What do practitioners need to know about financial market history? In July 2015, Cambridge Judge Business School hosted many of the world’s leading academics and a number of senior investment professionals to discuss this question. The result is Financial Market History: Reflections on the Past for Investors Today, edited by David Chambers and Elroy Dimson and published by the CFA Institute Research Foundation. In this interview, Chambers, academic director of the Centre for Endowment Asset Management at Cambridge Judge Business School, and Dimson, chairman of the Centre for Endowment Asset Management and emeritus professor of finance at London Business School, discuss the book and its contents.
Nathan Jaye, CFA: What inspired your book?
Elroy Dimson: David and I share an interest in financial market history, which goes back quite a long way for both of us. More recently, CFA Institute and CFA Society United Kingdom have highlighted the importance of financial history. Partly, it’s a concern about whether people understand where we’ve come from and where we’re heading; partly, it’s dealing with a new generation of investment professionals for whom recent events, even the onset of the global crisis, is a part of history. I think that inspired an interest in making sure that people, in thinking about the future, profit from an understanding of the past.
The conference was held with the book in mind?
David Chambers: Yes. In the aftermath of the 2008 global financial crisis, there were many instances of people reaching back into history to pull out examples relevant to today, particularly examples from the 1930s. There was a sense that we really ought to try to educate CFA charterholders more on the topic.
Dimson: The model of the conference, which resonated a lot with CFA charterholders, was to bring in scholars from around the world — not to present papers which we would then assemble as a conference proceedings, but to prepare contributions to a planned book. This generated enthusiasm within the Research Foundation, which was in the process of moving toward a new model. Rather than publishing research monographs or literature summaries, they’re commissioning more substantial books.
What was remarkable about the research?
Chambers: The obvious thing is just breadth of scholarship. The whole field of quantitative financial history has really exploded in the past 10–15 years. In part, that is the result of the advance of technology and the fact that there is more digitization of archival documents and more people assembling new databases. As a result, there is more material that scholars are working with.
How do you build data for your research?
Dimson: We do different kinds of research. Let me describe the work done by two colleagues at London Business School, Paul Marsh and Mike Staunton, and myself. From the end of the 1990s, we put painstaking effort into compiling long-term asset-returns series. When we started doing this, there was already a decent series beginning in 1926 for the United States, and we had done a bit of work for the United Kingdom from 1955 onward.
But if at the end of the 1990s you had looked at leading investment textbooks, they projected that the US in the future would experience equity returns and equity risk premiums like the past. That’s slightly dangerous. But, bizarrely, when books were produced for other markets — often editions of major textbooks that were adapted from the original English versions — they would still use the US financial market experience as a guide to what might be expected in other countries.
So we accumulated long-term stock market data for a number of countries. And it became clear that just as Paul and Mike and I had focused on creating a British series beginning in 1900, other people in other countries had done much the same sort of thing. Consequently, our data is a compilation of work done by researchers around the world.
Our contribution was to bring them together and create a unified database which was broad in coverage. We’ve extended it to 23 countries, which we monitor from the beginning of the 1900s to the beginning of 2017. The DMS (Dimson–Marsh–Staunton) dataset has become quite influential in terms of helping people make judgments about the past and what might be expected for the future.
What was discovered about historical IPOs?
Chambers: Researchers accessed IPO data by looking at prospectuses that were published in newspapers and stock exchange records. These kinds of studies have now been done for the UK and the US, as well as Germany, Italy, Belgium, and the Netherlands, among others. These studies show there was an awful lot of IPO activity 100 years ago and even stretching back into the 19th century. Interestingly, in the beginning, most of this activity was not in stocks but in bonds. Companies first and foremost issued bonds, primarily because that was what income-conscious investors demanded. And it was only as we entered the 20th century that the world began its move to a position where stocks dominated investor portfolios.
Researchers have also looked at underpricing (the tendency of IPOs to jump in price on the first day of trading). It is a fairly common finding in modern IPO markets. But it appears not to have been as prevalent in the past, if you go back 50 or 100 years. Something may have changed in the way that modern capital markets evolved to bring about more extreme underpricing. The relevant major shifts have been the growing power of investment banks and the rise of institutional investors, both of which benefit from the economic rents represented by higher underpricing.
What does archival information on investor portfolios reveal?
Chambers: The study of how investors behaved in the past is a relatively new area. It is critically dependent on the survival of investor records. Hence, our analysis on the economist John Maynard Keynes became possible because we were able to find full details of his portfolios and his stock transactions. One of the later chapters in the book examines the historical rise of institutional investors in the US and the UK. Another chapter looks at the related area of banks and at financial and banking crises. There are a multitude of bank records that researchers are accessing, both in commercial banking and investment banking. For example, scholars are examining prior episodes, when banks were permitted to conduct both investment banking and commercial banking business. Against the background of the recent global financial crisis, they are considering whether these universal banks are a good thing or not.
How has technology aided your research?
Dimson: If we had been researchers 10 or 20 or 30 years ago, we would have been getting old archival material or old newspapers brought out one by one. In most cases, you would not even have been allowed to put it onto a photocopier. It’s not to do with copyright that copying was prohibited. Bending the spine of a bound volume would have been unacceptable and bright light would have been equally undesirable. In many cases, all you could do was get out your pencil and paper and make notes.
Being able to take your laptop in made things easier. Being able to take high-quality photographs makes a big difference. So in many cases, even when material hasn’t been digitized, it’s possible to go home with a lot of material and sort it all out in a way which simply wouldn’t have been possible a decade or two ago. Assistants can now be employed to collect valuable material — tasks that would have been unachievable before.
Chambers: An awful lot of newspapers are digitized now, so you don’t have to go to the archives and reach down for a dusty volume from the shelf. Additionally, researchers are applying text analysis to newspapers to develop measures of investor sentiment and to assess their impact on security prices.
What are potential biases affecting historical financial data?
Dimson: On the work I did with Mike Staunton and Paul Marsh, we were initially interested in creating an asset-return history for the UK. We didn’t trust the then-standard data series for long-term UK investment returns. When we re-collected the data ourselves, the previous study was very misleading.
The bias worked as follows. The people who originally collected that data wanted it to resonate with users. At the time, the standard British stock market index was the geometric, equally weighted Financial Times (FT) 30-Share Index. They decided that, as of the 1935 birthdate of the FT 30-Share Index, they would create an index which was (correctly) weighted by market capitalization.
Then they had to ask themselves: How would they go backwards before 1935 when they were trying to make the constituents look the same as the FT Index? The answer was to use the shares which were destined in 1935 to enter the index. These were companies that by 1935 had become big, and they started the index at the end of 1918 and rolled it forward. They included companies — some big, some medium, and some smaller — which became 30 major companies later on. That’s what we nowadays refer to as survival bias.
When you put together an index using historical data, it’s crucial to replicate what a live commercial index would look like. There can be no looking ahead, because when a newspaper publishes an index, it doesn’t know what tomorrow’s newspaper will contain. So there’s a trap, which the original researchers fell into. Today, we would not make the mistake of creating an index biased by company survivorship.
Understanding potential biases helps people to be discriminating users of historical data. That’s why there is a complete chapter in this book addressing some of the dilemmas that can come about through blindly using data just because it’s there.
The solution is being vigilant?
Dimson: In our case, we were vigilant. We thought the data was wrong, and so the solution was not simply vigilance. The solution was collecting data all over again. We collected a broader index and one which didn’t suffer from these sorts of biases. It is vital for financial historians to know what the potential problems are and then if necessary circumvent them. It can be a task that requires a lot of effort.
What can we learn about the history of currency trading?
Chambers: This is interesting for today’s investors, because quite a lot of research and empirical analysis have been done on the returns to currency trading over the past two or three decades. In general, they’ve shown that seemingly naive trading strategies, like carry and momentum, do quite well.
The chapter on currency trading is based on work by Olivier Accominotti and myself. We looked back to the very dawn of modern currency trading, which was at the end of World War I in London. We can date the beginnings of the modern currency market from then, when forward markets were first established on a continuous basis and people stopped trading currency through bills of exchange and moved to trading by telephone directly.
By going back and adding another 50 years of data, we can do an out-of-sample test as to whether the carry-trade strategy does in fact generate profits. By testing this strategy against this new data from an earlier period, we conduct an out-of-sample test. In fact, we do find that it is profitable. The fact that this simple carry strategy makes money indicates that it’s robust; it is not just a feature of the way that modern foreign exchange markets have developed in the very recent past.
What did you learn about bubbles?
Dimson: The general belief is that if you’re in a bubble, it’s going to pop. So it’s a good idea to time the market and get out. We know there is a large amount of research saying that market timing is really difficult and advising against it. But, at the same time, we know that investment performance depends on your entry and exit point. Even people who believe that timing is difficult often believe that when it’s a time like late 1999, when you saw ridiculous prices being paid for dot-coms, then there’s obviously a bubble. Or if it’s late 1974 and companies are selling at incredibly low price–earnings ratios, it’s time to buy. The belief is that in these extreme cases, bubbles are identifiable. Will Goetzmann [a professor of finance and management studies at Yale School of Management] took on board this idea and thought it was testable.
What did Goetzmann find?
Dimson: He proposed a number of experiments. If stock prices double and halve they get back to where they were. If you take an interval in which stock prices have doubled, they must look rather expensive at the end. So the researcher can examine a market that has doubled and can then see whether the market is more likely to double again or to halve (and get back to where the portfolio started). Summarizing Goetzmann’s results, he finds there is pretty much no difference in terms of what happens afterwards, except extreme outcomes are more likely to follow years in which stocks doubled or halved. Obviously, times when stock prices have doubled or halved are volatile periods, so volatility continues.
When he looks more closely, he finds that after one of these extreme periods doubling is slightly more likely than halving. But that’s simply because you expect some reward for investing in equities. If equities had doubled and then halved, you would end up over two years receiving nothing, and yet we know that equities should — and do — give you a positive risk premium. Consequently, you would expect that after a period in which they had doubled they are a little bit more likely to double than to halve again. And that is what Goetzmann finds.
So it’s very difficult to spot a bubble that is going to pop, because the bubble may continue and you’ll then lose out from having engaged in what’s still a variety of market timing.
What surprised you about financial innovation?
Chambers: Some of the essential building blocks in financial innovation emerged very early on, when you would have thought markets were very unsophisticated. One of the most fascinating chapters in the book discusses the example of 18th-century Netherlands, which at the time was the most financially developed country in the world. That’s where the first examples of structured finance and mutual funds appeared.
One example of a structured finance instrument was the plantation loan. There were many plantations in the West Indies, where Dutch investors were involved. They had the idea of floating bonds on the Amsterdam Stock Exchange and investing the proceeds from those bonds in plantation mortgages. Collateral was in the form of the plantation estates but also the proceeds from the sale of the crops. In effect, what we observe is a very early form of mortgage securitization.
Similarly, mutual funds were first floated in the 18th century. This was the first time that individual investors were given the opportunity to buy a basket of securities. Prior to that, they had to pick their own stocks and bonds, but now they were presented with the opportunity to buy a share in a much more diversified fund. The fund might also invest in faraway places, like Russia, that investors wouldn’t have been able to access for themselves.
Both these innovations are examples of important principles in finance — what we would call liquidity creation and asset substitution. They provide us with examples of how such innovations can work to enrich economies and the well-being of citizens, as was the case in 18th-century Netherlands. In the recent 2008 crisis, some people pinned the blame on securitization per se. However, there is not something innately wrong with securitization. As we have seen, this innovation has been around for several centuries. Rather, we should look at the way that incentives can work when these innovations are exploited and the impact that leverage can have.
How much can the past inform the future?
Chambers: Think about the 2008 global financial crisis and then consider the 1929 crash and its aftermath. The characteristic common to both crises is the very sharp falls in asset prices. An asset-price collapse can have extremely deleterious effects on industrial and economic activity, essentially because it undermines investor confidence in firms, it undermines depositors’ confidence in the banks, and it undermines the banks’ confidence in the businesses that they’re lending to.
Initially, the 2008 crisis was following a similar path to the early 1930s. Ben Bernanke, an academic economist and economic historian, was then Fed chairman, and he knew how bad it was in the early 1930s. His research had shown him how much of a contribution asset-price deflation made to the Great Depression. So he set about trying to counteract this most recent rapid fall in asset prices by implementing quantitative easing, pumping enough money into the economy so that it would support asset prices on a scale never previously contemplated — certainly not in the 1930s. And he did that very successfully.
However, now history cannot help us very much. Now we’re in a situation that is different from the past, because we’ve never been through a period where central banks have pumped so much money into the economy. As a result of quantitative easing, we’re in a completely different game. And there’s not much that we can, on this particular topic, pull out of the history kit bag. Sometimes history can be extremely helpful, but sometimes it will be limited as to how it can inform us.
Dimson: People may ask very simple questions like, What have asset returns looked like at times when inflation was very low? If we look across countries and across periods, we can find lots of cases where inflation was low or where inflation was negative. About one-third of all country years for which we have the data are years in which inflation rates were about zero or negative. The same goes for real interest rates: About one-third of all country years had real interest rates that were zero or negative. So there are similarities, which can be helpful. They’re helpful when people have a view about what you might expect in those different conditions.
But, of course, there can be differences. For example, when real interest rates were low in the past, they were quite often low because inflation was very high at a time when interest rates were also large. The gap between the two meant that there could be a low or negative real interest rate. So there are differences in the macro-environment as well as similarities.
What directions might future research on financial history take?
Chambers: We know a lot about stocks, bonds, and cash. We even know (thanks to the work Elroy has done with Christophe Spaenjers) about how collectibles, such as wine, art, and stamps, perform as investment assets. The one major area where we still lack good-quality data is real estate. We know something about land prices. But what investors are interested in are the total returns on property, which includes income as well.
As the 20th century developed, property became not just simply a collection of investments in agricultural land. Residential and commercial properties have become very important. So we have an ongoing research project that aims to bridge the gap. By accessing archival records of important investors in real estate, we are trying to develop a total return series for real estate over the very long term.
That’s important, because real estate was a substantial part of people’s portfolios, along with stocks and bonds, years ago. It’s still an important part of some investor portfolios today. I’m talking about institutional investors, not only individuals. It was an important asset class and remains one today.
This article originally ran in CFA Institute Magazine.
If you liked this post, don’t forget to subscribe to the Enterprising Investor.
All posts are the opinion of the author. As such, they should not be construed as investment advice, nor do the opinions expressed necessarily reflect the views of CFA Institute or the author’s employer.
Image credit: Thomas J. O’Halloran, photographer, courtesy of the Library of Congress