f Sr. No. of Question Paper: 8537 HC Unique Paper Code 32377908 Name of the Paper Econometrics Name of the Course STATISTICS : DSE for Honours Semester Vv Duration : 3 Hours Maximum Marks : 75 Instructions for Candidates 1. Write your Roll No. on the top immediately on receipt of this question paper. Attempt five questions in all. Question 1 is compulsory. Attempt four more questions, selecting two questions from each of the Section A and B. (2) Fill in the blanks (Attempt any five) : () In a regression analysis, values are fixed for the ———_ variables. tt) Variables such as gender, marital status, colour of the eye are examples of scales. P.T.O.ly a (ii) Multicollinearity is essentially © phenomenon. (iv) A regression model that includes the lagged values of explanatory variables is called a —____J model. (v) The coefficients estimated in the presence of Heteroscedasticity are not ————— estimators, (vi) If the Durbin-Watson d test statistic is found to be equal to 0, this means that ———___ order autocorrelation is (vii) Structural parameters express the —____. effect of each explanatory variable on the dependent variable. ® (b) State whether True/False (Attempt any five): (i) The smallest possible multiple regression model comprises of two variables. (ii) Stochastic variables are non-random variable - ; =o can (iii) A hypothesis such as Hy: B, = Ps ~ tested using an F-test. ‘ os, We! (iv) Ifa qualitative variable has m categor’® to introduce (m- 1) dummies. av’ (vy) Errors of measurem, ent is not a cause for model SP€Cification errors. (vi) In Koyck method, B,= Bo AR k=0, 1,. ~and0<)<], (1~A) is known as Tate o f decay of distributed (vii) The measure of proportion of variation in ¥. explained by the explanatory variables X,, X;, aot jointly is given by R?, . (5) (c) Give short answers (Attempt any two): (i) Derive the relationship between the coefficient of determination and F-test used in the analysis of variance. (ii) From a cross-sectional data on 41 countries, the following regression was obtained : In Y;=f, + B,In X,, + B, In X3, + Uy; Running the auxiliary regression 655.8417 + 2.5629 In X,; + 0.6918 InX,; — 0.4081 (InX,,P ~0.049 (InX,,)? + 0.0015(In X,,)(In X,,) R?= 0.1148 Test the presence of Heteroscedasticity sta 9 U 0° White's test. (Chi-square value at 5% leve igni i .0705 Significance is 11.0705) preaq $537 2. ii) I terpret the following estimated regression ny, (iii) In ¥ = -1.9356 + 0.1608 X, + 0.1995 X, (0.033) (0.1858) .001 (0.300) R*=0x p-value = (<0.001) a SECTION - A Is. If (i) Define Structural and Reduced form models. Q4= By + BYP + BY + U Q,= a + o,P + V QQ = Q .g for is demand-supply model, then find the reduc? model. i i ‘| jn he (i) Stating clearly the underlying assumptions, obta oa : i arial restricted least squares estimator of B and its 2h : : i + i the set of linear restrictions given bY RP for the GLM as Nes ma XS 8 +U. ) What ¢ 7 to YPES of mode] Specification errors is one lik ©ncounter in 7 c ic! Tact yer hc “ice? What are the consed Televant wars. .o of i@ @ (i) = ( fly the Problem of Multicollinearity and its ©onsequences Describe any two Temedial measures. (7,8) Discuss the types of data available for analysis, giving Suitab econometric le examples. In the case of General Linear model Y=XB+ U, ‘ator (BLUE) of B, clearly Stating all the assumptions. Comment on the MLE of 8. (6,9) obtain the Best Linear Unbiased Estim SECTION - B What happens if Ordinary Least Squares Sstinstors are applied to a model suffering from oNteconseinet disturbances? Discuss its consequences on me Procedure. Describe Durbin-Watson test or of autocorrelation by stating its assumptions. ure for the Explain the Durbin two-stage proved (10,5) : ers. ‘stimation of autocorrelation parame ; be n should What are dummy variables? What votes? Undertaken in the use of these varia P.T.O.8537 (ii) What do you understan 7. (i) (ii) d by lag? In the Geometr Lag Model : where B,=ByAt, OS 2 <1, discuss Koyck’s sches ers to be estimated of reducing the number of paramet (69 +U For a generalized least squares model y=xpr! with E(UU')=072, © being known symmet" p and find | positive definite matrix, estimate 2 ator for %* variance. Also, find unbiased estim: it ite’s 1 Define Heteroscedasticity and discuss aa (61) ie for Heteroscedasticity,3 (ii) Heteroscedasticity is more common in time series data as compared to cross sectional data. (iii) In Run’s test we assume that the number of positive and negative residuals should be at least 10. (iv) The graph plottec ween e, and @, can be used to indicate the presence of heteroscedasticity in the data. (v) In simultaneous equation model we may have more than one endogenous variables. (vi) When qualitative variable has five categories then we need to use five dummy variables. ‘ss 2 (vii) Farrar Glauber test is a combination of ¥ test, ¢ test and F test. (5) (c) Give short answers (attempt any two) (i) The residuals ¢; obtained from a regression satisfy the following resultsif e 4. (a) Discuss the method based on Frisch Confluence analysis. How do you use “transformation of variables” and “addition of new data” as remedial measures of multicollinearity? (b) Elaborate the following with the help of appropriate examples : (i) Statistical vs. Deterministic relationships. Gi) Regression vs. Correlation. (10,5) be SECTION -B - (a) Consider a simple linear model : ¥=f+,X,+u, for all i =1,2,...,7 with E(u,)=0, Var(u,) =07 / A, and Cov(u,,u,)=0Vi# j where g? is known and 4's are known positive numbers. Assuming the heteroscedastic structure EW?) =0°X Ni, obtain the variance of OLS estimator and Aitken’s estimator for #,. Comment on the efficiency of the estimators. (b) What is the distributed lagged model? We ye the difficulties in performing OLS estimation a parameters in the distributed lagged model? Describe three main reasons for ‘lag’ in the a pTosources of heteroscedasticity? Discuss Sp “ ‘ earman’ rank correlation test for detecting heteroncadatl (b) Describe Durbin Watson test for the detection of autocorrelation specifying the assumptions under which the test is valid. (8,7) 7. (a) Consider the regression model ! ¥,=B,+B,D, + BX, +B(DX) +m where, Y is the expenditure on food, X is after tax income, D,=1 for female and D,=0 for male. Obtain the expressions for mean food expenditure of males and females respectively. Identify the ntercept and differential slope on the statistical cients, discuss the f expenditure — Also draw differential i coefficients. Depending significance of these coeffi nature of two lines of regression 0 for males and females respectively: the lines of regression in each case. i procedure fosestion paper contains 8 printed pages] Roll No. ree m.ofQ.Paper +: 636 I Paper Code —-: 32377908 of the Course : B.Sc.(Hons.) = : Statistics : DSE-I of the Paper : Econometrics636 }) Write True/False (Attempt any five) : () 5 In the multiple regression model the adjusted R? equals the square of the correlation coefficient ‘r’. (4) Reordering of observations with respect to explanatory variable is the first step in conducting Goldfeld-Quandt test for heteroscedasticity. (iii) If in the regression model, one of the se as dary variables included is the lagged value of the dependent variable then the model is referred to as autoregressive model, (iv) Coefficient of overfitted model will have piased coefficient. 3 ‘a P.T.O.636 (iii) Consider the model having only dumm variable fo y ¥, = 3176.833 - 503.1167 D, p-value (<0.001) (0.158) where Y, = Expenditure on food : {1, for female Bs 1 O, for male If one explanatory variable X, is added to the model, then the result becomes Y =1506.244— 228.9868 D,+ 0.0589 X, p-value (<0.001) (0.061) (<0.001) where X, = after tax income ,n = 12 What is the reason for the dummy coefficient to be statistically insignificant in the first case and significant in the second case ? Section - A 2. (a) Discuss, in brief, the methodology of doyeadping anieconpmenic model, 6 P.T.O.- Discuss Frisch’s Confluence ane cr ~ detecting the presence of wautlines i6. (a) Explain With suitabl fen, the variance of B(OLSE) and 5 (GLSE) Pen pasting that the variance of the distur- i cs © examples, the tative variable with Tegression of one quali More than two Categories. . 8 fb) Lety = xp +u, E(U) = 0, E( UU) = g2o 1/0 + 9 0 ye 0 Where 2 = i ios of |, a'sare known Ons Onis ae Positive numbers and 6? is unknown. Find € term is proportional to X i.e. E(U?) = oi 2 te 7 i 7, P.T.O.