ETC2400 Econometrics
Overview
"Econometrics ... consists of the application of mathematical
statistics to economic data" (Tintner,
quoted in Gujarati, D.N., Basic Econometrics (3rd edition),
1995.
Economic theory offers hypotheses that are mostly qualitative
in nature. It typically says very little about the numerical magnitude
of the relationships it studies. If we want to test a given theory, or
use it as the basis for policy formulation, or obtain forecasts, we need
a means of turning economic theory into empirical models. The discipline
of Econometrics is concerned with the measurement of economic relationships,
leading (possibly!) to the verification of economic theory. It combines
the data collected by the economic statistician with the models devised
by mathematical economics into empirical models of real-world behaviour,
accompanied by a measure of the associated uncertainty. Latterly, with
the increasing availability of stock exchange data, it has been extended
to the quantification of theoretical models of the behaviour of financial
assets.
This course focuses on one of the cornerstones of econometric
modelling, the linear regression model. It begins with an examination
of the model under "ideal" conditions, then introduces various
extensions to cope with more realistic modelling situations. It is required
for further work in theoretical Econometrics (ETC3400), and ETC2400 or
its close cousin ETC2410 Practical Econometrics are prerequisites for
most 3rd year EBS subjects (ETC3410, ETC3430, ETC3460, ETC3480, ETC3500,
and ETC3510).
Objectives
To introduce students to multiple linear regression methods,
including their use in estimating and testing the validity of models in
economics, finance and business. At the end of the course students should
have a thorough understanding of the properties of the linear model under
certain idealised conditions, and an appreciation of the model's strengths
and weaknesses. Students will apply the theory to a variety of economic
data sets to gain experience in estimating and evaluating economic models.
The aim is to provide a thorough grounding in the theory and practice
of linear regression, both as an end in itself, and as preparation for
more advanced work. Topics covered include the classical assumptions underlying
the linear model; the properties of the ordinary least squares estimator;
probability distributions and their application to interval estimation
and hypothesis testing; the generalised least squares estimator; an introduction
to the problems of heteroscedasticity, serial correlation, multicollinearity,
structural breaks and stochastic regressors.
Syllabus
Part 1: Introduction,
review of simple linear regression, extension to multiple linear regression,
review of basic matrix algebra, the matrix inverse, rank, and matrix differentiation;
the multiple linear model in matrix form, derivation of the Ordinary Least
Squares estimator, assumptions underlying the Classical linear model,
properties of the OLSE under these assumptions, sampling distribution,
estimation of the error variance, measures of goodness of fit, interval
estimation, tests of simple hypotheses (t-tests), tests of general linear
hypotheses (F-tests), point and interval prediction.
Part 2: The problem
of Multicollinearity, violations of the Classical assumptions, properties
of the OLSE under non-spherical errors, the generalised least squares
estimator, Feasible generalised least squares, FGLS in the context of
serial correlation, testing for serial correlation, FGLS in the context
of heteroscedasticity, testing for heteroscedasticity, shifts in the parameters,
structural change and dummy variables.
Prescribed text
Johnston, J. and J. Dinardo, 1997, Econometric Methods,
Fourth Edition, McGraw-Hill International Editions.
Coordinator
Dr Gael Martin
Teaching Material
Login to
WebCT
