11398 - Statistical Models for Economic Behaviour

Learning outcomes

The course's goal is to provide for deep knowledge of problems and technical methods about estimation of statistical models describing micro-economic behaviour. Influence of data generating process on estimation strategy is highlighted referring to economic contexts of labour market and consumer behaviour. Students are required to have mastery of concepts and statistical methods concerning estimation multivariate linear models.

Course contents

1) Fundamental concepts and definitions (remind):

• Economic theories, models and measures; macro and micro models.
• Definition of utility function, rational choice, bounded rationality.
• Choices and uncertainty, the limited information problem.

 2) Main estimation methods (remind):

• Data generating process.
• Fundamental assumptions of the  classical linear model
• Estimation procedure when they are violated: Generalized Least Square, estimation of variance-covariance matrix.
• Maximum likelihood estimates, algorithms and software.

 3) Model estimates from panel data

•  Panel data definition and organization
• Fixed effects models: between, within and LSDV estimators
• Error components models: random effects, GLS e FGLS estimators
• Diagnosing panel data models

4) Truncated and censored variables

• Discrete choice models:Probit e Logit.
• Turncated and censored distributions: moments and  Inverse Mill's ratio
•  Tobit e double hurdle estimates
•  Selection models: Heckmann e Amemya.

 5) Duration models

•  Time intervals and duration measurement, discrete and continuous time
• Censored data in duration models
•  Non parametric approach: Survival and mortality tables, discrete survival and hazard functions
• Parametric approach: continous survival and hazard functions. Exponential, Weibull, log-log distributions
• Semi-parametric approach: baseline and proportional hazard, Cox model
• Competing risks models

Readings/Bibliography

W.H. GREENE, Econometric Analysis, Mac Millan, London, Third Edition, 1997

Chap.: 6, 8, 9, 11, 12, 14, 19, 20.

Teaching methods

• Lectures
• During the course will be presented guided case studies aimed ad stretch short reports on various topic
• Later, case studies will be done individually and the reports will be delivered and corrected by the teacher and will be considered as a  part of the final evaluation

The case studies are organized as follows:

• The definition of a theme of analysis chosen among those in the program of the course 
• Estimation of models via algorthims implemented by the students in Excell 
• Graphical representations and statistical tables adequate at the analysis of the estimates
• Drafting of a report describing the step of the work and the main results

Assessment methods

The examination aims to assess the achievement of the following learning objectives:

• in-depth knowledge of the effects of data generating process on estimations techniques 
• ability to choose the right model in each case
• ability to implement elementary estimation algorithms 
• the ability write reports on the sources of information, on calculations carried out and the results obtained

The final evaluation consist of:

•  2 individual reports of the case study (according to the method described in the section "Methods") 
• 1 written test with multiple choice questions

Reports

• The reports should be done individually in writing on a topic agreed with the teacher. The assignment of topics will occur 15 days before the end of the first cycle of lectures (the first report) and 10 days before the end of the second cycle (second report)
• The reports should be sent by mail to the teacher and will be evaluated within 15 days of receipt. The evaluation will be expressed out of thirty and notified via email. The evaluations of the two relations will contribute to the formulation of the final score for the 70% (35%+ 35%)

Test

• The written exam will cover all the topics covered in the course, in particular the degree of assimilation of some theoretical tools and the ability to read statistical indicators and estimates of models. This test will consist of 15 questions with 4 possible answers and will be evaluated in thirty (assigning 2 points for each correct answer), contributing to the formulation of the final score for the 30%
• On the teacher website (www2.stat.unibo.it/drudi /) are available sample tests, with interactive compilation and correction

Scheduling

The written test will be performed at the end of the course, while the 2 reports must be delivered at least 10 days before the date of the written test. As a guide, the first report could be prepared after the end of the first cycle of lessons.

 The dates for the Test will be available at  almaesami.unibo.it

Teaching tools

In the classroom:

• PC equipped with video projector and Internet access

 In the teacher's internet site (www2.stat.unibo.it/drudi/):

• File of slides available on-line and downloadable
• Other teaching material available on-line
• Link to statistical databases • 
• Exercises available online with interactive correction 
• Fac similar texts and exercises in Excell

 

Links to further information

http://www2.stat.unibo.it/drudi/

Office hours

See the website of Ignazio Drudi