Topics in Applied Econometrics: Methods and Evidence ETR3 30 hours Claudio Lucifora coordinator Fede

1
Topics in Applied Econometrics Methods and
Evidence ETR3 (30 hours)Claudio Lucifora
(coordinator)Federica Origo, Paolo Ghinetti,
Simona Comi (teaching assistants)

• Object
• The course combines statistical methods and
  applied economics problems
• Awareness of the empirical approach to economics
  
• Experience in the analysis and use of empirical
  data in public economics
• Understanding the nature of survey data and
  methods of dealing with it
• The use of econometric software packages as tools
  of policy analysis.

2
Structure and assessment

• The course is structured in modules on a specific
  topic.
• (i) lecture
• (ii) paper presentation
• (iii) tutorial.  
• The assessment of the course involves
• final project, in which the student has to show
  his/her proficiency in writing up an empirical
  paper. This will contribute to 70 percent of the
  final mark.
• two intermediate assessments to be handed in at
  the middle of the course programme and at the end
  of it (5 pages max. for discussion of the
  results, and programme and log output in
  appendix). This will contribute to 20 percent of
  the final mark.
• Evaluation of the presentation. This will
  contribute to 10 percent of the final mark.

3
1. Carrying Out an Empirical Project

• Posing a question, the literature review, the
  theoretical framework
• Data collection, econometric analysis, a causal
  interpretation, testing and forecasting
• Writing an empirical paper

4
2. Multiple Regression Analysis with Qualitative
Information Binary (or Dummy) Variables

• The Economic Analysis of the Gender Earnings Gap
• The Union-nonunion wage differentials
• Evaluating Public-Private Wage Differentials
• The Ranson-Oaxaca decomposition
• Returns to ICT skills and using a computer

5
In-class work

• Paper presentation Blau F. D. and L. M. Kahn
  (2000), Gender Differences in Pay, Journal of
  Economic Perspectives, 14(4).
• Contents This paper uses cross-national
  comparable data from ISSP to analyse the patterns
  and consequences of wage discrimination and
  occupational segregation.
• Tutorial Evaluating Gender Wage Differentials
  Estimation of wage equations with gender dummy to
  identify the gender-wage gap. (i.e.
  discrimination)
• Data data Bank of Italy, 2002

6
3. Limited dependent variables

•   The optimal choice of the educational level.
•   Female Labour Market Participation, Child
  Care Assistance
•   The Effects of Taxation on Labour Supply

7
The empirical issue

• Objective to deal with truncated, censored or
  self-selected samples.
• Basic question How can we estimates models with
  non-random selected sample?
• Example We want to estimate a wage equation for
  women
• Log wixi?ui
• But wi is observed only if a woman works, i.e her
  reservation wage is lower than the wage market
  offered to her (we do not observe wage if the
  woman does not work).
• OLS biased, we need to take into account the non
  random nature of the sample.

8
Applications

• Estimating women wage equations modelling
  participation.
• Estimating the effect of taxation on labor supply
• Estimating returns to college

9
In-class work

• Paper presentation Bardasi and Gormick (2000)
  "Part-time work employment Workers' choice and
  wage penalties in five industrialized countries",
  LIS wp 223
• Contents This paper uses cross-national
  comparable data from LIS to analyse the patterns
  and consequences of part-time employment among
  women.
• Tutorial Estimation of models with limited
  dependent variables Women Participation and The
  part-time wage gap.
• Data households data from SHIW 2002.

10
4. Estimating Treatment Effects under Endogeneity
and Self Selection

• Policy interventions An analysis of treatment
  effects of interest and models with binary
  endogenous treatment
• Selection in economic choices a model with self
  selection and endogenous switching
• Public-Private Schooling Attendance and Family
  Background
• Pension portability and labour mobility
• Wage equations, and public-private union-non
  union differentials (revisited)

11
The empirical issue

• Objective estimating the causal effects of
  binary treatments (union membership, income
  support programs, policy interventions, etc.) on
• Continuous outcomes
• Binary outcomes
• Problems
• Lack of counterfactuals
• Lack of experimental data
• Non-random assignment (endogeneity or
  self-selection)
• Heterogeneous treatment effect in the population

12

• The strategy
• construct counterfactuals using observational
  data
• Use econometric techniques to recover and
  estimate true causal effects (free of endogeneity
  and selection bias)
• Key feature well-defined set of assumptions on
  how the causal effect can be identified
  (identification strategy). Alternatives
• Selection on observable heterogeneity only (PSM)
• Selection on unobservables (endogeneity or
  self-selection) ? role of unobservable
  heterogeneity

13

• focus on models with selection on unobservab.
• Two class of (linear) models will be discussed
• Binary endogenous treatment
• Homogeneous (constant) treatment in the populat.
• genuine endogeneity problem
• Alternative estimation techniques
• IV methods
• 2-step Heckman correction
• Model with endogenous switching
• Heterogenous treatment (individual-specific
  effect)
• Self-selection and comparative advantages
• 2-step Heckman correction
• augmented Oaxaca decomposition

14
In-class work

• Paper presentation Evans and Schwab (1995),
  Finishing high school and starting college do
  catholic schools make a difference?, QJE
• Contents the paper provides an in-depth and
  clear discussion on how the ideas developed in
  class modify when both the treatment (attending
  a Catholic high school) and the outcome
  (finishing the high school) are binary
  indicators, and, thus, probabilitic models have
  to be used.
• Tutorial Alternative methods to evaluate the
  public sector wage premium
• Data Bank of Italy SHIW.

15
5. Non-parametric and semi-parametric Estimation
an application to quantile regression

• Estimating the Returns to Education and Training
  Programmes
• Kernel estimates of wage functions
• Quantile regressions
• Evaluating the Effects of the Minimum Wage on
  Employment

16
The empirical issue

• Objective OLS estimate an average effect
  throughout the whole distribution. This might not
  be correct as returns may vary along the (wage,
  income, hours, etc.) distribution. Use
  nonparametric or semiparametric model
• Problems
• Choice of banwidth
• Choice of quantiles (quartiles, deciles,
  centiles?)
• Counterfactual distribution
• Heterogeneous effects in the population

17
In-class work

• Paper to be presented Abrevaya, J. (2001) The
  Effects of Demographics and Maternal Behavior on
  the Distribution of Birth Outcomes, Empirical
  Economics, 26(1).
• Tutorial Evaluating returns to education (data
  Bank of Italy, 2002)

18
6. Propensity Score Matching (PSM)

• Evaluating the Effects of Active Labour Market
• Programmes (ALMP) on (Un)Employment
• Evaluating the Efficacy of Training Programmes on
  Job Search
• Estimating the effects of an anti-poverty program
  The schooling cost of teenage wed-lock
  childbearing
• Estimating the union membership wage premium

19
The empirical issue

• Objective estimation of the net impact of a
  certain treatment (ex ALMP, training, ICT
  introduction, change in pay compensation schemes,
  etc.)
• Basic question How and to what extent does
  the treatment change the average outcome
  variable(s) for the individuals/firms who were
  actually treated (ATT)?
• The problem we cant observe the same individual
  in two different states at the same time and
  (usually) the treated are not a random sample of
  the initial population (self-selection into the
  treatment and/or endogeneity of the treatment)
• ? Need to estimate the counterfactual situation

20
Applications

• Evaluating the Effects of Active Labour Market
  Programmes (ALMP) on (Un)Employment
• Evaluating the Efficacy of Training Programmes on
  Job Search
• Estimating the effects of an anti-poverty program
  
• The schooling cost of teenage wed-lock
  childbearing
• Estimating the union membership wage premium

21
In-class work

• Paper presentation Dehejia, R. and Wahba, S.
  (2002), Propensity Score Matching methods for
  non-experimental causal studies, Review of
  Economics and Statistics, 84, 1, 151-161.
• Contents This paper is aimed at providing
  evidence on the validity of PSM to evaluate
  non-experimental causal studies (application to
  impact of training on earnings in USA, comparing
  PSM results with the benchmark results from the
  experiment).
• Tutorial impact of vocational training on
  (subsequent) employment probability in Italy.
• Data individual data from ISTAT, LFS April
  2003.