Are you a Switched-on Buyer?
Corporate insurance buyers who can understand and replicate their insurers’ and reinsurers’ rating processes stand to get the best results, say Raj Ahuja and Tom Hettinger. 
We have heard a lot in the trade press on both sides of the Atlantic about relations between risk managers and their insurers, and about how buyers can make their programmes more attractive in the eyes of underwriters. Much of the comment has rightly concentrated on issues such as demonstrably high-quality risk management procedures and accurate exposure information. Important as these are, there has been relatively little comment on another and often critical aspect of negotiation: the way re/insurers rate their risks.
There are two reasons why understanding this aspect of the business is beneficial to commercial buyers. Firstly, it provides useful insight into underwriters’ thought processes. Secondly, applying the same processes to their own organisations will enable them to better understand their risk profiles and exposures, and to identify the most cost-effective steps to acquiring greater stability. This is especially desirable at a time when insurance–buying strategies are changing, usually with substantially increased retentions.
These benefits apply whether or not there is a captive involved, and to negotiations with reinsurers as well as insurers. Such considerations are especially relevant now that deductibles are at historically high levels and that, whether or not you agree with them, the weather modellers are urging us to take a pessimistic view of future patterns. 
Property is not, however, the only important class of insurance to consider in this context. Liability (especially workers’ compensation) and auto are two other areas where losses and premiums have escalated, partly because of legislative developments and inflation.
The methods used by insurers and by reinsurers to rate large corporate risks have gone through a transformation in the past five years in response to rigorous methodologies, supported by rating and simulation software, and also to pressure exerted by security analysts and shareholders.
Until the end of the 1990s, most underwriters’ rating strategies could be summarised, admittedly with some degree of oversimplification, by the following formula:
| Old price, allow for insured’s recent loss experience, allow for market trends |
Of course, the ability to demonstrate good risk management and internal controls will have a bearing, but re/insurers generally believe that they have already responded in these areas; they will require evidence of further and material changes. Another way to help your cause is to provide sufficient detailed and reliable data in the format that the models require.
The vaguer the information, the more the underwriter is likely to take a cautious view: in other words, to increase premiums. Many insurance buyers believe they already provide the necessary data, but the underwriter often sees it differently. You would be well advised to consult your broker and/or underwriter about what they require long before you begin the next renewal process.
And that is one reason why large corporates should consider creating risk models similar to those of their re/insurers. Doing so would enable them to anticipate how their underwriters will respond to a particular package and what incentives or concessions might produce the most favourable response. You will get an early indication of what is causing the premium to be so high, allowing you time to provide solutions.
A more important implication for buyers is that they are being forced to accept much higher deductibles. In the property/business interruption classes, many corporates have started to retain huge amounts of non-catastrophic risk, either directly or through their captives. This relates to both property and casualty. Whilst this may be a sign of good and confident management, it increases the inherent uncertainty and generates a lack of predictability for the P&L. That is a second advantage to creating your own risk model; it will enable you to manage and predict your residual (uninsured) risk effectively.
Let’s turn now from the theoretical to the specific: a retailer (see Exhibit 1), has outlets across North America and physical assets worth slightly more than $1.25 billion, employing around 5,000 staff, and runs some 300 cars and commercial vehicles. Reinsurance is arranged via a captive and, after the 2005 storms, only kicks in at $1 million for all perils for property/business interruption. All other classes have each and every loss deductibles of $200,000.
The risk manager has in the past sought to estimate likely losses to within a margin of error of $0.5 million, but the new structure makes this a considerably more difficult challenge. He creates a model to help him to quantify his risk and give him a clearer idea of his retained exposures. A key output is shown in Exhibit 2, where the buyers can really consider the degree of risk retained. The higher percentiles prompt a review of risk carried on the balance sheet, and further, more complex risk structures are needed.
This is the sort of thing that a growing number of corporates are doing, especially in North America and Europe, now that it has become a lot simpler. The secret is to choose a model or modelling software that allows maximum flexibility, and to start with a relatively limited and straightforward project. Once the initial project has been successfully completed and the lessons have been learnt, you can then progress to something more ambitious and wide-ranging. ‘Fit for purpose’ is always a key consideration.
One of the outcomes of this particular model is that, although the different classes may diversify risk to some extent, there is also a tendency (in the extreme) for them to move together. In other words, if one class gets out of control, there is small chance that the others will go in the same direction, with pronounced effects on the P&L and balance sheet.
Overall, the risk manager decides that the probability of missing his margin of error is too high, with an unacceptable level of volatility. After running a series of scenarios through the model, he decides that the most cost-effective reduction strategy would be to buy an aggregate stop-loss policy around the net retained losses. Ironically, he does so through a subsidiary of the same reinsurer whose higher deductibles forced him down this road in the first place. The outcome is that he can be confident of predicting his net retained losses to within $1 million. It is a reasonable compromise and enables him to deliver a level of stability that his board will accept.
More effective re/insurance buying is by no means the only benefit of the exercise. It can be a central tool, for example, in devising a wider mitigation strategy. Nonetheless, since insurance is clearly the risk manager’s responsibility in most large organisations, this is most likely to provide the most immediate incentive for creating a model.
It can also improve relationships with your underwriters. This kind of modelling exercise is not confrontational; it is about greater professionalism. It helps you to understand your underwriters’ thinking and can foster mutual respect. If the spread of this kind of practice is one outcome of the market changes since WTC and Katrina, then some good will have emerged from them.
This article appeared in Bermuda Reinsurance Magazine in February 2007