The Intelligent Quarterly from the publishers of The Insurance Insider

Spring 2012

Search archive

Advanced search

"There's more to Solvency II than just the numbers, you know," says the nice man from the Financial Services Authority (FSA), plaintively.

Yes, we know. But with the publication of QIS5, numbers currently dominate the universe. Firstly, we've had about half a dozen versions of the spreadsheets, each with corrections and variations but no change to the submission deadline despite all the extra work. Secondly, the numbers add up to the scary conclusion that many smaller companies and mutuals just won't get over the capital bar if it remains this high.

Thirdly, we are all still waiting for a clear picture of what Pillar 5 is actually going to require of us. ("Pillar 5" is yet another bit of Solvency II jargon for Pillar 2 governance plus Pillar 3 disclosure, which = Pillar 5.)

Many mid-size companies are reluctantly coming to the conclusion that Solvency II really will be done unto us, and the semi-compulsory nature of QIS5 has brought home just how much it's going to hurt. "So raise capital," the regulator blithely tells us.

Whither capital?
If only it were so easy. We are in the midst of a major recession that has pretty much wiped out any investment returns an insurer may have garnered. The insurance pricing cycle is bumping miserably along the bottom, with some brave talk but no real sign of a recovery on the horizon.

With the notable exception of Brit (following the recommended Apollo/CVC takeover offer), most insurers' shares are trading at a discount to net asset value.

Trading at a discount is generally a reliable sign that investors don't like your results, your returns or your business model. And if investors don't like you, where exactly does the regulator fondly imagine insurers are going to raise capital? It is the capital markets equivalent of telling the fat frump at the back of the class to ask the school heartthrob for a date.

Easier said than done
The other option is to get regulatory approval to use an internal model for capital calculation rather than the punitive standard model underlying QIS5. That in turn brings us back to "more than just the numbers".

Obviously it starts with the numbers, as produced by the "calculation kernel", which is the modelling mathematical black magic supplied by the actuaries.

This kernel must be able to supply "an appropriate" calculation of the firm's capital needs, as well as satisfy statistical, calibration and validation standards. In plain English, the actuaries can't simply change the inputs or the parameters of the model because they don't like the outputs.

There has to be an acceptable, documented rationale behind the assumptions and parameters of the model. Validation is all about getting an independent critical third party to work it through and make sure the model all hangs together to produce answers that are realistic and relevant.

So far so good. But the calculation kernel is the easy bit.

The tests that a company must pass to gain permission to use an internal model - which will hopefully produce a capital number sufficiently low to be met within a company's existing capital base - go beyond mere number-crunching.

Firstly, the inputs and outputs of the model must reconcile with the company's financial reporting. That might seem bleeding obvious, but it's supposed to prevent companies from finessing their capital numbers.

Then the company's board and senior management must understand what goes into the model, how it works, and how it fits with their business model. A management cop-out such as "it doesn't fit our business model but we like the outputs so we'll use it anyway" does not fly.

Another apparent statement of the obvious is that the model must cover enough of the company's risks to make it useful for risk management and decision-making. Insurers have no wiggle room to create pretty models that ignore whatever elephant is in their particular room.

The model must also support and verify decision making in the company, and be regularly discussed at board level. If a company decides to change its business mix or strategic direction, the board must document that they have considered the likely impact on their capital requirement as produced by their internal model.

Also the model must translate into consistent day-to-day business decisions. As an aside in this regard, it is the board's responsibility to design staff compensation systems in a manner that will reward good risk management and good business results, so it is likely that pay scales will be under review over the next couple of years.

The model must be widely integrated into the company's risk management system - no bolt-ons, no Excel spreadsheets, no manual data dump. Management information and anything else required for the capital calculation must come directly from the company systems, such that an outsider could dump information straight from a company's systems into their model and produce substantially the same results as the company's own actuaries. For many companies, that means a huge IT spend to upgrade hitherto un-integrated systems.

Those companies brave enough to embark on the internal modelling journey are even now submitting their PAQC for the IMAP - or their Pre-Application Qualifying Criteria for the Internal Model Approval Process.

This is supposed to be a document of not more than 30 pages outlining the scope of the model, its inputs and validation methods. It must also include a list of workstreams launched to ensure that procedures are in place or being designed, and policies are embedded, or being documented. It must include a gap analysis, together with timetables to close the gaps identified.

There must be evidence of board approval for the (probably large) budget. And most especially, the PAQC must include a contingency plan: what will the company do if its model is not approved and it is forced to default to the standard model with its higher capital requirements?

Continental drift
At the last count, 110 UK companies of various sizes had thrown their hats into the ring, compared to five in Germany and three in France. It's not that UK companies are braver than their European counterparts, it's just that the UK has more medium-sized companies with more to lose from the standard model.

Even for those companies that stick with the standard model, there are still burdensome corporate governance requirements.

Many of them can be subsumed under the requirement to produce an annual Own Risk and Solvency Assessment (ORSA).

Nobody has yet published a believable format for ORSA, but bright minds at PricewaterhouseCoopers suggest that it should be a report from the board to the regulator in a "format that can be easily understood at a high level and contain all the relevant information for the firm and the FSA to make an informed judgment as to the appropriate capital level and risk management approach".

Pillar 3 serves up more acronym salad: RTS (Return to Supervisor) and the SFCR (Solvency and Financial Condition Report). RTS is a confidential annual report to the regulator discussing the company's financial condition and performance, as well as its risk profile present and future, and sensitivity to market volatility.

SFCRs will likely be published along with the annual report and accounts. Annoyingly, if you try to find out what exactly these reports are supposed to encompass, you get... very little.

The European Commission of Insurance and Occupational Pensions Supervisors (CEIOPS), which is charged with developing the guidance, says it is beavering away.

No doubt it is, but so far we have little more than vague hints and speculation about what will need to be disclosed. Deloitte has suggested that in years to come companies will be required to discuss the results of backtesting - i.e. comparing their model prediction with what actually happened. Ernst and Young have speculated hopefully that CEIOPS will require companies to publish metrics that will make a comparison between companies more readily achievable. Others merely shrug.

So here we are, lumbering towards the 2012 deadline, working up to the Great Leap Forward... with no idea of how high we have to leap. Perhaps QIS6 - which has suddenly appeared on the horizon even though QIS5 was supposed to be the last one - will bring greater clarity.