Optimizing investment returns in challenging markets

Conning consultants Mark Saunders and Alexey Botvinnik discuss the impact of current market conditions on the modelling of portfolios in light of their recent whitepaper on strategic asset allocation. 

Interview by Vincent Huck


Has the COVID-19-induced market volatility and the prolonged low interest rate environment skewed insurers' traditional approach to ALM, forcing them to put greater emphasis on market risk?

Mark SaundersMark Saunders: Well, it has certainly skewed asset returns and made people consider their portfolios in light of the increase in tail risk. Insurers have been contending with low interest rates for several years, and as a result, more firms have been forced to look at alternative asset classes in their search for yield: asset-backed securities, loans, private placements, and infrastructure, to name a few. As balance sheets become more diversified, specialized tools and modelling are required to understand the return profile of these investments and to manage the additional risk.

Alexey Botvinnik: COVID-19 has accelerated these underlying trends, both by likely extending the period of low central-bank rates and the motivation to look at other asset classes, and also by encouraging companies to review their approach to modelling risks when they set their investment strategy.

What alternative approaches can companies implement and what are the benefits of such models compared to more traditional SAA approaches?

Saunders: Conning advocates a holistic approach, looking across whole portfolios in detail and incorporating liability cashflow into the analysis.
Using asset distributions with asymmetric tails and higher correlations in times of extreme economic events results in realistic asset returns, especially when looking at tail risk.
Incorporating liability cash flows ensures the model captures how changes in interest rates impact the insurance company's economic value. The liability cash flows also affect the timing and quantity of asset sales, which affects the likelihood the company will be forced to sell illiquid assets or Held at Amortized Cost securities in times of crisis.

Alexey BotvinnikBotvinnik: In addition, this approach allows the use of alternative risk and reward measures for the optimization; for example, it is possible to filter out any asset allocations which perform below a certain level in the 99.5th percentile.

What are the challenges to appropriately calibrate the models and ensure the calibration is correct? For instance, how did the calibration perform against a COVID-type pandemic scenario?

Botvinnik: Our approach is to combine the views of our quantitative experts and portfolio managers, ensuring the model calibration produces returns with a realistic risk-reward profile that is consistent across different asset classes and economies.

Saunders: In addition to producing sensible mean and volatility of returns, we work very hard to ensure the model produces realistic tail events. For many asset classes, the data history only includes a handful of severe market drawdowns, and such data needs to be supplemented with expert judgement. In May 2020 we back-tested our calibration across twenty economies, covering interest rates, equity returns, credit spreads, and FX rates, and found that in 96% of cases the impact of COVID-19 was captured within the projection produced by the calibration.

How does the framework work in practice? How do stochastic returns and liability cash flow provide useful analysis?

Saunders: We typically implement the framework in four steps. The first is to produce a set of stochastic returns that is consistent with our clients' internal views of market return and volatility over the next few years. We then import liability cash flows for each time and path and decide on a set of constraints for the optimization, such as duration limits, maximum allocations to high yield credit or overseas assets, etc. The third step is to select an "objective function", such as Economic Value or Present Value of Earnings, and run our Allocation Optimizer tool, which calculates an efficient frontier by evaluating the risk and reward for hundreds of different asset allocations. The final step is to analyse a selection of the points from the efficient frontier in more detail, looking at metrics such as solvency, earnings, and so on, as well as applying specific stress tests that are used in the company's ORSA.


Download the whitepaper


About the Authors 

Alexey Botvinnik, DAV, is a Director at Conning where he is responsible for economic scenario generators, ALM and risk management, Solvency II solutions, market consistent valuations and asset modeling.

Mark Saunders, FIA, CERA, is a Director within Conning's European Risk Solutions operations, where he is responsible for providing ALM and SAA modelling and advice to insurance companies and pension funds. He also leads the product management of Conning's stochastic modelling software.