Rob Arnott, founding chairman and head of Research Affiliates, discusses portfolio positioning amid tightening U.S. credit spreads, while Brandon Kunz, senior vice president and portfolio specialist, provides insight into Research Affiliates’ forecasting framework. As always, their insights are in the context of the PIMCO All Asset and All Asset All Authority funds.

Q: U.S. credit spreads have tightened significantly over the past year. How have the All Asset portfolios responded, and what is your outlook on credit?

Arnott:  The recent run in the performance of U.S. high yield bonds – the essence of spread-tightening – has been remarkable: The Bloomberg Barclays U.S. High Yield Index has returned nearly 25% over the past 15 months, surpassing performance of all other asset classes but stocks. Relatively low default rates, paired with rising optimism about the durability of the U.S. economic recovery, are buoying prices for this asset class. Currently, the spread of U.S. high yield bonds in the Bloomberg Barclays U.S. High Yield Index over comparable five-year U.S. Treasuries is 3.8%, plunging from a peak of 7.8% in February 2016 and sitting well below the historical average of 5.7% since 1987. Narrowing spreads reduce the long-term forward-looking return prospects for high yield bonds: When yields peaked in February 2016, U.S. high yield bonds boasted a long-term real return expectation approaching 3.0%, according to our forecasts; today, we’re projecting roughly half as much real return, at 1.6%.

Before delving further into our response, a few simple truths bear mention. We’ve observed that what’s comfortable is rarely profitable. It takes courage to place the buy order when others are selling. It also takes courage to sell when the crowds are clamoring to buy and believe “the trend is your friend.” Sure, the trend is your friend … until it’s not! This is why we aim to patiently “average into” our positions. We believe the willingness to tolerate discomfort and maintain the patience to exploit market anomalies can be a reliable source of long-term investment success. This is why disciplined, value-oriented contrarian strategies like the All Asset strategies may add value over a fiduciary (i.e., long-term) horizon. This philosophy also sheds some light into how our strategies respond to market movements, as in the recent high yield market rally.

As Figure 1 shows, we have steadily sold out of high yield bonds as spreads narrowed. In February 2016, our exposure to high yield bonds was 8.2% for the All Asset Fund and 12.0% for the All Asset All Authority Fund. As of 31 May 2017, we had cut these allocations by more than half, to 3.5% for All Asset and 5.0% for All Authority. As we’ve taken profits on high yield bonds, we’ve also steadily increased our exposure to countercyclical long-term U.S. bonds, which have sold off by 7.3% since the election (source: Bloomberg Barclays U.S. Long Treasury Index). In so doing, we’ve been gradually trimming our risk profile and our equity beta. We were decidedly “risk on” in early 2016; we’re decidedly less so today.

These responses to recent market movements are consonant with the All Asset strategies’ mission. Our portfolios are designed to systematically rotate across asset classes, seeking attractive entry and exit points over a multiyear horizon while helping our investors diversify away from mainstream stocks and bonds. After assessing long-term market prospects, we gauge how much they’re diverging from current market perceptions and respond in a contrarian way. We aim to do this while maintaining the discipline to allow momentum to carry markets beyond fair value.

Today we are comfortable with our reduced exposure to high yield.  It’s also worth noting that we gain exposure to this asset class through the PIMCO High Yield funds, which are positioned defensively and have largely avoided defaults over the years, with average credit losses far lower than market norms (as proxied by the Moody’s high yield default rate). Could spreads continue to tighten in the coming months? Of course. If yields rise, we would presumably increase our allocations to high yield bonds, unless other markets offered better bargains. If yields continue to fall, we would likely keep reducing our allocation.

In January 2016, when our core Third Pillar markets (real assets, emerging markets and high yield bonds) were ending a three-year bear market and valuations were cheap, we saw rapid outflows. Far from seeking bargains, many of our clients were fleeing them. At the time, I wrote “Our clients may flinch, but we will not.” Today, we’re often asked if the opportunities of early 2016 have come and gone. To this we say, “How many bull markets end after 16 months?” True, the Third Pillar is nowhere near as attractively priced as it was in early 2016, but our valuations suggest it’s still cheap relative to conventional 60/40 balanced portfolios (as proxied by 60% S&P 500 Index and 40% Bloomberg Barclays U.S. Aggregate Bond Index). I think most investors anchored heavily in domestic stocks and bonds (both priced to offer bleak long-term returns of less than 1% above inflation) will be disappointed by their results over the coming five to 10 years. And I think our investors will be grateful to own diversifying strategies that we believe are, even now, attractively valued.

Q: Capital market return expectations are an important input informing asset allocation decisions for the All Asset strategies. What modeling framework do you use to develop return forecasts for stocks and bonds?

Kunz: We model stock and bond return expectations using a “building blocks” approach designed to collectively capture the drivers of return, the best known of which is the Gordon Growth model.¹ The model determined that the historical return for an asset is a function of two components: cash flow yield and growth in future cash flows.

If asset valuations (e.g., yields, spreads, price-to-earnings (P/E) multiples) were always constant, then combining the observable current yield with an accurate forecast of cash flow growth would be the investment management industry’s equivalent of the Holy Grail – an extremely reliable return forecast. However, valuations do move, which can add to or detract from the return forecasted by the first two building blocks. While valuation changes can dominate the total return equation in the short term, over longer horizons valuations tend to mean-revert, so they generally net out of the equation. This leaves yield and growth as the two dominant components of longer-term returns. 

To illustrate this simple yet often underappreciated reality, we conducted a basic analysis. We estimated the return for a standard U.S. 60/40 portfolio (proxied by 60% S&P 500 Index and 40% Bloomberg Barclays U.S. Aggregate Bond Index) for each decade since the 1870s, and then compared it to what actually happened. Our forecast used only two of the three building blocks, starting yield and growth; we explicitly ignored valuation changes.  For U.S. equities, we used the starting dividend yield, and for growth, the 40-year average real growth in earnings per share plus starting inflation. For U.S. bonds, we used starting yields and assumed no growth in principal (though we recognize high grade bonds do exhibit very modest negative long-term growth from defaults net of recoveries).

This analysis helps explains why, in the context of All Asset, we intentionally haircut our valuation change assumptions when developing our asset class return forecasts (see Figure 2, paying particular attention to the far right column, which shows the difference between the forecasted values and the actual results). Put simply, we don’t want the volatile and difficult-to-predict valuation “tail” to wag the “dog” of our return forecasts. And while no one possesses clairvoyance, it’s possible to produce a reasonably accurate longer-term forecast by incorporating today’s yields and average historical growth into Gordon’s framework.

So what do we see? First, we note that, on average, this simplistic forecasting approach tends to be fairly accurate. The full-sample difference between the forecasted values and the actual values is just 0.6%.  Second, we see that dispersion around expected versus realized returns has been fairly symmetrical in the 14 intervening decades, with model expectations higher than realized in six decades and lower in the remaining eight. Third, we see that the dispersion tends to be mean-reverting – decades when the model overestimated returns tended to be followed by periods in which it underestimated returns, and vice versa. This is the valuation effect – more on that in a moment. 

So, what does our model imply today? With today’s starting dividend yields of 2.0%, 40-year average real earnings growth of 1.3%, and implied inflation of 1.8%, the U.S. equity market looks to offer annualized returns of approximately 5.1% over the next 10 years. When coupled with current U.S. bond yields of 2.5%, the annualized nominal return expectations for a U.S. 60/40 portfolio stand at a meager 4.0% for the next decade. And that estimate may be optimistic!

In addition, given the nine-year bull market in mainstream stocks and bonds, we could surmise that realized returns in the coming decade may come in even lower than the model implies as equity valuations gradually mean-revert lower.

While Gordon’s initial “yield and growth” approach provides a reasonable framework for forecasting long-term returns, it admittedly ignores the additional “building block” of valuation changes. For this reason, we also employ an expanded model to account for changes in valuation, akin to the widely cited Grinold and Kroner model.² Last month, our Jim Masturzo published a piece titled “CAPE Fatigue”³ that compares this “valuation-dependent” approach to both a one-factor cyclically adjusted price-to-earnings (CAPE) model and Gordon’s initial approach. The takeaway? No single model is a silver bullet applicable to all markets all the time, but using the expectations from multiple models may reduce the odds of undershooting the return level necessary to meet a predefined spending need.

In our approach to forecasting, we at Research Affiliates avoid assuming historical returns will be indicative of future results. In forecasting the individual components of either Gordon’s “yield and growth” model or Grinold and Kroner’s expanded valuation-dependent frameworks, we also avoid using current observations or long-term averages if we don’t see them as representative of the future. Instead, where applicable, we seek to build expectations for future yield, future growth and future changes in valuation by relating observations or trends from past environments to not-yet-observed environments of the future. (That said, we employ the usual caution that past performance is not a guarantee of future results.)

Here’s a simple example. Consider the yield received from an investment in short-term Treasury bills. Maintaining constant exposure to these investments over a 10-year horizon entails rolling into another bill at the then-prevailing (and likely different) short-term interest rate dozens of times. This means using the simplifying assumption that current yields represent what average yields will look like over a 10-year investment horizon is unacceptable, as is using the long-term average short rate observed within an entirely different investment environment. Instead, we use the determinants of short-term interest rates (i.e., demographic trends, economic fundamentals and monetary policy) to generate a forecast for the average short-term interest rate over the coming decade. Our modeling creates forward-looking expectations for the individual building blocks of each asset class and 10-year forecasts across a variety of asset classes.

While these forecast models ­are an important component of the All Asset suite’s investment process, we incorporate additional proprietary models to create shorter-term forecasts that inform our tactical trading. Our business cycle model, for example, calibrates our buy/sell discipline to the state of the macro economy by refining our shorter-term expected returns across asset classes. Specifically, this model recognizes that decelerating macroeconomic conditions may justify lower equilibrium valuations, and vice versa. This may temper our appetite for what appear to be attractively priced risk assets when probabilities of economic slowdown are rising. Similarly, if an asset class has experienced strong returns, but the economy is expected to keep improving, our business cycle model recognizes the rising risk tolerance of the marginal investor and encourages us to slow our profit-taking. We also incorporate a momentum overlay to account for shorter-term persistence in asset prices.

This approach, particularly when diversified across markets and asset classes, seeks to deliver a significant return premium. Both our business cycle model and our momentum overlay seek to reduce the probability of buying or selling too early, thereby potentially increasing the value added by All Asset’s shorter-term tactical trades.

The All Asset strategies represent a joint effort between PIMCO and Research Affiliates. PIMCO provides the broad range of underlying strategies – spanning global stocks, global bonds, commodities, real estate and liquid alternative strategies – each actively managed to maximize potential alpha. Research Affiliates, an investment advisory firm founded in 2002 by Rob Arnott and a global leader in asset allocation, serves as the sub-advisor responsible for the asset allocation decisions. Research Affiliates uses their deep research focus to develop a series of value-oriented, contrarian models that determine the appropriate mix of underlying PIMCO strategies in seeking All Asset’s return and risk goals.

Further reading
Recent editions of All Asset All Access offer in-depth insights from Research Affiliates on these key topics:

• The case for diversification amid an aging bull market for U.S. stocks (June 2017)
• Continued opportunities in emerging markets (May 2017)
• Benefits of active investment management in “third pillar” assets (April 2017)
• Where conventional wisdom may be making mistakes in today’s market (March 2017)
• Contrarian strategies to enhance real return (February 2017)
• Long-term asset class forecasts (January 2017)