Wednesday, December 15, 2010

Risk Premium Doesn't Make Sense

The idea that risk begets return comes from the idea that people don't like risk. As risk is avoidable, by say putting one's money in high quality bonds, investing in risky assets must generate a positive expected return to compensate people for such risk taking. This is the risk premium. Yet, we see again and again that risk does not beget return (see here). Merely taking risk is not compensated.

I was reading a book The Decline and Fall of the British Aristocracy, by David Cannadine, which covers 1870 through 1930. They highlight that this was the strongest aristrocracy in Europe due to primogeniture and Britain's exceptional wealth, and so the combination of power, status and wealth was unequaled. In 1880 only 10,000 people owned 66% of the land. Yet land prices fell after 1880 due to a collapse in agricultural prices, the rising industrial sectors marginalized the rural areas that historically gave the nobility so much of their wealth, and new laws turned power irrevocably from nobles to number (the Third Reform Act of 1884).

Classical liberal Richard Cobden wrote that 'the battle-plain is the harvest-field of the aristocracy, watered by the blood of the people', meaning, the aristocracy prior to 1900 generated their legitimacy via their willingness to lead groups into battle (often needlessly). As the British aristocracy declined from 1880 to 1910 they thought that World War 1 would re-establish them as the top of their country. Many were eager to fight, mainly afraid the fighting would be over before they could prove themselves in battle. Yet, while they proportionately lost more of their sons than those of other classes in WW1, the decline of the aristocracy continued unabated after the Great War. Indeed, the loss of life sharply curtailed the supply of domestic servants, and inheritance taxes went from nothing to 60% by 1939. The lower classes felt no sense of gratitude towards the aristocracy, having lost enough themselves. Battlefield courage is admirable, but it is not sufficiently rare to generate privileged status; the lower classes were not party to such an exchange.

In retrospect, the aristocracy arrogated power via myth, tradition, and brute force, rationalized via their exception valor. The problem with their self-serving story is that many would accept a probability of death for such success, so this 'courage premium' was not a real equilibrium result, as WW1 showed. Courage was a necessary, not sufficient, condition for acquiring power. Similarly, risk taking is a necessary condition to achieving riches, not sufficient. In standard theory, you take risk, and on average you will get rich quicker than others. That's all you need, a willingness to experience the randomness.

Unfortunate jobs like cleaning sewers are generally not highly paid even though they are rather disgusting; people get used to a lot of things. In a similar way, risk taking, if all it takes is a willingness to expose oneself to some stochastic shocks, is too easy. As Dan Pink notes, researchers have documented that monetary incentives work pretty easily for straightforward mechanical tasks, like keeping one's band in hot water for extended periods of time. Monetary incentives do not work for tasks taking some kind of creativity. In a similar way, if mere willingness to undergo stochastic shocks generate higher expected returns, it would be too easy. In real life, it's much more complicated, as intelligence is very important. You don't get paid--statistically--for merely for taking risk, however defined.

If one can expect a 5% annual return to investing in equities over bonds, then over 30 years this means quadrupling one's money relative to those who are risk averse. If that is the trade-off, I doubt anyone would choose bonds. It's a profound misunderstanding, a signature mistaken principle in finance.


Brent Buckner said...

I think you're overstating the principle. I'm a much bigger fan of your empirical analyses elsewhere.

You write: "if mere willingness to undergo stochastic shocks generate higher expected returns, it would be too easy."

Substituting "defer gratification" for "undergo stochastic shocks" shows how a simple characteristic of willingness could reasonably lead to higher expected returns (see the term structure of real return bonds).

You write: "If one can expect a 5% annual return to investing in equities over bonds, then over 30 years this means quadrupling one's money relative to those who are risk averse. If that is the trade-off, I doubt anyone would choose bonds." Ah, but what about, say, 2%? Just because 5% strikes you as too big doesn't mean 0% must in principle be more reasonable.

Anonymous said...

Brent is right, the discussion of the reasonable level of the risk premium is different from whether it logically should (or does, themselves separate questions) exist.

Of course we don't care about stochastic shocks that go away soon leaving you with a compound 5% return at the end. The problem is we don't know they go away soon. If shocks aren't mean reverting, and the risk premium isn't big enough, then shocks can have effects, potentially large ones, on our terminal wealth (terminal being whatever point we chose to care).

Eric's original post might be an effective argument against a 5% ex ante risk premium (I think the academics with their equity premium puzzle have said effectively the same thing for a long time - that this seems like too good a deal). But, if you go all the way to no risk premium, you're at the idiocy of Dow 36,000.

Eric Falkenstein said...

Well, I want to state for the record I'm anti-idiocy. At least, in principle. I think the expected equity premium for the average investor is zero (after taxes, transaction costs--spread, market impact, commission--,adverse market timing, survivorship problems, geometric averaging).

Brent: investing in equities has a carrot deferring gratification does not have, which is sufficient for it to be a dominated investment strategy and still exist (see my crappy stocks post)

Anonymous said...

I'm anonymous from above.

To clarify, the idiocy of Dow 36,000 was saying that there was zero ERP but that you should still load up on equities as zero ERP is perfectly reasonable.

Eric's point, that perhaps there is a zero ERP because people make errors, is completely different.

Anonymous said...

Was the risk premium for 4% TIPS (a few years ago) also zero? Or 2% TIPS today? Or blue chip equities with a 3% dividend yield?

BTW, taxes don't decrease the ERP estimate, they increase it, because stocks are a tax favored investment.

Eric Falkenstein said...

taxes: See Gannon and Blum. They look at historical tax rates on equities with assumptions about capital gains vs. munis. Like transaction costs, data today is not applicable to the sample used to generate the 5%ish figure bandied about.

Anonymous said...

For the same ERP, the after-tax ERP is lower than the pre-tax ERP. That is all Eric is saying (I think). That taxes are less a problem for equities than say bonds is a different issue.