spx Earnings Yield Minus 90 day Treasury Yield Versus % Change in the SPX the next month, dividends excluded (to isolate capital gains or loss), Janua

Earnings Yield Minus 90 day Treasury Yield Versus % Change in the SPX the next month, dividends excluded (to isolate capital gains or loss), January, 1945, to February, 2000:

E Yld - 90d:	Avge rr next m:	n:	% of sample:
all:	0.78%	664	100%
<0	0.40%	144	22%
>0	0.88%	519	78%
>.01	0.81%	416	63%
>.02	0.89%	292	44%
>.03	0.89%	242	36%
>.04	1.04%	200	30%
>.05	1.18%	163	25%
>.06	1.27%	117	18%
>.07	1.01%	92	14%
>.08	1.30%	76	11%
>.1	1.12%	43	6%
<-.01	-0.56%	62	9%
<-.02	-1.23%	18	3%
<-.03	-1.99%	8	1%
<-.04	-2.81%	2	0%

Note a couple of things:

First, there is a strong linear relationship between the difference between the Earnings Yield (trailing 12 month earnings on the S&P500) and the 30 day yield (as a proxy for short-term interest rates) and the return for the next month of the S&P500. The average return whenever the S&P500 earnings yield was less than the 30 day yield was only half the average monthly return of the study period (0.40% v. 0.78%).

Second, any Earnings yield T bill difference greater than 4% resulted in an average monthly return of over 1% in the S&P500.

Third, any Earnings yield T bill difference less than -1% led to a negative return. (As this is being written, this difference is about -2.1%).

The stock market is a bad place to be when the Earnings yield is less than the 90 day Treasury yield.

A summary broken down by thirds (roughly) follows:

E Yld - 90d:	Avge rr next m:	n:	% of sample:
<.5%	0.57%	206	31%
.5%-4%	0.74%	255	38%
>4%	1.04%	200	30%

Ratio of 30 Year Treasury Bond Yield to 90 day Treasury Yield:

30y / 90d:	Avge rr next m:	n:	% of sample:
all:	0.78%	664	100%
>1	0.99%	566	85%
>1.1	0.99%	499	75%
>1.2	1.10%	416	63%
>1.3	1.09%	335	50%
>1.4	0.83%	249	38%
>1.5	1.01%	209	31%
>1.6	1.16%	167	25%
>1.7	1.30%	144	22%
>2	0.88%	103	16%
<=1	-0.48%	96	14%

Note this is less linear or staggering than the Earnings Yield - 90 day Treasury, but it is more useful perhaps as a negative: when the yield curve is <1,>

Next I divided the data into thirds:

30y / 90d:	Avge rr next m:	n:	% of sample:
<1.2	0.23%	246	37%
1.2-1.5	1.20%	207	31%
>1.5	1.01%	209	31%

Again, not a staggering correlation and in fact extreme yield curves are slightly worse on average than modest ones (1.2-1.5; 1.01% v. 1.20% average next monthly gain).

Earnings Yield v. Next Month's S&P500 Return:

E Yld:	Avge rr next m:	n:	% of sample:
all:	0.78%	664	100%
>.05	0.78%	578	87%
>.06	0.92%	402	61%
>.07	0.85%	313	47%
>.08	0.92%	248	37%
>.09	0.75%	186	28%
>.1	0.95%	141	21%
>.11	1.05%	108	16%
>.12	1.55%	79	12%
>.13	1.75%	45	7%
>.14	1.75%	22	3%
>.15	0.78%	8	1%

Note that raw earnings yield alone is not that useful; lower yields are tolerable in a low interest rate environment. However, earnings yields are roughly correlated with return next month, although not dramatically.

Dividing the sample into thirds illustrates that this relationship is far from straightforward:

E Yld:	Avge rr next m:	n:	% of sample:
<=.06	0.56%	259	39%
6-8%	0.93%	154	23%
>.08	0.92%	248	37%

Note that there were only 11 cases when the market was earning less than 3%, and the average next monthly gain was only .49%.