The End of An Era: Uarterly Etter
The End of An Era: Uarterly Etter
QUARTERLY LETTER
Jeremy Grantham, Chairman January 2006
It will probably be revealed in 10 or 20 years exactly how remarkable and abnormal these last 20 years or so have been. Above all they have allowed for an unprecedented increment of wealth, some real and some on paper. Since the market low of July 1982, we have never had it so good; average stock market returns have been 10.5% real against a long-term average of 6.5%. The bond market, relative to its lower risk, has been even more remarkable with an average yearly return of 7% real for the 10-year bond against a 100-year average of just 1.7%. Even residential housing has appreciated by 1.8% real (do you know anyone who made just 1.8%?) compared to a longterm average of just 0.6%. Given the leverage typical in housing with an 80% mortgage in 1982 how quaint by todays standards the returns to householders were important. It really is an amazing confluence of good fortune, but it owes an enormous amount, as all great rallies do, to starting cheap. In 1982, the U.S. stock market was 8 times very depressed earnings the 10-year government bond was 14% despite the consensus forecast of economists of long-term inflation of just under 5%. Similarly, the wonderful returns owe a lot to finishing the period expensive. U.S. stocks are at an above average p/e (19x compared to 16x), and on sensationally high profit margins, which will of course prove just as mean reverting as the very depressed margins of 1982 were. The 10year bond is down to a 4.4% yield, and the housing market sells at a national average of 4.5 times income with some dizzying peaks like Boston and San Francisco at over 6 times.
This Time it Really Was Different.
probably worse even than the accumulated loss in Iraq. But how on earth can we compare this risk with the Cold War in which most of us grew up? Thousands of nuclear warheads faced each other and nearly, we know now, got loosed in the Cuban Missile Crisis. That Russia has rejoined the world market at least for a while and released over a dozen countries to full-blown market democracies is clearly one of the great beneficial historical events. That China ceased being a doctrinaire basket case and has become by far the largest and most important of all the great economic break-outs is remarkable, as is India giving up socialist paralysis for a more pragmatic and infinitely more profitable approach. These three events, in turn, have encouraged greater economic efforts in many other developing countries. Collectively these countries have never come close to the breadth of economic success that now can be seen on the back page of The Economist every week. For example, it was recently reported that the lowest single GDP growth for the year ended in September of the 22 emerging countries listed was way over the growth rate of the European Union! The usual changing handful of GDP collapses that had been associated with emerging countries were all completely missing for over 2 years. The remarkable advantages of adding over 2.5 billion people to the capitalist system and the unprecedented strong and consistent global growth that resulted can be added to the unprecedented low inflation, low interest rates, and asset price rises already discussed. The world really has been different. Some of these benign differences will regress away, but many, if were lucky, will be permanent. Will the world ever be so successful in the broadest sense as it has been in the last two decades? Yet most of these positives affect economies more than investors and, remarkably, as we shall discuss later, the U.S. economy, in contrast to the global economy, has
How quickly we get used to major changes. We all complain, for example, at how dangerous the world is because of terrorism and yes, sooner or later, there will almost certainly be some even worse event in loss of lives than 9/11,
proven to be remarkably phlegmatic in the face of all these histrionics, refusing to budge materially from its long-term GDP trend, and to the extent it has budged at all, it has slipped a little. For investors, however, the most important factors have not changed. Asset classes in the longer run must still revert to sell around replacement value. Profit margins will still move around normal levels. The cost of capital must still, more or less, equal the return to capital. The markets will still be made up of ordinary humans and they will still be full of behavioral twitches, not the least of which are greed and fear, and we would add, career risk avoidance, and the tendency to extrapolate. The last two decades may have been very different, and in an important way, better than usual, but looking forward for U.S. investors, this time its the same, as it always is.
Exhibit of the Quarter
Exhibit 1 was used at our recent conference to show two things. First, how stable the battleship U.S. GDP is, and second, how risk reduces with time. If you have a long horizon you can, more or less, count on getting back to trend even if you hit the Great Depression. Note that the U.S. did not just get back to the old growth rate, it got back exactly to the old trend, making up all the lost
ground as if the depression had never occurred. Mean reversion for this series really lives! What I noticed though, as we presented this exhibit, was a second point: how the growth tails off from 1995 and slowly falls below trend. To find a place equally below trend, you would have to go all the way back to 1943! Remember, this is the time period that brought us a New York Times article by the Merrill Lynch economist claiming the 1990s as the greatest economic decade of the 20th Century. It also brought us Alan Greenspans cheerleading comments on the new golden era that we will come to later. The bullish bias baked into everything as always seems quite remarkable. As a note on this last point, a leading newspaper opened its reporting on November housing data with home prices continued to surge last month, a month in which the median house price declined. The comment was justified presumably by the fact that November prices were still above 12 months earlier. Really!
End of an Era in Profit Margins
The perfect environment for fat corporate profit margins is to have corporations stay conservative and nervous so that their capital spending and capacity additions are held back. The combination of 9/11, the recession of 2001,
9.0 8.5
Real GDP in Log Space
5.5 1882 1888 1894 1900 1906 1912 1918 1924 1930 1936 1942 1948 1954 1960 1966 1972 1978 1984 1990 1996 2002
Source: GMO, Federal Reserve As of 6/30/05
and the shock of the stock market break combined to do this admirably. Their caution was justified in some ways because job growth was less than normal on a sustained basis, and real earnings per hour continued flat, which together would normally guarantee anemic growth in consumption and some pressure on profits. This time, however, because of the unique, sustained drop in personal savings from 10% of income to -1% that had been caused by asset price increases and easy borrowing, consumption held up and supported U.S. corporate sales plus foreign exports. 2005 was, in fact, the first year since 1933 in which Americans spent more than they earned. There is nothing better for profits than sales that are consistently a little better than are budgeted for, and this was just what the sustained increase in personal debt had delivered. Profits have also benefited from the U.S. being in a phase of rapid outsourcing. Initially, outsourcing has a large positive effect on margins as the first outsourcer to, say, China, pockets some of the gains as extra profits and the displaced workers help keep labor costs down. Most importantly, the last 20 years have been uniquely favorable to financial profits, which have grown thanks to decreasing rates, a stable economy, easy money, and moral hazard. This environment gave the twin benefits to banks of increased leverage, especially for housing related loans, and close to zero write-downs for bad debts. In total, the financial sectors share of corporate profits has risen from about 15% in 1980 to over 30% in recent years. Not bad for a sector that makes not a single widget, and in the case of our own sub-sector money management is provably what is generously called a zero sum game, which is to say, zero before management fees and transaction costs.
and in House Prices
house prices will slow the U.S. economy and send the savings rate up, and it will happen very soon, but it will not necessarily cause any more than a mild disappointment in economic growth rates. Even a mild slowdown, however, will make for a tougher earnings environment than we have seen for a few years.
The Beginning of a New Era: Diversify or Bust
One of the characteristics of the next several years will be a let down in expectations for investors, either quite quickly or painfully slowly, and after 23 years of generally high levels of confidence and expectations, this will be hard to swallow. The p/e in 23 years went from 8 to 20, inflation from 13 to 3, 10-year bonds from 14% to 4.4%, and profit margins on sales more than doubled. It must be obvious to everyone that this increment cannot be repeated, and that some give back is inevitable, but we seem collectively reluctant to accept it, and Lord knows inertia always keeps the game going longer than we think. A growing influence in the next era will be the rapid move almost a stampede to greater diversification via non-traditional and more marginal asset classes. The remarkable success of the early movers, mainly the leading endowments and a few others, has led to a great desire to emulate them, for there has never been such a large and persistent performance gap between leading endowments and the institutional average as there has been in the last 10 years. In the last 5 years the gap is over 10% a year. If the leading, outperforming institutions own much less U.S. equity and U.S. bonds than normal, and own much more hedge funds, private equity, timber, and other peripheral investment categories, then so shall we, goes the argument. The probability that the average pension fund in 10 years does not have a more diversified portfolio is probably as close to nil as things ever get in investing, assuming of course that there still are pension funds around then. This tidal wave of diversification is having, and will continue to have, a powerful effect on the pricing of these newer, smaller asset classes it is pushing them up. It will be very easy to underestimate this pressure in the next few years, and consequently, it will be easy to sell out of forestry, emerging equities, and the rest far too soon in terms of relative performance. Imagine a world of only two asset classes. One asset class, U.S. equity, has 19 managers each with $100 billion, and the other, forestry, has 1 manager with $100 billion. Both groups have owned nothing but their own asset class, which, let us assume, conveniently, have iden3 Quarterly Letter January 2006
As discussed in earlier letters, U.S. house prices have played an important role in sustaining global demand, and we have kept a sharp eye out for a weakness. Well, 2 of the last 3 months, September and November, showed a decline in the month to month prices of the median home, seasonally adjusted. This possible peaking is 6 months earlier than I guessed at the beginning of last year using the British example as a guide and it may be a head fake, but I guess its not, given the broad-based increase in the inventory of unsold houses. In the U.K., the flattening of house prices was followed quickly by a material slowdown in the growth of consumption and in the GDP, but it was not followed by a recession. To make a simple story more complicated, the U.K. house prices have now ticked up for a few months, showing, at least temporarily, a determination not to collapse. My bet is that even flat
GMO
tical long-term returns and similar volatility, but as in the real world, they have very different flight paths in that forestry tends to do well when U.S. stocks do poorly. Both groups have been to a Jack Meyer class at Harvard Management on optimization and realize that both assets are more valuable because they are less risky when owned jointly. All 20 investors have worked out that the ideal portfolio using these two assets is 50/50. The 19 equity investors all bid for half of the timber portfolio and the timber investor bids for a small piece of each of their 19 portfolios. Who is the beneficiary of the increased value, lower risk, and higher efficiency (Sharpe Ratio) of the combined portfolio? It is my belief that in the intermediate term, almost all the extra value (say 19/20) will accrue to the original timber investor as the 19 bid against each other and only stop when the price of timber has risen so much, and the returns fallen so low, that the Sharpe Ratio is bid down to a level just above where it was in an equity-only portfolio. Do your own optimizing, but I think you can depend on the fact that the return of forestry will often be below the return on stocks, perhaps quite a bit below. Where managed timber used to be 7% to 9% real return kept up by investors dislike of illiquidity and non-traditionality (or career risk) my guess is that it will often fall well below our assumed equilibrium return for stocks of 5.7%. And what about emerging equity? This asset is still interestingly different from U.S. equity and with a similar average return profile. Will its merit as a diversifier, combined with its small size, not guarantee that sooner or later it will sell, at least from time to time, at a premium to U.S. equities? Over the next 10 years or so, as the diversification example is followed, will there not be a nearly guaranteed outperformance of all the relatively smaller, peripheral categories to the benefit of the early and second generation movers? There are also likely to be recurrent waves of overvaluation of the smaller categories, probably followed by sharp corrections. Conversely, there should be intermittent selling pressure on the large traditional asset classes. It should make for interesting times and easy miscalculations! The effect on the pricing of hedge funds, venture capital, and private equity of this steady flow of new money is harder to calculate, but probably very different. These categories are not true asset classes, but really just a repackaging of existing asset classes: stocks, bonds, currencies, commodities, and others. The near certain pressure of new money here seems likely, therefore, to be easily passed through to the underlying liquid assets. For the intermediaries the funds themselves the flow of new money seems likely to facilitate high fees in the short
4
term, as good managers will be in short supply. But rapid growth and higher fees will attract much greater talent than was ever the case 20 years ago. Hedge funds, in particular, clearly offer the new hot-shot MBA graduate the best daydream of near instant wealth, perhaps even the best real chance. This surge of talent will not only make it increasingly hard for hedge funds to add steady, strong returns above their benchmarks, it will also inevitably toughen the job in long only as well. Oh, for the good old days, when little talent came into our industry, and when good MBAs still believed the market was efficient and therefore boring and went into consulting.
Morals Are Hazardous or Stability Is Unstable
Hyman Minsky, the economist, said famously that stability is unstable and meant that long periods of stability cause all types of leverage and other risk taking to grow until they use up all the risk units freed up by the greater stability. This process can go on and on until finally something goes badly wrong. The boss of a Midwestern pension fund echoed this attitude when he was recently quoted in The Wall Street Journal as saying that he was reaching for risk because it was getting hard for him to use up all his available risk units in what he saw as a decreasingly risky environment. Twenty three years of anything is a lifetime, and will do a lot of conditioning or brain washing. Every time anyone reached to take more risk in this time period it paid off as asset prices rose and interest costs dropped. After several episodes of this, any decline is greeted like Pavlovs bell; investors salivate and buy more, on greater leverage, and at greater risk. To add to the attractiveness of risk taking, the global environment became more stable and more appealing with generally declining inflation world wide significant regional spikes were not maintained and declining interest rates. Global GDP also became less volatile. Perhaps macro economic management became better around the world, but I believe it was far more a favorable environment than a broad-based spike in the talent of central bankers. Greenspans decision to go with the flow and emphasize stability may have come at the cost of longer-term problems. His unwillingness to move against asset class bubbles, but to announce clearly his intentions to mitigate the economic downside of bubbles breaking, introduced an asymmetry to risk: heads you win, tails you might lose a bit, but I will be trying to bail you out. The net effects of this moral hazard combined with other favorable factors are not surprising: the emergence of a $1.3 trillion, highly leveraged hedge fund business; and increased personal debt to a substantial new high (over 110% of personal income), facilitated
Quarterly Letter January 2006
GMO
by new borrowing tools, especially easier and more generous mortgages, designed to tap into rising home equity. There has also been a dramatic increase in investments in other formerly fringe areas such as direct investment, venture capital, emerging markets, and commodities including forestry and oil and gas drilling. You can easily add to this list, but the net effect is that risk aversion on average drops and risk premia fall. Since we have never had such a long, drawn-out period of falling inflation, falling interest rates, rising market prices in all three great asset classes, and above all, explicit moral hazard delivered in person by the Feds boss, we are looking at possibly the greatest opportunity to test Hyman Minskys theory: we have had an unprecedented long period of good and stable times and we have responded by taking out unprecedented levels of debt leverage. The good news is we are engaged in an exciting real-life experiment for which there is no clear precedent. Guinea pigs of the world, unite! We have nothing to lose but our shirts!
End of an Era at the Fed: Greenspan Finally Mellows, but Then Leaves
with speculative fervor, and by September he had reached, this enormous increase in housing prices. This is getting close to sustained jaw-boning against an asset class bubble, and jaw-boning from the Fed Chairman can be a powerful tool. Greenspans mellowing even included the possibility of error in his bedrock policy of non-intervention against the 1999 equity bubble: Whether that judgment [to not intervene in the equity bubble] holds up through time has yet to be revealed. This is a far cry from the certainty he showed in the rightness of his action until very recently. As perhaps my last ever (sigh!) dig at pinball Alan, I cant resist digressing into his changing view on market efficiency. In the teeth of the bubble (now theres a metaphor) in 2000 he maintained a ludicrous, but still standard enough view, of market efficiency. To spot a bubble in advance requires a judgment that hundreds of thousands of well informed investors have it all wrong. That all those worthy investors could be wrong inconceivable! But by late 2005, he had become a believer, like most of us, in the market as a behavioral jungle: Human psychology being what it is, bubbles tend to feed on themselves and booms in later stages are often supported by projections of potential demand. Wow! I believe that this later opinion is much closer to his natural (honest) view of markets than the earlier quote, for surely no ones real views change that much, that quickly. Let me end my Greenspan Era with my favorite quote of his in his role as the Great Cheerleader from January 2000. The American economy was experiencing a oncein-a-century acceleration of innovation, which propelled forward productivity, output, corporate profits, and stock prices at a pace not seen in generations, if ever. He believed that the Internet had pushed back the fog of uncertainty for corporations and that lofty equity prices have reduced the cost of capital. The result has been a veritable explosion of spending on high-tech equipmentAnd I see nothing to suggest that these opportunities will peter out any time soon. All this within 1 week of the peak from which the NASDAQ declined by over 75%, the Internet sub-index by over 90%, followed by the U.S. entering a recession and a capital spending bust! All in all, I shall miss him!
The King Is Dead, Long Live the King
It truly is the end of an era when I find myself agreeing with Mr. Moral Hazard himself the man who encouraged the biggest bubble in U.S. history but increasingly in late 2005 that has been the case. History cautions, he said in July last year, that long periods of relative stability often engender unrealistic expectations of its permanence and, at times, may lead to financial excess and economic stress. Greenspan also echoed our warnings on the low risk premia last August. Vast increases in the market value of assets are in part the result of investors accepting lower compensation for risk History has not dealt kindly with the aftermath of protracted periods of low risk premiums. And by October he could have been writing my quarterly letter: Extended periods of low concern about risk have invariably been followed by reversal in asset prices. Even more explicitly, in September he added, The irony is that economic stability produces its own risk. Pure Hyman Minsky! Greenspan, despite his concern over potential irrational exuberance, in 1996 had backed off from any attempt to interfere with the rising market, probably influenced by his personal political concerns. As his time ran out last year, however, with less need to worry about politics, his comments on the housing bubble showed much less timidity. He started quite conservatively in April saying, For the nation as a whole, I do not believe that a bubble has developed, but the extraordinary gains in some local markets may not be sustainable. By July he was escalating to, Some regional markets have been charged
So what about Ben Bernanke? For those of us who fear the long-term consequences of moral hazard and believe that a greater degree of concern with major asset class bubbles is warranted, the news is entirely bleak. Bernanke sounds like an unreconstructed Greenspan, perhaps in spades. Helicopter Ben, named for his comment
5
about dropping money by helicopter if necessary, did not get his nickname by sounding like he would remove all traces of moral hazard. For the Fed to interfere with security speculation, he said in 2004, is neither desirable nor feasible, but if a sudden correction in asset prices does occur the Feds first responsibility is to protectto provide ample liquidity until the crisis has passed. As pure a statement of moral hazard as you could have. As for new housing bubbles, in complete contrast to Greenspan, he sounded utterly complacent late last year. Housing prices, he said, merely reflect strong economic fundamentals. This in the face of a price series that shows the average U.S. house prices are over 2 standard deviations overpriced (a 40-year event randomly) when viewed as a multiple of family income. Nor is Bernanke any more sympathetic to the view that we share with Greenspan on the dangers of a low risk premium, commenting in September last year that Risk premiums [in U.S. stocks] look quite normal. So, if you will forgive me: The Bernanke put is alive and well And where it leads no one can tell.
The Nature of 2005: the Revenge of the Risk Takers
based value model beat price/book, and, on average, our approach has a 2% a year edge over book. But it is this year that counts. It is, in fact, the most profoundly disturbing aspect of quantery that simple, old-fashioned models regularly clean our clock.
Lessons of the Last Year
The year came in conservatively and for 5 months the market went slightly down, and low volatile, high quality stocks did what you would expect in that kind of year they outperformed. However, for the next 7 months, the market drifted up, accompanied by unexpected strength in volatile and low quality stocks the kind of difference you might expect in a very strong year, not a nearly flat one. The year ended with a +9% spread in favor of the highest 25% of the market by volatility against the lowest 25%, a spread that had been -7% in May, so a 16% swing. Ouch! The gap in favor of the lowest 30% quality compared to the highest 30% was in a range of +6% to +11% depending on precise definitions of quality. The Russell 2000, which had been immense for 5 years, quickly fell 7% to the S&P by May and then recovered to dead flat, leaving the running to the Russell 2500 or mid cap, which ended 3% ahead of the S&P. Most measures of value versus growth gave a couple of points of outperformance to value in large cap and a draw in small cap, a very far cry from the over 15% a year spread for the last 5 years. The most irritating characteristic for GMO was that the best 25% of price/book, an antique, crude measure of value, and one that we believe is badly distorted by modern accounting, beat the S&P 500 by 5.2%, where our sexy, quality-and-everything-else adjusted dividend discount model lost by 2.9%. The quantitative model we used 30 years ago would have had an easy year outperforming. To be fair, we have had more years in which our broad
6
The two main lessons for me are simple: never underestimate inertia (or what investors typically refer to as momentum). Things are reluctant to change. Once moving, for example, they keep moving much longer than seems reasonable. This is a lesson I must have learned 27 times! We have done asset allocation now for well over 15 years, and we really have learned to let things run, and it has helped a lot. But, once again, we pulled the trigger too soon to play against low quality stocks that a year ago had already had 2 wonderful years. Despite being expensive, they had a third strong year, and hurt our performance. On the other hand, we did not reduce our very large bets on emerging market equities, an asset class that many strategists said a year ago had gone too far, too fast. In fact, helped by their being relatively cheap on our numbers, we didnt sell a share. Our strategy, led admirably by Arjun Divecha in Berkeley, was up 70%, 27%, and 40% in the last 3 years and is up 7% this year in 7 trading days. That turns a dollar into three and a quarter! Now, finally, we have to do some modest selling, but the moral once again is inertia lives; do not overmanage and let things run.
Hits and Misses in 2005
We predicted a -6% real return for U.S. equity and we had a plus 1%. We bears were unhappy, but so were the bulls with the normal +5% to +10% estimates. As mentioned, we bet on high quality and this was a painful loss. On the positive side, though, we got asset class ranking about right. Emerging equity, our favorite, was far and away the best. Foreign developed, our second choice, also beat the S&P handsomely by 8.5% in dollar terms, or 24% in local terms, and these two winning bets were big enough to carry multiple errors. I believed in January that oil prices might easily stay over $50 and they did. And I feared a 15% rally in the dollar, because even if the long-term story against the dollar were to turn out to be correct, as we believe it will, a 15% counter-trend rally after the 2004 dollar weakness would have been standard operating procedure. I said it was my worst fear, and it was exactly what we got, but foreign market strength in local currency much more than offset their deficit in currency returns. In implementation, emerging markets outperformed by 5%, and fixed income was generally up nicely, led by our emerging debt strategy, which curiously also outper-
formed by precisely 5%. Asset allocation, where we have 27% of our assets, had its sixth consecutive strong year, and our $3 billion of small cap U.S. was strong. Our very large foreign developed funds were collectively about a draw, and our U.S. large cap funds and our hedge funds were poor, but not disastrous, except in the case of our new quality strategy that underperformed the S&P 500 by 5.8%, but, for the record, was almost exactly flat against our index of high quality blue chips. All in all, perhaps adequate, since more of our assets won significantly than lost, but still disappointing.
Forecast for 2006
expensive halves based on value. Slicing the data in this manner, 2006 ends up in the expensive half of year 2s where only 1 of the 9 years being studied ended in positive territory, and had an average real return of -14%. Given that most of our sample lies in the pre-Greenspan and Helicopter Ben era, where there was a lot less moral hazard flowing around the markets, lets settle for a -8% return for the first 3 quarters of 2006.
Recommendations for (the first 9 months of) 2006
The past couple of years, the January editions of the quarterly letter have included a 1-year forecast for the S&P, based primarily on the Presidential Cycle. In keeping with tradition, were going to make our predictions for 2006, but before we get into forecasting, Id like to add the appropriate caveats. These 1-year forecasts are interesting and inform us on the margin, but our 7-year asset class forecasts are what drive our asset allocation bets. That being said, the 1-year forecasting methodology has captured the spirit of the exercise over the last 2 years when we have made these forecasts, so were going to try again. (To remind you, the forecasts for the last 2 years were +10.5% and -3.5% nominal, or +8% and -6% real.) 2006 is the second year of the Presidential Cycle, which has typically been a weak year for stocks. The average real return for year 2s of the election cycle since 1932 has been +3.2%, with a little more than half of the years having a negative real return. Now, the +3.2% real return points to a weak year for the market, but it fails to pick up on something very interesting that happens within the course of the average year 2. As it turns out, the year 2 effect only lasts through the first 3 quarters of the year, at which point, year 3 begins. (The third year of the Presidential Cycle is the strongest of the 4 years, with no down years since 1950.) Take a look at the following table:
Year 2 Return % of up years Jan - Sep -4.1% 44% Oct - Dec 7.6% 83%
Cash and more cash! Better yet, good cash substitutes like a conservative mix of hedge funds. We would suggest holding as much cash as your career risk will allow you to, which usually is not very much. We run a variety of asset allocation products with varying degrees of aggressiveness. The absolute return accounts where we manage portfolios to make real money, i.e., beat inflation by the widest margin possible, are moving towards a 50% total weight in hedge funds and cash plus strategies. Within equities, emerging is still our favorite asset class, followed by high quality U.S. stocks, which are starting to look pretty exciting on a relative basis following another year of poor performance. Actually, avoiding stocks of junky U.S. companies, or shorting them where possible, is probably an even better bet than buying U.S. high quality stocks. At the high quality end, our interest is substantial, but our enthusiasm is moderated by the large element of specific risk: the stocks are so large, and industry concentration so high, that there is substantial risk for something going badly wrong with, say, a large drug company or the drug industry itself. The junky portfolio, in contrast, is made up of smaller and more diversified bets. Within fixed income, our second favorite category after cash is the 10 year TIPS with a real yield of 2%. All in all, we are cutting risk across our portfolios and concentrating most of the available risk units on emerging equities. Taking as little risk as possible and living to fight another day seems to be the mantra for at least the next 9 months.
Ben Inker and Letters to the Committee
As you can tell from the average returns and hit rates above, the last quarter of year 2 is a much better time to be in the market than the first 3 quarters. If thats not enough data mining for you, lets split the January September portion of the second years into cheap and
This quarter and next our Letters to the Committee series will cover the nature of risk in investing. Ben Inker is covering the state of the art, admittedly with a GMO twist. Next quarter, I will do Part 2, the fun part, which will cover the limitations and drawbacks I see with the state of the art.
Disclaimer: The foregoing does not constitute an offer of any securities for sale. Past performance is not indicative of future results. The views expressed herein are those of Jeremy Grantham and GMO and are not intended as investment advice. Copyright 2006 by GMO LLC. All rights reserved.
GMO
SPECIAL TOPIC
Ben Inker, Director of Asset Allocation January 2006
You are probably used to hearing GMO lambaste conventional wisdom and the academics when it comes to investing. While being an effective investor almost by definition means ignoring or even defying convention, when it comes to risk control, the academic view and conventional wisdom are definitely useful. While we believe that neither tell the whole story when it comes to risk and risk control, they actually make an excellent starting point.
Conventional Recommendations for Risk Control
overvalued asset class. Because of this, diversification not only reduces the volatility of the portfolio, but also the important risk of a multi-year disappointing return. b) Rebalance back to your long-term targets in a disciplined fashion. Conventional wisdom and the academics like this for the same basic reason if the original targets made sense from a risk/return perspective, there is no reason to let your portfolio drift away. In fact, almost everyone but the most extreme efficient market believers is in favor of rebalancing, since the only real argument against it is to assert that the capitalization weighted global securities portfolio is the only efficient portfolio, and that market movements merely reflect a well-reasoned change to that portfolio. This assumption is sufficiently heroic that even the academics who firmly believe that the stock market is efficient are happy to concede that one might want to second-guess the global markets in this fashion, for the sake of maintaining a consistent risk level. Our belief, again, adds a nuance to this idea that makes it even more important from our point of view than the conventional one. Given that the risk of overpaying for an asset is so deadly and the benefits of underpaying can be so material, rebalancing has the substantial benefit of automatically moving your portfolio away from asset classes that have done well, and are therefore at more risk of being overvalued, and towards asset classes that have done poorly, and are therefore more likely to be undervalued. While the academics view rebalancing as a pure risk-control exercise, to our minds it also improves long-term returns, since asset class returns are mean reverting. c) Use an optimizer to build an efficient portfolio. This probably cant yet be called conventional wisdom, although it is old hat from the academic perspective, with Markowitzs paper on mean variance optimization for portfolios now over 50 years old.
a) Diversify across all asset classes, insofar as is possible. The conventional view on this can be summed up basically as dont put all your eggs in one basket. The academic view comes to the same conclusion, albeit with a lot more math. The argument is basically that, since asset classes are not perfectly correlated, a well diversified portfolio will have an expected return equal to the weighted average of the expected returns of the underlying assets, but a volatility which is significantly lower than the weighted average of the assets volatilities. We come to the same conclusion ourselves, although we would tack on an additional rationale for diversification. In our view, one of the biggest risks an investor faces is the risk of overpaying for an asset. For an investor with a long time horizon, the volatility of an asset may not be of overwhelming concern, as long as the pricing of the asset is such that the investor is being compensated for taking that risk. Recessions, inflation, war, pestilence and other calamities generally do not have all that much impact over a 10to 20-year time horizon, but buying the asset when it is expensive can be devastating. Diversifying broadly reduces this threat because it reduces the odds of having a large percentage of the portfolio in an
* The Letters to the Investment Committee series is designed for a very
focused market: members of institutional committees who are well informed but non-investment professionals.
The academic argument is mathematically complicated but conceptually fairly simple. As a riskaverse investor, your goal in putting together a portfolio is to get the highest possible expected return for a given level of risk, or equivalently, put together the lowest risk portfolio that meets your return requirement. Mean variance optimization provides a way to do this, taking into account your beliefs as to the expected returns, volatilities, and correlations of the available assets. The result of this exercise is an efficient frontier of portfolios, which are efficient in that they have the highest expected return possible for any given level of volatility. A simple version follows below, with three assets large cap stocks, small cap stocks, and bonds. The line is the frontier of portfolios made up of large caps, small caps, and bonds that provide the highest return for each level of risk. The important thing to note, particularly from the academic perspective, is that the squares for large caps and bonds are below the frontier. This means that at the risk level that each of these asset classes embodies, some particular portfolio made up of all three portfolios is superior to The holding that one asset class in isolation.1 academics love optimization because it adds an element of mathematical rigor to the portfolio construction process.
7% 6%
Practitioners are somewhat more lukewarm to optimization, because it has significant practical drawbacks. First, it requires you to actually have beliefs with regard to the expected returns, volatilities, and correlations of the assets, which is not a trivial exercise. Second, it can be quite sensitive to those beliefs, so that apparently small changes to expected returns or volatilities can lead to wildly different portfolios. And third, the resulting portfolios are often so different from conventional portfolios that investors cannot stomach the career or reputational risk involved in moving over to them, choosing instead to ensure that their allocations mirror those of their peers.2 We believe that optimization is quite often a useful tool for investors despite these drawbacks, since the practical solutions to the problems seem more likely to improve portfolios than the reverse. First, while coming up with expected returns, etc., is not a trivial task, it can be quite helpful to undertake, since it can cause the investor to think critically about why an investment has the characteristics it does, and what kind of expected return is fair given the characteristics of that investment. It is hard to see this type of analysis as a bad thing.3 Second, the resulting portfolios will tend to be more diversified than the
Efficient Frontier
Small Caps Large Caps
Return
Bonds
16%
18%
20%
22%
Small caps wind up on the efficient frontier for the simple reason that, as the riskiest and highest returning asset class, it is impossible to put together a portfolio as risky or with as high an expected return including the other two assets, barring leverage. There are more mathematical drawbacks to optimization as well. Mean variance optimization assumes that the variance of an asset is an acceptable definition of its riskiness, which is not true insofar as the distribution of returns is not normal. And even further, the variance that should be used is not really the unconditional variance of the returns to the asset, but the expected error around the forecast, which is somewhat different, and many techniques for determining the expected returns are not even capable of coming up with the necessary standard error, let alone the relationship between those errors, which is required for determining the correlations. There are some ways around these problems, but they come with their own baggage. At least it is very unlikely to be a bad thing if any material thought is put into it. If an investor chooses to plug in trailing 10-year returns to each asset, for example, as his expected returns going forward, the optimization will have the probably disastrous result of encouraging the investor to pile into whatever assets have been doing well recently the exact opposite of rebalancing.
GMO
conventional portfolio, with smaller allocations to the conventional choices and a smaller home bias than exists in most investors portfolios. And third, even though investors are unlikely to take the outcome of the optimization at face value and put 30% of their portfolio into, say, timber, seeing that the optimal portfolios contain large amounts of timber will tend to make an investor seriously consider at least a small investment in such niche asset classes, which is all to the good. d) Pay attention to the beta of your portfolio, and any assets you consider adding. Beta is the sensitivity of an asset or portfolio to movements in the stock market. The conventional wisdom and the academics are really on the same page here, although the academics take it a bit further. The conventional wisdom says that beta is the primary source of risk to your portfolio and you should keep a close eye on it, avoiding increasing beta unnecessarily. Some academics suggest that beta is the risk in portfolios, and the expected return to an asset or strategy is determined by how much beta it delivers. In the extreme, this could argue against diversification, which would be a negative. But most academics are less dogmatic on the topic, arguing instead that low beta assets with reasonably high expected returns are a real benefit to a portfolio, just as an optimizer would suggest. Our own view is quite similar to the conventional one. While it is the case that asset pricing is quite inefficient and high beta assets can easily have lower expected returns than low beta assets, beta is extremely important from a risk perspective, and knowing how sensitive your portfolio is to the stock market is extremely useful, even though there are other dimensions of risk that are very important as well. e) Calculate the value at risk of your portfolio. Its probably too soon to say that this is conventional wisdom unless you are an investment bank, but it is a
measure that is growing in popularity. Value at risk (VAR) is an attempt to calculate the risk of a portfolio in terms of the amount of loss to expect the portfolio could be subject to over a certain period of time at a particular confidence level.4 There are a number of different ways to calculate this measure, but the details are not necessarily that important for our purpose here.5 The academic merit of VAR is that it is potentially a more flexible way of measuring risk than variance, and the practical merit is that most investors are capable of understanding what it means to say that a portfolio has a 1% chance of losing 5% over a 10day period. Our view is that VAR can be a useful way to think of the risk of levered portfolios (and indeed the hedge funds and investment banks who are the most fervent users are quite levered) with great flexibility to build into your risk model any idiosyncratic view of the world you have. But the quality of the data coming out is no better than that of the data going in, so VAR can easily give a false sense of security. If your VAR calculation underestimates risk, it may encourage taking on excessive leverage and lead you into serious trouble, la Long-Term Capital Management. VAR can also lead to rather frustrating ex-post analysis, since if your portfolio actually loses 30% when its VAR was 10%, you dont know if the calculation was wrong or if it was actually a 6 standard deviation event.
Summary
Much of the conventional wisdom when it comes to risk control actually is wisdom. Paying attention to portfolio diversification, rebalancing, and beta are basic requirements for thoughtful portfolio management. While optimization and value at risk have more pitfalls in their use than the first three, they can also be helpful in constructing and analyzing investment portfolios. As long as their limitations are understood, they can be exceptionally flexible and useful tools.
For example, an investment bank might want to calculate the 10-day VAR of its holding at a 99% confidence level. The result would be a loss on its portfolio, which would be expected to occur once in every 100 10-day periods. There are probably almost as many ways to calculate VAR as there are firms using it, but three common ones follow: 1) a covariance-based VAR calculation turns the variance of a portfolio using the same methods as mean-variance optimization and turns it into a value at risk; 2) a historically-based VAR calculates how a portfolio would have done if held through some period of past history - for example the past year - looking at the worst 1%, 5%, or 10% of events; and 3) a Monte Carlo simulation VAR would determine the loss in a similar way to the historical-based VAR, while using some other method of generating the returns besides a simple covariance matrix or history.
Disclaimer: The foregoing does not constitute an offer of any securities for sale. Past performance is not indicative of future results. The views expressed herein are those of Jeremy Grantham and GMO and are not intended as investment advice. Copyright 2006 by GMO LLC. All rights reserved.
GMO