A macro risk-based approach to alternative risk premia allocation
Alternative risk premia are encountering growing interest from investors. They mimic strategies formerly available through investment in hedge fund vehicles but with more favourable liquidity and cost characteristics.
In this paper, we investigate the question of the allocation across a range of cross-asset alternative risk premia. For this, we design an active macro risk-based framework that aims to exploit varying behaviour in different macro regimes. We then build long-term strategic portfolios across economic regimes, which we dynamically tilt based on point-in-time signals related to regimes nowcasting and current carry. Finally, we perform backtests of the allocation strategy.
The analysis of Fama and French in 1992 is regarded as the starting point of studies on alternative risk premia. Their research showed that the performance of stocks can be explained by the exposures of securities to three main factors: the market evolution, the spread of performance between value companies and growth companies, and the spread of performance between small and large-capitalisation companies.
Equity indices provide imperfect exposure to such sources of return. By contrast, an alternative risk premia approach seeks to magnify these sources of return by removing the effects of market directionality and bringing the portfolio to a chosen risk target.
A discussion around the definition of alternative risk premia was covered in our paper: Alternative risk premia investing: from theory to practice. The next step is often to question their sensitivity to economic and market environments. Traditional risk premia display characteristics related to business cycles and regimes. Similar patterns have been revealed for some alternative risk premia such as trend following and carry strategies.
Alternative risk premia and economic regimes
We consider the economic and financial environment to consist of four regimes: recession, inflation shock, market stress and steady growth. For the world economy, recession periods represent roughly 15%. Around 12% of inflation shocks have occurred outside recessions. The frequency of market stress regimes is about 13%. As a by-product, the steady growth regime represents around 60% of occurrences.
Economic regime long-term probabilities
Source: Unigestion, April 2017. The figure represents the unconditional probabilities associated with the different regimes for each region. Recession, inflation shock and market stress regimes are estimated through Markov-switching models applied to economic activity, inflation surprises and equity indices respectively. Where none of these regimes are estimated to be prevalent, we classify it as steady-growth.
To illustrate the sensitivity of alternative risk premia to these macroeconomic regimes, we look at regime-conditional excess Sharpe ratios, i.e. the difference between the Sharpe Ratio of each alternative risk premium in each regime and their long-term (unconditional) Sharpe ratio.
Some strategies can be seen as being more “defensive”, like trend following, FX value, bonds carry, and equity quality, which tend to do better than average during periods of recession and market stress. Most carry strategies, as well as size and momentum equity factors, tend to deliver lower-than-average Sharpe ratios during those regimes. However, they have historically delivered better-than-average results in steady growth periods.
Regime-conditional excess Sharpe ratios for alternative risk premia
Source: Unigestion, April 2017. The figure displays the Sharpe ratios for each alternative risk premium in each of the four macroeconomic regimes (recession “R”, inflation shock “I”, market stress “S” and steady growth “G”), in excess of their full sample period Sharpe ratios. Sharpe ratios are defined as the annualised average regime-conditional excess returns divided by annualised volatility.
To further assess the performance of alternative risk premia in each of the four macroeconomic regimes, we also measure their hit ratios under each regime, and in the full sample. We define the hit ratio as the percentage of periods in the relevant regime where there is a positive excess return over cash. The table below displays the full sample and regime-conditional hit ratios of all alternative risk premia.
The more defensive strategies such as FX value, bond carry and equity quality tend to have higher hit ratios during periods of recession and market stress. Equity momentum has also experienced one of the highest hit ratios in market stress. Meanwhile, carry strategies, as well as equity size factor experienced higher hit ratios in steady growth. Emerging market FX carry, dividend carry and trend following had high hit ratios in inflation periods.
Full sample and regime-conditional hit
Source: Unigestion, April 2017. The table displays hit ratios of all alternative risk premia, both on the full sample, and under each of the four macroeconomic regimes. Hit ratio represents the percentage of positive excess returns over cash. Calculations are based on monthly USD returns. The sample starts in January 1999 with the exception of Credit Carry (November 2005), Dividends Carry (August 2008) and Volatility Carry (June 2004). The sample ends in December 2016 for all time-series.
A macro risk-based asset allocation framework
Taking into account each risk premium’s sensitivity to the macro regime, we next seek to define and implement a process to allocate between them. As with traditional portfolios, the allocation decision encompasses a strategic allocation and a dynamic (or tactical) allocation.
Building a robust strategic allocation means finding the optimal risk budgets allocated to each alternative risk premium. Rather than maximising the full-sample Sharpe ratio – which would be the result of a standard mean-variance approach – we adopt an approach that seeks to be more robust across economic cycles (“all-weather”) by scoring each risk premium across several dimensions, taking into account behaviour under different macroeconomic regimes and specific aspects of risk and practical implementation.
The dynamic allocation consists of implementing active deviations from the strategic risk-budgets by incorporating two main dimensions: (i) the conditional probabilities of the economic regimes (ii) the carry, i.e. expected return of an asset should its price remain constant, associated with each of the alternative risk premia.
Since the information on regimes is usually available with a lag that inhibits its usefulness for asset allocation purposes, a more rapidly available measure of these regimes must be constructed. Here, “nowcasting” is used to describe the practice of estimating current economic conditions, and we have developed three types of point-in-time nowcasters: a growth nowcaster, an inflation nowcaster and a market stress nowcaster.
To build our active risk-based portfolios, we estimate active returns from carry and nowcasters. The z-score, or standard deviations from the mean, is translated into active returns. Active portfolio weights are then computed by multiplying active returns by the inverse of alternative premia volatilities and scaled by the ratio between the strategic portfolio’s risk and its Sharpe ratio.
The asset allocation methodology we have described has been backtested from January 2005 to December 2016. The period from January 1999 to December 2004 is used to initiate the model. The table summarises the performance statistics of the portfolio.
Risk and returns for strategic vs dynamic allocation to alternative risk premia
Source: Unigestion, April 2017. The table displays descriptive statistics from four different simulations. “Strategic” is based on a fixed strategic risk budget. “Carry”, “Nowcasters”, and “Combination” use the dynamic allocation process, with expected returns estimated respectively with “Carry” signal only, “Nowcasters” signal only, and a combination of both. Tracking error and information ratios are computed relatively to the “Strategic” simulation. Calculations are based on USD monthly returns net of hypothetical transaction costs. The sample starts in January 2005 and ends in December 2016 for all simulations.
All strategies have delivered high risk-adjusted returns, with Sharpe ratios between 1.42 and 1.73. The strategic portfolio has benefited greatly from diversification, with improved risk-adjusted returns relative to individual alternative risk premia over the same period, and low drawdown compared to the level of realised volatility. Dynamic allocation based on the individual signals has added value through improved returns for comparable risk levels, and even more so when using them in combination. The reason behind this incremental performance from the combination lies in their diversification properties. Active returns from the dynamic signals have the interesting property of neither being correlated with the strategic portfolio, nor with one another.
While the behaviour of alternative risk premia has been the subject of many studies, we have here investigated the less frequently studied question of their allocation. Drawing on recent advances on active risk-based investing, we have designed a macro risk-based allocation methodology to distribute capital across alternative risk premia. For this, we have defined a long-term strategic risk budget allocation that we have dynamically tilted based on two major types of signals: nowcasting indicators of macroeconomic regimes and the current carry of individual alternative risk premia. We have backtested the strategy for a diversified range of alternative risk premia strategies. While the usual caveats apply, we believe the evidence shows that investors can enhance the results of static portfolios over the full economic cycle, particularly in bad times such as recessions.
This message has been prepared for information and for your personal use only. It must not be published, reproduced, distributed or disclosed (in whole or in part) by recipients to any other person without the prior consent of Unigestion. All information provided is subject to change without notice and should only be considered current as of the date of publication without regard to the date on which you may access the information. Past performance is not an indication of current or future performance. Data and graphical information in the attached document are for information only. No separate verification has been made as to the accuracy or completeness of this data which may have been derived from third party sources. As a result, no representation or warranty, express or implied, is or will be made by Unigestion as regards the information contained herein, and no responsibility or liability is or will be accepted in this respect. An investment with Unigestion, like all investments, contains risks, including the risk of total loss. Complied: 24 May 2017