Skip to content

Insurer CIOs get to grips with behavioural economics

Posted June 1, 2018

David Worsfold

The prospect of applying analysis derived from behavioural economics to create uncorrelated strategies that have the potential to protect against downside risks engaged an audience of insurance chief investment officers at the recent Insurance Investment Exchange roundtable, hosted by Axa Investment Managers.

The scene was set by Bob Swarup, principal at Camdor Global Advisors and founder of the Insurance Investment Exchange. He said there was an uplift of interest in behavioral economics and the part human bias plays in investment decisions. Technology was playing a big part in this uplift: “Technology changes but human nature doesn’t. However, we now have technology powerful enough to analyse human nature and how human nature makes constant adjustments depending on experience”.

As many insurers look to tackle the persistent backdrop of low yields and volatile market conditions by diversifying their portfolios away from traditional asset classes, there has been rising interest in new ways of harvesting some of the persistent structural dislocations seen in markets and the risk premia observed, said Swarup: “This strategy holds the promise of uncorrelated returns while also controlling volatility and at palatable regulatory costs. This is a holy trinity for insurance investment, if there ever was one”.

He said the market was at a potential turning point as human memory and direct experience plays a major part in decision making. Citing what he called the ‘Moses Principle’ – that after 40 years in the desert a generation had passed on and for the younger generation anything looked better than life in a barren terrain– he said attitudes to interest rates were showing the limits of human experience in the investment markets.

“After ten years at 0% or thereabouts, any uplift in rates will look good to people with little experience of the world before the financial crisis, even if they do not rise as high as before”.

This theme of the limits of human memory was taken up by Augustin Landier, a visiting professor of finance at Harvard Business School, who is head of research at AXA IM Chorus.

“The limits of human memory introduce biases and the brain often neglects smaller, but potentially significant, effects. This was explored by the economist Herbert Simon in ‘Bounded Rationality’ as long ago as 1978. The post-financial crisis era has seen a new interest in these ideas”.

Landier said that one of the reasons for the revival of interest in these ideas and the new enthusiasm for applying them to institutional investment strategies was the availability of new data sources. The first generation of behavioural finance analysis was based on prices and accounting reports. It was very statistical and the time lags because of reporting delays often led to a slow awareness of important positive and negative factors.

The new generation of data from forecasts, news and textual sources coupled with the speed of availability and analysis has transformed behavioural finance.

There are still limitations, however. There needs to be a very large base of company data for it to work as a strategy, typically between 1000 and 5000 firms. There is also a need to monitor potential over-reaction to certain variables: “Analysts overweight earnings against cash flow. This is because you need to analyse accounts to get a true picture of cash flow”.

The need to use broad datasets and to dig deep into them is vital, said Landier: “There are dangers of everyone using the same data and therefore creating a herd effect”.

The big challenges in developing uncorrelated strategies using this analysis are whether they can produce stable results across market cycles and whether they can aspire to produce alpha. Alongside these is the potential regulatory capital cost but by focusing on liquid assets and transparency, this can be carefully controlled, said Landier.

The discussion around the table explored how this approach could be applied to some of the challenges insurers face, and indeed some have already begun to.

Brexit was one example of an event where traditional analytical tools did not produce the right information. Pre-referendum, the expectation was that equity markets would tank if there was a No vote. This didn’t happen because of the FX factors. Why wasn’t this factored into forecasts? Was it counter-intuitive and therefore overlooked?

For some participants, Brexit was a potential game-changer because it exposed how little they understand about politics and the behavioural implications of major political changes, something that has potentially significant implications across entire portfolios and all asset classes.

Others highlighted the need to examine the behavioural implications of regulatory actions, especially as the current regulatory regimes pre-determine large bond holdings, itself a behavioural bias.

The conservatism of boards and investment committees – another behavioural bias – was a potential inhibiting factor in the widespread adoption of uncorrelated approaches. Insurers want stability of returns over longer durations and are often cautious about strategies that involve significant commitment to alternative assets.

But there was also a sense of paradigms shifting. The potential of the “industrialising and multiplying” of information sources, coupled with new analytical techniques such as textual analysis and machine learning, opens up the way for all of these behavioural biases to be taken into account and for insurers to evolve new strategies to complement their traditional reliance on fixed income.


Leave a comment

You must be logged in to post a comment.