Overview

This course covers empirical strategies aimed at answering economic research questions. The course illustrates several applied econometric models, and develops the skills needed to plan and carry on empirical research projects. Since most applied economic research examines questions with direct policy or managerial implications, this course will clarify the difference between correlation and causal links. Purely predictive methods, grounded in correlation and regression analysis will be presented and compared with alternative estimation techniques (such as instrumental variables) more focused on the analysis of causal effects. The goal of the course is to provide valuable econometric tools for students majoring in several fields, from micro/macro/international/industrial economics to finance, and management. The agenda includes regression, instrumental variables, differences-in-differences, probit, logit, tobit, survival analysis. One of the goals is to equip students with working knowledge of the tools of probability and statistics, skills in data handling and statistical programming, and an understanding of the models and methods of applied econometrics. To this aim, problem sets with both analytical and computer-exercise components will be a relevant part of the course. The cases will be illustrated and discussed using the open source econometric package Gretl (gretl.sourceforge.net). Students should be familiar with basic concepts in probability, statistics and matrix algebra. The course includes a brief refresher, just in case.

Course objectives

• To cover the basic theory and practice of modern econometrics at an appropriate level
• To teach the students the habit of thought, knowledge and understanding to be able to carry out good quality applied economic research with confidence
• To develop the critical insight to appraise econometric results obtained by other researchers
• To provide students with the ability to use econometric software in an effective manner

Main topics

• Basics refresher: estimation and testing theory, matrix algebra
• The linear multiple regression model under standard assumptions: estimation through Ordinary Least Squares, reading and interpreting the regression output, using the model for prediction.
• Checking assumptions of the regression model: functional form, multicollinearity, heteroskedasticity, autocorrelation, non-normality
• Using the model to investigate causal relationships: omitted variables, endogenous regressors and the method of instrumental variables
• Using the model to analyze the effectiveness of policies and managerial choice: the differences-in-differences method
• Analysis of binary variables (probit and logit models)
• Analysis of categorical variables (multinomial logit models)
• Analysis of ordinal variables (ordered probit models)
• Analysis of count variables (Poisson regression)
• Analysis of truncated and censored variables (Tobit models)
• Analysis of durations (survival model

Course organization

The lectures (36 hours) will introduce and explain the theoretical concepts; the practice sessions (6 hours) will be used to discuss and solve exercises; the computer lab sessions (12 hours) will be used to analyze cases, using suitable econometric software. The cases will be distributed a few days in advance, in order to encourage active participation of the students.

(i) Written exam (1+½ hours, weight 60%); the first part of the written exam may be taken as a midterm exam. (ii) Short oral exam (15 minutes, weight 40%) (iii) Non-compulsory team project work  (2 students): a paper about 10/15 pages long, where some of the techniques illustrated in the course is applied to a real problem; the paper has to be illustrated in a 30 minutes presentation, possibly via skype; the project work may increase the final mark by at most 2/30