* R. Carter Hill, William E. Griffiths and Guay C. Lim, * Principles of Econometrics, Fourth Edition, Wiley, 2011. * Chapter 14 SAMPLE 1 500 READ (byd.dat) / names * Chapter 14.3.1 Testing for ARCH effects, pages 523-4. STAT r / MEAN=mr * Jarque-Bera test for non-normality, page 524. OLS r / GF GENR ehat2=(r-mr)**2 * Estimate an auxiliary regression OLS ehat2 ehat2(1.1) * Calculate an LM test value as a test for ARCH effects. * Note: the auxiliary regression used 499 observations. GEN1 lm=(\$N-1)*\$R2 * Calculate a p-value DISTRIB lm / TYPE=CHI DF=1 GEN1 pvalue=1-\$CDF PRINT lm pvalue * ARCH(1) estimation, page 524. HET r / ARCH=1 RESID=E STDRESID=EH * Numeric differences in standard errors compared to the textbook * may be due to estimation algorithm differences. * Get the conditional variance function GENR h1=(E/EH)**2 * GARCH(1,1) estimation, page 526. HET r / GARCH=1 RESID=E STDRESID=EH PRESAMP GENR h2=(E/EH)**2 * Plot the conditional variance functions * ARCH(1) Figure 14.6, page 525. GRAPH h1 / TIME LINEONLY * GARCH(1,1) Figure 14.7 (b), page 527. GRAPH h2 / TIME LINEONLY * -------------- Model Variations -------------- * ARCH-in-mean model HET r / ARCHM=1 ARCH=1 NUMCOV COEF=B * GARCH-in-mean model * Use the estimated coefficients from the ARCH-in-mean model as * starting values. GEN1 B:5=0 HET r / ARCHM=1 GARCH=1 PRESAMP NUMCOV START=B RESID=E STDRESID=EH * Calculate and plot the conditional variance function GENR h3=(E/EH)**2 GRAPH h3 / TIME LINEONLY * Warning: The ARCH estimation results must be carefully checked for * satisfactory model convergence. STOP