John Hibbert of Barrie & Hibbert gave a talk on interpreting extreme events in the financial sector. Once more, a topic rather pertinent for these turbulent times.
His area of expertise is building models to stress-test the effect of external events on life insurers’ asset portfolios – although what he said probably works just as well for banks. Essentially, this is meant to help organisations’ managers effectively manage risk within their portfolio. The models are used to answer two different questions –
- what is the fair value of an asset (which is a risk-neutral approach for pricing policies)
- projecting what capital is required to maintain the asset (which is risk-based, using historical risk premia and observed probabilities of real events)
Like any models, the answers you get depend on the assumptions you put in. Insurers use these models in part because the regulator requires them to (to ensure they are adequately capitalised) and partly as part of their management responsibilities (to ensure they are pricing products correctly).
It is down to the insurers what assumptions they make: they are responsible for the models they use, and they have to report to the regulators on their capital, and are responsible to the organisations’ owners.
The kind of events that the models use are the “black swans”, made popular by Nassim Nicholas Taleb (with whom, though I found his book badly written to extent of being unable to finish it for irritation, I largely agree): outliers, events that happen rarely and so people rarely consider them.
Apparently, the assumptions most insurers used were based on the previous twenty years of market behaviour. Up to 2008, the previous twenty years in the financial markets were pretty stable. The 1987 crash, when the UK stockmarket (FTSE) fell 26% and the US market (DJIA) fell 23%, was out of the sample. The period did include the falls resulting from the bursting of the dot.com bubble and the September 11th attacks; but as Hibbert said, the period between 1988 and 2008 wasn’t severe for financial markets. Traders, analysts and fund managers had no experience of more turbulent markets.
The managers rejected extreme events when using their models. They used normal distributed events based on the historic data they felt relevant – a twenty year period. Hibbert says that he recommended using events falling in the 99.5 percentile – that is, events happening on average once every two hundred years – when using his models for planning purposes; that would have included the crash of 1929, when the DJIA fell by 40% (and continued to fall a further 80% before July 1932).
Managers commissioning these models didn’t use such events because they couldn’t really afford to: they would have had to set aside large amounts of capital and reprice all their products.
As it was, the markets in 2008 experienced “a perfect storm”: everything moved the wrong way. Markets in different places which should have provided diversification were correlated, and all fell together. Interest rates rose and then quickly fell; bonds fell.
The extent of the fall was dramatic – and, Hibbert said, probably a 1-in200 year event: in 1931, ten of the sixteen markets followed fell by over 25%; in 1974, 9 fell by over 25%; in 2002, 6. In 2008, all sixteen fell over 25% and most (including the UK and USA) fell by over 40%.
The normal distribution had fat tails – those rare, catastrophic events. Their assumptions came back to bite the managers.
As a result, the insurers have had to contend with dramatic changes in their portfolios and capital requirements. American Insurers Group (best known in the UK as sponsors of Manchester United) had to be bailed out by the US Federal Reserve – although this was largely down to its trading in credit default swaps and collateralised debt obligations, rather than its mainstream portfolio. In the UK, insurers haven’t suffered as many banks have. It is possible, however, that they may need to raise fresh capital.
Many products issued by insurers may have been mispriced, because their prices were based on the risks perceived – based on a limited, twenty year, period; and their capital may prove to be insufficient.
We can only wait and see.