs by 2010. Individuals switched PDPs in response to higher costs of their status quo plans, saving them money. Contrary to choice overload, larger choice sets increased switching unless the additional plans were relatively expensive. Neither switching overall nor responsiveness to costs declined over time, and above-minimum spending in 2010 remained below the 2006 and 2007 levels.

Patients rely on physicians to act as their agents when prescribing medications, yet the efforts of pharmaceutical manufacturers and prescription drug insurers may alter this agency relationship. We evaluate how formularies, and the use of information technology (IT) that provides physicians with formulary information, influence prescribing. We combine data from a randomized experiment of physicians with secondary data to eliminate bias due to patient, physician, drug, and insurance characteristics. We find that when given formulary IT, physicians' prescribing decisions are influenced by formularies far more than by pharmaceutical firms' detailing and sampling. Without IT, however, formularies' effects are much smaller.

Using nationwide county-level longitudinal data, we show that recent declines in housing prices are associated with an increased utilization of antidepressant prescriptions among the near elderly. Our results persist in difference-in-difference models using either all non-antidepressant drugs or statins as controls.