Relaxing the Assumptions of the Classical Linear Regression Model: Multicollinearity, small sample size, and heteroscedasticity
Cao Hao Thi
- Reading: DG3 Chapter 10-11
Problem set 1 Distributed
Relaxing the Assumptions of the Classical Linear Regression Model: Introduction to EViews
Cao Hao Thi
- Reading: DG3 Chapter 10-11
The course is structured in three parts. The first part reviews concepts related to the Classical Linear Regression Model (CLRM) and explores conditions under which the underlying assumptions this model fail, including multi-collinearity, heteroskedasticity, and autocorrelation. Then, students will learn econometric models with time series data, such as ARIMA and VAR models, and applications in economic forecasting. This part will also highlight important features of time series including non-stationarity, reverse causality, and spurious regressions. Class exercises will be conducted using EViews econometric software.
The second part introduces advanced models built up on the CLRM foundation. Restrictive conditions imposed on the classical model limits its applications in the real world. Therefore, this section opens up tools and methods to allow the estimation of many economic models when those conditions fail to hold. Student will learn the advantage of panel data over cross-sectional data. Two-stage regression and simultaneous-equation system models such as the demand-supply system can be used when economic decisions or public policies are determined endogenously. In the presence of limited dependent variables and self-selection problem, the course will introduce appropriate remedies including censored/truncated regression and Heckman correction. Class exercises in the second part will be conducted exclusively in Stata.
The last part will be an overall introduction of survey methods. Basic tasks of a real world survey from designing, implementing, to data processing and survey reporting will be presented. Basic statistics, sampling theory in empirical research are needed for students to apprehend the content of this part. Thus, the course will also spend sometimes on basic statistics in sampling theory and methods. The students will get familiar with the popular databases for analysis and research from the General Bureau of Statistics in Vietnam. The last lecture will be a practical exercise deploying actual data set and SPSS software.
