« Dealing with the Current Financial Crisis | Main | The Economic Journal of Negative Research Findings »

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Between the market top for the DOW in 1929 and the next major top in 2000, this index went from 400 to 1100 adjusted for inflation. This is 1.45% per year not including dividends and inflation. See http://www.dogsofthedow.com/dow1925cpilog.htm

Considering that our government has a tendency to undervalue the CPI, that the population growth is slowing and that the cheap energy is already gone, the future look bleak.

If you add to this that the DOW is constantly removing/adding companies, the typical mutual fund fee is 2-3%/yr and, let's not forget, the debt vs GDP is much higher than in 1980 of 1929, I don't see how a pension fund can assume a 6% rate of return.

Between 1929 and 1983, the DOW returned -1.7% (inflation adjusted).

"Allegedly, stocks generate, on average, higher real returns than bonds, that is, there is an equity premium."

Um....why allegedly? There's data on this isn't there? You could like, you know, calculate averages. You could even ask someone who knows how to test whether their average returns are statistically different.

Or did you have something else in mind?

Simon: "Um....why allegedly? There's data on this isn't there?"

If you want to demonstrate that there's an equity premium, start in 1932 and end at some point when the market is at a peak.

If you want to demonstrate that there isn't an equity premium, start in 1929, and end at some point when the market is in a trough.

Whether or not you find an equity premium depends upon the time period one chooses for the analysis.

Moreover, just because historically there has been an equity premium doesn't prove that there will be one going forward.

Normand - I'll take that as a "no" vote on the equity premium.

Here's a simple example of why you might want to buy stocks; consider the TD bank (ticker symbol TD.) Their stock (as of 2:18pm today) had a dividend yield of about 3.5%. That means that, if you bought the stock today and dividends stayed at the level they've been over the past, you would get $3.50 in dividend payments over the next year for every $100 you invested in the stock. If I instead buy a 1 yr Canadian Govt bond, I'll get just under $1.20 for every $100 I invest. According to the Bank of Canada, yields on Canadian Govt. Bonds of any maturity are all under 3.5%.

Historically, however, stock prices tend to rise over time. That means that I should expect not just to earn the dividend yield, but I should also expect to make some capital gains. (On average. Your results may differ.)

If you think that there is no equity premium, then you need to explain why the expected return on, say TD stock is down around 1.2%. That means that you must be expecting (a) the price of the stock to fall, or (b) that dividends are going to be cut a lot, or (c) both. In fact, some people are arguing that the very low interest rates we see currently make stocks look unusually attractive, in the same way that standing next to someone very ugly makes people seem more attractive.

Frances: "just because historically there has been an equity premium doesn't prove that there will be one going forward."

Okay....so you want to allow for a time-varying equity premium; that's fine (but not what your opening sentence says.)

Sold my TD.TO a couple of months back, after holding for decades. Agreed with my broker that they were, as he put it, "priced for perfection". Did well to sell them then. Was vaguely thinking of buying them back yesterday. But couldn't decide. Might still do so.

The best thing I read (or heard about) on the equity premium puzzle wasn't even supposed to be about the equity premium puzzle. That recent study comparing owning vs renting a house in the US. It said renting was almost always better, historically. But one of the assumptions was that the money saved from renting was invested in stocks. So essentially it was comparing investing in stocks vs investing in housing. Stocks beat housing. No surprise. Stocks beat bonds too.

Simon: "That means that you must be expecting (a) the price of the stock to fall, or (b) that dividends are going to be cut a lot, or (c) both."

Yes, that's right. Since the TSX is basically unchanged from August 2006, and well down from previous peaks, (a) certainly sounds plausible enough if you're thinking about the market in general.

But let's take TD - it's true, the dividend-paying stocks are the only ones on my screen that have that happy green writing beside them right now.

What if we are 99% confident that TD will continue to make the same dividend payments on an on-going basis. Then the sensible thing to do is to sell bonds and buy TD. TD stock values will soar. We'll see an equity premium for a while.

Then what? We're in a new equilibrium, where TD stock is worth, say, twice what it is right now. Going forward from there, will the equity premium continue? Why?

Simon - "but not what your opening sentence says."

No, it's in the title (Does the equity premium STILL exist?)

Simon,

Just a few days back, Yellow Media had a 35% dividend return. The share price keep falling and they finally cut the dividend from 0.65$ to 0,15$. That company is around for a very long time and most pension fund used to carry some of their shares. It could happen to any company, TD included.

I am following the 20 largest financial companies in the US since 2006. That list is now down to 9 (some went broke or are penny stock or in receivership). Out of those 9 you can count BAC, C, MS and WFC as bankrupt if normal accounting practises were back (market to market).

My broker offered me a 2% interest for all margins. For sure this has an influence on the market.

So for those of you who like fun and games with Excel, you can get high-quality long-term data on US stocks and bonds and inflation from Prof. Robert Shiller's (of "Irrational Exhuberence" fame) web site at http://www.econ.yale.edu/~shiller/data/ie_data.xls

To compare stock returns to bond returns, I don't have to worry about inflation (it affects them equally), so I'll just compare stock and bond nominal returns. Stocks peaked in Sept. 1929 at 31.30 and troughed in March 2009 at 757.13; that gives me a compound annual rate of return of 3.9% excluding dividends. Including dividends gives me another 4.0% compound annual rate of return. Bonds (in this case, 10 year US Treasury Bonds) gave me 5.3%.

So even when I try to look at things the way that Frances suggests, I still get an equity premium of a few percent (or about 50% more than the return on bonds.)

Frances, why do you think that the premium is zero? Is there some evidence to back that view?

Frances, agreed that the equity premium seems an artifact now. But what happens if this becomes the new normal? No one will buy equity anymore.Why would you when a bond, which has senior payment priority, promises the same return? It would be impossible for any company to raise equity financing anymore. Unless they start paying a premium again.

First a little history. While we associate the late 1940's and early 1950's with a period of unparalleled prosperity, the stock market of the time didn't reflect that. When looking at what market players and traders actually thought (some of them wrote about this time) the received wisdom was that there was going to be another Depression. The thought went that the Civil War was followed by the Panic of 1873 and the Long Depression, World War I was followed by the Great Depression and the latest war would be followed by a depression soon enough. Companies like Xerox whose business was booming trades a 3 times earnings. The P/E multiples of this time period were ludicrously low by today's standards.

The Dow didn't reach its high of 1929 until 1954. Further, please remember the Dow is not a proper index. It is a price-weighted list of 30 stocks. The S&P 500 is much broader and weighted by market capitalization and is generally considered a much better index methodologically speaking and a much better representation of US Equities and the market's view thereof.

Simon - "why do you think that the premium is zero? Is there some evidence to back that view?"

I'm not saying that the premium in the past has been zero. I agree that there are periods over which one can observe an equity premium.

I'm questioning whether or not we can accept as gospel truth that the equity premium will continue to exist in the future. As I said in the post: "If stocks had a higher return, on average, than bonds, surely any investor large enough to diversify risk would buy mostly stocks. This would bring up the price of stocks up until, in equilibrium, the returns to stocks would be close to the return to bonds."

As rogue says, there are still privileges to bond holding, which would justify a small equity premium. But do you sincerely, honestly, believe that stocks will continue to generate a 50% higher return than bonds going forward, indefinitely, forever? And given that firms will still choose to finance their undertakings through equities rather than bonds?

Norman: "Between the market top for the DOW in 1929 and the next major top in 2000, this index went from 400 to 1100 adjusted for inflation. This is 1.45% per year not including dividends and inflation. See http://www.dogsofthedow.com/dow1925cpilog.htm"

Sure, but if you're excluding dividends, you're excluding a big chunk of the return of the Dow, and equities more generally, over the past century(typically, for most of the past century, in excess of 4% a year).(see http://disciplinedinvesting.blogspot.com/2009/02/dividends-critical-component-of-total.html). That's like measuring the return on bonds and ignoring interest payments.

Interestingly, during the 1980's and 90's, we saw a decline in the dividend yields of the major American stock indexes. I couldn't tell you why that is (if you're a pessimist, it's because the market is still overvalued, if you're an optimist, it's because companies chose use retained earnings to boost shares prices, either through investment, or through share buy backs). I would not be the least bit surprised, though, if we don't see that trend reverse over the next decade.

Frances:

I sincerely, honestly expect that from today onwards stocks will have a higher rate of return than bonds, and that it will quite a bit larger than what is indicated by those calculations I did (which you'll recall you designed to give the lowest possible return for equities.)

I do not agree that "...in equilibrium, the returns to stocks would [or should] be close to the return to bonds." This is because (a) I think stocks add more risk to a diversified portfolio than bonds, and (b) investors are risk averse, so (c) they need an equity premium to entice them to hold stocks in their portfolios.

Yes, that premium may vary through time. This afternoon, I think many investors have decided that stocks are riskier than they realized, so they are requiring a larger equity premium on stocks going forward. If you think we're in equilibrium right now, I think you should expect a higher, not a lower, equity premium looking forward. (No, I have no idea whether we're in equilibrium.)

Didn't most countries, especially the US, UK and Canada engage in "Capital Repression" until the 1960's? The US debt-to-GDP ratio in 1948 was 120%. The US Treasury actually had to sit down and negotiate a deal with the Federal Reserve in 1948 to keep interest rates low, specifically for the short term so the US could begin to pay down its debt without facing higher interest costs but more generally because it allowed the debt to be managed through inflation and economic growth, both of which were enhanced by low interest rates.

With interest rates suppressed no wonder equities looked good. But interest rates are no longer repressed like that so it stands to reason equities and bonds should converge.

Isn't there a fundamental problem with trying to determine whether there is an "equity premium" on a retrospective basis? Isn't the equity premium an inherently prospective concept.

For example, my return on Yellow Media has been, thus far , far (far, far, far) less than the return on US treasuries that I might have invested in. But I didn't buy Yellow Media units on the expectation that their price would plumet precipitously (although I recognized that as a risk), I bought Yellow Media units on the expectation that they would outperform US treasury bills by a sufficiently high margin to ouweight the risk that they might not. Had you offered me Yellow Media units with an expected return of 1% (or whatever the US government is paying these days), I'd have told you to get stuffed. That's the equity premium.

The fact that investors (and the market generally) might made investment decisions that, with the benefit of hindsight, were wrong (i.e., in 1929, stocks were badly overvalued, in 1950, they were badly undervalued), doesn't prove that there isn't an equity premium.

And while I agree with Frances that bonds carry their own risks (some common to stocks, i.e., bankruptcy, others distinct from stocks, perhaps inflation), I don't think it's absurd to say that they are, as a class, less risky than stocks (since the latter is entitled to a fixed, and preferred, income payout and gets priority in bankcruptcy, liquidation, etc.). It might be interesting to look at the environment of large corporate issuers, and compare the returns on bonds, stocks, and various hybrid instruments with different degrees of debt/equity features (preferred shares, convertible debentures, etc.). You would expect to see the expected return at the time of issuance on those various instruments increase fairly smoothly as you move from the equity to debt side of the spectrum - although that may be complicated by inconsistent tax treatment of form of capital income.

Bob, I agree on the inherently prospective nature of the equity premium.

On risk premia:

The Canada Pension Plan Investment Board, according to its latest investment report, has $148 billion worth of assets. http://www.cppib.ca/Investments/Total_Portfolio_View/. You can't diversify away risk. I can't diversify away risk. But surely the CPP should be able to diversify away risk.

If there's an equity premium, the CPP investment board and any other investor with vast amounts to invest over a long time period should buy only stocks. Increasing demand for stocks. Raising the price of stocks. Until the only equity premium that remains reflects the risks that can't be diversified away by a $148 billion dollar long term investment fund - i.e. a very low level of risk.

Frances, you might want to take a look at The Cross-Section of Volatility and Expected Returns (Ang et al, 2006). I presented their findings in the financial econometrics class this past winter and what they argue is that while there is a return on stocks with idosyncratic volatilities greater than government bonds after a certain point portfolios with high volatilities show returns that are lower than theory would predict. What me and professor Wan concluded was that portfolios at the high end were being purchased by people that simply lacked the knowledge to properly price portfolio returns, ie investors that liked to gamble with their investments.

Now the data we were looking at was from before the recession but its quite plausible that if we are in a time global uncertainty there will be no market actors that can properly price risk and therefore there will be no equity premium.

"But interest rates are no longer repressed like that so it stands to reason equities and bonds should converge"

Really? You think US Treasury yields can be repressed much more than the 0.01% yield they're currently paying on 90-day bonds? Today's 10-year bond rate is roughly the same, or lower, than the 10-year bond rate through the 1940's and the 1950's.

"The Canada Pension Plan Investment Board, according to its latest investment report, has $148 billion worth of assets. http://www.cppib.ca/Investments/Total_Portfolio_View/. You can't diversify away risk. I can't diversify away risk. But surely the CPP should be able to diversify away risk."

Isnt there always going to be some amount of market risk that can never be diversified away unless you invest in government bonds?

Compared the the price inflation, wage increases and general economic growth being exhibited? In the 1950's we had strong growth with low interest rates. Now we have anemic growth with low interest rates.

So interest rates were low in 1950's compared to the rate of return offered by the economy, while today the interest rate is much higher compared to the economy's rate of return.

I am right if I say: Planet earth taken as a whole, productivity gains being neglected, the expected return on all assets (bond, stocks, GIC, etc) is the inflation rate, or 0% inflation adjusted.

Otherwise we are getting richer by printing more dollars which only means more inflation. Or where is that wealth coming from?

This is assuming a long enough time frame.

Does this make sense for a bunch of economists?

There are two arbitrage conditions, the (nominal) return on capital versus bonds, where the latter is a policy variable, and Tobin's Q. I don't see why the two must give the same price -- so there will be some form of arbitrage available. But the arbitrage is limited -- let's say households can only borrow against collateral when they want to purchase capital, and that as household leverage increases, they demand a higher spread in order to engage in the arbitrage, etc. Pick your favorite credit friction.

In that case, it makes sense that central banks would set the bond rate to be below the nominal return on capital, because they *want* to encourage borrowing at the bond rate and purchasing capital. They want to encourage investment, and they also want to maintain a positive inflation target.

All of that points to some form of long run equity premium apart from risk preferences.

Frances: your basic argument, as I understand it, is that bonds ought to have the same expected return as equities because it is possible to hold large portfolios of either, thus diversifying risk. But it is only the idiosyncratic risk of a security that can be diversified away; systemic risk cannot be eliminated by diversification. The systemic risk of equities is much higher than that of bonds. When you buy a large diversified basket of equities you are investing in the future real economic output of the sector or economy to which your basket belongs. A small change in your estimate of the future growth of this output has a large change in your estimate of the value of the basket. When you buy a large diversified basket of bonds, you are investing in a portfolio of cashflows that are nominally fixed. Yes there is default risk, but the component of default risk that is systemic corresponds to the Armageddon scenario in which no market instrument pays off. Yes, there is interest rate risk, but the equity portfolio is subject to the same risk. Yes, there is inflation risk, but in the economies that successfully issue large amounts of nominal bonds that risk is perceived to be small.

The very nature of a bond - an asset with downside but no upside - tells you what the market perceives it to be: a safe asset. High-yield bond account for a small fraction of the bond market. Remember that until Milken came along, there was no such thing as on-issue junk, only "fallen angels." The buyer of a bond is buying safety, and has to pay the price of the hedge - there's no free lunch.

It is also worth mentioning that the idiosyncratic risk of equities is so much higher than that of bonds that it makes a practical difference in portfolios. An equity hedge fund can happily cap its leverage at 2:1, but you can't make money in bond arbitrage without 10:1 or 20:1 leverage.

One last point: the riskier a bond is (considered as an individual security), the more it behaves like an equity.

If there's an equity premium, the CPP investment board and any other investor with vast amounts to invest over a long time period should buy only stocks. Increasing demand for stocks.

Well, let's turn that around. If there isn't an equity premium, why on earth would the CPP invest in stocks? If the long-run expected return is the same, and bonds are less volatile than stocks, they (and everyone else) would be crazy to invest in stocks.

In practice, of course, the CPP does invest signficantly in equity investments (though often not in publicly traded stocks). As of last spring, 54% of its portfolio was in equity investments, while 30% was in fixed income assets (the balance is held as what they call "inflation sensitive assets", i.e., real estate, infrastructure, inflation linked bonds).

Of course, you might expect to see the equity premium change over time as people's subjective perceptions of risk change. In the 1950's, I'd imagine that the generation of investors who were raised during the Great Depression had a healthy appreciation of the risk of equity investments relative to government bonds, so you'd expect the equity premium to be significant, in order to induce people to invest in equities.

Conversly, the generation of investors who may have taken a beating investing in debt in the late 1960s and 70's, might understandably be skeptical of the security of debt investments, and be willing to accept a lower (or non-existant) equity premium in the 1980s or 90s. In light of the volatility of the market over the past 3 years, we might be entering another era of high equity premiums, as aging boomers scarred by the last crash and worried about their retirement savings switch to safer investments.

"You can't diversify away risk. I can't diversify away risk. But surely the CPP should be able to diversify away risk."

No.

Just.....no.

I like you, Frances. And, as a friend, I think you really should think hard about reading an introductory textbook chapter on risk and diversification. There's lots of good ones out there.

Perhaps you could blog again after that.

"If there's an equity premium, the CPP investment board and any other investor with vast amounts to invest over a long time period should buy only stocks. Increasing demand for stocks.

Well, let's turn that around. If there isn't an equity premium, why on earth would the CPP invest in stocks? If the long-run expected return is the same, and bonds are less volatile than stocks, they (and everyone else) would be crazy to invest in stocks."

Um....there may be a good reason to invest in stocks even if (a) stocks are riskier than bonds, and (b) the expecteed rate of return on stocks is lower than that on bonds. It is quite possible that adding stocks to a diversified portfolio of bonds lowers the risk further still. (This is typically a mid-term exam question when I teach basic finance.)

The example that I used to use in class was gold; terrible rate of return (I think I need a new example this year -- maybe I should use yen) and very volatile returns. But many people argue that gold has a negative Beta; it does well when other assets do poorly (at least that seems to be holding up) and vice versa. That means adding some gold to an otherwise well-diversified portfolio can lower the overall risk.

Simon,

Fair point. Given that for the immediate future, the government doesn't need to dip into the CPP IB, that's probably why they hold bonds.

Simon -

Just to clarify - I wouldn't predict there to be no equity premium. Just a much smaller one than is generally supposed to exist. A dx sized premium.

I don't think that reading a basic finance textbook would help, because the issues I'm having here are not economic.

The first is philosophical: the problem of induction. As Bob Smith has pointed out, the crucial issue is whether or not the equity premium exists going forward. And knowing that an equity premium has existed in the past doesn't prove that one will exist in the future. It's like the inductivist turkey, who wakes up happy and well fed on Christmas Eve and figures "I didn't get killed yesterday, I should be just fine today."

The second is psychological. Based on the blog survey we did a while ago, the people who read this blog are overwhelmingly male and highly educated. I.e. precisely the demographic that financial literacy research suggests are highly confident of their ability to invest in the stock market. I'm just way less confident, I'm the type of person who says "You can give me a bond that will pay me a return if the market goes up, but will still protect my premium. Great!" (Actually that's not far off.) So I just don't believe that I'm going to be that lucky, that my stocks are going to fly like an eagle, carrying me to some kind of Freedom 55 paradise.

Tell you what, if you're right and my stocks do outperform my bonds by 50% over the next 25 years (or whatever you figure is a reasonable bet), I'll buy you a $100 bottle of wine. 2011 $. Because I like you, too. And if the market does that well, I'll be able to afford it.

But I'm guessing that by 2027, when the last of the boomers hits 65, the CPP starts winding down that big pension fund to pay for my retirement, every one of the boomers has switched from stocks to bonds to avoid that systematic risk, and there's very few people in the peak investing years (50 to 65) we'll see a huge sell-off in the market. Indeed we're already seeing the early boomers hitting the age when standard investment advice says sell stocks buy bonds, but that should be offset by late boomers coming into their peak savings years.

Some of the more productive enterprises are funded by shares, not bonds, because the expected cash flows are uncertain, the inherent nature of innovation. So it's not surprising that the vast majority of bonds cannot and do not play in this arena. It makes sense that those who invest in economically productive innovation should reap the spoils, but only if it's truly innovative, and thus get a premium for their trouble. It has turned out enterprise is generally innovative and I expect this to continue. This often gets confused with volatility which involves a whole lot of non-innovative activities but once we filter that out it shouldn't be surprising at all that equities have a premium.

I think the answer is that it isn't an equity premium. It's just a risk premium and it's just that most equity is more risky than most debt. So the common stock of a particular company would outperform its bond. However, there is no reason why junk bonds would not perform similarly to some stocks. As for diversification, it works for both debt and equity similarly. So while you can use diversification to reduce your equity risk to the level of buying a single bond, you can also reduce your debt risk by diversifying your bond portfolio too. So the argument you are making for stocks would apply to bonds too, part of their risk premium gets competed away through diversification. Now we are back to a world where the higher risk in equity is rewarded with a premium.

there is an equity premium and it is 3.25738492835983274%.

The volatility of the equity market portfolio is between 12-25% (obviously varying with time). The volatility of the bond market portfolio is between 4-6% (again varying). This has been observed for a long time. If there was no premium from holding stock, the market would be irrational.

As said above you can see equity as a (usually deep in the money) call option on the same underlying asset (the whole economy, obviuosly not perfect, but a reasonable approximation). That gives you leverage (as the value of the equity is just the premium of the option which is much smaller than the value of the underlying asset), which gives you excess return (and the possibility of losing everything)...

Considering that average levarge is about 200% (ratio of company asset to market cap), a doubling of the return is not unreasonable (should be less since you would need to remove the time value of the option).

If you refuse induction, you refuse the scientific method... It's like saying just because apples fall to the ground every day in the past it doesn't mean they will fall in the future... Literally true, but not very helpful in trying to understand the world...

Actually scratch my leverage comment, that's wrong. The time value should cancel that if we were risk-neutral... But people are not risk neutral so there is still an excess return...

Frances; Thanks for the offer of the wine; that's very sporting of you. (I like a nice bottle of Spatlese.)

But I think you do have a problem with the economics; I think you would find it useful to work through the mathematics of diversification -- particularly the difference between market and idiosyncratic risk. Understanding the idea of a stock's "beta" might give you a different perspective. Not that it rules out time-varying risk premia.....

Frances: A few more comments on your post from 6:47 pm.

Induction: I have no problem with your reluctance to embrace induction. But your theories still need to be able to fit the facts. Your comments about historical stock and bond returns show that you could be more familiar with those facts. (Again, introductory finance textbooks cover this stuff.)

Psychology: I'm not sure that I understand your point here. I don't invest with confidence either and I'm skeptical of the "Freedom 55" marketing that retail institutions bombard us with. So? What does that have to do with the equity premium?

Demographics: Yup, aging will tend to make retail investors switch from stocks to bonds. But lots of other factors may make those changes seem small. I've already mentioned the elasticity of supply argument. Remember also that financial sector can sometimes be pretty good at repackaging risk; most boomers would be happy to buy a pension from pension funds that invest in stocks and other assets. We've also seen big changes in retail investor tastes; stock ownership is much more widespread than it used to be (often through mutual funds or ETFs.) As a non-confident investor, how sure are you that the demographic effect will dominate?

Economics: Most finance professionals associate relatively high stock prices (relative to earnings or dividends) with low expected returns and vice versa. I think you're arguing for low stock prices. Why don't you think that means high expected returns? Given our low interest rates, doesn't that imply a high rather than a low equity premium?

I promise I'll let you have the last word.

Here's another way to think about it -- rather than from the buyer's perspective, think about it from the perspective of the seller (one who needs finance). Let's start as if there were no equity risk premium -- the price of equity and the price of debt risk are equal.

If you view your business as a call option struck at zero (where the premium is the npv of all the startup plus ongoing costs), would you rather finance some of your business by giving away potentially unlimited upside or repaying a known, fixed stream?

Even absent tax benefits of debt, it seems reasonably intuitive that sellers of equity/debt return (i.e.,business owners) would prefer selling debt to equity at the same price (of risk) and keeping the unlimited upside for themselves.

Another interesting question is to then think about what the equity risk premium should be for short option / debt-like companies (e.g., banking and insurance -- those businesses received mostly fixed, limited upside payments and have event risk that can take their enterprise value to zero very quickly).

intellectual hattips to Ian Ayres for "flipping it" and Nassim Taleb for "banks/insurance are short options".

err, to finish the previous thought -- marginal sellers of "debt return" --> lower price of "debt return" --> higher price of "equity return"

1. With 10-year TIPS yielding precisely zero, I sure as hell hope there's still an equity premium!

2. Of course there is (ex ante, right now). The S&P 500 is selling at about 20 times average real earnings for the past 10 years. Make the ridiculously conservative (relative to conventional forecasts) assumption that average real earnings on the S&P 500 will be the same in the future as they have been over the past 10 years. Given this assumption, and with bonds yielding zero in real terms, you would need really bizarre assumptions to get an equity premium anywhere near zero going forward.

3. Will the equity premium go to near zero eventually? I doubt it, though I think the trend may continue to be downward. But unless you make bizarre assumptions about earnings, a near-zero eventual equity premium implies that the equity premium today is huge, because you will need huge capital gains (relative to bonds) to get prices to the level where they are consistent with a zero equity premium going forward.

Equities outperform bonds 3 out of 4 calendar years. Therefore, the *risky* portion of your portfolio should always be 3:1 equities/bonds.

However, your portfolio should be divided into a risky and risk-free portion. Risk-free means something where the principal is guaranteed by a CDIC-insured entity.

How much goes into risky and risk-free depends on your risk tolerance, age, funding needs etc. But the 3:1 ratio of the risky portion should be constant, no matter what.

(Source: "Investments", Bodie, Kane, & Marcus)

"Equities outperform bonds 3 out of 4 calendar years. Therefore, the *risky* portion of your portfolio should always be 3:1 equities/bonds."

Sounds like a pseudo-analytical rule of thumb.

Whether you'll see an equity premium if you buy stocks today is, of course, unknown. But you should. Despite what you've written there is a higher risk to equity investing, and that risk should be compensated by higher expected returns.

Return of principle, plus interest, is a right for bondholders. The only way they're not getting exactly what they expect is if the underlying company or government goes bankrupt and defaults. The risk of bankruptcy is quite low, practically non-existent for many bond issuers. Even if they do go bankrupt, the bondholder will still likely walk away with something, as they have a more senior position.

Equities generate their return through dividends and capital gains. These are entirely contingent on a company's performance. While the risk of bankruptcy is low, the risk of underperformance is significant.

The same is true aggregated accross the entire market. If the overall economy underperforms, only a small number of companies will default on their debt, while the vast majority of companies will underperform. Bonds remain safer than equities, and therefore should generate a lower expected return.

Any world where there isn't an equity premium is a world where equities are overpriced.

In the literature it's called the equity premium "puzzle" for a reason. Brad DeLong and Konstantin Magin in a 2009 Journal of Economic Perspectives survey piece argue "degree of risk aversion needed to support the existing equity return premium seems extremely high."

In other words, all these arguments about risk - the ones that I'm guessing are in the intro finance textbook Simon would like me to read - explain a *small* equity premium, but not an equity premium of the size that has existed historically.

Perhaps it would clarify things by stating the equity premium hypothesis in various ways.

1. The expected return on equities is just enough higher than the expected return on bonds to compensate shareholders for risk. Actual returns may vary. I don't have any problems with that hypothesis, but given the long investment horizons and degree of diversification that can be achieved by large investment funds, the concentration of wealth in the hands of extremely rich people who can afford to take risks and also have long investment horizons, and the emergence of modern financial products, it seems that this would imply an equity premium smaller than, say, the 4%+ postulated by DeLong and Magin. Perhaps if I read Simon's intro finance textbook I could figure out exactly what the premium would be.

#1 seems to imply that the equity premium continues, but is smaller than the historical one.

2. The expected return on equities is more than is required to compensate shareholders for risk, because market imperfections prevent the average investor from buying a fully diversified portfolio. Financial advisers give poor advice, take large overheads, corporate governance is flawed (see Mike's recent post on this), people exhibit loss aversion and other irrationalities, etc. In this story, yes, there might be a large equity premium, but it doesn't do the average investor any good, because they're paying 2.5% mutual fund fees plus money to a financial adviser plus not minimizing their tax bill plus taking money out of the market at precisely the wrong time.

I'd believe that story too.

3. By buying the market through an exchange traded fund, it is always possible to beat the return on government bonds over time period X. That's the version of it that I have problems with. If it was true, everyone would buy ETFs until the equity premium disappeared.

Tyronen: "Equities outperform bonds 3 out of 4 calendar years. Therefore, the *risky* portion of your portfolio should always be 3:1 equities/bonds."

I'm with Andrew F on this one.

Simon, on the psychology/confidence thing, this G&M piece gives an overview of the literature, with links to some fairly respectable papers. (I don't choose the title).

http://www.theglobeandmail.com/report-on-business/economy/economy-lab/frances-woolley/men-money-smart-or-just-more-confident/article2053726/.

Maybe loss aversion plays a role here. People hate losing $1 more than they like making $1. This might explain why people prefer the lower volatility and risk of bonds, beyond what the risk premium would suggest is reasonable.

I think your deduction in the last paragraph (which is standard in the literature) is incorrect, or at least premature/incomplete. There is another possibility. Just because present empirical results imply an implausibly high degree of risk aversion to explain movements in the equity premium, it does not mean the premium is not related to risk. It could simply mean that the risk models being used are not correct, or at least inadequate (attempting to connect the premium/risk to standard notions based on volatility, such as the CAPM and Value at Risk models).

Some recent work suggests a connection between the premium over bonds and P/E or P/D ratios (and other measures of price swings relative to a fundamental benchmark). As expected returns over bonds increase (due to expectations of earnings/dividend growth), prices rise until the price relative to some benchmark (E or D) is high enough to offset the excess return. This is a notion of risk which relates to the possibility of a reversal, or downside risk, which would be potentially sharper for a given small move in earnings the further the P/E ratio. These swings in prices relative to benchmarks represent an alternative notion of risk which links it to price swings (rather than price volatility) and seemingly can well explain the premium. When expected earnings growth becomes less optimistic, P/E ratios fall, the "downside risk" falls, and the premium of stocks over bonds becomes lower, or even non-existent. This GAP between the P/E ratio and its "benchmark" long run levels provides the offsetting risk to compensate for differences in returns.

The same model has been applied to the foreign exchange market to explain the premium puzzle there, relating the premium to the GAP between the exchange rate and its Purchasing Power Parity Level (a fundamental historical benchmark similar to P/E or P/D ratios). Frydman and Goldberg (2007) and there after have found (in my opinion) convincing empirical support for such a model of the premium.

A "physicist question": Understanding that I am speaking in averages over many investments and this may not be true in specific cases, what is wrong with this argument?

With some extra cash I wish to invest I have a choice: lend money to a company (ie buy bonds) or invest in the company (buy stock). Presumably the company would expect to borrow the money (sell me a bond), invest the money, produce more, make money, pay me back and _still make a profit_.

If I put $1000 into a corporate bond I get the interest back. If I invest $1000 in a company I get the "interest" _plus_ the profit. Hence an equity premium.

Chris J - "If I invest $1000 in a company I get the "interest" _plus_ the profit. Hence an equity premium."

Not quite, unless I've misunderstood your argument.

The problem with that reasoning is that there is no reason why would the company ever allow you to invest in stock. Any company that's reasonably confident of being able to pay the interest will do so.

So a company that issues stock has a non-trivial possibility of not being able to pay the interest back.

Which suggests that there should be an equity premium to compensate you for taking on that risk. The only question is: is that premium small or large?

An exchange traded fund can buy shares of every company in the market. Some will go up and some will go down but, on average, the expected return should be fairly stable - the idiosyncratic risk should get washed away. Market risk - the economy and the market going up and down - should also wash out over time. This is why I've been arguing that the equity premium will, going forward, be smaller than it has been in the past.

Would be happy to be wrong, of course.

It is not necessarily....: That's an interesting comment.

@Frances, something to consider is that even in the "equity" and "bond" camps there is diversity, namely that there are various schemes of bonds that deviate from the normal vanilla variety, that put them more into the equity camp. Likewise preferred shares and the like start encroaching more on bond-like behaviour. My point here is that there is more a continuum of equity and bond investments than two discrete entities, and comparing them as classes is difficult. By looking at returns on smaller tranches of bonds and equities you might be able to resolve a slope on a return graph that will point to what's actually going on.

I agree with some of the above sentiment. Put yourself in COO shoes and figure out why s/he would choose issuing more common stock instead of a bond.

Stock returns are correlated. You can have a huge basket of stocks that doesnt make it risk free as we just saw a few years ago or this week. Stocks are more volatile than bonds.

Second, lets assume independence as you do. Further let's assume that the number of indepent stocks is finite and the number of a independen bonds is of similar number. Individually stocks have more risk. You suggest that the aggregate risk equalizes. Certainly not, it takes many more uncorrelated stocks to equalize the risk. That's just on your terms of debate.

Now I go back to my claim that the 'basis set' of stocks is actually quite small.


It's was precisely the neglect of correlated risk that led to the massive MBS loses. Structured finance can only reduce uncorrelated risks!

Frances, I think one crucially important point is that returns on equities may be greater over extended periods of time, but they can be a great deal smaller than bond returns over very individually significant periods of time (eg the last decade.) A prudent individual investor does not want to risk being exposed to large equity losses immediately before and after retirement and hence is wise to have a mixed portfolio nothwithstanding any putative difference in very long run returns.

I'm not even sure about an equity premium at all. I would have made out like a bandit if I had purchased long Canada government bonds back in the late 1980s/early 90s before the Bank of Canada went on their anti inflation crusade.

"The problem with that reasoning is that there is no reason why would the company ever allow you to invest in stock. Any company that's reasonably confident of being able to pay the interest will do so."

Ah, but no lender will lend money to a company with no equity (at least not at any reasonable interest rate - try getting a shell company a loan without guarantees from its shareholders). So to be able to borrow, the company has to have some initial level of stock. Once it's out there, there's nothing a public company can do to keep me from investing in it. However, we do observe the behaviour you describe in other forms. For example, companies buying back shares rather than repaying debt (or even borrowing money to buy back shares). It's also the logic of the debt-financed cash take-over bids of the last decade. Of course, just because that's the logic, doesn't mean that the people doing it are right (although I'd like to think that people playing with billions of dollars aren't completely out to lunch).

Is information insensitivity a factor here? Most high rated bonds are dominated by the single variable of interest rates. Not so for blue chip equities. ETFs have helped equities here though.

An asset with higher information insensitivity should command a higher price due to being more liquid. Long run returns don't matter one iota if the market is not forward looking enough, which in general it isn't, because if it was it would have priced

Dang. A post I can actually make substantive comments on, but posted when I'm on vacation. I'll get back to this in a week.

Almost every bond default by a publicly held company leads to bankruptcy. In bankruptcy, equities and subordinated bonds are almost always wiped out, while bond debts to general creditors generally received double digit percentage repayments. Moreover, there are many companies that never default on bonds that still have almost all of their equity wiped out in the face of poor performance by the entity.

Leverage and subordination makes stocks inherently more prone to downside risk and more volatile than ordinary corporate bonds. At the very least, an apples to apples comparison requires that one compare equity to subordinated bonds.

One important reason that this risk can't simply be eliminated with a diversified portfolio is that equity value losses and bond defaults in different firms are not independent of each other. Companies don't default on bonds and have their equity wiped out one at a time. Instead, during booms all equities generate positive returns and no bonds default, while in recessions the entire market capitalization of equities falls in value (disproportionately relative to firm level profits and revenues because they are leveraged) and all of the bond defaults that are going to happen tend to happen at once. Increasingly, these correlations extend even across national boundaries. But for the California housing bust, Iceland might not have defaulted on its national debt.

To the extent that changes in equity value and bond default rates are not independent, the risk involved in a diversified portfolio is the same as the risk of investment in a single firm, equities are riskier than debt, and the equity premium is justified. For example, when the investment banks collapsed in the financial crisis, Fannie and Freddie tanked, AIG came into government ownership and two of the three American automobile manufacturers in the United States went bankrupt, equity owners and subordinated bondholders were wiped out pretty much across the board, but general unsecured bond creditors were almost entirely unscathed.

Put another way, it is mathematically legitimate to decompose risk into firm specific risk and marketwide systemic risk (or if you prefer a more complex model, firm specific, industry specific, and marketwide risk). To the extent that there is any marketwide systemic risk in that decomposition, you can't escape it with portfolio diversification. And, if the average investor by value is risk averse (something that is empirically true) then investors will willingly trade the ability to avoid a downside risk for more than the expected value of the downside risk avoided.

There are legitimate reasons for some investors to be risk averse. One reason to invest is to save for anticipated future expenses, like college tuition or retirement. This purpose restricts the ability of an investor to engage in market timing. A typical investor's objective is not to, "maximize my rate of return", but to be able to answer the question: "how much do I need to save to be confident that I have $X in 2016." An investor with the latter objective should be more concerned about downside risk than the potential for upside gain that equities have and bonds do not have to the same extent.

Moreover, many of the risks faced by bondholders (e.g. interest rate changes) are marketwide risks that cannot be escaped by diversification, and are risks that are faced by equityholders as well as bondholders, in addition to the risks which are unique to equities.

Tax motives are a worthwhile consideration as well. U.S. taxpayers have historically paid lower effective tax rates on gains from equities than on gains from bonds, have more control over tax timing, and can sometimes defer paying taxes forever due to the step up in basis of capital gains at death. I would not be surprised if a significant share of the preference for equities over debt attributed to an excessive risk premium is, in fact, a form of taxation premium (similar to that seen between municipal and corporate bonds). One way to measure this would be to compare rates of returns for subordinated bonds (which have similar risk of downside loss to equities empirically but very different tax treatment) to the rates of return on equity investments in the same companies.

@Frances - thank you.

ohwilleke,

Differential tax treatment may explain a demand side preference for equity over debt, but the preferential tax treatment for issuer of debt over bonds (other, obviously, than governments), provides an offsetting supply side preference (since interest on debt is generally tax deductible, while dividends are not - at least in the US and Canada, some jurisdictions do allow the deduction of dividends). Investors may well prefer equity income, but issuers don't. It's hard to tell whether one preference would dominate the other, but I would suggest that since a significant chunk of the world's investors are not entitled to the preferred tax treatment of equity (i.e., non-taxable entities such as pension plans, and non-residents, whose home countries, like Canada do not generally provide shareholder level recognition for foreign corporate-level tax (at least in respect of portfolio holdings)), the supply side would dominate.

Indeed, what this suggests is that, in the absence of an equity premium, the pre-tax return on equity should be considerably less than the return on corporate debt, since equity income is taxed less heavily than debt income at the investor level (and therefore investors would be willing to accept a lower return on capital in the form of income from equity) and since interest, but not dividends, are deductible (and therefore issuers are willing to pay interest at a higher rate than dividends). Somehow, I doubt anyone will observe that phenomenon for any reasonably lengthy time period.

Andrew Jackson: "long Canada government bonds back in the late 1980s/early 90s"

If every other form of tax increase is off the table, the US will have to levy an inflation tax to finance their debt. In the short term that would reduce interest rates, but once inflation was well-established, we'd see nominal rates rising to compensate for inflation risk.

That would impact the equity premium.

(It's sweetly ironic, isn't it, that the 1980s/early 1990s "bad" CPP investment policy - lending funds to the provinces at very long-term fixed interest rates - has provided great returns relative to some of the more recent "good" CPP investment policies?)

The suggestions that Frances "read an introductory finance textbook" and the specific reference to Bodie, Kane, and Marcus (one of the most widely used finance textbooks) are a bit ironic in view of the fact that Zvi Bodie, the senior author of that textbook, advises a course very much like the one Frances advocates above in his investment advice book aimed at the general public:

See here:
http://money.cnn.com/2009/09/16/retirement/Bodie_stock_allocation.moneymag/

See also his book http://www.zvibodie.com/Worry_Free_Investing, which is endorsed by two economists who have won Nobel honors for their work in finance.

To go back to the original questions which interest me rather a lot since I am a middle aged boomer who will depend on my investments to fund my retirement:
1) It seems likely that part of the historical equity premium reflected transitional effects. In the long run the stock market cannot grow faster than the (world) economy.
2) There can still be a small persistent equity premium to reward those willing to tolerate volatility but timing issues and survivorship bias make the calculation of that premium very difficult. On the other hand bond holders have to be compensated for inflation risk which might outweigh the volatility risk of shares.
3) Most investment advice seems to be based on time horizons that are too short. A middle class Canadian who starts saving at 30 should have an ultimate planning horizon that includes the possibility that they (or their spouse) might live to age 90. Given the risks that inflation pose for bond investors, it seems prudent to have significant portions of your holdings in diversified equities until you are into your 70s at least.

This thread has done a good job examining the Securities View of stocks vs. bonds; one which takes the viewpoint of an outside investor and considers each to be independent and posits a choice between them.

But there is a fundamental flaw in this view: stocks and bonds are not independent. It's all in the balance sheet.

Firms possess equity, the net capital owned by shareholders. Firms borrow money to invest and earn a return and expect that the return on their investment minus the interest and principal on the borrowed funds will be positive. This is straightforward management.

That return is credited to the shareholder's equity. Since equity is a small amount compared to assets, the return on equity is higher than the return on assets. This basic leverage.

The end result is that the return on equity is higher than the return on assets and higher than the return on debt by balance sheet structure.

The most controversial point I will make is my next one. In the limit, the value of a stock will trend to the value of its current equity plus its net present value of future returns. The discount rate is the return on equity, which I previously established to be higher than the expected return on debt instruments. So the expected return from holding a stock is the return on equity. By the structure of the corporate balance sheet, it is higher than the return on borrowed funds corporate borrowing would be irrational.

This is the the "Inside" or Business Ownership view of stockholding rather than the Securities or "Outside" view of stockholding.

Determinant

You have a typing error in your penultimate sentence.

I agree that expected returns on equity are higher than expected borrowing costs but it is not true to say that returns on equity must in the result be higher than borrowing costs. One of the wild cards in all of this is the fact that interest and equity returns are taxed differently. To some extent the decision by a company to access debt or equity funding is driven by the existence of pools of different investors with different tax circumstances.

Possibly Joe. But the Dividend Tax Credit is designed to recognize the existence of Corporate Income Tax. DTC posits that as the owners of the company, the actual tax burden on a dividend is the combination of the corporate and personal income tax. The theory is that the ownership of funds during a distribution of dividends does not change and the DTC compensates for the CIT already paid since dividends are not deductible for tax purposes.

The net result is that dividend income tax rates must include the CIT already imposed to compare apples to apples. To ignore CIT completely and argue that this is a tax distortion is debatable and in my opinion fallacious.

Determinant,

I agree that the existance of the DTC eliminates the distortion caused by CIT, but it only does so for shareholders who can claim the DTC. But, of course, a good chunk of the shares of Canadian corporations are owned by persons who can't claim a DTC (either non-taxables, such as RRSPs or Pension Plans, or non-residents).

This is why income trusts took such a beating when the government started taxing them as corporations and allowing unitholders to claim a dividend tax credit on certain distributions (I'm simplying greatly, the SIFT rules are a mess). If you were a Canadian resident taxable shareholder, you should have been generally indifferent (on an after-tax basis) between holding units in an income trust or a corporation (although there might be a cash-flow advantage to an income trust), but if you were a non-resident or a non-taxable, income trusts were infinitely superior on an after-tax basis, since those investors aren't able to claim an ITC to offset entity level tax.

Moreover, the DTC doesn't apply to income from foreign stocks. Although most Canadians have a strong home bias in investment, this will tend to discourage investment in foreign equity. Also, the DTC doesn't provide recognition for entity level tax on retained earnings (although to some extent that is offset by the preferred tax treatment on capital gains).

RRSP's and Pension Plans are engaged in the business investing deferred taxes. This is explicitly recognized in the literature. The fact that they pay a reduced rate at present is a tax expenditure and social policy tool, again both explicitly stated in government policy and in the literature.

I have in the past signed a T1213 form which allows you to claim a tax reduction at source against a monthly RRSP contribution. The CRA will happily allow you to defer your taxes with this form.

Second since foreign stocks have fully taxable dividends, just like fixed income interest payments, we are again back at square one. Of course stocks promise dividend growth so there is a simple reason for an equity premium.

Determinant,

But the problem with RRSPs and Pension Plans is that you NEVER get recognition of the corporate level tax, non-taxables can't claim a DTC, and the income that they distribute to heir holders/beneficiaries is not eligible to claim DTCs. You may well pay tax at a lower rate when you withdraw money, but that's a function of having a lower income, not a recognition of corporate level tax (i.e., interest income earned by an RRSP or pension plan gets taxed at the same low rate.

In any event, I'm not arguing that there isn't an equity premium, in the absence of taxes of course there is, I'm arguing that the effect of the tax system might tend to offset it by creating a willingness for issuer to pay higher (pre-tax) returns on debt than equity, and conversely a willingness on some investors to accept lower pre-tax returns on equity than debt.

Read Eric Falkenstein. I think he puts together a pretty good case that the equity premium is an illusion.

The comments to this entry are closed.

Search this site

  • Google

    WWW
    worthwhile.typepad.com
Blog powered by Typepad