Special Topics in Applied Econometrics


The lecturer is Maoliang Ye, Assistant Professor at Xiamen University (P.R. China).


Maoliang Ye

Xiamen University
361005 Xiamen, Fujian Province
P.R. China


Students must have followed an introductory course to Econometrics, providing them a good knowledge of the linear econometric models.

Students must be able to use STATA, a standard econometric software package.

Lecture topics

Chapter 1 | The evaluation problem and random control trial (RTC)

  • The evaluation problem and its challenge (selection bias)
  • Econometric method for randomized control trial under perfect compliance
  • Econometric method for randomized control trial under imperfect compliance
  • Design and implementation of field experiments

Chapter 2 | ​Applications of instrumental variable (IV) method

  • Review of IV methods
  • Practical issues in using IV
  • Example papers
  • Practice

Chapter 3 | ​Difference-in-difference (DID)

  • DID methods and its different forms
  • Practical issues in using DID
  • Example papers
  • Practice

Chapter 4 | ​Panel data

  • Formulation: Fixed effects, random effects, within and between transformations
  • FGLS and ML estimation
  • Endogeneity problems
  • Introduction to dynamic models

Chapter 5 | Regression discontinuity design

  • Regression discontinuity design
  • Practical issues
  • Example papers
  • Practice

Course contents

The course focusses on the econometric methods for causal inference and can also be used in international economics:

  • The evaluation problem and random control trial
  • Application of the instrumental variable method
  • Difference-in-difference
  • Panel data
  • Regression discontinuity design

It also deals with example papers and applications based upon specific datasets.

Core reading

  • Joshua D. Angrist and Jörn-Steffen Pischke (2009): Mostly Harmless Econometrics: An Empiricist’s Companion. Princeton University Press
  • W.H. Greene, Econometric Analysis, 7th edition

Learning outcomes

Upon successful completion of the course, students should be able:

  • To formulate the model and to choose between fixed and random effects
  • To estimate the model by FLS

For panel data

  • To formulate the latent model and its link to the observable model
  • To estimate the standard LDV models by maximum likelihood: logit, probit, tobit, nested logit
  • To interprete the estimation results

For limited dependent variables

  • To estimate these models
  • To apply them to gravity models used in international trade theory

For counting and Pseudo-Poisson models


The course is spread over 2 weeks and 10 classes: 8 classes for teaching and 2 classes for exercises and Q&A.


Students are evaluated through problems sets and a written exam.