Table 5

Monthly excess stock market returns against conditional variances, implied variance and volatility innovations with uncertainty

Modelc(c1)d(d1)e(e1)c2d2e2Adj. R2 (%)
Panel A: Rolling window model
(1)0.0127*** (7.3580)−2.4713*** (−12.1310)−4.5140*** (−5.8518)   23.919
(1)-E0.0141*** (6.8807)−3.3810*** (−2.8929)−13.1511*** (−7.0153)0.0017 (0.4118)0.6644 (0.5879)9.6562*** (5.5191)33.200
(1)-V0.0122*** (6.3176)−2.7114*** (−5.6440)−7.9211*** (−4.0476)0.0036 (0.9818)0.1817 (0.4548)2.7468** (2.3258)28.109
Panel B: MIDAS
(2)0.0137*** (8.8590)−3.3620*** (−11.1890)−5.7707*** (−4.3743)   25.579
(2)-E0.0100*** (3.5257)−4.4473** (−2.1610)−10.8156*** (−3.4397)0.0074 (1.4790)1.2914 (0.6825)6.5161** (2.2780)30.525
(2)-V0.0070** (2.4858)−2.1715* (−1.8547)−7.5626*** (−4.9310)0.0098** (2.5979)−0.5778 (−0.8695)4.5875*** (3.5795)30.185
Panel C: GARCH(1,1)
(3)0.0112*** (6.1250)−2.2474*** (−6.9994)−4.0364*** (−6.3578)   21.463
(3)-E0.0069* (1.8890)−1.4596 (−0.4567)−8.9600** (−2.0007)0.0076* (1.6687)−1.1033 (−0.3623)5.5141 (1.2932)26.246
(3)-V0.0046** (2.2378)0.4845 (0.8697)−5.5790*** (−5.9684)0.0098** (2.3244)−2.9811*** (−5.9331)2.5081*** (2.8718)25.507
Panel D: GJR-GARCH(1,1)
(4)0.0111*** (4.7355)−2.4344*** (−3.5514)−6.3264*** (−3.7760)   26.805
(4)-E0.0072** (2.2234)−2.5108 (−1.0646)−16.2551*** (−3.5881)0.0081* (1.7952)−0.3699 (−0.1575)10.9728** (2.4313)33.315
(4)-V0.0016 (0.6390)2.4062*** (3.1317)−5.9096*** (−4.5766)0.0131*** (3.1319)−5.5449*** (−6.8486)−1.0096 (−1.1106)29.415
Panel E: Implied variance
(5)0.0059** (1.9835)−0.0136 (−0.1229)−0.1855** (−2.5653)   0.821
(5)-E−0.0042 (−1.3380)0.4625*** (4.6442)0.2939* (1.9550)0.0142** (2.1479)−0.5506*** (−3.4403)−0.4644*** (−2.6341)4.075
(5)-V−0.0033 (−1.1283)0.5035*** (4.3046)0.6882*** (3.1617)0.0134** (2.0459)−0.4561*** (−2.9663)−1.2897*** (−4.6225)11.665

Note(s): The table reports estimates from the regressions of monthly excess stock market returns against conditional variances, implied variance and volatility innovations. Models (1) to (5) are estimates from the regressions without a dummy variable of uncertainty. Models (1)-E to (5)-E ((1)-V to (5)-V) are estimates from the regressions with a dummy variable of FEPU (FVOV). Newey and West (1987) corrected t-statistics are reported in parentheses. *, ** and *** represent significance at the 10, 5 and 1% levels, respectively

or Create an Account

Close subscription notice
Close access options