Author Details

Monika Szikszai

Monika Szikszai
Associate, MSCI Research

Thomas Verbraken

Thomas Verbraken
Executive Director, MSCI Research

Social Sharing

Extended Viewer

Long-Horizon Risk: The Past 50 Years

  • We compared the effect of long-horizon multiperiod stress tests on different allocations to U.S. equities and bonds.
  • Even in scenarios with the largest initial one-year drawdown for a 60/40 stock/bond portfolio, the portfolio return was positive over a five-year horizon.
  • Although short-term drawdown risk is important, it does not take market recoveries into account and may result in overly conservative allocations for long-horizon investors.

Long-horizon investors — in this blog post, those looking out five years and beyond — might aim to ride out short-term disruptions in the market, and shorter-term risk measures may therefore be less appropriate for their investment horizons. In this analysis, we looked at the one- and five-year return distributions of hypothetical portfolios of different allocations to U.S. bonds and equities using 50 years of history. We saw that, although the average annualized returns were nearly identical over the different horizons, the distribution was significantly wider for one-year scenarios, because markets typically recovered within a five-year period. Imposing a risk budget on measures calibrated over a shorter horizon might have resulted in excessively conservative allocations that missed out on gains during market rebounds.

 

Simulating Portfolio Returns After One and Five Years

To investigate risk on longer horizons, we constructed historically grounded rolling five-year scenarios from the 1970s until today.1 We looked at the return of various hypothetical allocations to U.S. equities and bonds under these scenarios.2 Furthermore, with quarterly rebalancing, we reinvested income and rebalanced back into the initial asset allocation. This historical simulation shows five-year-long return paths for the various asset allocations, as if we entered the markets at various points in time.

The exhibit below shows the average (solid lines), as well as the 5th- and 95th-percentile (dashed lines), annualized returns as a function of the allocation to equities. As expected, the more equity-heavy portfolios had higher risk and higher return. For a specific allocation, the average annualized return was almost identical at the one- and five-year horizon. The return distribution was much more dispersed at the one-year horizon, however, because in most scenarios with a drawdown, recovery occurred within five years. Over the full 50-year period, the 60/40 portfolio’s 5th-percentile return was -11% over the one-year horizon, but remained positive (+0.5% annualized) over the five-year horizon.

 

Market Rebounds Came to Those Who Waited

 

The average and 5th- and 95th-percentile annualized one- and five-year returns for various allocation strategies to U.S. equities and bonds.

We then chose the 60/40 portfolio to zoom in on the trajectories behind the distributions. We first focused on the one-year drawdowns, as they can help investors understand shorter-term risk and the potential risk related to forced selling. The 60/40 portfolio experienced the largest initial one-year drawdowns in the scenarios starting in October 1973, April 2008 and October 2000, respectively. Although even these scenarios generated a positive return over five years because of market recovery, their five-year returns remained well below the average. One might be tempted to hedge tail risk to enhance longer-horizon returns, but this hedge may be costly to execute and has to be well-timed.

 

Even After the Worst Drawdowns, the 60/40 Portfolio Recovered over the Long Horizon

 

Cumulative return distribution of a 60/40 portfolio at various horizons for select scenarios (colored lines). The gray area represents the distribution of the portfolio’s cumulative return between the 5% and 95% percentiles.

Short-horizon risk measures can help long-horizon investors understand drawdown risk. But using those risk measures to gauge risk at longer horizons may potentially lead to an overestimation of long-run risk, as they do not take market recovery into account.

The authors thank Zach Tokura for his contributions to this blog post.

 

 

1We used five-year-long multiperiod market scenarios, whereby historically observed returns on the U.S. equity market and the 10-year U.S. Treasury yield were propagated to hypothetical portfolios of U.S. equities and bonds. We used the MSCI Multi-Period Stress Testing tool, which accounts for the reinvestment of income, rolling into constant-maturity bonds and rebalancing each quarter to the target asset allocation. In total, we ran 185 five-year scenarios, starting each quarter from 1970 to 2016. We used the shifted log-normal transformation to make historical yield changes applicable to today’s levels.

2The base asset allocation was a 60/40 allocation to equities/bonds, with the bond portfolio consisting of 50% U.S. Treasurys, 37.5% U.S. investment-grade corporate bonds and 12.5% high-yield corporate bonds. U.S. Treasurys and high-yield bonds are represented by Markit iBoxx indexes. The equity market is represented by the MSCI USA Index and U.S. investment-grade corporate bonds by the MSCI USD Investment Grade Corporate Bond Index. Based on Jan. 11, 2021, market data.

 

 

Further Reading

Stress Testing Multiperiod Inflation Scenarios

Did hedging tail risk pay off?

The Risk of Risk Limits

Regulation