IO of Healthcare and Consumer Finance Markets

We document new patterns in investment behavior using a comprehensive dataset of 401(k) plans from 2009 through 2019. We show that there is substantial heterogeneity in asset allocation across plans, and that these differences are systematically predictable by sector of employment and demographic characteristics. For example, higher income and education is associated with more exposure to equities, while a greater share of minorities and retirees is associated with less equity exposure. These patterns cannot be rationalized by differences in investment options or plan participation. To understand observed investment behavior, we use a revealed preference approach that allows us to recover heterogeneity in investors’ (subjective) expectations and risk preferences. We find that differences in expectations play an important role in explaining portfolios. Further, we show that investors appear to form expectations based on local sources of information such as county-level GDP growth, home prices, and employer past performance. Overall, our findings are consistent with a model in which heterogeneity in investor expectations reflects idiosyncratic experiences and local environments.

This paper examines the impact of dynamic pricing regulation on market outcomes and social welfare in the U.S. long-term care insurance (LTCI) market. We first provide descriptive evidence that the introduction of rate stability regulation, which limits insurers’ ability to increase premiums over the lifetime of a contract, improved rate stability while reducing product variety. To quantify this trade-off, we develop and estimate a dynamic equilibrium model of LTCI where insurers have market power and cannot commit to future premiums. Our estimates suggest that consumers’ demand for LTCI is relatively price inelastic. However, insurers do not charge a high markup due to various pricing regulations. Using the estimated model, we conduct counterfactual experiments to assess the welfare effect of dynamic pricing regulation. We find that a stricter rate stability regulation lowers social welfare as the benefit from improved rate stability is outweighed by the cost from reduced product variety.

This paper studies how banks compete amid digital disruption and the resulting distributional effect across consumers. Digital disruption increases the geographic coverage of banking services, bringing new entrants to local markets. However, as digital customers shift from branches to digital services, banks close branches, and the remaining branching banks gain market power among non-digital customers that rely on branches. Consequently, digital customers benefit from the intensified bank competition at the cost of non-digital customers who pay higher prices for branch services and face the risk of financial exclusion. We provide empirical evidence by exploiting the staggered expansion of 3G networks, instrumented by regional distribution of lightning strike frequency. Using a structural model, we further quantitatively decompose the benefit and costs of digital disruption resulting from banks' pricing, branching, and entry decisions. The results highlight the role of banks' endogenous responses to digital disruption in widening the gap in access to banking services.

This paper tests for the presence of asymmetric information in the originate-to-distribute mortgage supply chain. In this supply chain, a mortgage specialist originates a loan that she sells to a bank which securitizes the loan into a mortgage-backed security(MBS) that he sells to investors. Using a novel data set on auctions through which mortgage specialists sell loans to banks, we show that the mortgage specialists have private information about loan quality. Using data on mortgage securitization, we show that banks have more information about loan quality than the MBS investors and that the banks use this to their advantage. The presence of asymmetric information should put downward pressure on the resale price of mortgages in the supply chain, and thus raise the costs to consumers seeking mortgages.

High administrative costs in U.S. health care have provoked concern among policymakers, but many of these costs are generated by managed care policies that trade off bureaucratic sludge against reductions in moral hazard. We study this trade-off for prior authorization restriction policies in Medicare Part D, where low-income beneficiaries are randomly assigned to default plans. Beneficiaries who face restrictions on a drug reduce their use of it by roughly 24.2%. Approximately half of marginal beneficiaries are diverted to another related drug, while the other half are diverted to no drug. These policies generated net fiscal savings, reducing drug spending by $84.94 per beneficiary-year (3.17% of drug spending), while generating $9.28-$10.91 in paperwork costs. Revealed preference approaches suggest that the value of the foregone drugs is likely to fall below the cost savings.

Insurer competition in hospital networks generates incentives for risk selection. I model this type of competition between insurers to understand the effect of risk adjustment and premium setting on hospital network breadth and consumer welfare. I use data from Colombia’s health care system to estimate the model. Every aspect of the Colombian national insurance plan is regulated by the government except for hospital networks, which insurers can choose separately for different services. I find that insurers risk-select by providing narrow networks in services that unprofitable patients demand the most. Eliminating risk adjustment reduces average network breadth by 6.7% and consumer welfare by 2.2%. Improving the risk adjustment formula increases average network breadth by 4.6%-28.0% and consumer welfare by 2.9%-8.0%, depending on how many risk factors are included. Price and non-price competition are substitutes for risk selection as a zero-premium policy exacerbates underprovision of insurance coverage. Results highlight hospital networks as an important dimension of non-price competition and cream-skimming in health care markets.

Low-income patients tend to receive lower quality health care. They have limited access
to high quality options and—even conditional on access—are less likely to choose high
performing providers. We show that differential information about provider quality is an important determinant of this disparity. Our empirical strategy exploits the temporary presence of a website that publicly displayed summary star ratings of general practitioner (GP) offices in England. Regression discontinuity (RD) estimates show that, on average, patients respond sharply to the information on the website, and that this response is almost entirely driven by residents of low-income neighborhoods. We incorporate our RD moments into a structural model of demand that allows for heterogeneity in information in addition to consumer inertia and heterogeneous preferences. Our results indicate that a meaningful fraction of the income-quality gradient could be eliminated by removing informational differences.

Entry represents a fundamental threat to cartels engaged in price fixing. We study the extent and effect of this behavior in the largest price fixing case in US history, which involves generic drugmakers. To do so, we link information on the cartel’s internal operations to regulatory filings and market data. We find that collusion induces significant entry, which in turn reduces prices. However, regulatory approvals delay most entrants by 2-4 years. We then estimate a structural model to assess counterfactual policies. We find that reducing regulatory delays by just 1-2 years equates to consumer compensating variation of $597 million-$1.52 billion.

In household finance markets, inactive households can implicitly cross-subsidize active households who promptly respond to financial incentives. We assess the magnitude and distribution of cross-subsidies in the mortgage market. To do so, we build a model of household mortgage refinancing and structurally estimate it on rich administrative data on the stock of outstanding UK mortgages in June 2015. We estimate sizeable cross-subsidies during this sample period, from relatively poorer households and those located in less-wealthy areas towards richer households and those located in wealthier areas. Our work highlights how the design of household finance markets can contribute to wealth inequality.

This paper studies the intermediation of auto loans through auto dealers using new and comprehensive administrative data. The arrangements between auto dealers and lenders incentivize dealers to increase loan prices. We leverage details of the corresponding contracts to demonstrate that many consumers are less responsive to finance charges than to vehicle charges. Taking this behavior into account, we estimate an equilibrium model of dealer price setting and lender competition. We explore counterfactuals where dealers have no discretion to price loans and final rates are set by lenders instead. We find large gains in consumer surplus from such a policy.

In many markets, prices are determined via bilateral negotiations between firms with market power. In such negotiations, parties often seek to improve their bargaining leverage. A downstream firm may threaten to exclude an upstream firm from its network. An upstream or downstream firm may merge with a competitor or a negotiating partner. In this paper, we study preferred retail pharmacy networks in the Medicare Part D prescription drug insurance program for the elderly to measure the impact of market structure and exclusion on equilibrium networks, prices, and consumer welfare. We estimate a game of network formation, price negotiation, and premium setting. We use a Nash-in-Nash with threat of replacement (NNTR) bargaining model for negotiated prices (Ho and Lee, 2019). In the NNTR model, insurers and pharmacy chains negotiate over prices and preferred network inclusion, and insurers can threaten to eliminate pharmacy chains from the preferred network entirely or replace them with any other, non-preferred pharmacy chain. Insurers set consumer premiums simultaneously. Next, Medicare beneficiaries enroll in a plan. Finally, enrollees experience prescription drug needs and fill them at pharmacies, and pay lower out-of-pocket prices at preferred pharmacies. We model retail pharmacy demand as a function of distance and expected consumer out-of-pocket costs, given enrollee characteristics (Starc and Swanson, 2021). We then estimate plan demand as a function of premiums and the ex ante expected value of the plan’s network implied by the pharmacy demand model, given consumer characteristics. Finally, using the estimated pharmacy and plan demand models, we recover bargaining weights implied by the NNTR network formation and pricing game. We estimate the bargaining weights by matching predicted NNTR prices to prices observed in the data in a GMM framework, conditioning on the observed network structure. The bargaining weights are identified by the presence of substitutes to insurers and pharmacy chains, including the presence of pharmacy chains that do not have preferred status with a given insurer and hence, which can be used as replacement threat points. Our estimates show that, allowing for exclusion, in the estimation meaningfully impacts the bargaining parameter estimates. Our preliminary estimates indicate that pharmacy chains have an average of 10-20% of the bargaining weight relative to insurers. This indicates that pharmacy chains capture only a relatively small fraction of the available surplus on average. We also demonstrate that the NNTR approach is better able to match the patterns in our data than the more common Nash-in-Nash bargaining model, in which insurers can threaten pharmacies with exclusion but not replacement. For most cases, the Nash-in-Nash approach misleadingly implies average bargaining weights for pharmacies that are very close to zero. Having estimated bargaining weights in the NNTR game, we then perform a series of counterfactual exercises, exploring the impact of horizontal mergers between pharmacy chains upstream, and between insurers downstream. Finally, we examine pro- and anti-competitive effects of vertical integration in this market, focusing on the impact of a recent acquisition between CVS and Aetna.

While regulatory authorities have largely focused on the efficiency losses associated with the exercise market power, less attention has been given to the possible distributional implications. This issue may be particularly salient in settings where private intermediaries provide goods and services that are differentially-subsidized to consumers based on observable characteristics; the distributional objectives of the government reflected in the subsidy schedule may be distorted by market power. In this paper, we develop a conceptual framework for assessing the channels through which market power influences efficiency and equity, and apply that framework to the market for health insurance plans under the Affordable Care Act. We find that market power leads to a modest (2%) loss in total welfare, but a much larger (8.5%) loss in consumer surplus. Critically, the loss in consumer surplus is regressive, with consumer surplus and insurance take-up falling disproportionately among lower-income, older, and sicker consumers. We also document that consumers only receive 50% of the subsidy spending, with the remainder captured by firms. Under a utilitarian metric, we show overall surplus would increase with income-invariant subsidies. However, with sufficiently high social preferences for redistribution, means-tested subsidies may be optimal. We conclude with a brief discussion of what our findings imply for the anti-trust regulation in product markets.