The course main objective is to provide students with a solid grasp of the probabilistic and statistical underpinnings of the most common econometric techniques. The topics studied will be of practical use to any student who plans on confronting data in their thesis or wishes to read and precisely understand the econometrics typically used in empirical research published in academic journals. Econometric theory has two parts:
The first part focuses on the specification and estimation of the linear regression models and its extensions.
The second part focuses on the theory and the applications of time series methods in econometrics.
Throughout, we will focus on both understanding and doing. The understanding will come from lectures, class discussions, and problem solving. The doing will come from extensive statistical software use. This course requires a quarter-long commitment. Econometrics is best learned by doing, and I will require you to do a fair amount of hands-on work. Hands-on work requires the use of computer software. In this class we use Julia.
Lectures of the first part of the course will be mostly based on the following two graduate textbooks:
Lectures of the second part will be based on the following textbooks:
Another reference is:
There will be two in class exam. A midterm and a comprehensive final.
The midterm will be held on March 28, 2017.
The grade will consist of a weighted average of these two exams and the problem sets. The midterm will count 30%, the problem set 30%, and the final the remaining 40%. The final grade will be the largest between the weighted grade as described above and the grade on the final. More specifically, denoting F, M, and PS your grade on the final, midterm, and problem set, respectively, your grade will be
Final Grade = max(.4*F + .3*M + .3*PS, .8*F + .2*PS, F)
Note: The midterm and the problem sets will be considered only for the first two final exam dates.
The following topics will be covered:
Asymptotic theory: law of large numbers and central limit theorems for iid and time series data
Parametric and semi-parametric estimation techniques
Properties of estimators
Conditional expectation and related concepts with application to econometrics
Single equation linear model and OLS estimation
Instrumental variables estimation of single equation linear model
Limited dependent variable models
Panel Data models
Univariate ARMA processes with stochastic and non-stochastic trends
Vector Autoregression (VAR)
Forecasting with ARMA and VAR models
Special Topics: Generalized Method of Moments
There will be a weekly problem set. The assignments will involve both theoretical and empirical work. Group study and free discussion are encouraged. But you should submit your own answers. You will probably find the class very hard to follow if you fail to spend sufficient time on all of the problem sets. The problem sets are part of the final grade as explained in the next section below.
If you have any question on the problem sets, please ask me or TA’s during our office hours. Our office hours are for you. I prefer to talk to you in person. I feel that email is not a very efficient way to ask econometric questions.
Problem set answers are to be turned in on time. You can hand in the homework AFTER the class. Late solution will not be accepted!
The TAs will lead a weekly practice session which will be held in the computer labs (301 and 306).
The sessions are an important part of the course and regular and attendance is strongly advised. During these sessions, the TA will review the concepts introduced in class and solve applied problems using Julia. Sessions are also useful for asking questions about the problem sets.
The following table gives the distribution of students across the two sessions:
Julia is a high-level, high-performance dynamic programming language for technical computing, with syntax that is familiar to users of other technical computing environments. In particular, its syntax is similar to Matlab. The similarity of the syntax means that a lot of Matlab code will run almost unmodified.
Julia has many advantages over other languages and for this reason is being extensively used in industries and in research.
Recently, the Federal Reserve of New York has open sourced its macroeconomic model (used for producing forecast about key variables and to conduct policy experiment). The code is written in Julia. You can read about it here. The code is on github.
An excellent tutorial is Programming in Julia. The tutorial is written by Thomas J. Sargent and John Stachurski. Along with being a complete textbook with Julia code for macroeconomics, this also is a very good introduction to Julia
Instructor: Giuseppe Ragusa
Angino: W, 16:00-17:00
Perricone: W 10:30-11:30 (by appointment)
Ragusa: T 11:30-13:00