Does GDP Growth Predict The Direction Of Stocks?
Since then Mr. Mauldin has written to me to offer clarification, pointing out that he does not necessarily agree with everything he publishes in his columns. Though I’ve asked Mr. Mauldin several times asking him to state whether he DISAGREES with the theory that economic growth and performance are unrelated, he did not. So leaving aside the unresolved issue of whether this columnist is disagreeing with Mauldin or Grantham or both, the notion that growth and stock valuations are unrelated is an idea worth rebutting no matter who holds it.
One of the cognitive distortions which must be dealt with first, though, is the one which knocks down the straw man who says that economic growth precedes rises in stock market valuation. If one looks at a chart which shows the relationship between economic growth and stock market valuations one year later (and we happen to have one on hand), one sees that such an approach is worse than useless as a tool to anticipate stock market valuations; in fact it sends out a false signal, showing a small positive relationship between economic growth and future risk premiums in the stock market.
Not only are the correlations much more poor at 3.6%, but the slope of the line is wrong. If investors were paying attention to growth a year ago, then there would be an inverse relationship between risk premiums and growth. But the data above to the small degree that they show any relationship at all, show the opposite relationship.
Why is that?
Because stock and bond investors are always trying to look into the uncertain future and assess the risks and rewards associated with growth, shifting resources back and forth between fixed income and co-ownership, as they anticipate either bust or boom respectively. Please note the word ‘anticipate’ in the prior sentence. Equity risk premiums are much more a matter of anticipated future growth differentials then they are of past or even present.
Ignoring this fact can cause analytical flaws when it comes to studying the relationship between growth and equity premiums, because in this case the effect comes before the cause. Economists who think of their science as being in essence no different from natural sciences like physics tend to see the world in terms of cause then effect. This makes sense: atoms and molecules do not think, they do not look into the future and see another molecule headed towards them which will bump them in a new direction and respond by moving out of the way. So (leaving aside some strange new speculation in quantum physics) causes come before effects in the non-sentient world of things.
But the sentient world of people is different. Hovering between beast and God, humans attempt to look into the future and respond to what they think is coming. In statistics this effect is called Granger Causation, and it renders traditional cause and effect statistical tools a bit too crude to deal with all of the problems in the more nuanced sciences of human action.
If you look again at the chart in this previous column produced by Lattice Strategies, the one with the green bubbles, you’ll see some evidence of the effect associated with Granger Causation. The shade of green varies with the expected GDP growth rate. What you see is that those countries which are a darker shade of green (i.e. those countries which forecasters expect to have higher GDP growth in the future) tend to fall above the line, meaning that they tend to have higher returns even than one would expect given their past GDP growth. It appears that some of their performance is based on GDP growth in the past, but some of it is based on the anticipation of higher growth.
The chart above shows that there is a relationship between equity risk premiums and future GDP, and that the relationship is what we would expect it to be given the theory: the higher the risk premium, the lower the future growth. Mathematically stated: the two factors have a negative correlation of 35%. Visually, that means that the line which describes the relationship slopes downward as you can see above. Also, the data does not cluster much around the line. It is not a very good fit.
This is also understandable because this measures the relationship between trailing earnings and GDP one year later: in other words the data spans up to two years’ time. Anticipating the future is very difficult, and investors are always guessing and always updating their guesses, but almost never getting their forecasts of the future exactly right. Investors tend to anticipate future changes in GDP growth, but inexactly.
But the main point is that investment is a forward-looking process. If you reverse the order and look at how much GDP growth in the past anticipates risk premiums in the future, you find that they do not.
This is one reason why attempts to sever the relationship between growth and stock valuation miss the point: they ignore the forward-looking nature of the entrepreneurial/investor function. This is what happens when analysts get too used to looking at equations and lines and charts and forget that they are dealing with persons possessing human agency, not with billiard balls possessing only vector and momentum.
So, aside from all of the well-known criticisms about GDP and what it measures vs. what it purports to measure, we have the larger problems associated with trying to use it as a stock valuation tool. It always applies to the past, but stocks are oriented towards the future. Despite the well-worn cliché about the dangers of driving by looking in the rear view mirror, nevertheless the nature of backward looking statistics proves to be just too tempting.
Financial media need something to talk about, just like any other media, and people like to go on TV in the role of expert. So every month, when there’s a new ‘print’ on GDP, people go on the air to talk about what it means to the stock market. The answer that it means nothing to the stock market now, that the stock market cared about Q4 GDP during Qs 1-3, beforehand, in anticipation feels a little unwelcome.
The better answer is to stop chasing stats which are already dead the first time you hear them. Yes, you can learn a lot by studying the dead, you can see the past patterns of the world and learn from them. For example, you can learn from past statistics how destructive excess government spending is to growth, not just in the ‘end game’ but in real time (another area in which it appears that John Mauldin has taken a mistaken view; perhaps more on this in future columns). But the best way to anticipate future stock valuations is not to look at stale growth stats, but to look at national behavior. At demographics, cultural and policy conditions. But that takes far more careful thinking than simply reacting to stats on release day.
Article originally published on Forbes.com.