Bond ratings are letter ratings (Aaa=best) assigned to rms that issue debt. These ratings measure the quality of the rm from the point of view of the likelihood of repayment of the bond. Suppose you have been hired by an arbitrage house that wants to predict Moody's Bond ratings before they are published in order to buy bonds whose ratings are going to improve. In particular, your rm wants to distinguish between high quality, A-rated bonds and medium quality, B-rated bonds. Your boss is about to buy bonds based on the results of the following model: ^ Yi = 0:70 + 0:05Pi + 0:05PVi 0:02Di R 2 = :69 N = 200 where Yi = 1 if the rating on the ith bond =A, 0 otherwise; Pi = prot of the rm that issues the bond, in thousands of dollars; PVi = standard deviation of Pi over the last 5 years; Di = the ratio of debt to total capitalization of the rm that issued the bond; 1 (a) What model was used to run this equation? (10 marks) (b) What statistical problems exist in this equation? (10 marks) (c) What is the interpretation of the coecients of the 3 independent variables? (10 marks) (d) Do the signs of the estimated coecients make sense? (10 marks) (e) What would be the interpretation of the coecients if this was a Logit model? (10 marks) (f) What suggestions would you have for a rerun of this equation with a dierent specication? (10 marks) 2