Panel Data Analysis

This course introduces various econometric models and techniques that can be applied to panel data. It also emphasises the connections between these methods and causal inference, which is the primary goal of social science research.

Day 1
We begin with an overview of causal inference in the context of observational studies. We quickly survey the idea of causality and relevant concepts such as average treatment effect (ATE). Importantly, we will draw connections between these concepts and regression analysis. We also quickly review the set of assumptions on which OLS regression hinges.

In the second half of the day, we look at simple panel data methods. Specifically, we make a distinction between ‘independently pooled across section’ data and ‘panel’ data. The latter is the focus of this course. We’ll see that panel data differs in some important respects from an independently pooled cross-section in that a panel consists of the same entities (individuals, firms, countries or whatever) across time. We then study two basic panel data methods: two-period panel data analysis and first differencing.

Day 2
We focus on slightly more advanced methods for estimating the unobserved effects in the context of panel data analysis. We introduce the fixed effects estimator, which, like first differencing, uses a transformation to remove the unobserved effect prior to estimation. As a result, any time-constant explanatory variables are removed in the process. In contrast, we introduce the random effect estimator, which looks attractive when we think the unobserved effect is uncorrelated with all the explanatory variables. For example, when we have good knowledge about the factors affecting the dependent variable and have controlled for these factors in the equation, random effect estimator can sometimes be the preferred strategy. With these foundations, we then look into a relatively new correlated random effects approach, which provides a synthesis of fixed effects and random effects methods, and has been shown to be practically very useful. In the meantime, you’ll also learn that the usefulness of these methods critically hinges on the assumptions made with respect to the error term and the relationship between the regressors and the error term.

Day 3
We begin with a lab session during which we put into practice the methods introduced in the previous two days. We carry out these analyses in Stata and learn how to interpret the results. If you have brought your own data, this is a good opportunity to carry out the analysis for your current research projects. In the second session, we return to the classroom to learn more about the research designs and methods to which panel data can be applied. We start to relax the assumptions made in the previous discussion and learn methods designed to deal with various violations of these assumptions. The first is instrumental variable (IV) method, which deals with violations to the strict exogeneity assumption. Again, I emphasise the connection between this method and causal inference.

Day 4
I introduce several more advanced panel data methods that address further violations of the standard assumptions. Since panel data contains repeated observations of the same entity over time, we may have different error variances for different panels as well as correlation of the error terms across panels and/or time. To deal with these potential challenges, we study methods such as clustering and robust estimation, panel-corrected standard error (PCSE) estimates and dynamic panel methods (Arellano-Bond estimator). For these more advanced methods, we will not look into the technical details but rely more on an intuitive understanding of the challenges these methods address.

Day 5
In the first session, we return to the lab and put these more advanced methods into practice. We learn how to carry out panel IV analysis, obtain the PCSE and Arellano-Bond estimators and calculate cluster robust standard errors. Again, this is a good opportunity for you to carry out the analyses using your own datasets and check the robustness of the results across various model specifications. The last session is a seminar. Students who want extra credits can present their research or research proposals that use panel data methods, and receive feedback from the Instructor and fellow participants.

This course is a general survey of panel data methods. It attempts to achieve a balance between theories and practical/implementation. If you are interested only in implementing these methods, you may find this course less beneficial.

The limited time available means we won’t have the luxury of going deeply into methodological niches, such as presenting all the techniques for tackling autocorrelation.

This course builds on ordinary least square (OLS) regression and extends it to data with a time-series-cross-sectional structure.

You need

• familiarity with basic statistical concepts such as sample mean and sample variance as well as their properties
• basic knowledge of regression analysis (up to multiple regression)
• basic skills in Stata.

If you do not have this knowledge, I strongly encourage you to take one of the following, either before or concurrently with this course:

Introduction to STATA

Introduction to Inferential Statistics: What you need to know before you take regression

Multiple Regression Analysis: Estimation, Diagnostics, and Modelling

This course is designed for students who are somewhat familiar with the content covered up to Chapter 9 of Wooldridge’s 2016 textbook (see below) but do not have a good knowledge of the materials beyond that. This is an easy way to check if this is the right course for you.