4 Suppose we are estimating a model for the return on a bond rt of the form, rt=a + brt-1+ et where e is an error term. a) Explain the difference between the conditional variance and the unconditional variance of r. Which of the two is more relevant for financial decision making? (20%) b) If we suspect that the variance of e changes systematically through time what would be the consequences for standard OLS estimation. Outline the ARCH and GARCH models, which would allow us to deal with this problem fully. (30%) c) If you believed that the variance of e affects the return on the bond how would adapt the GARCH model to allow for this. (30%) e) If you were investigating a model such as the capital asset pricing model which used the covariance between the market return rm and the bond return r, How could the GARCH model be extended to allow for this case? (20%)