Longitudinal/Panel and Time Series Data Analysis using Stata

INTRODUCTION

Longitudinal or panel data are multi-dimensional data involving measurements over time. Such data are analyzed using dynamic model. Dynamic models have become increasingly popular due to their ability to take into account both short and long term effects and unobserved heterogeneity between economic agents in the estimation of the parameter estimates. Stata is very specialized in handling dynamic data.

This training course provides an overview of existing dynamic data analysis techniques. Participants will be taken through a series of illustrative examples, with a theoretical and applied overview. Recent issues in dynamic panel data analysis will also be covered. The course concludes by addressing the issues of; i) non-stationary in long panels, where the time series (as opposed to cross-sectional) characteristic of the data dominates; and ii) cointegration. The training will pay particular attention (using a combination of both official Stata and user written dynamic panel data analysis commands) to: i) evaluating which specific econometric methodology/specification is more appropriate for the analysis in hand; ii) selection of the appropriate instruments; iii) rigorous post estimation diagnostic/specification testing; and iv) the problems of inference resulted from weak-instrument bias, instrument-proliferation bias and small-sample bias. Special attention will also be given to the interpretation and presentation of results.

DURATION

5 Days

COURSE OBJECTIVES

By the end of this training, participants will become knowledgeable in the following:

• Usefulness and problems with Panel Data
• Opportunities and challenges of panel data.
• Linear models data analysis with dynamic data
• Logistic regression models with dynamic data
• Count data models with dynamic data
• Linear structural equation models with dynamic data

COURSE OUTLINE

Module 1: Introduction

Introduction to Panel Data

• Why Are Panel Data Desirable?

• Problems with Panel Data

• Examples of Time-varying and time-invariant variables

Opportunities and challenges of panel data.

• Data requirements

• Control for unobservables

• Determining causal order

• Problem of dependence

•Software considerations

Module 2: Linear models

•Robust standard errors

•Generalized estimating equations

•Random effects models

•Fixed effects models

•Between-within models

Module 3: Logistic regression models

•Robust standard errors

• GEE

•Subject-specific vs. population averaged methods

•Random effects models

•Fixed effects models

•Between-within models

Module 4: Count data models

•Poisson models

•Negative binomial models

Module 5: Linear structural equation models

•Fixed and random effects in the SEM context

•Models for reciprocal causation with lagged effects