B>G?, from Kim Zussman
Now that the "too low risk premium" problem has been ameliorated, this week the Fed has an opportunity to stabilize recently volatile markets. Since Bernanke took over the bank, has there been any change in market reaction compared to his predecessor?
Checked weekly SPY returns for weeks which contained FED meetings (dates from FOMC calendar site), and compared returns for the last eight (Bernanke) and the 17 prior back to 1/27/04 (Greenspan):
Two-sample T for Ben rt vs green rt
N Mean StDev SE Mean
Ben rt 8 0.0019 0.0155 0.0055 t=0.4 4/8 pos
green rt 17 -0.0005 0.0128 0.0031 9/17 pos
The difference is not significant, but interestingly while B weeks were positive, G weeks were slightly negative in the period. Here are Green and Ben weeks:
wk start green rt
01/30/06 -0.018
12/12/05 0.006
10/31/05 0.019
09/19/05 -0.017
08/08/05 0.001
06/27/05 0.005
05/02/05 0.012
03/21/05 -0.012
01/31/05 0.024
12/13/04 0.006
11/08/04 0.013
09/20/04 -0.015
08/09/04 0.003
06/28/04 -0.008
05/03/04 -0.009
03/15/04 -0.010
01/26/04 -0.008
wk start Ben rt
01/29/07 0.019
12/11/06 0.007
10/23/06 0.008
09/18/06 -0.004
08/07/06 -0.009
06/26/06 0.023
05/08/06 -0.025
03/27/06 -0.003
Also from Kim Zussman:
Based on MC's suggestion, for FED weeks since 2004 what is the effect of the prior week? Checked this with regression: Ind var = week before, dep var = week after:
Regression Analysis: wk ret versus wk before (all weeks, Ben and Al)
The regression equation is wk ret = 0.00040 + 0.108 wk before
Predictor Coef SE Coef T P
Constant 0.0004 0.0027 0.15 0.886
wk before 0.1075 0.1714 0.63 0.537
S = 0.0135695 R-Sq = 1.7% R-Sq(adj) = 0.0%
Shows slight, insignificant positive correlation. Next checked same
since Ben took over:
Regression Analysis: B week versus B-fore
The regression equation is
B week = 0.00533 - 0.996 B-fore
Predictor Coef SE Coef T P
Constant 0.0053 0.0058 0.92 0.392
B-fore -0.9963 0.7486 -1.33 0.232
S = 0.0146648 R-Sq = 22.8% R-Sq(adj) = 9.9%
Here there is a trend toward reversal of prior week's return (rsq pretty hi). In that Ben's weeks tended to reverse the priors, since 2004 Al must have had positive correlation:
Regression Analysis: G week versus G-fore
The regression equation is G week = 0.00008 + 0.172 G-fore
Predictor Coef SE Coef T P
Constant 0.00008 0.0032 0.02 0.981
G-fore 0.1716 0.1700 1.01 0.329
S = 0.0127503 R-Sq = 6.4% R-Sq(adj) = 0.1%
Hope springs eternal from so little n. Or at least next week.
