Podcast: Jan Rosenzweig on fat tails and LDI portfolios

An optimised portfolio can look very different when extreme moves are given more weight

UK pensions learned a thing or two about fat tails and extreme moves during last year’s gilt crisis, when 30-year yields jumped nearly 100 basis points over the course of two trading days. Jan Rosenzweig’s familiarity with market-breaking events dates back further, to the global financial crisis of 2008 and 2009.

“Once you’ve lived through something like that, that’s always at the back of your mind,” he says. “I’ve seen first-hand how some of these things develop and why they develop, and I’ve been thinking in those terms [ever since].”

Quantcast – a Risk.net Cutting Edge podcast · Jan Rosenzweig, 16/05/2023

 

The experience prompted Rosenzweig, a portfolio manager at Pine Tree Market Neutral and visiting senior research fellow at King’s College London, to explore portfolio optimisation methods that defy standard thinking.

In this podcast, Rosenzweig – who also consults for insurers and pension funds – discusses the optimisation of liability-driven investments (LDI). His latest paper, published on Risk.net, shows that an optimised LDI portfolio looks very different if the assumption that returns are normally distributed is dropped.

Rosenzweig’s method relies on the decomposition of portfolios into uncorrelated components, using a technique he introduced in 2021 called independent component analysis. In that paper, he also introduced a risk-adjustment optimisation based on kurtosis – the fourth moment of the return distribution – rather than volatility. This approach penalises tail returns more than volatility-based optimisation.

Related

Fat tails and optimal LDI portfolios

A portfolio optimisation technique for pension funds and insurance portfolios is presented

In his new paper, Rosenzweig generalises that method to allow the risk adjustment to use any positive exponent. A high exponent identifies securities with extreme returns and minimises the allocation to them. When returns exhibit extreme deviation, the exponent corresponds to infinity. In that setting, the portfolio manager’s aim is to minimise the maximum deviation.

Identifying independent components, and filtering out assets prone to extreme deviation, results in an optimised portfolio that may differ significantly from one optimised under the assumption that returns are normally distributed.

Could this approach have prevented last year’s LDI crisis? Not directly. The problems were primarily caused by a funding gap, rather than poor portfolio optimisation. “The cushion wasn’t big enough,” says Rosenzweig.

But he also sees other lessons in the crisis. “There was too much focus on neutralising small market moves, and that ate up too much capital, probably unnecessarily,” says Rosenzweig.

This is something his portfolio optimisation method could help with. The approach described in the paper “allows any hedging to be very focused”, says Rosenzweig. “You can focus on the part of the distribution, you can focus on the set of returns you want to hedge against, and that means the actual capital you need to allocate for hedging is sometimes quite a bit smaller.”

And if pensions need less capital for hedging, they may have more available for other purposes. “So, in that sense, you have plenty of capital left over for cushioning the impact of any funding gaps, for pursuing returns, for anything else you might need extra capital for.”

Index

00:00 Intro

02:00 Portfolio optimisation and LDI investments

07:50 The optimisation method

09:28 Extreme deviation

13:12 Diversification

16:36 Future research

17:43 The LDI crisis

To hear the full interview, listen in the player above, or download. Future podcasts in our Quantcast series will be uploaded to Risk.net. You can also visit the main page here to access all tracks, or go to the iTunes store, Spotify or Google Podcasts to listen and subscribe.