Myron Scholes on Time: Conundrums of risk management
In this episode, Myron tackles some of the toughest conundrums in risk management: when to de-risk, when to re-risk, and how to think about risk in a real, multi-period world. Myron, in conversation with Phil Maymin, critiques the standard 60/40 portfolio allocation, highlights the illusion of fixed correlations for risk management, points out subtle but common mistakes people make in estimating risks, and offers an innovative perspective on how to move forward.
39 minute listen
- Risk is not just a system of measurement but should be an active component of investment management.
- Correlations among broad asset classes change over time and especially in periods of stress.
- Explicitly managing tail losses and drawdowns may be a better approach in many instances, for example in the area of retirement.
Janus Henderson Podcast
Janus Henderson Podcast
Options (calls and puts) involve risks. Option trading can be speculative in nature and carries a substantial risk of loss.
Diversification neither assures a profit nor eliminates the risk of experiencing investment losses.
Fixed income securities are subject to interest rate, inflation, credit and default risk. The bond market is volatile. As interest rates rise, bond prices usually fall, and vice versa. The return of principal is not guaranteed, and prices may decline if an issuer fails to make timely payments or its credit strength weakens.
Volatility management may result in underperformance during up markets, and may not mitigate losses as desired during down markets.
Correlation measures the degree to which two variables move in relation to each other. A value of 1.0 implies movement in parallel, -1.0 implies movement in opposite directions, and 0.0 implies no relationship.
The Fed, or Federal Reserve is the central banking system on the United States.
Standard Deviation measures historical volatility. Higher standard deviation implies greater volatility.
Philip Maymin: Welcome back to the Myron Scholes podcast, On Time. Chief Investment Strategist at Janus Henderson, Professor of Finance at the Stanford Graduate School of Business, and Nobel Laureate in Economic Sciences, among many other accomplishments and responsibilities that would take too long to list. Myron shares his unique insights with us here. My name is Phil Maymin, and I have the pleasure of working with Myron here at Janus Henderson. These podcast episodes are aimed at sophisticated investors and those who wish to be sophisticated investors and they’re intended to be thought provoking and perhaps, even controversial. We hope you leave each episode with more questions than you started and we invite you to send feedback or questions to Myron at email@example.com. Today’s episode is about conundrums in risk management.
Myron Scholes: Thank you, Phil. I appreciate your introduction. In finance, we describe three tools of risk management. These are, A, reducing risk by holding a larger portion of your portfolio in assets such as near default bonds or money market accounts. B, diversification of risks by holding a cross-section of assets – a strategic portfolio or model portfolio, as I described in earlier blogs. And C, insurance of the downside by buying put options or holding cash assets, plus long call options on portfolios.
The interesting thing, again, is hold more cash reserves. When I started in finance way back when, when I was a young boy of five, but I was going to graduate school in the ‘60s, what we really only knew at the time was risk management was, how much reserves did you set up? Cash and equities, that was model one … reserves and equities. Then we learned about this thing called diversification and Markowitz model, and then Sharpe and his characterization of risks and diversification. And that led me to puzzle about the third component which is insurance. How do you insure the risk of a portfolio? That’s another risk management tool. We talk about insurance in the home, we talk about insurance of the cars, insurance about life, but no one really had talked about insurance in portfolios as a risk management tool. They talked about pricing of options and risk taking in literature, so I became very interested in that as a research topic and that’s where I got together and married these ideas with Fischer Black, who was a colleague and who we worked on these problems together, to come up with how to price insurance by using options. And that was how we got into that third component here of risk management.
The fun, however, begins when we move from a single one-period decision-making model to a multi-period model over time. So I have my risk management, I hold cash, and I hold risky assets. But risk changes. Outcomes change. Needs change, which makes solutions very difficult. What’s the dynamics? How do I change things over time?
I will discuss the cushion first, the reserves first. What cushion to build into investment management by holding safer assets and when to use the cushion to increase risk again. We will see that these are very tough problems. It reminds me of my first wife, who had cushions on the chair and she said, “The questions on these chair…” I said, “When can I use the cushions?” She said, “Never. These cushions are only decoration.” So I said, “Why do we have cushions? I want to use them.” But we never resolved that argument. We will see that these are tough problems to decide and how to use the cushion. We return to diversification issues followed by issues in ensuring the portfolios and how these are different or like each other. Reducing risk by holding more cushion.
Problem one: When to increase the risk again in the portfolio, when to reduce the cushion. If the market does go down and as a result, the investor suffers loss, less of a loss than holding market risk because of the cushion or the cash holdings, there is no theory which suggests when the investor should increase risk once again. With safe assets relatively stable in value and the risk of equity staying the same or increasing a time of drawdowns, the risk of the portfolio falls approximately linearly with the market drawdown. You end up with more and more cash reserves and less and less equities. What to do then is an important unanswered question. There are various rules of thumb, such as rebalancing the portfolio back to target once a month or once a quarter, etc. Some have argued that a rebalancing strategy is a good strategy regardless, for if market prices revert, that is, expected returns are greater with market risks as markets fall in value, to induce investors to come back into the market, what goes down goes up again. So-called buying the dip. The investor will make abnormal profits or realize greater returns, through the rebalancing and risk-taking, use more of the cushion at that time to do so. There is no guarantee, however, that rebalancing will produce abnormal returns for adding to risk if the market were to go down, might proceed further drawdowns and greater losses. So you use the cushion, but you get back some of the cushion again, but you have less cushion to use going forward. This is the problem. Some argue that central bankers such as Mr. Greenspan when he was head of the Federal Reserve Bank of the United States in the ‘90s and the 2000s, did issue a free put to the market protecting their downside. Economically, however, if everyone were to believe Mr. Greenspan would always do this and was a prophet in so doing, market prices would not have fallen in the first place. Smart traders would buy on the dip in every instance and no dip would occur. Given we see large market drawdowns mean that Fed or the Federal Reserve Bank or central bankers are not omnipresent, or the central bank, the government in the fiscal policy, is not that great either because we get huge moves as we’re seeing currently with inflation expectations changing and risks in the commodity world, etc., changing dramatically as a result of both fiscal policies, monetary policies, and political interchanges in the economy.
Problem two: Why hold equities and reduce risk by combining with safe assets such as government long-dated bonds? The optimal portfolio to hold for the next investment period is the portfolio that is gleaned to be tangent to the capital market line, that portfolio with the highest expected excess return per unit of risk. This is the Markowitz model. Take that portfolio that is the highest value or the one that is tangent to the safe asset so you combine holding risk in that risky portfolio with the safe asset to get the desired level of risk each period. With normal distributions, this comes from portfolio theory. Expect the returns and risk of the optimal portfolio change with new information about the economy affecting cash flows and risk premiums. Some argue that if the investor wants the risk of, say, 60% of the market and an allocation of 60% to equities and 40% to bonds, the returns on this strategy is commensurate with the risk. So it’s hard to believe that this static strategy is what investors really want. They always want 60% in equities and 40% in bonds. I never understood where this allocation came from in the first place. Where does it come from? I didn’t read it in the Scriptures, I didn’t read it in the pundits of the world. But somehow, however, it has become the default allocation strategy for myriad institutional investors.
There is another way to look at this strategy. If the expected risk premium of equities over bonds is deemed to be 5% a year, close to the historical average over the last 100 years, so when bond returns are higher, the premium on average is about 5% over bond returns, then holding 40% in bonds to reduce risk has an expected cost or insurance premium of 2% a year. That is 40% of 5% is about 2%. So to assume that you’re holding this portfolio of 60% in equities and 40% in bonds, cost in loss returns on average an expected loss of about 2% a year for that cushion, the cushion cost, that 2%. Some have argued that long run returns on risk assets are proportional to volatility. More volatility equates to a designated increase in returns. I don’t think this can be true in equilibrium, however. But believing this, some leverage bonds to have the same volatility as equities such that all asset classes were presumed to have the same expected returns based on volatility. The problem with this risk parity strategy is that the only way that all would have the same expected return is if returns were perfectly correlated – bond returns always move lockstep with equity returns. Obviously, this might occur only when central bank policies are affected returns and the economy’s in shock. Everything is highly correlated. So again, if all assets are redundant, then, because they’re all perfectly correlated, obviously, they’ll have the same expected returns other than for the increase in their systematic exposure. Bonds will have less returns. But once I know the returns on equities, I know the returns on bonds. Once I know the returns on risky, I know the returns on less risky equities. That’s what it means to have perfect correlation. And only in times of shock do we observe that. So the risk parity type strategy would only work then. Because otherwise, if volatility was an important criteria only on assets, then what about diversification? I would diversify and end up making a superior return by diversifying. So therefore, assets couldn’t have returns based solely on volatility; they’d have to be unsystematic risk components that are non-diversifiable. That’s what Bill Sharpe showed and he was awarded the Nobel Prize to show that, that it’s really the systematic exposure that counts, not the volatility that’s important. But in our world, we’re not looking at expected returns. What we’re looking at are the dimensions of risk and then volatility comes in, because volatility, although not coming in to expected returns, volatility comes in dramatically in how you risk manage the portfolio over time and compound returns.
Problem number three: Longer-dated bonds and correlations change among asset classes. Many don’t hold short durations bonds in their portfolio; they hold longer duration instruments, and some hold risky bonds with credit risks as well. Between 1981 and 2021, the returns on longer-duration bonds and equities have tended to be negatively correlated. That is, when risks increase unanticipatedly, investors fled into bonds and sold equities that they fled from to reduce risks, and bond prices tended to increase as equity prices fell, hedging some of the coincident losses in equities. Holding longer-duration bonds over this period provided terrific, unanticipated benefits to the portfolio. The returns on long-duration instruments were excellent. Interest rates fell from the teens to close to zero as real rates fell and inflation fell. Real rates were close to zero and even negative during the recent times. We estimate real rates by looking at the nominal rate, the rate you would receive on holding bonds, and subtract the inflation rate from the nominal rate to get the rate we observe to give an estimate of what the real rate is in the economy.
As we know, in Europe over the last number of years, and in the United States, real rates were close to zero or could have been negative, or negative in Europe and in Japan. Holding longer-duration risk assets increased returns and diversified risk through both the drift in rates and the negative correlation with equities. Over the last 40 years, there were only a few periods when the Federal Reserve and other central bankers had to intervene to stem anticipated inflation. They worried more about the effects of deflation on their ability to grow and can manage the economy. At times, when central banks need to intervene to stem inflation, the correlation between duration and equities might likely become positive. Because, basically, over those time periods, the Federal Reserve could overshoot or undershoot. Where like now, we’re under the question, are we going to have a hard landing, which both bond yields might increase because of inflation increasing, and at the same time have the economy getting into very bad shape? Or are we going to have a soft landing where the Fed is managing correctly so we don’t have much unemployment, the economy percolates along fine, but they do control inflation? Or whether we have, as I said before, it’s called stagflation, inflation and lag and recession in the economy? Or do we have it be the case the economy grows again and inflation comes under control?
Those are all things which change the risk distributions going forward. Empirically, over long periods of time – and this is a very fascinating statistic – over very long periods of time, the returns on equities tend to be very lowly correlated with the returns on bond. It’s only at times of shock that we get large, positive or negative correlation between the returns on bonds and the returns on equities. If equities were… Think if equities were always expected to be negatively correlated with duration, that’s the duration as the sensitivity of the bond returns to the changes in interest rates, duration would be a terrific hedge to holding equities and investors would bid up duration to offer low or negative expected returns. Because you can’t have it two ways. You can’t have something be a perfect hedge and make a high rate of return on that. This does not happen in economics. The future therefore is unlikely to see similar paths in correlations of bonds to equities over the past 40 years. Risks change. The dynamics change. And so we have to think about changing estimates of not only risk, but how assets are going to correlate with each other going forward.
Problem four: Expected drawdowns are described incorrectly. Many investment managers forecast downside loss and ascribe loss to their investors in terms of the number of standard deviation from the mean, a so-called Z score that could result. Two standard deviations below the mean for a normal distribution is used by many to calibrate tail losses on investments, say in equities. That is, if the S&P 500, a broad-based index that everyone uses as the benchmark in equities for equity returns, and as reported on Bloomberg and in the press daily, has an estimated volatility of 15% a year. And many would articulate a loss of 25% or 30% for equities, two standard deviations. And you listen to investment managers give presentations, that’s exactly how they would describe it. Our risk of loss is two standard deviations, and essentially a loss of 30%. For a 60% equity strategy and a 40% bond strategy, the tail loss would be estimated to be 15% because you’re 60%, 40% protected. These estimates did not integrate or take account of losses below the two standard deviation mark, for example. I don’t care what my two standard deviation is. What do I lose in excess of that? I don’t eat two standard deviation moves; I get fed on the bigger losses. That’s when I really suffer, and I wish I had a bigger cushion at that time. I wish my wife had bought more cushions to put on the couch so I could sell them later on. So this is deceiving, however. So for example, in 2007/2008, the 60/40 strategy lost 35%, far more than 15%. So the expected tail losses of more than two standard deviations for a normal distribution, even for a normal distribution, not for one with very fat tail losses or shock results, is two-thirds greater. So basically, you can end up closer to the 30% mark than the 15% mark. Quoting a 30% loss in equities would be far closer to the truth than it would be the 15% loss. Estimating a 50% maximum loss is closer to reality for equities. And losses of this magnitude were realized during the financial crisis, as I said, of 2008 and 2001 and 2002. My point is that tail losses and drawdowns are much greater than those articulated in standard thinking of a normal distribution. Even for a normal distribution, you should quote it correctly, which includes much greater expected maximum losses. Surprises, therefore, are much more common than presented to investors. Investors are disappointed and therefore lose trust. In addition, tail losses may be even greater implied by a normal distribution; as I discussed, risk may not be normally distributed at times of shock. I know that for being a fact. If I look at the distribution implied by the option market, you do not see times in which the … times the option market is forecasting normal distributed return. At times, the option market is forecasting much greater losses to the downside. At time, much greater upside potential than downside potential. Many use T distributions to estimate actual expected tail loss. It is my belief, as I just said, that the option market might provide much superior estimates of expected tail losses than these historical distributions. The market has information. They have information we can use to inform us as to what the losses might be on a 60% equity strategy and a 40% bond strategy, much greater information that tell us what losses might occur on 100% equity strategy. And it might give us information about how much cushion we might want to build. Should I sell those cushions or should I buy more cushions, you know, going forward? And that is a very interesting future direction of what time provides to you.
Problem five: Many investors reduce risks during drawdowns. This is interesting. Many investors reduce risk, they increase their cushion. Many advisors and academics claim that investors should do nothing and hold their portfolio positions during drawdown. Do not increase your cushion. Do not use the cushion. Investors should not ignore drawdowns, in my view. Obviously, holding and staying the course depends on the depth and direction of the drawdown. For many investors, risk is of paramount importance. This is something that really has come home to roost and very important to think about. Most of us have leveraged positions in the market. We don’t think we have leverage, but we do. I have promised myself consumption obligations. I promised myself the future cars, my vacations, my health care, my housing, etc. And now that I’ve gotten older, I realize that I’m also working for my family, my kids and my grandchildren, preserving purchasing power. These commitments are akin to leverage. As the market falls, leverage increases and the risk of not being able to meet future obligations increases. Although they can’t reduce their commitments, they can reduce their equity holdings to manage risk, a further risk of loss, and liquidate their portfolio partially or in full. Although some call this a panic response, this to me could be a rational response. No one knows when the momentum of the market will stop or when and if recovery will follow. Those who are rich, however, and not as affected by the drawdowns and are not as leveraged because their future consumption is guaranteed because of their wealth, provide liquidity to these investors who need to liquidate to increase their cushion as markets fall. Even the rich, however, reduce their demand as downside or momentum increases and equity values fall. In extreme crises, huge supply demand imbalances might be caused by the responses of market participants to increase the cushion and the inability of intermediaries to know exactly when to step in and to add value. After the fact, as we know, it is obvious when we should supply liquidity to market participants or step in. But market participants are constrained to reduce risk. But it’s not easy to explain or ascertain when crisis are evolving. So the interesting problem is, increase the cushion when you need to increase the cushion, but when should you sell some of those cushions or when should my former wife have told me that I can now sit on the cushion on the couch and use the cushion and use it to cushion my back and to help me weather going forward into the future?
Problem six: A systematic problem is risk-taking during retirement. There are those who claim buying annuities is an excellent choice for retirees. But this fixes consumption. It does not take account any bequest motives, it doesn’t take account of the next generation. It doesn’t take account of changes in consumption, it doesn’t take account of changes in needs. If a retiree builds wealth for retirement and bequest, the retiree might be investing for the next generation in addition to meeting their consumption needs. These become interesting an important planning function in risk management. And especially so as the baby boomer generation moves forward in retirement. This is the same problem that besets the spouses of the men who are dying more quickly. And as the wealth accumulation for women increases, they have to understand this whole same problem. What is it that I want to achieve? How do I want to work for the next generation? What do I want to accomplish for myself and my current consumption needs and my philanthropy and other needs? This is a growing problem that will be very important in the years ahead and needs to be settled. I believe that a systematic approach to helping investors in making these complex decisions is an important avenue of growth and research in the investment world. And we will add more colors as we proceed because time and how risk changes over time and how needs change over time with the changing risk and reward and realizations is the really important problem. Not the one-period model. The one period model is a stepping-stone, but it’s one step on a long path of stones to make the right choices and decisions.
Maymin: Thank you. Wow. You’ve totally decimated any sense of calm and happiness that we may have had with all this. We thought risk management was a solved problem, right? We thought we could rest easy at night, but apparently not. What are we supposed to do?
Scholes: Get more cushion, more pillows.
Maymin: 100% cash? What do we do?
Scholes: Yes actually, I think that in studying this problem, we have to really understand, investors have to understand the dynamics of risk or needs change over time and incorporate that into their investment decision. So many advisors are moving away from just the idea of, “How can I pick a portfolio for you to beat an index?” to thinking about the life management choices, incorporating tax, bequests motive, these things into the investment decision. That’s where the risk core comes from. Understanding, building trust with your clients, helping your clients make the decision, being the trusted advisor. That trusted advisor will marry together components of this, you know, what types of cushions to buy? What types of risk management strategies to buy? How do I marry together asset selection trying to garner the extra return if I can? And how do I do that? What matters to select, but how do I marry them together? How do I think of the dynamics? And where do I buy information and garner information to do a better job as my needs of my clients change, and that perception of these needs change? And that institutional knowledge changes and the risks that I can observe in the market change? So I’m saying here that the dynamics of risk is important. Risk is not just a system of measurement. Risk is not just saying, “Let me tell you what risk my portfolio has.” Risk is saying … is an active component of management, and that has to be incorporated. Just as asset selection is an active
component, risk management is an active component. It’s a strategic component. It’s a solution to what a client needs. And the client needs, because where technology is evolving to make it possible for the advisors to do more idiosyncratic needs, and build solutions more for the client, move more towards the solutions focus, and do it in a systematic fashion with, really, education, with really understanding of two parts. One is the systematic allocations based on how much cushion my client needs and the dynamics are risk management, which is the dynamic allocation based on time as needs change and market information as risk change.
Maymin: Of the six problems, are they all helped by looking at the market sources of risk, the option markets? Or do some of them still require additional self-examination of my own needs, or do they all require that?
Scholes: They all require … Everything is interrelated. I mean, basically, the leverage we have and the risk taking we have is a very important problem. I mean, right now in our economy, we have many people who are very leveraged. Our economy is leveraged, the central banks are leveraged, we have fiscal leverage not only the United States where the debt to GDP ratio is very high, our debt-to-value ratio is increasing. It’s the same thing in developing nations, and in China and Japan and Europe, and the likes. We have those … the leverage or debt is increasing and, you know, we have all the promises we made to ourselves, which is my implicit debt, it’s not explicitly the same thing. And individuals have taken on large amounts of debt, you know, either through mortgage contracts or leveraged positions and other things. So how to manage this risk is a dynamic problem. It obviously has to be managed in a systematic way and using any information you can to help you and assist you doing that. And option markets is one avenue of garnering crowd sourced information. It’s inexact, it’s not perfect, and you have to understand, it’s necessary to understand, as well, the desires and needs and how those are changing, the client base that one is advising as a trusted adviser.
Maymin: So many people or institutions, they choose an asset allocation strategy or portfolio, and then risk happens, drawdowns happen. Would you advocate for taking that and flipping it on its head and thinking, how much drawdown can I handle and then dynamically determining your investment?
Scholes: Yes, both. I think both. I mean, obviously, one has to react to outcomes and make changes based on outcomes. And in addition, one has to use the possible outcomes in the initial planning. You know, the same … it’s just part of life. And given the world is so complex and uncertain, it’s not an exact science. But one of the wonderful things about research and where the research is going to go, is how to use more and more information, how to use technology to help us garner more information about risks and how we get more information in the market prices. And with computers and crowdsourced information and remote markets, you know, we can then figure out how to use computers to incorporate more and more information in helping in decisions. So there’s a division of labor. I don’t expect the doctor who’s saving lives of patients to spend varying amounts of time in trying to figure out how to do that allocation. They should buy services from others. But it’s a much more expansive service than just the idea of trying to pick winners.
Maymin: It’s not one size fits all.
Scholes: That’s correct.
Maymin: Awesome. Thank you, Myron.
Scholes: You’re welcome.