Chris Adolph :: Panel @UW

Time Series and Panel Data for the Social Sciences

UW CSSS/POLS 512

A survey of regression models for time series and time series cross-sectional data. Emphasis on modeling dynamics and panel structures with continuous outcomes, as well as on interpretation and fitting of models. Topics vary and may include trends and seasonality, ARIMA models, lagged dependent variables, distributed lags, cointegration and error correction models, fixed and random effects, panel heteroskedasticity, missing data imputation, and causal inference using panel data.

UW CSSS/POLS 512

Time Series and Panel Data for the Social Sciences

Offered occasionally at the
University of Washington

↤ For my Essex Summer School
course, click Panel Data again.

Syllabus

Readings

Spring 2020

Class meets:
TTh 4:30–5:50 pm
Taught by Zoom

TA:	Inhwan Ko (UW Political Science)

Section meets:
F 1:30–2:50 pm
Taught by Zoom

Lectures PDFs of slides are best viewed in Adobe Acrobat, rather than in your browser.

Topic 1

Course Introduction / Review of Linear Regression and Simulation

Students should also review Topics 1 and 2 from my maximum likelihood course; I will answer questions about these lecture slides during the first week. Students looking to brush up on matrix algebra may want to read through Kevin Quinn's matrix algebra review in pdf format. Finally, in conjunction with these slides I discuss examples that can be found in slides on labor standards in Africa and intimate partner homicide in the US.

Topic 2

Basic Concepts for Time Series: Trends, Lags, and Cycles

The univariate time series simulation function for R mentioned in the lecture is available here; this function allows for deterministic trends, stationary and nonstationary ARMA processes, and additive or multiplicative seasonality. Also available is a simpler but less flexible R script showing simulation and diagnostics of stationary processes using built-in functions.

Topic 3

Modeling Stationary Time Series

Example code and csv data for estimating and interpreting ARMA models in R.

Topic 4

Modeling Nonstationary Time Series

Example code for estimating and interpreting ARIMA and ECM models in R. You will also need this helper function for plotting counterfactual time series using R base graphics.

Topic 5

Basic Concepts for Panel Data

For the curious, the R script used to construct the example plots in the first half of this lecture is here.

Topic 6

Panel Data Models with Many Time Periods

Panel ARIMA and fixed effects models in R: example code and csv data for estimating and interpreting panel models with a large number of periods and a small number of cross-sectional units.

Topic 7

Panel Data Models with Few Time Periods

Panel GMM models (Arellano-Bond/Blundell-Bond) in R: example code, helper functions, and csv data for estimating and interpreting panel models with a small number of periods and a large number of cross-sectional units. On Nickell bias: a script to plot the asymptotic results of Nickell (1981) as well as a helper file and two Monte Carlo scripts (large N / large β and large N / small β) to produce the finite sample results in the lecture slides.

Topic 8

Heteroskedasticity in Panel Data

Topic 9

In-Sample Simulation for Panel Data Models

Example code for simulating in-sample unit-by-unit from a panel data model. Uses the cigarette data and helper functions from Topic 7.

Advanced Topic 1

Missing Data and Multiple Imputation for Panel Data

Advanced Topic 2

Panel Data with Binary Dependent Variables

Student Assignments

Problem Set 1

due by email Tuesday 21 April 2020