Econ Essay

1278 WordsFeb 3, 20146 Pages
Question Two (a) MODEL 1: R&Di = α0 + α1 Profitsi + ѵi R&Di = 77.57696 + .3614021Profits R^2 = 0.4976 s.e (983.6063) (.0907899) t = (0.0789) (3.9806) MODEL 2: ln R&Di = β0 + β1 ln Profitsi + ѵi ln R&Di = -1.2551 + .9910lnProfits R^2 = 0.7064 s.e (1.3647) (.1597) t = (-0.9197) (6.2043) (b) MODEL 1: From the graph for Model 1, it is evident that errors are heteroskedastic. This is because e2 is increasing with Profits. MODEL 2: From the graph above for Model 2, we can see that by taking the log of Model 1 we are able to smooth the data and thus reduce variance within the model. However in the log-log model it shows that it is not clear to have heteroskedasticity. (c) MODEL 1: White's test for Ho: homoskedasticity against Ha: unrestricted heteroskedasticity chi2(2) = 14.03 Prob > chi2 = 0.0009 For Model1 the White test stat is bigger than the chi square value, at better than 1% significance level. Therefore we reject the null hypothesis Cameron & Trivedi's decomposition of IM-test --------------------------------------------------- Source | chi2 df p ---------------------+----------------------------- Heteroskedasticity | 14.03 2 0.0009 Skewness | 1.64 1 0.2006 Kurtosis | 4.76 1 0.0292 ---------------------+----------------------------- Total | 20.42 4 0.0004 --------------------------------------------------- MODEL 2: White's test for Ho: homoskedasticity against Ha: unrestricted heteroskedasticity chi2(2) =

More about Econ Essay

Open Document