| |
 July 2004 No Time Like the Present
Analysts continue to be divided on this issue. Some argue that it isn't really a bearish scenario at all. Rising rates are the result of a strengthening economy and improving corporate profits. They're more of an effect than a cause. It's earnings -- not interest rates -- that are predictive of share price movement. This is supported by a statistical study conducted by investment bank Ryan Beck and reported in the May 6th edition of The Wall Street Journal. It showed that over the past 42 years, earnings growth of the S&P 500 explained 90% of the index's price movement. The correlation with interest rates was considerably less. Yet this doesn't completely eliminate bearish concerns. Why? Because earnings themselves are affected by interest rates. As rates rise, the cost of borrowing increases. Businesses pull back on their investments and more cash flow is diverted to servicing debt. Demand also begins to slow in a tighter monetary environment. To be sure, profits don't immediately turn into losses, but they are harder to come by. So even if interest rates aren't directly related to stock prices, it stands to reason the they are at least indirectly related. That's why the bears fear last year's rally is about to evolve into this year's selloff.
Lessons from the PastIn an effort to get a handle on what to expect in the short-term, many analysts have looked back previous periods of Fed tightening. Most memorable to today's investors included the stock market crash in 1987, the bond market crash in 1994, and the recent three-year bear market. Each of these events occurred shortly after the Fed moved from an accommodative to a tightening stance.Quants rely heavily on market history too, but they also know that significant differences between periods can severely diminish the value of such studies. We would suggest there are substantial differences between the conditions of the 1987, 1994, and 2000 financial markets and those of today. At 46-year lows, interest rates are considerably lower than at the beginning of almost any previous tightening cycle. As a result, the initial moves should not have such an immediate damping effect as those in prior periods. The S&P's P/E was much lower in 1987 and 1994 and much higher in 2000 than it is today. Currently it's around 21x prior year's earnings while it was as low as 12x in 1987 and 14x in 1994. Conversely, it was at 22x at its lowest point in 2000 after peaking at 31x. Lower P/Es tend to indicate a sluggish economic environment. Rising rates slow it more. The Fed was able to minimize the effects of the 1987 selloff by pumping money back into the economy. They weren't so fortunate in 1994 when the bond market was caught off guard as rates doubled from 3% to 6% in just 11 months. The exact opposite was true in 1999-2000. Fed Chairman Alan Greenspan had already been concerned for several years about investors' "irrational exuberance". P/Es had skyrocketed from speculative buying and (in hindsight) falsely inflated earnings. The Fed moved to "prick the bubble" in an effort to minimize the damage that would inevitably occur.
The inflationary environment is different now, too. A little inflation is actually good for businesses since it affords them some pricing power. It can also inflate (pardon the pun) reported earnings as well. Inflation was running higher in 1994 and 2000 and although it was slightly lower in 1987, it was a temporary respite from the prior 10 years. Although picking up now, inflation is still well below the historical averages. But these differences don't mean we still can't get some insight from the past, only that it will take a little more work than simply comparing today's environment to those of the recent past. Actual RelationsRather than speculating about the effect of rising rates and how they should affect the markets, a better approach is to consider what relations they've had in the past, not just in specific periods. We used data from Ibbotson Associates going all the way back to 1926.Obviously two things to compare were short-term rates and stock price levels. To do this we started with the 30-Day Treasury Bill and the S&P 500 Index. Since this is a study on the effect on equity price levels, we only used the S&P 500's capital appreciation, not total return including dividends. Instead of the T-bill's nominal (reported) return we used its real yield. This factors in the effect of inflation that many analysts think has a bearing on the market. After all, it stands to reason that a tightening policy will have a greater effect if real rates are already high than if they start from a low or even negative level. To consider the effects of the P/E level, we calculated values using historical downloadable data available from Professor Robert J. Shiller's website. As you'll see in a moment, this served several purposes.
Finally, we looked at each set of relations under four different scenarios. The most obvious is to simply look at the effect of rising interest rates on current market conditions. Of course there's often a delay between rate movements and their actual impact on the market so we also considered the market effects 3, 6, and 12 months later. The results offer little support for the experts' oft-repeated prognostications. In fact, they don't suggest much of a trend at all. Over the 936 months observed, there were 42 instances where interest rates rose a minimum of three consecutive months after holding steady or falling. Chart 2 shows the frequency of returns on the S&P 500 3, 6, and 12 months after rates began rising. The only noticeable pattern here is one of greater dispersion as time passes. Returns are skewed positively three months after rates begin to rise but begin to move towards neutral after six months. But after twelve months have passed, returns are almost evenly distributed from +15% to -15%. The changes of the three time periods suggest that the effects are greater as time passes -- but not much more. Indeed, the fact that returns are so evenly distributed after twelve months indicates that there really is no trend at that point. In other words, when rates start rising, the market is just as likely to be up as down twelve months hence. We thought the P/E might have something to do with this, but it really doesn't. If anything, there's less impact on share prices when rates start to rise when the P/E is high than when it's low. Perhaps that's because high P/Es signify a booming economy and strong corporate profits. Under those conditions tighter monetary policy may have less of an immediate effect. Or, of course, the whole "relationship" could be spurious. It appeared strongest three months after rates started up, but when we looked at the correlation between rate increases and the S&P 500 price level there was no statistically significant relation at 3, 6, or 12 months. You can get a sense of this from Chart 3. This is a scatter plot of T-bill rates and S&P 500 appreciation with the latter being led by three months. Each point is an ordered pair with the T-bill value plotted on the horizontal axis and the S&P 500's lagged return on the vertical. As you'll notice, there's a big mass in the center of the chart with only a few outliers. The red line in Chart 3 is the "regression line", or the line of best fit. Believe it or not, it minimizes the distances between each point and the resulting line. (We'll spare you the calculation behind it.) When two series are highly correlated with one another, each point on the scatter plot falls on or very close to the regression line. That's certainly not the case here and remember, this is the most highly correlated of the time periods.
The coefficient of determination or R2 shows how much of the S&P 500's price change is explained by the changes in interest rates. Again sparing you the calculation, it's .001547 for the relation shown in Chart 3. In other words, changes in interest rates explain a whopping .1547% (that's not a typo) of the changes in the S&P 500's price level when the latter is led by 3 months. When relations are this miniscule, they usually aren't statistically meaningful. That's the case here too as the price changes in 3, 6, and 12 months all fail the so-called "F-test" for significance. You can always find a regression line and a description of a relation no matter how unrelated two series of numbers might be. The fact you can doesn't prove they really are correlated, especially if the relation isn't statistically significant. History is Not on Their SideRemember this the next time you hear one of CNBC's talking heads describing what will happen to stock prices as the Fed moves rates higher. They may refer back to similar situations in the past or imply history backs them up, but you know better.The fact that any relationship between interest rates and share price movement is so weak -- if it exists at all -- makes it almost a toss-up when considering what will happen this time. This is clearly illustrated by the return distributions of the final histogram of Chart 2. To be sure, some analysts will correctly predict the direction of the market in the months following the Fed's first rate hikes, but then again, they might be equally correct even if interest rates stay unchanged. In other words, they won't be right because of their acute knowledge of historical relations, but rather because they're simply better prognosticators. Or maybe they're just lucky, after all, even a blind hog occasionally finds an acorn. The one thing we can say with a fair amount of assurance is that for this issue, neither history nor statistics offers much support. Search this site! Just enter you key word or words:
Get current quotes or follow your own custom portfolio,
courtesy of E-Line Financials:
|
|||||||||||||||||||||||||
FTER MONTHS OF SPECULATION, IT'S now clear that interest rates are rising and will continue to do so, at least the short-term. What still remains to be seen is what impact this will have on share prices and the overall economy.
