UK insurers modelling the risk that interest rates move into negative territory are finding that the capital implications of doing so are less than straightforward. Hugo Coelho reports
Red ink is in high demand by analysts keeping tabs on the topsy-turvy world of government bonds in post-crisis Europe. Switzerland's entire stock of government bonds has been in negative territory for weeks, and investors in German debt paid more for a bund issued on 15 July than what they will receive in interest and repayment principal over the next decade.
So far, UK gilts have been a notable exception to the negative rate curse. In the aftermath of Britain's vote to leave the EU, yields on the benchmark 10-year UK government bond sank to a – very positive – record low of 0.975%. Insurers want to believe yields will remain above the 0% line in the near future, yet they are not taking chances.
According to an industry survey by consultancy Willis Towers Watson, ten out of 17 large UK life insurers allow for a negative interest rate shock in their Solvency II internal models. One year ago, just three firms did so.
InsuranceERM understands many other firms will follow when they review their models this year. This includes some of the most sophisticated property and casualty insurers, which have large holdings of short-dated bonds, the first to plunge into negative territory.
The rapid increase in the number of firms modelling negative interest rates reflects the higher probability that insurers are putting on this market event
Senior consultant Tim Wilkins says the rapid increase in the number of firms modelling negative interest rates reflects the higher probability that insurers are putting on this market event. The steep drop in the yield curve over the first half of the year, and in particular after the referendum, has only increased the perception of this risk.
This marks a clear departure from the Solvency II standard formula, which applies to most of the industry. The formula was devised at a time when negative interest rates were confined to the world of economic theory, and is blind to the possibility that the yield curve falls below 0%. In other words, it requires insurers to hold no capital against the risk that yields become negative or even more negative than they already are.
There is a broad expectation that this issue will be addressed when the European Commission reviews the standard formula by the end of 2018. Like in other areas of the directive, internal model firms might be just showing the way forward.
While internal model insurers are moving in the same direction, the approach to calibrating the stresses and the outputs seem to vary widely. One insurer in the survey said it is holding enough capital to withstand a fall in rates down to 3% at the two-year point. The shock used by another firm allows the yield curve to sink into negative territory until the 10-year point.
"Differences appear because firms have different risk models, they use different sets of data and they are looking at different time horizons," explains Sam Worthington, director at Willis Towers Watson and head of asset modelling.
The modelling of negative interest rate risk might be expected to drive up the capital that insurers hold against their investments, as it allows for greater volatility of outcomes. However, this is not always the case.
"Modelling negative rates increases risk at the centre of the distribution, but counter-intuitively it can lead to a reduction of tail risk"
To test the industry impact, Worthington has run representative asset portfolios through Willis Towers Watson's real-world economic scenario generator. He found that modelling negative rates does increase the capital required in the interest rate sub-module on a standalone basis, but when other market risk sub-modules are included in the analysis, the results are different.
"In times of market stress, we tend to see equities fall in value, corporate bond spreads blow out, and government bond yields fall, as result of the flight to quality. In tail scenarios, we found that the adverse impact resulting from falls in equity and corporate prices was more than offset by the rally in government bond prices," he explains, noting that this results in capital relief.
"Our analysis suggests modelling negative rates increases risk at the centre of the distribution, but counter-intuitively it can lead to a reduction of tail risk and therefore reduce capital requirements."
The different capital implications of a negative interest rate shock may help to explain why some firms have lagged behind others in modelling this risk. Wilkins says that the major concern for a number of insurers is an increase in interest rates, because of the way the value of assets and liabilities move.
Another reason to hold back is the technical hurdle of altering an internal model. Depending on the impact of the change in insurers' solvency capital requirement and their internal model change policy, modelling negative interest rates might equate to a major change, for which they must seek approval from the Prudential Regulation Authority (PRA). The process can take up to six months, and insurers will be tempted to revise other parts of the model at the same time.
A switch to modelling negative interest rates might equate to a major model change
InsuranceERM asked the PRA whether it has given guidance to firms to model this risk, but the regulator did not respond by the deadline.
In any case, the issue is unlikely to go away anytime soon. On 4 August the Bank of England's Monetary Policy Committee is expected to cut policy rates to 0.25% from 0.5% and unleash a new wave of quantitative easing, which will put downward pressure on yields.
If there is a lesson to learn from the experience of Japan and continental Europe, it is that low rates can always go lower.