**EC 228:
Econometric Methods (Fall 2017)**

02: Stokes Hall 295S: T Th (9 – 10:15)

03: Stokes Hall 295S: T Th (10:30 – noon)

Christopher Maxwell** **Maloney
Hall, 337

maxwellc@bc.edu Office Hrs: TTh 12:30 – 2

http://www.cmaxxsports.com & by arrangement

This is an introductory course in the use of econometric
methods, with an emphasis on empirical applications and cross-sectional
analysis. Our focus will be on learning ** to
do** econometrics, not just learning econometrics.

While the course will cover the development of the formal tools of econometric analysis (simple and multiple regression analysis, estimation, inference, qualitative variables, and so forth), we will also spend quite a bit of time on empirical methods (posing questions, building datasets, running regressions, supplementing datasets, running more regressions, etc etc … until we can confidently say something about the questions at hand). As such, an important part of the course will be a set of empirical Exercises and an empirical research project in which students will be building their own datasets and applying the various econometric methods developed in the course.

**Peer tutors**: Margaret Andersen & Kristian Singh @
Connors Family Learning Center

**Prerequisites**:
An introductory statistics course such as EC 151. No exceptions. I will also assume that students have an
understanding of basic *Excel* (which will be used at times to assemble
datasets and verify calculations) and basic calculus. I do not assume that anyone has previous
exposure to *Stata*, the computer language that we will be using to run
regressions (but you will be better for it if you have seen Stata before). (See below for more about Stata.)

**Course reference text**:

·
Jeffrey M. Wooldridge, *Introductory
Econometrics: A Modern Approach*, Cengage Learning.

I have deliberately not listed the edition; if you decide to
purchase the text, feel free to buy the 2^{nd} (2002), 3^{rd}
(2006), 4^{th} (2008), 5^{th} (2012), or 6^{th} (2015)
editions. A copy of the text will be
placed on reserve at the O’Neill Library.

**Important Note: ****I do not
follow the text closely. I will be
distributing lecture notes for most of the material covered in class.**

**Some additional texts:
**There is no need to purchase any of these (most are available at
O’Neill). I list them just because
sometimes it is useful to see a different presentation of the material.

·
Angrist, Joshua D. and Jörn-Steffen Pischke, *Mastering 'Metrics: The Path from Cause to
Effect*, 2014.

·
Stock, James H. and Mark W. Watson, *Introduction
to Econometrics*, 3^{rd} ed., Addison-Wesley, 2014.

·
Studenmund, A.H., *Using Econometrics: A Practical
Guide*, 6^{th} ed., Pearson, 2010.

·
Ashley, Richard, *Fundamentals of Applied
Econometrics*, Wiley, 2012.

·
Kennedy, Peter, *A Guide to Econometrics*,
6^{th} ed., Wiley-Blackwell, 2008.

**Grading** (Exams: 75%; Research project: 10%; Labs: 7.5%;
Exercises: 7.5%):

· Three exams: Two mid-terms and one final exam

§
MT #1 (20%;
mid-October… Oct 14^{th}?):
Ordinary Least Squares - SLR and MLR models

§
MT #2 (20%;
Dec. 7^{th} (the last class):
Estimation and Inference + additional topics

§
Final (35%; 9:00 AM, Mon Dec 18^{th}
and/or 12:30 PM Tues. Dec 19^{th} ):
covers full semester

All exam grades are curved. In each case, I am hoping to have both sections of the course sit at one exam. Stay tuned.

Exams are *open
cheat-sheet* (two cheat sheets allowed for the final exam; one sheet for
each mid-term).

·
__Final Exam: Optional.__ The Final Exam is optional. If you decide not to take the final, the
weights on your mid-term exam grades are scaled up so that each mid-term exam
counts for 37.5% of your course grade.
You must commit to taking the final exam at the time you pick up that
exam. To allow you to make a fully
informed decision about whether or not to take the final exam, conditional
course grades, assuming that you are not taking the final exam, will be posted
to Canvas by the end of the day, Tuesday Dec 12^{th}.

There are no make-up exams in this course. If you miss either mid-term exam, then you must take the final exam (exam weights will be adjusted proportionately).

·
Research Project (10%): Replicate and improve upon an econometric
analysis that has been published in an academic journal. The deliverable is a *PowerPoint* presentation, due Tues Nov. 21^{st}, just prior
to Thanksgiving Break. This will be a
team assignment, with three students per team.
I will assign teams. Grades will
be given for both individual as well as collective performance. More details below.

· EC 228 Labs (7.5%): EC 228 course-wide labs, focused on using Stata in empirical/econometric analysis. Lab scores will be curved.

·
Exercises (7.5%): There will be five Exercises over the course
of the semester. These will focus on
empirical applications of the tools developed in the course. Feel free to work together on these, but
please submit your own write-up (unless it’s a *team* assignment … more on
this below). Course grades for Exercises
will be curved. More details below.

**Canvas: ** Boston
College has in the past few years migrated away from BlackboardVista to a new
course management system called ** Canvas**. That migration has not gone smoothly for my
courses. Until Canvas offers

**Accommodations**:
If you are a student with a documented disability seeking reasonable
accommodations in this course, please contact Kathy Duggan (x2-8093;
dugganka@bc.edu) at the

**Academic Integrity:**
You will be held to Boston College’s standards of academic
integrity. If you have any questions as
to what that means, please go to http://www.bc.edu/offices/stserv/academic/integrity.html.

**Research
Project** (10% of total grade): This is an empirical research project, which
will kick off with team assignments after the first mid-term exam. There
are two phases to the project: you will first replicate an existing published
piece of econometric analysis (of your choosing), [1]
and then improve that analysis in some way (by adding more data, changing the
specification of the model, changing the estimation technique, and so
forth). Topics should showcase
interesting econometric analysis, and need not be restricted to topics in
Economics.

Your deliverable will
be a *PowerPoint* presentation (or the
equivalent). It should be concise and to
the point; ** shorter is always better**. I will say more about the format of the
deliverable when teams are assigned.
Hardcopy presentations are due Tues Nov 21

·
**Phase I
: Replication**

Replicate
both the summary statistics presented in the paper (to show that you have
indeed replicated the construction of the dataset) as well as ** at least
one set** of regression results of interest.

·
**Phase
II: Improvement**

** Your
turn!** ... your improvement to the published
analysis. This should be a lot of fun…
but again, it will not go quickly or smoothly, so budget your time accordingly.

Your
Phase I presentation should discuss your data sources and how your dataset was
constructed; credit will reflect in part the level of difficulty.[2] In some cases you may be able to obtain data
from the original authors, which obviously greatly simplifies the replication
phase. You can do that if you want, but
since building datasets is hard work, you won’t get as much credit for your
efforts as you would had you built the dataset yourself.

In the Phase II
portion of your presentation, you should brag about your outstanding
improvements to the analysis, and discuss as well aspects of your analysis that
did not pan out as expected.

I will
assign the research project teams, which will likely have three members
each. Students’ grades will reflect both
their individual performance as well as the quality of the final team
product. *Peer evaluation* forms will be distributed at the end of the
semester, so that team members can assess each other’s performance.

*Empirical work is slow
going. Be sure to leave yourself enough
time to complete the assignment to your satisfaction.*

**The Labs:** There will be six or so one-hour sections of
econometrics labs, taught by excellent BC Econ graduate student
instructors. The lab sections will focus
on the use of Stata in empirical analysis generally, and econometric analysis
specifically. Lab scores, which
will be curved, count towards 7.5% of your course grade.

**Exercises: **There will be five
empirical exercises, counting towards 7.5% of your course grade. These will typically be** **team assignments
(with 2-3 students per team) depending on how much work is required, and are
typically graded on a 10 point scale. I will
assign teams, which will differ from Exercise to Exercise. Teams should submit just one Answer Set. Final grades on Exercises are curved.

In some cases, the
Exercises are designed to give you practice with the techniques and tools we
have developed in class… other times,
they are designed to introduce you to new material, which we have not yet
covered in class. These will take some
time to complete, so please do not leave them until the last moment.

**Course Topics** [Wooldridge 4^{th} and 5^{th}
edition chapter numbers are in square brackets (there are just a few changes
with the 6^{th} edition)]

**A: Introduction to
Econometrics and Empirical Research**

1. The Nature of econometrics and economic data [Chapter 1]

- Estimating the relationship between x and y; causality v. correlation; data types; economic v. statistical significance; robust analysis (how many regressions did you run?); art v. science

**B: Introduction to
Ordinary Least Squares (OLS)**

2. Simple Linear Regression (SLR) Models [portions of Chapter 2.1-2.3]

*SLR analytics*(single explanatory variable):*in the beginning*(SLR.1: the data generation process); residuals and sum squared residuals (SSR); OLS as minimization of SSRs (FOCs and SOCs); sample statistics (sample means, variances, standard deviations, covariances, and correlations); Sample Regression Function (SRF) and predictions; economic significance (elasticity)*SLR assessment*: Sum Squared Explained (SSE) and Sum Squared Total (SST); SST = SSE + SSR (w/ constant term in the model); goodness of fit (Coefficient of Determination, R^{2}), Mean Squared Error (MSE) and Root MSE (RMSE)

3. Multiple Linear Regression (MLR) Models [portions of Chapter 3.1-3.4]

*MLR analytics*(adding, and subtracting, explanatory variables): SLR analysis continued; interpreting coefficients I –*ceteris paribus*(partial effects and the SRF); the*collinearity*regression; multicollinearity; interpreting coefficients II –*what’s new*; omitted variable bias (endogeneity)*MLR assessment*: SLR assessment continued; adjusted R^{2}

**Mid-Term Exam #1 about
here**

*Sidetrip: Getting started with that research paper*

4. Carrying out an empirical project [Chapter 19]

· The scientific method: Testing hypotheses; collecting data; running regressions; doing it all again; and again; until… conclusions

**C: Review of
Probability and Statistics**

5. Review of foundational basics: Probability and statistics [Appendices B & C] (some of this will have been covered in Part B. above)

· Probability: Probability distributions (probability density functions (pdfs); measures of central tendencies (expectations) and variability (variances); joint and conditional distributions; measures of association (covariance and correlation); conditional expectations and variances

· Statistics (Estimation and Inference): Populations; point estimators; sampling distributions; sample means, variances, covariances and correlations; unbiasedness; interval estimators; confidence intervals; hypothesis testing

§ Focus on estimation of the population mean, LUEs (Linear Unbiased Estimators), and BLUEs (Best Linear Unbiased Estimators)

**D: Estimation and
Inference in Regression Analysis**

6.
Estimation
with SLR and MLR models [remainders of Chapters 2 and 3]

- Estimation in SLR models: Gauss-Markov assumptions (SLR.1 – SLR.5); Population Regression Function (PRF); conditional means; means, variances, standard deviations and standard errors of OLS estimators (intercepts and slopes); unbiasedness (OLS coefficients; MSE); LUEs; homoskedasticity; BLUE: The Gauss-Markov Theorem
- Estimation with MLR model (what changes? … not much!): … now MLR.1-MLR.5; Multicollinearity and Variance Inflation Factors (VIFs); n-k-1

7.
Inference
in SLR and MLR analysis [Chapter 4, portions]

- Inference with SLR models: Add SLR.6 to the mix; normally distributed errors; variances, standard deviations and standard errors; t statistics; t-tests (Null hypotheses); p values; confidence intervals; hypothesis tests; economic v. statistical significance (elasticities v. p-values)
- Inference with MLR models (what changes? … not much!): Now MLR.6; n-k-1; F-tests of linear restrictions; reported F stats and p values; Chow tests

**E: SLR/MLR
Analysis – Further issues etc.**

8. Further
Issues I [portions of Chapters 6-8]

·
Heteroskedasticity [Chapter 8]: Issues (OLS standard errors no longer
correct; LUE but not BLUE); White-corrected standard errors (*robust *inference); working towards BLUE
(weighted least squares… but where do those weights come from?)

- Binary and categorical independent variables (Dummies) [Chapter 7]; fixed and interaction effects (intercept and slope dummies); average residuals (lazy economists); percentile dummies and more functional forms
- Binary dependent variables [Chapter 7]: linear probability models
- Functional forms (polynomials; logs; splines; etc) [Chapter 6]

9.
Selected Topics [Handouts + portions of Chapter
15]

*Differences-in-Differences*: Deflategate; NBA Referee Own-Race Bias*Regression Discontinuity Designs*: Highway Fatalities & Daylight Savings Time- Endogeneity and Instrumental Variables [Chapter 15]: The Oregon Health Insurance Experiment (Medicaid)

10. Further
Issues II [portions of Chapters 5 & 6]

· OLS asymptotics [Chapter 5]: Large sample properties; consistency (convergence in distribution)

·
Scaling effects and *beta* regressions [Chapter 6]

·
Forecasting and *prediction* intervals [Chapter 6]

11. Limited
dependent variables [Chapter 17]:
Maximum Likelihood Estimation (MLE); logit and probit models; censored
and truncated regression models

12. More
about specification and data problems [Chapter 9]

· Misspecified models; proxy variables; measurement error in dependent and independent variables; missing data; outliers; non-random samples

**Calendar**

I can’t guarantee how exactly we will progress through the material. But based on previous semesters, here’s a likely calendar (#s refer to weeks in the semester):

1. Syllabus; Introduction to the course; Econometrics examples

*Intro to OLS: SLR and MLR models*

2. *SLR Analytics*; Exercise #1 distributed
(due wk #4)

3. *SLR Assessment*

4. Finish Intro to OLS/SLR; Exercise #2 distributed (due wk #7)

5. *MLR Analytics* (focus on differences from
SLR analytics)

6. *MLR Assessment* (again, focus on
differences viz. SLR assessment); Distribute practice questions for MT #1

7. Finish
Intro to OLS/MLR; Mid Term #1 (Intro to OLS; Thurs Oct 14^{th} ?)

*Estimation and Inference*

8. Kickoff Research Papers; Review of Probability and Statistics; Exercise #3 distributed (due wk #9)

9. Return MT #1s; Intro to MLR Inference (reading regression results); SLR and MLR models: Estimation and inference; Exercise #4 distributed (due wk #11)

10. SLR and MLR models: Estimation and inference cont’d; Research paper tips

*Further topics/issues*

11. Consistency; Heteroskedasticity; Dummy variables; Exercise #5 distributed (due wk #15)

12. More about dummy variables; Functional forms

13. *Thanksgiving Week*: Research papers dues on Tuesday

14. Selected topics; Distribute practice questions for MT #2

15. Semester
wrap-up and review; Mid Term #2 (Thurs Dec 7^{th} (last class); emphasis on material since MT #1)

**16. **

**Stata @ Boston College**

There are a large number of statistical software packages that
you can use to do econometric analysis.
We will use Stata, one of the more popular packages and the package that
receives the most support at

I will be providing more details as the semester develops, but for now: Stata is available to BC students through the “application server”, which can be accessed at https://apps.bc.edu … once CitrixReceiver has been installed on your computer. (If you are not connected to the BC network, you’ll need to use VPN to access the apps server.) To learn how to access Stata through the apps server, go to http://www.bc.edu/offices/help/teaching/app_server.html .[3]

Alternatively, and to avoid traffic jams with Citrix and the
apps server, you may want to purchase a six-month *Stata IC* license for $75 (sorry, but *small* Stata will not sufficient for Ec 228). For details, go to: http://www.stata.com/order/new/edu/gradplans/student-pricing/
.

We will devote some time to learning how to use Stata to build datasets and run regressions. You will discover that building datasets is long, hard, tedious and unrewarding work… and running regressions is relatively quick, easy and fun.

As the semester progresses, you may find the following resources of interest:

Encountering Stata questions/issues/features?… just *Google*
it (always include “UCLA”). And here are
a few sites that might be helpful (the pdfs are posted to the course website):

· http://fmwww.bc.edu/GStat/docs/StataIntro.pdf

· http://www.stat.ucla.edu/labs/pdflabs/started.pdf

· http://dss.princeton.edu/training/StataTutorial.pdf

· http://isites.harvard.edu/fs/docs/icb.topic515962.files/GettingStartedWithStata.2109.pdf

· http://www.ats.ucla.edu/stat/stata

Examples and datasets (**bcuse** may be helpful here… I’ll
explain in class):

· http://fmwww.bc.edu/gstat/examples/wooldridge/wooldridge.html

· http://fmwww.bc.edu/ec-p/data/wooldridge/datasets.list.html

[1] Published here means published in an academic journal (so no unpublished senior theses, web blogs, or the like).

[2] If you
want a sense of *degree of difficulty*, just ask.

[3] And to learn about VPN, go to http://www.bc.edu/offices/help/getstarted/network/vpn.html .