The Great Moderation and the relationship between output growth and its volatility.

[not equal to] post-1982 t test 0.7024 Structural stability test for [H.sub.0]: pre-1982 = post-1982 unconditional variance [H.sub.1]: pre-1982 > post-1982 F test 3.7772 [0.0000] Standard errors appear in parentheses; p values appear in brackets. The measures of skewness and kurtosis are normally distributed as N(0,6/T) and N(0,24/T), respectively, where T equals the number of observations. LB Q(k) equals a Ljung-Box statistic testing for autocorrelations in growth up to k lags. LM (k) is the Lagrange multiplier test for conditional heteroscedasticity, distributed asymptotically as [chi square](k). ADF(n) equals the augmented Dickey-Fuller unit-root test with n lags selected by the Schwarz Criterion. The t statistic tests for structural change in the mean between samples i and j. The F statistic equals the variance-ratio test between samples i and j, asymptotically distributed as F([df.sub.i],[df.sub.j]), where df denotes degrees of freedom. * 5% significance level. ** 10% significance level. Table 2. GARCH-M Results without Structural Break [y.sub.t] = [a.sub.0] + [a.sub.1][y.sub.t-1] + [a.sub.2][y.sub.t-2] + [lambda][[sigma].sub.t] + [[epsilon].sub.t]; and [[sigma].sup.2.sub.t] = [[alpha].sub.0] + [[alpha].sub.1][[epsilon].sup.2.sub.t-1] + [[beta].sub.1][[alpha.sup.2.sub.t-1]

[a.sub.0] [a.sub.1] [a.sub.2] [lambda]

0.4267 * 0.2371 * 0.2002 * 0.0911

(0.1648) (0.0692) (0.0725) (0.1960)

LB Q (3) LB Q (6) LB [Q.sup.2] (3) LB [Q.sup.2] (6)

1.6828 4.2863 0.7517 6.7173

[0.6407] [0.6379] [0.8609] [0.3477] [[alpha].sub.0] [[alpha].sub.1] [[beta].sub.1]

0.0106 0.1606 * 0.8355 *

(0.0123) (0.0492) (0.0434)

Skewness Kurtosis Jarque-Bera LR

0.1388 0.8898 * 8.5442 * 0.0213

[0.3870] [0.0059] [0.0139] [0.8839] Maximum log-likelihood function value: -76.5102 ARCH-M Results without Structural Break [y.sub.t] = [a.sub.0] + [a.sub.1][y.sub.t-1] + [a.sub.2][y.sub.t-2] + [lambda][[sigma].sub.t] + [[epsilon].sub.t]; and [[sigma].sup.2.sub.t] = [[alpha].sub.0] + [[alpha].sub.1][[epsilon].sup.2.sub.t-1] + [[beta].sub.1][[alpha.sup.2.sub.t-2]

[a.sub.0] [a.sub.1] [a.sub.2] [lambda]

0.3891 ** 0.3223 * 0.1611 * 0.1130

(0.2247) (0.0755) (0.0651) (0.2497)

LB Q (3) LB Q (6) LB [Q.sup.2] (3) LB [Q.sup.2] (6)

2.6000 5.8491 1.0033 8.7188

[0.4575] [0.4403] [0.8004] [0.1904] [[alpha].sub.0] [[alpha].sub.1] [[beta].sub.1]

0.3661 * 0.3107 * 0.3730 *

(0.0564) (0.0883) (0.1190)

Skewness Kurtosis Jarque-Bera LR

-0.0293 1.2352 * 15.0378 * 2.6977

[0.8551] [0.0001] [0.0005] [0.1005] Maximum log-likelihood function value: -85.6076 Standard errors appear in parentheses; p values appear in brackets; LB Q(k) and LB Q2(k) equal Ljung-Box Q-statistics tests for standardized residuals and squared standardized residuals for autocorrelations up to k lags. LR equals the likelihood ratio statistic following a [chi square] distribution with one degree of freedom that tests for [[alpha].sub.1] + [[beta].sub.1] = 1 or [[alpha].sub.1] + [[alpha].sub.2] = 1. * 5% significance level. ** 10% significance level. Table 3. GARCH-M Results without Structural Break at 1982: I [y.sub.t] = [a.sub.0] + [a.sub.1][y.sub.t-2] + [a.sub.2][y.sub.t-2] + [lambda][[sigma].sub.t] + [[epsilon].sub.t]; and [[sigma].sup.2.sub.t] = [[alpha].sub.0] + [[alpha].sub.1][[epsilon].sup.2.sub.t-1] + [[beta].sub.1][[alpha.sup.2.sub.t-1] + [gamma] Dummy, where Dummy = 1 for t [greater than or equal to] 1982:I; 0 otherwise.

[a.sub.0] [a.sub.1] [a.sub.2] [lambda]

0.4427 * 0.2634 * 0.1757 * 0.0519

(0.1361) (0.0694) (0.0726) (0.1633)

LB Q (3) LB Q (6) LB [Q.sup.2] (3) LB [Q.sup.2] (6)

2.4247 5.4342 0.8279 4.9614

[0.4890] [0.4894] [0.8427] [0.5487] [[alpha].sub.0] [[alpha].sub.1] [[beta].sub.1] [gamma]

0.3948 * 0.0801 * 0.6271 * -0.3322 *

(0.1958) (0.0692) (0.1767) (0.1651)

Skewness Kurtosis Jarque-Bera LR

0.0944 0.1846 0.6862 17.5179 *

[0.5560] [0.5684] [0.7095] [0.0000] Maximum log-likelihood function value: -64.1225 ARCH-M Results with Structural Break at 1982:I [y.sub.t] = [a.sub.0] + [a.sub.1][y.sub.t-2] + [a.sub.2][y.sub.t-2] + [lambda][[sigma].sub.t] + [[epsilon].sub.t]; and [[sigma].sup.2.sub.t] = [[alpha].sub.0] + [[alpha].sub.1][[epsilon].sup.2.sub.t-1] + [[alpha].sub.2][[epsilon].sup.2.sub.t-2] + [gamma] Dummy, where Dummy = 1 for t [greater than or equal to] 1982:I; 0 otherwise.

[a.sub.0] [a.sub.1] [a.sub.2] [lambda]

0.4644 * 0.2880 * 0.1559 * 0.0254

(0.1366) (0.0751) (0.0714) (0.1619)

LB Q (3) LB Q (6) LB [Q.sup.2] (3) LB [Q.sup.2] (6)

2.3699 5.3571 0.7853 4.2135

[0.4992] [0.4988] [0.8529] [0.6478] [[alpha].sub.0] [[alpha].sub.1] [[alpha].sub.2] [gamma]

0.9886 * 0.1891 * 0.0959 * -0.8191 *

(0.1843) (0.0844) (0.1125) (0.1754)

Skewness Kurtosis Jarque-Bera LR

0.1307 0.1385 0.8607 18.1591 *

[0.6967] [0.4153] [0.6502] [0.0000] Maximum log-likelihood function value: -64.7860 GARCH-M Results with Structural Break at 1984:I [y.sub.t] = [a.sub.0] + [a.sub.1][y.sub.t-1] + [a.sub.2][y.sub.t-2] + [lambda][[sigma].sub.t] + [[epsilon].sub.t]; and [[sigma].sup.2.sub.t] = [[alpha].sub.0] + [[alpha].sub.1][[epsilon].sup.2.sub.t-1] + [[beta].sub.1][[alpha.sup.2.sub.t-1] + [gamma] Dummy, where Dummy = 1 for t [greater than or equal to] 1984:I; 0 otherwise.

[a.sub.0] [a.sub.1] [a.sub.2] [lambda]

0.4396 * 0.2365 * 0.1888 * 0.0516

(0.1296) (0.0728) (0.0671) (0.1621)

LB Q (3) LB Q (6) LB [Q.sup.2] (3) LB [Q.sup.2] (6)

2.0877 4.0421 1.2643 6.3881

[0.5543] [0.6709] [0.7376] [0.3811] [[alpha].sub.0] [[alpha].sub.1] [[beta].sub.1] [gamma]

1.0989 * 0.1025 * 0.0737 * -0.9326 *

(0.5406) (0.0893) (0.4186) (0.4615)

Skewness Kurtosis Jarque-Bera LR

0.0652 0.2646 0.8633 26.3497 *

[0.6830] [0.4115] [0.6494] [0.0000] Maximum log-likelihood function value: -62.2568 ARCH-M Results with Structural Break at 1984:I [y.sub.t] = [a.sub.0] + [a.sub.1][y.sub.t-1] + [a.sub.2][y.sub.t-2] + [lambda][[sigma].sub.t] + [[epsilon].sub.t]; and [[sigma].sup.2.sub.t] = [[alpha].sub.0] + [[alpha].sub.1][[epsilon].sup.2.sub.t-1] + [[alpha].sub.2][[epsilon].sup.2.sub.t-2] + [gamma] Dummy, where Dummy = 1 for t [greater than or equal to] 1984:I; 0 otherwise.

[a.sub.0] [a.sub.1] [a.sub.2] [lambda]

0.4360 * 0.2592 * 0.1806 * 0.0533

(0.1369) (0.0706) (0.0746) (0.1618)

LB Q (3) LB Q (6) LB [Q.sup.2] (3) LB [Q.sup.2] (6)

2.6621 5.0311 0.4254 5.1444

[0.4467] [0.5398] [0.9349] 0.5254 [[alpha].sub.0] [[alpha].sub.1] [[alpha].sub.2] [gamma]

1.0628 * 0.0913 * 0.0823 -0.8914 *

(0.1946) (0.0821) (0.1073) (0.1828)

Skewness Kurtosis Jarque-Bera LR

0.0759 0.1963 0.6057 22.1019 *

[0.6360] [0.5441] [0.7386] [0.0000] Maximum log-likelihood function value: -62.7772 Standard errors appear in parentheses; p values appear in brackets; LB Q(k) and LB Q2(k) equal Ljung-Box Q-statistics tests for standardized residuals and squared standardized residuals for autocorrelations up to k lags. LR equals the likelihood ratio statistic following a [chi square] distribution with one degree of freedom that tests for [[alpha].sub.1] + [[beta].sub.1] = 1 or [[alpha].sub.1] + [[alpha].sub.2] = 1. * 5% significance level. Table 4. Subsample Results with Structural Break at 1982:I AR(1): 1947:1-1981:IV [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ACSII] [a.sub.0] [a.sub.1] [lambda] 0.8160 * 0.3089 * -0.2937 (0.4157) (0.0900) (0.6735)

LB