If you’ve read any of my past blogs about Solvency II you will know that I have a love/hate relationship with it. I fully believe in and support what Solvency II is trying to do but find the way it has been implemented extraordinary. But you can see why we are where we are: a combination of good intentions, an over-belief in rules and procedures, and the desire to create a single set of regulations that fits a diverse marketplace.
Principles-Based vs. Rules-Based Regulation
A common principles-based regulation across Europe—encouraging good business practice, protecting the consumer but without putting a drain on the business—would have been wonderful. But principles-based regulation implies a lack of explicit rules; without explicit rules how can common standards across Europe’s diverse insurance industry be guaranteed?
It can work within one country with a strong, homogeneous regulator—as evidenced by the UK’s ICAS system—but what if regulators across Europe cannot be sure to interpret and implement principles-lead regulation in a common way?
So now we have an odd amalgam of rule-based and principle-based regulation; supposed flexibility but backed with metre after metre of documentation outlining formulae, methodologies, requirements and guidance with, to my mind, insufficient clarity of purpose or method. Nowhere is this better illustrated than in EIOPA’s 46-page Guidelines for the application of outwards reinsurance arrangements to the non-life underwriting sub-module.
Sympathy for the Regulator: Making One Size Fit All
This post is the first in a brief series of blogs that consider the standard formula within Solvency II. As I have discussed in previous blogs, rather than encouraging firms to develop their own internal capital models, with all that implies for better risk understanding, Solvency II does the opposite. The weighty, overwhelming internal model approval process within Solvency II discourages what should be the pinnacle of the risk management process: to develop an internal model to assess risk and return.
The true value of a model is the process of designing and parameterising it, making sure that all risks and their inter-reactions are modelled and understood (and, crucially, the limits of data and modelling are also understood). The value is not in the number that pops out the end – which at such an extreme return period as 1 in 200 is frankly pure speculation no matter how much data is used and how many actuaries are employed – the value is in the thought and challenge process that has to go into deriving that number.
But regulators need a number and need to be able to confirm that numbers are consistently calculated so they can be compared. Rating agencies have the same issue: simple capital formula, such as those used by rating agencies, cannot suit all firms. But rating agencies have the freedom to interpret their simple models, adding modifiers and capital add-ons as their expertise and experience leads them. Solvency II cannot allow national regulators the same level of discretion.
To try to get around this, the standard Solvency II solvency capital requirement formula has been made amazingly complex, to try to properly reflect the differing risks of the full range of European insurers:
- big and small
- national and international
- mono-line and multi-line
- retail and wholesale
But no formula can do this, despite its size and complexity it still suffers the inherent flaws of a simple capital model, particularly for reinsurance optimisation: one size does not fit all.
Premium Risk and Reserve Risk: Excess of Loss Gets Lost
Putting catastrophe risk to one side for a moment, like rating agency models, the core of Solvency II’s insurance risk solvency capital requirement formula is based upon risk-adjusted factors applied to simple volume measures: net premiums for premium risk, best estimate reserves for reserve risk. Those factors represent the relative riskiness of different classes of business. So if you write more business, or more risky business, or have higher reserve requirements for more risky lines, your assessed solvency capital requirement will be higher.
So far all well and good. The only way to reduce the charge is to reduce the volume measure or diversify across more lines. The easiest way to reduce a premium is to take out proportional reinsurance; the easiest way to reduce best estimate reserve levels is to arrange a loss portfolio transfer.
Reinsurers have been marketing structured quota shares as ideal Solvency II capital products. We shall see later that all may not be quite as straight-forward as it seems, but it is certain that the standard formula, ex-catastrophe, does not reward non-proportional products to anywhere near their true economic extent. Yes an excess of loss reinsurance will reduce net premium, and an adverse development cover may reduce best estimate net reserves, but no way near proportionate to the benefit of the protections at the 1-in-200 level.
In the last full quantitative impact study, QIS5 in 2010, an attempt was made to use a formula to show the true value of excess of loss covers on premium risk, but it was so complicated and had so many requirements attached that, as far as I’m aware, nobody used it. At least it tried, and at least one reinsurer developed products to optimise its value, but by the publication of the drafts for the latest stress tests it had gone.
Instead, for three classes of business, it is now assumed that excess of loss is bought, and a 20% reduction from the published factor allowed, whether or not any excess of loss reinsurance is actually purchased or not. For the remaining classes no discount is allowed, again whether or not any excess of loss reinsurance is actually bought.
As we will see later, there are ways around this, undertaking specific parameters (USPs) or internal models, partial and full. But USPs, like internal models, need approval and are limited by rules and strictures. We don’t yet know how the approval process will work in practice and how common approval standards will be across national regulators.
Catastrophic Reinsurance Guidelines
Unlike premium risk or reserve risk, QIS5 maintained a freedom within the catastrophe risk element of Solvency II’s standard formula to show the value of a range of different reinsurance types, proportional and non-proportional. There were some examples given but broadly speaking it was for the insurer to show the value and the regulator to agree or otherwise.
This was too liberal for the European Commission. Freedom to interpret could lead to different standards in different territories; rules had to be applied. Willis Re took an active part in the discussions within the European Insurance and Occupational Pensions Authority (EIOPA) – arguing for maintenance–indeed extension– of that flexibility. We were successful; the draft reinsurance guidelines when they emerged in June of last year explicitly allow firms to show the value of excess of loss contracts; including:
- even stop loss
But the guidelines as they emerged are a confusing 46-page document. I will talk more of these guidelines in the next blog, but as I mentioned in my last blog I love the fact that EIOPA, commenting on the results of the consultation exercise, effectively acknowledged that people found the guidelines impenetrable but refused a general request for examples despite acknowledging how valuable they would be, as they “…cannot be included as part of the guidelines due to the legal nature of the EIOPA guidelines”.
At many stages in the calculation process, firms can try to make the case to their regulator that a different methodology to that mandated should be employed. Again, it will be interesting to see how national regulators react to these arguments. Unfortunately, Willis was unsuccessful in our attempts to get the principle of undertaking specific parameters agreed for catastrophe risk, but if the standard formula truly does not work firms can attempt to get a partial internal model approved, with all that that entails.
Gaming the Rules?
The trouble with any set of rules is that people will work out how to get around them outside of the spirit and intention of the rules themselves. This is true of sport, tax and solvency capital requirements. It is even worse when the rules are over-complicated and poorly drafted; it is easy to game the rules. Just as an unscrupulous actuary can show that a particular deal is wonderful if offered by his firm but an appalling waste of money if offered by another, it is very possible to design reinsurance that works well in the standard formula but is at best sub-optimal (and at worse positively harmful) in the real world.
Regulators are alert to this; the guidelines explicitly say that reinsurance should not be geared towards optimising the formula but meeting the real risk appetite of the firm. Willis wholeheartedly endorses this view—a firm’s risk appetite must be central to their decision making—but I suspect that rather more formula gaming will be going on than many regulators expect and realise. For example, the standard solvency capital requirement formula is blind to commission terms, an aggressive sliding scale may undermine the true level of risk transfer. Similarly for event excess of loss, rather fewer reinstatements may be needed to optimise the formula than the firm would be comfortable with in reality.
A firm’s reinsurance programme must be structured to meet their real need but at the same time it is important that that programme also can be shown to show true value, or as close to real value as necessary, in the standard solvency capital requirement formula. In the next blog we dive a little deeper into the reinsurance guidelines and in the last look at strategies to ensure that help a firm design reinsurance that works both for it and Solvency II.
This post was originally published February, 16 2015.