The workshop focused on applied micro-econometrics and public policy evaluation, held from July 25-27, 2016, with objectives including discussions on causal inference and methodological solutions. Participants learned techniques such as matching methods, difference-in-difference, panel data models, and instrumental variables, applying them through Stata software. Key skills included descriptive statistics, multivariate estimation commands, and ensuring reproducibility through documentation and programming in Stata.

1. Overview and
Objectives of the
Workshop
Day 1, Introduction
By Ragui Assaad
Training on Applied Micro-Econometrics and
Public Policy Evaluation
July 25-27, 2016
Economic Research Forum

2. Objectives
• Discuss fundamental problem of causal
inference in program or policy evaluation
• Propose methodological solutions
• understand the underlying econometrics
• gain experience in applying the techniques
2

3. Policy/Program Evaluation
• Although all the techniques discussed are framed
in terms program/policy evaluation, techniques
can be used to study any endogenous treatment
so long as the assumptions underlying the
various methods are upheld
3

4. Day 1
• Fundamental Problem of Causal Inference
• Overview of Potential Solutions:
• Randomization
• Non-Experimental Methods
• Solutions (1) Matching Methods
• Propensity score matching and propensity score
weighted regressions
• Application in STATA (1): Propensity score
matching and propensity score weighted
regressions
4

5. Day 2
• Solutions (2)
• Difference-in-Difference Methods
• Applications in STATA (2)
• Difference-in-difference in practice
• Solutions (3)
• Panel Data Models: fixed and random effects
• Applications in STATA (3)
• Panel data models
5

6. Day 3
• Solutions (4)
• Control Functions and Instrumental Variable
• Applications in STATA (4)
• Instrumental variables
• Solutions (5)
• Regression Discontinuity Method
• Applications in STATA (5)
• Regression Discontinuity Method
6

7. STATA skills
• Commands to undertake microeconometric program
evaluation
• Descriptive statistics
• Graphing
• Multivariate estimation commands
• Post-estimation commands
• Commands to support reproducible results and
documentation
• Programming skills
7

8. Descriptive Statistics and Graphing
• Descriptive statistics present key characteristics of the
analysis in an accessible format
• Critical for:
• Understanding your data
• Providing background and motivation for a policy issue
• Checking assumptions of microeconometric models
• Can be presented graphically
• Appropriate descriptive statistics and graphs are
particularly important to help communicate results to
policy makers
8

9. Multivariate estimation commands
• Propensity score matching
• Propensity score weighted regressions
• Difference-in-difference estimation
• Panel data models
• Fixed effects & random effects
• Instrumental variables techniques
• Regression discontinuity designs
9

10. Post-estimation commands
• After estimating a model, can undertake post-estimation
commands
• Types of post-estimation commands:
• Saving your results
• Checking the assumptions of your estimator
• Testing hypotheses about coefficients
• Obtaining predicted values or marginal effects
10

11. Documentation and Reproducibility
• When you (or others) re-run your analyses, want to get the
same results
• A study is reproducible if all the numbers presented in the
study can be reproduced
• When re-running the STATA do files
• A study is replicable if a re-study repeats the finding
• Documenting your findings is critical to reproducing your own
results and allowing others to reproduce your results
• STATA commands should all be saved in do files to allow results to be
reproduced (re-run) later
• Log files and other forms of output (graphs, tables) saved from STATA
11

12. STATA programming
• STATA has many commands that are built in to the
program
• Newer releases expand the set of commands available
• Sometimes you will want to undertake econometrics that
are not built in to STATA
• Skills that allow you to go beyond STATA’s built-in
commands:
• Locating, installing, and applying user-written programs in STATA
• Based on .ado files
• Writing and using your own programs
• Creating your own commands through programs in .do files
12