Reinsurers’ ability to diversify has grown more sophisticated with the use of capital models. But unforeseen correlations can still occur
This is already shaping up to be a big year for catastrophes – both natural and man-made. The past five months have brought major earthquakes in Haiti and Chile, Winter Storm Xynthia in Europe, hailstorms in Australia and the Deepwater Horizon oil rig disaster in the Gulf of Mexico. Together it is thought these losses could amount to between $15bn and $20bn – a significant number when compared with last year’s total tally of $26bn for insured catastrophe losses. And the Atlantic hurricane season, which is anticipated to be the most active since 2005, has only just begun.
The good news is that most reinsurers were able to shore up their coffers following last year’s benign catastrophe year. The ongoing recovery of the global financial markets has also boosted the investment side of the balance sheet. And while it would have been impossible to predict the random sequence of events that have occurred so far this year, anticipating a steady stream of US landfalling hurricanes is somewhat simpler.
Since the 2005 season, when Hurricanes Katrina, Rita and Wilma (known as KRW) cost reinsurers around $60bn, there has been a push to diversify reinsurance portfolios by geography and line of business in order to avoid dangerous concentrations in US peak zones. As a result, capital models are being put to greater use as companies explore the likely impact of a new class of business or geography on their capital requirements.
“One of the big concerns insurers have when they’re looking at their book of business is trying to manage concentration risk, either from a geographic standpoint or line of business,” AM Best vice-president Ed Easop says. “Some companies’ strategy is to focus on a limited number of geographic areas, product types or product distribution and try to do that very well.
“Other companies say it’s better to have a mix of business, so that if there is any kind of event, it doesn’t bring the whole book of business down.”
Companies can diversify by line of business – by writing long-tail casualty business to complement a short-tail property portfolio, for example – or by geography. The basic concept behind geographic diversification is that big events are unlikely to occur in tandem in different parts of the world. An earthquake is unlikely to occur in California at the same time as one in Japan.
However, a reinsurer with an unhealthy exposure to US hurricanes can find itself subject to several losses within one storm season.
“If you have a lot of business in Florida or another area prone to hurricane or other catastrophic events, your balance sheet can get really hit in a short period of time,” Easop says. “If you’re overly concentrated, you run the risk that in a short period of time, you could have a major call on your capital.
“KRW was an extreme concentration of activity that hit one geographic area in a very short period of time. When something like that happens, it puts a great strain on your capital base and, given the short time frame, it’s hard to reload your capital,” he continues. “Companies try to diversify so it spreads out any catastrophic hit to their balance sheet from a time perspective as well as from a geographic perspective.”
While diversification is seen as a good way of spreading a company’s risk, it must also be done intelligently. In 2006 and 2007, the word ‘worsification’ was coined to describe companies that charged into new territories without performing adequate due diligence. The pressure was on to expand their book of business, but they did not always achieve the desired result.
“That’s one area where internal models can be very useful,” Easop says. “A company can look at all the different variables that drive their results throughout the business cycle and model various business combinations over a period of time.
“In effect, the models allow managers to say ‘what if’: ‘what if I take this block of business and add this other piece? Am I getting a benefit from diversification or am I doubling up my risk exposure because this new line of business or new geographic area has the same cyclical highs and lows as I already have?’.”
The theory behind diversification is a simple one. It allows you to write more business for the same amount of capital. “It goes to heart of the insurance concept itself: the premiums of the many paying the losses of the few,” non-executive director of Ultimate Risk Solutions and former chief executive of BMS Group, John Spencer, says.
“Diversification, whether by line of business or geography, is simply a further device for spreading risk. You don’t need to double the capital to write two uncorrelated lines of business because the?probability of them both?going catastrophically wrong at the same time?is low.”
So why concentrate on one line of business if it makes so much sense to diversify? While most insurance professionals would agree that diversification is a good thing, ultimately protecting businesses and policyholders against potential insolvency, a monoline business model can be very profitable in the low catastrophe years.
Nevertheless, monoline reinsurers fell out of favour after the active storm seasons of 2004 and 2005. The vendor catastrophe modelling agencies recalibrated their models to reflect a period of heightened hurricane activity and the rating agencies increased their capital requirements.
“The? KRW losses of 2005 had a predictably dramatic impact on the reinsurers that had chosen?to focus exclusively on catastrophe-exposed property business,” Spencer says. “However, that concentration of risk seemed to come as a surprise to the rating agencies and there was a subsequent push towards diversification as a way of?avoiding the alternative, which was the need to bump up capital requirements dramatically.”
While the rating agencies helped to drive diversification, investors and reinsurance buyers were also influential, pulling their capital away from carriers they saw as having too many eggs in one basket. Several Bermuda reinsurers opted to expand into new lines of business, such as buying US excess and surplus lines carriers, or turned to new markets by buying or setting up Lloyd’s insurers or moving into emerging markets.
“A lot of the Bermudians started off focusing on mainly property reinsurance in the USA and how do you diversify that? That’s where you start writing other lines, say specialty or casualty,” Willis Re chief actuary Ian Cook says. “Because they are uncorrelated, you can get a benefit to the reinsurer because the capital they’ve got to support their business can support both the property and the liability sides, or property in more than one country.”
Capital models can be used to inform decisions on diversification by calculating the likely impact on capital requirements. The idea is companies gain a ‘diversification credit’ if they are writing business that is uncorrelated.
“If you took a reinsurer that was just writing property reinsurance and one writing liability reinsurance, you could work out how much capital they would need,” Cook says. “But if you brought those two companies together, the amount of capital they need is less than the total sum they would have needed as two companies. It’s not going to go badly wrong at the same time and therefore you don’t need as much money to give you the same level of solvency confidence.”
While models can be used to determine how diversification will affect the company, decisions should be based on the risk appetite of management and the company’s shareholders, says Cook.
Making decisions successfully is also about understanding the potential pitfalls in models. “It’s easy to assume things aren’t correlated and do diversify, but what you tend to see in extreme situations is that things you thought were independent actually do have some dependency.”
Under Europe’s new Solvency II regime, re/insurance companies will either be able to develop their own internal capital models or use a standard formula to assess their capital requirements. While the internal model route requires more initial investment, it is more likely to provide a diversification benefit. As the standard formula is based on the ‘average’ company, it will not provide enough diversification credit for larger, more complex reinsurers, particularly those spread geographically.
Having a risk-based approach to capital is being emphasised as much by the rating agencies as the regulators. Outside Europe, reinsurers are also busy improving their internal models in a bid to enhance their enterprise risk management frameworks. Models are becoming more sophisticated as a result. They have better computational power, greater scenario testing and more complete datasets on which to test hundreds, if not thousands, of different assumptions.
But the models are not infallible. “You get to the point where you’ve got assumptions on top of assumptions on top of assumptions, and that’s where the variability in the models can come from,” Easop says. “You can have two different modellers look at the same baseline information and data, build the model to similar parameters but, due to on individual decisions based on these assumptions, you can get very different results; which is why companies have a lot more information now with the models. But you can’t run the company just based on the model.”
Using models to reflect the likely impact of diversifying a company’s book of business is just one part of the equation. Models are a useful tool for determining the right mix, but ultimately the board must make the decisions. Under Solvency II’s ‘use test’, senior management will need to demonstrate they use their internal models in day-to-day decision making.
If properly embedded in the organisation, models should be sufficiently flexible and easy to use? as part of management’s day-to-day decision making, rather than wheeled out on an annual basis, says Spencer.
“For example, an insurer may want to improve the balance of its? portfolio to ‘de-risk’ it and thus optimise?return on capital. A good capital model can help identify those parts of a portfolio that are having a disproportionate effect on capital requirements. And by non-renewing them, those requirements can be reduced considerably.”
There is a growing appreciation that overdependence on any kind of model is not a good thing. There are too many examples – in insurance and beyond – of where models failed to predict unforeseen concentrations of risk. “Many people feel there is too slavish a dependence on the output of models. It should be remembered they are just a tool to assist decision making and are only as good as the assumptions and quality of data that’s gone in. There are bound to be weaknesses,” Spencer says.
“9/11 is cited as the classic example of the insurance industry being taken by surprise,” he adds. “And that was a lot to do with the aggregation of risk from different lines of business that would previously have been viewed as uncorrelated.
“I remember the surprise at just how much fine art was housed in the twin towers – that was a major event for the fine art market. A lot of work has since been done by risk-modelling organisations?to try to measure those aggregations more accurately.”
The financial crisis is another example of where models – in this case value-at-risk models – were unable to adequately reflect extreme-tail risk. While there is less systemic risk in the insurance industry than in banking, reinsurers should continue to try to account for the unexpected in their modelling. Events such as the financial crisis or 9/11 tear up the rule book when it comes to uncorrelated risk.
“With the financial crisis, there was a major shutdown in parts of the global economy. Companies were struggling to sell assets that they thought would never decrease in value and it was hard for companies to raise additional capital,” Eason says. “It was almost like a perfect storm. People have started to realise these worst-case scenarios are more than just scenarios, they can actually happen.” GR