Edward Renshaw
Professor of Economics
State University of New York at Albany
Bad Septembers, what are they related to? Among the suspects are costly summer vacations, shorter days and cooler temperatures which may help some investors think more clearly, the way we finance education and some financial innovations which have made it easier and less costly for investors to switch retirement funds from stock to money market funds and back again.
In his book on Stocks for the Long Run (Chapter 17) Jeremy Siegel has noted that September is the worst month to have owned common stock on the average and the only month with a negative return. "Maybe the poor returns in late fall have nothing directly to do with economics but are related to the approach of winter and the depressing effect of the shorter days". Siegel goes on to suggest that, "Perhaps, the poor returns in September are the result of investors liquidating stocks (or holding off buying new stocks) to pay for their summer vacations."
An even more chilling depressant is the way we finance education. Some educators are paid on a nine or ten month basis and don't received their first pay check until about the end of September. School taxes are often due in September and some parents may be forced to liquidate stock to pay for tuition and the expense of sending their children to private schools, colleges and universities. September is also one of those months when a successful investor has to pay his or her estimated income taxes.
Financial innovations which have made it easier and less costly for some investors to switch retirement funds between stock and money market funds may have made it more rational for investors who are not burdened with educational expenses in September to delay their own stock purchases or move out of equities in the hope of being able to repurchase them at a lower price after the "correction" which has so often occurred in September.
In late August of 1995, Siegel's findings with regard to bad Septembers were widely publicized by Steve Sakson of the Associated Press. The S&P composite stock price index, instead of declining, continued to recover from the July 19th low of 550.98 and beginning on September 5th proceeded to achieve eight consecutive new historic closing highs. This was the first time that many new closing highs had been achieved in an eight day trading period since July of 1964.
One of the more important lessons to be learned from this type of experience is: never trust an average. It can conceal a lot of period to period variability. In Table 8.1S we show the monthly percentage changes in The S&P composite stock price index since it was extended back to 1928.
From 1928-94 there were 39 years when the S&P index declined during September and only 27 years when it increased. Before jumping ship or delaying stock purchases, however, it makes sense for one to look for relationships that might enable an investor to do a better job of differentiating between positive and negative returns.
In Table 7.2S it was noted that ten or more new highs in the second quarter of a calendar year have so far always spilled over and resulted in some additional new highs in the third and fourth quarters of the same year. For the six cases prior to 1995 (1955, 1964, 1965, 1968, 1983 and 1985) the September changes in the S&P index were: 1.1, 2.9, 3.2, 3.9 1.0 and -3.5 percent respectively. While this system did not protect one from a sizable loss in 1985, the average September gain for these six years was 1.43 percent versus an average September loss of 1.61 percent for the other 61 years from 1928-94.
In essay 6S it was noted that the stock market is more likely to suffer prompt corrections in a turbulent new high bull market than in a market which achieves numerous new highs without a lot of day-to-day volatility. One measure of volatility that can easily be determined from the data in Table 8.1S is the severity of the monthly downside corrections for the S&P index during a calendar year prior to September.
For the ten years from 1928-94 when the worst monthly correction prior to September was a decline of 2.4 percentage points or less (1944, 1955, 1958, 1959, 1964, 1972, 1976, 1985, 1987, and 1992) the average following September change was slightly positive.
Another factor that individual investors ought to keep in mind in appraising the September outlook is the possibility that money managers have become more forward looking in their behavior. For the 16 years when the S&P index increased in July and then declined in August (perhaps in anticipation of a bad September) the stock market has advanced more often than not in September and since 1959 has consistently increased in September.
Consecutive declines in the S&P index in June, July and August (which occurred in 1946, 1966, 1974, 1981 and 1990), on the other hand, were all followed by declines of more than five percent in September. The last three of these cases are associated with economic recessions.
The possibility that a very bad summer for the stock market might help to tip the US economy into an economic recession leads this analyst to conclude that the Federal Reserve ought to be encouraged to buy and sell stock index futures to help stabilize this market, reduce the risk of economic recessions and perhaps offset, to some extent, a saving and investment imbalance in September when people go back to work and school.
The stock market crashes of 1929 and 1987 have given October a bad reputation. One of the more interesting questions to be asked in connection with Table 8.1S is whether there can be a bad October without a disappointing September? For those cases where there was an increase in the S&P index of .1 percent or more in September the worst following decline in October was only 3.0 percent in 1955--a loss that was more than compensated for by an increase in stock prices of 7.5 percent in November.
For the seven years from 1928-94 when September gains were equal to 4.0 percent or more, the average following October change for the S&P index was a gain equal to .714 percent--in spite of three losses amounting to -1.5, - 1.9 and - .1 in 1939, 1954 and 1973.
From 1928-94 there were actually fewer monthly declines for the S&P index in October than in February, May, June, September and November. The safest months to have owned a portfolio similar to the S&P index have been December and January.
February used to be a bad month for stocks. Since the stock market correction of 1984, however, there have only been two declines in the S&P index during February. February's superior performance in recent years may be related in part to the much publicized January effect.
January and December are probably the most interesting months to study the stock market. By the mid 1970s statisticians were beginning to conclude that there might be some seasonality to the monthly rates of return from holding common stock due primarily to large January returns and by the early 1980s it was discovered that most of the exceptional returns in this month were associated with small firms.
While much has been written about the January size effect it is less well appreciated that this month often sets a tone for the market that will carry forward to the end of the year. From 1947-94 there were 25 years when the S&P composite stock price index increased by more than one percent in January and in every one of those years the financial return for the entire year was positive. (See those cases marked with an asterisk in columns (2) and (3) of Table 8.2S.)
The financial return for the entire year was also in excess of the January gain in all but two of these years. The first exception was 1987 when the S&P index, after the largest January gain in its history (13.2 percent), increased another 22.9 percent before getting involved in the most spectacular new-high crash in the history of the NYSE. The other exception is 1994 when the Fed launched a preemptive strike against the possibility of accelerating inflation beginning in February and continued to raise short term interest rates through February of 1995.
The identification of relatively safe years to have invested in the stock market can be improved somewhat by combining the gains for both January and the preceding December. For the 28 cases where the combined gain was in excess of three percent for these two months the full year returns for the S&P index are also positive.
The unresolved question is whether this success story is an accident or whether there is something special about January that enables investors to more accurately anticipate the behavior of stock prices in the remainder of the year.
It has been noted that January is the start of the tax year for investors and the beginning of the tax and accounting years for most firms. If this month gets off to a good start some funds that were taken out of the market in connection with tax-loss selling in the preceding year may be reinvested.
Preliminary and in many cases final announcements of previous calendar (fiscal) year's sales and earnings are often made in January. Some analysts have suggested that the extra information that is available at this time of the year from both government and the private sector may enable investors to more accurately forecast the future.
Except for the short lived recession from January 1980 to July of that year--when there was some mad speculation in US oil companies following the Iranian revolution--the US economy has never experienced a peak in business in any year since 1945 when the January gain in stock prices was over one percent.
Yr. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 28 -1.1 -1.8 10.8 3.2 1.3 -4.0 1.3 7.4 2.4 1.5 12.0 .3 29 5.7 -.6 -.2 1.6 -4.3 11.2 4.6 9.8 -4.9 -19.9 -13.4 2.5 30 6.2 2.2 8.0 -1.0 -1.6 -16.5 3.7 .8 -13.0 -8.9 -2.2 -7.4 31 4.9 11.4 -6.9 -9.6 -13.7 13.9 -7.4 .9 -29.9 8.4 -9.8 -14.5 32 -2.8 5.1 -11.8 -20.2 -23.3 -.9 37.7 37.5 -3.7 -13.9 -5.9 5.2 33 .7 -18.4 3.4 42.2 15.9 13.2 -8.8 11.5 -11.4 -8.9 10.3 2.2 34 10.6 -3.7 -.1 -2.7 -8.1 2.1 -11.5 5.4 -.5 -3.2 8.3 -.4 35 -4.2 -4.0 -3.1 9.6 3.2 6.8 8.3 2.2 2.4 7.5 3.9 3.7 36 6.6 1.7 2.5 -7.7 4.6 3.1 6.8 .9 .1 7.5 .4 -.6 37 3.8 1.5 -.9 -8.3 -1.0 -5.3 10.3 -5.5 -14.2 -10.2 -10.1 -5.0 38 1.3 6.1 -25.0 14.1 -4.4 24.7 7.3 -2.7 1.5 7.6 -3.3 3.8 39 -6.9 3.3 -13.5 -.5 6.2 -6.4 10.9 -7.1 16.5 -1.5 -4.9 2.4 40 -3.5 .7 1.0 -.5 -24.0 7.7 3.1 2.6 .9 3.9 -4.2 -.3 41 -4.8 -1.5 .4 -6.5 .4 5.3 5.5 -.9 -1.0 -6.9 -4.2 -4.5 42 1.4 -2.5 -6.8 -4.4 6.4 1.8 3.1 .7 2.7 6.4 -1.4 5.2 43 7.2 5.1 5.3 .1 4.5 2.0 -5.4 1.0 2.4 -1.3 -7.6 5.9 44 1.5 -.3 1.7 -1.2 4.0 5.1 -2.1 .9 -.3 .0 .4 3.5 45 1.4 6.2 -4.6 8.8 1.1 -.3 -2.0 5.8 4.2 3.0 3.2 1.0 46 7.0 -6.9 4.6 3.8 2.2 -3.9 -2.6 -7.3 -10.2 -.8 -1.1 4.3 47 2.4 -1.5 -1.7 -3.9 -.9 5.3 3.6 -2.8 -1.4 2.1 -2.9 2.1 48 -4.0 -4.7 7.7 2.7 7.8 .3 -5.3 .8 -3.0 6.8 -10.8 3.1 49 .1 -3.9 3.0 -2.1 -3.7 -.2 6.2 1.2 2.4 3.0 .1 4.4 50 1.7 1.0 .4 4.5 3.9 -5.8 .8 3.3 5.6 .4 -.1 4.6 51 6.1 .6 -1.8 4.8 -4.1 -2.6 6.9 3.9 -.1 -1.4 -.3 3.9 52 1.6 -3.6 4.8 -4.3 2.3 4.6 1.8 -1.5 -2.0 -.1 4.6 3.5 53 -.7 -1.8 -2.4 -2.6 -.3 -1.6 2.5 -5.8 .1 5.1 .9 .2 54 5.1 .3 3.0 4.9 3.3 .1 5.7 -3.4 8.3 -1.9 8.1 5.1 55 1.8 .4 -.5 3.8 -.1 8.2 6.1 -.8 1.1 -3.0 7.5 -.1 56 -3.6 3.5 6.9 -.2 -6.6 3.9 5.2 -3.7 -4.5 .5 -1.1 3.5 57 -4.2 -3.3 2.0 3.7 3.7 -.1 1.1 -5.6 -6.2 -3.2 1.6 -4.1 58 4.3 -2.1 3.1 3.2 1.5 2.6 4.3 1.2 4.8 2.5 2.2 5.2 59 .4 -.0 .1 3.9 1.9 -.4 3.5 -1.5 -4.6 1.1 1.3 2.8 60 -7.1 .9 -1.4 -1.8 2.7 2.0 -2.5 2.6 -6.0 -.2 4.0 4.6 61 6.3 2.7 2.6 .4 1.9 -2.9 3.3 2.0 -2.0 2.8 3.9 .3 62 -3.8 1.6 -.6 -6.2 -8.6 -8.2 6.4 1.5 -4.8 .4 10.2 1.3 63 4.9 -2.9 3.5 4.9 1.4 -2.0 -.3 4.9 -1.1 3.2 -1.1 2.4 64 2.7 1.0 1.5 .6 1.1 1.6 1.8 -1.6 2.9 .8 -.5 .4 65 3.3 -.1 -1.5 3.4 -.8 -4.9 1.3 2.3 3.2 2.7 -.9 .9 66 .5 -1.8 -2.2 2.1 -5.4 -1.6 -1.3 -7.8 -.7 4.8 .3 -.1 67 7.8 .2 3.9 4.2 -5.2 1.8 4.5 -1.2 3.3 -2.9 .1 2.6 68 -4.4 -3.1 .9 8.2 1.1 .9 -1.8 1.1 3.9 .7 4.8 -4.2 69 -.8 -4.7 3.4 2.1 -.2 -5.6 -6.0 4.0 -2.5 4.4 -3.5 -1.9
Yr. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 70 -7.6 5.3 .1 -9.0 -6.1 -5.0 7.3 4.4 3.3 -1.1 4.7 5.7 71 4.0 .9 3.7 3.6 -4.2 .1 -4.1 3.6 -.7 -4.2 -.3 8.6 72 1.8 2.5 .6 .4 1.7 -2.2 .2 3.4 -.5 .9 4.6 1.2 73 -1.7 -3.7 -.1 -4.1 -1.9 -.7 3.8 -3.7 4.0 -.1 -11.4 1.7 74 -1.0 -.4 -2.3 -3.9 -3.4 -1.5 -7.8 -9.0 -11.9 16.3 -5.3 -2.0 75 12.3 6.0 2.2 4.7 4.4 4.4 -6.8 -2.1 -3.5 6.2 2.5 -1.2 76 11.8 -1.1 3.1 -1.1 -1.4 4.1 -.8 -.5 2.3 -2.2 -.8 5.2 77 -5.1 -2.2 -1.4 .0 -2.4 4.5 -1.6 -2.1 -.2 -4.3 2.7 .3 78 -6.2 -2.5 2.5 8.5 .5 -1.8 5.4 2.6 -.7 -9.2 1.7 1.5 79 4.0 -3.7 5.5 .2 -2.6 3.9 .9 5.3 .0 -6.9 4.3 1.7 80 5.8 -.4 -10.2 4.1 4.7 2.7 6.5 .6 2.5 1.6 10.2 -3.4 81 -4.6 1.3 3.6 -2.3 -.2 -1.0 -.2 -6.2 -5.4 4.9 3.7 -3.0 82 -1.8 -6.1 -1.0 4.0 -3.9 -2.0 -2.3 11.6 .8 10.9 3.7 1.5 83 3.3 1.9 3.3 7.5 -1.2 3.5 -3.3 1.1 1.0 -1.5 1.7 -.9 84 -.9 -3.9 1.3 .5 -5.9 1.7 -1.6 10.6 -.3 -.0 -1.5 2.2 85 7.4 .9 -.3 -.5 5.4 1.2 -.5 -1.2 -3.5 4.3 6.5 4.2 86 .5 7.1 5.3 -1.4 5.0 1.4 -5.9 7.1 -8.5 5.5 2.1 -2.8 87 13.2 3.7 2.6 -1.1 .6 4.8 4.8 3.5 -2.4 -21.8 -8.5 7.3 88 4.0 4.2 -3.3 .9 .3 4.3 -.5 -3.9 4.0 2.6 -1.9 1.5 89 7.1 -2.9 2.1 5.0 3.5 -.8 8.8 1.6 -.7 -2.5 1.7 2.1 90 -6.9 .9 2.4 -2.7 9.2 -.9 -.5 -9.4 -5.1 -.7 6.0 2.5 91 4.2 6.7 2.2 .0 3.9 -4.8 4.5 2.0 -1.9 1.2 -4.4 11.2 92 -2.0 1.0 -2.2 2.8 .1 -1.7 3.9 -2.4 .9 .2 3.0 1.0 93 .7 1.0 1.9 -2.5 2.3 .1 -.5 3.4 -1.0 1.9 -1.3 1.0 94 3.3 -3.0 -4.6 1.2 1.2 -2.7 3.1 3.8 -2.7 2.1 -4.0 1.2 95 2.4 3.6 2.7 2.8 3.6 2.1 3.2 -.0 4.0 ?
Source of basic data: Standard and Poor's Security Price Index Record.
Percentage Change S&P Index Full Year
--------------------------- Financial Return
Year Preceding December January S&P Index
(1) (2) (3)
1947 4.3 2.4* 5.5*#
1948 2.1 -4.0 5.4
1949 3.1 .1 17.8#
1950 4.4 1.7* 30.5*#
1951 4.6 6.1* 23.4*#
1952 3.9 1.6* 17.7*#
1953 3.5 - .7 - 1.2
1954 .2 5.1* 51.2*#
1955 5.1 1.8* 31.0*#
1956 - .1 -3.6 6.4
1957 3.5 -4.2 -10.4
1958 -4.1 4.3* 42.4*
1959 5.2 .4 11.8#
1960 2.8 -7.1 .3
1961 4.6 6.3* 26.6*#
1962 .3 -3.8 - 8.8
1963 1.3 4.9* 22.5*#
1964 2.4 2.7* 16.3*#
1965 .4 3.3* 12.3*#
1966 .9 .5 -10.0
1967 - .1 7.8* 23.7*#
1968 2.6 -4.4 10.8
1969 -4.2 - .8 - 8.3
1970 -1.9 -7.6 3.5
1971 5.7 4.0* 14.1*#
1972 8.6 1.8* 18.7*#
1973 1.2 -1.7 -14.5
1974 1.7 -1.0 -26.0
1975 -2.0 12.3* 36.9*#
1976 -1.2 11.8* 23.6*#
1977 5.2 -5.1 - 7.2
1978 .3 -6.2 6.4
1979 1.5 4.0* 18.4*#
1980 1.7 5.8* 31.5*#
1981 -3.4 -4.6 - 4.8
1982 -3.0 -1.8 20.4
1983 1.5 3.3* 22.3*#
1984 - .9 - .9 6.0
1985 2.2 7.4* 31.1*#
1986 4.2 .5 18.5#
1987 -2.8 13.2* 5.7*#
1988 7.3 4.0* 16.6*#
Percentage Change S&P Index Full Year
--------------------------- Financial Return
Year Preceding December January S&P Index
(1) (2) (3)
1989 1.5 7.1* 31.2*#
1990 2.1 -6.9 - 3.1
1991 2.5 4.2* 30.0*#
1992 11.2 -2.0 7.4#
1993 1.0 .7 9.9
1994 1.0 3.3* 1.3*#
1995 1.2 2.4* ?*#
*Years when the January gain for the S&P index was more than one percent.
#Years when the combined gain for both Dec. and Jan. was over three percent.
Source: This table was first publicized in the April 6, 1989 issue of The Market Chronicle and then republished in The Practical Forecasters' Almanac. The basic data are from Standard and Poor's Security Price Index Record.
Go on to Essay 9S:
Return to the Introduction