Understanding the risks embedded in a portfolio is central to providing value to clients. Traditionally, risks have been measured statistically – with standard deviation or value-at-risk (VaR). The shortcomings of those metrics, however, have been well documented. In response, a new generation of analytical tools has emerged that allow advisors to assess risk through scenario analysis – looking at portfolio outcomes through the lens of a storyteller.
The last two books on the financial industry that I reviewed for Advisor Perspectives emphasized the importance, in attempting to understand economic behavior, of “stories” and “narratives.” In John Kay’s book, “Other People’s Money: The Real Business of Finance,” he writes, “We deal with radical uncertainty through storytelling, by constructing narratives.” Later, he elaborates: “The reality of market behavior … makes little use of probabilistic thinking but relies on conviction narratives – stories that traders tell themselves, and reinforce in conversation with each other. Such narratives are the means by which we cope with radical uncertainty – the unknown unknowns.” Nobelists George Akerlof and Robert Shiller, in their book, “Phishing for Phools: The Economics of Manipulation and Deception,” use the “stories” theme repeatedly. As they write, “people tend to think in terms of stories … ‘mental frames’.”
I will examine how one software product, HiddenLevers, has implement a scenario-driven methodology to portfolio analysis. But, first, let’s look at the advantages of scenario analysis over traditional risk measurement.
A story about narratives
A little over a year ago, I attended a conference in Hong Kong of INET, the Institute for New Economic Thinking, hosted by the Fung Global Institute (now the Asia Global Institute). INET is a reformist economic policy think-tank, founded by George Soros and others in response to the global financial crisis of 2007 and 2008.
On the day that I attended, John Kay was also at the conference. After an address by a luncheon speaker, Kay spoke up from the audience. He said, as he does in his book, that “risk” was something that takes the form of “narratives.”
After lunch I attended a session in which the lead presenter, I’ll call him Professor X, began by referring to Kay’s comment. Then he shrugged and made a gesture of waving his hands in the air. His clear point was, “What can you do with that?”
Professor X and a passel of his graduate students then launched into a presentation, as replete with mathematical formulas as any I’ve seen, on the subject – as I recall – of risk premiums in foreign exchange trading. I attempted to engage with them on the question of how there can objectively be risk premiums for foreign exchange trading when it is a zero-sum game; but my simple question was so buried in their talk under layers of mathematics – like the pea under the princess’s bed – that it was hard to discuss with them.
Professor X’s shrug and wave of the hands seemed to mean, “You can’t do mathematics with narratives.” To which I would say, so what? Mathematics is a tool; you use it when it is a good fit to your problem and can help solve it. But it’s not obligatory.
The scenario approach to envisioning the future
Nevertheless, mathematics – or at least arithmetic – does arise naturally in many contexts. The scenario approach to investment planning is a combination of narratives with mathematics when it is needed. If done right, the math doesn’t drive the approach; the tail shouldn’t wag the dog.
I’ll admit to some residual bias in favor of the scenario approach. I was an author of an article that appeared in The Journal of Portfolio Management back in 1980. At the time, I was actually employed doing research on renewable energy. It was during the first (and temporary) phase of widespread enthusiasm for the field, after nearly tenfold increases in the price of oil in the 1970s. I tell the story in my book The Big Investment Lie. It was my first attempt to get out of the deeply unsatisfying field of institutional investment consulting. Renewable energy seemed like a productive field to get into.
But before the switch, I had been a partner in a four-man consulting partnership. We were trying to tout the idea, with our mostly investment-manager clientele, of using scenarios to plan investment strategy. I personally thought it could be a way to make what we were doing less brain-dead. We held a conference on scenarios for our clients and invited – and paid – several fascinating speakers. Unfortunately, although our business was otherwise reasonably successful financially, the conference wasn’t. Furthermore, we hadn’t fully worked out exactly how we would use scenarios.
The stress-testing approach
More recently, scenarios have played an increasingly important role in financial risk management after the (I thought, inevitable) failures of mathematical models, like VaR, and formulas used to rate collateralized debt obligations (CDOs).
Scenarios in which financial institutions may have highly negative experiences are the stuff of the “stress testing” that banks are now required by regulators to do. The scenarios can be created mainly in two ways. They can either be an actual sequence of negative occurrences in the past, or hypothetical scenarios that are created by brainstorming.
If a stress scenario comes from a slice of historical time, it is relatively easy to run a stress test. All of the values of relevant parameters come directly from that slice of history. Since so much data is recorded and available, a stress-tester can employ those values, run the historical data through the current situation in a financial institution or a portfolio on a computer and see what happens. This, of course, requires a certain creativity, but it is more realistic and provides better hypothetical information than a model like VaR. It also enables you, or should enable you, to think seriously about world events, rather than just statistics.
Alternatively, a person or a team of people can create hypothetical stress scenarios without reference to actual history. These scenarios could be events that never happened before, like “Grexit,” the exit of Greece from the euro zone.
When you create a hypothetical scenario out of whole cloth, it’s much more challenging to run your financial institution through a stress test. You don’t have all of the relevant contemporaneous data, as you do if you use a financially stressed period in history; you have to backfill that data somehow. The obvious approach is to use adjusted data from some similar event that did occur in the past, for example the Russian default of 1998. You might also look at how the relevant variables co-varied over historical periods; then if you hypothesize a possible sequence of values of one of those variables in the future, you might be able to construct hypothetical sequences of simultaneous values of other variables.
The HiddenLevers approach
Several companies have, to one degree or another, started to offer the scenario stress-testing approach to investment managers and advisors. I looked at one of them, HiddenLevers. Others that offer some form of scenario-based stress testing, according to HiddenLevers’ cofounder, Praveen Ghanta, are Rixtrema, Bloomberg, and RiskMetrics. I interviewed Ghanta, as well as two of HiddenLevers’ clients, and watched a video of one of their presentations to clients that they group under the heading of “War Room Webinars.”
The idea is to test a client’s portfolio’s vulnerability to a variety of potentially stressful scenarios (and some that might not necessarily be stressful but might have a positive impact). HiddenLevers’ approach is to track the values of a range of macroeconomic variables called “levers,” such as U.S. home prices, the price of oil and corporate IT spending. They run regressions of the prices of individual stocks and groups of stocks (based on a common industry) against these levers (and against the market as a whole) to see how the stock prices have correlated with the values of the levers.
Then a series of scenarios is created. For example, “Iran Conflict,” “China Crash,” and “Hyperinflation.” For each scenario, values of the levers are conjectured. Since the stock prices have been correlated with the levers, each scenario gives you, more or less, the impact on stock prices, or at least on industrial groupings of stocks.
Initial skepticism
When I first looked at this concept, I was skeptical. It involves running literally millions of regressions, each lever against each stock, for each time period that could constitute a scenario. But at least HiddenLevers mentions right up-front their methodology for weeding out spurious correlations.
I also remembered a scenario that had happened to me and wondered if you can really predict anything. It must have been on the precise date January 17, 1991. I was on the way to my bank. I had recently realized that I finally had enough money, after fits and starts, to begin doing some serious investing in the stock market. I presume I was on the way to the bank to wire-transfer funds to invest in Vanguard’s S&P 500 fund.
As I was driving to the bank and listening to the radio, it was suddenly announced that U.S. Air Force planes were on their way to Baghdad to commence bombing. There had been no warning of this attack before the fact, but Saddam Hussein’s Iraqi army troops had advanced over the border of Kuwait and were occupying that country. The bombardment on January 17 was the first salvo in the Gulf War, the U.S.’s and allies’ ultimately successful effort to expel the Iraqi troops.
When I heard this, I thought, “Should I be making an investment in the stock market now? Isn’t this the kind of situation in which uncertainty and risk are elevated?”
But then I thought, “Nah, who knows?” and went ahead with it.
Well, in two weeks the S&P was up 10%, in a month it was up more than 15% and in two and a half months it was up more than 20%. Could this have been foreseen, even assuming the scenario of a war in the Gulf?
I doubt it; though I was surprised to notice that in a HiddenLevers war scenario, an equity portfolio did indeed benefit. Perhaps this is because it was taken from war scenarios in the past. It would not surprise me much if the typical response of equity markets to a war has been to rise.
Once HiddenLevers has constructed these scenarios and estimated the impact each of them will have on individual stocks and stock groupings, they can assess the vulnerability of any particular portfolio to any one of the scenarios. This is one approach to risk assessment, in my view, better than purely statistical measures like portfolio standard deviation or VaR.
Good information
The two HiddenLevers clients I interviewed, both financial advisors, were happy with the service for what seemed to be good reasons. They said it helped them to think about things, and also that the HiddenLevers website was a great source of information. Many of the displays in its analyses are clickable, leading to deposits and sources of interesting data. The clients were also happy with HiddenLevers “War Room Webinars.”
I watched one of those War Room Webinars, nominally on Obamacare, and was impressed. The analyses were thorough and deep and presented the right combination of qualitative and quantitative information. To do this well is an art; HiddenLevers’ cofounders, Ghanta and Raj Udeshi, gave a very informative presentation.
It might even help to identify hidden risks
HiddenLevers offers a way to think about risks, rather than necessarily something that delivers them cut-and-dried on a statistical platter as if they could be scientifically identified. Their sales, however, probably depend in part on the latter interpretation. Their presentation of the material could be better – it has too many numbers with too many significant digits – but unfortunately that’s become so standard that it’s a hard battle to fight. In my view, they should present a range of outcomes, such as what would happen to oil stocks in the event of armed conflict with Iran, in a form something like 20% plus-or-minus 5%, or even plus-or-minus 10%, instead of 21.86%.
I am favorably inclined toward anything that helps people think in a qualitative way about what the future may hold, how it may be influenced, how it will affect the investor and how to invest in light of that. It may not improve investment results – in the sense of a comparison with a benchmark – but at least it’s closer to what investing ought to be about.
And the stress scenarios that HiddenLevers explores should expose vulnerabilities to risk in an investor’s portfolio; perhaps even, exposures to opportunity.
Michael Edesess, a mathematician and economist, is a visiting fellow with the Centre for Systems Informatics Engineering at City University of Hong Kong, a partner and chief investment officer of Denver-based Fair Advisors and a research associate at EDHEC-Risk Institute. In 2007, he authored a book about the investment services industry titled The Big Investment Lie, published by Berrett-Koehler. His new book, The Three Simple Rules of Investing, co-authored with Kwok L. Tsui, Carol Fabbri and George Peacock, was published by Berrett-Koehler in spring 2014.
Read more articles by Michael Edesess
