Collaborations

Research Culture 

Dr. Christis Katsouris's research interests lie broadly in econometrics, particularly, time series econometrics, panel data econometrics and macroeconometrics. His research focuses on econometric theory and methods with economically relevant applications. 

• Research Software: The performance of econometric methods is assessed with both simulated and real data using econometric and statistical software. 

• Research Supervision:  I would be happy to supervise excellent postgraduate students with strong programming and quantitative skills on topics close to my research areas of expertise. From the time series econometrics literature topics include: robust econometric estimation and inference, forecast evaluation and equal predictive accuracy tests as well as variable and model selection in the presence of multiple structural breaks. From the macroeconometrics literature topics include: identification, estimation and inference of structural parameters in macro models. From the econometrics literature topics include: identification strategies, estimation and inference with economically relevant applications. 

Research Impact

As an Early Career Researcher my overarching goal is to sustain a high-quality research outputs and to get published in leading academic journals in economics and econometrics. My current research focuses on developing robust econometric methods for time series and panel data.

Specifically, for the validity of proposed econometric methods we aim to develop asymptotic theory for model estimators and test statistics that facilitate inference. Simulation studies can be used to examine the efficacy and robustness of methods to time series properties that violate classical econometric assumptions. In particular, conventional estimation and inference techniques can be affected by features such as serial correlation, heteroscedasticity, nonstationary volatility and heavy-tailedness. Econometric theory in time series regression models with persistent regressors under heteroscedasticity and weak dependence requires further study. Tackling finite-sample bias using techniques such as variable transformations, bias-adjustments and  bootstrap resampling, allows to obtain asymptotically valid inference inference and improved estimation. For example, a solution to the presence of asymptotic bias in inference procedures for Local Projections is the implementation of data-dependent selection rules for specifying VAR(h) approximations to VAR (∞) models. Depending on the econometric model valid inferences can hold without bias corrections. Therefore, the development of estimation and inference methods that are robust against features such as the incidental parameter problem and serial correlation in error terms, allow to address economic-policy driven questions.