Today, it is widely recognized that markets fluctuate between different regimes within which key aspects of investing, such as the risk/return relationship, volatilities, and correlations, vary greatly. Here, we discuss our iStrat Team’s approach for classifying regimes based on market behavior and the investment implications, which may be helpful in constructing a portfolio that is better able to withstand the ups and downs of different environments.

Regime identification

It is possible to establish a specific set of regimes up front (e.g., based on phases of the economic cycle) or to define them according to a set of market characteristics (e.g., inflation, growth, or volatility). These methods have merit and a role to play, but they require the imposition of a “regime map” on the markets, which can lead to behavioral biases. 

Instead, our behavior-based approach uses various characteristics (returns, volatility, correlations, price and yield levels) of multiple asset classes (equities, commodities, currencies, and fixed income) to define the market environment. We then apply a machine-learning technique in which an algorithm is used to discover hidden patterns without user intervention and then group market environments into natural clusters — i.e., regimes. 

Using this approach, we found that markets generally oscillate between four regimes: 

• Risk-on — Risk is rewarded, with the highest returns to growth assets and the largest underperformance by defensive assets. 
• Nervous — Returns for all assets are strong but fluctuating, with elevated volatility, a wide range of outcomes, and high drawdown risk.
• Panic — Risk is punished, with strong returns to defensive assets and the weakest returns to risk assets. 
• Uncertain — Slowing growth and rising inflation lead to muted returns and moderate volatility across asset classes.

Figure 1 shows these regimes over time and the sections that follow delve into the details.

Figure 1

Regime 1: Risk-on

This was the most common regime, having occurred around a third of the time, with the general theme being that risk assets were rewarded over defensive assets. As shown in the table in Figure 2, equities provided strong returns, emerging market equities outperformed developed market equities, and credit outperformed duration. The chart in Figure 2 emphasizes the pro-risk aspect of this regime in the upward sloping return-to-risk relationship. Drawdown risk for equities was the lowest of all the regimes (last column in the table).

Figure 2

Investment implications of this regime:¹

• Security level — Cross-sectional dispersion of global equities was lower than usual while the cross-correlation of global equities was higher than usual. In other words, equities tended to be less volatile relative to each other and to move more closely together, which may mean there was less benefit to taking active risk. 
• Sector level — This regime favored more cyclical sectors (e.g., energy, materials, industrials) over those with less-cyclical cash flows (e.g., real estate, utilities). 
• Factor level — Risk-seeking factors outperformed risk-aversion factors in both equities and credit, reinforcing the pro-risk nature of this regime.

Regime 2: Nervous

This regime, the third most common (about a quarter of the time), entailed elevated volatility and a wide range of outcomes for assets. As shown in the chart in Figure 3, there was a slightly positive return-to-risk profile, though not as strong as in the risk-on regime. Equities and bonds generated their second-best returns of all the regimes, indicative of a regime that was broadly pro-risk. At the same time, there were indications that this regime was fragile and prone to reversal, including higher volatility and larger drawdowns.

Figure 3

Investment implications of this regime:¹

Security level — Cross-sectional dispersion between global equities was higher than normal, while cross-sectional correlation was lower than normal. That is, in this regime, equities tended to be more volatile relative to each other and less correlated, further evidence of a “nervous” environment. 

Sector level — It was difficult to discern a clear overall bias in sector results, likely because of the volatile nature of the regime.

Factor level — Risk-seeking factors outperformed risk-aversion factors in equities and credit, given the broadly pro-risk nature of the regime. Mean-reversion factors performed well, implying that cheaper securities tended to outperform in this regime.

Regime 3: Panic

This was the rarest of the four regimes (about a sixth of the time), characterized by a negative relationship between risk and return, elevated volatility, and high drawdown risk. As shown in Figure 4, returns for risk assets (equities and commodities) were universally negative, and often by double digits. Their volatility and drawdown risk were also elevated. In contrast, defensive assets produced positive returns, with global government bonds in particular delivering their best return and lowest drawdown of all the regimes. Gold, often viewed as a safe-haven asset during times of market instability and high volatility, also achieved its best return. The downward sloping return versus volatility chart is perhaps the best illustration of this period.

Figure 4

Investment implications of this regime:¹

Security level — Within global equities, cross-sectional dispersion and cross-sectional correlation were higher than normal. In other words, equities tended to be more volatile relative to each other and more correlated. 

Sector level — Cyclical sectors (energy, materials, industrials) were relative underperformers in this regime.

Factor level — In keeping with the risk-off nature of this regime, risk-aversion factors outperformed risk-seeking factors in both equities and credit. 

Regime 4: Uncertain

This was the second most common regime, occurring just under a third of the time. It often occurred when economic activity was slowing but inflation was rising, and it was marked by muted returns (at or near zero), with just a couple of exceptions in the low single digits (Figure 5). Realized volatility was also low to moderate. As illustrated by the chart, markets hardly differentiated between assets on the basis of risk. This moderate environment also translated into below-average drawdown risk for almost all asset classes.

Figure 5

Investment implications of this regime:¹

Security level — Both the cross-sectional dispersion and the cross-sectional correlation between global equity securities were lower than normal. In other words, equities tended to be less volatile relative to each other and less correlated.

Sector level — Perhaps related to the security-level point above, the relative sector performances did not follow the usual pro-cyclical or anti-cyclical patterns; instead, we saw outperformance by sectors that are typically able to pass on inflation, such as energy, utilities, and real estate. 

Factor level — This regime’s lack of discrimination with regard to risk was evident in the performance of the risk-aversion and risk-seeking factors, with no clear bias between them within equities or credit, and with both producing a distribution of excess returns broadly around zero.

What’s the use case? Positioning for portfolio resilience, rather than regime timing

To summarize, our behavior-based research shows that markets have naturally fluctuated between four regimes, with risk-on regimes offering the most positive return-to-risk profile, followed by nervous regimes, with still-strong returns but also elevated volatility. Risk-taking was not rewarded in panic regimes, while uncertain regimes were characterized by muted results across the board.

While it may be tempting to use this analysis to position a portfolio for a regime transition, we think this would be challenging given the uncertainty over both the length of a regime and the nature of the subsequent regime — as shown in Figure 1, there was no clear pattern in the regimes over time. Instead, we think asset allocators would be better served by spending their time understanding how their portfolio is likely to behave during each of these regimes and ensuring that it is constructed so that it will be resilient enough to hold up in any potential environment. 

When building portfolios, allocators are generally comfortable employing geographic diversification, asset class diversification, and even some factor diversification. But allocators are generally not diversified across regimes, with portfolios often designed for a single specific set of conditions. We think our research in this area could help allocators build a more well-rounded understanding of regimes and construct more diversified and robust portfolios.

¹Sources: Wellington Management, Datastream | Sector and factor results based on returns in excess of cash (3-month US Treasury bill). | PAST RESULTS ARE NOT NECESSARILY INDICATIVE OF FUTURE RESULTS AND AN INVESTMENT CAN LOSE VALUE.

Important disclosures

Indices used for asset class characteristics
The asset class characteristics shown are based on the following indices:

DM equities: MSCI World (USD) Total Return Index
EM equities: MSCI Emerging Markets (USD) Total Return Index
Small cap: MSCI USA Small Cap Index
Global bonds: FTSE World Government Bond Index – Total Return
Global IG bonds: ICE BofA Global Corporate Index – Total Return Value Hedged 
US TIPS: ICE BofA US Inflation-Linked Treasury Index
US high yield: Bloomberg US Corporate High Yield USD – Month-to-Date Return
Commodities: Bloomberg Commodity Index Total Return
Gold: Bloomberg Gold Subindex Total Return

Neither MSCI nor any other party involved in or related to compiling, computing or creating the MSCI data makes any express or implied warranties or representations with respect to such data (or the results to be obtained by the use thereof), and all such parties herby expressly disclaim all warranties of originality, accuracy, completeness, merchantability or fitness for a particular purpose with respect to any of such data. Without limiting any of the foregoing, in no event shall MSCI, any of its affiliates or any third party involved in or related to compiling, computing or creating the data have any liability for any direct, indirect, special, punitive, consequential or any other damages (including loss of profits) even if notified of the possibility of such damages. No further distribution or dissemination of the MSCI data is permitted without MSCI’S express written consent.

ICE Data, its affiliates and their respective third-party suppliers disclaim any and all warranties and representations, express and/or implied, including any warranties of merchantability or fitness for a particular purpose or use, including the indices, index data and any data included in, related to, or derived therefrom. Neither ICE Data, its affiliates nor their respective third-party suppliers shall be subject to any damages or liability with respect to the adequacy, accuracy, timeliness or completeness of the indices or the index data or any component thereof, and the indices and index data and all components thereof are provided on an “as is” basis and your use is at your own risk. ICE Data, its affiliates and their respective third-party suppliers do not sponsor, endorse, or recommend Wellington Management Company LLP, or any of its products or services.