Sticky Luck

 Three:

What Can We Expect?

The Market-Expected Return

Pension sponsors, Peter Lynch and many others “expect” an equity premium.  What do we mean by expectations?  The probability model.  Puzzles of the equity premium, versus the historical fact.  Alternative histories and investor insurance.  How the future could play out.  Risk-adjusted returns and market-expected returns.  The bond-equity spectrum.  What does the market-expected return say to us?

As if attacking Buffett’s and Lynch’s records were not enough, let me now question an important business assumption of corporate America.  Many US companies sponsor pension plans, and the assets and liabilities of even single plans can be in the billions of dollars.  Under accounting standards that have been in place since 1987, the sponsoring employer must make a “best estimate” of the long-term return it expects to earn on its pension assets.  It must also select a return it knows it could earn with a high degree of certainty – in other words, through investing in a portfolio of high-quality bonds.  (This return is known as the discount rate, and is used for measuring the plan liabilities.)  By reading the public accounts of major US companies, we therefore have a window into their thinking on the additional return they expect to achieve through the stock market, compared with the safer bond route.

Surveys of these assumptions are conducted every year.  A 2006 survey by Watson Wyatt shows the returns companies assumed for 2005.  Over 600 companies were surveyed out of the 2006 Fortune 1000.  Their average, long-term expected return on their pension assets was 8.3% per year.  Their average “safe” return (the discount rate) was 5.6% per year.   Of the pension assets, on average 63% were invested in equities, 31% in bonds and 6% in real estate and “other.”  If we assume the bonds were on average of high quality, their return would be similar to the discount rate.  Simple arithmetic then leads us to the conclusion that these surveyed companies are expecting, on average, an annual return of about 9.5% from their equity portfolios.  That return is almost 4% above the “safe” rate of return.  It is more than 5% about the “risk-free” rate of return that could be earned on ten-year Treasury notes as of the same date.

It is reasonable to conclude that these companies were expecting an annual 5% “reward” for taking the risk of equities, rather than the smooth path of Treasury bonds.  How can they justify this assumption?  The best available answer is history.  If we look at returns averaged over most of the 20^th Century, we see that equities have returned at least 4% per year more than Treasury bonds, and often much more.  Smaller companies’ equities have averaged something like another 2% more on top.  This spread between equity and bond yields is known as the “equity risk premium.”  The name implies that it is the reward investors receive for taking equity risk.  Can we expect such an equity risk premium in the future?  It appears that most large pension plan sponsors think so. 

Peter Lynch also has a view on this topic.  I recall watching him on a TV commercial around the turn of the Millennium.  Though I may not remember the exact words, I believe Lynch said that he was often asked what return could be expected from equities over the next century.  He said his answer was about the same return that equities had delivered over the 20^th Century – roughly 12% a year.   I cannot be certain he said 12% - it may have been 11% or even 13% - but the precise figure does not matter so much for my discussion, so I will assume it was 12%.

Expectations and Risk

What do Lynch and large plan sponsors mean when they talk about expected returns of 9%-12% per year?  As an investor I want to know just how likely “expected” means, and what caveats, if any, surround it.  I am not trying to be pedantic here.  I think there is a real confusion about the words “expect” and “risk” when it comes to investment, such that it leads to real misunderstanding about the nature of investment and the danger of big losses that were not imagined.  As I said earlier, most people understand that the stock market involves risk.  But which of the following is an accurate description of that risk?

1. The risk of fluctuating asset value such that, on any selected target date, the stocks may have underperformed bonds.  However, in the long term (undefined), it is accepted that the market will inevitably recover and provide the superior return expected.
2. The risk, even in the long term, of not achieving the same return as bonds. 

We should take a moment to consider how fundamentally different are these two definitions of risk.  The first might be called “fluctuation risk.”  In other words, asset values go up and down, and may be inconveniently down at the same time that we need or want to spend our money.  However, it is assumed that such a situation cannot last forever.  No US bear market has ever lasted more than a handful of years, and market recoveries have always eliminated the bear market shortfall.  If the investor can only hold on without the need to liquidate, he or she will eventually earn the expected return.  The second risk is more fundamental and, for many investors, difficult to imagine.  Yet it is highly plausible to the Japanese public who, since 1990, has seen a lead taken by bonds which to some seems almost unbridgeable, despite some pullback in the last few years.

The concept of “the long term” needs more definition for self-centered mortals.  To an individual investor, the risk choices are better expressed as:

1. If I stomach the fluctuations, I can enjoy a superior sum in retirement and, if desired, pass on to my heirs.
2. The market may under-perform throughout my lifetime.

Most investors accept that they will want to hold something safer in their final years, so an investment horizon will typically be no more than forty years, and rarely more than fifty.  So our critical question about risk and expected return might be phrased as follows:

If I define my expected return on equities as the sum of all possible returns I may earn, weighted by the probabilities of getting those returns, then will this expected return exceed the expected return from bonds during my investment horizon?

For those unfamiliar with it, I think the best way to illustrate probability weighting is to consider a single die where one of the white dots has been scratched off.  Let’s say that it’s a dot on the six, so we now count that side of the die as another five.  So we have a die with numbers one through four, and two fives.  We are asked to roll the die and we will be given an amount of dollars equal to the number that is thrown.  What should we pay to play this game?

An essential feature of the answer is the concept of the “expected value” of the game.  In this case it is simple to calculate:

1 x ($1 + $2+ $3 + $4)   +   2 x $5  =  $3.33

                            6

I chose a damaged die because it allows me to weight the $5 by a factor of two, compared with the single weighting of the other results.  This example therefore shows how probability-weighting of the various outcomes gives me the “expected pay-off” of this game.  If I played the game enough times, my pay-offs would average $3.33 per game.  So if I was charged less than this amount to play the game I would eventually make a profit, even though for periods of time I may find myself in a loss position.  On the other hand, if I paid more than this per game I would eventually lose money if I played for long enough, even though I may be lucky enough to stay ahead of the average for a lengthy period of time.

Expectations Drive Prices

My definition of expected return on equities is based on the same idea.  In theory, every possible outcome in the future is weighted by the likelihood of that outcome.  We can illustrate that concept with a new biotech company waiting for its one drug to be approved by the FDA.  We’ve estimated that, upon approval, the company stock is worth $100.  If the drug is not approved, the stock is probably worth about $20.  Based on our knowledge of the facts, we decide the chances of approval are about 60%.  Therefore we think the stock is worth:

($100 x 60%) + ($20 x 40%) = $68.

So if the market price is currently below $68, we should be buying the stock and, if it is above $68, we should be selling.  The $100, $20 and 60% inputs are all themselves based upon some other probability weightings.  This is the only way that an analyst can assign a scientific value to a stock.  Typically, the probabilities are based on historic information about other stocks.  For this one stock, if we purchase it at $60, we will either make $40 on the drug’s approval or lose $40 on its failure.  However, if we have 99 other companies with different drugs but the same $100/$20/60% inputs, and sixty of the 100 get approved, as estimated, we expect to earn a profit of $8 per stock.  What looks like a game of chance with a single stock becomes a calculated business risk when spread across many stocks, just as a casino can make a predictable profit from the individual luck of the punters.

In a very crude way, I believe Peter Lynch was applying historic information for his “probability weighting” that arrived at the 12% number.  There is only one century of the modern US stock market where we have had a regulatory environment and large-scale trading that is comparable to today.  Lynch “expects” 12% in the 21^st Century presumably because he views the 20^th Century as the only data-point that is relevant and available.  Of course, returns could well turn out to be higher or lower, but it seems convenient to have all their possible probability weightings average out at 12%.  After all, why should the 21^st Century necessarily be different from the 20^th Century from an investor’s perspective?  Large pension-plan sponsors appear to do something similar, but add into their weighting the bear market at the start of this century and perhaps add a little greater weighting for the other major bear markets of the 1930s and 1970s. 

Though neither Lynch nor the plan sponsors define their use of the term “expected” in the way I did, it is difficult to draw a useful conclusion from their statements in any other way.  Both know their estimates are subject to risk and significant error.  So if their definition is not the same, it would mean their expectations were either of a more or less “hoped for or optimistic” category, or else of an “even on a pessimistic scenario” category.  Neither would have any usefulness, as we have no sense just how optimistic or pessimistic they intend to be.  We are always forced back on the conclusion that the probability-weighted version of “expected” is the only one that gives us actionable advice.  It does not give us a measure of risk, which of course we will need, but it is a vital first step.

Equity Risk Premium Puzzles

If we accept that the market has an expected return of 9-12% per year, we are now faced with several new puzzles.  First, it seems to follow from this point of view that the only purpose of holding bonds as a long-term investment is to dampen temporary fluctuations because, given sufficient time, we will earn a return of 4-7% more than bonds.  We remain concerned about the length of those fluctuations given our limited life spans but, nevertheless, 4-7% per year would seem a very high price to pay for smoothing out returns. 

Second, we know that dividend yields are typically under 2% so the rest of the equity return must come from capital gains.  So the equity market must grow in size by 7-10%, or typically 4-7% above inflation.  We know that 4% growth in the US economy is considered fast, and a significant portion of that is the result of an increasing population.  So what is this additional corporate growth feeding off?  We don’t expect it to steadily consume the government’s portion of the economy.  Growth from overseas is one source but, in today’s global economy, the greater worry of most Americans is that foreign businesses will take a greater share of our economy than we will of theirs, not the other way round.  Of course, we can invest in foreign equities, if we believe that their 21^st century growth will replicate America’s 20^th Century success.

But that observation leads us to a third, more complex puzzle:  Which equities are we talking about?  A large pension fund will typically hold mainly US corporations (which admittedly having their own foreign investments) with a profile that would not look greatly different from a US broad-market index fund.   Lynch was presumably referring to the market as a whole.  But, as discussed in the last chapter, the equity market is far from uniform.  In that chapter we examined the big difference between the returns on value and growth stocks.  I will take a slightly different approach to this criterion.  It is no coincidence that both categories’ names have an appealing ring to them – aren’t value and growth the two things we are looking for?  But let’s take a more objective view of the distinction.  “Book to price” ratio doesn’t tell the complete story.  If book value is to mean anything useful, it means the worth of the business’s debt-free assets.  If the market has placed a low capitalization on these assets, it is because it doesn’t see much business growth potential.  Yet debt-free assets still produce income or, if they don’t, could be sold off (under accounting standards, at book value) and the resulting capital used to produce income.   A less enticing, but more accurate description of value stocks would be “income” stocks.  Their earnings may not be growing, but right now those earnings are quite high relative to the market price.

Balancing Expectations Through Prices

One of the market’s ironies is the out-performance of such income stocks against growth stocks, at least since 1968.  Since all stocks have done well, this must be because of the earlier underestimation of income stocks rather than the overestimation of their growth counterparts.  In essence, income stocks have grown along with growth stocks, but have also added more cash income to the total return.  However, whatever has happened unexpectedly in the past does not tell us what to expect about the future.  The out-performing income stocks have been duly rewarded with price appreciation.  That appreciated price will reflect anticipated growth in future years.  The market may have been proven wrong in the past, but it will have accounted for that past error in today’s assessment.  It may have over-priced an income stock because it is now allowing too much for the history lesson. 

A premise of efficient-market theory is that, if the market were to leave some glaring error through not accounting for a prior lesson, there are millions of lines of computer code which in seconds would arrange portfolios to counter that error.  I am not using the term arbitrage here, where quick advantage is taken to make cash out of a price mis-match.  I am talking about the forecasters who conclude that certain stock classes or maneuvers will deliver steadily higher returns over time and who therefore overweight their portfolios accordingly.  We already know that many investors believe income stocks will always prevail over growth, and they have accordingly bid up these stocks’ prices.  The stronger the historical phenomenon, the stronger the believers and the greater the market “lesson learned” through a price adjustment.  It is the price adjustment that negates the additional expected return for the future.

Income stocks are typically well-established and generate lots of cash, but are generally not expected to achieve major “breakthroughs” in the scale of their business – except perhaps by acquisition.  Growth companies in contrast tend to rely upon change, pumping not only any available cash but additional capital into less tried and tested concepts.  Naturally, the second type of company carries more risk, and therefore has a much greater “concentration” of equity (versus income) as today’s investor would see it.  On the other hand, “income” companies can appear to have many features of a bond – such as utility stocks, real-estate investment trusts, and even tobacco companies.  We could make a related distinction between developed and emerging market equity markets.  Emerging markets offer more growth opportunities, but the less certainty about income because of greater economic or political risks.  Again, they have greater equity concentration.

In fact, if we try to “score” all equities by their “income” and “growth” qualities, we find we have a continuous spectrum all the way from the most speculative growth opportunity to the relatively safe source of shareholder dividend income that becomes closer to a corporate bond.  This phenomenon is well known, and is partly evidenced by a widely used measure known as “beta,” which is the relative movement of an individual stock to the market as a whole.  A high-beta stock will tend to move with greater volatility that the market; a low-beta stock with lesser volatility.  Hybrid securities such as preferred shares and convertibles provide a further “bridge” between equities and bonds.  Bonds themselves offer a similar spectrum, from “junk” with high yields because their credit-standing is closer to equity, all the way to virtually risk-free (in monetary terms) government debt.  In other words, if we consider any theoretical combination of income and growth, with an associated risk factor attached to both, we can find some quoted security somewhere which will match it quite closely.

An immediate question now pops out.  If equity has a higher expected return that debt, it would seem to follow that stocks with a higher equity concentration (e.g. high-beta stocks or growth stocks) have a higher expected return than stocks with a lower equity content (e.g. low-beta stocks or income stocks).   In other words, as reward for accepting their volatility, in the end we could expect superior performance.   Is there any evidence to support this conclusion, and does it even seem plausible?  

We have already discussed the historic out-performance of “value” (i.e. income) versus growth, which contradicts the conclusion.  A study by Fama and French covering the period 1963-1990 also showed no support for it.  It would also not seem like a conclusion that fits so comfortably with the traditional appeal of the equity-risk premium.  The attraction of equity has taken on a wholesome “share the value of human progress” aspect to it.  But layering on more and more risk starts to sound like risk of the speculative, Las-Vegas kind.  We are reminded of those who jumped into Internet stocks with both feet in the late 1990s, and who find themselves underwater some eight years later.

Junk bonds add to the doubts.  A large portion of investors avoid them, accepting that any additional yield earned would be fully offset by the risk of default.  And yet that risk is of a similar quality to equity risk.  If the equity remains healthy, the superior yield is retained.  If junk fails, the equity does also.  If investors see the probability-weighted return on junk as no better than government debt, why should there be a probability-weighted advantage of equities over bonds?  With equities, we have a good chance of earning more than bonds, but we have also a good chance of earning less.

So Why the 200-Year-Old Fact?

Yet the equity-risk premium is a fact, earned cumulatively over the better part of two centuries.  There seems to be plenty of data from which to draw a conclusion from and, like Lynch, we don’t have any reliable and directly comparable data that would point in another direction.  Coming back to my earlier point about the “wholesome” nature of equity investment, this greater return also seems to reflect, at least when measured in whole decades, the inexorable progress of Mankind towards a healthier, richer lifestyle.  We may experience a set-back at some point in the future for lengthy periods but, because invention is always additive, eventually we will reap the benefits when the business cycle does another rotation.

The problems with this analysis are that first, it looks backwards to what actually happened, rather than to what else might easily have happened, and second, in the context of what might have happened, we are reviewing a very short timeframe indeed.

A common pursuit of historians is to examine the root causes of great tragedies such as the two World Wars.  The American experience of these wars is not predominantly a tragedy, given the relatively small proportion of fighting forces (compared with the other main players) the very light civilian casualties, and the immediate and sustained economic benefits.  Yet, globally, from the fighting and resulting disease and famine, each war ended or badly damaged over one hundred million lives.  Economies collapsed, and a large portion of the world turned socialist with bleak outlooks for free-market capitalism.  A more speculative historical pursuit is to consider what might have happened, were it not for some small chance of fate.  Let me give two examples, while disclaiming any historical authority on the details.

• In 1939, Neville Chamberlain resigned the British Premiership and Winston Churchill was asked by King George VI to form a government.  Both Chamberlain and the King would have preferred Lord Halifax to have taken the position and, had Halifax been ambitious enough for it, he would probably have prevailed.  Halifax would almost certainly have sought accommodation with Hitler, and it appears that Hitler would have granted it.  Given Britain’s fear of a repeat of the 1914-18 conflict with its attendant slaughter, the country as a whole would not have pushed back against such leadership, especially as Nazi brutality was barely understood by most people at the time.  There would have been insufficient justification for a war in Europe for the USA, and the Nazis could well have dominated the continent indefinitely, eventually overrunning the Soviet Union which would not have had crucial Allied materiel donated to them.  There would have been no US nuclear deterrent.  In fact, another likely event would have been Nazi Germany’s success in developing the atomic bomb first and making Nazism virtually unstoppable, whether or not a full-scale world war actually took place.  In fact, it remained a puzzle to Albert Einstein – without whose initiative there would have been no Hiroshima bomb – that his former colleague, Werner Heisenberg (of the Uncertainty Principle) has not succeeded in Germany’s nuclear weapons project.  It now seems more than likely that Heisenberg was instrumental in delaying Germany’s essential research.
• During the Cuban Missile Crisis, Kennedy may have accepted military advice for an invasion of Cuba.  It is now known that the local Soviet military had standing orders to launch a nuclear strike on the USA in that event.  Given the likely US response and consequent Soviet counter-response, the world today would have been an entirely different place.

Looking back, there is a tendency to see today’s prosperous, free markets as inevitable.  In fact, there were many pivotal points that might have destroyed them completely, had a chance event intervened.  Churchill, Roosevelt and Stalin each personally made critical differences to the outcome of WWII which, under other leadership, would likely have been very different.  Under Stalin, Mao and others, capital markets were actually destroyed in other countries.  Investors who lost all would never recover, even with the eventually return of the free market.  During the US Depression, there was an appetite for Mussolini-style leadership that is unthinkable today, though Roosevelt refused to take advantage of such popular sentiment.  The world we have today was one possible outcome from all the potential worlds faced by earlier generations and, for investors, a surprisingly good one following some exceptionally lucky breaks for humanity.

Bonds as Insurance Policies

During much of the 20^th Century, long Treasury yields were in the 2-4% range, reflecting market expectations during what was generally a less inflationary environment.   We typically conclude that such expectations proved too pessimistic.  But if we characterize yesterday’s bond holders as pessimistic, we would also have to conclude that anyone who voluntarily purchases insurance, and does not end up making a substantial claim against that insurance, is similarly pessimistic.  The purchase of full auto insurance is a fairly good analogy, as most purchasers could financially survive the self-inflicted loss of their vehicle more than they could a sustained market collapse.  This insurance represents the buyer’s willingness to accept slightly less assets (after paying for the insurance) in place of the small chance of large drop in assets (the cost of auto replacement), even though – due to the need of insurance companies to add a profit loading – the expected cost of accidents would be lower if we drove without full insurance. 

Accepting a lower return to avoid large potential loss is a rational investment strategy, whether or not that loss occurs.  Only chance events will determine whether or not it is the more successful strategy.  Moreover, given the peculiar nature of the chance events, such as Lord Halifax’s state of mind, or Churchill’s near-fatal traffic accident when visiting New York in 1931, or Roosevelt’s dangerously high blood pressure which nevertheless allowed him to survive until 1945, there is no scientific way to measure such probabilities looking to the future.  The only available objective probabilities are those which the market implies to us.

Many people accept the qualitative nature of this argument, but reject the conclusion that the world’s major free stock markets could have earned a premium as high as 6% per year through the chance avoidance of severe, negative events.  To counter that position, we need to expand our case to include the great technological leaps forward that actually did take place, but were not expected to happen, or to have such an impact upon productivity.  The telephone, mass production techniques, the affordable automobile, passenger aircraft, radio, electronic computers and the Internet all advanced productivity much further than was commonly grasped, even well after their invention.  It is easy to be wise in retrospect, but none of this appeared inevitable at the time.  World capacity for the automobile was originally estimated at one million, as this was the estimated number of men who could be trained as chauffeurs.  What impact would the Internet have had if the web browser had proven too unreliable?  What if microprocessors had proven much too unstable?  The PC proved disappointing at raising productivity during the first decade of its life, and few people could reasonably conclude at the time that a warming-up period was needed.

If the Future Takes a Different Tack

We know there will be inventions in the future, but can we be sure they will have a similar effect upon productivity for the benefit of Western stock markets?  Might they instead actually encourage more free time and a reduction in spending as “quality of life” is measured differently?  Might equities actually suffer, even as people get closer to their personal goals through means other than additional consumption?  Many would argue that Japan went through such a transformation in the 1990s, as technology reduced the numbers of working hours needed to sustain a comfortable way of life.  There is some evidence that France has chosen such a direction.  As the US workforce ages, its values may not be immune to such a trend.  Even if technology continues to provide “break-though” changes to the workplace – and this is never guaranteed – we may choose not to translate them into greater GDP per capita.

Or they may be absorbed by non-wealth-creating activity.  America’s experience of war has been benign in terms of economic impact, but that is largely because it has generally harnessed excess capacity without seriously degrading peacetime production.  The First World War was a different story for its European participants.  Moreover, the culprit does not have to be total war.  Major changes in security requirements are activities that give rise to no meaningful net wealth, and restrictions on commerce are almost always a negative factor for all affected parties.  The West may in future view the Cold War, and perhaps the peaceful but limited decade that followed, as a period of relative stability, with its comfortable protection of home markets.  The Western markets, with their track record of 10%+ annual returns during the “American Century,” may be destined to decline, while Asian markets have no such track record to project forward.

Such ideas are pure speculation, but illustrate that a repeat of one century’s experience is not just unsure, but is instead only one of a vast number of scenarios.  Most human progress in the West was halted in the 14^th and 15^th centuries due to disease, showing that such progress is not inevitable if the challenges – now including avian flu and climate change – are severe enough.   These do not need to be doomsday scenarios, but ones where return on capital is gradually eroded by unexpected financial burdens.  Capital may be granted a shrinking share of the profitability pie, particularly if the Internet world reduces the value of large brands and individual talent can quickly steal markets away from the traditional combination of passive shareholders appointing executive management.  In such a world, all the spoils would go to the private entrepreneur, not the stock market.  How long will it take the Nikkei, currently around 17,000, to return to its peak of 40,000 some sixteen years ago?  Might the Nasdaq follow the same timeline?  Why should we imagine that Western markets are immune to such histories?

Many would accept that future equity returns will be lower but still think it unreasonable not to give equities at least a few percentage points advantage over Treasuries.  This position itself creates interesting questions.  If we are rejecting the returns of 11%+ achieved over the 20^th Century, where is the logic of settling on a 7-8% return?  I am aware that there are economic models which have been fashioned to support such return, many with sophisticated inputs.  But the end result is still driven by big assumptions about the way the future will unfold – economically, technologically, politically and naturally (climate, disease, natural disasters, etc.)  Moreover, no model’s sophistication can match the colossal size and complexity of real market forces, as they seek to balance the expectations of millions of investors and speculators every second.  Such models start to look like maps of the night sky drawn by an ancient priest who believed they help him understand the universe.

In fact, the modeled equity-risk premium is typically nothing more that the assumption that equity investors must be compensated with higher expected returns for taking on additional risk.  As I mentioned earlier, there is no clear evidence to justify this assumption.  And even if there were evidence that an equity-risk premium was inevitable in the past, could it be reversed over the 21^st Century?  To address that issue, we will look a bit deeper into the intuitive appeal of the assumption, and then consider the role of sticky luck on the landscape.

The Expected Equity Premium?

We know that, over time, equities have fluctuate in value to a scary degree, and sometimes wildly enough – as in the early 1930s and 1970s – to ruin many investors.  It is natural to concluded that the more intrepid investor should be rewarded for intestinal fortitude.  Second, surveys tend to suggest that the typical investor tends to be risk adverse, and therefore would be tempted into equities only through a premium yield.  Third, businesses would not borrow money unless they expected to earn a superior return.  Fourth, many investors have fixed obligations, such as insurance companies, so they will tend to bid up the price of the safer assets like bonds that they are required to hold – in other words, Treasuries have a scarcity value which reduces their expected yield.

None of these arguments stands up to much scrutiny.  It is insufficient to say that equity investors should be rewarded for taking risk – there has to be a market mechanism that delivers this reward.  As we have seen, investors in the late 1990s who took the greatest risk – through Internet stocks – have suffered the greatest.  Simply put, the market does not deliver on what “ought to be” unless we can find a behavioral basis.  The other three arguments attempt that.

It is true that experiments show people often fear losses more than they desire gain.  But there is a difference between losses that people can roughly quantify, such as auto or home loss, and the theoretical risk of stock market losses.  There is a strong belief, encouraged by many financial advisors, that market losses are essential “paper” losses which will eventually be recovered if the securities are held for long enough – in other words, the concept of the higher expected return in the long term. The investor is simply experiencing market fluctuations, not real losses.  This comforting thought is sufficient for people to get into the market, especially as it is rising steadily and they think they will “miss out.”  It is interesting that, when the losses have occurred, many investors, unable to maintain the mantra of “paper losses,” then choose to sell, later buying back in a higher market. Their risk-adverse behavior occurs at the worst time for their long-term performance.  But is can also be seem as a rational insurance cost against permanently lower markets, which start to seem like a reality to many people when things are at their worst.

Major sectors of the economy, such as gaming, suggest that people will actually accept a lower expected return, even a negative one, for the chance of big winnings.  In the financial world, nowhere is this better illustrated than in currency speculation.  Here, the expected return is zero because, by definition, trading currencies is a zero-sum game for all those involved in the transactions.  Yet technology has made currency speculation a vast new market for retail investors – a topic I will explore in later chapters.  The sheer scale of speculation, as opposed to the limited currency trading for international settlements, also illustrates the fact that “fixed obligation” investment is swamped by purely voluntary capital flows.  In the case of interest rates, the statutory requirements of insurance companies and the like are small indeed compared to the derivatives and hedge funds that “bet” on their direction.  In fact, the existence of derivatives squashes the “scarcity value” argument, as instruments that perform close enough to Treasuries can, for all practical investment purposes, be synthesized indefinitely.

Moreover, once we go beyond investors who have to match fixed-dollar obligations, reducing risk depends upon the investor’s personal circumstances.  The effect of inflation upon the real value of investments was particularly damaging in the 1970s.  We mentioned earlier that the Dow Industrials passed through the 1000 marker both in 1966 and 1982.  However, this can hardly be described as “flat” performance – price inflation eroded the value of the dollar by over two-thirds during that period.  Even with general price inflation relatively tame at the current time, healthcare inflation is much higher and a more important element for retirees.   Thus bonds are never risk-free in a “real money” sense.  Their value will decrease if inflation increases unexpectedly, and increase if inflation decreases unexpectedly.  They are therefore another “bet” on the future, and are priced by the market accordingly.  Though I continue to refer to the “risk-free” rate of return, lack or risk is limited to the return measured in a single currency. 

As for the borrowing of companies in order to produce greater results for their shareholders, this does, in fact, turn out to be an argument more in favor of a zero equity risk premium.  At any given time, companies have borrowed all the money they believe could generate a greater return to shareholders.  The fact they have ceased borrowing tells us that they no longer see further “risk adjusted” equity returns that are superior to borrowing costs.  The value of prior borrowing has already been neutralized by appropriate share price adjustment.  And this analysis leaves aside the natural bias of public companies to borrow.  Success is rewarded through higher leveraging of management’s stock options.  Failure cannot be punished to a comparable degree.

When we begin with today’s known fact of a successful free-market world, then it follows inevitably that equities have done better than people in the past would have expected, on a probability-weighted basis.  Many of those people believed socialism had a strong chance of taking hold of the world, or at least permanently in major parts of the world, and would prove particularly unkind to the investor.  Or that the tension between political ideologies would have more seriously inhibited world trade, or that the waste of blood and treasure in warfare would have sapped more economic growth.   Compare such fears with the experience of anyone under the age of forty whose only daily exposure has been the long bull market of the 1980s and 1990s, the relatively contained “technology bubble” and the resumption of capital growth through the time of writing.  For the individual human mind, a period of twenty-five years of day-to-day market experience is almost indelible, unless preceded or succeeded by similar periods of opposite experience.  Therefore investors not yet in middle-age find it difficult to imagine a bear market lasting five or more years.  Investors in their fifties and sixties recall the sickening combined effects of falling nominal markets and rampant inflation, massively eroding the real value of equities for the better part of two decades.  Investors in their eighties can never forget the despair of the 1930s, either first hand as children or through the views of their parents.  It is the universe of these investors, bulls and bears alike, which determines the expected return on securities.

Sticky Luck of the Mind

Each of these investor generations are potentially victims or unwitting beneficiaries of sticky luck.  By that, I mean perceiving a pattern in prior markets, perhaps holding true over long periods of time, that shape their expectations about the future.  Those whose investment experience is shaped by the 1930s and 1970s tend to see constant signs of bear markets emerging, and can find in historic data plenty of consistencies to make their point.  Those shaped by the 1980s and later see signs of recovery with each market dip, and also can point to repeated patterns that back their prognosis.  What is the poor retail investor to conclude when faced with both sets of arguments from famous advisors, backed by convincing charts and revealed trend-lines, pointing in exactly opposite directions?  Which group of advisors is mistaken?

According to efficient-market theory, both groups are misled.  No matter how appealing and consistent any market pattern may appear, it can tell us nothing about the expected returns of the future.  That is not to say that there is no way to determine whether or not market bubbles have occurred, but data patterns alone cannot provide that answer

I will broaden the market we are considering to include all freely traded securities, which includes all such bonds.   As I have mentioned, all investors at all times have the opportunity to go in and out of bonds at prevailing prices and yields.  We have also discussed how all investors wish to maximize their returns, and that so few are constrained to purchased a limited supply of any one type of security (e.g. Treasuries) that such constrained groups will have negligible impact upon market equilibrium.  So, if at any one point in time, market prices fully reflect all known information, how would it be possible for one type of security to have a higher expected yield, as determined by the market’s perspective, than any other? 

Let’s go back to the definition of expected return:  it is the sum of all possible returns I may earn, weighted by the probabilities of getting those returns.  How do we get any sense of those probabilities?  I have explained how the financial industry has generally used historical performance to measure the probabilities for the future.  However, under efficient-market theory, the relevant history we use is no more than a random event itself, and a seductive but wholly unreliable source of knowledge about the future.  What is there left, then?  We have only today’s market - in fact, this moment’s market.  That market contains within its data the “votes” of every investor as to what he or she imagines the future to deliver.  Never mind where they get their information from.  That information may be based on a fallacy, but their votes are a fact.  These investors are sincere in their votes as they have staked their money on them – the only way to register a vote in the market.  The future direction of the market will also depend upon the future voting pattern of these same investors, and the people they gradually pass their money and ideas on to.

Starting from Certainty

The one expected return we can know with certainty is the Treasury yield.  As it is regarded as risk-free, it requires no further adjustment.  All other bonds, which carry an element of default risk, also carry a greater arithmetic yield which is accepted as compensation for the addition risk.  It is therefore not difficult to conclude that the additional yield, assuming the bond survives to maturity, is the market’s expected loss from default on a large portfolio of such bonds.  In other words, the market’s expected yield on a portfolio of junk bonds would be the same as on a portfolio of Treasuries.  When the yield spread narrows between junk and Treasuries, the market had changed its expectations of that risk – no matter how irrational some commentators may judge the market to be.  This way of viewing the bond market is well-established under efficient-market theory.   We say that the risk-adjusted return on all bonds, using the market’s collective opinion to measure the cost of risk, is always the same, for any given period over which that return is earned.

At this point, behaviorists cry foul.  If we allow the market to define risk, then risk measurement is itself subject to irrationality.  To say that the risk-adjusted return is the same for different securities is nothing but a way of defining what we mean by risk-adjustment.  I would agree.  To many observers, the market may seem irrational in the way it measures risk.  However, each observer has his or her own way of measuring risk.  The market represents the consensus of all observers who also happen to be voting with the sincerity of their risk capital.  The market also consists of buyers and sellers with very different viewpoints.  We can therefore be sure that, at any given time, the buyers of riskier bonds believe that the market is overestimating risk and the sellers believe that it is underestimating risk.  Do we have any observers who always reliably forecast who is right and who is wrong?  To avoid sticky luck, we need observers who have a long and consistent track record predicting in both directions.  We have no evidence they have ever existed, and efficient-market logic tells us that, should they exist, they must possess an amazing, clairvoyant power that transcends the combined power of the world’s forecasting tools.

Now I introduce another important modification of efficient-market theory.  The market has its own timeframe over which it measures risk.  That timeframe is a combination of all buyers’ and sellers’ timeframes.  We will discuss timeframes is greater detail later, but for the moment simply state that they are extremely difficult to measure and also are likely to vary greatly.  Therefore it is quite possible for investors to have the same long-term view of risk as a critical observer, but to make their investment decisions based on a shorter timeframe. The observer may therefore be proven right in the long term, but that does not make the market wrong from its own, shorter-term perspective.

Market-Expected Returns

Instead of discussing risk-adjusted returns, I want to focus on what I call the market-expected return.  I define it as the weighted average of all buyers’ and sellers’ expected returns, where the weights are the sizes of the buys and sells.  Each investors’ expected return is assumed to be based on the type of calculation given earlier for the biotech company.  I realize that many investors do not approach investment decisions this way, and even more do not think explicitly in terms of expected returns.  Later, I will explain why I do not think these facts weaken the approach I am taking.  For the moment, I would like the reader to accept that, in some form, every investor has an explicit or implicit notion of expected return when choosing an investment.  If this were not the case then, assuming the goal of investors is to maximize investment returns, there would be no basis for choosing one investment over another.  Moreover, we must have some “absolute” notion of expected return, rather than just a relative one, as the risk-free rate of return is a known benchmark to or from which we would add or subtract a relative return.

Just as we can say that the “market risk-adjusted” rate of return on all bonds is the same, I believe we can also say that the market-expected return of all bonds is the same.  This becomes easier to accept if we try to imagine how some riskier bond might have a market-expected return greater or less than that of a Treasury bond.  Remember that expected returns are already adjusted for the investors’ perceived risk, so a greater expected return would mean that the consensus view of the buyers and sellers of the riskier bond is significantly different from the consensus view of the buyers and sellers of the Treasury bond.

Let us assume that the difference in risk between the two bonds is quite small.  The group of potential buyers and sellers of the two bonds will be largely the same investors, as both bonds will be close enough to their desired risk profile.  Moreover, sellers of one of the bonds are unlikely to buy the other bond, as transaction costs would be high relative to the risk difference.  The potential sellers of both bonds would tend to have the same risk preferences, and the potential buyers would tend to have the same risk preferences also.  If one of the bonds had a higher expected return among these investors, after allowing for perceived risk, it would be perceived as having greater value, and this will result in potential buyers of the other bond switching to actually buy the higher-return bond.  This activity would bid up the price of that bond, and the process should continue until the new price eliminated the excess return.  Hence the riskier bond would have the same consensus expected return as the Treasury bond.

The term “consensus” is trivial for the Treasury bond as it has a single, arithmetic return which is defined as risk-free and therefore requires no risk adjustment.  The term for the riskier bond describes the weighted-average of the buyers’ and sellers’ expected returns.  The sellers will view the return as too low for the perceived risk, and the buyers will view the return as relative good value for the perceived risk.  The differences may seem unimportant for bonds that have such close profiles but, just as with the different groups of investors, we can keep inching our way into riskier and riskier bonds.  Just as we assume many investors, so we assume many different risk levels for bonds so that each step to the next level of risk is a small one.  With each step, the potential buyers and sellers are essentially the same, so their consensus expected return is the same for the next-more-risky bond as the last-less-risky bond.

What Consensus?

Of course, with every step, the divergence of expectations widens, so the term “consensus” is purely technical – it is the averaging of increasingly divergent opinions.  Since the average doesn’t change, it remains equal to the risk-free rate of return, the point at which we started.  Later we will look more closely at how this process works when prices are in constant motion.  Right now, I think this explanation has reasonable intuitive appeal.  If we start from the point that prices have already adjusted to reflect supply and demand, then the expectations of those deciding to sell always balances the expectations of those deciding to buy.  The levels of those expectations will drive the volume of trading but, since buying always equals selling, the average of those levels will not change.  Buyers believe the security is superior to alternatives, and sellers believe it is inferior to alternatives – even if that alternative is cash.  The freely adjusting market price is the critical mechanism to balance all consensus expected returns.

The assumption behind the above process is that all bond investors have a free choice between one type of bond and another which is close in quality.  I have mentioned earlier the argument used sometimes that some bond investors are restricted to a precise bond type, particularly Treasuries.  I have argued in response that any such restricted investors will be small enough to have a negligible impact upon the market.  However, it is worth adding here that my assumption is quite flexible.  Even if all Treasuries were purchased by investors who had no alternative, then very high-quality corporate bonds would be our benchmark, their yield being reduced by their expected defaults to give a notional risk-free rate.  If high-quality corporates are restricted, we would simply go further up the debt ladder for our adjustment.  In fact, the defensible relationship between Treasury and corporate yields relative to transaction costs, duration, default and redemption risk is further evidence that there are no unique market conditions inhibiting choice at this end of the market.  Yield-spreads have fluctuated over time, reflecting market perception of future risk-spreads.  Even the colossal purchases of Treasuries by China and Japan has failed to create any unusual spreads, illustrating the power of derivatives, hedge funds and the rest of today’s vast financial sophistication to wash out the imprint of any one class of investor.

It is not difficult to see how “market risk-adjusted returns” and “market-expected returns” are really the same concept if we are looking into the future.  If we are averaging all investors’ expectations, we are averaging out all their perceptions of risk, because the risk probabilities are “built in” to the expectations.  The problem with current scientific discussions about risk adjustment is that they are focused on past data, so there is no way to measure the associated risk.  Once an event is passed, probabilities become certainties.  Forward-looking, market risk-adjustment is the right concept, but we have no simple way to define it. 

A Journey Across the Security Spectrum

The step-by-step process through the risk spectrum of bonds will take us all the way to junk or “below investment grade.”   Just as some investors expect higher returns by taking on more risk, others expect such exposure to deliver lower returns, either through defaults of loss of market confidence.  There is no point in the entire spectrum of fixed-interest where there would not be another bond sufficiently close in risk profile to any other chosen bond that investors would not consider the two as broadly equivalent.  There is therefore no “discontinuity” where it would be reasonable to conclude that the market-expected return should actually change and not remain constant.   Yet, as we move into the junk portion of the spectrum, we are starting to take on substantial equity risk.  In fact, such risk is present in all corporate bonds, but an investment-grade designation provides a reasonably clear limitation to such risk.  Junk involves full equity exposure, converting equity’s indeterminate yield into a fixed one, subject to avoiding default. 

So when we add equities to the mix of securities for which we are measuring the market-expected return, there is no special qualitative change.  Like bonds, equities are streams of future income to which levels of risk are attached.  Some income equities act much like bonds, and there is a complete spectrum up to the most speculative growth stock.   The market has set the price at which any holder of any security can switch to any other.  Let us now consider all securities which have the same expected term (i.e. duration, or time-length) of investment.  For the moment we will not try to get too precise about expected term, and just use some working assumption.  We might, for example, assume that most investors expect to hold each equity an average of one year, and that a one-year bond therefore has a similar duration.  By definition again, the risk-adjusted return on all such securities, based on the market’s collective opinion of measuring the cost of risk, no matter how irrational we imagine that collective opinion to be, is the same.  Otherwise, investment would flow to wherever the risk-adjusted return, as measured by the market, is higher. 

Since the market has within seconds fully reflected all known information into its pricing, there is no way to “beat the market” to the calculation of the market risk-adjusted return, other than with the gut-feel, rapid trading open to a few highly talented professionals, as discussed earlier.   Once the market has settled on the price of each security that achieves the equality of its risk-adjusted return with every other security, then only new information will cause it to change those prices.  That new information will be new to all investors.

News Re-Visited

And let us to clear again about the nature of new information.  Suppose everybody is confidently expecting the Federal Reserve to raise interest rates by a quarter-percent at its next meeting.  The fact that the Fed actually made this change is only new to the extent that it converts a very high probability into a certainty.   This “news” adds very little “new information” and therefore would move the market very little.  Therefore an investor who believed that the Fed raising rates would be a bearish sign to the market would likely be disappointed, as the market had factored that likelihood into its pricing as everybody formed the same opinion in advance.  So when we talk about news we mean either:

• Something unexpected or unanticipated, or
• Something less than a certainty becoming a certainty, the news being the “gap” between the two.

Given this definition of news it is clear that, short of illegal insider trading, it would be at least extremely difficult to have consistently superior knowledge of future news than the market as a whole.  In fact, in his book, “The Wisdom of Crowds,” James Surowiecki argues why, under normal conditions (i.e. absent bubbles) the consensus opinion of crowds is very likely to be superior to that of even the subject-matter experts among the crowd.  When it comes to actually predicting the future, even subject-matter experts usually have no special insight.  For example, an art expert may be able to predict the auction price of a painting today, but that gives him or her no special edge in predicting the future trend of auction prices or the levels of commission the market will bear, and therefore the share prices of auction houses.  A stock trader to whom all art is junk can do just as well looking at charts of sales, earnings and other trends.  Or, as our analysis suggests, just as poorly, as those charts are already factored into the current price and it is only the “new news” that will move the price.

So I believe it is a smooth transition now from market risk-adjusted returns to market expected returns, as applied to the whole, freely traded securities market.  This is, at first, a highly counter-intuitive conclusion, given all the professional parties who have used history to project an expected risk-premium of several percentage points per year.  Yet these are merely the opinions of a limited number of individuals, compared with the millions of investors who actually vote with their money.    Moreover, the market is not necessarily contradicting the view that equities, “in the long term,” will yield a risk premium.  The market simply has nothing to say about long or short terms, as each investor has his or her own investment horizon.  Those sellers who are weighting the market view downwards - compared with the buyers who are weighting it upwards – may have a very short time horizon for the lower returns they expect from the securities they sell.  We do not know, at least at this point in our analysis. 

What Does It All Mean?

Since the market-expected return is an average of such a huge range of contradictory expectations, is it reasonable to argue that it tells us nothing useful, like the average temperature of a man with his head in the ice-box and his feet in the oven?  I should emphasize again that there is nothing “right” about the market-expected return, any more than any other individual’s expectation.  My critical point is that we cannot improve upon the market-expected return, as we have no factual foundation to do so.  Because of sticky luck, no individual can provide practical proof of such improvement.  If we are told the average temperature of the body, and we do not know where it is lying, that average temperature is as good a guess for the temperature of the head and feet as any other guess.

There is some intuitive appeal to the concept of averaging the disparate views of the many.  Belief in democracy is based on such an appeal.  Each voter votes for a particular congressman.  45% of the population may vote with resolution for their Republican candidates.  But if the Democrats end up with large majorities in the House and Senate, we accept that the American people have given them a mandate.  Also, we tend to accept that the electorate chose the Democratic Party’s platform, giving it a mandate to implement the platform, even if every Democratic voter made a pure “gut” decision.  We do not demand logic from the voters as to why they chose as they did.  We assume that, in some way, the voters’ gut feel was a distillation of their exposure and reaction to all the debate on the issues, and they somehow reliably weighed up all related social, economic and geopolitical facts, as far as is humanly possible.  We would never trust a limited group of experts to choose our government, because of the risk they might occasionally get it very badly wrong. 

Despite the logic of a single, market-expected return for all freely traded securities, I know it is very difficult to overcome today’s emotional appeal of equities’ “natural” superiority in the long term.  The sticky luck of two centuries’ returns remains compelling to our psychological time horizons.  In particular, most people cannot even accept that the bull market of the last quarter-century is the coincidence of very few, unexpected and therefore chance events – perhaps even a single underlying chance event – because they do not experience chance events of this nature in other parts of their lives.  Investors in 1967, 1957 and 1927 would likely have had a similarly skeptical perspective.  Less fortunate investors in 1977, 1947, 1937 and possibly 1917 would probably have been more open-minded.  And as for the countless scenarios the world never experienced due the existence, health and decisions of Churchill, Halifax, Hitler, Roosevelt, Einstein, Heisenberg, Stalin, Mao, Truman, Kennedy and Khrushchev – the list of true history makers, and those who otherwise might have been, is endless – we shall never know.