Monday, November 22, 2010

Why the Equity Premium Isn't A Risk Premium

As I've said, there's a new consensus forming that while risk premia do not exist within asset classes they do exist across assets, with the signature example of cash, bonds, and equities. Yet, the premium dfrom cash to bonds is more like a cash premium, easily explained by the transaction benefits of cash, as opposed to any 'risk premium' in bonds. After all, you can pay for most things in cash, and that property is worth something. The equity premium of stocks over bonds, however, has for decades seemed like obvious proof of 'the risk premium'.

Yet, the equity premium for your average rube is hardly what the indices suggest, because taxes, transaction costs, adverse timing all make your average return much less than the S&P500. But even without that, consider what the equity return premium is. It should be a payment for risk, where risk is a covariance with something that proxies the happiness of investors.

If you look at the portfolio constructed by going long a beta 0.5 stocks, and short beta 1.5 stocks, the beta of that portfolio is about -1.0 (find such portfolios here). This is a costless portfolio, and its annual return of around -1% annually seems insanely cheap insurance. Basically for this modest fee, you can add it to the S&P500 and presto! Beta for your equity investing is now basically zero for only a 1% a year! Most of the 'Equity Premium' can be captured by adding this costless -1 beta hedge to the S&P500.

If the S&P500 premium is due to 'risk', the risk factor is orthogonal to the S&P500 itself. What risk factor would this be? The fact that it can only exist as a subtle and obscure latent factor makes it inconsistent because it is also so vivid it generates a ubiquitous and omnipresent price impact. It's not as absurd as a theory based on wanting to kill your father and have sex with your mother, but it's up there.


Anonymous said...

That the full ex post erp was not achievable is a fair point.

But involking the low vol/beta effect again to somehow disprove the erp seems odd.

Anonymous said...

Also, the necessary hedging to achieve the zero beta returns was even less achievable than the erp, a lot less, yet somehow that does not bother us? Tcosts, the lack of existence of futures, etc, make this way more theoretical than just long stocks.

Dave said...

"Basically for this modest fee, you can add it to the S&P500 and presto! Beta for your equity investing is now basically zero for only a 1% a year!"

For that to be the case, what percentage of your capital would have to go into the long beta 0.5 / short beta 1.5 portfolio and how much would go into the S&P 500?

Eric Falkenstein said...

Well, shorting is always a bit uncertain, because it is never really clear what you can short. But the nice thing about shorting 1.5 beta stocks, as opposed to the 'highest beta' stocks, is this is much more feasible. But doing this only makes sense institutionally, for a large firm, as an individual would have many hurdles making it much more onerous. Yet, it is possible institutionally, so at the margin, I think it's a realistic strategy that should affect how things are priced in equilibrium.

Anonymous said...

You could say precisely the same things for obtaining the equity risk premium (it can be done by an institution better), even stronger as shorting is more costly and skills at it less wide spread.

Your story, equity risk premium irrelevant as hard to really achieve, but beta effect really relevant, even when it's harder to have achieved, does not hold together.

Choose which side you want to let go of, that's the only choice.

Eric Falkenstein said...

anon: I'm arguing that to the extent the equity premium exist, it is independent of any covariance it has.

Anonymous said...

It seems like you want to replace one error, "risk is always rewarded", with a different error, "risk is never rewarded".

Markets aren't totally rational but they aren't totally irrational either. Reality is messy. If your theory isn't messy, it's wrong.

Anonymous said...

I'm the original top anonymous poster not the new anonymous poster, but I agree with him/her. I think the answer is somewhere in between. I completely accept that the ERP wasn't nearly as good as it looks for all the real world and survivorship bias problems (academia agonized over this for years and still does, with ideas like "habit formation" and other complex answers, when the answer is so much simpler, nobody really got that ERP!). But, I don't know how you can completely reject that part of the ERP has come from people realizing it's the risk that dominates their portfolio. And I don't see how the success of any anamoly (in this case the low beta one) fully obviates that.

I think the observation that the ERP isn't what people think is important, and the low beta anamoly important, but don't see the utter "complete risk is always priced at zero" conclusion.

I love the line "if your theory isn't messy it's wrong", but would add that if your theory is too messy it's probably not too useful as it's probably just a bunch of connected stories explaining (fitting?) the data. Of course the trick is to have a theory just simplified enough to be useful... When I figure out such a theory I'll post it...

Finally, while I've never been accused of being a hyper-bull (sometimes the opposite) I'll note that equity valuations aren't that much higher than historical experience (maybe a Shiller PE of 20 vs. 15). In that case, they are probably poised to deliver below, but not excessively below, historical experience, and now we are in a position to actually receive that return (low cost index funds). The USA as survivor argument still applies, but might this mean equities aren't a terrible deal right now? There are shades of the disaster Dow 36,000 in my argument, but only a tad.

Eccdogg said...

I always hear folks say how tough shorting is, but is it really that tough if you have some training in options trading.

One really simple way is to buy ATM puts and sell ATM calls. Options markets are pretty liquid for most S&P 500 stocks and buying and selling options cancels out the gamma/vega risk of just buying options. By put-call parity the position above is identical to a forward on the S&P 500.

Anonymous said...

The point is we are saying the erp on stocks historically is not very relevant as hard to achieve. Comparably clearly shorting is way harder, though not hard now.

Anonymous said...

"By put-call parity the position above is identical to a forward on the S&P 500."

But for hard-to-borrows, the forward price won't be the same as the carry-adjusted spot price.

$9,000,000,000 Write Off said...

I've been reading here for a good bit. The jargon often exceeds my grasp and the strategies exceed my wealth. But I've attempted to find a low-volatility equity fund for an individual investor to replace an old, unattended roll-over IRA (~$50,000) full of oddball stocks & funds.

Using this screener I selected equity funds with Morningstar Risk = Below Average & Low, Morningstar Overall Rating= Above Average, and, of course, 0 expense ratio/no Front Load.

That ghetto jalopy delivered a bunch of Vanguard Funds, especially their "Retirement Target" funds:
VFIFX (Morningstar Risk = Low)
VASIX (Below Average)
VTOVX (Below Average)
VTTHX (Below Average)
VTTVX (Below Average)
VTXVX (Below Average)
VFORX (Below Average)
FFFBX (Fidelity)(Low)
AONIX (American Century)(Low)

I'll probably mess around with Google Finance a bit while my sales settle and then buy one or more of the Vanguard funds, targeting retirement at 70 & 80 years of age.

I'm not sure a low volatility equity fund make sense for a tax deferred account (compared to income/ dividend investments), but I have no idea how to even figure that out, and the project is fun, so I ignored that concern.