Most of my works are a good combination of developing novel theories and applications with computational methods. One of my current research focuses on developing test statistics for the functional form of the regression functions under nonparametric nonlinear cointegrating regressions with regressors that are endogenous with long or semi-long memory innovations. Nonparametric cointegrating regression models have been extensively used in financial markets, stock prices, heavy traffic, climate data sets, and energy markets, where I use subsampling for these dependent and complex-structured data. My other research centers on the development of standard errors of point estimates from fine stratification designs with only one or two sampling units per stratum. This is a popular design since it permits the extensive stratification, but the standard error of its estimators can be biased or impractically large. I develop computationally fast hierarchical Bayesian approaches, which can perform better than the traditional methods. I also have done other works related to survey sampling such as developing predictors for area level positively skewed variables with measurement errors in small areas.
