Course Objective

(i) To gain insights in the computational challenges in Econometrics and
Data Science. From simple regression models towards time series and
nonlinear models, computations need to be implemented in a numerical
stable manner. (ii) To explore the ability of computer simulations to
gain insights into the properties and behavior of estimates and test
statistics.

Course Content

In this course we discuss numerical and simulation-based methods for
their use in econometrics and data science. In the first part we review
matrix computations, numerical methods for optimization and Monte Carlo
integration. We illustrate how these methods are used for estimation of
parameters in statistical models. In the second part we investigate
properties of estimators, test statistics and residuals using simulation
studies. In particular, we simulate distributions of parameter estimates
under different data generation processes, but also distributions of
test statistics such as unit-root tests, of R-squared goodness-of fit in
spurious regressions, and of model selection criteria such as Akaike
information criteria. We finally use simulations to verify the accuracy
of diagnostic tests related to normality and heteroskedasticity.

Teaching Methods

Lectures (4 hours, each week) and tutorial lectures (2 hours, each
week). The lectures will introduce the methods from a selection of books
and articles. The homework (mostly in the form of assignments) can be
discussed during the tutorials, together with Q&As.

Method of Assessment

Written exam (percentage P), Assignments (percentage 100-P), where P is
in [50,70].

Recommended background knowledge

(i) Introductory courses in Econometrics and Statistics; (ii) Basic
programming skills: some familiarity with one of [Python, MATLAB, Ox, R]