Macroeconomics 101

Panel Data Econometrics

Summer 2010

Professor: Freddy Cama, WH 222, 646-2126 (Macro dept: 646-3901), email: frojasca@gmail.com
Time and Location: T, Th 11:45 - 1:00, Universidad Nacional del Callao main building 108.
Tuesday 10:30 - 11:30
O¢ ce Hours: Wednesday 12:00 - 1:30
Thursday 1:30 - 3:00.
Textbooks: Arellano, M. (2003) “Panel Data Econometrics”[MA], and Baltagi, B. (2001) “Econometric
Analisys of Panel Data” [BB]

Auxiliary Textbooks: Greene W. “Econometrics” [WG]

Material:
This course is an applied one for macroeconomic research. The goal of this course is to give tools for
research and adapt the student to the working climate on macroeconomics using panel data. The course has
5 lectures and is strongly based on Arellano (2003) and Baltagi (2001), additionally we are including some
relevant papers from prestigious economic and statistical journals; we strongly recommend that students
check out these papers for some extensions or details.

Software: STATA 11 and Eviews6.0

Grading: There will be daily quizzes and a …nal exam. They will count toward the grade as follows.

Pop Quiz 60%

Homeworks 15%
Final 25%.

The …nal exam will be on August 11st between 3:30 p.m to 5:30 p.m.
In addition to these exams there will be dialy quizzes which would be graded the same day. they will
consist in basic calculations, stata codes or some mathematical proofs. After you learn basic techniques of
extracting information at the beginning of the course, you will be ready to have your …rst homework; indeed
after each lecture there will be a homework. These homeworks will barely a¤ect the grade you receive in the
course, but they will be critical for the …nal exam because the exan will cover the homeworks and theoretical
discussions. That is why you need basic mathematical skills to pass the course. Further information about
the exam will be provided later.

There is an introductory session which would cover topics as OLS, inference, maximum likelihood, het-
eroskedasticity and serial correlation. Likewise, there will be an introductory session of STATA, which will
be graded after the introductory session. It is necessary to be clear that we will not issue certi…cates for the
introductory session, just we will give certi…cates for the main course. Assistants Cindybell Gamboa and
Jose Luis Nolazco would help students by email replying questions about the basic commands in STATA for
performing statistical tests and handling time-series and cross-section data.
 Description of the Course
The course will be centered around seven main topics covering the notion of basic panel data, model
especi…cation (one and two ways), heteroskedasticity and serial correlation in the error component, Inference,
dynamic panel data models, nonstationary panels and binary response. The purpose of the …rst chapter (
We called this lecture or chapter 0 or basic) is to do a quickly motivation and introduction of basic panel
data theory and empirics you’ve already taken. This lecture will be 2 hours long, therefore we recommend
you to be familiar with all the subchapters or prepare some questions related to each topic. The daily quizz
will include some easy task usually consisting of basic calculations, stata codes or some mathematical proofs.
All the assignments will be done individually and the homeworks will be done in groups of 3 or 4 students.
Each group will turn in one write-up.
The purpose of written homework in this course is to use the tools we are learning in the course and
develop skills in understanding and communicating results from some particular questions in the …eld of
macroeconomics using panel data. It is not to give you busy work, drill or some . Don’t think of your
homework paper as a certi…cate proving that you have done or tried to do the best on the assignment.
Think of it as an exercise in learning and in reporting what you have learned, likewise you could try to
put on the paper some discussions related to the below topics, all they are welcome in the research …eld,
let’s discuss!. Please be clear on your statements because there is a lot of truth in the statement if you
can’t explain it, you don’t understand it, and it is true as well that (this is mine) if you learnt it withouth
some understanding, pretty sure you will forget it tomorrow, be e¢ cient!. Don’t write to the instructor (who
already knows how to do the problems), but explain your solutions to someone who needs help, perhaps a
classmate who has been sick. Start at the beginning and be clear, logical and complete. The ultimate test
of what you write is this: can someone learn from your paper? Easily? Remember, the reader will see only
what you wrote, not what you meant to say. So it must all be there, and be accurate. Make your paper
reader friendly. We will talk some about this in class.
The …rst chapter talks about how to deal with the error speci…cation; basically we need to consider the
implementation of two estimators: …xed or random e¤ects. The second chapter is designed for giving you
the basic of programming for implementing robust covariance matrix as long as the calculation of some
biases based in typical or standard estimations. The third chapter is aimed to know more about of testing
the speci…cation model; we will discuss the Hausman test and test for individual and time e¤ects. The
fourth chapter wants to introduce the dynamic panels; we will look through the classical Arellano and Bond
estimator as long as some empirical estimations. The …fth chapter we will have examples of cointegration
in large panels . Basically we will look through the International-Monetary-Fund methods for assessing the
dynamic of real exchange rate in low income countries. Basically we present three methods: the …rst one
based on disalignments of the current account which would be helpful when we want to know how much the
real exchange rate would need to appreciate for closing this current account gap, the second one is realted to
the external sustainability using a proxy of net foreign assets, the last methodology is the typical one which
is broadly used in the macroeconomic research …eld. The last chapter aims to use crisis models using logit
panel methods; we would have applications using historical …nancial-crisis data from southamerica.
The purpose of working in groups is twofold. First, by sharing ideas you will be able to learn from each
other, allowing you to clarify what you get out of the lecture and reading. Second is to get you accustomed
to working with other people, a likely situation in your future jobs. The goal for an assignment is to get each
group member to understand the entire assignment. Frequently a major part of an assignment will be to
summarize the various components. In order to do this you will need to understand the entire assignment.
Therefore you should not divide the problems among your group members, but have each person work on
each part and discuss what you come up with. Again the idea of the course is to give you tools, the duration
of the osurse is just a week but we expect you could learn much you can. Good luck!.

2
 Chapters
The course will be centered around several main topics covering the the one and two ways for describing
the error component of panel data models, feasible estimation and inference. We include the dynamic panel
data models as well,

Chapter 0: Introduction
1. Introduction: Panel Data Econometrics
2. Why should we use Panel Data?. ([7], ch 1)

2.1 Within-group e¤ect:

2.2 Random e¤ect

3. Panel data theory

4. Examples and empirics.

Chapter 1: Unobserved heterogeneity in the error component

1. To pool or Not to pool? ([6])
2. The One-Way Error Component Regression model ([7])

2.1 The Within-Group of Fixed E¤ects estimation

2.2 The Random E¤ects estimation

3. The Two-Way Error Component Regression model ([7])

3.1 The Within-Group of Fixed E¤ects estimation

3.2 The Random E¤ects estimation

4. Fixed or random e¤ects?

Chapter 2: Heteroskedasticity and Serial correlation in the error

component
1. Robust Standard Errors for within-group estimators
2. Optimal GLS with heteroskedasticity and Autocorrelation of Unknown Form.
3. Improved GMM and Minimum Distance Estimation under Heteroskedasticity and autocorrelation of
Unknown form.
4. Bias-Adjustment Heteroskedasticity-robust estimator when N or T increase to 1:[22]

Chapter 3: Speci…cation tests

1. Tests for Poolability of the data.
2. Test for Individual and Time E¤ects.
3. Haussman’s Speci…cation Test.

3.1 Size properties of Haussman’s test [10]

3
Chapter 4: Dynamic Panel Data Models
1. Asymptotics of Dynamic Panel Data Estimators. [2]
2. The Arellano and Bond estimator ([7], ch 8), [4]

3. The Arellano and Bover estimator ([7], ch 8), [5]

4. The Ahn and Schmidt estimator ([7], ch 8)
5. Dynamic Macro Panels ([8], ch 8)

5.1 From the theory to a suitable model.

5.2 Panel with homogenous dynamics.
5.3 Dynamic heterogeneity.

6. On Bias, inconsistency and e¢ ciency [16]

Chapter 5: Nonstationary Panels

1. Why is panel cointegration approach useful?
2. Examples of cointegration in macro panels.

2.1 Purchasing power parity [9], [20]

2.2 Income convergence
2.3 Conditional convergence
2.4 Real-Exchange-Rate misalignments

3. Panel Unit Root tests and choices of lag truncation methods

3.1 Homogeneous versus heterogeneous dynamics in panel unit root tests.

3.2 Treatment of cross sectional dependence in early panel unit root test.
3.3 Pedroni (1999; 2004) and kao (1999): Extension to standard unit root tests. [19], [21], [17] and
[15].
3.4 Levin, Lin and Chu panel unit root test.[3]
3.5 Im, Pesaran and Shin panel unit root test. [14]
3.6 Maddala and Wu panel unit root test. [18]
3.7 Reverse null tests for stationarity in panels.

4. Empirical Applications

4.1 Testing for strong PPP in panels

4.2 Bootstrapping for small sample corrections.
4.3 Income convergence tests
4.4 Real-Exchange-Rate equilibrium in low income countries.[23], [13], [1] and [12]

5. Spurious regression in panels.

4
Chapter 6: Binary Response Models for Panel Data and Cluster
Samples
1. Pooled binary-response model ([24], ch 15.8)
2. Unobserved e¤ects models under strictly exogeneity ([24], ch 15.8).
3. Dynamic Unobserved E¤ects Models ([24], ch 15.8).

4. Semiparametric Approaches ([24], ch 15.8).

5. Cluster Samples ([24], ch 15.8)
6. Applications

6.1 Sudden stops ([11])

References
[1] P. K. A. Abiad and J. Lee. Evaluating historical cger assessments: How well have they predicted
subsequent exchange rate movements? IMF Working Paper WP/09/32 (2009).
[2] J. Alvarez and M. Arellano. The time series and cross-section asymptotics of dynamic panel data.
Working Paper N9808 (1998).
[3] c.-F. L. Andrew Levin and C. shang J. Chu. Unit root tests in panel data: asymptotic and
…nite-sample properties. Journal of Econometrics 108 (2002) 1-24 (2002).
[4] M. Arellano and S. Bond. Some tests of speci…cation for panel data: Montecarlo evidence and an
application to employment equations. Review of Economic Studies, 58 277-297 (1991).
[5] M. Arellano and O. Bover. Another look at the instrumental-variable estimation of error-
components models. Journal of Econometrics, 68, 29-51 (1995).
[6] J. G. B. Baltagi and W. Xiong. To pool or not to pool: homogenous versus heterogenous estimators
applied to cigarette demand. Review of Economic and statistics, 82 (1) pp 117-126 (2000).
[7] B. H. Baltagi. “Econometric Analysis of Panel Data”. John Wiley Sons, LTD (2001).
[8] F. Canova. “Methods for Applied Macroeconomic Research”. (2007).
[9] S. W. G. Maddala and P. Liu. “Do Panel Data Rescue Purchasing Power Parity (PPP) Theory?” In
Krishnakumar, J., Ronchetti, E. (Eds.), Panel Data Econometrics: Future Directions, Elsevier (1999).
[10] P. Guggenberger. The impact of a hausman pretest on the size of a hypothesis test: The panel data
case. Journal of Econometrics (2009).
[11] A. I. Guillermo Calvo and L.-F. Mejia. Systemic sudden stops: The relevance of balance-sheet
e¤ects and …nancial integration. national Bureau of Economic Research (2008).
[12] P. Isard. Equilibrium exchange rate: Assessment methodologies. Working Papers (2007).
[13] J. D. O. A. P. Jaewoo Lee, Gian Maria Milessi-Ferreti and L. A. Ricci. Exchange rate
assessments: Cger methodologies. Occassional Paper IMF N 261 (2008).
[14] H. P. K. Im and Y. Shin. Testing for unit roots in heterogeneous panels. Journal of Econometrics
115, pp 53-74 (2003).
[15] C. Kao. Spurious regression and residual-based tests for cointegration in panel data. Journal of
Econometrics, 1999, vol 90 pp 1-44 (1999).

5
[16] J. Kiviet. On bias, inconsistency and e¢ ciency of various estimators in dynamic panel data models.
Journal of Econometrics, 68, 53-78 (1995).
[17] C. Lopez. A panel unit root test with good power in small samples. Econometric Reviews (2009).
[18] G. Maddala and S. Wu. A comparative study of unit root tests with panel data and a new simple
test. Oxford Bulletin of Economics and Statistics 61, pp 631-652 (1999).

[19] P. Pedroni. Critical values for cointegration tests in heterogeneous panels with multiple regressors.
Oxford Bulletin of Economics and Statistics, Vol 61, pp 653-670 (1999).
[20] P. Pedroni. Purchasing power parity test in cointegrated panels. Review of Economics and Statistics
83. pp. 727-731 (2001).

[21] P. Pedroni. Panel cointegration: Asymptotic and …nite sample properties of pooled time series tests,
with an application to the ppp hypothesis. Econometric Theory, Vol 20, pp 597-625 (2004).
[22] J. H. Stock and M. W. Watson. Heteroskedasticity-robust standard errors for …xed e¤ects panel
data regression. Econometrica, Vol 76, N(1) pp 155-174 (2008).
[23] A. P. Thierry Tressel, Lone Engbo Christiansen and L. A. Ricci. External balance in low
income countries. Working Papers (2009).
[24] J. M. Wooldridge. “Econometric Analysis of Cross Section and Panel Data”. The MIT press (2001).