英语论文网

Seminar 3: Testing CIP and UIP in Eviews
The principal aims of this seminar are to:
• Demonstrate misspecification testing in Eviews.
• Demonstrate the use of alternative estimators in Eviews.
代写留学论文To achieve these aims you will estimate CIP and UIP relationships for the UK/US foreign exchange market using OLS initially. Then you will subject the model to rigorous misspecification testing (testing whether the assumptions of the Classical Linear Regression Model (CLRM) are compatible with the observed data). This will include tests for:
• Heteroscedasticity.
• Autocorrelation.
• Incorrect functional form.
• Structural instability.
• Endogenous regressors.
The evidence for an endogenous regressor will lead you to re-estimate the model using an Instrumental Variable Estimator (IVE). The session will culminate with the estimation of a Generalized Method of Moments (GMM)/IV model to take account of both the endogenous regressor and heteroscedasticity/autocorrelation in the errors.
The learning outcomes from this analysis will be to develop your understanding of:
• Misspecification testing in Eviews.
• The importance of misspecification testing as part of any econometric analysis.
• The consequences of violations of different assumptions of the CLRM.
• Remedies for these violations including the use of alternative (non-OLS) estimators.
• How (and when) to implement these remedies/procedures in Eviews.
The data for this exercise are in the file cip_sem3.wf1 which I sent to you in an email. These data will allow you test both CIP and UIP relationships. This handout goes through the analysis for the CIP relationship in detail. Please attempt the estimation and testing for the UIP relationship in your own time to test your understanding of this analysis. For background on the CIP/UIP relationships read Cuthbertson and Nitzsche (2004) Chapters 24.3, 24.4, 25.1 and 25.2 (and see lecture 4). Further background on IVE can be found in Verbeek Chapter 5.
1. Initial Estimation using OLS
Create a variable for the (log) 3 month forward premium:
Click Genr on the workfile toolbar and type in:
fp_3m=log(uk_3mfrate)-log(uk_spot_rate)
Estimate the CIP relationship by OLS:
Click Quick/Estimate Equation on the main toolbar and type in:
fp_3m c uk_3mtbills-us_3mtbills
Dependent Variable: FP_3MMethod: Least SquaresDate: 02/04/07 Time: 21:23Sample: 5/10/2001 9/30/2005Included observations: 1147VariableCoefficientStd. Errort-StatisticProb. C-0.007260.000422-17.201410UK_3MTBILLS-US_3MTBILLS0.9497180.01685556.345830R-squared0.734945 Mean dependent va0.01585Adjusted R-squared0.734713 S.D. dependent var0.006551S.E. of regression0.003374 Akaike info criterion-8.543509Sum squared resid0.013037 Schwarz criterion-8.534712Log likelihood4901.702 F-statistic3174.853Durbin-Watson stat1.458456 Prob(F-statistic)0
In your own time, repeat the estimation for the UIP relationship. The dependent variable for UIP is the 3 month (12weeks×5days=60day) holding period return on sterling:
Click Genr on the workfile toolbar and type in:
uk_spot_3m=log(uk_spot_rate(60))-log(uk_spot_rate)
2. Misspecification testing
The background on these tests can be found in your notes for lecture 3 (and the references therein). Be sure to repeat these tests later for the UIP OLS model.
i.) White test for heteroscedasticity
There is no di