Friday, January 31, 2014

Predictability and correlation


Today another little note that I discovered while teaching. Warning: this will only be of any interest at all to time-series finance academics. I'll try to come back with something practical soon!

Does the predictability of stock returns from variables such as the dividend yield imply that stocks are safer in the long run? The answer would seem to be yes -- price drops mean expected return rises, bringing prices back and making stocks safer in the long run. In fact, the answer is no: it is possible to see strong predctability of returns from dividend yields, yet stocks are completely uncorrelated on their own.

I've been through three versions of showing how this paradox works. In  Asset Pricing the best I could come up with was a complex factorization of the spectral density matrix in order to derive the univariate process for returns implied by the VAR. In later Ph.D. classes, I found a way to do it more simply, by seeing that returns have to follow an ARMA(1,1), and then matching coefficients. This year, I found a way to show it even more simply and intuitively. Here goes.


Background: The mean and variance of two year log returns is
E(rt+1+rt+2)=2E(r)σ2(rt+1+rt+2)=2σ2(r)+2cov(rt+1,rt+2).

If returns are independent over time, the covariance term is zero, and both mean and variance scale linearly with horizon. The ratio of mean to variance E(rt+1+rt+2)/σ2(rt+1+rt+2) which (roughly) controls portfolio allocation is then the same at any horizon. If the covariance term is negative, stocks bounce back after declines, so the variance scales less than linearly with horizon, and stocks are safer for long-run investors. (Yes, I'm mixing logs and levels like crazy here, but this is a back of the envelope blog post.)

So, our issue is, are stock returns correlated over time? Stock returns are, in fact, predictable from variables such as dividend yields. But that does not mean they are predictable at all from past returns, correlated over time, and thus any safer for long-run investors. This is the  nice little paradox.

Sum up the simplest version of stock predictability with a vector autoregression
rt+1=brdpt+εrt+1Δdt+1=bddpt+εdt+1dpt+1=ϕdpt+εdpt+1

Here dp is the log dividend yield and Δd is log dividend growth. The return coefficient is about br0.1 and the dividend growth coefficient is about zero bd0, and ϕ0.94.

So, low prices mean high subsequent returns, and high prices (relative to dividends) mean low subsequent returns. It would seem that stocks are indeed much safer for long-run investors, as there really is a sense that low prices are temporarily low and will revert if you can wait long enough.

More seductively, if you plot the impulse-response function to dividend growth εdt shocks and dividend yield εdpt shocks, you see the former is a cashflow shock, giving a one-time shock to returns and a random walk in prices. (Top) But the dividend yield shock is an expected return shock, yielding a completely temporary component to prices -- green line bottom right. If prices go up and dividends go up too, the movement is permanent. If prices go up and dividends do not move, the price movement is completely transitory, and perfectly safe in the long run. Stocks are like long term bonds plus iid cashflow risk.

You would think therefore that just seeing prices go up, with no information about dividends, you would have something between the two; a partially transitory movement in stock prices that is somewhat safer in the long run.

You would be wrong.

Let's figure out the correlation of returns cov(rt+1,rt+2) implied by this little VAR. Use the VAR to write
rt+1=brdpt+εrt+1rt+2=brϕdpt+brεdpt+1+εrt+2

so
cov(rt+1,rt+2)=cov[brdpt+εrt+1,br(ϕdpt+εdpt+1)+εrt+2]cov(rt+1,rt+2)=b2rϕσ2(dpt)+brcov(εrt+1,εdpt+1)

The first term b2rϕσ2(dp) induces a positive autocorrelation or momentum, making stocks actually riskier for long term investors. dpt moves slowly over time, so if returns rt+1 are higher than usual, then returns rt+2 are likely to be higher than usual as well. This term makes stocks riskier in the long run.

The second term brcov(εrt+1,εdpt+1) is strongly negative. If there is a positive shock to expected returns, this sends current prices and hence current returns down. In this way stocks are like bonds: if yields rise, prices fall and current returns fall. This is the safer in the long-run term.

The fun part: In the standard parameterization, these effects almost exactly offset.

To show this fact, we need to find just how negative cov(εrt+1,εdpt+1) is. Campbell and Shiller's linearized return identity,
rt+1ρdpt+1+dpt+Δdt+1;ρ0.96

implies the VAR coefficients and errors satisfy the identities.
br1ρϕ+bd1ρϕεrt+1ρεdpt+1+εdt+1

Now, empirically, dividend growth shocks and dividend yield shocks are just about uncorrelated, cov(εdt+1,εdpt+1)0. So, the correlation we're looking for is
cov(εrt+1,εdpt+1)=ρσ2(εdpt+1).

Now,we have an expression for the covariance of return and dp shocks, so we can continue
cov(rt+1,rt+2)=b2rϕσ2(dpt)brρσ2(εdpt+1)=br(ϕ1ρϕ1ϕ2ρ)σ2(εdpt+1)

If we had ϕ=ρ=0.96=0, we get the result, cov(rt+1,rt+2)=0. Now, ϕ0.94 is the usual estimate. But ϕ is an OLS\ estimate of a very persistent series, so biased down. ϕ=0.96 is not that far off.

So, pretty much, dividend yield predictability coexists with complete lack of return autocorrelation. The momentum effect of a slow-moving forecast variable exactly offsets the bond yield effect that high prices mean low subsequent returns.

That does not mean that predictability is unimportant for long term investors. A properly done (Merton) portfolio theory isn't as simple as
w=1γE(Re)σ2(Re)
.
It turns out there is a market timing demand and a hedging demand. The hedging demand is positive in our case. So, it's still possible that long-run investors should put more into stocks, even though simple Sharpe ratios are not better at long horizons. I haven't yet found a simple way to calculate hedging demands however.

Update: A correspondent tells me this example is in a set of John Campbell lecture notes he "inherited distantly."

10 comments:

  1. "Does the predictability of stock returns from variables such as the dividend yield imply that stocks are safer in the long run?"

    Safer than what?

    ReplyDelete
    Replies
    1. Stocks in the short run. Are risks lower for people who can invest for a longer time.

      Delete
    2. There is no such thing as a long run stock versus a short run stock. Can't say the same thing with options, bonds, and other securities that are time sensitive.

      And so any "safety" derived from investing in stocks long term versus short term depends in part on what the investment time horizon of stock investors is.

      Contrast that with bonds. I can buy a 5 year bond and someone else can buy a 30 year bond. If I sell my 5 year bond, it will have no influence on the price of the 30 year bond bought by that other person. Bonds have time sensitive market segmentation built in.

      Delete
    3. But if stocks follow a random walk/noise in the short term yet follow 'logic'/discount cash flows in the longer term, you COULD say that LT investing is 'safer' (as in 'less random') than ST trading.

      Remember 'the market behaves as a voting machine in the short term and acts as a weighting machine in the long term'?

      Delete
    4. Frank, the point IS the investment horizon...are stocks safer for people that invest over 20 years or over 1 year horizons. Are you more likely to lose money if you invest for just 1 year in the stock market or 20 years in the same market.

      Delete
    5. Anonymous,

      But the investment horizon is not a property of the stock itself. That was the point I was making. It is a property of the investor group (investors with long term horizons versus investors with short term horizons). The safety comes from being with the majority group. If only one person was a long term stock investor and every other person was a short term stock investor, then investing in stocks for the long term is not safe.

      Other asset classes have an investment horizon built into the asset class itself (for instance bonds).

      So when someone says that "stocks are safer in the long run" what they should really say is that investing in stocks for the long term is safer than investing in stocks for the short term. The stocks themselves are not safer, the investment style is safer.

      Delete
    6. Frank, we know.

      Delete
  2. Could you say something about how does your calculations compare to the results of The Term Structure of the Risk: Return Trade-Off by John Y. Campbell and Luis M. Viceira (2005). It seems to me that in their case the coefficients do not cancel out each other and stocks have lower variance over long term (the term structure) and hence the stocks seems to be safer in long run. Is it a disagreement on the value of coefficients or am I missing something more fundamental?

    ReplyDelete
  3. All assuming, of course, that volatility is the correct proxy for risk.

    ReplyDelete
  4. Hello,

    I have a quick question. You say in " Discount Rates" that there is predictability in the long run. So let's say we can predict 5 years ahead but we can not predict 1 years from now. So why don't we run a regression that predicts 5 year returns, but run this regression 4 years ago. Would that be significant to predict 1 year excess returns ?

    Thanks for your time

    ReplyDelete

Comments are welcome. Keep it short, polite, and on topic.

Thanks to a few abusers I am now moderating comments. I welcome thoughtful disagreement. I will block comments with insulting or abusive language. I'm also blocking totally inane comments. Try to make some sense. I am much more likely to allow critical comments if you have the honesty and courage to use your real name.