Long-Term Asset Class Forecasting at State Street Global Advisors

Asset class returns can vary substantially through a business cycle, and a given asset class may have negative short-term return expectations (because of short-term obstacles) while still having overall positive return expectations over the medium or long term. As a result, State Street’s long-term asset class forecasts (LTACF) must explicitly incorporate both short- and long-term investment horizons with potentially diverging outcomes.

A key component of our strategic asset allocation process is the need to be forward-looking in the inputs used. While this brings a level of uncertainty into the process as with any forecast, it does give a better perspective on how markets are likely to behave and influence the ability of investors to meet their goals. And while historical price patterns serve as a guide for the future “equilibrium value” of asset prices, in most cases we utilize some forward-looking indicators. Such indicators may be based on views of our economists, aggregated views of the Street’s analysts, or our collective conjectures about future asset price behavior that we believe will come to pass.

Finally, our forecasts highlight the relative attractiveness of asset classes. For example, when we see 10 year bond return forecasts projected to be low or negative, we know that investors will need to consider assets further along the risk spectrum to achieve desired returns. Risk-return analysis per asset class also guides us in portfolio construction, highlighting which asset classes are eligible for inclusion given a client’s desired risk and return objectives.

Building Blocks Approach to Return Forecasting

We use a “building blocks” approach to return forecasting, focusing on drivers that have an inner frequency commensurate with the time horizon of the forecast. For example, when formulating forecasts for the next 5 or 10 years, we use building blocks that include 10 year growth and inflation views, as well as long-term historical averages for term premia and price multiples.

Equities

Future equity returns are calculated as a sum of our expectations for earnings growth, inflation, and income (equity dividends). The final forecast also includes a correction reflecting the propensity of earnings multiples to mean-revert over the long term—see Figure 1. Dividends and growth prospects are the foundation of this analysis as their combination is a known starting point for expected returns. A blend of current and forward dividend yields can be used to estimate income return to be received by equity holders. Real earnings growth will drive market valuations, and inflation adjusts to create a view on nominal growth. Finally, the experience of equity investors is highly influenced by the level of valuation at the time of purchase, especially at extreme price multiples relative to historical levels. As such, we will make adjustments for potential multiple expansion or contraction going forward.

This methodology is applied to 21 equity markets individually and then rolled up for more than a dozen regional and global index forecasts based on market capitalization of constituencies.

Sovereign Bonds

Our return forecasts for sovereign bond indices are derived from a comparison of current yield conditions with expectations for how the nominal yield curve will evolve going forward. For each bond index, the return forecast is calculated as a combination of price return and income return–see Figure 2.

When deriving the expected shape of the sovereign yield curve, we implicitly link that shape to our expectations of future growth. This is done by starting with the cash rate forecast (which is conditioned on growth expectations) and then adding a term premia component conditioned on the cash forecast.

Credit

For corporate bonds, we analyze credit spreads and their term structures, with separate assessments of investment grade and high yield bonds.

• For investment-grade bonds, we consider long-term credit spreads along with our forecast sovereign yield curve to generate the investment grade yield curve forecast. We then follow the same approach to calculating expected returns of corporate bond indices as we did for sovereign ones, but with one important caveat. We account for the fact that, because of bond downgrades, only a portion of the option-adjusted spread gets converted to income return.
• For the high yield universe, we consider the long-term high yield spread along with the forecast sovereign yield curve to generate the high yield curve forecast. One key difference from investment grade bonds is that, instead of bond downgrades, we incorporate a market-implied level of defaults that reduces total return expectation. The default impact is itself a function of the current high yield spread and bond maturity.

In both cases, we analyze bond indexes by country and type then produce forecasts for the regional and broad credit indexes by calculating market capitalization-based averages.

Alternatives and Smart Beta

We provide return forecasts for a range of alternative asset classes, including private equity, hedge funds, commodities, and real estate. We also provide a number of smart beta forecasts. Generally, our approach to forecasting alternatives aims to link alternatives to global market factors that we do have forecasts for – equity markets, levels of interest rates and inflation, credit spreads, etc. Linkage is not always trivial, and each forecast incorporates asset class-specific parameters. Detailed description of each is beyond the scope of this short summary, but we do plan to write a companion paper dedicated to this subject.

Risk Forecasting

Our focus so far has been on the expected return part of the forecasts. Most investors are also interested in risk and correlation expectations for a more complete picture of relative attractiveness of various asset classes going forward. For that reason, we provide risk and correlation figures alongside expected returns.

Our risk estimates rely heavily on analyzing long windows of historical data, spanning numerous market environments. Our long-term correlation forecasts are derived from analyzing long-term equal-weighted correlation from data going back to 1990, with negative long-term correlations systematically reduced in magnitude to moderate possible overly optimistic in-sample bias.

Recent LTACF Innovations

Several innovations were incorporated into the LTACF over the course of 2019-2020, focusing on the effects of ESG scoring, bond downgrades, and currency exposure.

ESG Score

We believe that the use of Environmental, Social, and Governance (ESG) criteria by financial markets is part of a multi-year readjustment of how institutional investors approach price formation and risk estimation of financial assets. We therefore have a responsibility to systematically and explicitly include ESG metrics in our investment analysis and decision-making process.

In 2019, State Street launched its ESG scoring system, R-Factor™ (R as in responsible investing). R-Factor scores are based on SASB’s financial materiality framework and draw from raw data provided by several high-quality ESG data sources as well as governance insights from State Street’s asset stewardship team. The result is an ESG view by company that is based on multiple perspectives seen through the lens of financial materiality and is therefore important to investors.

Beginning in Q3 2020, we incorporated R-Factor into our long-term equity asset class forecasts. Because some countries’ R-Factor scores are semi-permanently higher than others (due to structural differences in culture, law, or environment), we focus on how R-Factor scores rise or fall over time, rather than on absolute levels.

ESG score improvements may be rewarded by the marketplace in the form of (1) higher returns and (2) reduced risks. While ESG ratings may have a nuanced effect on returns, its impact on risks is perhaps most straightforward: Improved ESG R-Factor scores are likely to reduce tail risks associated with ESG issues, thereby delivering an overall lower level of risk (standard deviation). The inverse is true for those countries that have seen a deterioration in their ESG ratings. To account for these relationships, we built a framework that rewards higher-performing countries with lower risk expectations, and vice versa. To ensure that we do not change the risk expectation of an equity class as a whole, we subtract the weighted average of the normalized country scores (“global score”)—see Figure 3.

Effect of Bond Downgrades

Prior to Q2 2020, our process estimated an investment grade bond return forecast as a sum of return forecasts for a hypothetical sovereign bond index with matching cashflows and IG bond option-adjusted spread (OAS) estimated from the historical pattern. This methodology was a good first step, but it does not take into account bond downgrades. Some investment grade corporate bonds will be downgraded over their lifetime, and when they are, they will be removed from the index. Losses experienced by the holders of those bonds because of the downgrades would not be reflected in the index OAS, and the index OAS would in fact over-estimate the actual extra spread-related income that bondholders receive.

To account for this phenomenon, in Q2 2020 we introduced an adjustment factor that considers: (1) the proportion of OAS that shall be realized as return and (2) underlying bond tenures. We apply this factor, which is always less than 1, to historical index OAS values before including them in long-term forecasts for investment grade bonds—see Figure 4).

Currency Adjustment

Until recently, we only issued our forecast in local currency terms. Given State Street’s global footprint, it is often the case that our clients or strategists need to obtain forecasts in a currency that is different from the “native” (aka “local”) currency of a given asset (e.g., requiring the Australian Dollar forecast for the S&P 500). We have added this functionality to our forecasting framework.

When performing such currency transformation, the outcome depends on whether currency exposure is hedged or left unattended (“unhedged”). If one assumes full hedging, the resulting return forecast would be equal to the local currency version but with the adjustment for cost of hedging – the cash rate differential. The risk forecast would not change.

For the unhedged currency exposure case, we build and add the exchange rate forecast and subtract the inflation differential between the two countries in order to convert the local currency forecast into the unhedged one—see Figure 5. Risk figures for the unhedged currency case are higher than those for the local currency forecast, as additional, exchange-rate-related risks are in place.

State Street Advantage

Reasonable inputs are a cornerstone in developing an asset allocation strategy aligned with investors’ goals and objectives. Many asset management organizations around the world publish long-term asset class forecasts, but we believe that multiple aspects of our approach are differentiating.

Long-term asset class forecasting has been an area of focus for our Investment Solutions Group for over 15 years. Today, we provide long-term forecasts for over 150 asset classes on a quarterly basis following a systematic process that also relies on fundamental inputs from a team of deeply experienced investment professionals across the firm. Over the years we have constantly enhanced our process; this has included adding new asset classes as markets have evolved and refining the methodology to capture key developments that allow us to provide thorough and forward-looking asset class forecasts for our clients. Our clients have found this to be a key input to their long-term planning process.

Each quarter, State Street Global Advisors publishes its long-term asset class forecasts. This paper is best viewed in conjunction with those quarterly forecasts. Please reach out to your representative from the Investment Solutions Group or from Investment Strategy and Research if you would like additional information.