The Future of the Equity Premium: SUMMARY: PRELIMINARY AND INCOMPLETE

Konstantin Magin and I are closing in on the first draft of our "Future of the Equity Premium" paper...

SUMMARY: PRELIMINARY AND INCOMPLETE: DRAFT

The Future of the Equity Premium

J. Bradford DeLong and Konstantin Magin

December 2006

Suppose that, at the start of some year since the beginning of the twentieth century, you had taken $1,000,000 that you had invested in bonds and believed you would not want to touch for twenty years, and invested it insteade in a diversified portfolio of equities. (Or suppose you had been able to borrow $1,000,000 at the long-term government bond rate). And suppose you had then let both legs of that investment ride for twenty years. What would have been the results in dollars (adjusted for inflation) twenty years later?

The figure above tells you the answer, for each year from 1901, the start of the twentieth century, until 1986--twenty years ago.

In two years you lose: $50,000 if you invest in 1913, and $150,000 if you invest in 1929. In both years, you would have done better to invest in bonds. In all the other years--all 84 of them, 97.7% of the time--you win.

Often you win big. The average return is $4,356,000: more than four times the value of the initial long equity leg of the portfolio. 93% of the time you are up more than $500,000. 80% of the time you are up more than $1,000,000. 59% of the time you are up more than $2,000,000. 42% of the time you are up more than $4,000,000. 19% of the time you are up more than $8,000,000. And the champion moment to go long stocks and short bonds was 1942, just after Pearl Harbor, when your portfolio held until 1962 would have given you an edge of $15,600,000.

This is, in brief, the equity premium puzzle of Mehra and Prescott (1985). Throughout the twentieth century anybody with some initial wealth devoted to bonds and with a twenty-year or more time horizon had a near money machine at his or her disposal. And not a nickel-and-dime money machine either.

Rietz (1988) and Barro (2005) attribute the apparent equity premium to serious and severe risks of equity ownership that do not show up in the twentieth-century sample. The problem is that it is hard to think of what such risks might be. We have one Great Depression in the sample. Isn't that enough? And when the Day of the Revolution comes and the red flag flies from the towers of the Loop, investors in U.S. government bonds will be in no better shape than investors in corporate equities. At a more subtle level, Weitzman (2006) argues that the equity premium is generated by unknown unknowns--risks that we cannot characterize and that have the potential to have any conceivable effect on our expected utility. In Weitzman's framework, the anomaly is not that stock prices are so low but that they are so high: that the equity premium return is not still bigger than it has been. But once again there is the question of just what these risks are, and why they depress the relative price of stocks rather than bonds.

Once you put the possibility that the ex ante distribution of stock relative to bond returns had a big, important, substantial lower tail that just does not show up in the data ex post, attempts to explain the equity premium return as some kind of rational-expectations equilibrium for investors with reasonable degrees of risk aversion are likely to be in vain. You have to construct a scenario in which investors and markets are such that investors do not turn the crank on a money machine. And that is very, very hard to do.

An enormous variety of institutions and investors ought to have taken advantage of this money machine--sold their bonds and put their money in stocks: young parents of newborns looking forward to their children's college, the middle-aged looking at rapidly-escalating health-care costs, the elderly looking forward to bequeathing some of their wealth to their descendents or to worthy causes, workers with defined-contribution pensions, businesses with defined-benefit pensions, life insurance companies, the Social Security Trust Fund, the reserve accounts of the world's central banks, businesses with reputational capital that do not expect to be blindsided by new technologies, hedge funds, and so on. On the other side of the market, there are all the companies that appear underleveraged: replacing high-priced equity capital with low-priced debt capital would seem to have been as profitable a strategy for a long-lived company as investing in high-return equity rather than low-return debt is for a long-lived investor. Yet they have not done so, or have not done so on a sufficient scale to pick up all the $4,356,000 checks and the occasional $15,600,000 check left on the sidewalk. We can explain why any one particular group might have good reason not to try to take advantage of the equity premium, but it is much harder to come up with the many good reasons why all of the groups of potential investors, and institutions, and organizations passed the $4,356,000 checks by.

It's not that the existence of an equity premium return is a new discovery. In 1923 financial analyst Edgar L. Smith pointed out that diversified investments in American equities had far outperformed bonds of all types. Edgar Smith's (1923) Common Stocks as Long-Term Investments was not the first publication to point out that stocks earned higher returns than bonds on average-—that was well-known. But investments in equities were viewed then as the domain of the risk-loving and the entrepreneurial. Speculators—-either possessing inside information about fundamentals or market microstructure, risk-loving, or naïve—owned stocks. But prudent investors did not: the risk of ruin was seen as too high. What Smith did was to publicize the fact that equity diversification worked: diversified stock portfolios produced higher rates of return without bearing higher systematic risk than bond portfolios, especially once one took inflation risk into proper account. A portfolio invested in one or two individual stocks was overwhelmingly risky: "subject," Smith wrote, "to the temporary hazard of hard times, and [might] be permanently lost as a result of a radical change in the arts or of poor corporate management." But, Smith pointed out, these risks could be diversified away: "effectively eliminated through the application of the same principles which make the writing of fire and life insurance policies profitable." By contrast, portfolio diversification did not work for bonds, which were all "subject to the same hazards" which were "not reduced by increasing the number of different bonds held."

Some economists--Blanchard (1993) is probably the best advocate for this position--see the large premium equity return in the past as a mistake on the part of the market, a mistake that the market should be correcting. They anticipate that the equity premium is smaller in the present than it was in the past and that it will vanish or nearly vanish in the future. Blanchard sees the high equity premiums of the 1950s and the 1970s as a combination of excessive salience of the memory of the Great Crash and the Great Depression in investors' minds, and as the result of simple money illusion a la Modigliani and Cohn (1979).

But we do not see any signs that Ms. Market has moved to eliminate past mistakes. The real return on the 20-year U.S. Treasury inflation-protected bond is 2.1%, while the current annual earnings yield on the S&P composite stock market index is 5.7%. These numbers suggest an expected equity premium today of 3.6%, lower than the 6% premium equity return of Mehra and Prescott (1985), but far, far higher than the value of 0.25% per year of Mehra's (2003) baseline representative-agent model for a coefficient of relative risk aversion of 2. and an equity premium of 3.6% per year is enough to double your relative wealth over twenty years, and quadruple it over forty. Plus you get substantial immunity from long-run inflation risk as well.

The equity premium remains a puzzle. And there is no reason to think that it is a puzzle about to vanish. As Rajnish Mehra (2003) wrote:

The data used to document the equity premium over the past 100 years are as good an economic data set as analysts have, and 100 years is a long series when it comes to economic data. Before the equity premium is dismissed, not only do researchers need to understand the observed phenomena, but they also need a plausible explanation as to why the future is likely to be any different from the past. In the absence of this explanation, and on the basis of what is currently known, I make the following claim: Over the long term, the equity premium is likely to be similar to what it has been in the past and returns to investment in equity will continue to substantially dominate returns to investment in T-bills for investors with a long planning horizon.

Or as Warren Buffett wrote twenty years ago:

[T]he secret has been out for fifty years, ever since Ben Graham and Dave Dodd wrote Security Analysis, yet I have seen no trend toward value investing in the 35 years I have practiced it. There seems to be some perverse human characteristic that likes to make easy things difficult. The academic world, if anything, has actually backed away from the teaching of value investing.... It's likely to continue that way.... [T]hose who read their Graham and Dodd will continue to prosper.

Comments

[Michael L]

Not being an economist, I'm sure this has already been considered. Isn't at least part of the explanation that access to the market is, in general, sold; and that means salesmanship (which also skims off the top) has a lot to do with the choices being made?

[CJColucci]

So we know the equity premium exists, we don't know why it exists, it shows no sign of going away, and the usual explanations for it don't make sense. In the absence of any explanation that makes sense, therefore, the most reasonable conclusion is that it will continue, though we don't know why.
That actually makes sense, I think.

[Barkley Rosser]

[Inflation almost always gave real investments in government bonds a severe haircut. The default was implicit rather than explicit, but it was very real. At least in the cases I have examined, partial compensation of equities is better than bonds honored in inflated currencies.]

Regarding the revolution argument, in fact outside of the US there were many countries in the world during the 20th century, a solid majority in fact, in which there were nationalizations of privately owned firms, usually without full compensation. However, except in those cases where there was indeed a full blown revolution or takeover by outsiders, government debt remained honored. So, there has historically been a substantially greater risk of arbitrary loss of equity returns compared to the bond returns at the global level.

It has been one of my arguments that part of the reason we have seen an apparent permanent increase in price-earnings ratios in the stock market since the early 90s has been the collapse of the socialist bloc and the apparent ending of nationalizations pretty much everywhere, with a general drive to privatize appearing instead. Thus, this risk of nationalization has largely disappeared, although it may be reappearing again in some places such as in Latin America.

[James Killus]

In the late 1990s, I joined a company that had a 401(k) plan, and set mine up as a mix of a stock fund (S&P Index) and a bond fund (high grade corporate). Until 2001, the stock fund was ahead. Then, well, you know. And the stock fund has yet to catch up to the bond fund.

Maybe part of it is that the comparison is only to U.S. Bonds, since corporate bonds have a premium there also. I'd also like to see how the "diversified portfolio of equites" was chosen. And I'd like to see the results using a shorter time filter, like 2 years, 5 years, and 10 years.

[Brad DeLong]

Oh. Over shorter time periods, bonds can beat stocks. And they might beat stocks this time. Still: S&P value-weighted earnings yield of 5.7%, 20-year TIPS yield of 2.1%...

[anne]

The period from February 2000 to February 2005, was exceptional. These 5 years easily marked the biggest gap in favor of the long term bond index over the large cap stock index that I could find from the 1930s on. However such a period is self-limiting. As interest rates decrease dramatically there is in time little more in the way of capital gains that can be expected from long term bonds. So, portfolio managers looking for more than simply interest rate gains should have moved from bonds to stocks in 2004 or before.

[anne]

Remember, the 5 years from February 2000 marked a dramatic bull market in bonds and a feirce bear market in information technology or assorted growth stocks. A wholly singular period since the 1930s.

We are now in the midst of as dramatic an international bull market in stocks as I can find record of.

[anne]

Lawrence Summers has pointed out that the last 5 years mark the most profound period of international economic growth since 1945. No matter the ever bearish bears, international investment markets have reflected just this these 5 years.

Also, notice how well long term investors should have fared these last 10 years through a fierce bear market. The exception always being what I call the Japanese market-of-death.

[anne]

http://www.mscibarra.com/products/indices/us/performance.jsp
http://flagship2.vanguard.com/VGApp/hnw/FundsByName

Vanguard MS Fund Returns
11/30/96 to 11/30/06

Large Cap Index is 7.6
Large Cap Growth Index is 5.2
Large Cap Value Index is 9.7

Mid Cap Index is 12.1
Mid Cap Value Index is 14.8

Small Cap Index is 12.1
Small Cap Value Index is 14.0

Europe Index is 10.7
Europe Value Index is 13.8
Pacific Index is 2.6
Pacific Value Index is 4.5
Emerging Markets Index is 8.8

Energy is 16.7
Health Care is 16.5
Precious Metals is 13.7
REIT Index is 15.6

Long Term Bond Index is 7.7
Intermediate Term Bond Index is 6.4

[Notice the favoring of value indexes through the years including the bear market.]

[Aaron]

How about this?

Not all investors have 20 year horizons. I'm 61 now. In the past I had a very long investment horizon, but I didn't have any money then. (I did have a substantial investment in human capital, which was leveraged to the hilt and which carried lots of idiosyncratic risk. Even so my tiny portfolio was all in stocks. So do believe that the risk premium is too high.)

Now that I've got some money, my investment horizon has gotten shorter, and I notice that leveraged equities are not as consistent a money machine over shorter (and shorter) horizons as it has been over your 20 years. Take a look at the maximum three-year draw-down on your hypothetical portfolio. What's true of me is true of a lot of individuals, pension funds, and life insurance companies. They tend to have the most money to invest when their horizon is shortest.

So the high equity risk premium may only provide a free lunch to investors who don't age: endowments or extremely wealth families, for example. And in fact, the most astute of these tend to enjoy much better investment results than the rest of us. If so the equity risk premium is likely to remain high until Harvard, Yale, the royal families of the Gulf, etc., own an even larger proportion of the world's wealth than they do now.

For the present representative investor, however, Barro's focus on the disaster scenario may be on the right track.

[anne]

Aaron:

Increasingly, I am convinced that age should not be, need not be, a factor in portfolio allocation. Notice the selected mix of 10 year Vanguard returns, gained through a period of the most severe American bear market since the 1930s. Why should I have not been able to allocate a portfolio for myself, my parents and grandparents in 1996 in much the same way as in 2006? I realize I do not think of the age of my parents in investing. Hmmm....

[steve]

According to W. Bernstein, the equity risk premium (ERP) in the 19th century was only 1.5%. (The ERP for foreign markets is also lower than the 5% for 20th century US shares.) So, which number should the investor use to extrapolate through the next 20 years?

Bernstein attributes the very different results of the 19th and 20th centuries to inflation. He claims that in the 19th century inflation was not understood to be a threat to bond returns. At the beginning of the 20th century, bond yields were 5%, which looked pretty attractive, as it was assumed to be a real rather than nominal yield. The inflation brought on by two world wars and the death of the gold standard was disastrous for bond holders. On the other hand, equity returns managed to keep pace with inflation (companies could pass increased costs on to consumers). As a consequence, investors bid up equities and devalued bonds. But, you can't abandon hard currency twice, so perhaps the 21st century will be more like the 19th?

If a rational investor is unsure whether the next 20 years is going to be more like the 20th than 19th centuries, then she might think in terms of an ERP which is somewhere between 1.5 and 5 percent. Given that the investor can't be absolutely sure she won't need liquidity sooner than 20 years in the future, it doesn't seem so crazy to keep part of the portfolio in bonds. (Very bad to need cash when the equity market is going through a down fluctuation.)

Much of this is summarized (with links to Bernstein's work) here:
http://infoproc.blogspot.com/2005/01/risk-premium-and-equity-outperformance.html

My question for Brad: at what value of ERP does there cease to be a "puzzle"?

For example, at ERP = 2-3 percent I can see need for liquidity or behavioral dislike of volatility as easily enough of an explanation. For ERP = 2% stocks only beat bonds in about 60% of 10 year periods! (See figure at link above.)

[Aaron]

Anne:

Short-term volatility in investment returns doesn't feel consistent with a "worry free" retirement. (If your family is very wealthy, then they should be worry free no matter how you invest their/your assets as long as it's not completely crazy.)

But I agree that traditional rules of thumb about the percentage of bonds vs. the investor's age probably generate portfolios that are too conservative. So age should be a factor in asset allocation, but not as much of a factor as a lot of people think.

That said, it seems possible to me that the correlation of wealth with age (or the present value of liabilities) may help explain Brad's conundrum.

[Jonah Lehrer]

What about the loss aversion hypothesis put forth by Benartzi and Thaler?

[anne]

Aaron:

I understand, and realize I simply do not think in terms of portfolios and age but I need to think further and your argument is important.

Notice with a bond portfolio; I find advisors arguing for short term, short duration bond funds, for older investors. This seems nutty. Given a portfolio of fairly constant duration as Vanguard offers, I know my parents will always earn more holding the long term bond index than the short term over any 10 year period. What difference does short term price fluctuation makes with a bond portfolio unless the point is market timing which is another matter?

Always go long term index for income. But, not for analysts. Well, they either do not understand or are playing portfolio games.

[BillS]

Anne, the argument for short term bonds is that they have negative correlation with equities and therefore provide better diversification. Longer term bonds tend to correlate more strongly with equities. They tend to go down at the very same time that equities are under pressure. If you are going to take on more equity-like risk, it is more tax efficient to do that with equities rather than long term bonds because of the lower tax rates.

An alternative way of looking at it is that short term bonds have less risk and therefore you need less of them to diversify your portfolio. To get greater returns you are better off buying more equities than buying longer term bonds, again for tax reasons.

Even for non-taxable accounts there are reasons to prefer shorter term bonds. Long term bonds can be quite risky. If you have a 30-year bond and interest rates double, you lose half the value of your bonds. If you look at a normal, non-inverted yield curve, the steepest part of the curve is in the one to five year range. Beyond five years you get little additional yield for taking on substantially more risk. You would be better off buying equities. Long term bonds are generally more appropriate for institutions like university endowments and pension funds that have clearly defined long term commitments requiring steady income.

[anne]

Oh dear, oh dear, the most profound relative long term bond rally since the 1930s just took place precisely when stocks had peaked and were beginning the fiercest bear market since the 1930s.

Oh dear, oh dear, I am expressly adressing holding a relatively constant duration bond fund such as Vanguard offers.

Oh dear, oh dear, the beauty of a constant duration bond fund is that no matter what interest rates do through a duration period, the investor will make the beginning yield. So, as John Bogle taught, if you like the yield buy the fund and hold and hold. Well, long term yields are higher than short term.

For them what likes to hold bonds, Vanguard long term funds are the cat's pajamas :)

Duration for Vanguard bond index funds run about 2, 5, and 10 years for short, intermediate, and long term. Also, there is the inflation protected protfolio and there are assorted tax free bond funds.

[anne]

Duration Example: a 1 percentage point increase in interest rates will lower the price of a Vanguard short term bond fund by about 2%. However the yield of the fund will gradually increase by 1%. So, within 2 years an investor will have the earned the starting yield on the fund.

Unless the idea is to play for capital gains, which may be preferred, I could care less what interest rates do. So, for bill paying and the like an investor can simply simply hold a long term bond fund and write checks against the yield of the fund and no matter the price fluctuations.

That has been precisely what has been best for investors who always wish to hold bonds.

Vanguard long term investment grade bond fund has returned 8.93% a year since inception in July 1973.

[anne]

Duration, from what I sort of remember was understood in the 1970s and became a fine tool for controlling portfolio risk with no return sacrifice. Sure if an investor is worried about price changes than I can understand limiting duration, but as John Bogle would say a percent of return is more valuable than a percent of risk and controlling duration control risk wonderfully.

Also, know what is readily knowable. The idea of buying individual bonds never occurs to me. Vanguard can do that far better, readily controlling for risk on individual bonds.

[anne]

Bill, thank you, the comments help me think.

The problem for some investors is wishing more income than can be gained from dividends unless a portfolio is quite large. These are not dividend paying times. So, I easily understand holding a bond index or inflation protected bond fund in a taxable account. Also, there are the various tax free funds.

The Vanguard High-Yield tax free bond fund has returned 7.35% a year since 1978.

[Stephen Gordon]

There's also the state-dependent preferences explanation for the equity premium.
http://worthwhile.typepad.com/worthwhile_canadian_initi/2006/12/welfare_costs_o.html

[anne]

Oh, by the way, Vanguard High-Yield tax free bond fund is a fine investment grade fund. The fund was run by Citigroup briefly in the beginning before controlling for duration, and was scary. That changed at once when Vanguard took control.

[Stephen Gordon]

Poop. Can't get the link to work.

[Barkley Rosser]

Brad.

Point well taken regarding inflation and bonds.

OTOH, there are plenty of near 20 year intervals where equity performance was awful. My fave is anyone buying the Dow in mid-1966 when it first passed 1,000. Was more than 16 years before it got back there in just nominal terms.

[Yes. You need to have a *long* horizon before mean reversion in stock prices and the law of large numbers work in your favor.]

[anne]

No, no, no. Such an example is meaningless. There were ample dividends on stocks from 1996 and through the following 16 years, averaging well above 4%. Also, an investor does not suddenly set down baskets of money on the day the market index peaks having never invested before. A leaning to value around 1966 and after and an investor did just fine, and an investor had to lean because there were no index funds till September 1976.

[anne]

There is however a real scary story, for those wishing scary stories, and I do not understand this. The Nikkei index was at 39,000 in December 1989 and is in the 12,000s to this day with minimal dividends. Now, the climb to 39,000 after the 1985 Plaza Accord increased the value of the Yen was steep. The Yen is not much changed against the dollar since the bear market began in 1990. But, here is a market-of-death I do not understand. Japanese stocks always always always seem pricey.

[Elizabeth]

This is my thinking:

1) Part of historical premium was equities were too cheap- as noted current yields and TIPs yield is at 3.7%.

2) Part of the 3.7% is illusion - current profit rates way above historical norms and likely to eventually succumb to competitive pressures. Let's say profits are 20% too high - that takes yield down to approx 4.5%.
Now the equity premium is 2.5% - still high.

3) The rest of premium exists because bonds are way too expensive - a result of many institutional factors. Life insurance companies for instance face huge capital penalties for holding equities. Pension plans and other large institutional investors have historically held bonds and they don't understand this makes no sense. At the moment you have the central governments owning massive quantities.

4) The equity premium persists because of barriers to doing the long- short trade. The only people out there who have a long enough time frame who can short treasuries and who can withstand the ups and downs of the trades over potentially a decade are wealthy individual investors. Insurance companies - no. Pension plans - no. Even hedge funds cannot do a trade with a 20 year time horizon. And at this point, most feel they already have enough risk in portfolios without adding this.

[techy789]

"unknown unknowns"? - you sound like Donald Rumsfeld!
One point: when you characterize the typical outcome of stock investment, you say "all 84 of them, 97.7% of the time" the stock investors come out ahead. But these numbers are considering overlapping periods of investment, not distinct periods (i.e., the return on stocks from 1914 to 2006 will look very much like the return on stocks from 1915 to 2006 - but this correlation should not be used as evidence of the risk premium).

[jkb]

I'm surpised that no one has posted the joke,"assume a ladder." As others have noted most investors-even pension funds with annual contribution liabilities and benefits to pay-have a need for stability and current income.Dollar cost averaging is great on the way in,but can kill you during the distribution phase.Taxes also have a substantial effect on the re-investment rate. The more interesting question is the effect of computers on all this.For most of this history it was easy to maintain a constant reinvested pool of hundreds of different stocks. Just assume a ladder. It is only with modern computers and their ability to easily replicate a constant index that large numbers of investors have had the functional ability to do this. I suspect it is this ability that lowers the dividend rate,raises the typical PE ratio and breeds the desire for hedge funds.What Bunkie with is slide rule and seat on the exchange used to do at JP Morgan is now open to anyone with a pc. Hence Bunkie and the gang at the House of Morgan want more.Surprise!

[jkb]

An error in my previous post. Should be raises - not lowers - the PE ratio. Sorry!

[albrt]

I don't think you can solve this mystery by theorizing based on free investor choice. Both supply and demand depend on institutional factors. Business and government institutions simply will not tolerate higher interest rates over the long run because they do not believe businesses can produce returns satisfactory to the proprietors or sufficient to employ available labor. Although institutions cannot literally control interest rates under current conditions, they have many mechanisms at their disposal for knocking interest rates down - from direct central bank purchases to insurance company regulations to recommending that elderly small investors hold more bonds than they really need.

[shah8]

Ah, but this is going to be fun...Long post, quick conclusions at the bottom.

Professor Delong, if you enjoyed Freakonomics, you'll probably enjoy Stumbling on Happiness, by Daniel Gilbert. It's hillarious how much Dr. Gilbert *avoids* talking about more serious topics. If you read this after reading a book about racism/jim crow/classism, you get banged around the head with the implications of what the author is saying, while he stays with the jokes about infidelities...

Okay, now, a critique of your position.
1) The first objection, as other people above have remarked, is that you plotted graphs of the *profits* by stocks and bonds. However, there does not seem to be plots of *opportunities* to buy and *opportunity costs* of buying.

I mean, forget about people *having*different*needs*and*how*bonds*are*good*for*old*people, have you seriously thought about the impact of Keyenesian government policies? I mean, that big peak from the late 30s to the late 50s gotta be pretty indicative of the crowding out of private investment by large public debt and debt spending. Thus, isn't that equity premium *largly derived from the the *taxes* on *business opportunities*? Just as there was rationing at the consumer level for things like beef, tea, sugar, there were systems of rationing that occured in the business sphere, from nationalization/public works/high profit taxes.

Now, look towards the sixties. JFK lowered taxes. Many of the big public works programs are winding down. The big iron triangle politics of defense appropriations (with NASA's friendly face) starts up. This was an era where there were huge engineering firms thinking up all kinds of crazy stuff at IBM, GM, Bell Labs, Xerox, et al. For all those young 'uns who think today is the information age, well...no Yahoo's, or Microsofts, or anything like that, but people were just as excited back then at all the interesting things being added to the human experience during the sixties, not just the free love, you know...However, understand this, this was about value being added to a present industrial base!

Now look towards the seventies. Look at all that nasty inflation generated by the gun's in butter policy mix. More importantly, look at the consequences of the East Texas oil peak in 1972. With that little OPEC response to the Yom Kippur War, the US, especially the policy elite, got the shock of their lives. What they could do to Japan in 1941, Saudi Arabia could do to them, and put them in a straight-jacket in terms of power projection. What, did you think Nixon *enjoyed* visiting China and making polite noises to Mao? Damn right he didn't, but he had to, given the Soviet's oil reserves, which THEY could use as a weapon, to force more equitable access to materials. US simply *had* to take advantage of the Sino-Soviet split.

So leading to this next string of points: Control of the *excess oil* leads to the control of other crucial resources, by making power projection easier, and by giving preferential access to that excess oil to resource countries and any competing first world countries that might make a stink. Without Texas as a geopolitical weapon, the US had less leverage over the other oil producing countries to keep oil prices low, thereby keeping it cheap to replace labor with capital. Thus, a huge general commodities boom started. And one thing to note about commodities and extractive industries...They are *easy* to nationalize. So a wave of nationalizations occured. So many of the best, most profitable business enterprises went under the government umbrella. Now, how does anyone invest in the best stuff? By buying developing country bonds! So, in the US, not only are basic commodities are increasing, but investment supply is drying up, and that shuts down the value added economy that was present in the 60s. That most definitly has a negative impact on equities markets, and the debt markets got oversubscribed.
Also note that this where alot of the stagnation of wages started. It's not so cheap to replace labor with capital anymores, so the best way to be competitive is to keep labor costs down with a vengeance. This was the primary avenue that companies increased the attractiveness of their stocks after the doldrums of the mid-1970s. Eventually, though, oil became cheap again in the 80s with the over expansion in the third world and the openings of northern oilfields.

And going through these patterns, one can see this truth being *very*, **very** present! Equities is about the value of the company's *GESTALT*. Not about the company's leadership, nor about the underlying assets. It's about whether being organized is better than not being organized. Whenever companies are having a great deal of trouble finding demand for their products and services, or when they are having trouble feeding the organization with the raw materials of existence, then the underlying assets of the company is worth more than company itself, and the rational thing to do, at the time, is to invest in the major competing source of gestalt. That would be the government. And Bonds. Remember, it's the gestalt that matters for attractiveness, not necessarily profit or cashflow. People always want to be sure they get paid back at the drop of a dime.

2)Now, if it's *that* obvious that any diversified stock market strategy would work to ameliorate the risk spread inherent between stocks and bonds, why don't more people do so? Say, when the government is encouraging people to buy war bonds, say, HELL NO! I wanna buy stocks! If everyone is buying bonds, then STOCKS gotta be CHEAP! And while one can underestimate the social pressures that can be brought onto individuals to engage in economically unprofitable action (check out the particulars of what happened economically during segregation, it's fascinating), one should not also underestimate how stubborn people can be in pursueing their best advantage.

So we have to turn to sociology and psychology. Now, the very first thing that should be understood, is that the stock market is a casino, or at least have most of the properties of a casino, including the phenomenon of the House always winning. So, from the start, hell yeah! Buy index funds until your purse begs for MERCY! Automatically diversified, no brain, reasonably little worry of losing most of your investment, and few people nibbling away at your principle through service fees. What a wonderful thing, 'tis must be magic!

Okay, let's get real here. The main reason index funds are so great is because it's a relatively unknown option, compared to what it gets you. Not only that, it is about against everyone in the industry's interest to tell you about them. No money in fees that way. Okay, plenty of people know about index funds, but even many of the informed still like to feel in *control* of their investment. They have more fun that way, ever since Adam ignored the fact that he was having sex with a hot woman named Lilith, in favor of being petulant about having control about who gets to be on top. And for the same reason control was important to Adam, it is generally important for people to feel that they have control because it makes them feel important and avoid anxieties, even if they know it could mess them up. It's not as if people care that some random lottery ticket number has exactly the same odds as anything they could generate, even if they *know* that. What they want is a sense of affirmation that the world is responsive to their needs (even if it's in their daydreams). And that's where your friendly neighborhood stock broker comes in! (and somehow, the inherent need to feel that the world is reasonable and fair continues)

Okay, let's wave that wand...*Poof* I just fiated that everyone knows the wonders of diversification and index funds, and believes, oh yes BELIEVES in doing the right thing, and by God, let us invest in these miraculous tools of wealth! What do you think will happen? Yeah, the risk premium will decline. However, think about what else will happen...Wall Street as a whole would be moaning about horrible profits. Unfair losses! This must be remedied! So one of two things will happen: The costs of buying into an index fund will increase (well, you know the saying, supply and demand--never mind that there is an infinite supply of service, thanks to the miracle of modern computing) On the other hand, create a new investment market, with cool doodad lightings and cool pamphlets about how investing in this cool new category can make you cooler than that prissy Jones character. Same shit, different day.

Conclusions:
1) High spreads between stocks and bonds is a strong indication that there is an impediment to investing in stocks, whether by taxes, opportunity, or social issues. In other cases, it is an indication that assets (inside or outside of business) are worth more the union of disparate parts, for whatever reason, and continued investment faces considerable opportunity costs.

2) People are willfully stupid, and that the market will do what it has to do to keep it that way, because the profitability of the market for the serious/and/or/long term folks is dependent on the importation of fresh, squealing, pigs for the slaughter.

[anonymous]

Prof. DeLong,

Suppose that, at the start of some year since the beginning of the twentieth century, you had taken $1,000,000 that you had invested in bonds and believed you would not want to touch for twenty years, and invested it instead in a diversified portfolio of equities. (Or suppose you had been able to borrow $1,000,000 at the long-term government bond rate). And suppose you had then let both legs of that investment ride for twenty years. What would have been the results in dollars (adjusted for inflation) twenty years later?

This paragraph and the following graph are a bit unclear. When you say "both legs", are we to assume that you keep $1 million in stocks and $1 million in bonds? Or is it $1 million in stocks and short $1 million in bonds?

Also, I think, based on the subsequent paragraphs, that the plot is of the profit one would make on 1 million, but if the question is the risk premium, shouldn't there be a plot of how much more one would make by owning $1 million in stocks for 20 years than if one owned $1 million in bonds for the same time period?

[BillS]

Hi, Anne

Regarding long term bonds, you're right that the last few years have been a good market for bonds. But consider the opposite case. Assume the hoped for soft landing in the trade deficit doesn't work out. Suddenly short term rates are at 10% and your long bond fund is stuck at 5%. You can't sell without losing half your principle. Meanwhile 8% inflation is eating away at your 5% yield and in a few years you're eating Alpo. The point is that a long bond fund is volatile and risky, just like stocks.

The thing you should ask yourself is why you are holding bonds. What purpose do they serve in your portfolio? If you are looking for high returns and are willing to take risk, which must be the case since you are holding a long bond fund, they why don't you instead hold the Vanguard Total Stock Market which has a higher expected return than long bonds.

In Modern Portfolio Theory (MPT) the general idea is that you load up on stocks in various asset classes to provide a higher expected return and balance that with just enough short bonds to allow you to sleep through the volatile roller coaster of the stock market. Short bonds preserve principle and respond fairly well to inflation. You can get more inflation protection by dividing the bond portion of your portfolio between a short bond fund and a TIPs fund.

Since you sound like a Vanguard devotee you might find it interesting to hang out at the Vanguard Diehards forum at Morningstar. A few weeks lurking there is well worth the time. www.diehards.com

[Ed D]

What would happen if the Chinese wised-up and began dumping all those USA bonds they've been accumulating and "reinvested" the proceeds into our "can't lose" stock market?

[Raph]

Dear all

I look forward to reading your paper. This puzzle puzzles me. I just have a couple of thoughts:

(1) Most investors can't leverage at the 20y US bond yield.

[Every investor and institution that owns long bonds can. And they are the people with the money.]

Financing costs will be much higher, most likely making the trade less interesting. Downpayments, margin calls etc also have to be taken into account.

(2) Looking at past US data the ERP may well look too high, but I somehow feel there is a substantial survival bias, the US economy and its corporate sectors having been such successes. Who knows what the future holds however.

(3) What if the government was to issue debt to invest it in the stock market? Wouldn't that be a form of nationalization the consequence of which would require an ERP?

[Robert Waldmann]

An aside, there seems to be a word missing. I think it is "aside"

"Once you put the possibility that the ex ante distribution of stock relative to bond returns had a big , important, substantial lower tail that just does not show up in the data ex post [?????], attempts"

[anne]

Oh dear, for them what wants to be even more conservative the Vanguard inflation protected bond fund is the ultimate in conservatism, though even here the duration will rightly be between 6 and 7 years or between an intermediate and long term term bond fund.

There is really no danger of China selling significant amounts of American bonds, nor is there a realistic danger of short or long term bond rates going from 5% to 10%. However, were interest rates to double the Vanguard long term bond index would experience a price decline that would be compensated in about 8 years with additional yield by which time interest rates would be back to 5% anyway and there would have been no long term problem.

No need to hold any bond funds, though they are useful at the least for a checking reserve and some investors will want to have a steady cash income and there is real diversification in longer duration bond funds. But, worry about a fairly constant duration Vanguard bond fund is simply needless.

[vtcodger]

****Rietz (1988) and Barro (2005) attribute the apparent equity premium to serious and severe risks of equity ownership that do not show up in the twentieth-century sample. The problem is that it is hard to think of what such risks might be.***

Two experiments that might shed some light on that.

1. Draw a similar chart for the Japanese markets (priced in yen, not dollars). And note that barring a sudden explosion in Japanese equity prices, the next four years are likely to be really dismal as they will reflect stocks purchased at the peak of the Japanese bubble that ended in 1989. The problem here seems to be that with all the tools of modern economics to hand, Japan has been remarkably slow to recover from it's stock market bust two decades ago. Couldn't possibly happen in the US of course.

2. Redo the charts using inflation adjusted bonds instead of real bonds. I'm ill equipped to do that, but I expect that the chart for equity purchases in the 1960s will not look very bright. If so, would that imply that part of the apparent "equity premium" comes from an inadequate risk premium for fixed rate bonds. i.e. the equity premium should be measured relative to an optimal low risk investment, not an investment whose risks are underestimated?

===

And finally, someone mentioned 4% dividends included in the figures. Well and good. But if memory serves me correctly, the current S&P500 dividend payout rate is under 2%. Does that imply that current US equity prices are inflated by a factor of two?

[Barry]

Ralph: "(2) Looking at past US data the ERP may well look too high, but I somehow feel there is a substantial survival bias, the US economy and its corporate sectors having been such successes. Who knows what the future holds however."

Survival bias - Brad, don't the index prices for stocks hide the fact that some stocks dropped out of, and others 'rose up' into the index?

[Peter vM]

The ERP has historically been much lower outside the USA - in Denmark and a few other European countries it has been close to zero, postwar.

In Japan, if you start investing 100 years ago you lose all your wealth in 1941-5

[Once again: the equity premium is a premium return on stocks relative to bonds. You would have lost all your bond wealth in Japan due to inflation as well.]

and then lose 75% of it in 1989-2003.

The US ERP may reflect survivor bias - equities across the world have earned nowhere near the kind of premium the US has, and -100% losses (war, expropriation) are not uncommon.

[Once again: the equity premium is a premium return on stocks relative to bonds. Survivor bias only works if there are some non-survivors where bonds do well and stocks to badly.]

[anne]

The Japanese stock market is different in relative valuation in a way I have not understood. Japanese stocks always are over-valued. they were over-valued in 1985 and 1989 and 1995 and 1999. This is different than any other developed market, and decidedly different than the American market in which there was fine value about even in 1999-2000.

Notice however that we are passing through a profound broad and deep international bull market in stocks that is especially hopeful for having made sense in terms of value.

[anne]

Diversity can be real and really simple....

http://flagship2.vanguard.com/VGApp/hnw/FundsByName

Vanguard Fund Returns
12/31/05 to 12/27/06

S&P Index is 16.3
Large Cap Growth Index is 9.6
Large Cap Value Index is 22.9

Mid Cap Index is 14.4

Small Cap Index is 16.9
Small Cap Value Index is 20.3

Europe Index is 33.8
Pacific Index is 12.4
Emerging Markets Index is 29.1

Energy is 20.2
Health Care is 11.3
Precious Metals is 34.1
REIT Index is 34.3

Long Term Bond Index is 3.0
Intermediate Term Bond Index is 4.2

[SPXsurvivor]

[There is no survivorship bias in the long time series S&P or Cowles Commission indices.]

Would someone please address the issue of survivor bias...average lifetime of an SPX component is down to 10-15 years now according to Mike Mauboussin, Legg Mason strategist. The index selects stronger companies and boots weaker ones, how do I do this as a passive 20 year holder? Has anyone really cleaned up the equity returns data to account for survivor bias?

[anne]

The Vanguard-Morgan Stanley indexes are superior to the S&P indexes for broader base and less turnover, but as for survivor bias in indexing there is no adverse significance though mutual fund managers and advisers will claim otherwise to forever justify higher fees. Survivor bias for whole funds is significant for matural and hedge fund companies however.

[Aaron]

Anne, et. al.:

On the question of short- vs. long-term bonds. After a lot of work and thought on the subject, I've concluded that it doesn't matter very much. In the context of equal-volatility mixes of bonds and stocks, the difference between the optimal bond portfolio duration in the worst is measured in handfuls of basis points.

So my recommendation is to think about something else.

Aaron

[anne]

No; you are entirely wrong, the difference in return over long periods between short and long term bond indexes is significant and substantial indeed. Invest where we will, but know history and know it right.

Match the Vanguard long term investment grade bond fund returns of 8.93% since 1973, with any short duration fund. Not even close. Please.

[anne]

There is the reason cost cost cost is so important as Berkshire Hathaway teaches. A few basis points less cost here and there and add time to the mix and lean to value and "wow."

[steve]

Survivor bias is a non-issue. How do you think Vanguard manages to replicate the S&P500?

As the market cap of a company falls below 500th place, it will cross the market cap of the 501st largest company. At that point, sell all your shares in the falling company and use the proceeds to buy shares in number 501 (soon to be 500). This allows you to be fully in equities at all times, but only own the top 500. Replace 500 by some other number N if you want to own more or less of the market.

It is not difficult (esp. today) to replicate the performance of an entire equity market.

Also, regarding the bond portfolio duration question, I think it's not the main issue.

The real question is why investors are willing to give up significant expected return to trade historical "equity-like" behavior for "bond-like" behavior. "equity-like" means more volatile, with greater prob. of big losses. (Giving up some return is reasonable, but how much is justified "rationally"?)

However, it's not clear it's a well-posed question since, e.g., in comparing the 19th to 20th centuries in the US, or US vs. abroad, the relative behavior of bonds vs equities can vary tremendously.

If you could show that investors are strongly confident that 20th century returns are indicative of future returns, then it is puzzling that they would give up an ERP of 5%. But I don't see why they (investors) should accept the historical case as being as strong as Brad thinks. (In which case, no puzzle!)

Keep in mind that inflation has been the main enemy of bonds and central banks in the 20th century didn't seem as capable of (credible at?) fighting it as they are now. Going forward, I doubt the realized 21st century ERP will be as large as 5%. For an analysis by the Economist on this, see the link below. They predict equity returns as low as 5-7%, leading to an ERP of roughly 0-2% !

http://infoproc.blogspot.com/2004/12/future-investment-returns.html

[Raph]

[Once again: think that investments in government bonds in high-inflation countries do very badly as well. I haven't been able to get survivorship bias to work as an explanation--and I've tried.]

Hi all

My comment on survivorship bias was not related to the SP500 construction methodology but was rather meant to be taken more broadly. You are only looking at the US which over the sample period has been the most successful economy with fantastic businesses. Were you to look at world equities at the beginning of the 20th century and compute an ERP over the overall opportunity set that ERP may well look much lower. Think Argentina, Egypt and what else.

Happy new year

[quartz]

I expect that if a larger and larger group of people did practice longer buy and hold of index funds that the risk premium might well decrease. I agree, however, that indexing only works if a goodly number of people don't do it, otherwise, there would not be an effective market.

[anne]

Oh dear, I am lost lost lost. There are a number of successful approaches to investing, mine, as was my father's, has been to know how to invest simply and pay no real attention to the chatter that makes no sense to me as I go happily about my day job. I have an income stream, I have a sense of relative value, I can wait indefinitely for value to be fully realized and wait some more. I know how not to compete with professionals.

[anne]

"I agree, however, that indexing only works if a goodly number of people don't do it, otherwise, there would not be an effective market."

Say what? Indexing would work if I alone were the only indexer. John Templeton cleverely found that out in the 1940s.

[Hal Varian]

From http://www.ischool.berkeley.edu/~hal/people/hal/NYTimes/2001-05-31.html

[And if you had put your money into the corporate bond market, or into Treasuries, in each of those 39 countries, your return would have been a lot less than 3.11%. In 4 of the 7 G-7 countries--Germany, France, Japan, and Italy--you would have been wiped out by inflation. My guess is that your real return on those 39 government debt markets would have been 0.]

Two financial economists, Philippe Jorion and William N. Goetzmann, asked the following question: Suppose your grandfather put $100 into each of the 39 major stock markets in the world in 1921, what yearly return would that investment have earned by the end of the 20th century? The answer is a shocking 3.11 percent -- essentially the same as the return on United States Treasury bills. This figure represents only the capital gains component of the return, since dividend data is awfully hard to come by, but it is still remarkably small.

On reflection, it's not hard to see the reason for this low return: a lot of bad things happened in the 20th century. The German stock market disappeared twice, in 1931 and in 1944. Most European markets in occupied countries were liquidated during World War II. In the early 1960's, the Egyptian and Argentine stock markets collapsed. In fact, according to Mr. Jorion and Mr. Goetzmann, this nation's stock market is quite an anomaly, scoring the highest of those 39 markets in both longevity and average return. And even the United States market looked pretty shaky in the 30's

[MrM]

What about the country/market bias? How does the equity premium vary across countries?

If you had started in Russia in the early 20th century, all of your investments would have been wiped out.

[And all of your investments in bonds would have been wiped out too. The equity premium puzzle is the premium return of stocks relative to bonds. You need some risk that is harmful to stocks but leaves bonds untouched.]

What would similar analyses forGermany or Japan look like?

[anne]

http://www.ischool.berkeley.edu/~hal/people/hal/NYTimes/2001-05-31.html

May 31, 2001

With Privatization, Market Risks Could Put a Hole in the Social Security Safety Net
By HAL R. VARIAN - New York Times

[Where Hal Varian was warning of a danger to the Social Security safety net from privatization in 2001, Chile was being used as a fine model for privatization. However, Chilean returns from pension investments are running at half the predicted levels and Chile is now designing the pension system anew.]

[anne]

http://www.nytimes.com/2006/12/26/world/americas/26chile.html?ex=1324789200&en=098ccf5916d505f8&ei=5090&partner=rssuserland&emc=rss

December 26, 2006

Chile Proposes to Reform Pension System
By LARRY ROHTER

SANTIAGO, Chile — Responding to growing complaints that the privatized pension system here is failing to deliver adequate benefits, the Chilean government has recommended that it be supplanted by a system in which the state would play a much larger role. The current system is a favorite of free-enterprise enthusiasts, including President Bush.

The changes, part of a reform package scheduled to go to Congress early next year, include a guaranteed minimum pension for the country's poorest citizens, even those who have never contributed to the private system.

The proposal also contains measures meant to stimulate competition, reducing the high costs to contributors and the extraordinarily high profits for pension fund administrators that analysts blame for some of the current problems.

"This is a radical reform, because it moves us from a system based solely on individual savings to one that includes a pillar of solidarity based on one's rights as a citizen, and not contributions," Labor Minister Osvaldo Andrade told reporters when the plan was announced on Dec. 15. "We are integrating systems that are fundamentally different."

Social Security in the United States, like national pension systems in many countries, is based on a pay-as-you-go arrangement in which workers, employers and the government all contribute. Under the Chilean system, in contrast, workers are required to pay 10 percent of their salaries into private investment accounts that they control; employers do not participate, and the state's contribution has been reduced.

In recent years, that pioneering privatized system has been emulated by a score of other countries and praised by leaders of many others. Mr. Bush, for example, proposed using the Chilean model as the basis for a reshaping of Social Security, calling the system here "a great example" and saying the United States could "take some lessons from Chile." ...

[anne]

Looking through the index records, the Chilean stock market has been a puzzle for 20 years, offering lower returns than other Latin American markets though the economy has been generally robust. The markedly lower index returns in Chile are puzzling and disturbing because of the importance of the returns for pensions, but Chile offers brilliant returns to financial companies handling pension accounts. Costs matter, but there is more here.

[steve]

Shorter solution to Equity Risk Premium "puzzle":

Past performance is not a guarantee of future returns :-)

[Shah8]

Anne, the chile example is an excellent approach to my first point, in the latter part of which, where I said that the premium is dependent on the relative gestalt between business and governments. Chile is completely dominated by mining and agricultural companies, with a large focus on the export market. The relative value of raw or relatively raw commodities decrease the value of organization, as all you need is the rump organization to provide vertical integration. And with all that commodities going elsewheres in the export markets, for value to be added elsewheres, and little to local companies, then the government, either because it owns the enterprise, or that it gets royalties from the resource extraction, offers the greatest value of organization! It is the one most able to garruantee it's own existence, and keep flows passing through it the best. Profit is as if you see an electric current. Long term, it's about how regular that flow is (hence the Fanny Mae scandals) No matter that there is some chicago school trained economist with awesome ideas for a corporation, the government crowds out investment, because it has preferential access to men and materials in a commodities based economy. And it does so *naturally*! So privatization of state resource wouldn't work.

If they want to improve diversity in the economy, then they gotta look to japan and korea...which is kinda dissapointing, because the *best* way, is fair labor laws and strong education effort. And that takes a very long time before the fruits come forth...

[anne]

Shah, that is most interesting but I did not understand where the earlier argument would take us. I like both arguments, and am thinking.

[steve]

Alphie,

The point of the ERP "puzzle" is not that buying and holding a certain *fixed* set of equities is likely to outperform buying and holding a particular bond. As I wrote before:

** The real question is why investors are willing to give up significant expected return to trade historical "equity-like" behavior for "bond-like" behavior. "equity-like" means more volatile, with greater prob. of big losses. (Giving up some return is reasonable, but how much is justified "rationally"?) **

"equity-like" could be an index, or whatever. The point is that people are trading away too much expected return for a decrease in vol (assuming of course bonds really are less volatile over time).

To have a puzzle you start with 2 asset types (e.g., a stock portfolio and abond portfolio) with different avg. return and vol characteristics, and then you argue that things are out of whack - people are paying too much return to reduce vol, why should they not prefer one portfolio to the other, etc. All the other stuff about survivorship bias, etc. is a red herring as long as I can realistically construct the two portfolios. The index algorithm I gave doesn't require any magical foresight - Vanguard really executes it :-)

The problem is that to say things are out of whack you must mean historically (based on some past performance), but you better be sure you didn't cherry pick the data (US equities, 20th century), because then instead of a rationality puzzle you just have some period of historically anomalous performance.

Actually it is even more subtle than that because the vol alone does not characterize the fluctuations in price unless you assume Gaussian (normal) behavior. If there is a fat tail (i.e., occasional disasters, which do exist) then you have to take that into account as well. Who is to say that 20th century US investors weren't lucky to avoid the fat tail, and that investors still in their guts know that?

[Dean Baker]

At the risk of being redundant, it seems that Brad has left the real economy out of his story thus far. In this respect, let me point out that CBO's latest projections show real corporate profits being almost 17 percent lower at the end of 2016 than in 2006 [http://www.cbo.gov/ftpdocs/74xx/doc7492/08-17-BudgetUpdate.pdf].

Current PEs are around 20, which allow for a dividend payout/buyback rate of approximately 3.0 percent. If the CBO projection proves right, and the PE stays constant, then real stock prices will fall at annual rate of approximately 2.0 percent over the next decade, giving a real yield of 1.0 percent, approximately 1.0-1.5 percent less than the real return on ten-year treasuries.

So, those looking for an equity premium must believe:

1) CBO's projections are hugely off the mark (if so, I guess we should throw out the budget projections derived from their economic projections);

2) PE ratios will rise enough to still provide a substantial equity premium (if so, please write down your end of period PE and see how seriously people take you); or

3) the economy has become so globalized that the profits of U.S. corporations bear little relationship to profits in the U.S. (if so, please write down the implied share of the profits of U.S. corporations that will come from outside the United States).

Equity returns don't come out of air. If people believe that we will always have an equity premium, then it is important that this claim be tied to the real economy. If the implications are not plausible, then it is not plausible that a substantial equity premium will persist into the future.

[RB]

Gannon makes an excellent case for the dangers of extrapolating historical patterns from 1925-1995 for the years after. Brief summary: the PE based on earnings over a 15-year period for any year from 1995-2005 is higher than anytime in the 70 years that preceded it. (And lower your expectations for equity returns in the future).
http://www.gannononinvesting.com/2006/12/in_defense_of_extraordinary_cl.html

[shah8]

Okay, it's late at night, but I've had another idea...

Why not do a set of graphs, equities spread of a equities/bond spreads vs *chief financial officer*'s period in office? Do this with a wider pool of countries. So you would do a graph of the equities/bond spread in japanese financial markets during the period Fukui is leading the administration. The same with Volker, so forth and so on.

Doing it this way, you can apply a number of statistical best fits, and use that to figure out if *government policy* is the source of the premium (like the very basic simply matching similar graphs to each other, and then do an intensive study of action (or by flagging the number of some kind of standard activity, such as raising interest rates) undertaken by the general ministry/gov't.

/me really wishes he could walk to a *real* library...

[Alex Tolley]

I'm not clear what data you used but here are my 2 cents:

1. Until the '60's there were no mutual funds to diversify stocks - so stocks had to be bought individually - not easy for most people and therefore less easy to diversify .

2. Transaction costs for stocks were very much higher in the past - pre 1975 in the US and pre 1986 in the UK, around 5%. Not something that you saw with other vehicles.

3. For most people, the choice is not vs. 20 yr long dated bonds vs equities , but short-term deposit accounts. Yes they are awful, but not so bad during periods of inflation, guarantee a positive return (before inflation) and they offer high liquidity. And let's not forget that most people did not know how to deal with inflation - I well recall in the UK a financial columnist speculating whether it was better to stop saving or increase saving during the 70's inflation.

4. Human behavior fighting the last experience. Depression in the 30's, hyperinflation in interwar Germany, inflation again in the 70's. Experience might well count against a rational expectation.

5. Behavioral finance experiments show that humans are not "rational" about returns - we hate losses. This biases choices. On this basis, the problem is not that there is an equity premium puzzle, it is just that the model doesn't fit human behavior.

[zinc]

Stocks are roughly equal to (cash dividend)+ dividend growth) discounted at some equivalent market rate of return. Corporate cash flows also tend to increase with inflation, a natural hedge.

The macro trend of the last 20 years:

Since 1983, the value of stocks have increased naturally with declining discount rate (bond yields) as inflation subsided. The relative P/E ratio expanded from 8 to 20. At the same time. equity prices have increased with earnings. The cash flow from bonds doesn't increase. As long as corporate earnings are increasing and the market discount rates are steady or declining, the price of the asset grows with the increase in cash flow.

The parabolic rise in equities since the mid-80's has been great. Actually, returns from equities since 1998 have been sub-par and would have much worse without the 17 % linear levitation we have seen since June 2006.

Rather than the simple 20 year period model, a more focused approach would ask "Are the future cash flows from equities greater than the future cash flows from bonds ?" Is the sceanario of the last 20 years repeatable over the next 20 years? Can P/E ratios continue to expand, interest rates fall, and inflation decline.

The nuance of this article suggest it will. The activities of the private sector and government are generally focused on providing an environment of increasing prosperity, enhancing the probabilities of increasing corporate earnings over time.

Sometimes it gets out of whack, though, and a reset is required. NASDAQ 2000. Japan.

[Randolph Fritz]

I think Aaron is onto something; as an individual investor, one has to assume one will want one's money out at an unexpected time--illness, job loss, what have you. And one can get thorougly screwed in the the short-term if one doesn't have some money in an investment with steady value. Small investors, in other words, cannot take full advantage of the equity premuim. Perhaps part of the issue here is that some investors don't realize when they've become "large". (If one buys a house in the USA, one necessarily has an investment outside of stocks.)

In the long term, the numbers one sees here reflect the overall growth of the US economy through the 20th century; I can see how to capture this in investment in stocks, but not in bonds, and I think the equity premium reflects that. That period, however, saw the emergence of the oil economy, two post-war booms, and enormous expansion of international trade. Do we believe that such growth will be sustained in the 21st century, with the abandonment of the oil economy (the entire planet's economy will shrink if this is not done) and huge climate and ecological disasters? Perhaps growth in China and India will offset that. It does seem to me that if there is an equity premium to be had in the 21st century, it is to be had by investment in international equities, if this is even possible.

[steve]

Brad,

Doesn't the *tone* of the discussion here (independent of the actual arguments) undermine the claim that there is an ERP puzzle?

You have over 50 comments from dozens of commenters, with varying degrees of economic and financial literacy, but including at least several experts (at least two well-known economists!). A significant fraction seem unwilling to accept that the 20th century US data set is indicative of future performance.

Given this reaction, isn't it reasonable to conclude that a large fraction of actual market participants might feel the same way? Many commenters here, even after reading your nice exposition, aren't confident enough to bet on your money machine with their own portfolios!

If indeed the ERP persists at 5% over the next 20 years, perhaps it is better to regard those that take advantage of it to be simply more canny investors than the market as a whole.

If you bet on equity outperformance and turn out to be wrong over the next 20 years, well you just made the wrong bet on a misleading signal.

Either way, there is no rationality puzzle for the market as a whole. Rational people seem to be able to look at your data set and come to different conclusions.

[David Levine]

Brad:

I always thought there were two answers.

1. A century is a lot of data for economists. At the same time, it is only 5 20-year non-overlappying periods. It is risky to make big predictions on 5 data points. (I realize there are more than 5 data points, but far fewer than 100 independent 20-year periods.)

2. To look at more data, you can look overseas. Naive observations find a big equity premium, I believe.

BUT... Lots of stock markets have shut down. Thus, looking at stock returns in the U.S., where the market remained open, has survivor bias. Bonds also have that bias (that is, many nations have defaulted and some have ceased to exist for periods of time).

[Yep. The equity premium is not due to survival bias.]

Nevertheless, it is worthwhile to look at more nations' data, not just the U.S.

As Niels Bohr once said, "Prediction is difficult, particularly of the future." As you know, stock returns are notoriously difficult to predict. It makes sense to hold lots of equity, but I do not recommend that any nation or person borrow on the bond market to buy stocks.

[Laurent GUERBY]

If you look at historical spreads between other asset classes you can see convergence. For example obligations and housing rents have converged to the exact same return during the past 20 years in France. See graph on the upper right of page 7 of this PDF:

http://www.credit-agricole.fr/IMG/pdf/EI_061018.pdf

[Anthony Damiani]

So, what you're saying is, "don't buy bonds"?

[Dean Baker]

Since Brad was good enough to repost this at the top of his page, let me again see if I can get a simple answer out of the promulgators of the equity premium view.

As I noted, CBO projects that real corporate profits will be 17 percent lower in 2016 than in 2006. This means that if PEs stay constant, real stock prices will fall by almost 2 percent annually over the next decade. With a 3 percent dividend/buyback rate, this implies an annual real rate of return of approximately 1 percent.

So, I would like to know where my equity premium will come from?

Is CBO off the wall -- should be we be doubling those profit projections for 2016? If so, I guess that we have to change the deficit projections as well -- happy days are here again.

Alternatively, will companies pay out more than all their profits as dividends or in share buybacks? interesting story.

Maybe PEs will rise enough to maintain a decent equity premium. A PE of around 33 or 34 come 2016 should do the trick (don't ask about returns in years beyond 2016).

Or, we could see an unprecedented uptick in the rate of globalization of capital so that the bulk of the profits of U.S. corporations are coming from abroad in 2016.

These are the possibilities I see on the table. I would hope that one of the equity premium promulgators can tell us where we think our premium will come from in the next decade. It would make me much more comfortable about having my retirement money in the stock market.

[Barkley Rosser]

Dean,

Actually, last year when the social security privatization debate was in full public swing, some of the more sophisticated defenders of it fessed up that the only way to do it and maintain the likely returns, especially if one was going to go along with the horribly pessimistic forecasts of GDP growth in the infamous mid-range forecast from the SSA Trustees, was to buy foreign stocks, particularly those in China and India. That is what all those Gen-Xers are supposed to be counting on 50 years from now, after today we cut their future benefits from social security.

[Porlock Junior]

Randolph Fritz made a point that looks rather like another paradox: one that might reolace that of the ERP.

"In the long term, the numbers one sees here reflect the overall growth of the US economy through the 20th century; I can see how to capture this in investment in stocks, but not in bonds, and I think the equity premium reflects that."

This sounds very reasonable. It also takes into account the element of selection bias and survival bias inherent in using US 20th-century numbers rather than those of economies that have been less fortunate relative to those long-tail disaster possibilities.

But what if bond markets tried to capture the effects of such growth? J. P. Morgan says, "You want me to lend to the government at 2%? According to 20th-century trends, I can make 8 or 9 without any effort. Applying MPT and some reasonable assumptions, I insist that you pay me at least 7.153%. Fork over, Andrew, or no money for you." And his friends (and his more numerous enemies) all say the same.

So, if risk-free bond interest rates had reflected a true relation to equity returns all that time, could we possibly have financed that high growth rate?

[anne]

Dean Baker:

After considering the estimate on corporate profits from the Congressional Budget Office, I have no idea why we should pay the least attention to it. The idea that real corporate profits will be lower in 10 years time than now is just nutty. Look to fine corporate profits in weastern Europe and resulting gains in stock prices in domestic currencies and dollars. Investors are rightfully paying no attention to the CBO estimate.

My estimate going forward after John Bogle is to allow for a constant price earning ratio, assume a conservative 7% increase in corporate profits and 1.5% dividend for an increase of about 8.5% a year or, say, a range of 7.5% to 9.5%.

Even a 7% increase in profits with a 2.5% increase in dividends and stock buybacks gives 8.5%.

[steve]

Here's a calculation much like Dean Baker's, which appeared in the Economist some time ago. It leads to a predicted ERP of 0-2% over the next 20 years. The main assumptions are that corporate profits can't grow faster than GDP, and that P/E ratios aren't going up from where they are now. Of course, it is possible that US corps. will get most of their earnings growth abroad from faster growing economies (e.g., India and China), but that is a rather aggressive and speculative assumption. Brad's money machine can't continue to operate over the next 20 years without at least *some* nontrivial assumption.

Economist: "...Assume that America's nominal GDP grows by 5% a year (3% in real terms, plus 2% for inflation). If the share of profits in GDP is constant, profits will grow at the same rate. However, profits could do much less well, because in America, Japan and the euro area their share of GDP is close to a record high. They might well be expected to fall.

Suppose, though, that profits do rise in line with GDP and that p/e ratios stay the same. Then, Mr Barnes estimates, the total nominal return on American shares over the next decade will average 6.8% (5% profits growth, plus dividends), half the figure for the past 20 years. If profit margins fall modestly and the p/e ratio reverts to its long-term average, returns will average 4.9%—well below investors' expectations. Surveys suggest that individuals expect returns of more than 10%."

http://infoproc.blogspot.com/2004/12/future-investment-returns.html

[anne]

There is no reason to believe that an economy that grow sufficiently to insure a 7% growth in corporate profits for 60 years, cannot continue to grow sufficiently. Again, a conservative 7% increase in profits along with 1.5% in dividends after Vanguard index fund costs is reasonable. Not to mention stock buybacks, and implied dividends of corporations such as Berkshire Hathaway which pays no dividends and does not buyback stock. Not to mention the increasing advantage in profits from international operations.

There is no reason to assume a return to American stocks of less than 7.5% to 9.5% at a constant price earning ratio over the coming decade.

Notice also however how investors have leaned to value these past years in looking for returns.

[anne]

The point of end-of-the-world estimates for future American economic growth as far as I can tell are simply an attempt to undermine support for social insurance and benefit programs such as Social Security, Medicare and Medicaid....

Productivity growth is not ending, western Europe is growing nicely even with the dread goblin of, shudder, older Europeans. Growth will be fine, especially so if we can ever be sane enough to stop spending what is now $14 billion a month, yes, $14 billion a month for the tragedy Iraq.

[anne]

http://www.mscibarra.com/products/indices/stdindex/performance.jsp

National Index Returns [Dollars]
12/30/05 - 12/31/06

Australia 32.5
Canada 18.4
Finland 31.0
France 35.4
Germany 36.8
Hong Kong 30.4
Japan 6.3
Netherlands 32.4
Norway 46.3
Sweden 44.6
Switzerland 28.2
UK 30.7

[anne]

http://www.mscibarra.com/products/indices/stdindex/performance.jsp

National Index Returns [Domestic Currency]
12/30/05 - 12/31/06

Australia 23.3
Canada 17.9
Finland 17.2
France 21.1
Germany 22.4
Hong Kong 30.7
Japan 7.3
Netherlands 18.5
Norway 34.6
Sweden 24.4
Switzerland 18.8
UK 14.6

[anne]

What is expecially interesting is to notice how well cautious value-focused investors have done these 10 years or this year. Evidently, the more cautious the better the general portfolio return.

When the Economist offers investment suggestions, my sense is to hide the children.

[stebe]

Ann,

If corporate profits grow at 7% but GDP at 5% (3% in real terms - slightly above long term trend), then after 20 years the corporate share of GDP will be substantially higher than ever before. Note the corporate share of GDP at the moment is at a historical high, so your projection puts us in uncharted territory - it's not a conservative projection relative to long term history.

It seems more likely that the corporate share of GDP will revert to its long term average, which means an even lower ERP (perhaps zero, if P/E ratios also revert to average values).

These projections all depend on whether you regard the recent run up in equities to be normal or exceptional. Regarded in a broader context, the recent period (say, last 60 years) is exceptional and not likely to be repeated.

I don't know what will happen, but rational people can come to very different conclusions.

[David Amann]

Someone wrote: As others have noted most investors-even pension funds with annual contribution liabilities and benefits to pay-have a need for stability and current income.Dollar cost averaging is great on the way in,but can kill you during the distribution phase.Taxes also have a substantial effect on the re-investment rate.
============================================

This is where the popular books on investing and indexing are a little sparse. Too often, popular investing advice focuses primarily on return without any discussion of volatility.

While this is not as important in the accumulation phase of an individual's investment life, volatility becomes extremely important in the income distribution period of an individual's investment life.

In order for retiree's to live a comfortable retirement, they must have some exposure to equities; however, if they can limit the ups and down of that exposure with a portion in bonds, and a good asset allocation strategy, retirees can capture a significant portion (though not all) of the return of the market without having to liquidate a huge number of shares in the periodic downturns.

In other words, when you need to live off of your money rather than your labor, you're going to have to learn a lot more about managing money.

David