Delivering Solvency II internal models

32px-Flag_of_FranceInternal Models are as necessary as they are problematic. They enable insurers to capture and manage complex risks, yet remain vulnerable to misinterpretation and misuse. The more complex the model the more vulnerable it can become. In November 2014 research conducted by Solvency II Wire revealed that approximately 175 insurance and reinsurance entities across Europe were in pre-application for a Solvency II internal model or a partial internal model. Survey participants, mostly National Competent Authorities (NCAs), were also asked what they considered to be the biggest challenges for insurers preparing for an internal model. They identified four broad themes: handling the volume of work and amount of documentation; ensuring data quality; applying the use test; and adapting the model to local specificities.

Counting IMAPs

The Solvency II Wire research counted the number of firms in an Internal Model Application Process (IMAP) in each country. EIOPA, however, uses a slightly different methodology and counts the pre-applications by insurance groups, not individual entities: EIOPA considers that there are just over 100 pre-application processes under way. Delivering Solvency II internal models 3These processes can be broadly divided into three types. About a quarter of the entities are cross-sector groups. EIOPA is particularly interested in these entities, as issues arise from the inconsistencies between different supervisory authorities. Another quarter are domestic groups; the legal structures of these groups vary by country and as a result some countries have mostly domestic groups, whereas in others these are mostly solos. The remainder, about half, are domestic solos, which only involve one authority. Andrew Candland, Head of EIOPA’s Oversight Unit, told Solvency II Wire that he believes the number of pre-applications processes was likely to change over time. “I think the overall trend for the number of processes in EU is downward. Not all the processes will lead to applications for day one approval.”

Volume and documentation

Not surprisingly the sheer volume of technical work and related documentation required for producing an internal model is proving demanding for firms. The challenges identified by NCAs include managing the tight timescale and a lack of human resources and expertise. The latter is partially exacerbated by the increasing burden on specific functions and fulfilment of the four-eyes principle (stipulating that at least two people need to be involved in any key decision taken by the undertaking). The associated challenges around documentation include meeting the extensive documentation requirements and filling the EIOPA application template. For cross-border groups the challenge is more onerous as they have to manage the expectations and specific requirements in the different Member States where they operate. “The different standards across the various European supervisors are leading to a multitude of documentation [requirements] for the same aspects of the internal model,” Robert Lempertseder, Head of Risk Analytics and Reporting, Munich Re, said. This was particularly so for the more qualitative aspects of the documentation such as the use test and expert judgment where a range of standards seem to exist, he added. “Currently a group needs to address the different requests of all solo supervisors of the internal model. This issue can be addressed by better coordination within the college of supervisors or by more precise guidance from EIOPA on documentation standards.” EIOPA recognised the amount of work and technical challenges involved in preparing the models and has set up a number of initiatives, such as the Centre of Expertise in Internal Models, to help manage the process for both regulators and industry. Mr Candland points out that despite the onerous requirements there are a number of benefits associated with the work and data collection. “It is worth remembering that even though Solvency II has been a catalyst for a lot of this investment, the investment was probably always needed or would have been needed. Insurers are in the business of taking on risks and charging for them. From a pricing point it is in the interest of insurers to collect the data in order to ensure that their pricing is in line with the risks they accept and that they will continue to be profitable. This will help insurers to avoid insolvencies and, thus, improve protection of consumers.”

The importance of data quality

A number of NCAs expressed concerns that insurers would struggle with the stringent data quality requirements in Solvency II. And there is an expectation for very high standard of justification and understanding of where the assumptions for the model are coming from. Delivering Solvency II internal models 1“I think that [expectation] must be right because most models are extremely complicated and the numbers can vary considerably just by moving individual assumptions around,” Mr Candland, said. “In order to deal with that, it is necessary to really think through and have good processes to understand where the assumptions are coming from and to get confidence that the number coming out is based on sound processes and expert judgement.” Solvency II may actually encourage the data collection where there is currently a lack of reliable data (for example, for operational risk). EIOPA’s approach is “better late than never”, as Mr Candland explains. “You could say: ‘we cannot quantify this risk, we haven’t got the data’, and you keep saying that for twenty years. Or you say: ‘ideally I would have started collecting twenty years ago!’ But at least if you start collecting now, then in twenty years’ time we will have a reasonable set of data.”

Use test, the board and culture

Beyond the technical challenges, regulators singled out a number of challenges associated with application and integration of the model and its outputs into the business, decision making, risk management and governance processes: the so called ‘use test’. Vesa Ronkainen, Head of EIOPA’s Centre of Expertise in Internal Models, told Solvency II Wire that it was not surprising the use test requirements were a challenge. “These are complicated models and it is a new type of work for many board members,” he said. “The general idea is that requirements have to be proportionate.” Delivering Solvency II internal models 4“Of course the model developers need to understand basically everything about the assumptions. For board members one of the key issues is about validating and checking that the model is fit and appropriate for purpose. For some board members it may be challenging.” Mr Candland added that the issue is a key priority for EIOPA. “This is often discussed in our internal models meeting. We have some good practice for national authorities of what is an appropriate level of understanding of the board.” Justin Elks, Director of Risk and Solvency II Lead at the insurer Just Retirement, notes that applying the use test could be impeded by the need for Board members to learn and internalise the dynamics of the new Solvency II numbers after years of familiarity with the existing regime (the ICAS in the UK) -all before Solvency II calibrations are finalised and the regime is in force. “Management has been using the ICAS numbers for several years and as such are more familiar with, and understanding of, the current models and calibrations,” Mr Elks said. “They have a more innate understanding of the impact of changes and of what levers can be pulled.” Mr Elks believes it will take time to get a ‘feel’ for the Solvency II numbers, even where stress positions are looked at and understood. “This could impact management’s confidence level in the model and in making decisions.  Management also need to be aware of the limitations of models at different confidence levels, and the need to ensure calibrations work at 99.5% as well as the other confidence levels which can, and should be, used for risk-based decision making.” Michael Wainwright, Partner at the law firm Dentons, believes that the regulator will recognise that individual board members will need time to become familiar with the internal models used in their business. He also draws a distinction between incumbent and new board members. “A new member might start by asking why the business has invested in developing a particular internal model and what advantage they gain from using it, as opposed the standard formula,” Mr Wainwright explained. “They might also enquire about the parameters and data that drive the operation of the model, where these derive from and how they relate to the business-as-usual operations of the company. A longer standing member will be expected to understand the significance of the outputs of the model for management decisions in the business: for example, where to invest and where to rein back.”

Embedding a risk-based view

Introducing an internal model and applying the use test go hand in hand with the need to disseminate and embed a risk-based view and culture throughout the organisation. An area that survey participants identified as challenging. Delivering Solvency II internal models 6Lutz Wilhelmy, Director, Group Risk Management at Swiss Re, which has been using internal models since 1994, said that in order to embed the risk-based view across the organisation senior management needs to have its own definitive internal risk assessment that it believes is most suitable for the business. This assessment should form the basis for deriving all other risk-based regulatory views in a consistent and transparent way. “The most adequate view should be used to assess the risk-adjusted performance of all units and to determine risk-based compensation. Moreover, steering directions and limits set should be consequentially derived from it. Living and breathing the model embeds it in the firm.” In order to successfully integrate the risk-based culture into the organisation the firm must consider other forms of measurement that may be less appropriate to its own business model. These include the different accounting views, rating agency views, and regulatory views that are less advanced than Solvency II. In the case of Swiss Re this includes the Swiss Solvency Test. “These [views] have to be taken into account as boundary conditions. In some cases they might even be more decisive steering tools,” Mr Wilhelmy, said. “It has proven useful to handle conflicts between the different assessments transparently and to take conscious decisions when departing from an economic risk-based position.” To overcome these difficulties the firm is focusing on the key uses of the model and applying a number of strategies to engage the Board and senior management. “We are looking to build and further improve management familiarity with the models below the Board level and create ‘pull’ for model use through training and engagement activity,” Mr Elks said. “At the same time we are seeking to create some “push” pressure for use of the new model from the Board by acting as a demanding customer: for example, refusing to look at initiatives if Solvency II capital has not been considered appropriately. We have also successfully used peer pressure in management committees to improve engagement in Solvency II capital.” To really embed the Internal Model in the business, Mr Elks believes a “service culture” needs to be built around meeting the needs of business areas for information to support decisions, as opposed to simply producing regular reporting information, which has traditionally been the focus of actuarial teams. “Ultimately I’d expect Boards to be looking for opportunities to use the models in decision making in any case given the amount of investment and resources committed by the firm. Passing the use test should be secondary to getting a return on the Board’s investment in its Solvency II model.” Consideration should also be given to the fact that a change in culture may impact other areas of the business. The change to an economic risk-based steering will inevitably impact the profitability of some lines of business, according to Mr Wilhelmy. “This means that some businesses will grow and others will shrink or even become no longer viable. The change to a risk-based culture also entails other significant changes that need to be well-managed.”

Challenging and challenges for the Board

Solvency II will bring the role of senior management under scrutiny, for all insurance undertakings. But for those using an internal model the challenge for board members and non-executive directors (NEDs) is particularly acute. The increased responsibility and need to be at least familiar with the some aspects of the internal model have led to concerns that inevitably, as the responsibility placed on NEDs increases, less technically minded Board members might be less willing to take on the work. “Managing an insurance company is an increasingly complex business,” Mr Wainwright, said. “That does not seem to be deterring existing NEDs or new candidates yet. In time, the new senior management regime that the PRA is introducing for insurers may cause people to have second thoughts about the industry, especially if the regulators are more successful in their efforts to target enforcement action at managers, as opposed to imposing penalties on firms. For the time being, the sense seems to be that the banking industry will bear the brunt of this development, if and when it comes.”

Local specificities

The challenges discussed above apply to all internal model firms, but large cross-border groups and their local subsidiaries face an additional set of challenges in adjusting the model to fit local markets of varying size and industry sophistication. Delivering Solvency II internal models 2“We often see a challenge where you have a highly centralised group model, which is sent off to the far reaches of the group,” Mr Candland, said. “For example, if the local board was not directly involved in the Internal Model, it struggles to explain the model. That is clearly an issue that needs to be addressed by the group.” The main concerns expressed by survey participants include difficulties in understanding complex models and the appropriateness of the model for local conditions. “You could see such examples in non-life insurance where there are particular risks that are geographically relevant, for instance, flood risk or earthquakes. If the model was built in the group headquarters in another country, it might not be adequately capturing those particular risks faced by some smaller subsidiaries.” Some of Europe’s largest groups such as Swiss Re and Munich Re are aware of these challenges and have developed a number of solutions over time. Swiss Re implemented a consolidated group model for internal steering purposes more than fifteen years ago, Mr Wilhelmy, said. “Since then, the model is constantly further developed and refined to generate adequate consistent assessments for new business and also for smaller legal entries.” Typically the on-going refinement of the model does not lead to material changes at the group level and local boards have the ultimate decision to adopt model changes. “In order to facilitate a smooth process, they are involved early in the model development process and are requested to endorse any model change before the group committee authorises them for group-wide use and makes the model change available for local adoption.” Munich Re engages with local entities to assist them with tailoring the model to specific conditions. “We provide guidance to our subsidiaries as to where they can deviate from group standards and how this needs to be governed and documented,” Mr Lempertseder said. “Any change in these adoptions also fall under the model change policy. This approach allows local specifics and enables central control.”

Quest for the holy BAU grail

At present most firms are engaged in preparing to submit their internal model application, having completed the pre-application stage. Substantial resources have been spent designing and implementing the models, and, as with most things Solvency II, the aim is to move into business-as-usual (BAU) operations as soon as possible to start reaping the benefits of implementation. EIOPA recognises that the process will take time and may vary by country and company size. “It will not be unusual for companies to need a couple of goes through the cycle before the process can be considered BAU,” Mr Candland, said. “I imagine the next couple of years will continue to be challenging and then we will gradually go down the slope of effort needed.” Or as Mr Ronkainen puts it, “Of course we could assume that the first application for internal model is the most difficult one. The next time you can use the experience. So in future that would hopefully decrease the amount of work that is needed.” BAU may be the holy grail but it is most certainly not the end of the quest. Mr Candland emphasises that model work is always work in progress. “It will be important that the on-going model validation work continues, so firms spot when they need to improve their model to reflect changes in the external environment as well as internal changes to the firm and the mix of risks on its balance sheet.” — To subscribe to the Solvency II Wire mailing list for free click here. Delivering Solvency II internal models 5]]>

Print Friendly, PDF & Email

3 thoughts on “Delivering Solvency II internal models

  1. I’m troubled by the statement that capturing 20 years of data will make some risks more quantifiable. Think back 20 years ago and think about how relevant those conditions are to today. 20 years ago the internet was not what it is today, processes are not close to what they are today, the idea of a smartphone had just barely been introduced, etc..
    Any analysis of risk has to consider how much the environment is changing to evaluate how much data collection is possible (under the old environment) before the old data becomes irrelevant.

  2. Should we stick with pillar 1 or face the risks stemming from pillar 2 aswell? should we stick with ORSA or use an IM instead? which is better for the undertaking and as well for the regulatory purposes? Pillar 1 SF is enough for all the risks impacting the business? Can we quantify (modelling methodology) all the risks impacting the business?
    International (EU wide at least) internal model = Standard Formula.

Comments are closed.