Recent expansion in the Federal Reserve balance sheet appears likely to send stocks down or keep them flat in the next few months, rather than send prices up. The belief that the bull market of the last four years came mainly from Fed action (balance sheet expansion / reserve creation / money printing) appears to be based on simplistic and likely wrong analysis of the data. Hope that the Fed can and will prevent any significant decline in stock prices, sometimes referred to as the "Bernanke put," appears to be baseless. The unraveling of that hope may begin in March.
We will look at data leading into the crisis as well as after to add context to what has happened in the last four years. We'll look beneath the upward trend in both the Fed balance sheet and stock prices, at their rates of change, where the actual relationships, if any, are revealed. It should become obvious the Fed does not have the power to stop a bear market in stocks.
Here are several of the monetary variables in the last seven years. The Fed balance sheet is weekly data, the others are monthly.
Leading into the crisis M1 and currency in circulation were growing at less than their long-term trend. Since the crisis they have been increasing at a faster rate than their long-term trend and have more or less caught up to where they would be if they had grown at the historical rate the last eight years. The growth in M2 on a long-term basis has roughly stayed in line with its historical growth rate, although the annual growth rate has varied significantly since the crisis.
As the crisis unfolded the Fed dramatically increased its balance sheet and the monetary base. Since that explosion in late 2008 these series have become much more volatile even though their long-term growth rate seems to be only modestly higher than before the crisis.
Here are weekly data points for the S&P 500 (average of daily closes for each week) and the Fed balance sheet. On the time scale the weekly data points are plotted as if they occurred on Wednesday.
If you begin looking at the data from the March 2009 low the balance sheet and stock prices have an r-squared of 0.74. I have seen someone portray this relationship with an r-squared of 0.90 and claim 90% of the gain in stock prices is due to Fed money printing. I suppose if you were a little more selective with the beginning point and perhaps smoothed the data a bit you could get from the 0.74 I calculate to 0.90. That seems a bit inappropriate since the relationship from the October 2007 high has an r-squared of 0.00.
Just because two data series move in the same direction for a few years does not mean the moves are related. When two data-series both trend up, they have a high statistical correlation or r-squared. However, to understand if there is a significant relationship requires looking at de-trended versions of the data, perhaps using rates of change or first differences. To understand why this is so let's consider something we have more experience with, inflation.
In the last 75 years stock prices and the consumer price index both trend up and have an r-squared of 0.84 using monthly data. If this represented the actual relationship between the two series you would expect high rates of inflation to correspond with raging bull markets. However, it is just the opposite where high rates of inflation tend to correspond with bear markets.
Below is the balance sheet and stock data presented as 13-week rates of change (roughly a quarter of a year).
The most prominent feature of the chart is the 13-week period ending 11/19/2008 where the Fed balance sheet expanded 143%. In these same 13 weeks the Fed Funds rate was cut from about 2% to almost 0% and the S&P 500 lost more than 36% of its value, which was its worst 13-week period during the crisis. The most extraordinary stimulus in Fed history did not protect against a bear market.
The rocketing of the balance sheet growth rate to 143% and back would skew any statistical correlation, so we are going to look at the data following that rocket ride with four different lead times. Each of the four charts has two parts, a scatter plot on the left and a time-series plot on the right. The scatter plots have a best fit line that shows the best statistical relationship between the growth in stock value and the growth in the balance sheet. The time-series plots show the same data on a time scale. The scale for the balance sheet in the time-series plot is calibrated to the correlation shown in the best fit line on the scatter plot. The time series also have some shaded data that is not in the scatter plot, but is provided for context.
First we will look at the concurrent plot where there is a modest positive correlation between stock prices and the balance sheet. The r-squared is 0.08. Each point on the scatter plot represents a 13-week period where the growth in the balance sheet is shown on the horizontal axis and the change in the stock price is shown on the vertical axis. For example, the highest point shows stock prices rising 35.3% in the 13 weeks ending June 3, 2009. The balance sheet expanded 9.3% in those 13 weeks. It is easy to pick out this highest point on the time-series plot on the right.
When data series are concurrent it is difficult to say which one influences the other. The answer is clearer if one variable moves and the other responds after a lag. Sometimes really significant relationships will oscillate where there is a positive relationship with one lead time and an inverse or negative correlation with a different lead. A baby boom, for example, has multiple positive and inverse correlations with stock prices during the course of the life cycle. It is conceivable such a significant relationship has developed between the Fed balance sheet and stock prices. The correlation above is pretty modest and it gets weaker if you start looking at lead times for the balance sheet. The correlation gets stronger if the stock market leads.
The chart above shows the balance sheet lagging 13 weeks relative to stock prices plotted on the time scale. So for example the highest point still shows a 35.3% increase in stock prices for the 13-week period ending June 3, 2009, but it is now aligned with the 0.4% increase in the balance sheet for the period ending September 2, 2009. The implication is that if stock prices go up it has a positive influence on the Fed balance sheet 13 weeks later.
As I mentioned earlier the correlation got weaker if the balance sheet leads. As the lead increases, the relationship inverts. The strength of the inverted relationship peaks with the balance sheet leading 16 weeks and an r-squared of 0.26. This inversion is shown on the time-series plot with an inverted red scale for the balance sheet, where minus 40 is at the top and 40 at the bottom.
So the increased growth rate in the balance sheet over the last 16 weeks now appears to become a downward influence on stock prices for the next 16 weeks. The stock markets' 13-week rate of increase appears to have peaked in the week of February 13 at 11.4%. The rate for the week of February 20 was down to 8.7%. As I write on Monday the 25th it looks like the rate of change will continue weakening.
If you continue increasing the lead time the inverse correlation weakens and then turns positive again. This positive correlation becomes strongest with a 49-week lead time. It is only strong relative to lead times shorter or longer; it is actually pretty weak with an r-squared of only 0.09.
What little influence this chart implies suggests the increase in stock prices will weaken from here and stay weak till October.
No firm convictions should be drawn from the four years of data where the relationships between the Fed balance sheet and stock prices while modest appear to be more significant than they were before the financial crisis. Four years of data is not enough time to conclude the nature of the relationship has materially changed from what it was before the crisis, which was rather insignificant.
The tentative conclusions are that the most significant relationship is inverse, where rapid balance sheet growth reduces stock market (SPY) advances about 16 weeks later. The two correlations with the balance sheet leading both suggest a weaker stock market in the next few months.
I find no convincing evidence the Fed manipulated stock prices higher in the last four years, or that it can prevent a bear market. The perception the Fed has pushed stock prices higher, will continue to push them higher and can and will prevent any significant decline may have helped inflate a market bubble. When a bear market comes whoever hopes to be saved by a Bernanke put will probably hope in vain.
Additional disclosure: There is no guarantee analysis of historical data their trends and correlations enable accurate forecasts. The data presented is from sources believed to be reliable, but its accuracy cannot be guaranteed. Past performance does not indicate future results. This is not a recommendation to buy or sell specific securities. This is not an offer to manage money.