Reality Versus the Expectations of Reality

Journal of Financial Planning; March 2012


Mark W. Riepe, CFA, is a senior vice president at Charles Schwab & Co. Inc. and president of Charles Schwab Investment Advisory in San Francisco, California.

A frequent question I get from casual investors is why the stock market rose (fell) when the economic news of the day is bad (good). They often assume that bad economic news should result in an immediate negative movement in stock prices, and vice versa.

There are a lot of reasons this relationship isn’t nearly as tight as investors suppose. Just a few examples are:

  • Multiple factors. There are many things going on in the world on a given day, and many of these events have at least some influence on stock prices, so a client shouldn’t get overly fixated on any one piece of data.
  • Different interpretations of the data. The interaction between buyers and sellers ultimately determines the price of a security. Investors and traders are humans who interpret new information differently. After all, if everyone agreed about everything, we wouldn’t have so much trading volume.
  • Skepticism about data quality. Investors don’t always believe the numbers reported, knowing that all economic data is subject to revision.

However, one reason seems to baffle many casual investors—the market moves not so much in reaction to news but instead moves in reaction to news inconsistent with what was expected.

My impression is that the financial media do a good job of positioning new economic news relative to what numbers were expected, but the mainstream media, which devote a few minutes in a given broadcast to financial markets, rarely go to this trouble.

So for this column I decided to conduct a little empirical study to see whether it really was the case that expectations matter more than the actual numbers themselves, in hopes of creating some clear evidence that could be used as a teaching example.

Raw Numbers

I chose weekly initial jobless claims as the economic data point to study. I could have picked any economic data series, but I picked this one for several reasons:

  • It is a variable both professional and casual investors pay attention to. Right now there’s so much focus on jobs as a symbol of the economy’s health. In fact, every Thursday, almost without exception, a lead story on the all-news station I listen to going to work is the number of people who have initially filed for unemployment claims the preceding week.
  • Clear link between it and the market. Examples that illustrate a concept are best if they’re clear. In this case, investors pay attention to this number because it provides an early glimpse (in other words, a leading indicator) into the health of the labor market and subsequently the economy. The state of the economy has at least some bearing on stock prices.
  • High frequency. This variable comes out weekly, and that means there will be lots of data points to observe. This is important when trying to gain some semblance of statistical significance.
  • Expected data are available. Not every economic data series has a large number of people who create expected numbers for the series available to analysts such that one can get a reliable sense about the consensus of the market before the newest results were announced.

Before analyzing the data it is best to formulate a hypothesis. A reasonable hypothesis a client might formulate is that if the claims for a given week are lower than the preceding week, that is good news for stocks, and prices will rise. Using statistics jargon one would expect a negative correlation between the change in initial unemployment claims and stock prices on the day the data are announced. If claims go up (down) then stocks prices go down (up).

Does that actually happen in the real world? Yes, but the correlation is weak and inconsistent. If we start with claims data from October 2009 to January 2012 and correlate that with the movement in the S&P 500 on the day the claims data were announced, the correlation is slightly negative but far from statistically significant.

I used the October 2009–January 2012 period because I wanted recent data, and a shorter period would make it virtually impossible to come up with a statistically significant result.

Not surprisingly, the hypothesis is rejected, and that is the source of confusion for many casual investors. What they are missing is that the market is a forward-looking animal. At any given moment prices reflect a consensus of sorts regarding the expectations about future conditions. To the extent historical or backward-looking numbers are used, it is only because the investor expects the past to have some bearing on the future.

Numbers Couched in Expectation

So let’s push this teaching example further and formulate a better hypothesis. We need one that incorporates the notion that the market reacts when the actual news is at odds with what it was expecting—in other words, when important news comes in at odds with the market’s assumptions, it adjusts prices accordingly to take into account the new information. If we want to gauge the impact of initial unemployment claims on stock prices we need to compare the actual claims numbers to what was expected. The specific hypothesis is that if the weekly number is higher (lower) than the consensus, that’s bad (good) news, and stock prices will fall (rise).

Using the same time period, we get a much better result. The correlation is negative as before, but it’s actually even more negative, –0.16 versus –0.10, and just barely misses being statistically significant at the 95 percent level. In fact, if I had gone back and modified the claims number to include the number originally announced instead of the final, revised numbers in the government databases, I suspect the results would have been stronger.

This, of course, is just one variable, but it wouldn’t surprise me if we were to get similar results doing a study much like this one across a wide variety of closely followed economic data points.

The lesson for your clients: stock market reactions to the daily flow of news are not so much about the news itself—good or bad. They are about whether that news was expected, leading to a reaffirmation of existing market valuations, or a surprise, leading to a reassessment of current valuations.

Investment Planning