Regression Analysis and Marks

BRUNEL UNIVERSITY insure of Science Degree examination Specimen Exam Paper 2005-2006 EC5002 manikin Financial Decisions and Markets EC5030 Introduction to Quantitative Methods Time completelyowed 1. 5 hours Answer every of question 1 and at least two other questions 1. needed Provide brief answers to all the following (a) A sample of 20 observations corresponding to the model Y = + X + u, gave the P P P following data (X X)2 = 2154, (Y Y )2 = 869, and (X X)(Y Y ) = 10604. Estimate . 5 marks) (b) plant that r2 = byx bxy , where byx is the least-squ bes (LS) slope in the atavism of Y on X , bxy is the LS slope in the regression of X on Y , and r is the coe? cient of correlation surrounded by X and Y . (5 marks) (c) Present four resource in ation/unemployment regressions. (5 marks) (d) Give one reason for autocorrelated disturbances. (5 marks) (e) Explain how we might usance the Breusch-Godfrey statistic to test estimated residuals for serial correlation. (5 marks) (f) The fol lowing regression equation is estimated as a production function for Q lnQ = 137 + 0632 lnK + 0452 lnL, cov(bk bl ) = 0055 0257) (0219) where the well-worn errors are given in parentheses. sieve the speculation that capital (K ) and labor (L) elasticities of outfit are identical. (5 marks) continue (Turn over) 1 event TWO QUESTIONS FROM THE FOLLOWING 2. (a) Economic theory supplies the economic interpretation for the predicted relationships between nominal (in ation) uncertainty, real (output growth) uncertainty, output growth, and in ation. Discuss ve testable hypotheses regarding bidirectional author among these four variables. (25 marks) + yt b) An investigator estimates a linear relation for German output growth (yt ) yt = 1 + ut , t = 1850 1999. The values of ve test statistics are shown in Table 1 Discuss the results. Is the above equation in good order specied? (10 marks) 3. (a) i) Show how various examples of typical hypotheses t into a common linear framework Rb = r, where R is a (q k) matrix of cognize constants, with q k, b is the (k 1) least-squares vector, and r is a q -vector of known constants. ii) Show how the least-squares estimator (b) of closely . an be used to test various hypotheses iii) The test procedure is whence to reject the hypothesis Rb = r if the computed F value exceeds a preselected critical value Discuss. (20 marks) (b) The results of least-squares estimation (based on 30 quarterly observations) of the regression of the actual on predicted stakes evaluates (three-month U. S. Treasury Bills) were as follows rt = 024 + 094 rt + et RSS = 2856 (086) (014) where rt is the observed interest rate, and rt is the clean expectation of rt held at the end of the preceding quarter.FiguresX parentheses are estimated standard errors. in X (rt r )2 = 52. The sample data on r give rt =30 = 10, According to the rational expectations hypothesis expectations are unbiased, that is, the average prediction is equal to the observed realization of the variable under investigation. Test this claim by reference to announced predictions and to actual values of the rate of interest on three-month U. S. Treasury Bills. (Note In the above equation all the assumptions of the classical linear regression model are satised). 15 marks) Continued (Turn over) 2 4. (a) What are the assumptions of the classical linear regression model? (10 marks) (b) Prove that the variance-covariance matrix of the (k 1) least-squares vector b is var(b) = 2 (X 0 X) 1 , where 2 is the variance of the disturbances and X is the (n k) matrix of the regressors. (15 marks) b (c) In the two-variable equation Yi = a+bXi , i = 1 n show that cov(a b) = 2 X= X)2 . (10 marks) X (X 5. (a) Explain how we might use White statistic to test for the presence of heteroscedasticity in the estimated residuals. 10 marks) (b) A specied equation is Y = X +u, with E(u) = 0 and E(uu0 ) = where =diagf 2 1 Derive White correct estimates of the standard errors of the OLS coe? cients. s (15 marks) (c) Explain how we might test for ARCH eects? (10 marks) 2 2g . 3 Table 1. Test statistic Value of the test p-value White heteroscedasticity test 50. 72 0. 00 Box-Pierce Statistic on 82. 263 0. 00 Squared Residuals Jarque-Bera statistic 341. 754 0. 00 ARCH test 65. 42 0. 00 Ramsey test statistic 39. 74 0. 00 4