Is asset allocation at a tipping point?

14 November 2022

Nikola Vasiljevic, Head of Quantitative Strategy, Zurich, Switzerland; Lukas Gehrig, Quantitative Strategist, Zurich, Switzerland

  • Equity-bond correlations have shot up over the past twelve months, eroding the diversification benefits of bonds
  • Is 50 the new 60? With recent correlations, the longstanding 60/40 equity-bond portfolio benchmark has struggled, and a 50/50 benchmark may be preferable
  • Our analysis of the equity-bond correlation shows that the correlation increases when inflation is not in sync with the growth cycle
  • If the terminal US federal funds rate were to land between 5-7%, we would expect a further increase in the correlation of equities and bonds of 0.1-0.2, tipping the five-year average into positive territory

Since the turn of the century, negative equity-bond correlations have survived several macro and bear-market regimes. However, the relationship has flipped, turning diversification strategies on their head, with bonds and equities crashing in 2022 amid soaring inflation and surging interest rates. What are the main drivers, will the current trend last, and how does this impact asset allocation?

Asset allocation

The relationship between equity and bond returns holds the key to asset allocation. A negative equity-bond correlation implies that the two asset classes will move in opposite directions, on average. In such a scenario, bonds can bring balance to a portfolio by mitigating the effects of equity drawdowns. 

When the correlation becomes positive – as happened in 2022 after two decades of negative correlations – diversification benefits erode as bonds provide less protection during periods of equity-market turmoil (see chart). Arguably, 2022 has been a prime example of the potential dangers at such times in terms of the US 60/40 portfolio performance1.

Finding the ideal asset mix

Despite the tough macro landscape, bonds’ shine has not evaporated. Theoretically, as long as the two asset classes are not perfectly correlated, diversification benefits exist (albeit they’re relatively muted). This can be shown by calculating the portfolio volatility across equity-bond mixes (see chart) for different levels of equity-bond correlation. 

A perfect positive correlation signifies no diversification, and in that case the investor should choose the asset with the higher return. Conversely, when two assets are perfectly negatively correlated, diversification is maximised, and an asset mix with no volatility can be found. The optimal asset weights for that riskless portfolio are inversely proportional to their respective volatilities.

In practice, we do not see such extreme values. For most asset classes, correlations are typically between -0.5 and +0.75, although equities can have higher intra-asset-class correlations, especially during sell-offs.

In our chart, the maximum volatility reduction with zero correlation is about 2.3% (relative to where there are no diversification benefits). If correlation drops to -0.25, the maximum risk reduction reaches 3%. However, when correlation flips to +0.25, the diversification benefit reduces to about 1.7%. This effect can be described as a “diversification pull” – decreasing correlation improves diversification and reduces portfolio volatility (or pulls down the volatility curve).

The 60/40 portfolio is dead, long live a 50/50 portfolio?

Assuming that an investor’s risk tolerance remains the same, when the equity-bond correlation flips from negative to positive, the optimal asset allocation also has to shift in order to remain within the pre-defined risk budget. 

In the earlier example, this means that the equity allocation should be cut by about 1.5-2% per correlation increase of 0.1. Therefore, a correlation jump to +0.25 from -0.25 – which is a realistic scenario for US equities and government bonds – would probably result in a new 50/50 bond-equity portfolio mix, replacing the 60/40 portfolio. 

Bringing macro regimes to the fore

Historically, changes in macro regimes and uncertainty matter for the long-term performance and co-movement of equities and bonds. To explore the implications, we consider two macro regime frameworks – one based on inflation only, and one that combines inflation and growth. 

In what follows, we measure the performance of equities and bonds relative to their respective average returns over the period reviewed. Therefore, all reported numbers and discussions regarding performance are based on long-term de-trended data calculated on a quarterly frequency and annualised. 

This approach allows us to better demonstrate which macro environment is favoured by equities or bonds. We also show equity-bond correlations, which are calculated using regime-dependent quarterly returns, without any adjustments.

We define four different inflation regimes as follows: 

  • Deflation: Low and falling inflation
  • Disinflation: High and falling inflation
  • Reflation: Low and rising inflation
  • Inflation: High and rising inflation

Our findings show that bonds tend to outperform relative to their long-term average in deflationary and disinflationary environments, whereas equities favour reflation and disinflation (see chart, regimes 1).

Annualised average quarterly performance of MSCI USA Net Total Return index and Bloomberg US Treasury Total Return index in different inflation-only (regimes one) and inflation-growth (regimes two) regimes from March 1973 to September 2022. The regime-dependent performance is calculated relative to the respective average return over the full sample

These results are intuitive. Bonds usually react positively (negatively) to falling (rising) inflation, which ultimately results in lower (higher) interest rates. Equities are typically boosted by monetary and fiscal stimuli during reflationary episodes. They also outperform in disinflationary regimes, since inflation stabilisation should aid future profits while reducing discount rates. Equities tend to underperform in deflationary periods, which often coincide with recessions and spikes in uncertainty. Both asset classes typically suffer losses in inflationary regimes, as seen in 2022.

In the regimes 2 panel of the chart, we consider four inflation-growth regimes:

  • Deflationary bust: Falling inflation and real GDP growth
  • Stagflation: Rising inflation and falling real GDP growth
  • Disinflationary growth: Falling inflation and rising real GDP growth
  • Inflationary boom: Rising inflation and real GDP growth

This setting allows us to gain further insights by linking equity performance to the economic growth. Our findings indicate that bonds and equities favour disinflationary growth phases. Bonds also tend to outperform in deflationary busts, whereas equities do well in inflationary booms. 

Shifting tides of equity-bond correlation

Overall, the two proposed macro regime frameworks distil coherent conclusions. The latter unveils that equity-bond and inflation-growth correlations tend to move in opposite directions. As such, equity-bond correlations tend to increase when inflation is not in sync with the growth cycle.

This can help us to understand the recent moves in the equity-bond correlation and form a view on expected diversification benefits of bonds over the next twelve months.

The current macro backdrop can be classified as “inflation” in the inflation-only regimes 1 setting, and “stagflation” in terms of that seen in inflation-growth regimes 2. Such an environment is harmful for diversification as both equities and bonds tend to underperform in sync. If inflation keeps crawling up and the growth weakens further, we are likely to see continued equity and bond drawdowns. 

The negative performance of the US 60/40 portfolio since its peak in December 2021 until September 2022, is already on a par with that seen during the dot-com bubble burst of the early 2000s and the oil price shock in 1974. 

In a scenario where persistent inflation, coupled with further rate hikes over the next six to 12 months exacerbates growth weaknesses, the fall in 60/40 portfolio returns could be extended by another 10%, overtaking 2009 as the worst historical drawdown over the past half century.

However, if inflation finally cools off and rates stabilise, assuming a mild recession, the macro regime would likely switch from “stagflation” to a “deflationary bust”. Historically, that regime was supportive of bonds but not equities. 

In that case, the equity-bond correlations are expected to come down. When growth picks up the speed at a later point, a strong rebound in the 60/40 portfolio performance (and especially in equities) could be expected based on historical performance. The equity-bond correlation would increase again during that phase.

Breaking down correlations using risk premia

In a final step, we now seek to explain the drivers of the equity-bond correlation from a risk premia angle. 

We first decompose returns into nominal short-term interest rate and risk premia (such as term premium or equity risk premium) and other factors (such as real earnings growth and inflation)2

Second, we apply the mathematical covariance decomposition into covariance of all interaction terms. This procedure results in an equity-bond covariance factorisation via variances of individual terms which impact both equities and bonds (such as short-term interest rate and term premium uncertainty) and covariances (cross-asset risk premia interaction). 

Finally, by scaling the covariance by the product of equity and bond volatilities, we can offer insights on the dynamics behind the equity-bond correlation (see chart).

The first order effect comes from the short-term interest rate and term premium uncertainty (measured as their respective variances). Both factors contribute positively to interest rate uncertainty lifting the equity-bond correlation. These forces are balanced by the interactions between the short-term interest rate and the risk premia changes (in particular, the term-premium changes), which contribute negatively on average. 

Another decisive factor is the cross-asset premia interaction, or the covariance between the term- and equity-risk premium, which behaves broadly in line with the inflation-growth regimes discussed earlier. 

Finally, the covariances between the real earnings growth and the short-term interest rate and term-premium uncertainty changes, represent a second-order effect. 

An additional important insight offered by this framework – which is not directly observable in the chart – is that the average equity valuation over the observation period represents a scaling factor in our calculations. An elevated price-earnings ratio can amplify the effects of the short-term interest rate and term-premium uncertainty, and contribute to equity-bond correlation distortions.  

Therefore, equity market valuations and macro conditions can reinforce each other and result in significant correlation swings, especially in the short-to-medium term. 

Understanding the recent correlation shift

To better understand our quantitative framework and put it into the current macro context, let’s look into the past twelve months and focus on changes in the correlation determinants. This analysis reveals the key drivers of the rising equity-bond correlation (see chart). 

The main positive contributions come from the elevated short-term interest rate and term premium uncertainty, and the increased co-movement between the term and equity risk premia changes (that is, both premia have decreased recently). 

The joint contribution of the covariance between the two risk premia changes and the short-term interest rate uncertainty change boils down to the impact of the rising short-term interest rates on the realised equity returns (in excess of cash), which has been negative3.

An outlook for the equity-bond correlation

A relatively strong relationship between the short-term interest rate uncertainty and most other factors allows us to focus on the former when analysing future scenarios. To estimate possible dynamics of the equity-bond correlation next year, we consider three scenarios for the evolution of the future US policy rate.  

Using the Bloomberg Consensus data, we focus on the median and high forecasts (terminal policy rate at 4.4% and 5.25%, respectively), and we add one stressed scenario where the policy rate reaches 7% at the end of 2023. 

This simplified scenario analysis indicates that the equity-bond correlation will at best remain around the current level, but it could also increase further by about 0.1-0.2 in case the stressed scenario should materialise. As discussed above in this article, the outcome ultimately depends on the macro regime which will dominate next year – stagflation or deflationary  bust.

Related articles


Global Outlook 2023

As investors near the end of a tough year, full of twists and turns, our bumper Outlook 2023 takes a look at prospects for financial markets next year.