Answer Happy

15.17 The Federal Deposit Insurance Corporation (FDIC) releases data on bank failures. Following are data on the number of U.S. bank failures in a given year and the total amount of bank deposits (in $ millions) involved in such failures for a given year. Use these data to develop a simple regression forecasting model that attempts to predict the failed bank assets involved in bank closings by the number of bank failures. Compute a Durbin-Watson statistic for this regression model and determine whether significant autocorrelation is present. Let a = .05. Year Failures Failed Bank Assets 5 Woo va UlWN- 19 11 7 34 45 79 118 144 201 221 206 159 108 100 42 11 9 10 11 12 13 14 15 16 17 18 8,189 104 1,862 4,137 36,394 3,034 7,609 7,538 56,620 28,507 10,739 43,552 16,915 2,588 825 753 186 27 5 1 71 DURBIN-WATSON TEST Že - 6-1² 2 D= 11 Σe: t=1 where se n=the number of observations
15.17 The Federal Deposit Insurance Corporation (EDIC) releases data on bank failures. Following are data on the number of U.S. bank failures in a given year and the total amount of bank deposits (in $ millions) involved in such failures for a given year. Use these data to develop a simple regression forecasting model that attempts to predict the failed bank assets involved in bank closings by the number of bank failures. Compute a Durbin-Watson statistic for this regression model and determine whether significant autocorrelation is present. Let a = .05. Year Failures Failed Bank Assets 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 11 7 34 45 79 118 144 201 221 206 159 108 100 42 11 6 5 1 8,189 104 1,862 4,137 36,394 3,034 7,609 7,538 56,620 28,507 10,739 43,552 16,915 2,588 825 753 186 27 DURBIN-WATSON TEST Σ D = 14- 12 11 where n=the number of observations Note from the formula that the Durbin-Watson test involves finding the differenc between successive values of error (e- .). If errors are positively correlated, this differ ence will be smaller than with random or independent errors. Squaring this term elim nates the cancellation effects of positive and negative terms. The null hypothesis for this test is that there is no autocorrelation. For a two-tailed the alternative hypothesis is that there is autocorrelation. H:p = 0 Hip + 0 autocorrelation is positive autoc Sena