Question5 A model has been estimated using data for different markets and the fo

Question

Question5 A model has been estimated using data for different markets and the following diagnostic results have been obtained for each. Explain what the problem for your model is and what you would do to rectify it, in each of the following scenarios Note: all scenarios are based on different series, ie there is no connection between problems) a. 19E 248376 s0103884 1049 247028[5 marks] 02 .1 00 0 02 03 04 b. White Heterosk Test: F-statistic 3.394480 Probability 0.003734 Obs'R-squared 20.50718 Probability 0.008578 [6 marks] C. Ramsey RESET Test 57.14435 Probability 0.000000 Log likelihood ratio 93.92522 Probability 0.000000 (6 marks] d. Plot of residuals from original regression: 0.04 006 [6 marks] Question 5 Continued o

Explanation / Answer

a) it gives summary of data.Summary of data means minimum point of data, maximum point of data, its average, standard deivaition etc.

b)heteroscedasticity is the variation of variable is unequal across the range of value of a predictable variable. since, Probability is less than 0.05 we reject Ho. so, heteroscedasticity is there. One of the imp assumption of regression is that, there should be no heteroscedasticity. in short, variance of residuals should not increase with fitted values of response variable. You can rectify it with Box-Cox transformation.

c) Regression Equation Specification Error test. It states that Higher the F value high chances of wrong Functional form. it also tells you you can improve specification of your model.you can improve it by adding Squared term.

d) Residual plot ia graph that shows residuals on y axis and independent variable on x axis. if the points are randomly dispersed around x axis then linear model is best, otherwise non linear model will be best.

e)Since, Probability is greater than 0.05, We do not reject Ho. so, there are chances of LNFX has unit root.

f)we can say that from first column that series are both persistent, that autocorrelation functions that only die away slowly. autocorrelation is imp assumption of time series model not for regression model

Related Questions

Navigate

• Browse (All)
• Browse Q
• Subjects

---

---

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.