Macro-factors revisited: an evolving approach to portfolio resilience

This paper is motivated by two key aims. The first is to improve our initial work on macro-factors (see ‘Building robust portfolios with private assets: the importance of macro alpha and beta (2024)’) that highlighted the interaction of asset class returns with the economic environment – with a focus on unlisted infrastructure. This is initially from a methodological perspective and a desire to improve our modelling to reflect advances we have made in this space. Our second aim is to incorporate internal and external feedback to address key questions raised. We believe these innovations, which to a large degree support our initial conclusions, allow us to better understand the interaction of private market asset classes, with the economic cycle.

Factor model redux

We introduce several innovations to extend the framework introduced in our 2024 publication. These innovations fall into two broad categories. The first is a refinement of our methodology by applying insights from our ongoing research of analytical possibilities. The second is building on specific findings from our 2024 paper around how factor choice and construction impact model outcomes. Our overall aim has been to improve upon existing literature with a focus on the estimation of robust results. We outline our innovations below.

Improving the factors

We based our framework in our previous paper on growth, inflation, interest rates, credit, and commodities. These are well-recognised macro factors from the literature. Since then our thinking has evolved, facilitated by our improved methodology, and has led to the following updates:

Economic growth: Our original factor in this space was cyclical equities less defensives, a widely used, high-frequency proxy for economic growth. This variable has become increasingly problematic as a proxy for GDP growth, however. Reasons for this are unorthodox monetary distorting valuations, the increase in share buybacks impacting momentum, the rise of passive investing and ETFs reducing signal and the market thematics driving index gains that do not reflect economic activity (notably tech/AI). We replace this metric with the OECD Composite Leading Index (CLI) which is far more aligned (conceptually and statistically) with actual trends in economic activity.

Interest rates: We have improved on a relatively simple approach that was to take the first differences of US 10-year yields. We found that this factor definition, while well recognised, was too narrow for global markets and portfolios. To convey more information from the entire term structure of the yield curve we construct a principal component from real yield curves of major advanced economies (constructed using nominal yields and break-even inflation swaps). The approach synthesises the information into a clearer signal, better linking real rate dynamics with asset returns. 

Credit: We also seek to get a clearer signal out of the credit curve. Our original factor was the return on a portfolio long investment grade (IG) corporate credit and short government bonds. This is altered to be the portfolio return of long high yield (HY) corporate credit and short IG corporate bonds. This better captures credit conditions and risk perception/appetite and reduces the interaction with base rates and liquidity effects.

Inflation: This factor remains actual advanced economy CPI inflation and is a driver distinct from real rates and growth. Its inclusion supports a cleaner decomposition across growth, inflation, interest rate (real), credit and liquidity channels.

Liquidity: This is a new factor aimed at capturing the illiquidity premium that we would expect in private market assets and that in part characterises their risk-return profile. It is seeking to quantify underlying market conditions in markets that affect pricing power and exit risk.

We are also seeking to separate liquidity risk from credit risk where the former is difficult to differentiate from the latter in terms of the impact on returns particularly in stressed markets (where correlations rise). To construct what is a composite indicator for this factor we are guided by two key studies: Pastor–Stambaugh (2003) who proxy liquidity via capturing the price impact of order flow, that is how much prices move in response to trades; and Amihud (2002) where illiquidity is measured  as the price impact per unit of volume –  higher values mean less liquid assets. This process is detailed in the Technical Appendix along with a graphical exposition of the factor.

We should note here, somewhat prefacing our results discussion, that this liquidity factor is constructed from identifiable market dynamics. By this definition we are able to get an estimate of how liquidity flows or ‘cyclical liquidity’ impact returns – this is particularly true of private market asset classes and those in closed end funds. What is less clear, and less identifiable via a constructable proxy, is the impact on returns of ‘structural liquidity’, which is the illiquidity premium – again an important concept for private market assets. This is a discussion we will return to based on our results.

For more, download the full white paper.

[1] For more information on the models and testing used, please refer to the white paper’s Technical Appendix.  

The information presented on this webpage provides an overview of the whitepaper that can be downloaded above and below. For full details, including all disclaimers applicable to the data contained herein, please refer to the complete whitepaper.