Mar 2006

The Temporary Revenge of the Pessimists

I've just worked through Dimson, Marsh and Staunton's magnum opus, "Triumph of the Optimists. 101 Years of Global Investment Returns", which must now supersede all previous surveys of such matters, so comprehensive and magisterial is its achievement. Everything anyone might wish to know about stock, bond and currency returns since 1900 is there and a lot more besides.

I can summarise the book in a couple of sentences. "Equities good, government bonds disappointing." and "Anglo-Saxon economies have been good places to commit long term capital." This is all uncontroversial stuff, albeit of value when presenting "the case for equities" to first time investors. As always though, the challenge for professional investors is to work out how to make actual forward-looking money from the backward-looking data and I want to highlight the facts in the book that surprised me and made me think harder about that challenge.

The authors analysed returns from 16 capital markets between 1900 and 2000 (101 years!). These markets made up c 90.0% of global equity market capitalisation in 1900 - interestingly not far from their combined weight at the end of the period too. The other 10.0% of global market capitalisation in 1900 comprised then fashionable emerging markets like China and Russia, which shortly after crashed and burned to nothing. Things will be different this time round - won't they? The real return on equities was positive for all 16 countries, typically at a level of 4-6.0% per annum compounded. Belgium was the poorest performer, with an annualised real total return of 2.5% and Sweden the best, returning 7.6% per annum real. Nothing much good has ever come out of Belgium. The difference between these two performances is eye-popping. A local unit of Belgian currency invested into domestic equities in 1900 grew to 12.3 units by 2000, meanwhile, a Swedish currency unit would have attained a value of just under 1,700. The gap illustrates the power of our old friend - compound interest. Relatively small divergences in annual, compounded returns over many years result in dramatic differences in terminal wealth.

Over the period as a whole it appears that resource rich countries with an Anglo-Saxon business culture did best, with the top five markets being, in ascending order, Canada, US, South Africa, Australia and Sweden. The UK and Netherlands were 6th equal, with annualised total real returns of 5.8% - both trading-oriented societies. Away from former British colonies, I didn't know that the oldest, formal stock exchange in the world is the Paris bourse, established in 1724, 63 years before Amsterdam. Stocks had been dealt in an organised way in London since 1698, but it wasn't until 1801 that the LSE obtained its constitution and own building. The early establishment of Paris did not guarantee competitive returns in the Twentieth Century, though, with France's 3.8% per annum rate only half that of the best performer. As a general proposition, the worst equity markets were those associated with countries that lost wars and particularly with the inflation that accompanies lost wars. Italy, Germany and even Japan were amongst the poorer performers. Add France - arguably also the loser of two Twentieth Century wars - to those three and you have the countries in the sample of 16 with the worst inflation experiences in the century. Not surprisingly, the government bonds of these four also did much worse than the average.

I assumed that the relevance of natural resources had waned with the century, as the real prices of many commodities fell during the period. The recent gains in commodity prices and these competitive returns from resource rich markets are making me rethink that assumption. It looks clear that access to, or control of, key resources is a good correlate for strong capital market returns. Both Germany and Japan are famously short of natural resources, particularly energy. Indeed, their bellicosity, that in the end damaged their capital market performance, was in part inspired by their desire to gain preferential access to strategic reserves of resources. It would be a brave bet to assume the 21st century will work out differently, but one man who seems to be prepared to make that call is Sheik Ahmed Zaki Yamani, the former Saudi oil minister. As chairman of the Centre for Global Energy Studies, he recently said "We have seen the future: the main source of energy will be hydrogen. It will be the end of the oil era." Mind you, at the same time he also noted of oil - "The price will remain high for some time until the major oil consumers will be able to be independent from oil, especially from the Gulf region." BP is not yet a short, then, except perhaps on a fifty year view.

It is worth questioning the future of dominant industry sectors, like oil, though, because the 101 years saw radical changes in the constituents of stock markets. In 1900 one sector made up 50.0% of the value of the 100 largest UK corporations and 63.0% of the top hundred US. By 2000 these weights had declined to 0.3 and 0.2% respectively. The industry in question is, of course, railroads - the great growth industry of the Nineteenth Century. Back in 1900, Oil, Pharmaceuticals and Information Technology had nigh on zero weights in the US/UK markets, but over 25.0% of the whole by 2000. What really interests me is not so much which sectors emerged over the course of the period, because I am not so optimistic about my ability to pick winners in, say, the nanotechnology space. What really interests me is which industries saw their share of market capitalisation hold reasonably stable over very long periods, because these have, therefore, provided the most certain and least risky areas through which to have participated in the long term propensity for equity markets to go up. Three UK sectors really stand out as having "legs". These are Banks and Finance (15.4% of the 1900 market by value and 16.8% by 2000), Utilities (3.1% and 3.6%, with a distorting period in public ownership, of course) and Breweries and Distillers (3.9% and 2.1%). I expect that in another 100 years time these crucial economic functions - financing the growth of national and international economies, supplying energy and water and slaking the thirst of the workers - will still be rewarded with substantive stock market weightings. As is well known, dividends are a crucial component of long term returns. The book reinforces this truth. In the US, nominal total returns per annum from equities were 10.1% and almost identical for the UK. However, with dividends excluded those nominal returns fall to 5.4% per annum for the US and 5.1% for the UK. Nearly half the nominal returns have been delivered in the form of dividends. Analysis of dividends and, particularly dividend growth does present something of a wake-up call for investors though, at least for those hoping to get rich quick. The fact is dividends grow more slowly than I expected - meaning that long periods of time are often required for steady dividend growth to work into really attractive long run returns. Of the 16 countries in the sample, only 7, less than half, delivered any real dividend growth at all over the period. Sweden came out top, with real dividend growth for the 101 years of 2.3% per annum. Perhaps not coincidentally, Sweden was, as we have seen, the best performing market over the period. What is it that I've missed about Sweden, an economy I've only ever registered in the past as the unholy spawning ground of ABBA and Sven? The worst dividend record amongst global stock markets is that of Japan. Japanese dividends actually declined by an average of 3.3% real per annum over the century. Real dividend growth in the US averaged 0.6% per annum and 0.4% in the UK - not obviously exciting, but much better than the average.

Again, the proposition that dividends can not grow faster than real GDP for any sustained period (otherwise profits as a proportion of GDP would rise indefinitely) is confirmed. Indeed, there was only one economy that delivered real dividend growth above GDP per capita growth over the century - South Africa. Nowhere else was high GDP growth associated with high dividend growth at all. This is another way of observing the counter-intuitive truth, that high GDP growth is not necessarily correlated with high returns from capital markets. Over the twentieth century, the three fastest growing economies among the sample were Japan (3.9% per annum GDP growth per capita), Italy (2.8%) and Spain (2.6%) - yet real dividend growth was negative for all three and each delivered total returns for the period toward the lower end of the sample. As the authors say - "GDP can grow without generating wealth gains to equity holders." This is a very important consideration and one thinks of the poor returns from Chinese equities over the past decade as further corroboration. The Chinese Shanghai Index is still below its 1997 levels, despite all that economic growth. Capital-friendly social attitudes and laws, as well as decent growth, are required to create wealth, at least for minority owners.

The United States - perhaps the most capital-friendly society on the planet - poses major questions for global investors today. The Twentieth Century very much belonged to the USA. Over the period, the US equity market delivered 6.7% per annum real, turning $1 into $711. The World excluding the US returned 5.2% per annum real, growing $1 into an impressive, but much smaller, $162. Meanwhile, the US Dollar rose in value against 13 out of the 15 other currencies. Only the Swiss Franc really made headway, rising 1.2% per annum on average against it. Sterling declined against the mighty greenback by 1.2% per annum since 1990. Is it obvious why Sterling's bear market should have come to an end? We know that many UK institutional investors are significantly underweight US assets. This may well turn out to be correct, but it goes against the lessons of history.

Dividends matter not only to total returns, though. The evidence from the book strongly supports the received wisdom that yield and value-oriented investment strategies outperform. Owning low price/book, high dividend yielding equities has been the way to maximise wealth over the last century. Between 1926-2000, US high yielders gained 12.2% per annum nominal, compared to the market average of 10.6% and a low yield index of 10.4%. One dollar in those high yielders grew to $4,948 over the period, again startlingly more than one dollar in the low yielders, which made $1,502. For the UK the figures are even more extreme. This time between 1900-2000, UK high yielders turned £1 into £61,235, growing at 11.5% per annum, while the lowest yielding amongst the top 100 companies managed 8.6%, turning £1 into £4,046. Why should this be and will the superiority of "value and yield" hold into the new century? It appears that in the long run the bird in the hand - the certainty of the cash dividend- is worth more than the reinvested earnings of the "growth" stock, that investors undervalue today's cash return, compared to tomorrow's possible capital appreciation. This appears plausible, but the effect has been so well known, for so long that I find it hard to understand why it has not been arbitraged away.

I could double the length of this précis by reviewing bond and bill market performance. I content myself with a couple of observations. First, perhaps the real surprise is not that government bonds have performed so poorly relative to equities, as everybody knows. UK equities returned 5.8% per annum real total returns between 1900-2000, while UK gilts managed only 1.3% per annum (at least a positive real return). What is more surprising is that bonds did so badly relative to cash, or treasury bills. Across the sample of 16 countries, bonds beat bills by only 0.5% per annum - in the US by 0.7% per annum and the UK a measly 0.3%. In other words, the return on bonds looks miserable, given the very considerable volatility of their capital values, not only compared to equities, but also against the certainty of just leaving savings on deposit. Whatever, we must concur with the authors' conclusion that "real returns achieved on bonds over the Twentieth Century turned out to be lower than investors' ex ante expectations." This is the reason that the name of the book is "Triumph of the Optimists". Risk takers have been rewarded, while "safe" investments have been poorly rewarded. You have to wonder why anyone would ever own a bond.

One answer to that last question, of course, is that returns depend on the price that you pay to access any asset. The authors of the book had previously worked on an analysis of the "small company effect" both globally and in the UK and their conclusions led, in part, to the launch of the Hoare Govett Small Companies Index, back in 1987. They note ruefully that the fanfare that greeted their conclusions about the superiority of investing in small capitalisation companies coincided uncannily with a temporary peak in small cap performance and they proceeded to underperform through the 1990's. Of course, a similar effect is apparent in the relative performance of all stocks and bonds since 2000. For instance, the UK gilt market has a total return of some 44.0% since the start of the new millennium, while the FT All-share has returned only 12.0%. "Not fair", the equity bulls might claim - stocks had got overpriced by the end of the 1990's. But these starting points really do matter. As Dimson et al point out, the dividend yield on the world equity market in 1990 and again in 1950 was over 5.0%. In 1950 equities were certainly offering a dividend yield higher than available on government bonds. US governments yielded as little as 1.9% in 1946, before embarking on a 36 year bear market that carried yields all the way up to 15.0% in 1982 and destroyed government bonds' reputation for "safety". By 2000, the world equity index was itself yielding an all time low, coincidently of 1.9% too and government bonds were a pariah asset class.

Our bet is that markets are some way through working out the overvaluation of Equity back in 2000 and the concurrent undervaluation of the "certainty" associated with Government Bond investing. The government bond content in the Lindsell Train Investment Trust is down to c20.0%, the lowest since we launched in early 2001. As long term equity bulls, with our convictions strengthened by this terrific book, we are sure our bond exposure will decline further over the next few years. As pragmatists too, though, we are grateful that we have earned high real returns from our fixed interest holdings over the past 5 years, while equities adjusted after the fin de siecle excesses.


Nick Train
Apr 2006