Does a Faulty Barometer Herald a Storm for Stocks?

Summary: "The January Barometer" simply states that "As goes January, so goes the year," and it's racked up a seemingly remarkable forecasting record since well before Yale Hirsch of Stock Trader's Almanac first popularized it as early as 1972.

Should you fire your financial advisor and hire a month in order to optimize your asset allocation? Probably so, if you believe proponents of a time-honored indicator of future stock market performance known as "The January Barometer." The Barometer simply states that "As goes January, so goes the year," and it's racked up a seemingly remarkable forecasting record since well before Yale Hirsch of Stock Trader's Almanac first popularized it as early as 1972. Since 1938, the direction of change of the benchmark S&P in the first month out of the gate has matched the year as a whole more than a whopping 80% of the time, making January by far the most predictive month on the calendar. The results are similarly impressive if you use the Dow Jones Industrial Average (DJIA) as a yardstick and, although it somewhat diminishes the accuracy of the forecasting tool, if you assess efficacy over the next 11 or 12 months to avoid double-counting January's moves in the periods it's supposed to foreshadow. Dating back to the inception of the NASDAQ Composite Index in 1971, January achieves the greatest success of any month in anticipating the movement of OTC stocks throughout the following 11 or 12 months, and ranks second only to April in its correlation with calendar-year outcomes. Starting from 1950, an up January has meant about a 13% gain in stock prices through the remainder of the year, while opening with a down month presaged about a 1% loss. Criticisms of The January Barometer The historical evidence looked even more compelling at the start of this decade, but The January Barometer laid an egg in 3 of the past 5 years. In 2001, a positive January called a premature end to a bear market that got ugly after Al Qaeda suicide hijackers attacked the World Trade Center and Pentagon. In 2003, stocks declined in January, continuing a deep correction in the wake of a sharp initial rally off the final bear market low of the previous October, but turned higher in springtime to climb 26.4% by year-end, still the biggest annual gain since the 1990s. Last year, the market fell again in January, only to see the S&P 500 eke out a 3% gain for all of 2005, although the Dow edged down a fraction of a percent. However, the lackluster display by the blue chips actually understated the effect of the Barometer's error in a year in which smaller stocks outperformed for a 6th straight time and the average equities mutual fund returned a total 9.5%. Supporters of The January Barometer sometimes point to the 20th Amendment, a piece of Depression-era legislation also known as the "Lame Duck Amendment," to explain why it works. The 20th Amendment mandates that presidential terms, as well as those of senators and representatives, shall conclude in January, and calls for congress to convene on January 3. Formerly, they didn't throw the rascals out until March. Despite ratification in early 1933, the amendment didn't take effect until 1934. Hence the nation was forced to endure 4 months of lame-duck leadership from a by then wildly unpopular Herbert Hoover after the 1932 election, as the Great Depression deepened and Wall Street surrendered the vast bulk of its spectacular gains achieved during the summer of '32, following the stock market bottom. Now, the president delivers his State of the Union Address, highlighting priorities for the year ahead, and unveils his proposed budget in January, making the month particularly influential, or so the theory goes. Of course, they don't hold national elections every year, and almost all of the leaders are incumbents or politicians with already well-known agendas. If the timing of the presidential inauguration is so important, why didn't a "March Barometer" foretell stocks' future before 1934? From 1897 through 1933, the direction taken by the DJIA in January corresponded to the full year's results 23 times out of 37, versus just 20 of 37 for March. The record throughout that interval stays the same even if you substitute the S&P for the Dow beginning in 1928, the first year they tabulated daily prices for the S&P. Staunch defenders of the January Barometer like to commence their record keeping in 1938, citing the especially lopsided congressional margins enjoyed by Democrats earlier under the FDR Administration. This smacks of classic backfitting, however. Could the real reason behind the 4-year delay in implementation of their pet prognostic technique instead be the disastrous performance shown by The January Barometer in the 1934-1937 timeframe? In 1934, the S&P jumped a robust 10% in January, only to slide 19% during the next 12 months. If you sold on January's 4% dip to kick off 1935, you missed a roaring 57% advance. And if a 4% rise in January 1937 enticed you to bite, the stock market's October 1937 crash left you licking your wounds amid a 41% plunge. Another benefit to choosing 1938 as a starting point, while ignoring the entire 1897-1937 period, rests in the fact that most market years are up years, and the more recent era captures the secular bull markets of 1949-1968 and 1982-2000, leaving out the worst years of the Depression and the relatively dull markets of the first 20 years of the 20th century. In 1897-1937, stocks went up only 23 out of 41 times (56%), compared to 47 of 67 (one year was unchanged), or 70%, subsequently. January historically ranks as the second-strongest calendar month for stocks, trailing only December. January Barometer's Notable Failures Still, in over a century since the advent of reliable daily stock averages, the January Barometer boasts a 72% (78 of 108) success rate, including a level of accuracy approaching 80% during those years in which the market closed higher in January, as was the case this year. Yet the S&P 500, through Friday, February 10, 2006, remains over 1% lower this month after hitting new bull market highs a few short weeks ago. Accordingly, this seems like a good time to examine some of the January Barometer's most notable failures following those occasions when it appeared to call for further stock price appreciation. 1902: The DJIA established a final bull market peak in June 1901 and continued to edge down slightly in 1902. 1903: Railroad stocks had risen for over 6 years, more than tripling without a serious setback, when they topped in September 1902. Their yearlong bear market was just getting started when 1903 rolled around, and their eventual collapse would drag down the industrials. 1906: Final bull market high in late January, and the DJIA was nearly cut in half before the end of 1907. 1914: A 5-year bear market, which began with an unsuccessful assault on all-time highs in 1909, climaxes in July 1914 when authorities shut down the New York Stock Exchange at the outset of World War I. 1917: After stocks more than double to a November 1916 final top in the first couple of years of the War, in which America gets rich supplying the Allies in Europe, the market drops 40% by December 1917, as direct U.S. involvement in the conflict looms. 1929-31: Stocks crash after an explosive rally in the summer of 1929 caps an 8-year bull run, ushering in the Great Depression. Optimistic investors prematurely bid stocks higher to begin each of the next 2 years, only to regret it. 1934: After more than doubling in less than a year, the new bull market stalls following fresh highs in February 1934. 1937: A March top culminates an advance of nearly 5 years and 372% in the DJIA before the short but severe 1937-38 "Roosevelt Recession," which saw industrial production fall faster than during 1929-32 and cut the Dow in half. 1946: A last thrust higher following a 10% February correction merely postpones the inevitable. The 129% DJIA gain in a span of more than 4 years culminating in a May 29, 1946 peak grossly understates the extent of the advance leading up to the high. The S&P does significantly better than that, and other averages leave the blue chips in the dust. Railroad stocks nearly triple, and the Dow Jones Utility Average quadruples. 1966: Another bull market launches in the second year of the decade, only to die in the 6th, as the Dow touches 1000 for the first time en route to a February 9, 1966 closing high. 1994: On February 2, the anniversary date that preceded the 1946 correction, and also in the 4th year of a bull market, stocks begin a 10% correction, as in 1946. This time, however, rather than quickly racing to a final top after the correction is over, the stock market trades in a narrow range throughout the rest of the year before busting out higher in 1995. 2001: The 1990s bull market amazingly lasts over 9 years, taking the NASDAQ Composite from a mere 325 to above 5000 in March 1990. After a run like that, the ensuing bear market wasn't nearly complete despite a reflex rally in early 2005. What Can be Learned? Are there any lessons we can take from the 14 notable failures of the January Barometer described above? Six of the examples (1902, 1903, 1917, 1930, 1931, 2001) involve false January rallies that developed in the early stages of bear markets. Clearly, we don't fit into this category. The bear market following the late 1990s tech-stock mania bottomed on October 9, 2002. Our market attained its subsequent high-to-date just last month. Could we have already seen the final top, or might the entire advance since 2002 represent nothing more than an elongated bear market rally? The latter possibility would be essentially unheard of, given the amount of time elapsed since the low. Nevertheless, bull markets have been known to expire in a shorter time than the 3 years and 3 months required to trudge to the January 11, 2006 closing highs in the DJIA and S&P. Almost half of all previous misleadingly bullish Januarys came late in long or powerful bull markets, during the years (1906, 1929, 1934, 1937, 1946, 1966) of their final tops. The latter 3 such cases, like our present situation, all unfolded following "second-year lows," but served up lengthier and more energetic advances than the 2002-06 bull market so far. The 2-month, 12% bounce in the S&P from its low last October 13 would represent an uncharacteristically brief and anemic concluding bull leg, especially anticlimactic on the heels of a flat year. Unlike 1946, 1965-66 and 1994, we haven't seen a 10% market decline in some time. The largest correction the market could muster in 2005 was on the order of 7%. The less-than-stellar 52% maximum improvement in the closing price of the Dow since its October 9, 2002 trough is also tepid by bull market standards. As in 1942-46, the S&P is ahead of the DJIA, and broader indexes have crushed both blue-chip measures, but the S&P's reluctance thus far to challenge its all-time high, unlike the Dow after it was similarly cut in half 100 years ago, further attests to the underachieving nature of the existing bull. Still, this bull market is undeniably long in the tooth, and enough time remains in 2006 to set up a final top and then possibly stage a decline big enough to make a liar of The January Barometer for a 4th time in 6 years.