Understanding the role of bonds in retirement portfolios

Conventional wisdom and popular personal financial advice suggest that asset allocation should shift toward bonds as investors move toward retirement. Rules of thumb like “100 minus your age” in stocks similarly suggest a decreasing stock allocation with increasing age, and target-date funds managing billions of dollars for Canadians follow a glide path toward a heavy fixed-income allocation in retirement. The Fidelity ClearPath Retirement Portfolios shift from a 92% equity allocation when retirement is many years away to a 33% equity allocation in retirement.

This common asset allocation advice stems from the idea that bonds are safer than stocks due to their lower volatility, and that retirees should not take too much risk with their investments. While volatility will always pose an important psychological challenge, it may fall short as an objective measure of risk for long-term investors, which retirees are because a typical retirement can easily last 30 or more years. One of the challenges with understanding how assets behave for long-term investors is a lack of historical data.

To address data limitations, the authors of the 2023 paper Beyond the Status Quo: A Critical Assessment of Lifecycle Investment Advice use a block bootstrap procedure to sample from 38 developed countries from 1890 to 2019. This approach creates nearly 2,500 years of country-month return data. Using this data procedure, the authors run one million simulations of a U.S. couple’s lifecycle. They compare the impact of various asset allocation strategies on retirement wealth accumulation, retirement consumption, probability of financial ruin and bequest.

Contrary to the conventional wisdom, the authors find that a constant allocation of 35% domestic stocks and 65% international stocks for the full lifecycle, including retirement, is optimal across all comparative measures. This finding suggests that despite stocks’ greater volatility, they produce better outcomes for retirees. The all-equity portfolio produces better income replacement rates on average and better left-tail outcomes for retirement consumption than a target-date fund, a “120 minus age in stocks” heuristic, and a 60% stock/40% bond balanced fund.

The main finding is not that stocks are safer in the long run than previously thought, but that bonds are riskier. Over the long horizons that most investors are concerned with, bonds become more correlated with domestic stocks, and they have bad downside risk in real terms, pushing retirement portfolios toward stock ownership.

This research, based on one of the most comprehensive data sets available, unequivocally suggests that investors should hold internationally diversified all-equity portfolios through the full lifecycle, including throughout retirement. Even adding 5% or 10% in bonds to the equity portfolio results in underperformance on all metrics considered.

Despite its dominance in the long run, the all-stock portfolio is more volatile and has larger average drawdowns, which could be particularly difficult for retirees to endure psychologically.

The psychological aspect of living with a volatile portfolio should not be ignored. While some evidence suggests that investors become more risk-seeking with increasing wealth, the persistent gap between investor returns and fund returns suggests that volatility causes behavioural challenges with real financial costs. Assessing how much risk an investor should take involves, broadly speaking, an assessment of their psychometric risk tolerance, their capacity to endure short-term declines in the value of their investments, and their composure through past experiences with market volatility.

Financial advisors do seem to boost the share of stocks in investors’ portfolios. Trust in a financial advisor may also reduce the perceived riskiness of investments and allow risk-averse investors to earn higher expected returns with a financial advisor than they would on their own.

A final and important note is that, while these findings are based on comprehensive historical data, they are far from conclusive. It is unclear whether the time series return characteristics carefully preserved by the bootstrap procedure the authors used will persist in the future, and it’s even unclear how confident any investor should be that the equity risk premium will show up during their lifetime.  

Another recent paper, Stocks for the Long Run? Sometimes Yes, Sometimes No, goes back further in time for the U.S. market only and finds that in the sample period 1793 to 1862, stocks beat bonds in not one of the 30-year rolling periods in the sample. This speaks to parameter uncertainty, one of the challenges with any model used to predict future returns.

Current research suggests that bonds are much riskier than previously thought and may not fill the “safe asset” role in lifecycle asset allocation, pushing retirees toward heavier stock allocations. Regardless of long-run risks, investors are myopic and concerned with volatility; an objectively optimal portfolio can quickly become disastrous if volatility results in bad investor behaviour. Even still, it is wishful thinking to believe that we can determine the objectively optimal portfolio ex ante due to parameter uncertainty.

Where does this leave retirement asset allocation?

It is important for investors at all life stages to understand that volatility is not the only way to think about risk, and that less volatile assets may be hiding greater long-term risks. It is also important for investors to understand their psychological constraints when building their portfolios. And finally, it is important to remember that, despite our best efforts, we cannot predict the future.

Investors need to be comfortable living with a portfolio knowing that its long-term outcome cannot be known. The best we can do is shift the distribution of outcomes in their favour.