英语论文网

Key Revision Checklist: Lectures 4-6
The following is a revision checklist covering the key concepts from lectures 4-6. You should understand and be able to explain:
写留学论文• The consequences of autocorrelation and heteroscedasticity for OLS estimators.
• The Newey West HAC variance-covariance matrix (as one solution to the problems of autocorrelation and heteroscedasticty in the Classical Linear Regression Model (CLRM)):
o Contexts in empirical finance where you might use a Newey West HAC variance covariance matrix. For example, where observations on the dependent variable ‘overlap’ (overlapping data problems – as in the UIP model). (See also below in the context of weighting GMM sample moment conditions in the presence of autocorrelation and/or heteroscedasticity).
• The principle underlying Method of Moments Estimation (and its relationship with OLS). The use of Generalized Method of Moments (GMM) when the model is over-identified (i.e., when there are more moment conditions than unknown parameters).
• The consequences of endogenous regressors for OLS estimators.
• Solutions to the problem of endogenous regressors: IV and GMM estimators (including 2SLS):
o The use of a Newey-West HAC variance-covariance matrix to estimate GMM weights where there are also problems of autocorrelation and/or heteroscedasticity in the model.
• Applications of GMM/IV estimators in empirical finance (e.g., testing CIP and UIP: see Seminar 3).
• Misspecification tests for the CLRM (see Seminar 3). Tests for:
o Heteroscedasticity.
o Autocorrelation.
o Incorrect functional form.
o Parameter instability.
o Endogenous regressors.
• The Wold Decomposition Theorem (as the theoretical basis for ARMA modelling).
• Stationarity and invertibility conditions for ARMA models.
• The shapes of the ACFs and PACFs for different types of ARMA models.
• The Box Jenkins methodology for ARMA modelling: identification; estimation; and testing.
o The use of information criteria to aid model selection.
• The utility of ARMA modelling (as compared to structural modelling) in empirical finance (i.e., for potentially superior out of sample forecasting: see Seminar 4).
• The motivation for using GARCH models in empirical finance.
• The structure of the basic ARCH/GARCH model:
o The conditional mean equation (specified according to finance theory or a statistical model such as ARMA).
o The conditional variance equation: GARCH(1,1).
写留学论文o The assumption of Gaussian (NID) standardized residuals (⇒a Gaussian conditional error distribution).
• Methodology for identifying, estimating and testing GARCH models:
o The use of Maximum Likelihood to estimate GARCH models.
o The importance of misspecification testing based on testing the NID assumption for the standardized residuals – see Seminar 5.
• Extensions to the ‘plain vanilla’ GARCH with applications:
o Multivariate GARCH: applications in estimating dynamic hedge ratios and time varying CAPM betas.
o Asymmetric GARCH (TARCH and EGARCH): applications in modelling leverage effects (see Seminar 5).
o GARCH-M: applications in modelling time varying risk premia (see Seminar 5).
Key Revision Checklist: Lectures 7-9
The following is a revision checklist covering the key concepts from lectures 7-9. Based on these lectures, you should understand an