Thomas G. Koch's Homepage

Research

Fields of Interest

Primary Research Interests: Medical insurance and the labor market.
Secondary Interests: Insurance and Health Economics, Labor market costs of disease, Unemployment Insurance in models with search frictions.
Primary Fields of Study: Labor Economics and Macroeconomics

Current Research

Public Insurance, Crowd-out and Health Care Utilization: Public health insurance expansions increased health care utilization and health in the 1980s and early 1990s. Did more recent expansions do the same? I estimate the causal effect of eligibility on the health care utilization of children, using the discontinuous rules that determine eligibility. I find that the more recent public health insurance expansions actually decreased total expenditures, office-based visits, and pharmaceutical use, while increasing ER use. These patterns are consistent if less generous public insurance crowds out private insurance. Using household information on parental insurance offers, I find evidence in support of this "crowding down" hypothesis.

Pareto Gains from Hiding Information: Asymmetric information can lead to adverse selection and market failure. In spite of this, asymmetric information may be a Pareto improvement. Asymmetric information mitigates the variance of fairly priced insurance contracts faced by an agent behind the veil of ignorance. This result first is demonstrated in a static three risk-type model. In a dynamic setting, asymmetric information limits potential future price variation. This welfare gain offsets the costs of adverse selection. A dynamic model, when calibrated to the US medical insurance market, finds large Pareto gains from asymmetric information

The Shifting Shape of Risk: Endogenous Market Failure for Insurance: Micro-foundations models typically assume that most or all of the idiosyncratic risk a household faces is uninsurable. This paper considers an economy where risk is insurable, but endogenous market failure renders some risk uninsured. I find that both the level and the shape of the distribution of risk are important in determining the amount of market failure. In particular, changes to the distribution of medical risk since 1996 should have induced more risk sharing across types. Changes in the distribution associated with changing prices of medical care are primary. Changes associated with a shifting age distribution are not as important.

Optimal Taxation of Insurance Markets with Adverse Selection: Subsidizing insurance can be Pareto improving when the market suffers from adverse selection. I construct a model of insurance choice with adverse selection and endogenous pricing to measure the welfare consequences of the tax subsidy of medical insurance. When matched to cross-sectional medical insurance and expenditure data, there are no Pareto improvements. Increasing the subsidy rate would increase average welfare, though with limited returns. The current subsidy rate, 26%, is efficient if insurance induces $1,100 of wasteful spending by the insured.

A Hidden Welfare Cost of Saving Medicare: A popular solution to the looming entitlements crisis in the United States is to increase the eligibility age for Medicare. Using an empirical model of insurance choice with endogenous prices, I find that this would have a modest welfare cost. The losses are concentrated among those who lose Medicare coverage. For the non-elderly, the benefit of lower taxes is mitigated by a higher price for medical insurance.

Bankruptcy, Medical Insurance and a Law with Unintended Consequences : The number of consumer bankruptcies has grown dramatically. Meanwhile, the proportion of workers with medical insurance has fallen, while its price has grown. This coincidence began around 1986, when Congress passed the Emergency Medical Treatment and Active Labor Act (EMTALA), guaranteeing a standard of medical care, thus limiting potential medical liability. I construct a tractable general equilibrium model of the medical insurance market, and find that repealing EMTALA would reverse these trends. Further, I find that changes in the distribution of medical risk should have led to increasing medical insurance rates after 1986. It actually fell over this period.

Measuring Hedonic Markets: Equalizing Differences and the Tax Subsidy of Medical Insurance: In the United States, the cost of private medical insurance is not taxed as income. The focus of this paper is: how many workers will forgo private medical insurance if those costs are taxable? Using medical expenditure data available in the Medical Expenditure Panel Survey (MEPS), I construct each household's willingness to pay for medical insurance, given its expected medical expenditure and the variance of their medical expenditures, conditional on demographic and medical information. I find that the number of workers without private medical insurance would increase by half if its cost were taxed as income.

Other Research

Optimal Replacement Rates in a Search Economy: Endogenizing UI: I consider why UI systems are different in their generosity to the unemployed. I first posit a model of optimal UI in a frictional labor search. This yields a first-order condition which optimizes the level of UI provided by governments. Cross-country calibration of UI levels reveals the importance of Nash bargaining to the optimal level of UI. A simple GMM procedure using US UI data reinforces these results.

The Total Cost of Disease, joint with Brett Wendling. Using a rich collection of data on disease, treatment and labor market outcomes, we measures the cost of disease, both in expenditures on treatment, and lost labor force productivity.

Unemployment Duration, Unemployment Insurance and Human Capital: Longitudinal Survey Data and Methods, joint with Richard W. Evans. In this study, we estimate the effects of human capital on the unemployment hazard using longitudinal survey data. We find that the level of human capital has a negative effect on the unemployment duration in specifications that use panel data to account for unobserved individual heterogeneity, but has a positive effect in specifications that do not directly account for unobserved heterogeneity.

Note: Much of the work listed above is in progress; titles and descriptions are subject to change.