The utility of empirical learning in economics is conditioned by theory and method. Historically, these threads have been pursued with considerable separation. With the emergence of micro-level data, theory of agent level behavior and choices has been reconsidered to direct empirical study of data characterizing discrete and discontinuous choice by households and producers.  From a methods perspective,  as micro-level data has increasingly become available, the limits of models that predict only conditional means or averages have been increasingly recognized.  What we want is to explain is individual variation!  Why do people make different choices? We have the data, but how can we use it to learn about the determinants of heterogeneity ~ differences across agents in the economy?

Standard linear, parametric estimation methods are seriously limited in their ability to explain variation in behavior in such micro-level data sets. In many cases, resulting estimates are biased, inefficient, and difficult to interpret conditional mean responses. While these may be of interest to some, they are based on equally-weighted parameter estimation and provide little guidance with respect features of processes that may drive performance of systems of interest. In place of such aggregate or average methods, a variety of new methods have been developed.  This course is paired with 5956 Microeconometrics - Theory.  In this course, we will examine applications of methods presented in 5956.  Thus, applications of microeconometrics will be the focus of this course.  We will read and discuss applications to see how consumer and producer choice problems are conceived and how microeconomic theory is used to specify and motivate estimation of empirical models.  We will also consider the full range of specification decisions researchers must resolve in estimation including: sampling, optimization methods, and numerical methods for deriving estimates.  We will also consider how robustness of resulting estimates can be evaluated and learn to appreciate the intrinsic subjectivity of estimates given the set of specification decisions from which they are derived.  We will consider parametric and some nonparametric approaches.  We also use exercises to provide a basis for learning "how-to".  Students will be required to develop, write, annotate, and implement code to estimate models.  Recommended languages include R, Matlab, and similar.  

The goal of this course is to provide graduate students with exposure to a wide range of applications of microeconometrics as reported in contemporary literature.  Through reading, discussion, and exercises students will gain ability and confidence in the active use of microeconometric methods including numerical methods for nonlinear optimization, integration over highly dimensioned functions, sampling from multivariate density functions, and simulation.  

How will we learn?  A combination of some lectures to provide settings, active student led discussion of readings, and preparation of assignments will provide a basis for learning. 

Unless the student has a solid understanding of the statistical and econometric theory behind microeconometric modeling, this course should not be taken independently of Microeconometrics Theory 5956.

Students will be able to develop a microeconometric model drawing on consideration of an economic choice setting and available data.  They will be able to control the complexity of the model and to specify estimation methods to enable estimation.  They will be able to write out models and estimation methods and translate their mathematical notation into a computer code such as R or Matlab that enables estimation.  They will be able to examine the sensitivity of derived estimates to the specification decisions employed and to take steps to derive robust estimates.  

Readings will be used to provide examples of applications.  Students will be expected to read in detail papers assigned.  Exercises will provide a basis for individual work using methods covered, as well as for group discussion of approaches taken.

Mathematical theory of optimization 

Linear econometric theory

Multivariate calculus

Matrix algebra

Office hours will be 

Tuesday 13h-15h

Limited other appointment times will be set as needed.