17 September – 8 October 2021
Early bird registration deadline: 26 July 2021
Registration deadline: 16 September 2021
The course will cover the basics of panel data analysis and some more advanced extensions, focusing mainly on microeconometric settings with a large number of cross-sectional observations. The statistical package Stata will be used to illustrate all of the methods, including applications to the banking sector.
The common estimators – random effects, fixed effects, and first differencing will be discussed, with emphasis on robust inference and specification tests. During the course modules, the instructor will present in details:
- The extensions that allow heterogeneous slopes and trends
- Instrumental variables methods
- Estimation of dynamic models also will be covered.
- Fixed effects estimation and inference with a large number of time periods, applicable to more aggregated data.
- The problem of unbalanced panels and how to test for nonrandom sample selection
The panel data methods will be applied to estimate bank cost functions as well as estimating the effect of foreign ownership on market power, as in Delis, Kokas, and Ongena (2016, JMCB). Also, difference-in-differences methods will be illustrated by studying the effects of changes in banking regulations, such as the European Bank and Recovery Resolution Directive, on credit default swaps.
- Random Effects, Fixed Effects, First Differencing
- Robust Inference and Robust Specification Tests
- Instrumental Variables
- Heterogeneous Trend and Slope Models
- Dynamic Models
- Large-T Panels
- Correlated random effects approaches to panel data
- Unbalanced panels and detecting sample selection problems
- Nonlinear Panel Data Models
- Bank Cost Functions
- Effect of Foreign Ownership on Market Power
- Difference-in-Differences Methods
What you will learn
- You will learn to use Stata to estimate basic linear panel data models by random effects, fixed effects, first differencing, and instrumental variables versions of these.
- You will learn how to use robust specification tests to choose among estimation methods.
- You will learn what happens when additional heterogeneity is introduced into the basic model.
- You will be introduced to large T panel data sets.
- You will understand the consequences of unbalanced panel data sets.
- You will understand the tradeoffs between pooled and joint estimation methods for nonlinear models
- You will learn about the effects of the Bank Recovery and Resolution Directive (BRRD) on Credit Default Swap (CDS) spreads.
- Introduction and Overview of Panel Data Methods
- The Basic Linear Model and Assumptions
- The Common Estimators: Pooled OLS (POLS), Random Effects (RE), Fixed Effects (FE), and First Differencing (FD)
- Comparing POLS with RE and FE. Correlated Random Effects
- Choosing Between RE and FE
- Choosing Between FE and FD
- Combining Fixed Effects and Instrumental Variables (FEIV)
- Specification Tests for FEIV. Comparison with Random Effects IV (REIV)
- Combining First Differencing and Instrumental Variables
- Estimation Under Sequential Exogeneity
- Heterogenous Trends and Slopes
- Difference-in-Differences Designs with Panel Data
- Considerations with Unbalanced Panels
- Panels with a Relatively Large Number of Time Periods
- Probit and Logit Models for Binary and Fractional Reponses
- Exponential Models for Count and Other Nonnegative Outcomes
Format of the course
This course consists in a balanced mix between self-paced material and live online activities. This format will bring to your own devices the course material and interactions with instructors, teaching assistants and other participants.
During the first and the second week in the course, you will be guided in the course material via video lectures and live classes with the course instructor. In addition, you will apply the theory into practice through homework assignments.
The third week of the course will consist in lab sessions, in which the instructor and teaching associate will guide you in practical exercises in STATA, during a set of live online events.
The course format will give ample room to question times and collaboration. You will benefit from close guidance and throughout the whole course, with multiple occasions for individual feedback and interactions with the instructor and teaching assistants.
The course will require 20 hours to be completed
17 September: Opening of the course
Access to all course modules
17 September – 4 October: Self-paced progression throughout lectures and homework exercises
(total time required: 12 hours)
21 September (3-4:30 PM CET): Optional: ‘Brush up’ session
4 – 8 October: Live online sessions
(total time required: 8 hours)
24 September (3-5 PM CET):
First live class: recap of modules 1-6
1 October (3-5 PM CET):
Second live class: recap of modules 7-16 + Q&A (3-5 PM CET)
5 October (3-5 PM CET)
Third live class:
- Lab 1
- Office hours (5-6.30 PM)
7 October (3-5 PM CET)
Fourth live class:
- Lab 2
- Office hours (5-6.30 PM)
Closing of the course
Meet the instructor
Jeffrey Wooldridge is University Distinguished Professor of Economics at Michigan State University. He is a Fellow of the Econometric Society and the Journal of Econometrics. He is the author of two textbooks in econometrics: Introductory Econometrics: A Modern Approach, 6e; and Econometric Analysis of Cross Section and Panel Data, 2e. He has served on several editorial boards, including as editor for the Journal of Business and Economic Statistics as co-editor of Economics Letters. While an assistant professor at MIT, he won the graduate teacher-of-the-year award three times. He has given dozens of econometrics short courses around the world.
MA degree in business, economics, or statistics. Or, a BA in economics with training in linear algebra and multivariable calculus.
A knowledge and understanding of ordinary least squares, generalized least squares, instrumental variables (using matrix algebra), some asymptotic theory is required to follow the course.
Participants are required to have the STATA software installed on their own devices.
FeesEarly bird fees apply only for private sector and public authorities until 26 July
EARLY BIRD fee– Public Authorities (Standard fee 1.100€) – e.g. National Competent Authorities, Central Banks and European Institutions.
EARLY BIRD fee– Private Sector (Standard fee 1.200€)
800€ – Full-Time Professors, PhD Students, Research Associates. Please submit a certificate attesting your status of Professor, PhD Student or Research Associate to email@example.com before registering. FBF secretariat will provide you with a code to register. *Seats for academics are limited.
Please note that the payment must be settled one week before the start of the course.
A certificate of attendance will be provided to all participants after the course.
- Registered participants can cancel their participation and ask for a refund until three weeks (15 February) before the start of the course, by sending an email to the FBF secretariat. Past that date, refund requests will no longer be accepted by the Secretariat (unless for compelling and motivated reasons).
- In case a course is cancelled, registered participants can request a total refund or request a voucher to attend another FBF course.
- In case a course is postponed to another date, registered participants have the following three options: request a voucher to attend another FBF course, transfer their registration to a colleague or request a refund.
For more details, please contact firstname.lastname@example.org
- Delis, M. D., Kokas, S. and Ongena, S. (2016) Foreign ownership and market power in banking: Evidence from a world sample. Journal of Money, Credit and Banking, 48(2-3), pp. 449-483.
- Pancotto, Livia and ap Gwilym, Owain and Williams, Jonathan (2019) The European Bank Recovery and Resolution Directive: a market assessment. Journal of Financial Stability, 44. 100689. ISSN 1572-3089
Panel Data for Banking Sector Analysts 2021
Send download link to: