B030722 - ECONOMETRICS LAB

ENGLISH

Students are requested to verify, by means of Monte Carlo simulations, some properties of the methods whose "theory" had been studied
in the previous courses of Econometrics (such as Econometrics of the BA, Microeconometrics, Macroeconometrics, Statistical Inference).
Some lessons will "refresh" (or integrate) the theory already studied in the previous courses of Econometrics.

Calzolari G. (2012): "Econometric Notes". https://mpra.ub.uni-muenchen.de/85396/17/MPRA_paper_85396.pdf

Greene, W. H. (2008): "Econometric Analysis" ($6th$ edition). Prentice-Hall, Inc. Upper Saddle River, NJ.

Matching theory with practice and computational metods in Econometrics.

Statistical Inference, Microeconometrics, Macroeconometrics.
Moreover, students "must" already have a working knowledge of
(and be "moderately" fluent with)
a programming language of their choice
(Matlab, R, C, Fortran, Python are all suitable).

Some lessons (chalk and blackboard) will "refresh" (or integrate) the theory already studied in the previous courses of Econometrics.
NO LESSON will deal with a particular programming language. Each student will adopt and use a programming language of her/his choice (therefore, she/he "must" already have a working knowledge of, and be "moderately" fluent with, a suitable programming language, not simply be able to apply tools provided by the available commercial computer packages). As students will be requested to "program" the Monte Carlo simulation algorithms, some sort of "flow-chart" of the algorithms will be discussed during the lessons. Computational experiments will be done and discussed during the lessons.

To get the "idoneity", a student must produce "convincing" results of her/his experiments.
Being a "laboratory", results of the experiments of each student will be discussed every week during lesson-time.

Students are requested to verify, by means of Monte Carlo simulations, some properties of the methods whose "theory" had been studied
in the previous courses of Econometrics (such as Econometrics of the BA, Microeconometrics, Macroeconometrics, Statistical Inference).

Examples are:
Unbiasedness and efficiency of ordinary least squares in a regression model satisfying "classical" assumptions (Gauss-Markov theorem).
Bias, inconsistency and/or consistency of ordinary least squares in presence of endogenous regressors.
Instrumental variables estimation method.
Likelihood, score and information matrix.
Maximum likelihood estimation method,
Numerical techniques to maximize the likelihood,
Computational efficiency of maximization techniques: Newton-Raphson versus BHHH.
Alternative estimators of the information matrix and of the variance-covariance matrix.
Pseudo-maximum likelihood (or quasi-mximum likelihood) of misspecified models and robust estimator of the variance-covariance matrix.
Simulated maximum likelihood.
Indirect inference.

Monte Carlo simulations will be done on simple linear and non-linear regression models,
microeconometric models (logit, probit, tobit),
models for time series (AR, MA).

Some lessons will "refresh" (or integrate) the theory already studied in the previous courses of Econometrics.

The course will "not" include lessons on a particular programming language.
Students will be requested to "program" the Monte Carlo simulation algorithms,
not simply to apply tools provided by the available commercial computer packages.
Therefore, students "must" already have a working knowledge of
(and be "moderately" fluent with)
a programming language of their choice
(Matlab, R, C, Fortran, Python are all suitable; different students may use different programming language;
the purpouse of the course is to experiment econometric methods,
"not" to learn a programming language).

To pass the exam ("idoneous") a student must produce "convincing" results of her/his experiments.