Equity Premium Puzzle

For background, this is the Wikipedia entry for Equity Premium Puzzle.

"The equity premium puzzle refers to the phenomenon that observed returns on stocks over the past century are higher, by approximately 6%, than returns on government bonds. Economists expect arbitrage opportunities would reduce the difference in returns on these two investment opportunities to reflect the risk premium investors demand to invest in relatively more risky stocks. The puzzle arises as the observed difference in returns implies an implausibly high level of risk aversion..."

While I find the entire theory dubious at best, I have neither the interest nor the expertise to address that question here.

What seems to me undeniable is that historical data must depend on historical conditions. Thus a theory that makes a prediction about, and an analysis of, the relationship between historical stock and bond returns without any qualification or description of the political, economic and demographic conditions under which the historical data being considered was obtained would seem to raise a red flag. In any case, economics is not physics or mechanics, and is not generally subject to controlled experiments.

The question is whether there is some set of historical political, economic and demographic conditions over the last century which would have likely led to a substantial premium for stock returns over those for government bonds.

I believe that the answer is probably yes, and that it involves a fundamental difference in the characteristics of stocks and bonds.

Making no attempt to be exhaustive, the history of the US over the last century includes an increase in population, an overall increase in the prosperity of the population, and as a result, a general increase in the surplus of income and wealth over a merely life-sustaining level.

This increased surplus has inevitably led to an increase in the demand for investment vehicles of all kinds, specifically including stocks and government bonds.

In addition, the historical reality of an inflationist Federal Reserve has eventually made cash an obviously poor long term savings vehicle, leading to an additional shift towards investment in stocks and bonds.

Now the question is whether this increase in demand for stocks and bonds affects their reported returns differentially.


For a government bond, an increased demand will reduce the interest rate that the government must offer as it auctions off new bonds. While this reduction in interest rate will boost the market value of previously existing bonds in the secondary markets as their future return to maturity must be forced down to match the new bonds, it seems unlikely that the long term reported returns on government bonds will differ substantially from that expected from the interest rates to be earned from new bonds. This is just saying that government bonds primarily provide interest income, and not capital gains, taking the long view over a century.

In addition, it seems highly unlikely that government bonds have even managed to earn their nominal interest rates when adjusted for inflation.

Summarizing for government bonds, the long term real rates of return have likely been substantially reduced by century-long trends of both increased demand and increased inflation.


In contrast to bonds, an increase in demand for stocks results in an increase in their price and return. Also in contrast to bonds, the immediate response to an increase in demand does flow through to the reported results as a capital gain, whether realized or not. Even though the immediate increase in value may reduce the future returns if valuations matter to someone, as would likely happen with a forcing of Social Security payroll taxes into the stock market, the increased returns on existing holdings due to an increase in demand for stocks will be reported in the historical returns for stocks.

In addition, the Federal Reserve does not just inflate the money supply and inflate prices, but it also suppresses effective interest rates below their natural levels. This means that whatever future streams of cash flows are used to value a stock results in a higher present value after discounting. This also shows up in the reported results for the real rates of returns of stocks.

Summarizing for stocks, the long term real rates of return have likely been substantially increased by century-long trends of both increased demand and increased inflation, and suppressed interest rates.

Overall Summary

It seems overly simplistic to only consider risk when comparing the long term real returns of stocks and government bonds. It seems likely that the fact that stocks and government bonds can be expected to respond differently to known historical political, economic and demographic trends is at least partially responsible for the differential historical real returns reported.


It wouldn't take a particularly alert reader to note that I have said nothing about supply. I have no numerical evidence for either demand or supply. I am making the assumption that the net historical trends are for an increase in demand relative to supply. In a small way a hand-waving argument would say that increases in the supply of government bonds are often correlated with higher levels of inflation, leading to reduced real returns even if supply temporarily offsets demand increases for government bonds. For stocks, the reported returns are primarily those for large cap stocks. Their supply was relatively under control, especially for the pre-stock-option-era.

Share this

Seniority of the instruments

Seniority of the instruments counts towards risk, as well . Bond holders will get sumpthin' stock holders will get nada.what about options? 90% expire worthless. I think a part of it is the more 'cowboy' types . type A, if you will, make stocks more volatile, while conservatives tends to run the bond market. "Sexy and Dangerous' has gotta count for at least a percent or two, ?:grin: good question.
BTW the long term data user assumes the variances you spoke of wash out.

Don, The most basic


The most basic difference with these two types of securities is that one is valued purely quantitatively and the other is valued subjectively and quantitatively.

You can determine the value of any government bond on any day by knowing these things: maturity, coupon, the rate at which you receive payments (annual, semi annual, etc.), face amount and current interest rates. Since all of these are fixed except current interest rates, that is the only thing you need to focus on for current and future valuations. You can do this with just about any calculator.

The cash flows of equities are not fixed and require you to make many subjective determinations. Will the new marketing campaign work?, how will a competitor's product do?, will the new CIO use technology as a weapon or just spend money?, how will XYZ law affect them?, etc. Then, of course, you need to be able to understand accounting to determine an earnings estimate (assuming that the company’s financials are honest).

Even after you have come up with an earnings estimate, then you need to apply a subjective P/E ratio to come up with a market price.

This is a different take to the original question, but I think that there is a "difficulty of valuation premium" built in to the risk premium.