This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Find out more here

Moody's Analytics: what's in store for economic scenario generators

Companies: Moody's Analytics

Harry Hibbert, director, enterprise risk solutions, at Moody's Analytics discusses the evolution of economic scenario generators (ESGs) in light of dramatic market changes

Harry HibbertHow have low or negative interest rates affected insurers?

In terms of the modelling, low, and negative interest rates have led to the review of a lot of assumptions. In terms of the assets they hold, insurers need to seek higher yielding assets and also review the strategies they are using to try to hedge their liabilities.

What have been the most significant challenges for ESG users in recent years?

ESG users have seen challenges on two main fronts. First, new regulatory standards on the use of models, and actuarial risk management practices have meant that the workload has increased significantly. Stringent governance standards are being applied to how models are being used.

Second, as those governance standards have kicked in, there has been a requirement to review many of the fundamental features of models and the approaches that risk managers take to using market data and applying models to their balance sheets.

A good example of this is recent changes in how you use implied volatilities from market data. Lognormal models are poor at processing the low and negative rates that we are seeing these days. So, we have seen a shift in convention towards normal volatilities and that has triggered many significant reviews of modelling assumptions and practice for our clients.

If rates rise, will the approach revert to lognormal, or change to something else?

It would be feasible to change this again, but normal volatility works in a range of market conditions and is the convention now, so I think it is unlikely.

What does the industry’s shift towards higher-yielding investments mean for Moody’s Analytics?

Our clients are demanding more asset classes to be modelled, so we have developed a range of models covering assets such as infrastructure and property.

But our clients are also asking for more depth of analysis in all our models. This is especially the case for credit modelling, where we see a real demand for more granular models. At Moody’s Analytics, we have access to amazingly granular data and modern modelling methods.

We are also seeing a great need for modelling new investment strategies. For example, you do not necessarily need to change a portfolio to seek higher yields; you can instead change the strategies you employ to invest in those assets. We model the strategies to produce a kind of ‘realized book value’ that they might achieve, which gives insurers the opportunity to assess their options better.

How do you cope with modelling illiquid asset classes where data is relatively sparse?

If you do not have as much reliable data, you need to rely on theory more, but often there’s less research available in illiquid asset classes. So the real challenge is applying expert judgment: seeking the academic consensus and seeing what your expert clients are doing, and interpreting and deploying that in a robust and justifiable way.

Modelling liquidity itself is a big question for academics, and it is an area we are investigating closely.

What other initiatives are you pursuing?

We are developing a unified financial and macroeconomic simulation tool, which will allow us to get more macroeconomic variables within our models and support our clients in analysing narratives within scenario simulation.

The actuarial world has been focused on statistical, rigorous analysis of data that is mathematically founded, but has not typically looked at what-if scenarios. We are seeing insurers increasingly interested in what-if analysis. They want to know how an event impacts the models, and whether they can condition their scenario sets on that.

Another area is around more targeted solutions for the investment side and asset managers who are mandated by insurers. This entails having real-world models and premia assumptions embedded in them, and having the outputs readily available so they can be more useable for that client base.