Economic Research

Allianz Group

Dresdner Bank

No.: 6, February 20, 2004

Authors: Dr. Jürgen Guhe, Group Risk Controlling

Dr. Helmut Kesting

____________________________________________________________

Deregulation in the mid-1990s made way for the single European market for insurance. A level playing

field demands the greatest possible harmonization of supervision. Equally, efficient supervision must

also guarantee the protection of policyholders. The European Commission therefore began by revising

the requirements dating from the 1970s on the solvency margin for life assurance and non-life insurance

undertakings (see Box 1).

Solvency and solvency margin � a brief explanation of the terminology

�Solvency� means the ability to pay. The �solvency margin� is the minimum amount of extra capital

prescribed by EU Directive that an insurance undertaking must have to fall back on in unforeseen

circumstances in order to cover losses.

The �Solvency I� package encompasses the entire present European system of prudential supervision

aimed at assessing an insurance undertaking�s overall financial strength. In addition to a company�s

financial resources, particular significance also attaches to its technical provisions and to cover for

these provisions by suitable assets.

�Solvency II� refers to the current project of wide-ranging renewal of prudential insurance regulation.

The concept of �overall solvency� aspires not only to endowment with financial resources adequate to

support the business written, it also integrates qualitative elements into the supervisory process.

These include the development of active internal risk management by companies on the one hand and

the promotion of market discipline with appropriate disclosure requirements on the other.

It was obvious from the outset that this revision of the regulations in force would not suffice in the long

term.

solvency margin aspect. The introduction of international accounting standards � as from 2005 in

the EU, all publicly listed companies must prepare consolidated accounts according to IAS �

makes it possible to harmonize other supervisory tools such as insurers� technical provisions.

volume (in the case of non-life businesses, for example, on the premium and claims index). At

best, this reflects the company�s risk only indirectly.

insurance undertaking lowers its premiums, all other circumstances being equal it can improve its

solvency margin at short term even as its overall financial position deteriorates.

and major claims (natural disasters, industrial accidents) on the one hand and volatile eq-uity

markets and historically low interest rates on the other are weighing heavily on their earnings

and posing new challenges to supervisors.

At the beginning of 2000 the competent Commission Services therefore launched the �Solvency II�

project with the aim of fundamentally overhauling insurance supervision. The similarity in terminology

to Basel II is by no means accidental. On the contrary, the intention is to examine whether, and to

what extent, the standards developed by the Basel Committee can also be applied to the insurance

industry.

Given the complexity of the task, the project was divided into two stages. The first was completed in

September 2003 with agreement on a fundamental framework of principles for the new regulatory

concept. According to this, supervision will in future assess an insurance undertaking�s overall solvency.

For this reason a three-pillar strategy is recommended along the lines of Basel II, but taking

suitable account of aspects specific to insurance. In particular, technical reserves are to be retained as

a supervisory tool. Overall, the following fundamental objectives are pursued:

In the second stage begun in December 2003 this basic concept must be fleshed out. The following

picks up on various key issues and problems that have to be answered and solved in the process.

Basel II is composed of three elements: first, risk-sensitive capital requirements (standardized approach /

internal models); second, the supervisory review process; and third, market discipline (transparency).

These three �pillars� make up an overall concept. Given that a capital requirement cannot be

assigned to every individual risk, the first pillar has to be enhanced by comprehensive examination

Checking the effectiveness of internal risk management systems and controls also has an important

part to play here. Finally, the market is taken as a complementary �regulatory authority� alongside

state supervision, in compliance with the free-market principle �as much government regulation as

necessary and as little as possible�.

At this abstract level the Basel procedure exhibits no banking-specific features whatsoever and can

therefore basically also be applied to the insurance sector. In practical terms, this makes sense for

various reasons:

the United States.

single capital requirement.

system.

becoming increasingly important in insurance business. Moreover, as the traditional boundaries

between the financial sectors become less distinct, cross-sectoral competitive pressure is mounting.

Both encourage market discipline.

industry makes it easier to secure cross-sectoral consistency in accordance with the precept

of similar regulatory requirements for similar risk.

The concept of supervision resting on three pillars thus also provides a future-proof framework for

insurance undertakings. Now it is a matter of implementing this concept in such a way as to accommodate

the specifics of the insurance business.

with the introduction of a minimum capital requirement and a target capital level. Basically, the latter

would correspond to the capital a company requires to run its business with a low probability of failure,

whereby this probability has yet to be quantified.

This gives rise to the question as to how the target capital level can be justified and how exactly it is to

be determined (what is meant by �low probability of failure�). Basel II is of little help here because of

the marked difference in the respective regulatory objectives (risks). Banking supervision is concerned

first and foremost with the avoidance of systemic risk. This is taken to mean liquidity and solvency

risks (long-term lending, short-term financing) involving the danger of runs on banks, possible knock-on

effects as a result of close inter-dependencies (interbank market) and the potential threats to payment

systems or macro-economic financing in the event of major bank failures.

Insurance business does not entail systemic risk of this kind. There are, for example, no such close

interrelationships between insurance providers. Nor is there any imminent danger of sudden outflows

of liquidity, given that insurance policies cannot be cancelled at liberty. That said, the activities of some

big insurers have recently begun casting doubt on this traditional view as they take over credit risks

from banks (credit derivatives), for instance, or transfer their own risks to the financial markets (ART �

alternative risk transfer - products). The IMF recently addressed this issue (in its Stability Report June

2002) but clearly ruled out the danger of systemic repercussions from insurance insolvencies, partly

because of the negligible volume of these new financial market activities.

The supervision of primary insurers therefore continues to focus on individual protection, in other

words the protection of policyholders. Theoretically speaking, the optimum level of the regulatory capital

requirement is given when the following (necessary, but not sufficient) condition is satisfied: the

cost of a marginal increase in the level of capital must be consistent with the benefit (the added security).

In practical terms, however, this equilibrium condition is of little significance, as it cannot be operationalized.

An obvious solution is therefore to guarantee the original goal (protection of the policy-holder)

through the instrumental goal of solvency protection.

An important conclusion can be deduced from the above considerations. The concept of �solvency�

should not be interpreted too narrowly. Under no circumstances is this about securing an insurance

undertaking�s economic existence, and that would indeed be incompatible with regulative principles.

Basically, all that must be guaranteed is that if a company becomes financially distressed its commit-ments

towards its insurance customers can be discharged. This is also guaranteed, incidentally, in the

event of orderly liquidation (say by another insurance company taking over the failed undertaking). To

avoid unnecessary regulatory capital burdens, the benchmark for the above probability of failure

should not therefore be set �too low�.

The philosophy underlying the first pillar is as simple as it is convincing. Higher risks should be backed

by more own capital than lower risks. As with Basel II, it is planned to use a risk-based standardized

approach for calculation of the target capital. Alternatively, companies will also be permitted to use

their own internal risk models, provided these satisfy regulatory requirements. Basically, there are

presumably only two possibilities for the standardized procedure:

The architects of Solvency II stress the importance of balance sheet valuation with reference to market

values as the foundation of a risk-based solvency system. Given that insurers� technical liabilities are

not as a rule traded, near-market valuations must be estimated using suitable discounting methods.

The main problem lies in the valuation of embedded options and guarantees. This ties in with IAS

debates on fair value accounting.

RBC models have been in use for quite some time in the insurance business. Familiar variants are the

US National Association of Insurance Commissioners� (NAIC) supervisory model, the Standard &

Poors rating model and the life and non-life supervisory models of the GDV German insurance industry

association building on these. So this methodology automatically puts itself in the running, so to

speak, as a possible standard procedure.

RBC models begin with the own funds available at the insurance undertaking, which are composed of

the equity capital, and further asset and liability items of a capital nature on the balance sheet (TAC =

total adjusted capital). The TAC is then compared with the capital required.

The solvency requirement is subsequently determined with reference to the respective insurance undertaking�s

size and risk profile. To begin with, the risk capital required is calculated separately for

each risk category (investment risk, default risk, tariffication risk, �) using a simple method, the multiplication

of volume factors (e.g. premium levels) by risk factors. As a rule, within an individual risk

category a distinction is made between risk factors for specific insurance lines. For example, when

computing the tariffication risk the size of the risk multiple will depend on the business line considered.

In the S&P model, the risk capital requirements thus calculated are added up to the overall capital

requirement without any regard for diversification effects. In contrast, the NAIC and GDV models make

� at least partial � allowance for diversification effects.

When combining the individual risks into an overall risk, account is also taken of the correlations between

the risks. Since regulatory models generally tend to be conservative in design, the correlation

coefficients are often notionally set at one. In this case the individual risks are simply added together,

which is effectively tantamount to a sort of worst-case scenario, as no allowance is made at all for

diversification effects. If the correlations are set at zero, we obtain a kind of lower limit for the overall

risk (strictly speaking, this is only conditionally true, as correlations can also assume negative values).

An alternative to the RBC models described in the above are scenario-based methods for calculation

of the relevant target capital. An insurance undertaking�s future losses, which must be backed up with

own funds, are based on three components:

costs, commissions etc.)

The supervisors now define a certain number (trade-off between precision and practicability) of scenarios,

for which the insurance undertakings establish their losses (risk). First, the relevant risk factors

are determined for each of the risks that are to be considered. In a simple two-factor model, for example,

the market risks could be modeled with the aid of an interest rate curve and a share index. The

determination of specific values for the risk factors (10 % drop in share prices with a parallel 0.5 %

shift in the interest rate curve ) defines a certain scenario.

Certain requirements must be made of these scenarios in terms of their consistency. They should, for

instance, be �plausible�. Plausibility (in the narrower sense) is deemed given when the likelihood of the

an additional requirement could be that the values of the risk factors produce a coherent overall

picture. Moreover, the scenarios should be �adverse� or risk-heightening. From a risk aspect, there

can naturally be no interest in �fair-weather scenarios�. Scenarios can also be characterized in terms

of their complexity. With �simple� scenarios the regulators set explicit values for the risk factors. In the

case of �complex� scenarios the value ranges are prescribed only implicitly (for example by a certain

event such as an earthquake of a specific magnitude) and the undertakings must make their own

specification in conformity with their portfolios. Evaluation of the most adverse outcome of the scenario

then produces the target capital required.

Under certain conditions, to be set by the supervisors, insurers� internal risk models will also be recognized

for regulatory calculation of the target capital. If applied consistently, this will mark a revolutionary

innovation for the insurance sector. This so-called internal model approach was applied for the first

time in 1996 to market price risks (on the trading book) in the banking sector. Basel II additionally pro-vides

for the use of internal models for credit risk (albeit on a limited scale) and also for operational

risk.

Here, too, the underlying philosophy is simple: given the increasing complexity of financial products

and their valuation methods, companies themselves are best placed to assess their own risks correctly.

This implies supervision taking a more qualitative stance, with less monitoring of compliance

with certain balance sheet ratios and more attention to the structure and efficiency of the undertaking�s

risk management as a whole.

Internal models aim to develop tailored approaches and methods to assess company-specific risks.

Within a framework determined by just a few parameters (time horizon, scale of risk and confidence

level) risks are modeled with the aid of theoretical probability models using in-house data. The uniform

framework makes the different risks comparable, turning risk capital into a kind of common currency of

risk. Most internal models use a one-year time horizon, or several one-year periods for the consideration

of longer timelines. Value at Risk (VaR) on a high level of confidence is often taken as a risk

measure. The banks� internal models also use this risk measure, but with a much shorter time horizon

for market risk. Given certain theoretical weaknesses in VaR, so-called Tail VaR or the Expected

Shortfall is now frequently put forward as an alternative risk measure. As the name suggests, this de-notes

the loss expected in the event of loss (chart).

************

Like standardized models, internal models take a �bottom up� approach, beginning with the assessment of risk capital requirements in individual risk categories or sub-categories. Unlike standardized

models, this is carried out on the basis of stochastic models taking in-house data as their parameters.

While the risk factors in standard models represent an average undertaking, internal models translate

individual claims experience and risk exposure into risk capital requirements. Moreover, the risk factors

in standard models do not react directly to alterations in market parameters; as a rule they can be

adapted only with considerable time lags.

In contrast to standard models, individual models make consistent allowance for diversification effects

when aggregating the various risk categories into the overall risk capital for a company or an entire

group.

At present, certain categories of risk are not modeled, or not modeled explicitly, using standardized

approaches and can only be represented adequately by internal methods. In non-life business these

include the risk of natural perils, which is not explicitly valued in standard models even though it is the

most important driver of risk in some business lines. Finally, only internal models are in a position to

value explicitly embedded options and guarantees by life assurers.

The specific form that Solvency II takes will be determined not least by the supervisory authorities�

fundamental approach. To simplify matters, we can distinguish between two basic positions:

The third Basel II consultation paper is consistent with the demanding approach. The capital requirements

for each individual transaction are determined down to the very last detail on the basis of �best

practices�. The result is a complex set of rules and regulations on a hitherto unprecedented scale,

implementation of which will entail high costs.

We should always bear in mind, though, that company-specific situations can be modeled with the aid

of standardized processes to only a certain extent. Besides which, it would be a misapprehension to

believe that statistical methods could make risk management, or even risk measurement, entirely objective.

Material limits therefore exist to the possibilities supervisory authorities have to regulate insurance

undertakings� internal models in detail.

One illustration of this is the �fundamental dilemma of risk management�. For the assessment of future

risks over a certain time horizon, it makes a great difference whether the assumption is one of �nor-mal�

conditions or periods of stress. Depending on the scenario on which an assessment is based,

different methods will already be used for risk measurement ( in the case of market risk the variance-covariance

method, say, or extreme value statistics and stress testing). This gives rise to the question

as to how to handle what are generally extremely varied results. There is no simple mechanical

method of meaningfully combining the different risks (capital requirements). This dilemma arises with

practically all risk. The Basel Committee also addressed this issue in depth � but ultimately to no effect. Capital requirements for extreme events cannot be introduced as standard practice, because this

would imply having to hold far too much capital in �normal� periods. Shifting stress testing to the second

pillar does not solve the problem either as long as the consequences this has for capital are not

clear. That leaves us with a truth as simple as it is fundamental: companies must take risks and develop

a strategy for handling them appropriate to their objectives.

Moreover, the discussion surrounding Basel II (with procyclicality as the buzzword) shows that �over-regulation�

can trigger counter-productive reactions. If all market participants were to measure and

respond to risk in the same way, such homogeneity of behavior in periods of stress (market turmoil)

could even intensify the risks.

The moderate approach toward the supervisory philosophy would therefore appear to have quite a lot

in favor of it. The supervisory authorities cannot and should not relieve management of the responsibility

for steering pivotal corporate processes. Regulatory capital requirements are no substitute for company-

specific risk management. Regulatory capital can only ever serve as a kind of lower limit for the

capital actually required. Supervisors should adopt an open-minded and not too restrictive stance on

the admission and regulation of internal risk models. This is, moreover, consistent with the general

logic of the three pillar approach, since in this case the second and third pillars �mitigate� the first.

Specifically, this means openness and diversity of models instead of strict regulation.

In the insurance sector at least, an approach of this kind would be quite adequate. Experience teaches

us that very seldom do insurance companies suddenly become so financially distressed as to jeopardize

their very existence. Liquidity outflows do not occur �overnight�, so to speak, and even undertakings

incapable of meeting their long-term liabilities on a permanent basis are usually in a position to

generate positive cash flows (through new business, for example) in the short run. This leaves regulators

sufficient time to intervene.

It is undisputed that the broadening of the regulatory authorities� remit envisaged in Basel II will also

be applied to the insurance sector. Unlike Basel II, however, the intention is evidently to lay down

more clearly the possibilities for intervention by the supervisory authorities. The projected minimum

capital, which could be calculated using a simple method along the lines of the existing solvency mar-gin,

sets the first intervention threshold (chart). If an insurance undertaking�s financial resources fall

below this threshold, the supervisory authorities are obliged to take suitable steps to put the company

back on course towards the restoration of its target capital resources.

Pillar 1: Capital requirements

Target capital

(internal models/standard model)

Minimum capital

(simple model based on

existing Solva-model)

Intervention zone

for supervision

Required solvency capital

In this context a look across the water is instructive. The United States set up a regulatory review procedure

back in 1991. In contrast to Basel II, US regulators are obliged to take �prompt corrective action�

under certain clearly defined conditions. At the same time, the action they are permitted to take is

laid down explicitly (Box 2).

A rule-based procedure would also be a logical option for the EU, given that for the foreseeable future

national governments are likely to remain responsible for insurance regulation. That way, national

supervisors� discretionary scope could be restricted, making it easier to guarantee a level playing field.

The third pillar of Basel II, the expanded disclosure requirements, has at best been incidental to the

debate on supervision so far. Where differences of opinion have emerged, they have generally revolved

around the issue of whether the desired protection of business secrets can still be guaranteed.

This may be because recent financial scandals have triggered a general trend toward transparency.

But this still leaves the question of effectiveness. Effective market supervision requires more than the

mere availability of important information. There must also be people with a vital interest in this information who are, moreover, in a position to influence insurance undertakings by their behavior. Individual

policyholders are scarcely capable of this, because they lack both expertise and market power.

The remedy could lie in an approach that originated in the United States and is now gaining recognition

here as well. By this we mean upgrading subordinated external funds, a possible option for banks

and insurance providers alike. Subordinated external funds in the form of bonds issued by insurance

undertakings form part of regulatory capital. Holders of insurance bonds have powerful incentives to

track the issuer�s solvency very carefully, as they would automatically be �taken on board� if the company

were to get into financial difficulty. Most of the bondholders are, moreover, big institutional investors

in possession of the necessary know-how.

These considerations lead to the following proposal. Insurers should be obliged to hold a certain mini-mum

proportion of their capital in the form of subordinated bonds structured in such a way that new

bonds are issued at regular intervals (say half-yearly). Changes in the interest rate spread would then

offer intelligence on how the market assesses the respective company�s risk. They would also have a

direct impact on refinancing costs. Both have a disciplinary effect.

Implementation of the new supervisory regime would not be all that far off if the relevant Directive were

submitted in 2005. But given the complexity of the project and the many still unresolved issues, it

seems rather doubtful that the regulators will be able to keep to their official time schedule. That said,

the new supervisory concept is already foreshadowing events, with the big insurance undertakings at

least starting to adjust to the new scenario. So what changes can be expected on the insurance markets?

of supervisory systems and methods in the insurance sector. That improves the chances of

an international level playing field, taking insurance markets further down the road to global integration.

allocation of expensive capital will become an increasingly important factor of success.

The winners will be those insurance companies that succeed in building up company-wide integrated

risk management in which all the major risks are captured (comprehensive risk aggregation)

and managed at group level (maximum risk diversification).

become the key competitive factor.

the resources needed to introduce modern corporate governance methods. This will improve

their risk management and add value. They could, for example, reduce the volatility of their insurance

portfolio by increasing the number of mutually independent risks. This would help them save

capital in comparison to other companies that do not diversify their volatility risk in the same way

(the diversifiable or non-systematic part of volatility risk is not covered by fair premiums and must

therefore be backed by capital).

with its large number of small insurance undertakings, consolidation of the insurance market

is still in store.