Final Exam (0900-1200hr, January 29, 2003)

2940603 Advanced Econometrics (Assoc. Prof. Pongsa Pornchaiwiseskul)

Instructions:

a) Textbooks, lecture notes and calculators are allowed.

b)  Each must work alone. Cheating will not be tolerated.

c)  There are four tests. Attempt all the tests. Use only the provided test-books.

d)  All the hypothesis testing will use 0.05 as the level of significance.

TEST#1 (20 points)

During week t, weekly stock return (RBt) of banking sector is assumed to follow the following process:

RBt = b0 +b1Dt-1+ b2RMt + ut

[st]2 = a0 + a1[ut-1]2 + a2Dt-1

where

RMt = market return in period t

Dt = 1 if sectoral return is less than the market return

in period t

= 0, otherwise.

ut = independently distributed error term

[st]2 = Var(ut)

Use printouts 1.1-1.3 to answer the following questions.

1.1) Give a valid estimate for the above model. Explain in details.

1.2)  It has been claimed that the expected return and the return risk of the banking sector do not depend on whether the observed sectoral return is above or below the market return. Test this claim.

TEST#2 (20 points)

Unemployment rate (UE) and domestic inflation rate (DI) form a VAR(1) as follows:

UEt = b10 + b11 UEt-1 + b12 DIt-1 + e1t

DIt = b20 + b21 UEt-1 + b22 DIt-1 + e2t

where (e1t , e2t) ~ iid bi-variate of shocks

Use printouts 2.1-2.2 to answer the following questions.

2.1)  Estimate the model. Check for the validity. Explain in details

2.2)  Test whether the long-run correlation between the unemployment rate and the inflation rate is negative. Explain in details.

2.3)  How much will the future shocks contribute to the variance of the unemployment rate in 2 periods from now? This is a variance decomposition question with simultaneous effects.

TEST#3 (20 points)

Let Pt be the international price in term of local currency in period t. It is assumed that P follows the following model:

lnPt = b1+ b2lnEt + ut

where

Et = the foreign exchange rate in period t

ut = stationary ARMA(1,2) error term in period t

ut = rut-1 + et + q1e t-1+ q2e t-2

e t = white noise with variance of s2

Based on Printouts 3.1 and 3.2, answer the following questions:

3.1)  Estimate all the parameters and the variance of the estimates. Check for validity.

3.2)  Given the following information,

T / lnPt / lnEt
26 / 2.0 / 2
27 / 2.3 / 2.5
28 / ? / 2.3

calculate the best prediction for lnP28 and its prediction interval.

TEST#4 (20 points)

We are suspecting that endogenous time series (Y1t,Y2t) follow VAR(1) with exogenous variables (X1t,X2t,X1t-1,X2t-1). Explain how you will test it.

PRINTOUT 1.1

======

Dependent Variable: RB

Method: ML - ARCH

Date: 01/28/03 Time: 13:14

Sample(adjusted): 2 200

Included observations: 199 after adjusting endpoints

Convergence achieved after 22 iterations

======

CoefficientStd. Errorz-Statistic Prob.

======

C -0.002699 0.145566 -0.018543 0.9852

RB(-1)<RM(-1) -0.152600 0.137169 -1.112497 0.2659

RM 24.08926 23.91289 1.007375 0.3138

======

Variance Equation

======

C 1.009517 0.074876 13.48244 0.0000

ARCH(1) -0.098778 0.042461 -2.326308 0.0200

RB(-1)<RM(-1) 0.046743 0.180000 0.259685 0.7951

======

R-squared 0.019704 Mean dependent var 0.053044

Adjusted R-squared -0.005692 S.D. dependent var 1.004412

S.E. of regression 1.007267 Akaike info criteri2.837820

Sum squared resid 195.8153 Schwarz criterion 2.937116

Log likelihood -276.3631 F-statistic 0.775859

Durbin-Watson stat 1.977487 Prob(F-statistic) 0.568203

======

PRINTOUT 1.2

======

Dependent Variable: RB

Method: ML - ARCH

Date: 01/28/03 Time: 13:16

Sample: 1 200

Included observations: 200

Convergence achieved after 32 iterations

======

CoefficientStd. Errorz-Statistic Prob.

======

SQR(GARCH) -0.682372 0.562348 -1.213432 0.2250

C 0.545380 0.554730 0.983144 0.3255

RM 33.48573 23.21488 1.442426 0.1492

======

Variance Equation

======

C 1.036749 0.101894 10.17474 0.0000

ARCH(1) -0.085844 0.030997 -2.769479 0.0056

(RESID<0)*ARCH(1) -0.045022 0.047190 -0.954060 0.3401

======

R-squared 0.015897 Mean dependent var 0.052978

Adjusted R-squared -0.009466 S.D. dependent var 1.001886

S.E. of regression 1.006617 Akaike info criteri2.844992

Sum squared resid 196.5758 Schwarz criterion 2.943942

Log likelihood -278.4992 F-statistic 0.626776

Durbin-Watson stat 1.820597 Prob(F-statistic) 0.679525

======

PRINTOUT 1.3

======

Dependent Variable: RB

Method: ML - ARCH

Date: 01/29/03 Time: 13:16

Sample: 1 200

Included observations: 200

Convergence achieved after 26 iterations

======

CoefficientStd. Errorz-Statistic Prob.

======

C -0.033348 0.138380 -0.240988 0.8096

RM 22.53198 23.51115 0.958353 0.3379

======

Variance Equation

======

C 1.037565 0.102964 10.07699 0.0000

ARCH(1) -0.098655 0.010456 -9.435508 0.0000

======

R-squared 0.007892 Mean dependent var 0.052978

Adjusted R-squared -0.007293 S.D. dependent var 1.001886

S.E. of regression 1.005533 Akaike info criteri2.839077

Sum squared resid 198.1748 Schwarz criterion 2.905044

Log likelihood -279.9077 F-statistic 0.519739

Durbin-Watson stat 1.861711 Prob(F-statistic) 0.669179

======

PRINTOUT 2.1

Vector Autoregression Estimates

======

Date: 01/28/03 Time: 13:22

Sample(adjusted): 2 60

Included observations: 59 after adjusting

endpoints

Standard errors & t-statistics in parentheses

======

UE DI

======

UE(-1) 0.484077 -0.186778

(0.10345) (0.08903)

(4.67931) (-2.09803)

DI(-1) -0.312070 0.335431

(0.10708) (0.09215)

(-2.91425) (3.63996)

C 0.542541 0.227590

(0.18083) (0.15561)

(3.00031) (1.46253)

======

R-squared 0.382675 0.263981

Adj. R-squared 0.360627 0.237694

Sum sq. resids 67.11876 49.70590

S.E. equation 1.094783 0.942128

F-statistic 17.35697 10.04248

Log likelihood -87.52069 -78.66066

Akaike AIC 3.068498 2.768158

Schwarz SC 3.174136 2.873796

Mean dependent 0.973097 0.106559

S.D. dependent 1.369150 1.079059

======

Determinant Residual Covaria 0.950494

Log Likelihood -165.9369

Akaike Information Criteria 5.828370

Schwarz Criteria 6.039645

======

PRINTOUT 2.2

Johansen Cointegration Test

======

Date: 01/28/03 Time: 13:28

Sample: 1 60

Included observations: 58

Test assumption: Linear deterministic trend in the data

Series: UE DI

Lags interval: 1 to 1

======

Likelihood 5 Percent 1 Percent Hypothesized

Eigenvalue Ratio Critical ValuCritical ValuNo. of CE(s)

======

0.389833 34.99342 15.41 20.04 None

0.103549 6.340073 3.76 6.65 At most 1

======

*(**) denotes rejection of the hypothesis at 5%(1%) significance level

Unnormalized Cointegrating Coefficients:

======

UE DI

0.075458 0.138147

0.090885 -0.039561

======

Normalized Cointegrating Coefficients: 1 Cointegrating Equation(s)

======

UE DI C

1.000000 1.830790 -1.172253

(0.44837)

Log likelihood-165.0593

======

PRINTOUT 3.1

======

Dependent Variable: LOG(P)

Method: Least Squares

Date: 01/28/03 Time: 13:39

Sample(adjusted): 2 27

Included observations: 26 after adjusting endpoints

Convergence achieved after 19 iterations

Backcast: 0 1

======

Variable CoefficientStd. Errort-Statistic Prob.

======

C 0.556897 0.167547 3.323828 0.0032

LOG(E) -0.404976 0.073326 -5.522935 0.0000

AR(1) 0.578664 0.301539 1.919035 0.0687

MA(1) -0.977222 0.497775 -1.963178 0.0630

MA(2) -0.011509 0.362611 -0.031740 0.9750

======

R-squared 0.637016 Mean dependent var 0.423882

Adjusted R-squared 0.567876 S.D. dependent var 0.283021

S.E. of regression 0.186047 Akaike info criter-0.354592

Sum squared resid 0.726884 Schwarz criterion -0.112650

Log likelihood 9.609693 F-statistic 9.213453

Durbin-Watson stat 2.015074 Prob(F-statistic) 0.000184

======

Inverted AR Roots .58

Inverted MA Roots .99 -.01

======

PRINTOUT 3.2

Coefficient Covariance Matrix

======

C LOG(E) AR(1) MA(1) MA(2)

======

C 0.028072 -3.90E-05 0.030094 -0.064432 0.033179

LOG(E) -3.90E-05 0.005377 -0.000615 -0.005995 0.003812

AR(1) 0.030094 -0.000615 0.090926 -0.134770 0.099009

MA(1) -0.064432 -0.005995 -0.134770 0.247780 -0.171672

MA(2) 0.033179 0.003812 0.099009 -0.171672 0.131487

======

End of Exam

MEcon Thai Program, Faculty of Economics, Chulalongkorn University Page 2/2