Is it by coincidence that this past year has been filled with events bearing resemblance to the market in the late 1990s? I personally have battlefield experience in the technology capital markets during the late 1990s, and the market trade is too analogous to ignore. Show me social networking companies going public with sky-high valuations like Twitter (NYSE:TWTR) and LinkedIn (NYSE:LNKD) or other tech companies with high cash-burn rates and uncertain revenue models doubling in price on the first trading day post-IPO, and I say deja vu. Need I point back to the many companies that were taken public with weak earnings models, only to fold in the late 90s boom like Sycamore Networks, Pets.com, Webvan.com. The stream of IPOs is a function of the flow of funds in the capital market, triggered to a large degree by Fed policy and U.S. fiscal policy and the resulting interaction between the foreign capital markets and the United States. Not by coincidence, many of the government bureaucrats from the Clinton administration have had a hand in architecting the Obama economic plan. So, you can imagine why we are seeing comparable market results today.
One difference is emerging as we enter 2014 in the midst of the tech IPO frenzy. The equity market forecast sentiment continues to be resoundingly negative. The stock market climb (NYSEARCA:SPY) (NYSEARCA:DIA) from the depths of the 2009 lows has been one big wall of fear, and many have likened it to the most "hated market increases in history."
Goldman Sachs has already claimed the ominous position that the market will post a pullback in 2014 (Goldman Sachs 10% Pullback Forecast). Specifically, David Kostin, a Goldman Sachs equity strategist says, "We estimate a 67% probability of a 10% drawdown at some point in 2014…" He goes on to say, "We are not forecasting a decline in the index (he forecasts a 6% gain), but providing an estimate of the lowest point it may reach on its way to our future target." It is an interesting position. Essentially, "weebles wobble, but they don't fall down."
And, Goldman is not the only financial firm to jump on the impending market pullback bandwagon as we approach the New Year. Other firms/analysts already positioned with negative forecasts include: Albert Edwards of Societe Generale, who forecasts a U.S. recession triggered by declining productivity leading to lower corporate profitability; Nomura strategist Bob Janjuah, who projects a 25-50% global market decline; and Bill Blain, a senior fixed income broker at Mint Partners, who sees a U.S. stock sell-off triggered by the cessation of the Fed bond buying program (Negative 2014 Market Forecasts).
At this time, my survey of the media finds more negative or "hedged" forecasts for 2014 market returns than outright, "this market is headed higher" perspectives.
What does the historical trading pattern of the U.S. stock market actually show concerning the probability the market will rise or fall in 2014 following a banner year in which it is up over 20%? This article takes a look at the question, and provides the data to allow you to make your own assessment.
S&P 500 Returns - Abnormally Distributed, Positively Biased
One of the more interesting aspects of the Goldman prediction about the equity market in 2014 was its precision in setting the odds of a 10% correction sometime next year at 67%. I find it odd because in strict mathematical terms, predicting the probability of market returns assumes either the distribution of market returns is normal; or if it is not normal, means you have very good model that can predict the factor(s) that make market returns "abnormal."
With respect to the question of whether stock market returns are normally distributed, the answer is definitely not. Stock market returns over time are skewed negatively in a statistically significant way - which is a fancy way of saying results are typically positive, and above the average, until they suddenly are not. And, when stock returns do fall, they typically decline rapidly and substantially, creating a long negative tail in the distribution.
To show this point, the graph below contains the distribution of yearly returns on the S&P 500, recorded at the close of every month from the beginning of 1952 through November of 2013.
The graph above shows the breakdown of returns over the time period, segregated by different levels of return. The S&P 500 average or mean return from 1952 through 2013 was 8.8% plus dividends. But, more interesting is the span of the distribution. Assuming you could just randomly choose a time period to invest in the S&P 500, over 25% of the time you should expect that a year later the share would be worth less. Likewise, almost 25% percent of the time you should experience a return of more than 20%. The remaining 50% of the time the investment return would be between 0% and 20%, with a bias above the mean.
For math astute investors reading this article, the time series is not normally distributed, so inferring probabilities on future returns for investments made today using the distribution is problematic. The distribution passes the normality test using Excess Kurtosis as a test, as the statistical measure is close enough to zero when error adjusted. If you plot the entire time series end to end, rather than aggregated on the tails as I have done for illustrative purposes, the distribution appears normal. The primary issue is that returns are skewed to the right over time. The skew statistic of -.32 is more than 2 standard errors from 0 in the sample, meaning that it is significant. The skew is indicative of forces which influence the return pattern over time. In addition, the pattern of returns is very biased toward a positive result sequentially, until it breaks down, meaning prior results tend to influence the next result.
Given that you cannot expect that market returns will be random, inferring an expected return in the coming year with probability using normal probability theory is not possible. The only means of making an assessment is to obtain a good understanding of what causes the skew in market returns, and how the factors are expected to change over the next twelve months.
What Goes Way Up, Does Not Always Come Down
Most people instinctively expect that if a stock moves upward dramatically in value, eventually it overshoots in value and will pullback. The same rationale leads to expectations about the future direction for the stock market as a whole. To test whether this expectation is actually valid, I have extracted a subset of the market return data from 1952 through 2013. The subset is based on buying a share of the S&P 500 at any month end close when the market registered a year over year return of greater than 20%, and holding the share for 1 year. The 20% previous 12-month gain break point is very appropriate now because currently the market from November 2012 to November 2013 is up a whopping 27.5%. The size of the data subset is, maybe surprisingly to many, large at 174 out of 743 points in time since 1952.
If you compare the above graph visually with the comprehensive data reviewed in previous section, you might be convinced that the two data sets are nearly identical. In other words, making an investment in the S&P 500 after the market is up 20% or more over the last year will not negatively impact your expected return. In fact, the mean return of 9.8% plus dividends for this sub-set of historical market returns is slightly better than the entire time period. And, returns less than 0% happened only 26% of the time instead of 27%, and gains of greater than 20% happened more often at 27% of the time.
What is different between the two data sets? Statistically, the distribution of returns changed from being negatively skewed, to more evenly distributed. In simple terms, the data says the year after the market goes up by 20%, whatever was causing the data to be skewed in the aggregate, subsides. The removal of the skewing influence does not immediately affect market returns any more than would be expected at any other point in history.
Up 20%, Usually a Good Sign of a Slowdown within 2 Years
After the market records a gain of 20% or more, most investors expect some degree of mean reversion. If a pullback or major market correction does not happen in the year immediately after the market makes a run upward, when does it happen?
In the table above, you can clearly see that on average, there is a significant market adjustment within two years. The overall two-year CAGR for the market after a 20% gain year skews definitively more positive, as the average return 66% of the time is less than 10%, and less than 6%, 37.5% of the time. The data also shows, however, that 34% of the time the market two years after a major move up provided returns of greater than 10%. The data, although helpful, still seems to have an undercurrent which is counter intuitive.
Late 1990s Time Period Affects Return Distribution
Anyone who is remotely familiar with statistics has probably heard the saying, "figures lie and liars figure." If I had presented the historical data on returns one year after posting a 20% gain, the data might lead you to believe there is a good chance that 2014 will be another banner year. Possible, but in order to make that assessment for yourself, there is additional data which you might want to consider.
When the stock return data over the last 60 years is analyzed on a timeline, the 1996 to 1998 time period stands out as an outlier. A string of 20% plus year over year returns were posted for over 3 years. These returns happened following major market increases in1995, a rebound year after the near recession in 1994.
In order to share what happens if the "skewing" impact the bubbling 1990s had on the historical stock market returns, I excluded the data from the time series containing returns one year after a 20% gain was posted. The results are shown in the graph below.
Surprise, surprise - what most investors think should happen in the market after a major market advance of greater than 20%, actually statistically shows up in the data if you just void these 35 consecutive months of stock market history. The market returns are heavily skewed toward low or negative returns, and the average market return falls to 6.6% plus dividends.
Fed Bubble Machine Fighting the Inevitable Adjustment Process
In the above graph, the return distribution although significantly different from the entire historical time period, contains an additional data sub-set that as an investor you may find logical to extract given the present market scenario. Most cyclical market moves start with a bounce after a correction and propel forward until there is a market pullback. The last market pullback of any significance was the 2008-2009 "Great Recession." The current market scenario is more likely akin to a mid or late cycle up turn. In order to provide an understanding of what normally happens under this set of circumstances in the market, the data which corresponded to the returns registered the year after the market posted negative returns (i.e. a correction), were eliminated from the time series. The new time series is shown in the graph below.
This graph reflects the historical odds for stock returns the year after a 20% gain if the market was not recovering from a correction, and Fed policy was not channeling investment into the market by taking U.S. Treasuries out of circulation and holding interest rates below inflation. The average return in this historically simulated scenario is 3.1% plus dividends. But more importantly, over 50% of the time in the past under these circumstances the market posted negative returns 1 year after posting a 20% gain. The data did exhibit some above average positive returns under these circumstances. These returns happened primarily in the 1950s. It is worth noting, the time period recorded current trade account surpluses, not deficits.
Raising capital in the market during the year 2000 in Silicon Valley, the peak of the fundraising activity in the 90s boom, I recall plentiful capital and multiple investment firms, U.S. based and foreign, competing for individual deals. What I did not comprehend at the time was the heavy influence that U.S. fiscal and monetary policy was having on the market. In the late 1990s the U.S. curtailed its deficit borrowing and the Fed increased its holdings of publicly-traded U.S. Treasuries from 10% in 1995 to over 14% by the year 2000 - a significant quantitative easing move. A similar phenomenon is happening in the market today as government spending is not growing, and the Fed ownership of Treasuries (NYSEARCA:TLT) (NYSEARCA:TRSY) (NYSEARCA:SHY) (NYSEARCA:TLH) (MUTF:PTTRX) (MUTF:PTTAX) (NYSEARCA:BOND) is on the increase. How long the artificial stock market stimulus will continue is a function of when the excess market liquidity, which skews market returns, is forced to dry up. The market reversal in the late 1990s was slow to materialize as the Fed did not hastily retreat from the market.
The 1990 scenario began to break down in the year 2000. All that was required to whittle away at the soft underpinnings of the market valuation at that time was the Fed moving to a neutral position on asset purchases - no big purchases or sales of Treasuries, the balance sheet stayed fairly constant in the year 2000 which constituted tightening conditions. The big difference between then and now can be seen in interest rates and the economic statistics. The Fed was raising rates above 6% on Fed Funds, nominal GDP was above 6% and the unemployment rate was 4%. These are all conditions that the Fed continues to reiterate it does not see today, and therefore, it will continue to keep the Fed Funds rate low and continue to support the White House economic plan with its asset purchase plan.
Investing based on expected political outcomes is risky, as the last graph in the article shows. Putting new money to work this late in an economic cycle without major Fed support is putting your money at serious risk of visiting "money heaven." However, the likelihood as we head into 2014 of monetary tightening without some long tail event occurring that forces the Fed to stop QE is low - by definition these events are low probability. The obvious long tail events that would force a Fed adjustment are ones generated from the geo-political space such as a major foreign Treasury bond sell-offs (foreign ownership actually increased in September) or an oil shock (market sentiment is definitely toward price containment, not increase). On the domestic front inflation increasing (still no CPI juice) or even a sudden increase in government spending (Obamacare - what is the real cost?) are likely culprits, but again the signs remain benign. Every market signal that I follow continues to indicate deflation or negative tendencies, except the Fed and the stock market.
Personally my recommendation is to keep new money put to work in stocks (MUTF:VITSX) on a very short leash in this market over the transition to the New Year as the major Investment Houses push the pull-back theme and investors start to do year-end assessment. However, until an event occurs that truly knocks the Fed policy off course emerges, it is not time to bail out of the market. Although the market, in my opinion, continues to be detached from fundamental reality, the Fed is not going to dramatically change course under the present circumstances, meaning the available investment alternatives in the U.S. market will continue to be priced at a premium.
Daniel Moore is the author of the recently published book Theory of Financial Relativity. All opinions shared in this article are expressly his own, and intended for information purposes only.