EC486: Econometric Methods – Autumn Term (AT)

Dr Ragvir Sabharwal
LSE Fellow
Department of Economics
[email protected]

Overview: The AT part of EC486 comprises the first half of a Masters level course in
econometrics. Students enrolled on this course are expected at the very least to have
strong foundations in mathematics (i.e. calculus and linear algebra), probability theory
(e.g. conditional expectations), and statistical inference (e.g. construction of a t-statistic).

It is not compulsory for students to have completed a Bachelors level introductory course
in econometrics, but it is highly desirable. While we will cover regression modelling right
from the start, students without any prior econometrics knowledge can expect to find
EC486 very intense. The reason is that we will proceed quickly through any material that
would typically be covered in an introductory course. In addition, much of the material
covered after the first few weeks is likely to be new for students irrespective of any prior
training in econometrics.

Specifically, we will cover the basics of regression analysis as well as some key extensions
(such as instrumental variable methods and methods for panel data). A full list of topics
is provided further below. Our general focus will be on estimating causal effects - e.g. can
our econometric methods successfully deliver answers to questions such as: What is the
effect of a particular marketing program on sales?

To this extent, our emphasis will be on understanding how to use econometric tools
effectively, rather than on analysing their theoretical properties alone (as is often the
focus in other Masters level econometrics courses). Nevertheless, in order to properly
achieve our goal, we will have to engage with at least some of the mechanics behind the
estimators we are interested in. For this reason, while the level of mathematical
sophistication in this course will not be particularly high, we will frequently rely – as
previously mentioned – on concepts from areas such as multivariate calculus, matrix
algebra, probability theory, and statistical inference.
Format: We will meet each week for a two-hour lecture and a one-hour class. The
lectures will be formal in style with presentations using pre-prepared slide packs. Classes,
on the other hand, will be very informal in style, and are designed to provide a forum for
active, informed discussion of the course material (e.g. case studies, replication of papers,
etc.). The lectures and classes will cover different but complementary material, so it will
be important that you attend both regularly. For class discussions to work effectively, it
will be important for you to read any relevant material beforehand.

Requirements for class: There will be weekly problem sets. Some of these will be
paper-pencil exercises, and others will be computer exercises. You will have to solve the
former by hand and the latter using the statistical software Stata together with data which
will be provided on Moodle. Please be ready to discuss problem sets in class each week
with your class teacher (and each other).

Readings: There is no single ideal textbook for this course. The suggested readings
are from Stock & Watson (S&W), Introduction to Econometrics, which is a standard
undergraduate textbook, but bear in mind that its difficulty level is slightly below that of
this course. Wooldridge (W), Introductory Econometrics, is also a good introductory text.
If something doesn’t make sense to you after the lectures and after reading S&W, then
W is a good resource to which you can refer. (Students who would prefer a slightly more
advanced mathematical treatment of econometrics can refer to Verbeek, A Guide to
Modern Econometrics.)

Formative and summative assessments: Students have the opportunity to take a

mock exam in Week 10 and receive feedback in Week 11. The final marks for this course
will be based on a final exam paper in the Spring Term (65%) and a take-home exam in
the Winter Term (35%).
List of Topics:

1. Introduction
Causal questions and the selection problem
S&W ch. 1

2. Bivariate Relationships in Data

Scatterplots, the conditional expectation function (CEF), linear regression, and basic
properties of the CEF and linear regression
S&W ch. 4, 17

3. Samples and Estimation

Sampling and OLS estimation, the sampling distribution, standard errors and hypothesis
tests, homoskedasticity vs. heteroskedasticity
S &W ch. 5

4. Multivariate Regression
Regression and causality, the conditional independence assumption, short versus long
regression and omitted variables bias, the regression anatomy formula
S&W ch. 6, 18

5. Inference in Multiple Regression

t-tests, F-tests, R-square, tests involving multiple coefficients and joint hypotheses
S &W ch. 7

6. Functional Form in Regression

Nonlinearity in variables, dummy variables, interactions, saturated models
S&W ch. 8

7. Assessing Regression Studies

Internal vs. external validity, omitted variables bias, the role of controls and bad control,
measurement error
S&W ch. 9

8. Regression Details
Weighting regression, binary dependent variables, logit, probit, and marginal effects
S&W ch. 11

9. Panel Data
Introduction to time series S&W ch. 14 & 15
Panel data S&W ch. 10 +W ch. 13 & 14
10. Instrumental Variables, Part I
IV as solution to OVB and simultaneous equations problems, IV language, grouped data
S&W ch. 12

11. Instrumental Variables, Part II

Weak instruments, measurement error and bad control revisited
S&W ch. 12

12. Difference in Difference Estimator and Synthetic Control

Assessing causal questions and the selection problem in non-experimental data. Parallel
trend assumption
Other readings + A&P ch. 5

13. Regression Discontinuity

Assessing causal questions and the selection problem in non-experimental data
Lecture notes + A&P ch. 6

Textbooks:

Required: James H. Stock and Mark W. Watson, Introduction to Econometrics, 3rd or 4th
edition, (S&W)

Optional: Jeffrey M. Wooldridge, Introductory Econometrics, A Modern Approach, 5th

edition, Southwestern, 2013 (W)

Optional: Angrist, J. D.; Pischke, J. S. (2008). Mostly Harmless Econometrics: An

Empiricist's Companion. Princeton University Press (A&P).

Other readings for topic 12:

Abadie, A. (2005). "Semiparametric difference-in-differences estimators". Review of

Economic Studies. 72 (1): 1–19.

Bertrand, M.; Duflo, E.; Mullainathan, S. (2004). "How Much Should We Trust
Differences-in-Differences Estimates?". Quarterly Journal of Economics. 119 (1): 249–
275.

Card, David; Krueger, Alan B. (1994). "Minimum Wages and Employment: A Case Study
of the Fast-Food Industry in New Jersey and Pennsylvania". American Economic
Review. 84 (4): 772–793.