现在不是天启：不可避免的痛苦被推迟

GMO Quarterly Letter January 2005 Jeremy Grantham, Chairman Apocalypse Not Now: Inevitable Pain Postponed Letters to the Investment Committee II Barron's Interview Reprint Great Bear Markets Take Their Time continued through the first leg of the bear market, Yes, it already seems like we’ve been waiting a long time allowing – indeed setting up – a sustained bear market for the market low, but actually this wait is far from rally that has been one of the longest in history, and one remarkable on a historical basis. In fact, a low in these characterized by extreme speculation of a breadth not next 24 months would be the most typical timing we seen in prior new bull markets. But is it really a surprise could expect from history. But this market cycle has been that the market’s “animal spirits” – Keynes’s term for interestingly different from earlier bubbles and probably willingness to take risk – were not thoroughly broken this we should not expect a typical experience. time as they were following the bubbles in 1929, 1965, and Japan? Don’t bigger actions need bigger reactions?

The first difference is that the market in 2000 went far Doesn’t the removal of such excessive bullishness and higher than ever before: 34 times trailing earnings versus faith in a new era need either more than typical bad news 21 times at the market tops of 1965 and 1929. Even or more time? September 11 might have broken the Japan, adjusted for cross holdings, was not materially market’s heart, but the Fed and the Administration moved higher. Second, the sustained phase of declining interest against the negative fallout of September 11 to an rates and easy money has no historical US parallel, either unprecedented degree, with low rates and easy credit and in extent or duration. Third, stock ownership was far also with tax cuts that would never have passed without broader than ever before, so that the great majority of the that disaster. readers of large circulation magazines, for example, owned stocks this time, while they did not in the other But now, finally, the credit cycle is turning down and bull markets.

Other media, notably TV, joined in, interest rates are being raised carefully, although they culminating at the top of the market in 1999 – 2000 in a remain far lower than normal and are therefore still blizzard of coverage that was overwhelmingly bullish, for stimulative. Moral hazard is being removed equally bullishness in a bull market provably sells magazines and carefully by sprinkling a few cautious comments around; attracts TV audiences, whereas bearishness does not. the recent Fed’s rate setting committee (of December 14) Fourth, bullishness is also good for business at financial suggested that sustained easy money “might be” firms, and bullish recommendations far outnumber encouraging “potentially excessive risk taking” and bearish ones, but that is by now an old story. Fifth, we pushing up asset prices (they noticed!), and separately had unprecedented moral hazard in this cycle from a Greenspan suggested that speculators would be well Federal Reserve boss who was not only acting as advised to prepare for higher rates.

Doesn’t this all feel cheerleader for the internet revolution, high productivity, like an ordinary beginning of a stock market downturn? and high profit margins, but was also making it clear 2 years ago that speculation and the carry trade was a free The normal tendency is for the length of the market lunch. decline following a bubble to take about the same time or a little less to get back to trend as the market increase took I review this old ground to point out how powerful the from trend line to peak. For this down cycle from March forces were that pushed us in 2000 to such substantial 2000, that would lead us to expect a market low this year new highs in P/Es. Perhaps more interestingly, the or next. And admittedly there is quite a bit of noise favorable credit cycle and moral hazard of the upswing around this average, although less than you’d probably think.

A low in 2009 and 2010 would be just within the caveat, my title should probably have read “Apocalypse normal range at the longer end and 2002 would equally Probably Not Now”.) This does not mean that we should have been just within the short end of the normal range. expect these next 2 years to be anything but dreary, as Indeed, in September 2002 when the S&P tumbled to we’ll get to later, but a ‘dreary’ environment just seems 775, which was half its peak price, the market looked for unlikely to get the market to fair value that quickly. And a couple of weeks like it might just make the real break come 2007, we will be re-entering the Presidential third and slice through the then fair value of 700+/- and year pre-election stimulus. Betting on a market decline establish a real low, say in the low 600s, and get this between late 2006 and December 2007 is a very low whole unpleasantness out of the way.

That would have percentage game indeed. But although the market is felt like the S&L bloodbath of the late 1980s, in which unlikely to do badly in 2007 and even 2008, just the there were dramatic and rapid declines in the entire US steady passage of time allows for the great bubble real estate market to way below fair value, followed by an mentality to become a more distant memory and for high equally rapid gathering of the vultures that came flying in global growth rates and fat profit margins to regress to to make fortunes and set the cycle turning again. That more average levels, which would set the scene for a final event was nasty and brutish and short, but how much market low in 2009 and 2010. better that was than, say, the drawn-out Japanese land cycle at the other extreme, where land prices have Greenspanner* in the Works: Inadequate Pensions declined for 13 consecutive years!

As I boringly repeat, The consequences of this drawing out of the market we all make more money when asset classes are cheap decline would be many and painful. Probably the worst and we should welcome falling asset prices, not rising would be for retirement accounts. Corporate defined prices, if we have a decently long time horizon. And benefit pension plans would probably be looking at equally, we should prefer a rapid market decline to a another 6 years of investment returns far below actual slow one. You simply end up with more money if you hit assumptions. They would be deep in a crisis of the market low more quickly and then have more time underfunding and would be facing much increased annual compounding higher returns at the lower prices. pension costs as a drain to earnings. Many firms facing that would close their DB plans, which have clearly been So a low in late 2002 would have been the best.

A low the jewel of the retirement world. Unfortunately, this this year would be more painful both psychologically and seems likely to happen and it will be a bitter blow for to the pocket book than in 2002, but much less painful many retirees. Returns on individual retirement accounts than dragging this thing out to 2009, 2010, or later. will also be far below expectations at exactly the time that Unfortunately, this last alternative seems increasingly retirement rates hit the baby boom. There will then be a likely to me and I wish for us and everyone else that it not growing realization that we are collectively saving far too be the case and that the market low comes soon. Above little, for by then we will have been undersaving for at all, particularly with a looming retirement crisis, we need least 10 years.

Individuals, seduced by overpriced the high compound returns that come from lower prices. markets and overpriced hype into believing in the great But the painful fact is that now we come once again into bubble, will have taken an even longer ‘pension holiday’ the unfavorable first 2 years of the Presidential Cycle than the DB corporate plans. During the bubble, with two challenges: animal spirits are artificially high aggregate personal savings fell to almost zero from the and the global economy is artificially strong. (Animal formerly normal level of 8% of income, and have not yet spirits are ‘artificially’ high because they have been materially recovered, just as many DB plans made no raised by the long favorable environment for risk taking, contributions at all for several years. When the smoke and global growth is ‘artificially strong’ because the clears, it will probably be seen that for well over 10 years sustained US credit expansion has been allowed to flow personal pension savings averaged under 3% of income around the world through our sustained and growing US instead of the normal and necessary 8%, creating a total trade deficit.

This deficit in turn has stimulated the shortfall of over 50% of 1 year’s income. The shortfall of economies of the surplus countries like China as they DB plans is likely to be similarly painful. Even if the responded by increasing their money supplies.) Both reality of undersaving is then appreciated and personal components – animal spirits and global growth – just savings moves up by, say, 2% a year over the old normal seem too strong for the market to get to trend line and below (725 on the S&P 500) in the next 2 years, unless * For you American dudes, a spanner is proper English for a there is an unexpected crisis. (Given this important monkey wrench, and the expression means totally fouled up.

GMO Quarterly Letter – January 2005 2 to a total of 10%, two things would be true. First, the of -6% would also apply to next year, 2006, unless there extra savings would be a slight drag on parts of the is a dramatic market decline, but let’s concentrate on this consumer oriented economy for many years to come, and year first. The problem with this estimate is the volatility. second, it would still take 25 years to catch up with where Unlike predicting 2004, when volatility in the accumulated savings should have been! In this way, pain Presidential Cycle’s fourth year is famously low, this first from the bubble will have been pushed very deep into the year’s volatility is moderately above average. So the future where, given the population profile, funding range around our -6% real, which is intended to include retirements will be even more difficult than today. In this two-thirds of all years (one standard deviation) is a sense there really is no free lunch. There is a certain modest ±21%, or a range of +15% to -27%! As for the amount of pain from moving back to more normal debt remaining one-third of the years that lie outside this and interest rate levels, and ‘Greenspannery’ can range, we recommend benign neglect, since it is just too postpone the pain and spread it out over a long time, painful to be reminded of the real uncertainties involved Japanese fashion, but it cannot remove it. Indeed, in our business. uniquely for pensions, delaying the time it takes for asset prices to reach fair value does not just cause the same Forecast for 2005 from February 1 amount of pain to be moved into the future, it causes This is where the January effect rides to the rescue, as it incremental pain. Higher asset prices not only suppress not only has considerable forecasting ability, but has also your savings rates because you feel rich, but also served in the past to substantially narrow the yearly compound your accumulated savings at a lower rate than distribution of returns following bad Januaries. In year’s would occur with more reasonable asset prices. 1 and 2 of the Presidential Cycle, when prior Januaries have been below their average returns of +1.0% and the 2004 Predictions and Results market has been in the worst quarter of overpricing as it In 2003’s fourth quarter letter we estimated that for 2004 is now, the balance of the year over the past 72 years has the S&P 500 would return 10.5% ±5.5%, if January last risen only 22% of the time and has delivered an average year showed a market gain, which it did. As a suitable real return of -9%. The best performance was only reward for having the chutzpah to make a 1-year forecast +6.5%, and the range itself narrowed to ±10% at one for the first time in 26 years, the market ended up standard deviation. +10.9%! Last year we forecast very low market volatility and a strong value year, and both were also correct, owing In substantial contrast, if the prior January has been a lot to our research on the Presidential Cycle, which has above average in these overpriced years, the rest of the been running exceptionally true to form recently. We also year has been up 56% of the time, and the average return predicted moderate outperformance for foreign has improved to -2%, but with a normally wide developed equities which came in 9.5% higher than the distribution range.

S&P, and very strong outperformance for emerging Caveats and Forecasts for the Next Several Years which, due to a brilliant November, came in 17.5% ahead of the S&P. There were no other predictions. We have three inputs that all seem to give an edge in making 1-year market predictions: the January Rule, the It’s blindingly obvious that we should retire this 1-year Presidential Cycle, and the current market valuations. forecast and settle for a perfect record. However hubris These three are embodied in our estimates above, but we being what it is, and the Presidential Cycle as well as urge readers, probably unnecessarily, not to put a lot of value and the January effect being on a roll, we – Nick weight in them. History, even 72 years of it, is just Nanda and I – will try again. waiting to trap us in the short term. The range around these estimates is, as indicated, mind boggling. More Forecast for 2005: The Attack of the Drearies importantly, the principle of focusing on a 1-year horizon Value matters in the first 2 years of the Presidential Cycle, is probably a bad idea. You should also know that GMO unlike the third year, where almost all years go up. If in does not materially use this short-term input in its asset the first 2 years you are in the worst quarter by value (and allocation process. We are intrigued by it and are we are currently deep in the worst quarter), then the informed by it on the margin, but overwhelmingly our market has delivered on average since 1932 a negative decisions are driven by our 7-year asset class forecasts, real return almost exactly two-thirds of the time and always available at www.gmo.com, which are based on delivered on average a real return of -6%. This forecast slow, steady regression to the mean. Bets made on this 3 Quarterly Letter – January 2005 GMO 7-year basis are highly likely to win, based on long-term believe one or two minor factors going wrong is business history and actual 35-year experience, although timing as usual. It will probably take one or two major problems uncertainty in contrast can be nearly fatal. So far, only to bring this market to its knees in the next 2 years. one important long-term bet has failed, which is that emerging equity should beat the S&P 500. Even that bet, GMO Performance in 2004 when these two asset classes eventually reach fair value, will be well ahead since inception. The single most After 4 fairly consistently good years, GMO had a important prediction we have is still what it has been for completely mixed one. Starting at the worst end, our US the last 8 years: that the S&P 500 is overpriced and will large cap value strategies were behind their benchmarks go back to trend line fair value sooner or later. This is by almost 4% and our US core and US large cap growth about 730 on the S&P 500 compared to almost 1200 strategies were behind by about 1%. Emerging market today. To us this is far more serious input than last year equity and small cap international quant both ended up being likely to outperform by 10.5% or this year to behind by about 1.5%, but both had outperformed by over underperform by 6%. The only seriously prudent 13% the year before. The $4.5 billion in our 11 hedge position in our view is to lower risk and prepare for strategies was a bit disappointing, but not disastrous, with substantially lower prices. The rest is simply a range from +1% to +15% absolute return. On the more interesting, short-term gamesmanship and career risk positive side, our large cap international equity management. But to one degree or another, almost all of strategies – by far our largest pool of assets – were ahead us have to deal with these real life considerations. of their respective benchmarks by 1% to 3%, and our US and foreign bond strategies were both nicely ahead.

Risks to the System Asset allocation had its fifth consecutive good year with a lead of over 3% for our global balanced strategy.

Our forecast of -6% real for 2005 is based on nothing Finally, our emerging debt strategy once again was ahead unusual going wrong, but an unusual number of things of its benchmark by an eyebrow raising 7 percentage could go wrong. Let us count the ways: points this year.

The main problem for Possible Problems for the Next 2 Years … and Suggested Odds us in the US equity „ China stumbles .15 market in both long only „ Major terrorism or dramatically bad Iraqi developments .15 and hedge strategies was a sustained „ Other major political problems (Russia or totally unexpected) .10 outperformance of low „ Unexpected major economic weakness in Japan and/or Europe .15 quality stocks. GMO’s „ Substantial increase in US inflation .15 strategies tend to do best „ Major increase in commodity prices, including oil .20 in declining markets „ Minor or major financial crisis .30 that tilt to quality, next best in rising markets „ Rapid decline in dollar (over -10%) .15 that tilt to quality, and „ All other major problems out of left field .10 worst in rising markets „ Odds of none of the above going wrong (assuming above odds were right) .20 where low quality wins, which unfortunately Other than the fact that our odds are probably only worth describes both of the last 2 years in the US. Since high what you’re paying for them, what can be said about this quality is now cheap and since our outlook this year is for picture? The odds of each problem are not exceptionally a moderate, but perhaps sustained bear market, we expect high and many strategists would put them higher. The to do much better across the board in the US, other things possible exception would be the odds of at least a minor being even, which I suppose they are about two–thirds of financial crisis, which at .30 are pretty serious in our the time. opinion. The number of problems, though is unusually high, and that is one important reason why one should Recommendations for ‘Staying Alive in 2005’ tread very carefully for the next 2 years, for the odds of something going wrong are 4 to 1, if my component odds We recommend as much avoidance of risk as is possible, are correct. In the interest of full disclosure though, I given the constraints of career risk management; for one GMO Quarterly Letter – January 2005 4 of the paradoxes of our business is that reducing or Exhibit 2 avoiding real risk in portfolios can seriously increase High Volatile Stocks (25% by Market Cap) career and business risk, which rises with any deviation Are Painful in Bear Markets* from standard behavior. For global equities, the biggest Return of high volatile stocks, relative to the market portfolio risks are likely to come from any exposure to small cap and low quality, particularly those that are Today highly volatile and have high levels of debt. This is likely to be particularly true in the US. Conversely, investors should emphasize large, high quality blue chips.

Fortunately both quality stocks and large cap stocks following 2 very poor years are substantially cheaper than -5.3% the market in the US and moderately cheaper in EAFE, Return Relative to and both are likely to outperform in bear markets.

Exhibit 1 shows GMO’s measurement of the long-term -12.8% ebb and flow of the relative value of US quality stocks Market in Bear Markets against the rest of the market. Low on this exhibit is good, and as you can see high ‘quality’ today is very well positioned. The attractive relative pricing of high quality and large cap stocks is a much improved situation from a year ago when the ‘skating on thin ice’ phase started. -28.5% Exhibit 2 shows in contrast how dangerous the most volatile quarter of the market is in bear markets, and how Volatile Stocks Volatile Stocks different the pain is, according to how relatively cheap Are Cheap Are Expensive volatility is. Today, as you can see, it is in the worst third, * Peak to trough, a real return of -9% or less. where it has underperformed in the past by a stunning Source: GMO. Data from 1/1/1964 - 9/30/2004 -28.5% a year on average. For bonds, we also recommend an emphasis on quality and below average in most cases it should be seriously considered. duration. Cash is hard to own, given its low yield, and it Conservative hedge funds are an easier and better is particularly career-threatening when you’re wrong, but alternative if you have access to them. Up to 10% in commodities is a great diversifier since they are Exhibit 1 positively correlated with inflation, in complete contrast High Quality Stocks Are Cheap Today to stocks and bonds, which are both negatively correlated. Quality is cheaper today than it has been for 95% of the We do not particularly fear inflation and have no special time since 1965. insight into it, but an acceleration in inflation is clearly 1.3 one of the many risks we face in the next 2 years, and rapid global growth with increasing demands for Relative Valuation of Quality Stocks commodities led by China and India could well be a 1.2 cause. Timber (of course!) is a fine commodity in this respect too, for China is using rapidly increasing quantities 1.1 Yes, But What Do You Do?

(Quality Equity / Largest 500 U.S. Stocks) 1.0 Perhaps not surprisingly, having a broad line of asset allocation strategies with different levels of aggressiveness gives rise to criticisms that if I believe the 0.9 US market will go down, how can I with a clear conscience simultaneously be associated with a strategy 0.8 like our global balanced strategy, which has over 25% in Dec- 65 68 71 74 77 80 83 86 89 92 95 98 01 04 US stocks. The point is that we cannot tell a client how much career risk to take and having absolutely no US Source: GMO. As of 12/31/04 5 Quarterly Letter – January 2005 GMO stocks can often be extremely risky for a career or a 50% total equities or about 25% US equities. But given business, for it is an uncertain world where almost these constraints, we build the most efficient portfolio we anything can happen for a year or 2, and 2 years is a long can using our 7-year forecasts for return, volatility as a time in investing.

Everyone’s situation and willingness to measure of risk, and traditional correlations to measure take risk is different. For myself, in my own account, I the differences between asset classes. The attached am net short US stocks, but then I am confident I will not Barron’s article has more on this topic, including an be fired and I can stand the lost opportunities of a rising admission that this type of account is likely to lose a little market. I know that to be true since I did have huge lost money if our forecast for the next 2 years is right. It adds opportunities in 1998 and 1999. the advice that if you wanted to lower the risk of substantial loss to a very low level, in the event of our By exceeding the client’s tolerance for short-term bearish view being correct, you should hedge such a underperformance and getting fired, you commit a traditional strategy by going short the S&P 500 through substantial disservice to yourself as a manager, but even futures or ETFs (exchange traded funds) to a level of more so to the client, who has not only underperformed about 30% of the face value of the portfolio. with you, but then in addition often transfers into a stronger performing alternative, usually doomed to suffer All other GMO asset allocation strategies fall between the consequences of a mean-reverting world by falling these ‘no holds barred’ and ‘traditional’ extremes.

The badly just as the investment strategy they terminated bottom line of this unfortunately convoluted section is recovers. And we have certainly been there and done that there is no one-size-fits-all in this business and never that! will be, and we have no choice but to try and pick our So everyone has to manage his or her own career risk, and way through the minefield. But we can say one thing being too extreme or ‘wrong on your own’ has cost many with certainty: quite a few clients who ask for people their jobs, however right they might eventually aggressiveness in asset allocation are also aggressive in have been. In asset allocation we do not therefore try to firing managers for underperforming more successful push our own instincts on aggressiveness or risk competitors in the short term.

(On the other hand, the job preference; we just try to set the scene in terms of risk and pays well and is a whole lot of fun!) return as accurately as we can. All GMO strategies have fairly precise job descriptions. Our equity strategies must GMO’s Current Positioning and Potential for stay fully invested and within their category for example. Outperformance Our asset allocation strategies also come in various levels of aggressiveness designed to suit different purposes and All of our individual equity strategies are fully invested different risk preferences. At the aggressive end, we have within their respective universes, so our US small cap our ‘mean reversion’ strategy for example (now closed to growth, for example, is not going to sidestep much of the new clients) that is designed for clients who share our disappointment we see coming from the market.

And view that some asset class bets almost certainly will win, some of our riskier strategies, like emerging equity and but with very uncertain timing: This of course requires debt, may have some real weakness if we have even a clients who understand the need for patience, whiff of a financial crisis, which is of course a distinct occasionally of the extreme variety. In this strategy we possibility. However, all of our equity strategies are now are short the S&P 500 and low quality US stocks and long tilted towards quality and large cap from their normal emerging and international stocks, as well as some very positions, even emerging equity, and by the end of March high quality US stocks. or April we will be about as defensive as we ever get or feel we can get within our constraints.

Over 2 dreary At the more traditional end of asset allocation, we have years this should certainly save something. Whether it’s our global balanced strategy that is a fund of funds and a lot or a little will depend on the particular spin of the the type of portfolio we also run for Evergreen. In this decline and global economic and political developments style we do not go short and do not expect to drop below which are more or less unknowable.

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.

GMO Special Topic January 2005 Jeremy Grantham, Chairman Letters to the Investment Committee II Replacement Cost: The Bedrock of Value However hard I find it deciding what to believe about the c) Similarly, inflation changes cannot affect the real market, I always find it harder persuading others that replacement value. Yes, investors hate inflation and what I believe makes sense. Several aspects of how the P/Es tend to rise and fall with inflation, but that is market actually works seem initially counter-intuitive. very different – a purely behavioral response that can Many arguments that seem to me provably wrong in be exploited with a value manager's longer-term contrast, seem to be easily believable and appealing. orientation. For stocks are real assets and pass through inflation, or deflation for that matter. If that The total market must sell at about the total cost of were not the case, rapid inflation countries like Brazil replacement. It may be hard to calculate, but if we could and Turkey would have been selling at under one know the true replacement value it would be the fair time earnings long ago. value of the market. If assets in the market sell away from replacement or fair value, an arbitrage takes place d) Even imbalances in supply and demand, which will that is central to the effective working of capitalism. of course change short-term stock prices, cannot affect replacement value. Take for example the If the market sold at three times replacement value for supply demand imbalance suggested by the looming example, companies would sell a billion dollars of stock, baby boom retirees. (See the article authored by Rob build a new billion dollar plant, and have it sell Arnott and Anne Casscells in the March/April 2003 immediately at $3 billion in the market. Hugely Financial Analysts Journal titled “Demographics and encouraged, they would sell more stock and build more Capital Market Returns”.) There may be more retired plant until they drowned in fiber optic cable, for example, sellers of big houses and stock portfolios and fewer at which point profits and stock values would crash back younger buyers than there used to be, but to replacement cost or below. replacement cost does not change. Only if you never need any new houses or any new chemical plants can Conversely, if the market sold at half replacement cost, they sell below replacement value for any length of which company would build a greenfield plant when it time. Any demand for a new chemical plant, for could buy a competitor's plant in the market for half the example, will require that the system moves back to price? No new plants would be built until eventually a replacement cost. (Although I'm willing to concede shortage developed and then profits and stock prices that land, which really has no replacement cost, may would rise, until the new profits justified a new plant be an interesting special case.) selling in the market at full price. e) If the market averages replacement cost in the long If fair value in the stock market equals replacement cost, term for the arbitrage reasons given above, then the then it surely follows that: market must also equal the average profit margins of a) changes in short or long interest rates cannot change the corporate system times the average P/E. This is replacement cost and are therefore irrelevant to fair how GMO approaches fair value: normal margins market value, in contrast to the easy appeal of the times normal P/E. This method results in a fair value “Fed Model” (surely the Fed doesn't actually believe very close to the Federal Reserve's Replacement this model?) that argues exactly the opposite. Value (or Tobin's Q), both around 725 on the S&P, currently trading just under 1200. Another b) Tax changes such as capital gains tax rates are also clearly reasonable way of estimating fair value is to take a irrelevant to replacement cost and fair market value. long-term normalized P/E. Robert Shiller of Yale suggests taking the last 10 years' earnings after around the two black trend lines are remarkably inflation adjustment divided into the market price. similar and should be. For the record, the 10-year This should also come close to Tobin's Q. Exhibit 1 P/E is more predictive of future market performance shows the actual market since 1925 compared to than 1-year P/E as you might expect, and Tobin's Q is these two measures. The top panel shows the market the most predictive of all. Exhibit 2 shows the price moving around the Fed's estimate of predictiveness of Tobin's Q over the following 10-replacement value, and the bottom panel shows the year holding periods. Every month since 1925 the market against the price to 10-year trailing earnings market's price to replacement cost ratio is put into calculation. Two more clearly mean-reverting five bins from cheap to expensive. The results, as series would be hard to find. The deviations you can see, are surprisingly good, especially for expensive markets: 20% of all the time since 1925, Exhibit 1 identified as expensive on Tobin's Q, investors only U.S. Equities Are Still Materially Overpriced received 1.2% a year real return!

Tobin's Q: Price to Replacement Cost f) If fair value = replacement cost and since total profit of U.S. Corporate Assets (Federal Reserve) margins are very different in booms and busts, it 3.0 follows that: Replacement cost = high profit margins × low P/Es 2.5 or = low profit margins × high P/Es 2.0 If P/Es don't move exactly opposite the changes in profit margins, then the market could not equal 1.5 replacement cost.

This is just another way of saying that you can't use P/Es 1.0 -39% to meaningfully compare high profit margin periods with low margin periods. When you read in the paper today 0.5 that the P/E of the market is reasonable, it completely misses the point that margins are unsustainably high. In 0.0 the next 2 years, I quite expect more market pain from 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 01 falling margins than from falling P/Es. 1.4 Standard Deviations Expensive Exhibit 2 Price/10 Year Average Real Earnings 50 Quintiles of Tobin’s Q to Predict 10-Year Returns 12.2% 10-Year Real Return 35 9.7% 9.7% 25 Today 20 3.8% 15 -34% 1.2% 5 Cheapest Second Third 20% Fourth 20% Most 20% of 20% Expensive 0 History 20% of 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 01 History 1.3 Standard Deviations Expensive Tobin's Q for S&P 500 When Purchased Note: Tobin's Q renormalized to average one.

Source: GMO, Standard & Poor’s, Federal Reserve. As of 9/30/04 Source: GMO, Standard & Poor's. Data: 1926-2003.

GMO 2 Letters to the Investment Committee II, January 2005 It is worth noting though that the requirement for Exhibit 3 aggregate profit margins to move against aggregate P/Es How the Market Actually Works is counter-intuitive for many investors because it is obviously not the case for an individual firm. If a single P/E Ratio of U.S. Stock Market company in an industry has consistently higher returns 40 Record Margins (Claim ed) than its competitors (say, Microsoft) then we can all agree it is worth a consistently above average P/E. But for the 35 whole economy it works the other way, and one of the important jobs for financial analysts is to normalize for Record the economic cycle. The difference is that the profit Margins Record margins of the total economy are very reliably mean- Margins reverting, as new money floods in, attracted by higher 25 returns, whereas a great franchise like Microsoft always has an element of monopoly that resists regression. 20 A cross check on this logic is to realize that if return on capital is unusually high for the whole corporate sector, 15 then to be in balance and remove the arbitrage opportunities, the system must also offer an equally high 10 return to stock investors, and the only way returns to investors rise is for the P/E to fall. It may be superficially counter-intuitive, but that is how it works for the financial system to be in balance. Depressed Margins For fair price to equal replacement cost, then it must Dec- 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96 01 follow that P/E should be perfectly negatively correlated Actual correlation betw een profit m argins and P/Es is +31.8% with profit margins. Exhibit 3 shows how the market actually works. Investors extrapolate good times into the Source: GMO. As of 9/30/03.

indefinite future by assuming high margins will remain. Extrapolation, he explained, is the “convention” we adopt In this way they double count: they take peak margins to deal with an uncertain future even though we know times peak P/Es (instead of low P/Es) and they multiply from personal experience that it is not the case. depressed margins times depressed P/Es. This is particularly true in great bubbles, when the market For the record, extrapolation of today’s conditions gives appears to believe in golden new eras. The market peaks analysts a sure way of clustering together and avoiding of 1929 and 1965 both had near record profit margins and the potentially terminal career risk of being wrong on near record P/Es. But neither came close to the recent your own. The double counting that extrapolation causes bubble of March 2000 where margins were at a new is the reason that the market is far more volatile than it world record and P/E went to 35 times actual earnings should be. Robert Shiller noted over 15 years ago that if compared to a previous high of 21 times. (Not a bad you had clairvoyance about the future the market would increment!) The market low of 1982, before our current be quite stable because the economy is mostly a bubble began, was exactly the reverse, with very battleship that is reluctant to move off trend, resulting in depressed corporate profits being multiplied by the a surprisingly stable stream of earnings and dividends. lowest P/E – 8 times – in 40 years. The actual measured He noted that the real market never learned how stable correlation between profit margins and P/Es is far from a reality is. Greed and fear, career risk, and above all, perfect -1. In fact, it cannot even get the sign right! The extrapolation and the P/E profit margin double counting measured correlation is +0.26. have introduced several times the necessary modest volatility that must be caused by an uncertain future.

The market is therefore on average extrapolating, not normalizing. This extrapolation which takes place, Finally, if you have time, read Valuing Wall Street: instead of the normalization that is required to make the Protecting Wealth in Turbulent Markets by Andrew market stay close to fair value, is pure behavioralism and Smithers and Stephen Wright, the final long word on the was noticed (as always, it seems) by Keynes. importance of replacement cost.

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.

Letters to the Investment Committee II, January 2005 3 GMO