Springer Texts in Business and Economics

The backup regressions are given below: These are performed using SAS

OLS REGRESSION OF LNC ON CONSTANT, LNP, AND LNY Dependent Variable: LNC

Analysis of Variance

RootMSE 0.16343 R-square 0.3037

DepMean 4.84784 Adj R-sq 0.2713

C. V. 3.37125

b. Regression for Glejser Test (1969)

Dependent Variable: ABS_E MODEL: Z1=LNYA(—1/

Analysis of Variance

MODEL: Z4=LNYA1 Dependent Variable: ABS_E

Analysis of Variance

c. Regression for Goldfeld and Quandt Test (1965) with first 17 obervations

Dependent Variable: LNC

SPEARMAN RANK CORRELATION TEST

OBS R T

e. Harvey’s Multiplicative Heteroskedasticity Test (1976)

Dependent Variable: LNE_SQ

HARVEY’S MULTIPLICATIVE HETEROSKEDASTICITY TEST (1976)

OBS HV. TEST

1 2.90997

f. Regression for Breusch and Pagan Test (1979)

Dependent Variable: X2

Analysis of Variance

RootMSE 1.30804 R-square 0.1272

DepMean 1.00001 Adj R-sq 0.1074

C. V. 130.80220

BREUSCH & PAGAN TEST (1979)

OBS RSSBP

1 5.48535

g. Regression for White Test (1979)

Dependent Variable: E_SQ

Normality Test (Jarque-Bera) This chart was done with EViews.

SAS PROGRAM Data CIGAR;

Input OBS STATE $ LNC LNP LNY; CARDS;

Proc reg data=CIGAR;

Model LNC=LNP LNY;

Output OUT=OUT 1 R=RESID;

Proc Plot data=OUT1 hpercent=85 vpercent=60;

Plot RESID*LNY=‘*’; run;

***** GLEJSER’S TEST (1969) *****;

*****************************************,

Data GLEJSER; set OUT1;
ABS_E=ABS(RESID);

Z1=LNY**-1;

Z2=LNY**-.5;

Z3=LNY**.5;

Z4=LNY;

Proc reg data=GLEJSER;

Model ABS_E=Z1;

Model ABS_E=Z2;

Model ABS_E=Z3;

Model ABS_E=Z4;

TITLE ‘REGRESSION FOR GLEJSER TEST (1969)’; LABEL Z1=‘LNY“(-1)’

Z2=‘LNY“(-0.5)’

Z3=‘LNY“(0.5)’

Z4=‘LNY“(1)’;

run;

***** GOLDFELD & QUANDT TEST (1965) *****; *******************************************************.

Proc sort data=CIGAR out=GOLDFELD;

By LNY;

Data GQTEST1; set GOLDFELD;

If_N_<18; OBS=_N_;

Data GQTEST2; set GOLDFELD;

If _N_>29; OBS=_N_-29;

Proc reg data=GQTEST1;

Model LNC=LNP LNY;

Output out=GQ_OUT1 R=GQ_RES1;

TITLE ‘REGRESSION FOR GOLDFELD AND QUANDT TEST (1965) w/ first 17 obs’;

Proc reg data=GQTEST2;

Model LNC=LNP LNY;

Output out=GQ_OUT2 R=GQ_RES2;

TITLE ‘REGRESSION FOR GOLDFELD AND QUANDT TEST (1965) w/last 17 obs’;

run;

***** SPEARMAN’S RANK CORRELATION TEST *****; ***************************************************************.

Data SPEARMN1; set GOLDFELD;

RANKY=_N_;

Proc sort data=GLEJSER out=OUT2;

By ABS_E;

DataTEMP1; setOUT2;

RANKE=_N_;

Proc sort data=TEMP1 out=SPEARMN2;

By LNY;

Data SPEARMAN;

Merge SPEARMN1 SPEARMN2;

By LNY;

D=RANKE-RANKY;

* Difference b/w Ranking of —RES— and Ranking of LNY; D_SQ=D**2;

Proc means data=SPEARMAN NOPRINT;

Var D_SQ;

Output out=OUT3 SUM=SUM_DSQ;

Data SPTEST; Set OUT3;

R=1-((6*SUM_DSQ)/(46**3-46));

T=SQRT (R**2*(46-2)/(1-R**2));

Proc print data=SPEARMAN;

Var STATE LNC RANKY ABS_E RANKE D;

TITLE ‘DATA FOR SPEARMAN RANK CORRELATION TEST’;

Proc print data=SPTEST;

Var R T;

TITLE ‘SPEARMAN RANK CORRELATION TEST’; run;

‘HARVEY’S MULTIPLICATIVE HETEROSKEDASTICITY TEST (1976) **;

A**********************************************************************************.

Data HARVEY; setOUTI;

E_SQ=RESID‘‘2;

LNE_SQ=LOG(E_SQ);

LLNY=LOG(LNY);

Proc reg data=HARVEY;

Model LNE_SQ=LLNY;

Output out=OUT4 R=RLNESQ;

TITLE ‘HARVEY’ ‘S MULTIPLICATIVE HETEROSKEDASTICITY TEST (1976)’;

Data HARVEY1; set OUT4;

HV_TEMP=LNE_SQ‘‘2-RLNESQ‘‘2;

Proc means data=HARVEY1;

Var HV_TEMP LNE. SQ;

Output out=HARVEY2 SUM=SUMTMP SUMLNESQ;

Data HARVEY3; set HARVEY2;

HV_TEST=(SUMTMP-46* (SUMLNESQ/46)‘‘2)/4.9348;

Proc print data=HARVEY3;

Var HV_TEST;

TITLE ‘HARVEY’ ‘S MULTIPLICATIVE HETEROSKEDASTICITY TEST (1976)’;

***** BREUSCH & PAGAN TEST (1979) *****; ***************************************************,

Proc means data=HARVEY;

Var E_SQ;

Output out=OUT5 MEAN=S2HAT;

TITLE ‘BREUSCH & PAGAN TEST (1979)’;

Proc print data=OUT5;

Var S2HAT;

Data BPTEST; set HARVEY;

X2=E_SQ/0.024968;

Proc reg data=BPTEST;

Model X2=LNY;

Output out=OUT6 R=BP_RES;

TITLE ‘REGRESSION FOR BREUSCH & PAGAN TEST (1979)’;

Data BPTEST1; set OUT6;

BP_TMP=X2**2-BP_RES**2;

Proc means data=BPTEST1;

Var BP_TMP X2;

Output out=OUT7 SUM=SUMBPTMP SUMX2;

Data BPTEST2; set OUT7;

RSSBP=(SUMBPTMP-SUMX2**2/46)/2;

Proc print data=BPTEST2;

Var RSSBP;

***** WHITE’S TEST (1980) *****;

*************************************.

Data WHITE; set HARVEY;

LNP_SQ=LNP**2;

LNY_SQ=LNY**2;

LNPY=LNP*LNY;

Proc reg data=WHITE;

Model E_SQ=LNP LNY LNP. SQ LNPY LNY_SQ; TITLE ‘REGRESSION FOR WHITE TEST (1979)’; run;