Amanda Kowalski, Associate Professor of Economics at the Yale University Department of Economics and Faculty Research Fellow at the National Bureau of Economic Research (NBER), is a health economist who specializes in bringing together theoretical models and econometric techniques to answer questions that inform current debates in health policy.
Professor Kowalski's recent research advances methods available to analyze data from experiments. Applied to the Oregon Health Insurance Experiment, these methods show that future insurance expansions could increase or decrease emergency room utilization, depending on the individuals who sign up for coverage. Her other recent research explores the early impact of the Affordable Care Act and the long-term impact of Medicaid expansions. In previous research, she examined the impact of the Massachusetts health reform of 2006 on hospital care, labor market outcomes, and adverse selection in the individual health insurance market. She has also studied the price elasticity of expenditure on medical care and the marginal returns to medical spending on at-risk newborns using new estimation techniques. Her research has received the Zellner Thesis Award, the HCUP Outstanding Article of the Year Award, the Garfield Economic Impact Award, the National Institute of Health Care Management Research Award, and the Yale Arthur Greer Memorial Prize.
In 2014, Professor Kowalski was honored with a CAREER Award from the National Science Foundation. The National Institutes of Health, the Robert Wood Johnson Foundation, and the W.E. Upjohn Institute have also supported her research, which has been published in peer-reviewed journals, including the American Economic Review, the Quarterly Journal of Economics, the Journal of Health Economics, and the Journal of Public Economics. Her research has also been featured in the popular press, including The New York Times, NPR, and The Wall Street Journal.
Professor Kowalski holds a PhD in economics from MIT and an AB in economics from Harvard. Before joining Yale, she held a post-doctoral fellowship in Health and Aging at the NBER. Her interest in health policy has led her to spend two years in Washington, DC, one as a research assistant in health and labor at the White House Council of Economic Advisers, and another as the Okun Model Fellow at the Brookings Institution. She spent the 2015-2016 academic year on Associate Professor Leave from Yale as a Visiting Associate Professor at the Stanford Institute for Economic Policy Research.
"How to Examine External Validity Within an Experiment." Working Paper. Latest Version: August 2016. Prepared for the Journal of Economic Perspectives, under review.
A fundamental concern for researchers who design and analyze experiments is that
the experimental result might not be externally valid in another context. Researchers
have traditionally attempted to assess external validity by comparing data from an
experiment to other data. In this essay, I use insights from my recent work to show
researchers how to begin the examination of external validity internally, within the
data from a single experiment. My insights rely on overlooked information and minimal
"Doing More When You're Running LATE: Applying Marginal Treatment Effect Methods to Examine Treatment Effect Heterogeneity in Experiments." NBER Working Paper 22363. Latest Version: July 2016. Under review.
I examine treatment effect heterogeneity within an experiment to inform external validity. The local average treatment effect (LATE) gives an average treatment effect for compliers. I bound and estimate average treatment effects for always takers and never takers by extending marginal treatment effect methods. I use these methods to separate selection from treatment effect heterogeneity, generalizing the comparison of OLS to LATE. Applying these methods to the Oregon Health Insurance Experiment, I find that the treatment effect of insurance on emergency room utilization decreases from always takers to compliers to never takers. Previous utilization explains a large share of the treatment effect heterogeneity. Extrapolations show that other expansions could increase or decrease utilization.
"Medicaid as an Investment in Children: What is the Long-Term Impact on Tax Receipts?" (with David Brown and Ithai Lurie). NBER Working Paper 20835. Latest Version: January 2015. Revise & resubmit, American Economic Review.
[BKL Medicaid Calculator]
We examine the long-term impact of expansions to Medicaid and the State Children's Health Insurance Program that occurred in the 1980's and 1990's. With administrative data from the IRS, we calculate longitudinal health insurance eligibility from birth to age 18 for children in cohorts affected by these expansions, and we observe their longitudinal outcomes as adults. Using a simulated instrument that relies on variation in eligibility by cohort and state, we find that children whose eligibility increased paid more in cumulative taxes by age 28. These children collected less in EITC payments, and the women had higher cumulative wages by age 28. Incorporating additional data from the Medicaid Statistical Information System (MSIS), we find that the government spent $872 in 2011 dollars for each additional year of Medicaid eligibility induced by the expansions. Putting this together with the estimated increase in tax payments discounted at a 3% rate, assuming that tax impacts are persistent in percentage terms, the government will recoup 56 cents of each dollar spent on childhood Medicaid by the time these children reach age 60. This return on investment does not take into account other benefits that accrue directly to the children, including estimated decreases in mortality and increases in college attendance. Moreover, using the MSIS data, we find that each additional year of Medicaid eligibility from birth to age 18 results in approximately 0.58 additional years of Medicaid receipt. Therefore, if we scale our results by the ratio of beneficiaries to eligibles, then all of our results are almost twice as large.
"What Do Longitudinal Data on Millions of Hospital Visits Tell Us About Public Health Insurance as a Safety Net for the Young and Privately Insured? NBER Working Paper 20887. Latest Version: January 2015. Under Review.
Young people with private health insurance sometimes transition to the public health insurance
safety net after they get sick, but popular sources of cross-sectional data obscure how frequently
these transitions occur. We use longitudinal data on almost all hospital visits in New York from
1995 to 2011. We show that young privately insured individuals with diagnoses that require
more hospital visits in subsequent years are more likely to transition to public insurance. If we
ignore the longitudinal transitions in our data, we obscure over 80% of the value of public health
insurance to the young and privately insured.
Video Presentations on Health Reform and the Labor Market:
We model the labor market impact of the key provisions of the national and Massachusetts "mandate-based" health reforms: individual mandates, employer mandates, and subsidies. We characterize the compensating differential for employer-sponsored health insurance (ESHI) and the welfare impact of reform in terms of "sufficient statistics." We compare welfare under mandate-based reform to welfare in a counterfactual world where individuals do not value ESHI. Relying on the Massachusetts reform, we find that jobs with ESHI pay $5,350 less annually, approximately the cost of ESHI to employers. Accordingly, the deadweight loss of mandate-based health reform was approximately 2% of its potential size.
Efforts to control medical care costs depend critically on how individuals respond to
prices. I estimate the price elasticity of expenditure on medical care using a censored
quantile instrumental variable (CQIV) estimator. CQIV allows estimates to vary across
the conditional expenditure distribution, relaxes traditional censored model assumptions,
and addresses endogeneity with an instrumental variable. My instrumental variable
strategy uses a family member's injury to induce variation in an individual's own price.
Across the conditional deciles of the expenditure distribution, I find elasticities that vary from -0.76 to -1.49, which are an order of magnitude larger than previous estimates.
Insurance induces a tradeoff between the welfare gains from risk protection and the welfare
losses from moral hazard. Empirical work traditionally estimates each side of the tradeoff
separately, potentially yielding mutually inconsistent results. I develop a nonlinear budget set
model of health insurance that allows for both simultaneously. Nonlinearities in the budget
set arise from deductibles, coinsurance rates, and stoplosses that alter moral hazard as well
as risk protection. I illustrate the properties of my model by estimating it using data on
employer sponsored health insurance from a large firm. Within my empirical context, the
average deadweight losses from moral hazard substantially outweigh the average welfare gains
from risk protection. However, the welfare impact of moral hazard and risk protection are
both small relative to transfers from the government through the tax preference for employer
sponsored health insurance and transfers from some agents to other agents through a common
In this paper, we develop a new censored quantile instrumental variable (CQIV)
estimator and describe its properties and computation. The CQIV estimator combines
Powell (1986) censored quantile regression (CQR) to deal with censoring, with a control
variable approach to incorporate endogenous regressors. The CQIV estimator is obtained in
two stages that are nonadditive in the unobservables. The first stage estimates a nonadditive
model with infinite dimensional parameters for the control variable, such as a quantile or
distribution regression model. The second stage estimates a nonadditive censored quantile
regression model for the response variable of interest, including the estimated control variable to deal with endogeneity. For computation, we extend the algorithm for CQR developed by Chernozhukov and Hong (2002) to incorporate the estimation of the control variable. We
give generic regularity conditions for asymptotic normality of the CQIV estimator and for
the validity of resampling methods to approximate its asymptotic distribution. We verify
these conditions for quantile and distribution regression estimation of the control variable.
Our analysis covers two-stage (uncensored) quantile regression with nonadditive first stage as an important special case. We illustrate the computation and applicability of the CQIV
estimator with a Monte-Carlo numerical example and an empirical application on estimation
of Engel curves for alcohol.
We develop a model of selection that incorporates a key element of recent health reforms:
an individual mandate. Using data from Massachusetts, we estimate the parameters of the
model. In the individual market for health insurance, we find that premiums and average costs
decreased significantly in response to the individual mandate. We find an annual welfare gain
of 4.1% per person or $51.1 million annually in Massachusetts as a result of the reduction in
adverse selection. We also find smaller post-reform markups.
Radio Interview on Early Impact of the ACA:
I examine the impact of state policy decisions on the early impact of the
ACA using data through the first half of 2014. I focus on the individual
health insurance market, which includes plans purchased through exchanges
as well as plans purchased directly from insurers. In this market, at
least 13.2 million people were covered in the second quarter of 2014,
representing an increase of at least 4.2 million beyond pre-ACA
state-level trends. I use data on coverage, premiums, and costs and a
model developed by Hackmann, Kolstad, and Kowalski (2013) to calculate
changes in selection and markups, which allow me to estimate the welfare
impact of the ACA on participants in the individual health insurance
market in each state. I then focus on comparisons across groups of states.
The estimates from my model imply that market participants in the five
direct enforcement states that ceded all enforcement of the ACA to the
federal government are experiencing welfare losses of approximately $245
per participant on an annualized basis, relative to participants in all
other states. They also imply that the impact of setting up a state
exchange depends meaningfully on how well it functions. Market
participants in the six states that had severe exchange glitches are
experiencing welfare losses of approximately $750 per participant on an
annualized basis, relative to participants in other states with their own
exchanges. Although the national impact of the ACA is likely to change
over the course of 2014 as coverage, costs, and premiums evolve, I expect
that the differential impacts that we observe across states will persist
through the rest of 2014.
Video Presentations of Massachusetts Research:
In April 2006, Massachusetts passed legislation aimed at achieving near-universal health insurance coverage.
The key features of this legislation were a model for national health reform, passed in March 2010. The
reform gives us a novel opportunity to examine the impact of expansion to near-universal coverage state-wide.
Among hospital discharges in Massachusetts, we find that the reform decreased uninsurance by 36%
relative to its initial level and to other states. Reform affected utilization by decreasing length of stay and the
number of inpatient admissions originating from the emergency room. When we control for patient severity, we find evidence that preventable admissions decreased. At the
same time, hospital cost growth did not increase.
We implement an empirical test for selection into
health insurance using changes in coverage induced by the
introduction of mandated health insurance in Massachusetts. Our test
examines changes in the cost of the newly insured relative to those
who were insured prior to the reform. We find that counties with
larger increases in insurance coverage over the reform period face
the smallest increase in average hospital costs for the insured
population, consistent with adverse selection into insurance before
the reform. Additional results, incorporating cross-state variation
and data on health measures, provide further evidence for adverse
Response to comment on "Estimating Marginal Returns to Medical Care: Evidence from At-Risk Newborns"
In Almond, Doyle, Kowalski and Williams (2010), we describe how marginal returns to medical care
can be estimated by comparing patients on either side of diagnostic thresholds. Our application
examines at-risk newborns near the very low birth weight threshold at 1500 grams. We estimate large
discontinuities in medical care and mortality at this threshold, with effects concentrated at "low-quality"
hospitals. While our preferred estimates retain newborns near the threshold, when they are
excluded the estimated marginal returns decline, although they remain large. In low-quality hospitals,
our estimates are similar in magnitude regardless of whether these newborns are included or excluded.
We estimate marginal returns to medical care for at-risk newborns by comparing health outcomes and medical treatment provision on either side of common risk classifications, most notably the "very low birth weight" threshold at 1500 grams. First, using data on the census of US births in available years from 1983-2002, we find evidence that newborns with birth weights just below 1500 grams have lower one-year mortality rates than do newborns with birth weights just above this cutoff, even though mortality risk tends to decrease with birth weight. One-year mortality falls by approximately one percentage point as birth weight crosses 1500 grams from above, which is large relative to mean one-year mortality of 5.5% just above 1500 grams. Second, using hospital discharge records for births in five states in available years from 1991-2006, we find evidence that newborns with birth weights just below 1500 grams have discontinuously higher costs and frequencies of specific medical inputs. We estimate a $4,000 increase in hospital costs as birth weight approaches 1500 grams from above, relative to mean hospital costs of $40,000 just above 1500 grams. Taken together, these estimates suggest that the cost of saving a statistical life of a newborn with birth weight near 1500 grams is on the order of $550,000 in 2006 dollars.
This study examines the impact of state health insurance regulations on the price of high-deductible family and individual polices in the nongroup market. We use a unique and rich data set on actual insurance policies sold through a large Internet health insurance distributor to examine the impact of various regulations on policy prices, controlling for policy characteristics, demographic characteristics of the purchasers, and state-level demographics. We also use data from a single major insurance firm that provided offer prices for a family policy from a set of randomly selected zip codes. Both datasets suggest a strong statistical relationship between regulation and insurance prices.
Selected Work in Progress
"Politics, Medicare Payments, and Hospital Behavior." (with Zack Cooper, Eleanor Neff Powell, and Jennifer Wu). Latest Version: Coming soon.
"Returns to Medicare Spending: Evidence From Variation Across Physicians." (with Joseph J. Doyle and Heidi Williams). Latest Version: Coming soon.
"Medicaid Expansions and Health Spending Growth." (with Mikhail Golosov and Rebecca McKibbin). Latest Version: Coming soon.
Call for Randomization
"The Use and Evaluation of Experiments in Health Care Delivery"
"CQIV: Stata Module to Perform Censored Quantile Instrumental Variable Regression." First Version: December 2010. Latest Version: June 2012. (with Victor Chernozhukov, Ivan Fernandez-Val, and Sukjin Han). [Link to CQIV at RePEc]
The simplest way to install this command is to type the following at the Stata command prompt:
ssc install cqiv
You can also download the .ado and .sthlp files [Download CQIV Stata ado file][Download CQIV Stata help file], and then copy them into your personal ado directory [How to find your personal ado directory].
If you are updating from a previous version, type "net uninstall cqiv" at the Stata prompt before installing.
Problem Set on "Adverse Selection and an Individual Mandate: When Theory Meets Practice." First Version: January 2016. Latest Version: October 2016. (Developed with Austin Schaefer, Jack Welsh, and Megan Wilson).
This problem set is designed for graduate students and advanced undergraduates. It does not require Stata or Excel.
Download the PDF file with instructions.
Answer key available, instructors only please. Email Amanda Kowalski.
Problem Set on "Mandate-Based Health Reform and the Labor Market: Evidence from the Massachusetts Reform." First Version: May 2012. Latest Version: October 2016. (Developed with Toby Chaiken, Jonathan Kolstad, and Megan Wilson).
This problem set is designed for graduate students and advanced undergraduates. It requires Stata and Excel.
Download Stata, Excel, and Word Files as a zip file. The Word file contains instructions.
Answer key available, instructors only please. Email Amanda Kowalski.
Problem Set on "Estimating Marginal Returns to Medical Care: Evidence from At-risk Newborns." First Version: February 2012. Latest Version: October 2016. (Developed with Tiffany Fan and Michael Cunetta).
This problem set is designed for undergraduates. It requires Stata.
Download Stata and Word Files as a zip file. The Word file contains instructions.
Answer key available, instructors only please. Email Amanda Kowalski.
When available, research assistant positions are posted in the following locations:
Full-Time Research Positions paid through Yale
Full-Time Research Positions paid through the NBER
Part-Time Research Positions for Yale Undergraduates
Summer 2016 Research Team at NBER Summer Institute (from left): Jennifer Wu, Yen Tran, Rebecca McKibbin, Matthew Tauzer, Samuel Moy, Ljubica "LJ" Ristovska, Rae Staben, Amanda Kowalski.
Summer 2015 ISPS Health Research Team (from left): Amanda Kowalski, Edward Kong, Megan Wilson, Cindy Zheng, Samuel Moy, Rebecca McKibbin, Maggie Zhou, Jennifer Wu, Stuart Craig (not pictured: Aigerim Kabdiyeva, Saumya Chatrath).
Summer 2013 Research Team (from left): William Bishop, Martin Hackmann, Kate Archibald, Tiffany Fan, Amanda Kowalski, Gerardo Ruiz Sanchez.