SimCorp Insights: Navigating Risk Across Multi-Asset Portfolios
Can you tell us a little bit about yourself?
I am responsible for SimCorp’s Investment Decision Research in Europe, focusing on multi-asset class (MAC) risk analysis and stress testing. I am the author of the weekly Axioma MAC highlights newsletter, blog posts, and whitepapers, and a regular speaker at webinars and industry events.
What are some of the key risk regime shifts you're tracking right now?
After moving in lockstep for almost three years, stock and bond markets are now decoupled, as investors in each asset class appear to focus on different themes. Since Donald Trump took office in January, share prices have been following the ups and downs of trade negotiations, recently trading at or near record highs as major deals were struck. Treasury yields, on the other hand, have been locked in a tight range, as traders and central bankers keep evaluating the impact of existing tariffs and recent fiscal announcements on inflation and economic growth. The Fed seems set on further easing monetary conditions over the coming months, but how markets will react depends on the underlying reasons for the rate cuts. Easing consumer prices will be welcome news, whereas a cooling labor market or too much pressure from the White House are likely to trigger further turmoil.
What do you think are the most challenging aspects of building scenario analysis frameworks that capture risks across multiple asset classes?
The risk and return of a multi asset class portfolio strongly depends on the interaction between the different instruments in it, often more than on individual volatilities. When constructing a risk model or a stress test, it is therefore of the utmost importance to identify a calibration period that closely resembles the desired correlation regime. A good historical precedent can also provide plausible suggestions for which variables to shock and by how much. And having the right correlations allows the user to specify fewer variable shocks, which significantly reduces the risk of over-specifying the stress test.
Quant teams are increasingly managing portfolios with both liquid and illiquid exposures. How do the Axioma solutions address the fundamental challenge of risk aggregation across public and private markets?
Quant models rely on high-quality, high-frequency data, which is why they are very well suited for equity markets, where we have an abundance of observable pricing and fundamental information. Private assets, on the other hand, with their infrequent and often subjective and overly stable valuations, are notoriously hard to model. One common approach that we use in our Axioma solutions for integrating private investments into quantitative risk models is to replicate their behavior through a combination of public market factors, such as momentum, size, value, and credit spreads. In order to do this, we use cashflow data from private equity funds to imply their monthly returns, which are then regressed against those public factors. This allows multi-asset investors a consistent, holistic view across their total portfolio and to identify and, if desired, to hedge the associated risk exposures.
There's been significant evolution in fixed income markets – from rates volatility to credit spread dynamics. How are you adapting your factor models and risk attribution frameworks?
Traditionally, fixed income investors have been modeling portfolio risk through factors such as interest rates and credit spreads. The latter can be either at issuer level or rating and/or sector aggregates. This way of modeling credit risk corresponds to how bond portfolios are managed. But what if there are other common factors at work, similar to style factors in the equity world? Feeding our robust issuer credit spread curves into cross-sectional regression models, similar to those we use for our equity risk models, we have identified a range of style factors (e.g. market sensitivity, momentum, size, value), many of which carry discernable risk premia.
What are the most critical data signals and market indicators that you think traditional risk models consistently miss when capturing cross-asset dependencies?
In most traditional risk models, a portfolio is represented as a vector of sensitivities to a set of risk or pricing factors. The portfolio’s predicted volatility or tracking error is then calculated by multiplying those exposures by either past factor returns (e.g. historical or Monte Carlo VaR) or a factor covariance matrix (e.g. parametric risk). This means that the model will only capture and highlight sources of risk that the assets in the portfolio are exposed to. For example, an equity portfolio will usually show no risk contribution from interest rates, even though the underlying asset may well be exposed to changes in monetary policy. This is why it is important to use stress tests in addition to traditional risk models, as they can uncover those hidden sensitivities to other macro factors, such as commodity prices or inflation expectations.
Catch Christoph Schon live at Quant Strats (October 14-15) on the 11:45 AM panel: "Multi Asset Strategies - Finding a Functional Risk Model for Diverse Markets".
