Table 2: Regressions on Δ (log(S&P500)).

Sample 1960M03-2013M11 1960M03-2013M11 1960M03-2013M11 1960M03-2013M11 1960M03-2013M11
Sample size 645 645 645 645 645
Conditional Mean Equation
Constant 0.009330 (3.357484) 0.009392 (3.856645) 0.011152 (4.204454) 0.010966 (4.300682) 0.010794 (3.769422)
Coefficient on Δ (k) -54.30041 (4.844677) -57.93669 (6.006794) -50.99889 (4.301666) -55.61610 (5.626876) -55.66469 (5.627213)
Coefficient on Δ (log(RDISP)) 0.617094 (3.042664) 0.511563 (2.638545) 0.597931 (2.991092) 0.495394 (2.564833) 0.495110 (2.564089)
Coefficient on Δ (log(CPI )) -0.364320 (0.487215) -0.136211 (0.219542)
Coefficient on the predicted Δ (log( CPI )) -0.904306 (1.197264) -0.681087 (1.020066) 0.697952 (1.025360)
Coefficient on the residual of Δ (log( CPI )) -2.047505 (1.415218) -1.128097 (0.955703) -1.127707 (0.951314)
Coefficient on the absolute value of the residual of Δ (log( CPI )) 0.213096 (0.166536)
Coefficient on the average 3 lagged Δ (log(MS)) -0.766744 (3.017124) -0.870243 (3.545438) -0.780723 (2.921106) -0.864396 (3.513246) -0.867843 (3.530020)
Conditional Variance Equation
constant 0.0000742 (2.142612) 0.0000743 (2.122371) 0.0000743 (2.126372)
RESID(-1)^2 0.138277 (4.641334) 0.137034 (4.551326) 0.137237 (4.561186)
GARCH(-1) 0.828629 (19.92015) 0.829500 (19.68682) 0.829275 (19.71075)
Adjusted R-Square 0.068926 0.067982 0.071548 0.069972 0.068303
Log likelihood 1135.765 1164.633 1137.179 1165.307 1165.321
Durbin-Watson statistic 2.028738 2.028756 2.026901 2.026807 2.026609
Akaike Information criterion -3.506249 -3.586459 -3.507532 -3.585449 -3.582390
Schwarz information criterion -3.471604 -3.531026 -3.465957 -3.523087 -3.513099
Hannan-Quinn information criterion -3.492807 -3.564950 -3.491400 -3.561252 -3.555504

Notes: See notes under Table 1. CPI stands for the US consumer price index (all items). Newey-West heteroscedasticity and autocorrelation consistent (HAC) standard errors and covariance are computed for the regressions without a conditional variance equation. Bollerslev-Wooldridge robust standard errors and covariance are computed for the regressions with a conditional variance equation.