Ziling Jiang and Cyril Bosse-Platiere explain how institutional investors can develop strategic asset allocations using asset classes adjusted to provide higher ESG ratings
The environmental, social and governance (ESG) ratings of investments, such as their levels of climate risk exposure and carbon emissions, have become important for global institutional investors. However, incorporating an ESG preference into the strategic asset allocation (SAA) – ‘ESG tilting’ – requires thought.
SAA involves setting target allocations at asset class level and rebalancing the portfolio periodically. Each asset class is assigned an assumed mean return and variance, and an efficient frontier is then constructed to determine the optimal investment strategy for any specified level of portfolio risk, such that the highest expected return is achieved for that level of risk.
We cannot simply perceive one asset class to be more ESG-friendly than another, for example investment grade versus high yield corporate bonds. On the other hand, we know that ESG tilting will reshape the profile of our asset classes in two respects:
- At industry sector level, the weights of the most carbon-intensive sectors will likely be slashed in an ESG-tilted asset class. As risk/returns differ significantly between industry sectors, such sector re-weighting will lead to a risk/return change for each of those asset classes, and a different SAA outcome.
- At security level, there will be trade-offs between the most ESG-friendly securities and the most risk/return-efficient ones. Such trade-offs become more obvious when the two cohorts deviate to a substantial extent, and become complicated when we are faced with the task of selecting a limited number of securities and determining their weights from a vast index.
ESG-tilted assumption setting is rather complicated in practice. For example, our credit analysts may advise that there are only several hundred corporate bonds within our ‘investible universe’, and our end portfolio may contain less than 100 bonds due to the portfolio size limit. If we want the key attributes (duration, quality, solvency capital requirement, sector weights except for the carbon-intensive ones) of our end portfolio to remain close to the index, and we want some ESG tilting, there will be intricate optimisation mechanisms. How do we perform a fine calibration?
ESG tilting in practice
We will use a worked example to show how ESG tilting at SAA level can be implemented using a linear optimiser. Our example is based on investment-grade corporate bonds – usually the weightiest asset class for an insurer, and one for which there is often an abundance of ESG information available.
The first step is to identify the investible universe, which for simplicity we will assume is the whole of the ICE BofA US Corporate Index (C0A0), containing about 9,100 bonds. The key attributes broken down by industry sector-level are shown in Table 1.
The index volatility of 4.9% is calculated by mixing sector-level volatilities using a correlation matrix, all calibrated using historical total return data.
You will notice there are a few very carbon-intensive industry sectors (highlighted in green), but otherwise the ESG and climate change theme scores are similar. If we halve the weights of these carbon-intensive sectors and pro-rata the excess weights to the remainder, the characteristics of the index are shown in Table 1’s row B).
To better reflect the multi-step filtering process in portfolio construction, we use a linear optimiser to build a model portfolio (MP) with the objective of maximising the running yield, and the following constraints:
- Duration, rating and issuer concentration are to be broadly the same as the original index
- Retain 60% of the securities in the index
- Industry sector weights are to be within ±5% of the index’s sector weights, except for the carbon-intensive sectors that we adjusted
- ESG and climate change theme scores are to each be enhanced by at least 10%
- Carbon intensity is to be reduced by at least 30%.
The outcome portfolio – row C) – has the attributes shown in Table 1.
As an optional final step, suppose that, unlike the simplified assumption we made for the sector-weight adjustments, our credit analysts instructed us that the investible universe was only a small fraction of the index – several hundred bonds. Our end portfolio (MP’) then needs to be measured against this much reduced starting point.
A ‘yield-optimised, but not ESG-optimised’ similar model portfolio is built in the same way, except that only the non-ESG guidelines constraints are used:
- Duration, rating and issuer concentration to be broadly the same as the original index
- Retain 60% of the securities in the index
- Industry sector weights to be within ±5% of the index’s sector weights.
The expected return assumption of our ESG-tilted investment grade corporate bond asset class is calculated as:
RESG-tilted index = Roriginal index + (RMP – RMP’)
By repeating these steps for all asset classes, we have the risk/return assumptions we need for an ESG-tilted SAA. If we only focus on credit asset classes (where ESG information is more easily available), that usually still covers more than 50% of a typical insurance balance sheet.
We believe that our ESG-tilted approach to strategic asset allocation offers a practical solution to establishing an efficient frontier with improved ESG ratings.
Notes for table 1:
* Yield-to-worst: the lowest possible yield that an investor could receive on a bond that does not default.
** Carbon intensity is the volume of carbon emissions (tons of CO2) per million dollars invested for the scheme’s assets.
*** Climate change theme score (on a scale of 0-10) is the weighted average of all climate risk-related attributes of the issuer, including carbon emissions, energy efficiency, product carbon footprint, insuring climate change risk and financing environmental impact.
Ziling Jiang is head of insurance analytics, EMEA at Neuberger Berman
Cyril Bosse-Platiere is an insurance strategist at Neuberger Berman
Image Credit | Getty