Mr Grund, what experience has insurance supervision gained with regard to internal models during the first 18 months of the new supervisory regime?

We have gained a lot of experience and it has been consistently positive. Our expectations have been confirmed: internal models reflect the individual risks for insurance undertakings even better than the standard formula. The phenomenon of negative interest rates, for example, can be tackled much more effectively.

We have, of course, also experienced – little to our surprise – that the process entails a lot of effort for both undertakings and supervisors. It is therefore all the more important that BaFin enjoys the benefits of an integrated financial supervisory authority: because banks have already been using internal models since the end of the 1990s, we can draw on the expertise of our colleagues from banking supervision and the cross-sectoral department for quantitative risk modelling when carrying out our assessments.

Is the effort worthwhile? It is clearly only a minority of insurers that use internal models.

If it were purely a matter of the number of undertakings concerned, then this question would be entirely justified. Germany has more than 300 Solvency II undertakings and only 36 of them use an internal model. In addition to this, there are seven group models with which we work as home supervisors in some cases and as host supervisors in others.

If, however, you look at the market share affected, a very different picture emerges: it is predominantly very large undertakings that are concerned here, accounting in total for around 50% of the German insurance market. With this in mind, the effort involved is absolutely appropriate.

BaFin received a flood of applications for approval in the first few months. Are things quieting down now?

That great wave of approval applications is definitely over. But one thing is clear: every application precedes another one. This is because internal models need to be continuously adjusted, be it due to changes in business activities, external circumstances or the ongoing technical development of the model itself. BaFin will therefore have to continue working intensively with internal models. However, these intensive efforts are also worthwhile.

How is it worthwhile? What are the advantages of internal models from your point of view?

The advantages for BaFin are obvious: we can observe the undertakings, their risk profile and their risk and capital management much better than we can with the standard formula and, if necessary, we are able to respond accordingly in a more targeted manner. Apart from this, internal models are also an important management tool for the undertakings. The intensive work associated with these models results in much better identification of potential risks and countermeasures.

It goes without saying, of course, that internal models cannot replace plain common sense, nor can they reflect all risks. Things will always turn out differently in reality. Therefore, I would never go so far as to say that internal models are a panacea for all ills. Nonetheless, I view them as the premier league of risk management in insurance undertakings.

Why is it so important to supervise internal models on an ongoing basis?

An internal model must not only be suitable at the moment that the application is approved, but rather at all times – for so long as the undertaking is conducting business. The approval of internal models and ongoing supervision are thus inseparably linked to each other.

Ongoing supervision must also ensure that, when they are developing their models, undertakings do not take account of only those factors that will prove beneficial to them with regard to capital requirements. In order to be adequate, modelling must cover all components, the entire spectrum – and this also includes any potential negative developments necessitating additional capital-related measures. By conducting ongoing supervision, we are preventing internal models from being abused solely for purpose of capital optimisation.

Are there any indications that undertakings are attempting to do this?

It is not something that I would rule out. However, we are confident that our ongoing supervision of the models means that we have any such tendencies under control.

What precisely does ongoing supervision mean?

Ongoing supervision means, first of all, that there is constant communication with the undertakings about their internal models and business. If irregularities arise, we intervene swiftly. Furthermore, in Germany we have set up – in addition to standard reporting – a dedicated reporting system for internal models. This is not yet the case elsewhere and it is an area where we would like to see more European harmonisation. In addition, we give particular attention to conducting cross-comparisons between different models to gain useful insights into the appropriateness of the parameters used.

What are the greatest challenges for BaFin when it comes to internal models?

Internal models endeavour to measure undertakings' risks with tailor-fit precision. This means that BaFin requires employees who are capable of working on an interdisciplinary basis to assess the appropriateness of such models. BaFin is excellently positioned in this regard. Our interdisciplinary collaboration between modelling experts, economists and lawyers has developed to a highly advanced level, even by comparative international standards. This is something we can be proud of at BaFin, but of course it is also something we take as an incentive to spur us on to further improvement. Another – quite appealing – challenge is, of course, the international character of our work in the supervisory colleges for the large groups.

In terms of content, the main challenge I see is the comparability of the results. Whatever the individual specifics of business models, business and risk strategies or channels of distribution may be, the key information provided by internal models must remain comparable. This holds all the more true with Solvency II as we are dealing with a flexible solvency regime that makes it difficult to compare figures. There is still quite some work to be done on convergence in this area, in particular at the European level.

What role does the European Insurance and Occupational Pensions Authority, EIOPA, have here?

EIOPA has the main role in this regard. Its paramount task is to ensure that supervisory principles are applied in the same way throughout Europe. This is a matter of particular importance because some supervisory authorities in which cross-border groups are operating are navigating unfamiliar waters when it comes to assessing internal models. Therefore, EIOPA always has a seat at the table in the colleges concerned. Furthermore, it develops tools, such as a manual for ongoing supervision, and conducts specific projects to facilitate consistent supervisory practices, such as comparative studies on individual subjects.

At BaFin, we are committed to contributing to this process and provide considerable impetus in this regard. After all, it is only if internal models are consistent throughout Europe that they will achieve sustained acceptance as a supervisory instrument.

What level of acceptance have internal models gained at the undertakings themselves?

That is not an area where I have any worries. I don't think that any of the undertakings currently using internal models will ever turn their backs on them. Their ability to accumulate knowledge and their risk management capabilities have improved so substantially thanks to these models that they are no longer in the position to abandon them. This is particularly true for the large, international insurance groups.

Let's finish up by talking about your predictions for the future. What developments lie on the horizon for the supervision of internal models?
There will be some extension applications for existing internal models. These might relate to risk categories, such as operational risks which are not yet covered, or perhaps additional group undertakings from other countries. That, by the way, is another convergence issue that we need to work on, because internal models are still not very popular with some national supervisors.

Furthermore, there are also a lot of questions that we have to deal with in relation to Solvency II in general and we will, of course, pay particular attention to these when it comes to internal models, for example negative interest rates or the correct calibration of risks for government bonds.

And what developments do you expect on the international front?

The first reviews of Solvency II are coming up over the next few years. Both the solvency capital requirement under the standard formula and the long-term guarantee measures will subject to scrutiny. For us, it is important that we share the findings from our supervision of internal models here. At the same time, it is also a matter of developing a common European supervisory culture. Convergence will remain an important topic for the supervision of internal models and one where we will be making ourselves heard.

Furthermore, at the International Association of Insurance Supervisors (IAIS), it is currently being discussed whether internal models should also be permitted in the International Capital Standards for risk measurement. This is another area where we are very clear in expressing our opinion: it is our view that, as the premier league of advanced risk management, internal models are an indispensable component of any supervisory regime. However, some major countries are sceptical about this. They worry that in extreme cases this will mean arbitrary self-regulation by the insurers. This danger can only be counteracted by supervising internal models intensively and uncompromisingly. We are confident that we are on the right track in this regard.