All posts by Sellwood Consulting

Modeling the Reality of Risk in 2019

“The greatest scientists are artists as well.”
-- Albert Einstein

Two questions:

  1. When was the US Equity market riskier for an investor:
  • March 2009. The S&P 500 has just fallen 55% from its peak and the trailing ten-year standard deviation of S&P 500 returns was 21%. The market trades at a valuation of 13 times trailing 10-year normalized real earnings. Or…
  • March 2019. A decade later. The S&P 500 has just experienced a 10-year, relatively uninterrupted bull market in which it cumulatively returned 415% without declining 20% along the way. The index’s trailing ten-year standard deviation of returns has been 15%. The market trades at a valuation of 30 times trailing 10-year normalized real earnings.
  1. At what point in time – March 2009 or March 2019 – would most risk models say that the equity market was riskier?

A Tale of Two Markets

Early 2009 was a fascinating time to evaluate risk. Just about all information available to an investor suggested that the US equity market was riskier than normal. Historical measured risk was higher than average, the market had gone nowhere but down in recent memory, and every sensory input suggested “sell.” But at the same time, valuations were low, and long-term investors can afford to tune out the noise.

Early 2019 is an equally interesting time to model risk. On one hand, following one of the longest bull markets in history, after which prices have accelerated faster than fundamentals, it would seem that the market would be more, not less, risky than in 2009. Higher valuations represent higher risk, all else equal. They leave less room for error. But on the other hand, any model that calculates risk (defined as standard deviation) looking back ten years will, for the first time in a decade, not include the experience of 2008. Any such model will likely understate risk, relative to reality.

One of these market environments appeared riskier than the other; the other actually was riskier than the other.

Looking forward, is an experience resembling 2008 less likely simply because it happened eleven years ago now, rather than ten years ago? Of course not. Behavioral science (specifically, the availability heuristic) suggests that human beings would tend to think so. This is why we invent quantitative models — to help our human brains evaluate hard questions with better context. The problem is: in March 2019, any quantitative model with a ten-year lookback period would have suddenly “forgotten” about 2008. Without adjustment, the model would fall victim to the same inaccurate analysis that imperfect human beings do, as the last salient market event recedes into the past.

You Can’t Model Risk with a Formula

Careful readers of our 2019 Capital Market Assumptions may notice that this year, the “qualitative adjustments” we made to the expected risk (standard deviation) of several asset classes were higher than they have been in the past.

When we calculate a forward-looking risk assumption for each asset class, we model the “risk” by calculating an average of the asset class’s standard deviation of returns over the last ten years, and over the longest period for which we have reasonable data.

As an example, this simple approach would calculate a “risk” for US Equity (the standard deviation of the Russell 3000 Index) of 14.7%, as of the end of 2018. That index, in the last ten years, experienced a standard deviation of its returns of 12.4%. Over the longest time period we can examine, the index experienced a standard deviation of its returns of 17.1%. Averaging these two numbers yields 14.7%.

This risk number, 14.7%, quantifies how uncertain we are about future annual returns. Using our expected annual return for US equity (6.9%, in arithmetic terms) and a standard deviation of returns of 14.7%, our assumptions imply that returns in any future year worse than -22.5% (a “two-standard-deviation event”) would be extremely unlikely, and a return worse than -37.2% (a “three-standard-deviation-event”) would be nearly impossible.

We can draw a bell-shaped curve depicting these parameters. Everything under the curve is possible under the assumed distribution of returns. Real-world events should not fall to the left or the right of the curve – outside the realm of “possible” under our assumptions.

In the real world, it was only eleven years ago that we saw the US equity market decline by 37.3% in a single calendar year. Assuming market risk as calculated above would imply only a 0.1% chance of the 2008 experience happening. It would be saying that 2008’s real world decline should happen only once every 700 hundred years. This would be non-sensical. The US stock market, after all, has only existed for 201 years, and declines worse than -37% have actually happened in three of the last hundred years (in 1931, 1937, and 2008).

This is a problem. Nobody should forecast a level of risk that implies that something that has happened three times in the last 100 years should only happen once every 700 years. When the model tells you something that actually happened couldn’t have happened, it’s not reality that’s wrong; it’s the model.

Our process for designing risk assumptions solves this problem by forcing the model to adapt to actual observed reality. Rather than blindly accepting the output of a calculation, we have always applied a “qualitative adjustment,” upwardly adjusting the assumed risk until the actual worst-case scenario we have experienced, in the modern real world, had at least a 1% chance of happening, and has at least a 1% chance of happening again in the future. This implies that the worst-case scenario for a market, usually 2008, was a one-in-a-100-year event.

To recap, a naïve assumption of the US equity market’s risk, based on calculations of past returns, would yield an expected risk of about 15% and imply that 2008’s experience had a 0.1% chance of occurring. By focusing on real-world downside, and by making a reasonable assumption that 2008 was a one-in-a hundred-year event, we update our risk assumption to better reflect the real world. By increasing the risk assumption by 4.3 percentage points, we increase the validity of our risk assumption – the likelihood of 2008 happening inside of it – a hundredfold.

Still, a 4.3-percentage-point adjustment to the expected risk of a major asset class is unusual for us. It reflects the uniqueness of where we sit today, in early 2019. For the first time in a long time, a major market correction is so far in the rear-view mirror that it doesn’t show up in traditional models.

A year ago, under Sellwood’s 2018 assumptions, half of the risk calculation — the last-ten-years half — included the 2008 market experience. Our calculated risk for US Equity was 18.44%, making the real-world worst-case experience reasonably likely without adjustment.

But as the calendar turned into 2019, the 2008 experience fell off the calculation, and the calculated risk went down. Therefore, a higher adjustment was necessary to bring our assumed risk high enough such that a worst case in the real world is reasonably likely under our assumption set.

If we did not make this qualitative adjustment, our assumptions would tell an unlikely story: that at the tail end of one of the longest bull markets in history, risk has gone down, not up; and that the experience we all lived through just eleven years ago was so unlikely that it won’t happen ever again — not only in our lifetimes, but those of our children and grandchildren as well. We are uncomfortable designing an assumption set that makes these assumptions.


”Risk” is ephemeral, and not easily calculated. Nobody knows what the actual risk of an asset or asset class is. All anybody can do is guess, typically by examining what the past has delivered. Of course, we aren’t designing portfolios for the past; the ultimate goal is to develop reasonable expectations about what the future will deliver. Examining the past is of limited help in meeting this goal.

Our assumptions change as markets do, but the way we construct them includes guardrails to ensure that market anomalies or recent experience won’t contort the assumptions into saying the impossible.

To do otherwise is to surrender judgment to a formula.

Good investors know something that bad investors don’t: that it is often when things appear bleakest that prospects are brightest, and that when things look riskiest that they are safest. Models can help reinforce this truism, but only if the models are intelligently designed, blending art and science. Otherwise, using the models isn’t any more helpful than asking a bad investor for investment advice.

First Quarter 2019 Market Snapshot

Download (PDF, 284KB)

First Quarter 2019: March Madness, or Just an Incredible Rebound?

After a turbulent fourth quarter, equity markets quietly rallied, largely erasing 2018 losses. The S&P 500, which plunged 13.5% in the fourth quarter, gained 13.6% and is once again within striking distance of all-time highs. While risk assets posted gains across the board, the hardest-hit segments of the market during the fourth quarter were the largest rebounders, with small-cap and growth stocks leading the rally. Emerging and developed international equity, which had sold off slightly less than the U.S. market, bounced back nicely as well, both posting approximately 10% returns for the first quarter. Real estate and high-yield junk bonds also rallied.

While it has been a fabulous start to the year, this turbulent turnaround is the latest of several gyrations since volatility returned to equity markets last year. After one of the longest low-volatility rallies on record, investors began fretting early in 2018 about a laundry list of possible triggers that could derail global growth. Investor concerns over corporate profits, tightening central banks, and trade wars have whipsawed market sentiment. Adding to the market drama has been the will-they won’t-they trade negotiations between the U.S. and China, which have recently been outdone by the will-they won’t-they Brexit negotiations playing out within the United Kingdom Parliament.

One source of investor strain was removed this quarter however, thanks to Federal Reserve Chairman Jerome Powell. In a press conference early in the year, Mr. Powell relayed that the Fed would be “patient” when raising rates going forward. In March, Mr. Powell signaled that the Fed was likely done raising rates for the year. These statements represent a far cry from the official Fed projections from 2018, which had been showing the possibility of two to three rate hikes in 2019 alone. Fixed income markets rallied as rates dropped across the yield curve. The 10-year U.S. Treasury yield fell to a 15-month low and the yield curve, depending on which short-term rate is used for measuring, has flirted with inversion.

Welcoming Our Newest Shareholders

Sellwood Consulting LLC is pleased to welcome our newest shareholders, who have demonstrated lasting value to the firm and our clients, by exemplifying Sellwood’s core values of experience and integrity.

NICK WOODWARD, CFAConsultant, Principal

RUTHIE ZIMMERMANAssociate, Principal

These individuals join Sellwood’s four existing employee shareholders: Charlie Waibel, Kevin Raymond, Ryan Harvey, and Ashlee Moehring. Sellwood remains 100% independent and owned by its active employees.

2019 Capital Market Assumptions

 Download our 2019 Capital Market Assumptions White Paper.

Sellwood Consulting’s 2019 Capital Market Assumptions are available. These 10-year, forward-looking assumptions of asset class return, risk, and correlation are the key input variables for our asset allocation work on behalf of clients.

Our Assumptions Change as Markets Do

We update our assumptions annually. Over the course of 2018, most bond yields rose and equity market valuations declined. For this reason, our 2019 assumptions forecast higher returns for almost every asset class.

As a result of these updates, our forecasted “capital markets line,” depicting opportunities for investment return at varying levels of risk, has steepened: we believe that over the next ten years, investors will receive more compensation for taking risk than they were poised to a year ago. In the case of fixed income, higher real yields bring greater opportunity for return with lower risk. For risky assets, like global equities, the market experience of 2018 brought lower valuations, which increase the potential for future return.

Methodology Changes

While most changes to the assumptions simply reflect new market inputs, we made a few changes to our methodology this year.

First, we have slimmed down the assumption set, to include a more compact set of assumptions. The new assumptions are now mostly non-overlapping and for the largest segments of the global capital markets. For example, whereas in the past we have published assumptions for (all-cap) US Equity, Large-Cap US Equity, and Small-Cap US Equity, we now publish only an all-cap US Equity assumption. Asset allocation analysis is a blunt tool, and we believe that input assumptions should not be more granular than the methodology can support. Using a more limited set of assumptions reduces the risk of false precision when implementing them.

Additionally, we have updated our methodology for Non-Core Fixed Income, specifically our forecast for default rates for emerging markets bonds. In the past, we assumed that historic default rates would persist into the future. We have observed, however, that default rates have been much lower, after adjusting for ratings quality of the issuer, in the recent past. We now average default rates for the last decade and the longest period of data available, which assumes lower default rates, increasing the expected return.

We have also created new assumptions for Global Equity and Managed Futures.

Modeling Risk in 2019

The beginning of 2019 is an interesting time to model risk. On one hand, following one of the longest bull markets in history, it would seem that market risk would be higher, not lower, than in the recent past. Higher prices represent higher risk, all else equal.

On the other hand, any model that calculates risk looking back ten years will, for the first time in a decade, not include the experience of 2008.

Our approach to modeling risk has never changed. It has always looked back ten years, and longer, but we have never simply assumed that historical risk will repeat in the future. Instead, our approach forces years like 2008 to be acknowledged. For that reason, the “qualitative adjustments” to risk in our process are much higher in our 2019 assumptions than they have been in the past. For more detail, please see pages 27-30 of our white paper.

Fourth Quarter 2018 Market Snapshot

Download (PDF, 183KB)

Fourth Quarter 2018: The Bull Gets Trampled

The bull market that began in early 2009 finally lost its legs in the fourth quarter. The S&P 500 fell nearly 20% from September highs, finishing the full year in the red for the first time since 2008. Misery loves company, and markets in 46 out of the 47 countries classified as developed or emerging by MSCI declined in the year. The average country posted a 15% decline while the one outlier, Qatar, posted a 25% gain. While the magnitude of equity market drawdowns is significant, it is important to note that it comes on the back of broad market advances. Even after the correction, the S&P 500 is back to where it was just 15 months earlier in September 2017.

The Federal Reserve hiked rates four times in 2018, lifting the top range of the federal funds rate to 2.5%, and official projections show two to three rate hikes in the coming year. The markets, however, are skeptical and have priced in only a 10% chance that the FOMC hikes rates again in this cycle. Fed Chair Powell, like his predecessor Janet Yellen, is learning on the job about how best to communicate the rationale for future rate decisions, a job further complicated by President Trump’s public criticism of the latest tightening. The confusion over future rates and political uncertainty pushed long-term Treasury yields lower in the fourth quarter, flattening the yield curve. 10-year Treasurys offer only 30 basis points of incremental yield compared to 2-year bonds.

Traditional diversifying assets didn’t provide much relief for investors for the year as commodities and global real estate declined on growing global growth and trade war concerns. Oil dropped significantly as crude prices once again dipped below $50 a barrel. The decline in commodity prices has helped to drag market-based inflation expectations and inflation-protected Treasurys lower as well.

Happy Holidays from Sellwood Consulting

In the spirit of the holiday season, every year Sellwood Consulting celebrates a local charity.

This year, we made a donation on behalf of our clients to Community Vision, one of Oregon’s largest providers of individualized support for people with disabilities. Community Vision’s mission is to make Oregon a place where people with disabilities can live, work, and thrive in the communities of their choice.

Our staff also joined Community Vision for a “Volunteer Work Party.” While there, we made cookies with several of Community Vision’s self-advocates, and we were treated to a personal show by DJ Lamar, Portland’s best DJ.

We are honored that our financial support and engagement will help individuals with disabilities achieve independence.