QIS 5 - Making political capital

With the fifth quantitative impact study looming, Ceiops has released its final advice for Solvency II’s level two implementation measures to howls of indignation from the industry. The SCR standard formula’s calibration risks driving a wedge between Ceiops and the EC, as the latter comes under pressure from its industry constituents.

“Just because something is proposed by Ceiops, the European Commission is not obliged to accept it. We will get the views of the stakeholders and make the decision.” The comments by Karel Van Hulle, head of insurance and pensions at the European Commission’s (EC) internal markets division, encapsulates the division faced by the European insurance industry as it gets ready for the fifth quantitative impact study (QIS 5).

While the EC is the driving force behind the directive, it is the Committee of European Insurance and Occupational Pensions Supervisors (Ceiops) that is charged with advising it on the actual rules – and the two groups are not seeing eye-to-eye.

In particular, the formulation of the standard formula for a floor for the solvency capital requirement (SCR) for a company’s assets, below which they are supervised more heavily, was the remit of Ceiops. The formula was to be on a consistent economic basis, but sufficiently severe as to incentivise the use of internal models, whose bespoke nature it was thought would lead to better risk management.

However, in the fourth quantitative impact study (QIS 4) carried out in the first half of 2008, the calibration of this formula came into the spotlight when it undercut most firms’ internal models, thereby destroying the incentive to run them. This was partly because of two late changes by the EC over Ceiops’ head – the use of an interest rate swap for the risk-free discounting curve in the technical provisions, and the use of a lower 32% shock in the equity risk submodule of the formula.

As the stakeholders geared up for QIS 5 – the first such study since the worst of the financial crisis – Ceiops began work on a new calibration of the formula. While Ceiops will offer official guidelines on the QIS 5 specifications in March ahead of its August–November implementation, its advice to the EC for the level two implementation measures of the Solvency II directive is widely seen as the first blueprint.

The draft advice released in November last year saw complaints from the industry that it was guilty of over-prudence. This charge has not disappeared with the January publication of the final advice, despite numerous changes to the calibrations, notably in a weakening of the correlation specifications at the behest of the Chief Risk Officer Forum, an industry lobby group.

“My feeling is that Ceiops’ proposals are linked to the financial crisis from the banking side. It is clear the insurance industry is paying for the sins of the bankers,” says Yanick Bonnet, secretary-general of Gema, the French industry body for mutual insurers.

Although at the time of going to press, Ceiops is still to release advice on key elements such as the calibration of the technical provisions and the non-life risk module, essentially the calibration of the standard formula for life insurers’ SCR is announced. And it does not look good to much of the industry, which is again looking to the EC to provide a way out.

Tougher
“QIS 5 will be tougher than QIS 4, partly because the SCR calibration is more conservative, partly because firms’ asset values have fallen, and partly because of more prudent liability valuations,” says Andrew Smith, a partner at consultancy Deloitte. “The industry is certainly concerned about the results looking bad, but there is scope for the EC to push back. As a result a lot of people are expecting QIS 5 to be weaker than Ceiops’ final advice. There are a few key assumptions, such as the equity stress that have a large effect on the overall capital charge, and these may be the focus of adjustments.”

Carlos Montalvo, secretary-general of Ceiops, is confident that the EC will heed the bulk of his body’s advice when setting the parameters for the next study, due to take place between August and November this year, although he acknowledges there will be differences: “I can expect that the QIS 5 parameters will be broadly consistent with our final advice, but the EC is likely to make some changes, as in the end it is its responsibility to come up with the final specifications.”

However, Van Hulle is critical of the final advice. “On the calibration advice we are not totally in agreement with Ceiops. The feeling is it has been overly prudent, in applying extreme scenarios to the insurance industry. It wasn’t the cause of the crisis, and we do not want insurers scrambling around for capital just when the banks are undercapitalised.”

The apparent opposition of Ceiops and the EC has spooked some in the industry. “The fact that the EC appears to be listening to the industry is clearly welcome, but the fact there is still disagreement between it and Ceiops is a concern,” says one senior UK industry figure. “It would be worrying for the insurance industry if we entered 2011 and there wasn’t agreement between the EC and Ceiops, as this makes the implementation of Solvency II much more challenging.”

Formula
Like its predecessor, QIS 5 will require insurers to produce SCR numbers in line with the Solvency II standard formula, so the industry has to take note. But there is disagreement over how important this figure really is, given that most of the larger groups will be using an internal model when the directive comes into force.

Jorg Schneider, chief financial officer (CFO) at Munich Re, is dismissive of the idea that a conservative standard formula calibration could add to his firm’s capital requirements by the excessively prudent treatment of its primary insurance subsidiary, Düsseldorf-based Ergo. “We don’t believe it – we are running internal models. The most recent discussion has focused on the calibrations for the standard formula and this is not an issue for companies using internal models. So, for some individual companies there will be a higher standard capital requirement, but not for Ergo.”

Bruce Porteous, head of Solvency II and regulatory development at Edinburgh-based insurer Standard Life, is also clear that the standard formula is not essential. “The [UK] Financial Services Authority is not saying the standard formula is a benchmark for internal models. It’s not irrelevant though – it needs to be calibrated correctly to allow an internal model incentive.”

But others see the standard formula as playing a vital role in helping local supervisors to gauge capital levels. “In practice, supervisors will be most familiar with the standard model,” says Jeroen Potjes, chief insurance risk officer at the Hague-based ING, “and they will likely consider this as a useful reference benchmark. Therefore, if the standard model is inappropriately calibrated, this will make the differences that much bigger and that much harder to gain approval.”

Gema’s Bonnet is critical of those who would seek to ignore the formula. “I understand some big groups do not want to publish the standard formula result because they do not want to compare the reduction. But you should always have to use the standard formula, and compare it to your internal model,” he says. “Three words: level playing field. If you want to be able to compare companies across borders, you have to be able to compare the standard formula numbers. That’s why we are fighting against too high a standard formula.

“The main question is that you have a lot of submodules in the standard formula. For a life company, 80% of the risk is market risk, so if you have a good partial internal model, you can get a good reduction.”

Disastrous
The expectation of the final advice providing a blueprint for QIS 5 has given way to a feeling in the industry that the EC will have to step in. “The commission told the industry that the QIS 5 SCR would not be significantly bigger than QIS 4,” says Gregor Pozniak, secretary-general of Amice, the European industry body for mutual and co-operative insurers. “It is hard to see how this will hold if the Ceiops calibration is used.”

Industry figures, including one member of the CRO Forum, had put the potential impact of a QIS 5 based on the November draft advice as doubling the QIS 4 SCR. The January revision seems to have scaled back these doomsday scenarios, but not by much.

ING’s Potjes expects a “significant increase” over QIS 4, while Markku Paakkanen, CFO at Finnish insurer Tapiola, still thinks a QIS 5 informed by the Ceiops calibration would be disastrous. “It’s a rather worrying situation. The November advice would have doubled the QIS 4 SCR for life insurers and the failure rate would perhaps be as high as 40–60%. With the adjustments, I would expect the SCR to be around 150% of the QIS 4 level, with failure figures of around 20–30%,” says Paakkanen.

“To my mind, this is too high a figure. It is at the [practical] limit, because it might be politically difficult to have a favourable view in Parliament of a directive that pushes that much of the industry into insolvency.”

By contrast, 12% of QIS 4 participants failed to meet the SCR requirement, down from 16% in QIS 3.

The effect of the conservative calibration is compounded by the perilous state of many insurers’ assets, according to Gema’s Bonnet. “We have a sharp increase of around 50% in the SCR over QIS 4, and at the same time a decrease in own funds due to the financial crisis. With this situation, there will be a large proportion of insurers in Europe failing to meet the SCR, perhaps not far from 50%. It is not credible to propose a calibration like that, and it would not be politically correct for the EC.”

Standard Life’s Porteous has a less dramatic take. “Saying it will double is probably going too far, but it will be materially bigger. If Ceiops’ final advice all went through as it stands, it would not be a pretty picture. The [EC authored] level two text will hopefully be more reasonable.”
While the three players disagree on the precise nature and scale of the impact, it is clear they all agree on the solution: politics.

Failure
“The job of the EC is to take a position between the industry and Ceiops – it’s right to have compromises some of the time,” says Bonnet. And the industry is wasting no time in communicating its fears to the EC.

“A 50%, 60%, 70%, 80% failure rate. I hear these numbers all the time from the industry,” says Van Hulle at the EC. “Ceiops always goes to the worst-case scenario, and that is not appropriate. We have to be reasonable. Of course the EC will not back regulation that will push half the insurance industry into the sea. It is preposterous.

“Ceiops gives advice to the EC, but it is up to the EC to draw conclusions. So all these draconian and pessimistic scenarios are not justified. I do not see any reason why QIS 5 should result in a significantly higher SCR result than QIS 4.”

Noting that there were 1,400 participants in QIS 4, but 5,000 companies in Europe, Tapiola’s Paakkanen sees the EC as the best hope of preventing the collapse of the smaller players of which quantitative data is so far lacking. “With stricter rules we may have 30% of those [QIS 4] participants failing, plus 3,000 who haven’t participated at all so far. It is enough to make your hair stand on end,” says Paakkanen.

“So I wonder what the commission will decide with respect to these companies – you must keep the whole system going. There must be some transitionary rules for those in serious difficulty. It could lead to mass insolvencies across the industry. It is hard enough for some companies. How hard, we shall see.”

Ceiops’ Montalvo concedes these kinds of nightmare outcomes would be unacceptable. “It’s a valid point – we do not want a system implemented that would create an unacceptable failure rate. It would be a failure of the Solvency II project, and therefore also of Ceiops, if it led to 50% of the industry becoming insolvent. I am confident this will not be the case.”

However, he remains sceptical of the industry complaints. “It’s still premature to come up with concrete figures, as sometimes they can cause alarm. I’d like to see the data backing those failure rate numbers. There is always technical data supporting our advice, and we don’t always see that from the industry. It’s fine to suggest changes, but the problem comes when claims are made without data, as is sometimes the case. Where concerns have been supported with data, Ceiops has moved and adjusted its advice.”

But as is so often in politics, there may be an element of subterfuge. Deloitte’s Smith has a dark take on the power plays. “You need to recognise that with the way firms respond to QIS 5, there is a bit of a game going on here. If there is a massive failure rate, then the likely outcome would be a big stink at the EC and the requirements weakened.

“So everybody can see that there is no incentive to minimise their stated QIS 5 requirements, as this would weaken the argument for weaker requirements in the final version. I don’t think the industry will be trying as hard as possible to pass QIS 5.”

Van Hulle doesn’t dispel this notion when he speaks of the doomsday failure rates he hears from the industry. “I listen to them when they say that, and I smile and they smile back, because we both know it isn’t true.”  

Equity shock
There are several candidates for change in the Committee of European Insurance and Occupational Pensions Supervisors’ (Ceiops) calibration of the standard formula for the Solvency II solvency capital requirement (SCR) – the credit spread stress, the yield curve stress, the correlation matrix – but one aspect that has proved controversial is the design of the equity risk submodule of the market risk capital charge.

QIS 4 was enacted pre-crisis, when the conventional wisdom was that a 32% stress test was sufficient for the one-in-200 year value-at-risk for equity markets. With the German DAX index plummeting 40% during 2008, this assumption could not possibly fly in the current prudential focus in regulation. Accordingly, Ceiops moved to a stricter 45% level for exchange-listed equities, and 60% for unlisted, in the November draft advice.

“On market risk, the industry cannot say we don’t need to increase it, because it is not true – we saw the financial crisis,” concedes Yanick Bonnet, secretary-general at Gema, the French industry body for mutual insurers. “It’s clear that market risk was underestimated in QIS 4, especially on equities.”

After protests from the industry, Ceiops relented on the unlisted equity shock – pulling it back to 55% – but insisted a majority of its members supported the 45%-listed equity figure, although a minority was in favour of a lower 39% stress. One member – widely believed to be France’s insurance regulator Acam, although it refused to confirm this – maintains support for a 32% shock. “The stress is now far in excess of QIS 4, and this is something the industry has been commenting on all the time,” says Markku Paakkanen, chief financial officer at Finnish insurer Tapiola.

Perhaps more controversial than the figures themselves has been the methodology used to support them, which one industry expert claimed was so against the economic basis of Solvency II that it would not have been sufficient to pass Ceiops’ own guidelines for the statistical backing of internal model procedures.

Andrew Smith, a partner at consultancy Deloitte, agrees that the approach is somewhat less than scientific. “You can’t prove 45% is wrong, but equally it’s probably not what an objective researcher would come up with if they were asked to figure out the 99.5th percentile of the distribution. It would probably be more. The only way you’d get an equity stress in the low 30s, as some wanted, is to ignore the big falls.”

Carlos Montalvo, secretary-general at Ceiops, is dismissive of such criticism, and hits out at what he sees as tinkering with the standard formula. “The standard formula is different from what an undertaking can do for the purpose of internal modelling. We cannot imagine that a company would apply for an internal model by copying the standard formula. The question shows very well that the requests from undertakings for changing the standard formula ultimately end up in making the standard formula looking more like a standard model, which Ceiops tries to avoid.”