EC 3064, lecture 2, PAGE 1 EC3064, Applied Econometric Project Lecture 2: some empirical models List of the topics: 1 Modelling consumption demand 2 Modelling money demand 3 Empirical analysis of the purchasing power parity hypothesis 4 Modelling house prices 5 Testing the Fisher effect 6 Terms structure of interest rates 7 Modelling inflation persistence 8 Last minute addition: are we really cheated at the pumps?
EC 3064, lecture 2, PAGE 2 Some general ideas: 1. In order to test parameters, the model has to be stationary (all variables have to be transformed to stationarity) 2. Even if model is stationary, the non-normality of the residuals makes testing approximate, and not very reliable 3. Long-run (nonstationary) equations can be estimated, but testing is either impossible, or tricky) 4. The general algorithm is then: a) Test for stationarity (Dickey-Fuller, etc.) If the variables are non-stationary, then: Estimate long-run relationship and test for it (cointegration).
EC 3064, lecture 2, PAGE 3 If it exist then: Formulate, estimate and test a short-run model (with variables transformed to stationarity) with error correction (equilibrium correction) term. Check for the normality of the residuals and, if exists, formulate testing conclusions strongly; if it does not exist, with some caution. If it does not exist: Formulate, estimate and test a short-run model (with variables transformed to stationarity) without error correction (equilibrium correction) term. Check for the normality of the residuals and, if exists, formulate testing conclusions strongly; if it does not exist, with some caution.
EC 3064, lecture 2, PAGE 4 Methods you are expected to know: Obligatory (all students should know this) ? Ordinary least squares applied to a stationary dynamic model, in particular: ? Testing of the significance of parameters (individual and joint) ? Assessing properties of the residuals (autocorrelation, heteroscedasticity, etc.) ? Testing for outliers and…