Data collected over both units (e.g., individuals, states, countries) and time (e.g., days, months, years) — known as time series cross-sectional data or panel data — are common in the social sciences. By gaining leverage both across units and over time, these data help us answer important questions that would be difficult if we only looked at a single point in time (e.g., cross section) or single unit (e.g., time series): the relationship between growth and democracy, whether or not the resource curse exists, or how economic perceptions shape support for the government. Despite these advantages, panel data often show types of heterogeneity and dynamics that make standard regression approaches inappropriate.
This course is designed to survey some advanced topics in panel data. After a review of panel data fundamentals, we will cover topics such as panel unit root and cointegration tests, panel error correction models, and approaches to modeling dynamics in panel data with a small T.
Fee: Members = $1500; Non-members = $2800
Location: Simon Fraser University -- Vancouver, British Columbia - Vancouver
Date(s): July 2 - July 4
Time: 9:00 AM - 5:00 PM
- Andrew Philips, University of Colorado at Boulder
- Mark Pickup, Simon Fraser University and the University of Oxford
Day 1: Panel data fundamentals, including testing and modeling temporal dependence, and unobserved heterogeneity
Day 2: Panel unit root and cointegration tests, panel error correction models, and the pooled mean group estimators
Day 3: Modeling panel data with small T and overcoming Nickell bias (e.g., GMM, transformed-likelihood estimators)