In pursuit of the holy grail

Multiasset-class investing strategies today must combine upside participation with downside protection, the so-called holy grail.

From 1980 to 2000, investors could have placed 60% of their assets in an equity index and 40% in a bond index, fallen asleep for 20 years and awakened to discover the portfolio had appreciated by more than 10% per year. Eureka! Since 2000, however, the experience has been quite different, as the same portfolio twice has endured drawdowns in excess of 30%. A new approach is needed.

What caused this change and how should investors respond?

The Bank for International Settlements, in its 85th Annual Report (a top-10 reading for all investment professionals), zeroes in on the sources of financial instability. Historically, macroeconomists have built models around the belief that if inflation remains in check, financial stability naturally tends to follow. Fortunately for all of us, the BIS, in its annual reports, takes a different approach. Its view is that, especially when inflation is dormant, other imbalances (excessive debts, elevated debt service to income and overvalued assets) tend to build until they can no longer be sustained, at which time they reverse. Financial market booms in turn become busts. This boom-bust dynamic has become increasingly commonplace since financial markets were liberalized and deregulated, beginning in the mid-1980s. The so-called “Greenspan put” (which placed a movable floor on equity market valuations) artificially kept this dynamic in check for some time, but ultimately culminated in a far more severe financial collapse in 2008.

The BIS argues, correctly in our view, that discussions of macroeconomic stability must incorporate the financial (boom-bust) cycle. We agree with this “real world” assessment. Consistent with it, we have developed a “market state” concept. The market state perspective identifies macroeconomic and financial market indicators (macrofinancial indicators) that drive asset class valuations. These indicators can be combined using a statistical process to generate market states. The rationale for this approach is intuitive and reflects the current reality that asset valuations periodically decouple from underlying real economic activity. Finance is not “neutral” – it does more than simply facilitate productive activity. At times, financial markets anticipate events, leading real economic activity, either up or down. In other (“deflation scare”) circumstances, rapid credit growth and overly loose monetary policies propel asset price appreciation even as real economic growth remains weak (2009-2015). In this situation, the tendency for financial markets to decouple results in a disconnect between asset valuations and real economic activity that periodically triggers sharp reversals (as in 2007-2009), with significant losses to investors. In our view, a market state approach more effectively captures drivers of asset valuations than a static buy-and hold approach.

The first step is to assess more than 100 economic and financial data sets commonly monitored by financial market participants. This data is then distilled into two indexes. The first index, called the Financial Conditions index), summarizes the current state of the financial markets. The second index, the Economic Conditions index), reflects the state of the real economy. This two-dimensional macro-financial environment index is what we call the market state.

The FCI and ECI are illustrated in the figure below on the X and Y axes, respectively. For example, both the FCI and ECI are positive (+,+) in a boom and negative (-,-) in a bust; the two decouple in a slowdown (-,+) and recovery (+,-). The normal diamond in the middle reflects periods in which there is no clearly defined market state (about 30% of the time).

The market state vector is recalculated every month as new information and data become available. The regimes thus evolve and become dynamic.

In addition to the market-state concept, there is also a tail-risk vector, or TRX, which shifts in response to changes in volatility and correlations. Use of the TRX is limited to “bust” regimes and only turns on when markets are under stress. When used in combination with a “bust” regime, the TRX tends to provide early warning information ahead of a crisis. When this occurs, the portfolio automatically moves to safety (U.S. Treasuries, cash and gold).

Of course, no strategy is infallible; periodic false positives are inevitable. The objective is to protect against major drawdowns. False positives are an inevitable price to be paid to obtain this protection.

The combination of these monthly signals (FCI, ECI and TRX positions) are illustrated below from January 2000 to April 2015:

This sample multiasset portfolio must pass the holy grail test in order to be considered viable. That is, it must generate higher risk-adjusted returns and also protect against downside risk more effectively than its benchmarks. In our back tests, we calculate monthly excess returns for 12 asset classes — U.S. domestic stocks, international developed markets stocks, emerging market stocks, real estate investment trusts, the Goldman Sachs Commodity index, gold, oil, corporate bonds, U.S. Treasuries, high-yield debt, Treasury inflation-protected securities and emerging market debt — with the following constraints:

Targeted volatility = 9%

High yield + emerging debt ≤ 30% of total assets

Total bond exposure ≤ 65% of total assets

U.S. Treasuries and T-bills ≤ 75% of total assets

We then compare performance for this multiasset index with three benchmark portfolios:

Equal-weighted among 12 classes, called EW;

S&P 500 index; and

The optimized benchmark return for the 12 asset classes with identical constraints but no market state framework, called BN.

Relative to these benchmarks, to be considered our portfolio must generate:

Higher risk-adjusted returns

Lower correlations with traditional asset classes when they matter most

Smaller drawdown risks.

To evaluate the risk-adjusted returns, we compare Sharpe ratios from January 2000 to May 2015 in the back tests. This portfolio generated a Sharpe ratio of 0.88, which compares favorably with the three benchmark portfolios (EW, 0.45; S&P 500, 0.03; BN, 0.36) It is clear from a Sharpe ratio standpoint that this portfolio passes the test.

Lower correlations when they matter most

In the crisis of 2007-'09, correlations between risky asset classes quickly moved to 1, given panic selling by many financial firms. Even the values of risky assets that were not directly affected suffered, often sold because more troubled assets were illiquid and/or because investors needed to meet margin calls. The correlations of various risk-on asset classes to the S&P 500 rose throughout this period.

However, given the combination of a “bust” (-,-) and the TRX, this portfolio moved to safety; thus, correlations between it and risky asset classes actually declined during the crisis. In the back tests, the portfolio added 18% during this difficult period (primarily due to plummeting yields on U.S. Treasuries).

Significantly smaller downside risks

In comparing downside risks for this portfolio with the three benchmark portfolios, we examine the frequency and magnitude of negative returns and maximum drawdowns in the back tests. We begin by examining negative six-month rolling returns by decile (0 to -10%, -10% to -20%). The table below lists frequencies of negative decile returns for rolling six-month performance. The shaded area reflects portfolio losses of more than 20% over a rolling six-month period for each portfolio. The sample portfolio outperforms the others in terms of frequency and magnitude of negative returns.

In addition, the maximum drawdown (peak to trough) for this portfolio in the back tests was -11%, which compares favorably with the three benchmark portfolios (EW = -29%, S&P 500 = -54% and BN = -40%). It is clear this portfolio outperforms the three benchmark portfolios when evaluated in terms of downside risk.

In an article titled “On the Holy Grail of Upside Participation and Downside Protection,” published in the Winter 2015 issue of the Journal of Portfolio Management, Ed Qian, chief investment officer at PanAgora Asset Management, offers a new measure for quantifying risk-taking. It assesses the downside and upside risks of a strategy, relative to a benchmark. It does this by calculating the average downside and upside risk of a strategy relative to its benchmark.

From this measure, a positive (participation ratio difference) means that the strategy protects capital or participates more broadly in upside performance … or, perhaps, both. The chart below compares the sample portfolio with the three benchmarks. The portfolio loses less than a dollar (73 cents, 26 cents and 49 cents) for every dollar lost by the three benchmarks (EW, S&P 500 and BN, respectively). It gains $1.23, 74 cents and $1.08 for every dollar gained, yielding it a net gain (= P+ - P-) of 50 cents, 48 cents and 59 cents, respectively. From an upside/downside perspective, the sample portfolio improves markedly on the performance of all three benchmarks.

The participation ratio difference confirms the index generates upside returns and reduces downside risk in comparison with all three benchmarks. It appears to meet the holy grail test.

The revival of the financial (boom-bust) cycle presents new challenges to asset allocators and individual investors. Integrating macroeconomic and financial market information is a key consideration in the 21st century, given dramatic changes in market structure and behavior. Portfolios should be designed keeping the macrofinancial dimension in mind. Ignoring either is likely to imperil future returns.

John Balder is co-founder of Investment Cycle Engine Inc., Newton, Mass.