Local Composite Quantile Regression for Regression Discontinuity
Journal of Business & Economic Statistics, vol. 40, pp. 1863-1875, 2022

PDF Code

Towards Profitability: A Profit-sensitive Multinomial Logistic Regression for Credit Scoring in Peer-to-peer Lending
Proceedings of the Future Technologies Conference 2022, vol. 1, pp. 696-718, 2022

Conference Proceedings Link

Leverage and Asymmetric Volatility: The Firm-Level Evidence
Journal of Empirical Finance, vol. 38, pp. 1-21, 2016

PDF Journal Link

Quasi-maximum Likelihood Estimation of Multivariate Diffusions
Studies in Nonlinear Dynamics & Econometrics, vol. 17, pp. 179-197, 2013

PDF Journal Link

Quasi-maximum Likelihood Estimation of Discretely Observed Diffusions
The Econometrics Journal, vol. 14, pp. 241-256, 2011

PDF Journal Link

Panel Vector Autoregression under Cross-Sectional Dependence
The Econometrics Journal, vol. 11, pp. 219-243, 2008

PDF Journal Link

Finite Sample Properties of FGLS Estimator for Random-Effects Model under Non-normality
Contributions to Economic Analysis, vol. 274, pp. 67-89, 2006

PDF Book Chapter Link


I have developed the following R packages for some of my research:

  • boostvar: An R package to compute the estimates, standard errors, and p-values for the least-squares boosting method in vector autoregression models. This package implements the method in Huang (2022) and provides the standard error and p-value for all parameter estimates at every boosting step.
  • lboost: An R packages that implements lassoed boosting for high-dimensional linear regressions. Huang (2021) shows that combining the lasso with boosting can yield good accuracy and sparcity recovery in estimation.
  • rdcqr: An compaion R package for the estimation and inference in sharp and fuzzy regression discontinuity designs in causal analysis. This package implements the local composite quantile regression method in Huang and Zhan (2021) and can give accurate estimation and inference results when data are non-normal.


I offer several courses at Kennesaw State University.

  • Econ 2200: Principles of Economics - Macroeconomics
  • Econ 4490: Computing Methods for Business Data (an R programming class with business applications)
  • Econ 4610: Macroeconomics
  • Econ 4710: Econometrics
  • Econ 4760: Business Forecasting (an applied time series class)
  • Econ 7710: Statistics for Business Analysis (special topic, MBA level. We use R in business applications.)
  • Econ 8700: Econometrics and Forecasting Methods (MBA level)
  • Econ 8900: Statistical Learning of Financial Markets (special topic, Ph.D. level)
  • DBA 9104: Applied Methodologies in Finance and Accounting (DBA level)

I also teach a time series course at the Department of Economics in Georgia State University.

  • Econ 9740: Time Series Analysis (Ph.D. level)

Recent Posts

In the online supplement of Huang and Zhan (2021), we provide two simulation tables to demonstrate that local composite quanitle regression can be used to estimate sharp kink regression discontinuity (RD) and sharp RD with covariates. Links to the R code to replicate these tables are provided in this post for the intereted reader to download. The R code to replicate the simulation table for sharp kink RD can be downloaded here.


It is not very common to have a long folder path in setwd() in R. However, when you do have a long working folder path and you would like to make several breaks while typing the path inside setwd(), it will be nice to have a simple way to do that. The problem is after a line break, R automatically inserts a new line sign and several space immediately before the next path character.


There are various ways to organize your regression output tables in the project. The R packages such as stargazer and pander give simple solutions. I give an example of using the stargazer package to present a regression table in word documents. In Rstudio, go to Tools and Install Packages. Install the stargazer package. Let us use the subprime mortgage dataset as an example in the RMD file. Please download the data here.