Working Paper

• Analyzing Dynamic Multiple Spell Durations using Counting Processes,

  I propose a counting process approach to analyze multiple spell duration data. These data are doubly stochastic in a sense that both durations for an individual and the number of durations within a fixed period are random. I allow unobserved individual heterogeneities to enter into the model as fixed effects. In addition, conditional on the individual fixed effect, durations are state dependent. A first-difference transformation is developed to cancel fixed effects, and a minimum distance estimator is re-introduced with simplified proofs. Finite sample properties are investigated in simulations. The approach is applied to studying an individual’s work absence decisions.

• Consistent Test for Conditional Moment Restriction Models in Reproducing Kernel Hilbert Spaces (with Xiaojun Song),

  In this paper, we represent Integrated Conditional Moment (ICM) tests in Reproducing Kernel Hilbert Spaces (RKHS). There are several advantages of doing so. First, reproduc-ing kernels embody dimension and integral measure, and hence, are effective dimension reduction tools. This phenomenon can be explained by the isometrically isomorphic relationship among infinite dimensional Hilbert spaces. Second, the test statistics, expressed in terms of kernels, have analytic closed forms, making them easy to compute in practice. Third, one can generate kernels easily and massively from existing kernels. Each kernel corresponds to an ICM test, thus, for certain models, one may obtain a more sensitive test than by using conventional ones. We further propose projection-based kernels to eliminate estimation effect, leading to a simple multiplier bootstrap procedure to obtain critical values. A minimum distance estimator is developed as a byproduct. Monte Carlo exercises are performed to examine the finite sample performance of the proposed test, and an empirical application is studied.

• Static and Dynamic Incentives in Individual Outpatient Claims: Identification and Quantification (with Rui Cui),

  This paper investigates the impact of static and dynamic incentives on patient behavior in the context of health insurance with deductibles. Using data from the Rand Health Insurance Experiment (Rand HIE) for analysis, we propose a novel approach that focuses on healthcare events rather than healthcare expenditures to identify and quantify patients' incentives from those of physicians. We utilize a conditional shadow price and explicitly specify the state-dependent structure of impacts of previous events on subsequent ones. The study's main findings reveal that patients respond to both nominal and shadow prices, but on average dynamic incentives have roughly four times greater impact compared to static incentives. Furthermore, incentive effects are not uniform across different individuals, with static incentives having a greater impact on ''heavy users,'' and dynamic incentives affecting ''light users'' more. Lastly, we find that patients time their healthcare needs and exhibit retaliatory behaviors after reaching their deductible limits.

• Goodness-of-Fit Tests for General Counting Processes (with Xiaojun Song and Rui Cui).

  We propose an omnibus goodness-of-fit test for general counting processes. We show our test is consistent against (almost) any deviation, and can detect local alternatives tending to the null at a √n rate. We contribute to the literature in the following aspects. First, the test statistic is constructed based on an empirical process rather than a sequence of transformed event times. Second, we explicitly take the estimation effect into consideration when bootstrapping the critical value. Third, the proposed framework is valid for both the one-observational counting process as well as the n-observational process. Monte Carlo experiments results suggest good size and power properties of our test, and a simple empirical application is also studied.

Working in Process

• A Dynamic Analysis of Spanish Youth Job Turnover (with Miguel A. Delgado)