This paper studies consumers' demand for quality in the nursing home market, where information frictions are a source of concern. I start by estimating quality of nursing homes in California, before using these estimates as inputs into a structural demand model. I find that despite substantial local variation in quality, average demand for quality is low, even after accounting for unobserved supply side constraints arising from selective admissions practices by nursing homes. Patterns of demand heterogeneity point to information frictions being a major reason for this low demand. Counterfactual simulations based on estimates of the structural model and a competing risks model suggest that eliminating information frictions may reduce deaths by at least 8 to 28 percent in the short run, and potentially even more in the long run.
In regression discontinuity designs with multiple running variables (MRD designs), units are assigned to treatment base on whether their value on several observed running variables exceed known thresholds. In this design, applied work commonly uses analyzes each running separately: for example, when financial aid eligibility depends on GPA and family income, researchers separately consider the sample of students with low enough family income and study at the GPA threshold, and the sample of students with high enough GPA and study the income threshold. However, this approach does not fully exploit the richness of the data. I propose a new estimator for MRD designs using thin plate splines which improves upon the applied practice in two ways. First, the estimator provides efficiency gains by using the entire sample, and second, it may be used to estimate the conditional average treatment effect at every point on the boundary separating treated and untreated units. I provide an automated procedure for undersmoothing which eliminates the asymptotic bias of the estimate as well as Bayesian confidence intervals for the estimator, and derive theoretical results justifying the use of these methods. I find in a simulation study that the estimator is roughly unbiased in finite samples and that the confidence intervals have close-to-coverage empirical coverage. Finally, I demonstrate the performance of my estimator in an empirical application from Londoño-Vélez, Rodríguez, and Sánchez (2020), which studies the effect of a large financial aid program on higher education in Colombia. R code for estimation and inference is available.
Selection on unobservables is an important concern for causal inference in observational studies, and accordingly, previous papers have developed methods for sensitivity analysis for OLS, binary choice models, instrumental variables, and movers designs. In this paper, I develop methods for sensitivity analysis for a setting that has not been previously studied — discrete choice models. In particular, I derive bounds for the omitted variables bias under an assumption about how much the consumer values the omitted variable(s) relative to the included control variables, and about the relationship between the omitted variable and the variable of interest. I provide theoretical results for my bounding procedure, and demonstrate its performance in simulations. Finally, I show in several empirical applications that my procedure produces economically meaningful bounds.
Research in Progress
Racial Segregation and Choice Disparities Across Nursing Homes: Does the Distinction Matter for Policy?
Minority residents are disproportionately concentrated in low-quality nursing homes, and similar patterns of segregation and choice disparities are also present in other settings, such as school and neighborhood choice. However, while there is a link between segregation and disparities, policymakers may care about these issues for distinct reasons, and it is unclear whether policies that reduce racial segregation and disparities are one and the same. In this project, I take advantage of an administrative data set on the universe of nursing home residents to study several forces driving racial segregation and disparities in nursing home choice. Event study results show that a positive shock to the share of minority admissions in a nursing home results in a persistent increase in future share of minority admissions, consistent with in-group preferences. In addition, minority residents tend to live further away from high-quality nursing homes than white residents, and nursing homes are less likely to admit minority residents when capacity is strained. Next, to assess the relative contributions of these different forces towards racial segregation and choice disparities, I estimate a structural model that incorporates these factors and conduct counterfactual simulations. The simulations show that information interventions targeted at minorities may reduce choice disparities but are less effective at eliminating racial segregation across nursing homes, and vice versa if we only eliminate in-group preferences or discriminatory admissions practices.
Challenges in Measuring Mental Health Trends
Public awareness of mental health issues has grown in recent years, and there is a common perception that mental health in the population is worsening, concerns that are supported by descriptive evidence. However, changes in public attitudes towards mental health make the interpretation of these trends challenging: low response rates to mental health surveys make their results sensitive to changes in sample selection bias, and even diagnosis rates in comprehensive data sets such as the Medicare data are a function of individuals' willingness to seek professional help. In this paper, I address these measurement issues using a comprehensive data set on all nursing home residents, so it does not suffer from sample selection bias. Similar to Medicare data, it contains information on whether each resident was diagnosed with depression in the recent past, but in addition, it contains a rich set of psychosocial measures, which provides us with a detailed picture of the resident's underlying mental health. I find that while depression diagnoses at admission increased from 19 to 24 percent for residents admitted to a nursing home in California between 2000 and 2010, underlying mental health of these residents (based on observed psychosocial behavior as well as machine-learning predictions using hundreds of covariates) was roughly constant over the same period. These results illustrate the perils of inferring mental health trends from survey evidence or diagnosis trends alone.
Selective Admissions and Discharges by Nursing Homes
Previous research has shown that as a consequence of capacity constraints, nursing homes selectively choose which types of residents to admit (Gandhi, 2019; Cheng, 2022), and when to discharge residents (Hackmann, Pohl, and Ziebarth, 2020). I provide a microfoundation for a structural model where arrivals of different types of potential residents and the evolution of “discharge readiness” of existing residents follow certain stochastic processes, and nursing homes choose optimal optimal admission and discharge policies that maximize expected present discounted value of future profits. The solution to this problem yields testable implications, and shed light on identification of the structural model – intuitively, nursing homes’ admission and discharge policies are identified by differences in the characteristics of residents they admit and discharge during times of high and low occupancy. I estimate the model using an extension of the Gibbs sampler in Agarwal and Somaini (2022) and Cheng (2022), with data augmentation on residents’ indirect utility and latent variables that determine nursing homes’ admission decisions for potential residents and discharge decisions for existing residents.
The issue of fake news has been hotly debated in recent years, with some commentators claiming that it played a role in US presidential elections and the Brexit vote. Despite these claims, there has been limited evidence to date linking fake news directly to voting behavior. In this project, I seek to provide credible evidence on this question by using big college football games as an instrument for fake news consumption. I find that search volumes for pro-Trump fake news terms were lower in counties close to college football teams that played a big game shortly before the election, and also that these counties were less likely to vote for Trump. The magnitude of these estimates suggest that a one-standard deviation increase in search volume for pro-Trump fake news terms increased Trump’s vote share by about 4.5 percent. Finally, I do not find evidence that fake news affected overall turnout rates, or that fake news resulted in down-ballot effects.
This paper studies how cigarette consumption responds over time to changes in tax rates. Using a panel of state data, I estimate that the cumulative effect of an excise tax rise on consumption is larger than the cumulative effect of an increase in sales tax, in line with a theory of tax salience. In addition, I find that consumption falls in advance of an excise tax hike, whereas it only falls in the year after a sales tax increase. The pattern of consumption response to sales taxes is also consistent with consumer learning over time.