The long-term performance of the global market index (GMI) reached an annualized pace of 6.3% in June, slightly above the previous month’s estimate. This change marks the second month of firmer estimates. The forecast is based on the average estimate of three models (defined below) and remains close to the lower range of performance achieved in recent history, based on a 10-year rolling return.
GMI is an unmanaged market-weighted portfolio that holds all of the major asset classes (non-cash). The underlying components of the GMI continue to show high forecasts relative to their current 10-year yields, a condition that involves tilting allocations towards the higher ex ante estimates. The US stock market continues to be the extreme exception. US equities are expected to generate a significantly lower return than over the past decade. Meanwhile, GMI’s current forecast of 6.3% is slightly above its 10-year lead, suggesting trimming US equity holdings, especially if the weighting is above its target for a given portfolio.
Also note that GMI’s ex ante performance is slightly lower than its 10 year return. It is an index to manage lower expectations in terms of what a globally diversified multi-asset class portfolio will deliver in the coming years compared to the past decade.
GMI represents a theoretical benchmark of the optimal portfolio for the average investor with an infinite time horizon. Based on this, GMI is useful as a starting point for asset allocation research and portfolio design. GMI’s track record suggests that the performance of this passive benchmark is competitive with most active asset allocation strategies, especially after adjusting for risk, trading costs and taxes.
Keep in mind that some, most, or perhaps all of the predictions above are likely to be off the mark to some extent. GMI’s projections, however, should be a bit more reliable compared to estimates for different asset classes. Forecasts for specific market components (US stocks, commodities, etc.) are subject to greater volatility and tracking error than forecast aggregation into the GMI estimate, a process that can reduce some of the errors over time.
For context on how GMI’s realized total return has evolved over time, consider the benchmark’s track record on a 10-year rolling annualized basis. The chart below compares GMI’s performance against the equivalent for US stocks and US bonds over the past month. GMI’s current 10-year yield is 6.8%. This is up from recent levels last year, but well below the highs of the last five-year window.
Here is a brief summary of how the forecasts are generated and the definitions of the other statistics in the table above:
BB: The Building Block model uses historical returns as a proxy to estimate the future. The sample period used begins in January 1998 (the earliest date available for all asset classes listed above). The procedure involves calculating the risk premium for each asset class, calculating the annualized return, and then adding an expected risk-free rate to generate a total return forecast. For the expected risk-free rate, we use the latest 10-year TIPS (Treasury Inflation Protected Security) yield. This return is considered a market estimate of a real risk-free (inflation-adjusted) return for a “safe” asset – this “risk-free” rate is also used for all the models described below. Note that the BB model used here is (loosely) based on a methodology originally described by Ibbotson Associates (a division of Morningstar).
Equalizer: The Equilibrium model reverses the engineering of expected return through risk. Rather than trying to predict return directly, this model relies on the somewhat more reliable framework of using risk measures to estimate future performance. The process is relatively robust in that it is slightly easier to predict risk than to project return. The three entries:
* An estimate of the market price of expected risk for the entire portfolio, defined as the Sharpe ratio, which is the ratio of risk premia to volatility (standard deviation). Note: “portfolio” here and throughout is defined as GMI
* The expected volatility (standard deviation) of each asset (market components of GMI)
* The expected correlation for each asset relative to the portfolio (GMI)
This model for estimating equilibrium returns was initially presented in a 1974 paper by Professor Bill Sharpe. For a summary, see Gary Brinson’s explanation in Chapter 3 of The Portable MBA in Investment. I also review the model in my book Dynamic Asset Allocation. Note that this methodology initially estimates a risk premium, then adds an expected risk-free rate to arrive at the total return forecast. The expected risk-free rate is described in BB above.
AD: This methodology is identical to the equilibrium model (EQ) described above with one exception: forecasts are adjusted for short-term momentum and longer-term mean reversion factors. Momentum is defined as the current price relative to the moving average of the last 12 months. The average reversion factor is estimated as the current price relative to the moving average of the last 60 months (5 years). Equilibrium forecasts are adjusted for current prices relative to 12-month and 60-month moving averages. If current prices are above (below) moving averages, estimates of unadjusted risk premia are decreased (increased). The adjustment formula simply takes the inverse of the average of the current price to the two moving averages. For example: if the current price of an asset class is 10% above its 12-month moving average and 20% above its 60-month moving average, the unadjusted forecast is reduced by 15% (the average of 10% and 20%). The logic here is that when prices are relatively high relative to recent history, equilibrium forecasts are reduced. On the other hand, when prices are relatively low compared to recent history, the equilibrium forecast is raised.
Avg: This column is a simple average of the three forecasts for each line (asset class)
10 years of retirement: For a perspective on real returns, this column shows the 10-year annualized total return for the asset classes up to the current target month.
Broadcast: Average forecast of the model minus yield over 10 years.
Learn how to use R for portfolio analysis
Quantitative analysis of the investment portfolio in R:
An introduction to R for modeling portfolio risk and return
By James Picerno
