Chang Ching Lin

Estimation of Panel Data Models with Parameter Heterogeneity when Group Membership is Unknown

Abstract This paper proposes two methods for estimating panel data models with group specific parameters when group membership is not known. The first method uses the individual level time series estimates of the parameters to form threshold variables. The problem of parameter heterogeneity is turned into estimation of a panel threshold model with an unknown threshold value. The second method modifies the K-means algorithm to perform conditional clustering. Units are clustered based on the deviations between the individual and the group conditional means. The two approaches are used to analyze growth across countries and housing market dynamics across the states in the U.S.

On the Robustness of Symmetry Tests for Stock Returns

In this paper, by using a generalized asymmetry measure with the heteroskedasticity autocorrelation consistent estimation method and a long-run variance eliminating method, we propose two generalized symmetry tests in the presence of unknown distributions and serial dependence. The proposed tests encompass existing skewness tests, and generate new symmetry tests that are robust to both the heavy-tails and the serial dependence of stock returns. We also utilize the concept of an augmented distribution to establish an asymmetric distribution family that encompasses Pearsons type-IV distribution, and we use this distribution family and the score test principle to discuss the choice of asymmetry measures for testing symmetry. In this study, we also compare our tests with existing tests using a Monte Carlo simulation and an empirical example, and show that the robust tests outperform existing tests for checking the symmetry of stock returns.

The Behaviour of Real Exchange Rates: The Case of Japan

The study examines the convergence rate of mean reversion by contrasting the estimated half-life of real exchange rate (RER). We employ an extensive monthly consumer price index (CPI)-based product price’s panel for Japan (the U.S. as the num´eraire). We find that the disaggregated RERs are persistent due to the cross-sectional dependence problems. By controlling common correlated effects, the estimated half-life for all goods may fall to as low as 2.54 years, below the consensus view of 3 to 5 years summarized by Rogoff (1996). After correcting the small-sample bias, the estimated half-life of deviations from purchasing power parity (PPP) increase by 1.03 year. Our findings also support that the half-life of mean reversion of RER is about 3.55 years for traded goods, about 0.11 year lower than non-traded goods. We also show that traded goods and non-traded goods perform distinct distributions of persistence.

On long-run covariance matrix estimation with the truncated flat kernel

Despite its large sample efficiency, the truncated flat kernel (TF) estimator of long-run covariance matrices is seldom used, because it occasionally gives a non-positive semidefinite estimate and sometimes performs poorly in small samples, compared to other familiar kernel estimators. This paper proposes simple modifications to the TF estimator to enforce the positive definiteness without sacrificing the large sample efficiency and make the estimator more reliable in small samples through better utilization of the bias-variance trade-off. We study the large sample properties of the modified TF estimators and verify their improved small-sample performances by Monte Carlo simulations.

Reserve Prices in Online Auctions:Theory and Evidence

The paper proposes a simple model of auctions with an impatient seller who chooses the reserve price and the buy-it-now (BIN) price to maximize revenue. The three main sales channels in the online auction (the pure auction, the BIN auction, and the fixed-price sale) are each shown to be an optimal solution for the sellers with certain degrees of time-preference. The theory also predicts that the optimal posted price for the fixed-price listing is greater than the optimal reserve price for the BIN auction, which in turn is greater than that for the pure auction. This prediction is confirmed by data from eBays auction of iPods. Using the sellers inventory as a proxy for his degree of time-preference, the data also confirm the theoretical prediction that there is an inverse relationship between the optimal reserve price and the sellers time-discount factor.

How Did Workers Fare in Labor Mediation – An Empirical Assessment from Taiwan

Using newly collected data on labor disputes in Taiwan, we study how laborers fare in labor mediation. Our strongest result is that in the cases involving small claims, workers obtained a substantially high portion of their claim, a result unlikely to occur if those laborers had to resort to litigation. Overall, our study indicates that labor mediation is effective in helping the financially disadvantaged individual workers to obtain substantial recovery from their employers.