Description
The goal of this course is to continue building theoretical foundation and practical skills needed for independently conducting an econometric research project. This course is built around techniques of analysis of cross-sectional and panel data. First, we will be dealing with specification and data problems followed by the panel data models. We will then move onto the consequences of endogeneity and instrumental variable estimation and the systems of equations estimation. In the second part of the course we will cover estimation techniques of nonlinear models (maximum likelihood). In this context we will consider some limited dependent variable models (Probit, Logit) along with other advanced topics.
Will be interesting for
- Analysts working with cross-section and panel data, students, professors.
- Everyone who wants to understand the basics of the methods used to study causal relationship using cross-section and panel data.
After course completion, you will be able to
- Apply some most widely used econometric tools to analysis of cross-sectional and panel data;
- Understand the role of empirical evidence in evaluating economic problems;
- Discover and comprehend large segments of the applied economics literature through active problem-based learning;
- Effectively perform data management and analysis techniques using software (cross-sectional and panel data);
- Understand the issues researchers face when they analyze nonexperimental data and ways these problems are solved.
Prerequisites:
Probability theory and statistics, simple linear regression, multiple regression analysis
Faculty:
Solomiya Shpak – Ph.D. in Public Policy from George Mason University and an MA in Economics from Kyiv School of Economics and the University of Houston.
Language
English
Education format
Twice a week for seven weeks
Start
Tuesdays ad Thursdays from 5pm to 6:20pm
Price
10 000 uah
Course outline
1. Endogeneity. Specification and Data Problems
2. Pooled cross sections and Panel data models (fixed and random effects models)
3. Difference-in-differences model
4. Instrumental Variables (IV)
5. Simultaneous equation model (SEM)
6. Generalized Least Squares (GLS and FGLS). Seemingly unrelated equations (SURE)
7. Estimation of non-linear regression models. Maximum Likelihood.
8. Binary outcome dependent variables: linear probability model. Logit and Probit models
9. Limited Dependent Variables: Censored (Tobit models; duration models) and Truncated Regression Models
10. Limited Dependent Variables: Count variables (poisson model)
11. Multiple outcome dependent variables: multinomial/conditional logit/probit, nested logit, mixed logit, ordered logit/probit models
