in the context of health insurance
Bas Groot
Master’s Thesis to obtain the degree in Actuarial Science and Mathematical Finance University of Amsterdam
Faculty of Economics and Business Amsterdam School of Economics Author: Bas Groot Student nr: 0448702
Email: basg86@hotmail.com Date: October 26, 2013 Supervisor: Prof. Dr. R.J.A. Laeven Second reader: Prof. Dr. R. Kaas
Abstract
This thesis discusses the quantitative aspects of the Own Risk & Sol-vency Assessment (ORSA) for insurance undertakings in order to find out what can be learned from these aspects. The ORSA looks ahead multiple years, which is dealt with by the projection of future balance sheets and approximations of the future required capital. A quantita-tive example is performed for a (fictional) Dutch health insurance un-dertaking looking at multiple scenarios as a sensitivity analysis, result-ing in insight with respect to the main risk drivers for the undertakresult-ing and the development of the corresponding capital requirement. The position of the management regarding the ORSA is essential, ranging from input to management actions according to the ORSA results. Finally also possible improvements of the ORSA model used and in-centives for the ORSA are discussed.
Keywords Solvency II, Own Risk and Solvency Assessment, ORSA, Forward Looking, Sol-vency Capital Requirement, SCR, Balance sheet projection
Preface vi
1 Introduction 1
2 What is ORSA? 3
2.1 The goals of the ORSA . . . 3
2.2 The ORSA process . . . 5
2.2.1 The preparation stage of the ORSA . . . 8
2.2.2 Medium Term Planning . . . 8
2.2.3 Required Capital . . . 8
3 ORSA modelling 9 3.1 Forward Looking . . . 9
3.1.1 Balance sheet projections . . . 9
3.1.2 Capital requirements . . . 12
3.2 Starting position . . . 16
3.3 Assessment of the Overall Solvency Needs . . . 16
3.4 ORSA base case & scenarios . . . 17
3.4.1 Base case . . . 17
3.4.2 ORSA scenarios . . . 18
4 Results & analysis 20 4.1 Base case results . . . 20
4.1.1 Base case conclusion . . . 23
4.2 Liability Scenario . . . 23
4.3 Asset Scenario . . . 25
4.4 HRES Scenario . . . 26
4.5 Contracts Up Scenario . . . 28
4.6 Contracts Down Scenario . . . 30
5 Improving the ORSA model 33 5.1 Forward Looking . . . 33
5.2 Asset Liability Management . . . 34
5.2.1 Investment portfolio . . . 34 5.2.2 Premium policy . . . 34 5.2.3 Capital requirement . . . 35 5.3 Management actions . . . 35 5.3.1 Incentives . . . 36 6 Conclusion 37
Appendix A: Article 45 of the Directive 39 Appendix B: Base case 41
Appendix C: Liability scenario 45 Appendix D: Asset scenario 48 Appendix E: HRES scenario 51 Appendix F: Contracts Up scenario 54 Appendix G: Contracts Down scenario 57
I would like to dedicate this thesis to my daughter Eva, who was born during the time I was writing this thesis. Due to a turbulent pregnancy and a difficult period after birth, the process of writing has been delayed for a significant period. This makes the completion of my Master’s Thesis in order to obtain the Actuarial Science and Mathematical Finance degree even better, as I look forward to having more spare time. At least until the Actuarial Practice Cycle of the AEMAS programme starts.
I would like to thank my supervisor Roger Laeven for his guidance and support and also my employer Co¨operatie VGZ U.A. for providing extra time in order to work on my thesis. I would also like to address my gratitude to my wife and family for their support.
I hope you enjoy reading!
Introduction
The Own Risk & Solvency Assessment (ORSA) is a part of the future regulations for (re)insurance undertakings called Solvency II. Compared to the current regulation (Solvency I) which is volume (premium, claims) based and can differ between European countries due to the different requirements set by national supervisors, Solvency II aims for risk-based and harmonised European regulation. The basis of the new Solvency II regulation is the Solvency II Directive in 2009, but the European Insurance and Occupational Pensions Authority (EIOPA, previously CEIOPS: Committee of European Insurance and Occupational Pensions Supervisors) has been giving advice on the new regulation since 2004. The effectuation of Solvency II was originally intended for 2013, but has been postponed to 2014 until recent developments stating that Solvency II will come into effect on January 1st 2016. In the mean while EIOPA provided guidelines in order to maintain overall progress of implementation, but the national supervisors have to decide if and to what extent these guidelines need to be implemented.
Solvency II is based on a three pillar structure:
• Pillar 1 is about the calculation of the Solvency Capital Requirement (SCR): the amount of capital needed for undertakings such that the undertaking has enough capital to meet their insurance obligations on a 1 year horizon with a 99.5% probability. The calculation of the SCR can be done by a standard model or a (partial) internal model which has to be approved by the national supervisor in order to use it for Pillar 1. As most insurance undertakings use the standard SCR model and in particular the focus of this thesis, Dutch health insurance undertakings, do use this standard model, the focus will be on the usage of the standard model.
• Pillar 2 focusses on risk management and governance, which includes the ORSA to assess the financial position of the undertaking on a longer horizon, using own defined risk scenarios. More details about the ORSA will be given in the next chapter.
• Pillar 3 is about the reporting aspects of Solvency II, both to the general public as to the supervisor. The reports include balance sheet information, the SCR as calculated in Pillar 1 and the ORSA performed for Pillar 2.
As this thesis is written in the context of health insurance and in particular a Dutch health insurer, a little background information about the specifics of the Dutch health insurance system is needed. In 2006 the system was changed from a national health insurance (‘ziekenfonds’) and a private health insurance to the current obligatory basic health insurance (‘basisverzekering’) which can be complemented by supplementary insurance (‘aanvullende verzekering’). The benefits described in the policy of the basic health insurance are determined by the Dutch government, such that the fundament of the insurance policy is the same regardless of the undertaking. The basic health
insurance is financed in two ways: a non-differentiable (due to the obligatory nature of the insurance) premium paid by the insured themselves and a budget provided by the College for Health Insurers (‘College voor Zorgverzekeringen’ (CvZ)). This CvZ income is based on the characteristics of the people insured and aims to cover (together with the direct premium) all claims for the insured. The CvZ income is calculated during the insurance year, which is January 1st of every year and ends December 31. The Dutch basic health insurance also makes use of a special component of the standard SCR model to calculate the capital requirement called the Health Risk Equalisation System (HRES), in order to reflect the decrease of risks faced by undertakings due to the way their incoming cashflows, and in particular the CvZ income, is determined. HRES is used as an alternative parameter to calculate the (lower) capital requirement using the standard model. The usage of a HRES parameter is subject to strict conditions, making the situation for the Dutch health market rather unique.
The central question of this thesis is: what can be learned from the (quantitative results of) ORSA? In order to answer this question one of the first aspects is to find the different (quantitative) components of the ORSA and investigate its goals as part of the overall Solvency II framework. After this an ORSA model is set up, which is used to assess the financial position (under Solvency II) of a (fictional) Dutch health insurance undertaking on a four year time horizon. This model provides the quantitative results of the ORSA for analysis, providing input to answer the central research question. Using the knowledge of the ORSA model and its results some improvements of both the ORSA model and possible input variables, all input is present to answer the central question. It is observed that the position of the management when it comes to ORSA perfor-mance is essential, as it is important that they are involved throughout all stages of the ORSA, ranging from (facilitating) input parameters to defining management actions according to the results.
In chapter 2 the available literature on the subject of ORSA is studied, with a focus on the quantitative aspects of the ORSA and not the governance nor reporting aspects as seen in Pillar 3, to find goals and requirements of the ORSA. Chapter 3 then continues by constructing a model for ORSA performance, including future cashflows, balance sheets and capital requirements to assess the financial position of a Dutch health insurance undertaking. Also some of the ORSA requirements will be met and several ORSA scenarios will be defined. In the fourth chapter the results of the scenarios using the model of chapter 3 will be analysed, whereas chapter 5 discusses possible improvements of the ORSA model and challenges some of the underlying assumptions. The conclusion of this thesis is written in the final chapter.
What is ORSA?
The ORSA is mentioned in the Solvency II Framework Directive, Article 45 as a part of the risk management system (see Appendix A for the full text of Article 45), which is the basis of the future Solvency II regulation. The Directive only states the require-ments for the ORSA which are open to a wide spectrum of possible interpretations, while EIOPA has released guidelines for the implementation of the ORSA. These pro-vide more details on the requirements of the ORSA and how the requirements stated in de Directive are to be interpreted. On November 7th 2011 EIOPA released a public consultation paper (EIOPA-CP-11/008) ‘On the Proposal for Guidelines on Own Risk and Solvency Assessment’, asking for comments on the proposed guidelines, and recom-mendations for the ORSA. After the consultation closed on January 20th 2012, EIOPA released a Final Report on the public consultation on July 9th 2012. Although it is called a final report, the presented ORSA guidelines must be considered to have a draft status, in order to reflect possible future developments. In the Netherlands the Dutch Association of Insurers (“Verbond van Verzekeraars”, from now on DAI) presented a second version of their ‘Vision on ORSA - Good Practice’ document in February 2012, focussing on examples and possible ways to implement ORSA, not being a guideline nor a template. Latest developments are the release of yet another consultation paper from EIOPA (EIOPA-CP-13/009) on the ‘Proposal for Guidelines on Forward Looking assessment of the undertaking’s own risks (based on the ORSA principles)’. With the ‘Forward Looking assessment’ EIOPA essentially narrows the focus of these Guidelines to the quantitative aspects of the ORSA, providing guidance to assess the future finan-cial position of insurance undertakings. This consultation closed on June 19th 2013 and resulted in a final report on September 27th2013 together with the EIOPA final Guide-lines for the preparation of Solvency II. A few days later on October 2nd the European Commission released a draft version of an amended Directive in which the introduction date of the Solvency II regulation is set to 1-1-2016 as a definitive date. The EIOPA final Guidelines are meant to increase preparedness of the undertakings as well as the national supervisors once the new Solvency II framework is applicable.
2.1
The goals of the ORSA
The ORSA has several goals, but let’s start with one of the goals of Solvency II in general, before moving towards the goals of the ORSA itself to provide a better context. A goal of Solvency II is to provide a uniform EU regime that focusses adequately on risk management and to provide incentives for (re)insurers to measure and properly manage their risks. This in turn will create a level playing field for (re)insurance un-dertakings, whereas in the current regime the supervisory requirements with regard to risk management vary widely across EU Member States. Under the current regime of Solvency I, the supervision is mainly volume-based rather than the risk-based approach intended by Solvency II. The ORSA can play an important role in the improvement of
an undertaking’s risk management: measuring risks and manage these risks.
The ORSA aims to enhance the company’s Board responsibility not to take more risks than their capital base allows. Following this thought, the ORSA is positioned as a management tool and not a supervisory tool as means to calculate the regulatory capital requirement as done in Pillar 1 of Solvency II. As a result, undertakings should decide for themselves how to perform the ORSA appropriately, given the nature, scale and complexity of their risks. The ‘Own’ in Own Risk & Solvency Assessment is a very important aspect, as it forces undertakings to develop their own ORSA process regarding their organisational structure and risk management system. Note that for the calculation of the future required capital for ORSA purposes, the standard Solvency Capital Requirement (SCR) model of Pillar 1, which calculates a capital requirement for a 1-year horizon, can be a good starting position as it covers the most general risks faced by insurance undertakings. The undertaking itself must however determine whether such a standard SCR model is appropriate. This will also be discussed in the next section.
In the EIOPA guidelines, it is stated that it is important to focus on what is to be achieved through ORSA and not how to achieve these goals.
With the ORSA being a management tool, several goals of the ORSA process are management related. One of them is that the Administrative, Management or Super-visory Body (AMSB) must be aware (or more aware) of all material risks faced by the undertaking, whether these risks are included in the SCR calculation or not, or whether they are easily quantifiable. For risks included in the SCR calculation this is probably the easiest, as there were several Quantitative Impact Studies (QIS) with a focus on the SCR calculation. For the Dutch insurance industry, there were also 2 additional impact studies: RiSK (2010) the year after QIS4 and the Parallel Run (2012) the year after QIS5 using specifications provided by the De Nederlandsche Bank (DNB). These QIS exercises have been a good opportunity for the AMSB to learn about several risks modelled in the SCR calculation. As for looking beyond the SCR calculation, which is made for an ‘average’ (re)insurance undertaking, there is also the possibility of a (partial) internal model which is a tailor fit for the undertaking and hence can include risks not included in the SCR calculation. As many undertakings have decided not to use an internal model, this thesis will focus on usage of the standard SCR calculation. A downside for the ORSA is that these undertakings probably did not put much effort in possible risks faced not included in the standard SCR model. A possible example is reputational risk, which can cause significant damage in day-to-day business and the number of insurance contracts. Closer to the standard SCR model is the interpreta-tion of operainterpreta-tional risk, which is captured by the standard model using premium and provision numbers, but not actual operational events to calculate the possible impact.
While ORSA does not require undertakings to implement a (partial) internal model, they do have to perform an Assessment of the Overall Solvency Needs. The assessment is described in Guideline 8 of the ‘EIOPA Final Report’ of July 9th 2012. Note that this assessment does not necessarily call for a complex approach, although a first look at the requirements may let you think it does. Methods used for the assessment may range from (simple) stress tests to a, more or less, sophisticated economic capital model, as long as the undertaking demonstrates the underlying rationale. When an economic capital model is used, it does not need to meet the reporting requirements of Pillar 3 regarding an internal model for the SCR calculation of Pillar 1. Using an internal model for Pillar 1 is a complex process where the national supervisor needs to approve the model before it can be used for the calculation of the capital requirement. Another difference to the SCR calculation is that the Assessment of the Overall Solvency Needs considers all material risks, including long term risks it could face within the timeframe determined by the business planning period. Also the SCR only considers quantifiable risks, whereas the assessment should also identify and assess which non-quantifiable
risks are part of the undertakings risk profile and ensure they are properly managed. An undertaking can also decide not to use a capital buffer for all of its quantifiable risks, but manage and mitigate some of those risks instead, which can introduce a new risk for the ORSA: is the mitigation effective?
The Assessment of the Overall Solvency Needs should be in both quantitative as qualitative terms, where the quantification needs to be complemented by a qualitative description of the risks. It should cover all possible risks faced by the undertaking, but at least a) risks arising from all assets and liabilities, b) potential external stress, c) the quality of processes and inputs, in particular the adequacy of the undertaking’s system of governance where risks that may arise form inadequacies of deficiencies are considered, and d) the undertaking’s management practices, including the use of risk mitigation techniques.
After the undertaking has identified all the risks it is exposed to, a decision has to be made whether the risks will be covered by capital, managed with risks mitigation tools or both. If the risks are covered by capital, the undertaking needs to estimate the risks and identify the level of materiality. If the risk turns out to be material, a capital requirement has to be determined and/or there needs to be an explanation of how the risk will be managed. For risks managed with risk mitigation techniques, the undertaking has to explain which risks are going to be managed by which technique and explain the underlying reason.
Another goal of the ORSA is that the AMSB must take an active role in the ORSA process: for example challenging scenarios and making sure that the results of the ORSA and the insights gained in the process are taken into account when another ORSA is started. To perform this active role, the AMSB needs to understand the risks faced by the undertaking and how they translate into capital needs or risk mitigation. The active role for the AMSB also stresses the position of the ORSA as a management tool and provides the necessary awareness regarding all material risks faced by the undertaking.
2.2
The ORSA process
Now the goals of an ORSA are known, let us look at the process of running an ORSA. The Guidelines issued by EIOPA do not explicitly prescribe the steps an undertaking needs to take in order to perform an ORSA, which is consistent with their statement that it is an Own Risk & Solvency Assessment and thus undertakings need to decide for themselves how to perform the ORSA. The DAI however did put an effort in describing examples and possible ways to implement ORSA. Before we take a look at the process, they mention some essential elements the undertaking should keep in mind. The most important one is that the ORSA process, and hence all its components, should be proportional to the size and complexity of the undertaking, which is also pointed out in the EIOPA Guidelines. For a good ORSA performance, undertakings should also have the following elements in place, quoted from ‘Vision on ORSA - Good Practice’ of the DAI:
1. A Governance Framework consisting of a sound Three Lines-of-Defence model and risk committees in which those Lines-of-Defence are represented.
2. A risk culture in which senior management sets the appropriate tone-at-the-top. 3. A well-thought out risk appetite framework: this sets the boundaries for
(accept-able) risk taking in regular business operations.
4. An integrated risk management framework constitutes the context in which ORSA is performed. That framework must contain risk identification processes. At a min-imum, processes must be in place for determining business scenarios and specific
risks. Scenario analysis and planning can be a powerful tool for helping a man-agement board in assessing the resilience of the (re)insurer and their objectives to internal and external changes.
5. The qualitative results of risk identification and scenario planning must be trans-lated into quantitative results: measuring (likelihood/impact) and calculation mod-els (accumulation/diversification) will result in an undertakings risk/capital pro-file.
Some of these items relate directly to some of the goals mentioned earlier: the re-sponsibility of the AMSB links to (2) and the Assessment of the Overall Solvency Needs is related to (4). The Three Lines-of-Defence model mentioned by the DAI provides the following general structure, quoted from ‘Vision on ORSA - Good Practice’ of the DAI: • The First Line is the Business Unit, with accountability and responsibility for performance, operations and daily risk mangement, management control and first line monitoring activities;
• The Second Line consists of central Staff Units who are responsible for encouraging and challenging sound risk management throughout the organization, providing guidelines, methods and techniques, and supporting the first line in making proper risk-return trade-offs;
• Internal Audit are the Third Line and responsible for providing additional assur-ance by independently monitoring the effectiveness of control measures as well as monitoring the effectiveness of financial, operational, compliance and risk man-agement.
In this general model the First Line of Defence is responsible for performing the ORSA, where the Second Line facilitates the ORSA process and may also draft the ORSA report. The Third Line then performs a review of the ORSA process. Most insurance undertakings will have such a Three Lines-of-Defence model in place, but it may deviate from the general model described above. This deviation probably will be between the interpretation of the First and Second Line of defence, for example some staff units can be positioned in the First Line. Also note that the Third Line might be outsourced depending on the size of the undertaking.
When looking at the ORSA process, the following elements should be taken into account according to the quoted list below from the ‘Vision on ORSA - Good Practice’ of the DAI, noting that the quantitative ORSA process itself is in essence forward looking, as ORSA assesses the position of the undertaking using a horizon of multiple years:
1. A preparation stage is needed as starting point for an ORSA process.
2. A Base Case which is the outcome of the Medium Term Planning (MTP) process covering a period of 3-5 years. Business scenarios should be defined and directed at major potential threats to the MTP objectives. Stress testing is an essential part of the ORSA process and helps to understand the impact of scenarios and more specifically which circumstances can lead to a violation of internal or external capital ratios.
3. Comparison of the insurer risk profile to the assumptions underlying the SCR/ internal model in order to assess whether these assumptions are (still) appropriate for the organization-specific risk profile. In addition, a reconciliation of the risk profile to available capital, and an analysis of the quality of capital is needed.
4. The ORSA process is a trigger for management actions. Risks beyond a (re)insurers appetite should be brought to within acceptable levels. Additional capital will not reduce the risk, it will only provide a financial buffer for the period until the risk is brought to within acceptable levels, preferably through the use of proactive man-agement actions. Such actions can include transferring the risk (reinsurance, co-insurance), terminating the risk generating activity (divestments) or treating (mit-igating) the risk (implementing controls to decrease the likelihood and/or impact of the risk).
5. The ORSA process includes an analysis of all relevant information and performed calculations, as well as important management considerations. The outcomes of these analyses and any considerations must be documented in an ORSA Internal Report. Level 3 guidance suggests that an ORSA Internal Report with an appro-priate level of detail may be equivalent to the regulatory ORSA report required by Pillar 3. If so, it will be filed with the supervisor. In writing the report, it is im-portant to realise that ORSA is basically part of the strategic management process, not a mathematical exercise.
While the focus of this thesis is not the reporting requirements of Pillar 3 or the ORSA Internal Report, the last item contains an essential message: ORSA is part of the strategic management process and not a mathematical exercise. It is also pointed out that ORSA is a regular process, which should at least be performed annually, but significant changes in the undertakings risk profile may also result in an ad-hoc ORSA process out of the regular cycle. Defining external and internal triggers for such an ad-hoc ORSA is recommended, in order to perform the ORSA without any delay following any significant change in the risk profile as required by the Directive. This requirement can have a huge impact on the day-to-day workload of the undertaking, as the ad-hoc ORSA will become a first priority for a couple of departments of the undertaking. To be able to point out the ‘right’ priorities, some additional guidance from EIOPA or national supervisors is needed to indicate an appropriate time frame in which the ad-hoc ORSA needs to be performed. Also the quality level of (parts of) the ad-hoc ORSA is likely to be dependent of the nature of the significant change in the risk profile. The management actions as described in (4) can also be input for incentives, which will be discussed in chapter 5.
• Strategic management process • Medium Term Planning process • Risk Appetite
• Capital plan
• Enterprise or Integrated Risk Management • Scenario planning
• Stress tests • Risk assessments • Reporting procedures • Control Statements
• Communication & training Figure 2.1: Building blocks. Although the ORSA looks
like a lot of extra workload for (re)insurance undertakings, most undertakings already have existing elements or “building blocks” (see Figure 2.1 for the building blocks mentioned in ‘Vision on ORSA - Good Prac-tice’ of the DAI) in some form. An ORSA process can in a way be seen as a way to integrate these elements as stated below, which should lower the work-load according to the DAI. In theory this is true of course, but in practice many of those building blocks may not have a formal status or are not as exhaustive as required by the ORSA and Solvency II in gen-eral, increasing the initial
work-load significantly. This will in turn support undertakings to increase the performance of their risk management, achieving a general goal of Solvency II.
2.2.1 The preparation stage of the ORSA
The first step of the ORSA process as described by the DAI is the preparation of the ORSA. As the ORSA brings together business strategy, risk management and capital management, some questions should be answered, which is easier if the ORSA has already been performed before. The undertaking should assess whether there have been material changes in the business strategy and targets. Has their been a change in the risk appetite? Which material risks are identified, thus has the risk profile changed? Is the model for the calculation of the capital requirement (Standard or Internal Model) still fit for the risk profile? Have there been interal or external developments? What is the capital position?
All these questions have to be answered and on the other hand all the relevant data has to be collected like portfolio positions, cash flow projections, yield curves etc. Once done, it can form the basis of the risk assessment in terms of likelihood/probability and input for the calculations.
2.2.2 Medium Term Planning
The MTP process delivers the second step of the ORSA process, which mainly consists of the Base Case of the ORSA. In the MTP process it is likely that there is output besides the base case, which can range from slight deviations from the base case to significant other scenarios using other assumptions and/or policy during the business horizon. All these can be input for ORSA scenarios besides the base case, where the slight deviations of the base case may not achieve the actual status of an ORSA scenario due to the relative small impact. Some of MTP output might also share characteristics with other ORSA scenarios or stress tests, which are more likely to have a larger impact.
2.2.3 Required Capital
Part of the ORSA is the Assessment of the Overall Solvency Needs as described in the previous section. This assessment might result in risks that are not captured in the standard SCR model while the undertaking does not have a (partial) internal model for the calculation of the capital requirement of Pillar 1. Part of the ORSA is the assessment of the financial position of the undertaking, which is mainly determined by the capital model used for Pillar 1 and assumed to be the standard SCR model for most undertakings. For these undertakings it is also important that they compare their risk profile to the general risk profile underlying in the assumptions of the standard model to see whether the standard model is fit to the organisation. This can be done as part of the Assessment of the Overall Solvency Needs which can also result in the conclusion that the undertaking is subject to other significant risks not captured in the standard model. For these risks a model can be used which does not have to meet the strict requirements an internal model, but should be proportional to the size, nature and complexity of the risks as stated before.
ORSA modelling
This chapter is about performing an ORSA, taking a closer look at the assumptions and techniques used to assess the future position of the undertaking, the so-called forward looking assessment of ORSA. To do this, forward looking is dealt with by looking at balance sheet projections and calculating future capital requirements. In this process several assumptions need to be made and supported, as well as modelling several input parameters. A Dutch health insurance undertaking is taken as an example for the actual performance of the ORSA, hence only relevant aspects with respect to this type of insurance will be covered. The rationale behind the modelling can be used in a wider context, but some additional modelling might be required.
3.1
Forward Looking
As the ORSA is about forward looking, some assumptions are needed to do a quan-tification of the ORSA base case and the scenarios. Note that the ORSA base case as well as the scenarios are deterministic, making the ORSA a sensitivity analysis using predefined input parameters. The possibility of a stochastic approach will be discussed in Chapter 5. The forward looking asssessment has two quantitative components: future balance sheets and future capital requirements. The projection of future balance sheets will be dealt with first.
3.1.1 Balance sheet projections
To construct future balance sheets, a starting point is needed. The fictional balance sheet below is a simplification of a balance sheet of a Dutch health insurance undertak-ing with only one line of business: the mandatory basic health insurance (“basisverzek-ering”). While the balance sheet is a simplification, it contains the most important elements for ORSA purposes. The assets of the balance sheet are straightforward con-taining stocks, government bonds, corporate bonds, deposits, accounts receivable and cash. The deposits all have short durations with a maximum of 0.5 year so that these can be classified the same as cash under Solvency II. The accounts receivable consist of receivables from intermediaries and advances issued to health care institutions to cover their expenses prior to declaration. The liabilities of the balance sheet consist of Techni-cal Provisions, accounts payable and the Own funds. The TechniTechni-cal Provisions in turn have three components: Claim Provisions, Premium Provisions and a Risk Margin. The Claim Provisions contain the Incurred But Not Reported (IBNR) claims of the reporting year and the years before, while the Premium Provisions contain the present value of all future cashflows related to existing insurance contracts in the coming years. The Risk Margin serves as an add-on to the Claim and Premium Provisions to approach actual market value of the Technical Provisions as a whole and is a result of the calculation of the capital requirement. The accounts payable are related to insurance obligations
while the Own funds are the result of the asset total minus the Technical Provisions and accounts payable.
Balance sheet t
Stocks 30,000 Technical Provisions
Gov bonds 210,000 Claim Provisions 266,400 Corp bonds 60,000 Premium Provisions 42,172 Deposits 115,000 Risk Margin 8,143 Accounts receivable 120,000 Accounts payable 60,000 Cash 30,000 Own funds 188,285 565,000 565,000
For the future balance sheets assumptions have to be made for the different components: how do they develop in the future, what is the impact of the technical results of the insurance undertaking? As for future development, there can be 2 factors: how does the component develop by itself, for example stock return, and how should the component develop given the link with insurance contracts. An example of the latter is the cash needed to pay insurance claims.
For the asset total, the following formula holds, where the indices indicate the end of year position or the total amount of cashflows during the year indicated:
Asset T otalt+1 = Asset T otalt+ T echnical resultt+1+ Investment resultt+1
where the Investment resultt+1 is given by
Investment resultt+1=
X
i
asseti(t) ∗ returni(t)
for asset classes i = stocks, gov bonds, corp bonds, deposits, cash.
The technical result is more complex, as the cashflows during the year must be con-sidered: the technical result at the end of a year is equal to the cashflows during that year. In the equation below all indices are therefore the same. Note that the technical result used here differs from the technical result used in accountancy, as there for ex-ample also changes in the Technical Provisions are taken into account and not just the cashflows as done here.
T echnical resultt+1= W ritten P remiumt+1+ CvZ incomet+1
− Claims P aidt+1− Expensest+1 Where the Written Premium is given by
W ritten P remiumt+1= P remiumt+1−U nearned P remiumt+U nearned P remiumt+1
and the Claims Paid is given by
Claims P aidt= Claimst−2,t+ Claimst−1,t+ Claimst,t.
So basically the technical result is given by the premium, with some timing cor-rections, the CvZ income, the expected profit in future premiums, the claims paid and the expenses. The timing corrections of the premiums, the unearned premium of the reporting year and the year before, work in the following way: at the end of a year a certain amount of the premium of the next year has already been paid. In accounting terms the written, but not yet earned premium. So in year t + 1, the written premium is the premium of year t + 1, but of this premium a part has al-ready been paid in year t: the U nearned P remiumt.. On the other hand, also a
written premium of year t + 1, hence the incoming premium cashflow, is equal to P remiumt+1− U nearned P remiumt+ U nearned P remiumt+1 as stated above.
As for the claims: the notation Claimst,t+1 indicates the claim amount from
insur-ance year t, but paid in year t + 1. Note that the total claims of insurinsur-ance year t are given by:
Claims(t) = Claimst,t+ Claimst,t+1+ Claimst,t+2
where
Claimst,t= a ∗ Claims(t)
Claimst,t+1 = b ∗ Claims(t)
Claimst,t+2 = c ∗ Claims(t)
with a + b + c = 1.
Knowing the T echnical resultt+1and Investment resultt+1, the assets at the end of
year t+1 can be projected. Note that the accounts receivable and payable do not generate any return. The Deposits, Accounts receivable and Cash are related to the insurance contracts and can be assumed to be proportional to the Claims, which are related to the medical inflation: Claims(t + 1) = Claims(t) ∗ (1 + ht+1) where ht+1 denotes the
medical inflation of year t + 1 and consists of the inflation and an increase of the volume of medical treatments: ht= it+ volumet. Note that the volume of medical treatments
does not need to be only a volume component, but can also reflect price movements of medical costs that deviate from inflation.
So the following relation holds, also for Deposits and Accounts receivable: Casht+1=
Casht∗ (1 + ht+1). For the assets, the only balance sheet items left unknown are Stocks,
Gov bonds and Corp bonds. For the projection of these items it is assumed that the asset mix is kept constant: 10% − 70% − 20% respectively for Stocks-Gov bonds-Corp bonds. So for example the value of Stocks in year t + 1 is equal to:
Stockst+1= 10%∗(Asset T otalt+1− Depositst+1
− Accounts receivablet+1− Casht+1)
When looking at the Liabilities, the Accounts payable are also related to the insurance contracts, hence the same relation holds as for Accounts receivable: Accounts payablet+1=
Accounts payablet∗ (1 + ht+1).
For the Technical Provisions some more work needs to be done, beginning with the Claim provisions. The Claim provisions in year t + 1 are equal to the claims that still need to be paid in year t + 2 of insurance years t and t + 1: Claim P rovisionst+1 =
Claimst,t+2+ Claimst+1,t+2+ Claimst+1,t+3. The Premium Provisions of year t + 1 are
related to the insurance year t + 2, but no years after that due to the condition that the Premium Provisions are equal to the present value of future cashflows related to existing contracts. For the Dutch health insurance the contracts can be lapsed by the policyholder at the end of every year. So at the end of a year the future policyholders are known for 1 year only:
P remium P rovisionst+1= P remiumt+2− U nearned P remiumt+1
+ CvZ incomet+2− Claims(t + 2) − Expensest+2
Note that Claims(t + 2) is used in this equation as it describes all claims related to that particular insurance year and not only the cashflows of the insurance year described by Claimst+2,t+2. By definition of the Expected Profit In Future Premiums, the expression
for the Premium Provisions can be simplified:
The final part of the Technical Provisions, the Risk Margin, is part of the SCR cal-culation. Therefore the Risk Margin and the Own funds are be determined after the calculation of the capital requirement.
List of assumptions/data needed so far:
1. Portfolio return: Stocks, Gov bonds, Corp bonds, Cash, Deposits for s = t + 1, . . . , t + 4
2. Inflation is for s = t + 1, . . . , t + 4
3. Volume medical treatments V olumes for s = t + 1, . . . , t + 4
4. Claims(s + 1) = Claims(s) ∗ (1 + hs) with hs= is+ V olumes
5. CvZ incomes+1= CvZ incomes∗ (1 + hs)
6. Expensess+1= Expensess∗ (1 + is)
7. P remiums= (Claim(s) + Expensess− CvZ incomes)/α
8. U nearned P remiums= 0.1 ∗ P remiums
9. Claimss,s = a ∗ Claims(s) with a = 0.64 constant
10. Claimss,s+1= b ∗ Claims(s) with b = 0.34 constant
11. Claimss,s+2= c ∗ Claims(s) with c = 0.02 constant
12. α = 0.995 constant
13. Assumption t+5: equal to t+4 where needed, in for example P remium P rovisionst+4
3.1.2 Capital requirements
Now the balance sheets have been projected, except from the Risk Margin and therefore the Own funds, the capital requirements can be calculated. Note that this calculation requires more information than just balance sheets, so the list of required data and assumptions is getting larger. The capital requirements calculated are a simplified ver-sion of a complete SCR calculation. This is allowed according to the Good Practice document of the DAI and also necessary as not all data is available in our projections. The simplified SCR calculated using the projected balance sheets is based on the Parallel Run specifications issued by the DNB. These are based on the QIS5 specifica-tions, with some adjustments. The SCR for the health undertaking is determined using 3 main risks, Market (SCRM arket), Health (SCRHealth) and Default (SCRDef ault),
re-sulting in the Basic SCR (BSCR). Then Operational Risk (SCROp) is added resulting
in the SCR. For the capital requirement of the BSCR, the 3 main risks are added using a correlation matrix, resulting in a lower capital requirement than just simply adding all capital requirements. The difference between using the correlation matrix and the simple sum over the individual risks is indicated as ‘Diversification’. This diversification also occurs within the 3 main risks, but not when adding Operational Risk as mentioned before. The specifications of the SCR used are based on the Parallel Run (PR), issued by the DNB as a follow up of the fifth Quantitative Impact Study (QIS5). Since then there have been some updates issued by EIOPA, but for the Dutch health insurance industry these have marginal effects. As the method of calculating the capital requirement for the ORSA is not prescribed, the choice was made to stick with known specifications.
Market Risk
Market risk consists of several risks for which a capital requirement is calculated: interest (SCRInterest), equity (SCREquity), spread (SCRSpread), concentration (SCRConc) and
countercyclical premium (SCRCCP).
For interest rate risk there are two possibilities: the interest goes up or down. The case in which the Net Asset Value (NAV) decreases the most is taken. The NAV is determined using the balance sheet items that are sensitive to interest shocks: Gov bonds, Corp bonds, Claim provisions and Premium Provisions. The deposits are not taken into account due to the short duration and therefore the possibility to consider these deposits as cash. To dertermine the NAV, the duration of the assets and of the provisions is used, multiplied by the size of the shock which is duration dependent. So the value after shock is given by
V alue ∗ (1 + shock(duration)) ∗ duration
where the shock can be either shockup or shockdown, as given in the PR specifications.
Equity risk is given by a shock factor applied to the value of equity on the balance sheet. The shock is given by 34%, as the equity portfolio is assumed to consist of only of the ‘Global equity’ category.
Spread risk is determined using duration, rating and of course the market value of the relevant assets. Only corporate bonds are considered, as government bonds are excluded from this calculation. As for the rating the corporate bonds are assumed to have rating A for half the portfolio and rating BBB for the other half. The spread risk is then given by
SCRSpread=
X
rating
V alue(rating) ∗ duration ∗ shock(rating).
For concentration risk it is assumed that one third of the deposits and half of the corporate bonds are situated at the same counterparty with an A rating. The rating determines the ‘counterparty threshold’ (CT) and the risk factor (rf). The capital re-quirement for concentration risk is then given by:
SCRConc = (V alue − Assets ∗ CT (rating)) ∗ rf (rating)
where the Assets is the Asset Total minus the assets that are taking into account for the counterparty default risk, which is described later.
The last Market Risk is the countercyclical premium (CCP), which is a substitute for the illiquidity premium in QIS5. The capital requirement for this risk is calculated in a simplified way as
SCRCCP = P rovisions ∗ CCP ∗ duration,
where CCP is the percentage used to increase the term structure used to calculate the present value of the provisions.
Using the market risk correlation matrix, corresponding to which interest rate shock had the largest effect on the NAV, ‘up’ or ‘down’, the total Market Risk can be calcu-lated.
Health Risk
The next main risk of the SCR is Health risk, which consists of Premium & Reserve risk (SCRP rRes) and Catastrophe risk (SCRCAT).
The Premium & Reserve risk is calculated in the following way: SCRP rRes =
3 ∗qσ2
using σ(prem,s)= σ(res,s) = 3.6% for all s because of the HRES parameter. The premium volume V(prem,s)is equal to the maximum of the earned premium of the reporting year s and the expected premium of the existing contracts for s + 1. The reserve volume V(res,s)
is equal to the Claim provisions at the end of year s. Note that using this formula implies a correlation between the premium risk and the reserve risk of 50%.
CAT risk for health insurance is based on 3 (independent) scenarios, for which 2 apply for Dutch health insurance (Medical Expenses). These scenarios are Mass accident and Pandemic Medical Expenses. The capital requirement of the Mass accident scenario is determined by multiplying the ratio of people involved in the mass accident, the fraction of those people who receive, in our case, medical treatment, the value of this medical treatment and the number of people insured. The ratio of people involved in the mass accident is 0.15% for the Netherlands and the fraction of people who receive medical care is 30%. The costs of the medical treatment is set to 5, 000 for the current year and is assumed to increase according to inflation.
The capital requirement for Pandemic Medical Expenses is the multiplication of a fixed ratio 0.4, the number of people insured and the average value of the corresponding medical treatment received, which is set to 100 for the Dutch basic health insurance. On top of that the calculated capital requirement can be decreased by 80% due to the nature of the Dutch healthcare system.
The total capital requirement for CAT risk is then calculated as SCRCAT =
p
M ass accident2+ P andemic M edical Expenses2
The total capital requirement for Health risk is then calculated using the correlation prescribed by the specifications (25%) between Premium & Reserve risk and CAT risk.
Default Risk
The (counterparty) Default risk consists of two types of default: for type 1 the counter-party is likely to be rated and consists of the exposure related to the cash position in our case. The type 2 exposures are not likely to be rated and in our case consist of re-ceivables for intermediaries and cash outstanding (advances) at health care institutions like hospitals.
For type 1 exposures the information needed to calculate the capital requirement is the actual exposure per counterparty and the rating of the counterparty. The exact calculation is described in the QIS5 specifications, section SCR.6.2. As mentioned with concentration risk, all cash is situated at a single counterparty with an A rating.
For type 2 exposures the only input needed is the exposure and whether the receiv-ables are due for less or more than 3 months. In our case 75% of the accounts receivable are taken into account for the exposure with a due date of less than 3 months, for which 15% of the exposure is taken as the capital requirement. There is no exposure with a due date of more than 3 months. The remaining 25% of the accounts receivable does not apply for default risk as a result of contract management with health care institu-tions allowing to strike a balance between the advances and the health care costs for treatments that have been performed, but not yet claimed.
Now all main risks of the BSCR can be calculated, the BSCR itself can be calculed using a correlation matrix prescribed by the specifications.
Operational Risk
The Operational risk is calculated using the earned premium of the last year, the year before and the Claim en Premium Provisions. On top of that the operational risk is at
most equal to 30% of the BSCR:
SCROp = min{0.3 ∗ BSCR; max{0.03 ∗ P remiumt+
max{0; 0.03 ∗ (P remiumt− 1.1 ∗ P remiumt−1)};
0.03 ∗ max{0; Claim P rovisionst+ P remium P rovisionst}}}.
As the premium is higher than the provisions, the premium is the most important factor to determine the capital requirement for Operational risk, of course keeping in mind that the risk is capped by 30% of the BSCR.
The BSCR and the Operational risk together give the SCR. MCR
The SCR is not the only capital requirement needed: there is also the Minimum Capital Requirement (MCR). If an undertaking has enough capital to cover the MCR but not the SCR, a recovery plan is needed. If the Own funds are not enough to cover the MCR however, the undertaking is practically bankrupt and the national supervisor takes control. The MCR is capped between 25% and 45% of the SCR and with an absolute minimum of 2,200 thousand euros, but with the following underlying calculation:
max{α ∗ (Claim P rovisions + P remium P rovisions); β ∗ W ritten P remium} so
M CRt= max{2, 200; min{max{0.25 ∗ SCRt;
max{α ∗ (Claim P rovisionst+ P remium P rovisionst); β ∗ W ritten P remiumt}};
0.45 ∗ SCRt}}
with α = 13% and β = 5%. Risk Margin
For the Risk Margin of the Technical Provisions another version of the SCR calculation is needed: the SCRRU for a reference undertaking. Note that the introduction of a
reference undertaking was needed to break the circularity constructed: the Risk Margin is input for the SCR, but the Risk Margin itself is based on the SCR. To solve this the SCRRU was introduced without this dependency. For the reference undertaking less risk
categories are taking into account: unavoidable market risk, health risk and operational risk. The unavoidable market risk can be considered nil for short term non-life insurance like health insurance, hence SCRRU = SCRHealth+ SCROp. The Risk Margin is then
calculated using the best estimate provisions, the duration of these provisions and the Cost-of-Capital (CoC) rate, which is equal to 6%:
RM = CoC 1 + r1
∗ duration ∗ SCRRU where r1 is the 1-year interest rate.
With the calculation of the MCR and the Risk margin all aspects of the capital re-quirements are determined. Along the way some more assumptions were made regarding the calculation of the SCR, so a continuation of the list of assumptions is needed:
14. The specifications of the SCR calculation are fixed 15. Interest rates are constant
16. Duration of the technical provisions is constant 17. Duration of the bonds is constant
18. Bond ratings are constant
3.2
Starting position
Recall the balance sheet discussed earlier: Balance sheet
Stocks 30,000 Technical Provisions
Gov bonds 210,000 Claim Provisions 266,400 Corp bonds 60,000 Premium Provisions 42,172 Deposits 115,000 Risk Margin 8,143 Accounts receivable 120,000 Accounts payable 60,000 Cash 30,000 Own funds 188,285 565,000 565,000
The risk margin is a result of the SCR calculation, which can be done using the 19 assumptions listed in the previous section. The breakdown of the SCR with all rele-vant subrisks is given below:
Solvency Capital Requirement
Market Risk 19,024 Interest rate 3,471
Equity 10,200 Spread 3,510 Concentration 11,547 Counter cyclical premium 4,096 Diversification -13,800 Health Risk 67,566 Non-SLT Health 66,732 CAT 3,075 Diversification -2,241 Default 15,068 Type 1 2,012 Type 2 13,500 Diversification -444 Diversification -21,351 BSCR 80,307 Operational 24,092 SCR 104,399
Note that the SCR calculated above can be different from the SCR calculated in Pillar 1 for reporting purposes to the national supervisor (part of Pillar 3) due to the assumptions made. This of course also holds for the balance sheet and hence the SCR ratio. If there is a big difference between the SCR (ratio) calculated in Pillar 1 and for the purposes of the ORSA in Pillar 2 it would be wise to reconsider the assumptions made: maybe they are too limited for the characteristics of the undertaking.
3.3
Assessment of the Overall Solvency Needs
The Assessment of the Overall Solvency Needs should result in two different aspects: is the standard model fit for the undertaking and does the undertaking face any signifi-cant risks not captured in the standard model? For the health insurance undertaking discussed here the answer to the first question is ‘yes’ due to the following reasons listed per risk category:
Market risk
Market risk reflects the risks posed by these assets. Also note that the asset classes only contain ‘simple’ assets and no complex assets for which the actual risks faced is more difficult to determine.
Health risk
Health risk consists of two components: Premium & Reserve risk and CAT risk. For the Premium & Reserve risk the capital requirement is determined by the HRES parameter, which is calibrated by the DNB for the Dutch health insurance. Hence there is no reason to believe the insurance portfolio of the undertaking has a significant deviation from this parameter. Catastrophe risk is based on two scenarios, each calibrated to different countries including the Netherlands, which represent the most important risks in a catastrophic scenario. Hence capital re-quirement for Health risk is an appropriate approximation for the undertaking. Operational risk
As stated before, the Operational risk is in practice depends on the premium volume, capped by 30% of the BSCR, which seems like an arbitrary number. A model based on actual operational events would be preferred, but since such a model is not available (to the undertaking) yet, nothing can be said about whether the standard model is a good approximation or not. As the operational characteristics of the undertaking do not significantly differ from other insurance undertakings the model used seems appropriate by default.
Despite the statement that the standard model is fit for the undertaking, the remarks made when it comes to Operational risk could trigger management actions (see also Chapter 5) to construct an own capital model for this risk when it comes to ORSA.
With respect to the second question whether the undertaking faces any significant risks besides the ones used in the standard model, the answer is a bit more difficult. As the basic health insurance benefits are determined on a yearly basis by the Dutch government, they have a large influence on the size and number of the claims by in-cluding or exin-cluding certain compensations. These changes also have their effects on the premium of course, but the risks arising from premiums is partially handled by CvZ: they also calculate the effects of these changes on the premium and hence the CvZ income. This risk is faced by the entire line of business, making it unlikely that changes will be effectuated such that it will endanger the financial position of a large portion of the undertakings in the line of business. Due to the strong financial position this risk is considered to be relatively small, pushing the answer to the question to ‘no’ as there are no other significant risks not captured by the standard model.
3.4
ORSA base case & scenarios
It is time to look at the actual ORSA base case and the scenarios to be evaluated by the model described in section 3.1, which means the list of 19 assumptions needs some input. The ORSA scenarios will then be constructed as variations of the base case, using rather extreme but still realistic deviations. Keep in mind that all input parameters are deterministic and hence all scenarios as well. The base case is a reflection of the best estimate future developments.
3.4.1 Base case
Most of the assumption have already been stated, but for the other assumptions these values are used:
1. Portfolio returns
return for the whole horizon. Note that the return for deposits and cash is taken identical, as the deposits are classified as cash.
Equity 7% Government bonds 1% Corporate bonds 3% Deposits 0.1% Cash 0.1% 2. Inflation
The inflation is taken to be 2% for the entire horizon. 3. Volume medical treatments
Based on a negotiation agreement (“Onderhandelingsresultaten Schippers met ziekenhuizen, medisch specialisten, zelfstandige behandelcentra, GGZ en huisart-sen” [2013]) with a Dutch minister the total medical costs for the bulk of the market should be limited to 1.5% in t + 1 (corresponding to 2014) and 1% for t + 2, t + 3, t + 4. Hence the volume should be equal to V olumet+1= −0.5% and
V olumes= −1% for s = t + 2, t + 3, t + 4.
The other assumptions already have been discussed in the relevant balance sheet or SCR paragraphs. The most important aspect of the base case is the medical inflation, which is based on two components: inflation and a volume for medical treatments. The medical inflation determines the future claim amounts and as a result also the premiums, which are input for the most important risk of the SCR: Health risk. Due to the modelling assumptions the claim amounts also affect most of the balance sheet items and thus also Default risk, making the impact on the (B)SCR even larger and making medical inflation a very important element for the assessment of the future position. The assumptions regarding medical inflation (and thus Volume medical treatments) based on the negotiation agreement provide a best estimate for this parameter, as it is unlikely that the growth of medical expenses will be (significantly) less than the agreed maximum.
3.4.2 ORSA scenarios
The 5 ORSA scenarios defined below are variations of the base case, using different assumptions. As the ORSA is performed during year t + 1 in this case, the balance sheet and SCR calculation at the end of year t is fixed and also a lot is known about the current year t + 1. Because of this all ORSA scenarios will begin to deviate from the base case as of year t + 2, causing the balance sheet and SCR calculation at the end of year t + 1 to be equal to those in the base case for all scenarios.
Liability scenario
The Liability scenario describes an unexpected increase of the health care costs, as the undertaking is unable to limit the medical costs as agreed according to the base case assumption 3. The medical inflation turns out to be 3.3% higher than in the base case as of year t + 2, without the possibility to increase the premiums as other insurers are able to meet the reduction in medical volume. Douven et al. (2006) describe an increase of medical volume of 2.8% which is taken as a reference, combined with the target of −0.5% of the base case in t + 1 leading to a 3.3% increase. The medical inflation as of year t+2 will thus be 2%−1%+3.3% = 4.3%. Asset scenario
The Asset scenario deals with losses in the asset portfolio as well as a decrease in value of the Cash and Deposits position. These losses only occur in year t + 2:
Equity −40% Government bonds −5% Corporate bonds −5% Deposits −25% Cash −25% HRES scenario
The HRES scenario describes the event that the HRES parameter may not be used for the Dutch health insurance anymore. The HRES parameter can be used because of the mitigating effects of the Dutch health care system, but as these mitigating effects have been reduced and will probably be reduced even more, the level of the HRES, or even the usage of the HRES, becomes questionable.
Contracts Up scenario
In the Contracts Up scenario the undertaking has to deal with an (unexpected) increase in the number of insurance contracts by 70,000 which means an increase of almost 19% of the insurance contracts due to a competitive position in the insurance market. The increase is modelled as of t + 2 with the assumption that from this point the number of contracts stay constant. This means that the pre-miums, CvZ income and claims will increase by the same rate of almost 19%. For the expenses the same increase holds. For balance sheet items that are insurance related, Deposits, Accounts receivable and payable and Cash, it also means that they grow according to the increase in claims instead of just increasing by medical inflation.
Contracts Down scenario
The Contracts Down scenario is the opposite of the previous scenario: an un-expected loss of 70,000 insurance contracts due to a highly competitive market. The modelling is the exact opposite of the Contracts Up scenario, except from the modelling of the expenses: the expenses will stay at the same level, indicating that the staff cannot be reduced. This also means that the premiums are no longer based on the principle of an expected profit of 0.5%, as the premiums decrease by the same rate as the reduction in insurance contracts.
Results & analysis
In this chapter the quantitative results of the ORSA will be presented and analysed, consisting of the base case and 5 scenarios defined in the previous chapter. The scenario results, consisting of cashflow overviews, balance sheets and SCR breakdowns, are in-cluded in the appendix and will be presented in this chapter in a graphical form. Note again that the Risk Margin in the balance sheets is a result of the SCR calculation. The 5 ORSA scenarios will be discussed individually after the base case results.
4.1
Base case results
The base case results (and the results of the other ORSA scenarios) will be presented using three figures reflecting:
1. the cashflows of the technical result and the investment result during the ORSA horizon, see Figure 4.2;
2. the balance sheet development of both assets and liabilities, including the SCR as a direct comparison to the Own funds, see Figure 4.3;
3. the SCR using a breakdown of the BSCR, the diversification effect on BSCR level and Operational risk, see Figure 4.4.
The numerical results of the base case and the ORSA scenarios can be found in Appendix B-G and an overall view of the results, expressed in the SCR ratios of the scenarios, is displayed below in Figure 4.1. The base case (blue line) shows a more or less linear growth from an SCR ratio of 180% at year t to a ratio of 201% in year t + 4. The base case cashflows (Figure 4.2) show decreasing technical results, while the investment results are increasing. The latter is due to an increasing investment portfolio, causing a higher expected result. Recall that the technical result is the sum of the following components:
T echnical results= W ritten P remiums+ CvZ incomes
− Claims P aids− Expensess
providing an explanation to the decreasing technical results. Both the Written Pre-mium as the Claims Paid are not just dependent of the current insurance year, as respec-tively the Unearned Premium and the run-off scheme of the Claims are involved. The development of the claims is based on a 1.5% − 1% − 1% − 1% growth for t + 1 − . . . − t + 4 and as a result the Premiums shows a similar development. The Claims Paid consist of
Claims P aids= Claimss−2,s+ Claimss−1,s+ Claimss,s,
showing that the Claims Paid for year t+1 are based on the claims of years t−1, . . . , t+1 using the growth denoted above (for year t−1, hence Claimst−1,t+1, the assumption was
Figure 4.1: SCR ratios of all ORSA scenarios.
made that they equal Claimst,t+1). Hence the increasing claim amounts are partially
paid later than the corresponding premium income, for which the Written Premium is higher than the premium for the insurance year due to the Unearned Premium effects. These timing effects cause decreasing technical results, especially comparing t + 1 to t + 2 due to the increase of claims by 1.5% instead of 1% the years after.
Figure 4.2: Base case cashflows.
The increase of the investment portfolio can be seen in Figure 4.3, which is equal to 12.3% (to 336, 750 in t + 4) over the whole horizon, along with an increase of the Own funds by approximately 17.2% to 220, 625 in t + 4. The Deposits, Cash, Accounts receivable and payable increase according to medical inflation (4.6% over the whole horizon), which is also the main driver for the Claim and Premium Provisions that increase by 4.5% (to 278, 436) and 3.3% (to 43, 556) respectively. The asset total of 613, 876 at t + 4 is 8.7% higher than at the starting point due to the cashflow results. The figure also shows the SCR, which increases by 5.2% to 109, 800 resulting in an SCR ratio of 201% at the end of year t + 4.
More details of the SCR can be seen in Figure 4.4, showing the main risk categories of the BSCR: Market, Health and Default risk, along with the diversification and Oper-ational risk. The underlying risks and diversification effects are shown in Appendix B. The increase of the SCR breaks down in an increase of Market risk by 11.2%, Health
Figure 4.3: Base case balance sheet.
risk by 4.3% and Default risk by 4.6%. As Default risk is related to certain balance sheet items which are in turn depending on the medical inflation, it makes sense that this risk increases by the same percentage as the medical inflation. The increase of Market risk is caused by the growth of the investment portfolio, but increases less due to a link with the provisions and the assets that are related to claim amounts. Health risk mainly depends on the provisions as well as the premiums, both linked to medical inflation. As the premium, and hence also the Premium Provisions, is also related to expenses (increasing according to inflation), the effects are not identical to the increase in medical inflation. The higher level of Market risk partially disappears as a result of diversification, which increases by 7.5% over the whole horizon.
4.1.1 Base case conclusion
At the end of the ORSA horizon the SCR ratio for the base case is equal to 201%, an increase of approximately 5% per year starting from a 180% ratio. The SCR increases by approximately 5.2% over the whole horizon versus an increase of the Own funds by 17.2% while the asset total increases by 8.7%. The results shown are of course consequences of the input for the model and can be estimated without using the ORSA model, like Health risk which is related to the medical inflation put into the model (using the principle input = output). The added value of the ORSA model is the calculation of all these effects together, supported with calculations and not just a rough estimate done on a few pieces of paper. While a rough estimate might give the same (more or less) results, this is probably the effect of pluses and minuses during the estimate that cancel each other out.
The premium policy to maintain a 0.5% profit margin in combination with a similar absolute investment results provides the undertaking a strong and growing financial po-sition when looking at the SCR ratio. It is recommended to determine a desired level or bandwidth for the SCR ratio, such that whenever deviation from that level/bandwidth becomes to large management actions can be taken. This should also include manage-ment actions in the case that the SCR ratio is higher than determined, which can be done by lowering the profit margin for example to create a stronger competitive position.
4.2
Liability Scenario
With the liability scenario the intended reduction of the medical inflation cannot be made by the insurance company, while the competition is able to do so. Hence the claim size is larger than in the base case, while the premium and CvZ income stay at the same level. As the scenarios start in year t + 2, the effects are visible as of year t + 2. The complete overview of cashflows, balance sheet and SCR breakdown can be read in Appendix C.
Following the scenario description, the total amount of claims for year t + 2 rises from 758, 611 to 783, 645, divided over 3 years due to the run-off scheme. Hence the effects for this year only are visible for the whole remaining horizon, causing negative technical results in years t + 2 and t + 3 and an even lower technical result for year t + 4, see Figure 4.5. As the claims for years t + 3 and t + 4 also turn out to be higher, these effects are stacked, causing a technical result of −20, 239 in t + 4. Due to the negative results, the amount of assets is decreasing, causing a lower investment result (in t + 4: 5, 702 in the scenario versus 6, 702 in the base case) and an even lower (negative) total result in combination with the technical result.
Another factor that lowers the investment result are the balance sheet items that are related to the real medical inflation and hence the claim size: Deposits, Accounts receivable/payable and Cash. These turn out higher than in the base case, leaving less room for the stock and bond portfolio and thus causing a lower investment result. The higher medical inflation for the undertaking also causes higher claims and with the run-off scheme used, the Claim Provisions are also higher than in the base case: at t + 4 this results in Claim Provisions in the scenario of 287, 624 versus 278, 436 in the base case causing the Own funds to decrease even more besides the negative cashflow results, see Figure 4.6. Note that the Premium Provisions remain at the same level as the base case, as the next year’s expected results are still based on the intended reduction of the medical inflation.
Figure 4.6: Liability scenario balance sheet.
Another component of the Technical Provisions is the Risk Margin, which is SCR related and turns out slightly higher (8, 558 at t+4) than in the base case (8, 514 at t+4), implying a small increase in the required capital. The increase is related to Health and Default risk caused by the higher amount of claims: Health risk rises to 71, 278 versus 70, 495 in the base case while Default risk rises from 15, 757 to 16, 277 at t + 4, see Figure 4.7. The lower amount of assets on the other hand results in a dampening effect as Market risks drops from 21, 162 to 17, 179 at t+4, where the Interest rate risk reduced by more than 50% due to a lower mismatch between the higher claims and the lower value of the bond portfolio. The interesting result is that the SCR of the scenario turns out to be lower (108, 554 at t + 4) than in the base case (109, 800 at t + 4) while the higher Risk Margin indicated otherwise. This is the result of taking the calculation of the SCR for the reference undertaking for the Risk Margin, in which the Market risk can be neglected and is only dependent of (the increasing) Health risk and Operational risk.
The SCR ratio decreases to by approximately 17%-points per year to 131% at t + 4: 70%-points lower than the base case SCR ratio at that moment and almost 50%-points lower than our starting position. As the SCR ratios remain well above the 100% level, the financial position is able to endure the impact of the scenario, but the occurred losses require the attention of the management in order to constrain the medical costs to sustain a competitive position.
Figure 4.7: Liability scenario SCR.
4.3
Asset Scenario
The Asset scenario describes a financial crisis in year t + 2, where the insurance un-dertaking faces losses in both the investment portfolio as in the cash/deposits position. The details of the cashflows, balance sheets and SCR breakdowns starting from year t + 2 can be found in Appendix D.
Figure 4.8: Assets scenario cashflows.
The impact of the scenario is visible in the negative investment result in year t + 2 of −63, 047 resulting in a total result for that year of −57, 392, see Figure 4.8. This directly impacts the stock/bond portfolio and the Own funds that decrease by 61, 283 to 134, 619 in one year, see Figure 4.9. As the lower stock/bond portfolio also results in a lower Market risk (16, 402 versus 20, 084 in the base case) for the SCR, a slight dampening effect is visible, see Figure 4.10. The decrease of Market risk only results in a decrease of the SCR by 2, 310 due to diversification and a dominant position for Health risk in the structure of the SCR. As a result the SCR ratio drops by 57%-points to 128% caused by the investment losses.
In the following years the overall picture is very similar to the base case with an increase of the SCR ratio by approximately 5% per year, but with a lower investment
Figure 4.9: Assets balance sheet.
result due to a smaller stock/bond portfolio. At the end of the horizon the SCR ratio equals 138%, which is more than 60%-points lower than the base case. The financial position is able to handle this extreme shock if it is the only event with negative impact, if other (operational) losses occur the SCR ratio probably comes very close to 100% and thus a requirement for the undertaking to construct a recovery plan.
Figure 4.10: Assets scenario SCR.
4.4
HRES Scenario
The HRES scenario originates in the discussion around the SCR parameters for Pre-mium & Reserve risk, which are equal to 5% in the specifications, but for the Dutch basic health insurance a HRES parameter of 3.6% can be used. The level of this
