Question5 You are estimating a time series model for stock market returns of UK
ID: 3350229 • Letter: Q
Question
Question5 You are estimating a time series model for stock market returns of UK insurance companies. The dataset comprises of weekly returns on the insurance index and returns on the FTSE100 over 1986:01 2001:12. You assume that insurance returns (Rinsur) can be adequately explained by the current FTSE returm, (FTSE100, and market volatility (VOLAT), respectively. Accordingly your model and the corresponding EViews output using OLS estimation are as follows: RInsur, = 0+,(FTSE100), + 2(VOLAT),+, Dependent Variable: RINSUR Method: Least Squares Date: 03/13/06 Time: 18:22 Sample (adjusted): 1/10/1986 12/21/2001 Included observations: 833 after adjustments Variable Coefficient Std. Errer t-Statistic Prob. FTSE100 VOLAT 0.113777 0089407 .102629 0038110 0.082345 0.038090 1.272571 28.93243 2.161879 0.2035 0.0000 0.0309 R-squared Adjusted R-squared S.E. of regression Sum squared resid Log likelhood Durbin-Watson stat 0.503237 Mean dependent var .502040 S.D. dependent var 2.568870 Akaike info criterion 477.248 Schwarz criterion 1966.380 F-statistic 2.254550 Prob(F-statistic) 0.074034 3.640367 4.728404 4.745421 420.4092 0.000000 Breusch Godfrey Serial Correlation LM Test: (4 lags) F-statistkc 8911 Probability 0.0000 Bera Test F-statistc 86.1963 Probability 0.0000 White Test F-statistic 9.2531 Probability 0.0551Explanation / Answer
a). The p value of independnet variables FTSE100 and VOLAT and the that of F statistic are signinficant at 95 % confidence levels since p value of each is <0.005
The DW statisitc is 2.254 .
For n =16*52 =832 data samples (16 years ,52 week each) for 2 variables the DW criritcal is ~1.70
Clearly, the DW statistic> DW crtical hence there is negetive auto correlation .(DW statisitc > 2 has nenegtive auto correlation).
c). The residual plot is quite random (+ve ,-ve in random) indicating that hte linear model provides for a descent fit.
The residuals get larger as the prediction moves from small to large (or from large to small). ie in other words, the residual graph is indicating heteroscedasticity.
Error is limited to -5 to +5 with large number of outlier values while moving towards month .Also observe the R2 value. It indicates that only roughly half of the data set variance is accounted for by the regression model.
The problem with the residual plot is htat owing to the large variablity ,it tends to shift the regression line .
d).We can improve the specifications of the model by:-
a). converting it ot Log-Log model.
b). Choosing explanatory variables
c).Include dummy variables to improve on approximation.
d). Keep on refining the regression model by including the errors.
e).Refine predictors and check model fit
f).Check for
• Multicollinearity
• Outliers and influential points
• Missing data
• Truncation and censoring
and remove.
g).If two or more predictors overlap in how they explain an outcome, that overlap won’t be reflected in either regression coefficient.
h).Start with univerate discriptives and then move to biveriate discriptives .
Related Questions
drjack9650@gmail.com
Navigate
Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.