Tuesday and Thursday, 2:30 – 3:45. Social Science 6104
Bruce Hansen's Webpage
Department of Economics
This is new course. The focus will be nonparametric
methods in econometrics. We will be following the new textbook Nonparametric Econometrics: Theory and Practice
by Qi Li and Jeffrey Racine. We will also devote a third of the course to the study of model selection,
shrinkage and averaging.
Coursework will consist of weekly problem sets, largely taken from the exercises in Li and Racine's book.
These exercises are both theoretical and applied, and will count for half of your course grade.
There will be a Take-Home Exam at the end of semester.
This will count for the other half of your course grade.
Topics and Lecture Notes:
Nadaraya-Watson and Local Linear Regression
Conditional Distribution Estimation
Conditional Density Estimation
Conditional Quantile Estimation
Semiparametric Methods, Partially Linear Regression
Single Index Models
Nearest Neighbor Methods
Nonparametric Time Series
Some other textbooks, which are useful for reference include:
Silverman, B. W. (1986) Density Estimation for Statistics and Data Analysis
Scott, D. W. (1992) Multivariate Density Estimation
Fan, J. and I. Gijbels (1996) Local Polynomial Modelling and its Applications
Fan, J., and Q. Yao (2003) Nonlinear Time Series: Nonparametric and Parametric Methods
Pagan, A., and A. Ullah (1999) Nonparametric Econometrics
Hardle, W. (1990) Applied Nonparametric Regression
Ruppert, Wand and Carroll (2003) Semiparametric Regression
Hardle, Muller, Sperlich and Werwatz (2004) Nonparametric and Semiparametric Models
Yatchew (2003) Semiparametric Regression for the Applied Econometrician
Koenker (2005) Quantile Regression
Bosq (1998) Nonparametric Statistics for Stochastic Processes
The books by Silverman and Hardle are classics.
Pagan-Ullah is the first econometrics book on nonparametrics, and in this sense is similar to Li-Racine.
Fan-Gijbels is a thorough treatment of local linear and local polynomial methods.
Bosq is a theoretical treatment of kernel methods for dependent data.
Fao-Yao is a summary of their extensive contributions to time-series nonparametrics