This is the repository for the first half of Econometrics I, taught by Raul Riva. The second part will be taught by Andrea Flores.
- Instructor: Raul Guarini Riva (raul.riva at fgv.br)
- TA: Taric Latif (tariclatif at gmail.com)
- Classes: Tuesdays and Thursdays, 11:00-13:00
First of all, it's a mandatory class. But let's set that aside...
Most of you will be doing empirical work very soon. This class will cover topics that are useful for applied work in several fields like Labor Economics, Development Economics, Urban Economics, Macroeconomics, Finance, Health Economics, etc. After learning the basics in a cross-sectional environment, we will tap into other data structures like Time Series and Panel Data. Both are ubiquitous in applied work nowadays. Mastering the topics in this class will help you hit the ground running with empirical work in the next quarters.
I have four broad learning goals for this first half of the class. Ideally, after Week 5, you will be able to:
- Understand the motivation and implementation of standard non-parametric kernel-based methods;
- Understand the main idea behind bootstrap methods for i.i.d. data and their implementation;
- Deal with time series data, understand concepts like persistence and autocorrelation, estimate ARMA models, create forecasts, etc.;
- Implement GMM estimation under a general heteroskedasticity and autocorrelation structure (very useful for panel methods!);
A very tentative schedule below:
(In construction)
| Week | Tentative Schedule |
|---|---|
| Week 1 | Non-parametric estimation + (i.i.d.) Bootstrap |
| Week 2 | Intro to time dependence + AR and MA Models |
| Week 3 | ARMA Models - covariance structure, estimation, and forecasting |
| Week 4 | LLN and CLTs for time-dependent data + HAC covariance estimation |
| Week 5 | GMM computation + asymptotic theory |
I will prepare slides for each lecture and, ideally, I will make them available to you before class (maybe just before class). Lecture slides will be uploaded to this repository under lectures. The slides will have content mainly from two different books:
- Econometrics, by Bruce Hansen;
- Time Series Analysis, by James Hamilton;
I will highlight, for each lecture, what the relevant chapters are. I stress that reading a textbook is mandatory -- you should not rely only on slides. There is only so much we can cover during class time and through slides.
I try to have all my meetings on Fridays. I will reserve the 9am-10am slot for general office hours, but we can also schedule another time by email if that time does not work for you. I am always happy to meet to talk about the material.
Taric will host weekly sessions with you as well. He has full discretion regarding how to use that time. If you want a specific type of TA session, please negotiate it directly with him. My experience as a TA tells me that the best use of TA session time is going through common mistakes on the problem sets, talking about specific issues that came up during your own study time, and doing extra problems.
It is good etiquette to email the TA beforehand and let him know about your specific doubts beforehand. I strongly suggest you email Taric about anything that is unclear regarding the class material and you'd like to see covered during the TA session.
There is good and bad news here:
- Good news: attendance is not mandatory at all. In any case, I like to believe I will be able to add value to your learning experience. Hence, I suggest you come. But there will be absolutely no retaliation if you choose not to -- I mean it.
- Bad news: if you decide to come, you have to be on time. There is a 10-minute grace period. I reserve the right to deny your entry after that, unless you have an extremely good reason to be late, or you let me know in advance that you will be late.
- There will be three problem sets and one exam for this first half of the class;
- These four pieces of evaluation will be graded out of 10. Your grade for the first half of the class will be a weighted average of these three grades out of 10. One third of your grade will come from problem sets, and two thirds will come from the exam;
- The exam will be in-person and individual. I will make sure you have enough time for the exam. I do not want you to be time-constrained;
- Every student is allowed to bring one A4-sized sheet with notes, on both sides. You cannot share notes during the exam, but you can bring anything written on your sheet.
Aside from learning the material, we will use problem sets to get you up to speed with GitHub. It is a very useful tool for collaboration, version control, and sharing code. You will use it a lot in your academic and professional life. If you haven't created an account yet, please do so here.
Important stuff:
- All problem sets are posted under
problem_sets. The necessary data files and instructions are also there. - You should form groups of up to 3 people. You should keep the same group throughout the first half of the class. Andrea will let divorce and remarriage happen in the second half.
- We will use GitHub Classroom to handle the problem sets. For each problem set, there is a link for the associated assignment through GitHub Classroom. Check the instructions stored under the
problem_setsfolder. - These problem sets will have both theoretical questions and empirical questions. You need to submit a PDF report answering both the theoretical part and the coding part. You do not need to include any code in this report.
- You can use any reasonable language to solve the empirical parts, as long as it does not come with pre-made routines. For example: Python, R, Matlab, Julia, even C or Fortran if you are a very hardcore person. No, you cannot use Stata and you cannot use pre-packaged routines, unless it is something super standard such as numerical optimization. For example, if you are solving something with the GMM estimator, I do not want you using the
gmmpackage in R. Just be reasonable. - Submit code and your PDF reports as separate files. Add as many figures and tables to your report as you deem necessary to answer the empirical part.
This is really important:
- This is the first time I will be teaching this class. The end goal is making your learning path easier;
- I need ongoing feedback, and you do not need to be shy. If there is something I do that makes the lectures worse, please tell me. If there is anything you like a lot, please tell me.
- Feel free to email me, and talk to me in person. Feedback, even if super negative, will not impact your grade, obviously.