Publications
"Strategical Interactions on Municipal Public Safety Spending with Correlated Private Information", (with Lung-fei Lee), Regional Science and Urban Economics, forthcoming,2017.
"Social Interactions under Incomplete Information with Heterogeneous Expectations." (with Lung-fei Lee), Journal of Econometrics, 198. pp.65-83, 2017.
"Career Concern and Tax Preparer Fraud,” (with Liansheng Wu and Xianhui Bo), Annuals of Economics and Finance, Vol.11, Issue 2, 355-379, 2010.
"Strategical Interactions on Municipal Public Safety Spending with Correlated Private Information", (with Lung-fei Lee), Regional Science and Urban Economics, forthcoming,2017.
"Social Interactions under Incomplete Information with Heterogeneous Expectations." (with Lung-fei Lee), Journal of Econometrics, 198. pp.65-83, 2017.
"Career Concern and Tax Preparer Fraud,” (with Liansheng Wu and Xianhui Bo), Annuals of Economics and Finance, Vol.11, Issue 2, 355-379, 2010.
Working Papers
Tobit Models with Social Interactions: Complete vs. Incomplete Information
(with Xi Qu and Lung-fei Lee, submitted to Quantitative Econometrics)
Abstract: In many network data sets, the outcomes of interest are equal to zero for some agents and strictly positive for others. They can be analyzed by Tobit models with social interactions. Under complete information, all the observables and unobservables are publicly known. Under incomplete information, unobservables and some covariates observed by econometricians can be private information for agents. A Cox-type test is proposed for model selection. For property tax rates among adjacent municipalities in North Carolina, significant competing effects are found under both incomplete and complete information. However, the Cox Test is in favor of the complete information model.
(with Xi Qu and Lung-fei Lee, submitted to Quantitative Econometrics)
Abstract: In many network data sets, the outcomes of interest are equal to zero for some agents and strictly positive for others. They can be analyzed by Tobit models with social interactions. Under complete information, all the observables and unobservables are publicly known. Under incomplete information, unobservables and some covariates observed by econometricians can be private information for agents. A Cox-type test is proposed for model selection. For property tax rates among adjacent municipalities in North Carolina, significant competing effects are found under both incomplete and complete information. However, the Cox Test is in favor of the complete information model.
Social Interactions under Incomplete Information with Multiple Equilibria
Abstract: Socially interacted behaviors under incomplete information can be modeled as equilibrium outcomes of a simultaneous move game. Parameter identification and estimation can be based on the equilibrium expected outcomes. When there are asymmetric information on the exogenous characteristics, the equilibrium expectations are heterogeneous, varying with both individual's traits and the private information used to make predictions. When there are multiple equilibria, the set of equilibrium expectations are a set of functionals defined on the space of private information, which has not been fully characterized in the previous literature. Utilizing the intersection theory of differential topology and functional fixed point theorems, we find that when all exogenous characteristics are public information and only the idiosyncratic shocks are privately known, the set of equilibria is composed of a finite number of vectors and can be computed via the homotopy continuation method. When some exogenous covariates are private information, the equilibrium set is compact in a Banach space and can be approximated by a finite number of equilibria. Thus, it can be numerically computed using basis functions, Gauss-Legendre quadrature, and the homotopy method. Attaching a probability mass function to this approximated set, a computationally-feasible approximation of the complete sample likelihood is derived. Estimation is achievable by either maximizing the likelihood function or using simulated moments. This paper supplements the economic theory on games with multiple equilibria and extends the all-solution method for estimation of discrete choice games to a general framework incorporating discrete and continuous choices, bounded and unbounded outcomes, as well as di erent types of incomplete information. This method is especially useful for the model with peer effects, where the dimension of the equilibrium conditional expectation functionals can be reduced. We analyze the binary choice models in detail. Monte Carlo experiments show that our estimation method performs well. In addition, large estimation biases can occur if imposing
equilibrium uniqueness, either the assumed unique equilibrium is computed through contraction mapping or Newton's method.
Abstract: Socially interacted behaviors under incomplete information can be modeled as equilibrium outcomes of a simultaneous move game. Parameter identification and estimation can be based on the equilibrium expected outcomes. When there are asymmetric information on the exogenous characteristics, the equilibrium expectations are heterogeneous, varying with both individual's traits and the private information used to make predictions. When there are multiple equilibria, the set of equilibrium expectations are a set of functionals defined on the space of private information, which has not been fully characterized in the previous literature. Utilizing the intersection theory of differential topology and functional fixed point theorems, we find that when all exogenous characteristics are public information and only the idiosyncratic shocks are privately known, the set of equilibria is composed of a finite number of vectors and can be computed via the homotopy continuation method. When some exogenous covariates are private information, the equilibrium set is compact in a Banach space and can be approximated by a finite number of equilibria. Thus, it can be numerically computed using basis functions, Gauss-Legendre quadrature, and the homotopy method. Attaching a probability mass function to this approximated set, a computationally-feasible approximation of the complete sample likelihood is derived. Estimation is achievable by either maximizing the likelihood function or using simulated moments. This paper supplements the economic theory on games with multiple equilibria and extends the all-solution method for estimation of discrete choice games to a general framework incorporating discrete and continuous choices, bounded and unbounded outcomes, as well as di erent types of incomplete information. This method is especially useful for the model with peer effects, where the dimension of the equilibrium conditional expectation functionals can be reduced. We analyze the binary choice models in detail. Monte Carlo experiments show that our estimation method performs well. In addition, large estimation biases can occur if imposing
equilibrium uniqueness, either the assumed unique equilibrium is computed through contraction mapping or Newton's method.
Sample Selection with Social Interactions
Abstract: I correct the estimation biases for socially interacted actions arising from endogenously determined social groups by explicitly modeling agents' selections about whether to participate in the activity. This relates to a two-stage game with both discrete and continuous choices under incomplete information. The model framework incorporates three types of interesting effects: social interactions, selection biases, and inter-temporal effects. With possibly multiple equilibria, I devise a stochastic rule to complete the model and estimate parameters by maximal likelihood estimation.
Abstract: I correct the estimation biases for socially interacted actions arising from endogenously determined social groups by explicitly modeling agents' selections about whether to participate in the activity. This relates to a two-stage game with both discrete and continuous choices under incomplete information. The model framework incorporates three types of interesting effects: social interactions, selection biases, and inter-temporal effects. With possibly multiple equilibria, I devise a stochastic rule to complete the model and estimate parameters by maximal likelihood estimation.