Gestion Actif-Passif en assurance : Partie I

David CLERMONT, SwissLife France

Lyon, 3 Avril 2015, Formation d'actuariat ISFA, support de cours

Plan

• Introduction

• La nature des passifs d’assurance

• Les grands équilibres bilanciels d’une compagnie d’assurance

• Outils « classiques » de la gestion actif-passif

• Valorisation des actifs financiers et des passifs d’assurance

2
Introduction

ALM = Asset Liability Management.

ALRM = Asset Liability Risk Management. Plan DAY1

• Introduction
• La nature des passifs d’assurance
• Les grands équilibres bilanciels d’une

BSRM = Balance Sheet Risk Management. compagnie d’assurance
• Outils « classiques » de la gestion actif-passif
• Valorisation des actifs financiers et des
passifs d’assurance

BSRM = Economical Balance Sheet Risk Management

Normes : French GAAP, SII, MCEV

Actifs : Placements financiers

Passifs : Provisions Techniques,

Fonds Propres, valorisation.
3
Introduction

L’objectif de ce cours est de présenter la théorie et la pratique de la gestion actif-passif (ou ALM :
Asset Liability management) des entreprises d’assurance.

La gestion actif –passif a été toujours au cœur du métier de l’assureur mais elle a connu un
développement rapide au cours des 20 dernières années.

La sophistication des instruments financiers, le développement des outils de modélisation, les

nouvelles exigences des parties prenantes (régulateurs, mais aussi actionnaires, analystes financiers,
agences de notation et mêmes assurés) y ont fortement contribué.

ALM s’inscrit désormais dans une approche plus globale de gestion des risques (approche dite ERM:
« Entreprise Risk management »), par opposition à l’approche classique « par silos ».

Par conséquent, ce cours aura comme objectif d’élargir la vision de la gestion actif-passif au-delà de la
simple adéquation entre les actifs et les passifs d’un scenario économique central et soulignera
l’importance de la prise en compte des interdépendances entre les risques.

De même, la cohérence globale entre ALM, normes comptables, allocation d’actif, gestion de
l’exigence en capital et gestion des produits sera abordée.

De manière générale, une plus grande attention sera accordée à l’assurance vie où la problématique
ALM est plus complexe. Toutefois, le cours sera complété par des exemples Non vie ou santé&
prévoyance.

4
The importance of Risk management : is the insurance
industry different from the others ?

In the insurance sector, the production process is inverted : the policyholder pays a certain premium before potentially
receiving a contingent claim payments later.

In contrast, an “ordinary” firm invests an initial amount in order to acquire inputs (e.g. raw materials). The raw materials
are transformed and sold to the customers.

This feature explains why the ALM management of an insurance company relies on a prospective view and on
stochastic modeling techniques. In fact, short- term cash flows are of limited use as a means of understanding
of the business model.

However, whether the life insurance industry is fundamentally different from the banking industry is debatable (see [1]).
At this stage, we notice :

• There are similarities between some life insurance guarantees and the financial options.

• The insurance (even non life) policy could be seen as a contingent loan : the policyholders lend an amount
(the premium) to the insurer. The policyholders will be reimbursed in the case of claim (contingent loss).

• However, in contrast to the insurance firms, commercial banks can create money under a fractional-reserve
banking system (most of the systems worldwide).

5
The importance of Risk management :is the insurance
industry different from the others ? (2)

Because of this inverted production cycle :

The insurance industry is regulated in order to protect the customer.

The regulators require the insurer to lock a capital. The amount of the capital depends more
or less on the risk profile (Solvency I vs Solvency II vs economic capital). Note that even on a
completely unregulated market, the insurer will need to a lock a capital, but a different amount
from the regulatory one.

The risk can be financed through equity or hybrid debt, mitigated, avoided (risk selection),
or transferred (e.g. reinsurance).

The asset-liability management (ALM) requires a prospective and stochastic view .

6
The importance of Risk management

Before discussing insurance ALM in ERM context, it is worth recalling the importance of risk management and
why risk management creates value.

This nontrivial question was debated in the sixties and in the seventies. In fact, the Modiagliani-Miller theorem (1958)
states that under the following “perfect market” assumptions :

(i) neutral taxes;

(ii) no capital market frictions (i.e., no transaction costs, asset trade restrictions or bankruptcy costs);

(iii) symmetric access to the credit markets (i.e., firms and investors can borrow or lend at the same rate);

(iv) firm financial policy reveals no information,

the firm value does not depend on its corporate structure.

Consequently, many risk management activities and the risk transfer cannot add value, since the value of the firm
cannot by changed through financial transactions.

This conclusion is also in line with the CAPM (Capital Asset Pricing Model, Sharp 1964). Under the assumption of
“perfect” markets, the company should not focus on the idiosyncratic risk (i.e. company-specific) but only on the
systematic (or beta) risk.

The intuitive reason is that the CAPM assumes the shareholders can always reduce the specific risks themselves
through a diversification w/o cost.

Although theoretical, these conclusions are important. These frameworks show that the Risk management
matters precisely because the assumptions of the Modigliani-Miller theorem and CAPM are partially violated in
practice.

7
Why does Risk management matter ?

The Modigliani-Miller theorem shows why Risk management matters in reality :

Transaction costs

Asymmetric information

Bankruptcy costs

8
ALM and ERM (Enterprise Risk Management) (1)

Over the last decade, a new approach emerged : Enterprise Risk management (ERM)

ERM is essentially driven by the following changes:

• Inherent complexity of the risks, illustrated by high –profile corporate failures

• Pressure from the stakeholders (shareholders, regulators, analysts, rating agencies….)

• Move from “silo” to “global” (or “holistic”) approach

• Increasing sophistication of quantitative models combined with “qualitative” tools

• Risks are seen not only as a threat, but also as an opportunity

9
ALM and ERM (Enterprise Risk Management) (2)

However, various definitions of ERM are available.

These definitions provides insightful information on the organizations’ philosophies and ambitions (see
words in bold letters) :

CASACT (see [4]) : ERM is the discipline by which an organization in any industry assesses, controls, exploits,
finances and monitors risks from all sources for the purposes of increasing the organization's short- and long term
value for its stakeholders.”

COSO (see[5]): ERM is a process, effected by an entity’s board of directors, management, and other personnel,
applied in strategy setting and across the enterprise, designed to identify potential events that may affect the
entity, and manage risk to be within the risk appetite, to provide reasonable assurance regarding the
achievement of entity objectives”

S&P (see[6]) : S&P defines risk management as the process companies use to identify and monitor significant
risks and to set limits for these risks that reflect their risk appetite, competencies, and resources. Enterprise risk
management extends these core risk management processes so they can be adopted consistently across all
risks and in a way that supports the company’s overall corporate strategy.

ALM should be embedded in this broader ERM approach.

10
 Asset allocation and the Risk management cycle

The asset allocation needs to be embedded in an integrated risk management framework.

Identification
Monitoring/Reporting
All significant risks have to
be taken into account
Asset allocation
Asset-Liability management
Product management
Risk transfer (e.g. reinsurance)
Etc..

Manage/Implement/Exploit
Assessment/measurement
Risk management :
Quantification : statistical distribution,
avoid, accept,
aggregation methodology, assumptions.
mitigate transfer….
Assessment methods for « qualitative »
risks (strategic, operational, legal…).

Asset allocation and ALM are just two of the several dimensions under consideration in each step.
Solvency II requirement scan be mapped to each component of the risk management cycle.

11
Asset allocation and the Risk management cycle (2)

Besides quantitative techniques, operational, strategic and business risks have to be considered as
well. Some illustrative examples of potential loopholes in the risk management framework are :

Liquidity risk
The Traditional quantitative frameworks consider that all assets can be sold immediately which may not be the
case in distressed market.
From ALM life insurance perspective, liquidity risk is important since in some cases the local regulation or the
contractual clauses allow the policyholders to require the immediate payment of liability cash flows.
The traditional tools need to be complemented by stress and reverse stress scenarios in order to avoid an
erroneous view on the risk-reward asset balance.

Model risk
In general, tail correlations are key for the assessment of the capital need. Tail correlations between various
risk factors (not necessarily only related to the market risk) can be easily under estimated.
A classic illustrative example is a pandemic scenario combining the following effects:
• equity market drop, widening spread motivated by fears about the long term economy trends,
• payments of death and disability benefits and/or payment of medical expenses,
• issue with the employees’ availability leading to operational losses.
Another example is not modeled future guarantee resulting in under-priced products and an erroneous view of
the risk taking capacity on the asset side.

12
Definition of ALM in the insurance sector

Various definitions of ALM in the insurance field are available:

INTERNATIONAL ASSOCIATION OF INSURANCE SUPERVISORS [see 7]

“Asset-liability management (ALM) is the practice of managing a business so that decisions and actions taken with
respect to assets and liabilities are coordinated. ALM can be defined as the ongoing process of formulating,
implementing, monitoring and revising strategies related to assets and liabilities to achieve an organisation’s
financial objectives, given the organisation’s risk tolerances and other constraints”.

From [see 8 ] “Asset & liability Management (ALM) is the practice of managing an insurer so that actions taken with
respect to assets and liabilities are designed to address the broad set of financial risks inherent in their joint
behaviour.

The key concepts are highlighted. The “ERM” logic is visible.

13
Plan

• Introduction

• La nature des passifs d’assurance

• Les grands équilibres bilanciels d’une compagnie d’assurance

• Outils « classiques » de la gestion actif-passif

• Valorisation des actifs financiers et des passifs d’assurance

14
L’origine des passifs d’assurance

Les passifs d’assurance (la valorisation de l’engagement net de l’assureur

vis-à-vis de l’assuré)

naissent

des GARANTIES et OPTIONS du CONTRAT d’assurance

La connaissance des contrats est PRIMORDIALE,

INDISPENSABLE et (quasi) PREALABLE à l’ALM. !
15
 Life insurance : the nature of the financial
guarantees

The “market consistent” valuation approach is a common tool in Life insurance.

In fact, various guarantees sold to the policyholders represent financial options. For example (French Life
Insurance market) :
• Minimum guaranteed rate (MGR, “Taux minimum garanti”)
• Contractual profit sharing (« Participation aux bénéfices (PB) contractuelle »)
• Regulatory minimalprofit sharing (« Participation aux bénéfices minimale »)
• Lapse/surrender features (partial/ total)
• The right to invest additional premiums
• Switching option (« Droit d’arbitrage entre support Euro et support UC des contrats multisupport ») between the
Euro and the UL support in the so called « multisupport » products
• Option of future conversion into annuity
• Death guarantee for Unit- linked contracts , or more generally all GM*B

These guarantees represent financial « options » :

The policyholder has the right but not the obligation to surrender, switch, invest an additional
premium and so on..

In addition, the liabilities of the standard “French” saving products depend on the discretionary bonus policy
(“taux servi”, above the regulatory & contractual profit sharing). The insurer’s bonus policy depends on the
competitive pressure which is particularly difficult to model.
16
Life insurance : the nature of the financial
guarantees (2)

Most of the traditional European life contracts offer a combination of Minimum Guaranteed rates and some
kind of profit sharing.

In addition, various complex GM *B guarantees are offered through UL products on different markets (US,
Japan, some developments in Europe):

• Guaranteed Minimum Death Benefit (GMDB)

• Guaranteed Minimum Accumulation Benefit (GMAB)
• Guaranteed Minimum Withdrawal Benefit (GMWB)
• Guaranteed Minimum Income Benefit (GMIB)

GM*B are not common on the French market where saving “multisupport” products and retirements
products (e.g. “Madelin”) are sold. Floor death guarantees for Unit-linked are common (“garantie plancher en
cas de décès”).

17
Life insurance guarantees (1) :illustrative examples

Minimum guaranteed rate (« Taux minimum garanti » )

0% net guaranteed in the recent products:

« Les versement en euro font l’objet d’une garantie…

This guarantee is also required by « Code des assurances ». Note that even such low guarantees induce a cost for
the insurer in a low interest rate environment.

Lapses (“rachat”) :
« Vous pouvez à tout moment demander le rachat total de votre contrat et recevoir la valeur de rachat de votre
contrat. La valeur de rachat de votre contrat est égale à la valeur atteinte sur le contrat, telle que définie à
l’article « Calcul des prestations », diminuée des avances consenties (principal et intérêts) et non
remboursées ainsi que des éventuelles primes restant dues »

Profit- Sharing
« Le taux de participation aux bénéfices effectivement attribué au titre de l’exercice précédent est égal à 85 %
du rendement net réalisé dans le fonds en euros diminué des frais de gestion, il ne peut être inférieur au taux
minimum annoncé en début d’année »

Optional flexible premiums : « Les versements libres sont possibles à tout moment »

Optional future conversion into annuity

18
Life insurance guarantees : past developments

In addition to the regulatory and the contractual profit sharing, the French insurance companies can
serve a discretionary (or « commercial ») rate.

This implicit option is correlated to interest rates and in combination with the potential lapses
represents also a financial option.

19
 The nature of “non hedgeable” guarantees : life and
non life business

In the beginning, the insurance has been focused on the following risk drivers :
• mortality (death capital), longevity (annuities)
• disability risk (morbidity) and medical expenses
• non life risks (eg fire, motor , third party liability etc.. etc…,

These risk are “non hedgeable” in the sense that there is no deep and liquid market (although there are
some mortality and longevity bonds traded).

As a consequence, we can not rely on the risk-neutral valuation approach.

In this case, the liability are valued through an indirect approach : Cost of capital.

The capital amount is based on the capital required for the given lines of business (i.e. under Solvency II
it corresponds to the SCR w/o the hedgeable market risk).

20
Plan

• Introduction

• La nature des passifs d’assurance

• Les grands équilibres bilanciels d’une compagnie d’assurance

• Outils « classiques » de la gestion actif-passif

• Valorisation des actifs financiers et des passifs d’assurance

21
 FR GAAP vs bilan économique

La logique du FR GAAP est une logique de valorisation basée essentiellement sur le coût historique et
la prudence comptable.

A fur et à mesure du développement de la norme, des éléments correctifs (exemple: « Provision pour
risques d’exigibilité, PRE) ont été ajoutés, visant à rendre la norme à la fois plus « économique » et
prudentielle.

Néanmoins, la logique rétrospective du FR GAAP reste assez différente d’une approche économique
prospective de valorisation (et notamment de celle de Solvency II : introduction d’une notion de
valeur de marché ou Market Consisten Valuation).

Pourquoi étudier les règles FR GAAP dans le cadre de la gestion actif-passif ?:

• Pour certaines sociétés, le FR GAAP reste une contrainte managériale et doit être intégrée.

• En assurance vie, les règles de participation aux bénéfices contractuelles ou réglementaires sont basées
sur un résultat financier et/ou technique issus FR GAAP. Dès lors, ces règles ont un impact sur le Best
Estimate.

• Le résultat fiscal est très proche du résultat FR GAAP, l’impôt est une composante importante dans la
valorisation des fonds propres.
• Car le bilan French Gaap est le point de départ incontournable et de référence à déformer pour construire
un bilan économique : connu et habituel pour tout le monde (management, régulateur).

Voir [9] pour plus de détails sur le lient entre gestion actif-passif et provisions FR GAAP.

22
FR GAAP vs bilan économique

Pour plus de détails, se référer aux

rappels présentés en cours

Valorisation des actifs :

• Amortissement surcote/décote pour la plupart des obligations

• Action et OPCVM au coût historique (mais règles de PDD et PRE, voir ci-dessous)

• Etc…

Valorisation des passifs (voir ci-dessous)

23
FR GAAP vs bilan économique en assurance vie
Pour plus de détails, se référer aux
rappels présentés en cours
Pour le lien avec le bilan IFRS, voir le
cours [35]

FR GAAP Solvency II

Actif Passif
Actif Passif
Immobilisations
Fonds propres Own Funds
Dette hybride Dette hybride

Actifs financiers Actifs financiers

nets de PDD Provisions en valeur de marhé
techniques vie Best Estimate vie

Risk margin
PM UC
Actifs UC PM UC
Actifs UC
Bas de Bilan Bas de Bilan Impôts différés
Bas de Bilan Bas de Bilan

24
FR GAAP : La Provision Mathématique (PM)

Inscription dans les comptes d’une évaluation des engagements contractés par les assureurs
envers les assurés.

En vie, le poste PM représente environ 90% du passif. Différents principes de valorisation sont
appliqués selon la nature des provisions :

• Principe comptable : les engagements sont évalués avec une table réglementaire et
actualisés au taux défini à la souscription
• Principe coût historique : les engagements sont comptabilisés à leur coût historique
• Principe économique : les engagements sont évalués avec une table d’expérience et
actualisés avec un taux d’intérêt.

Des provisions spécifiques existent si les taux descendent en dessous du taux d’actualisation à la
souscription.

Pour les Unités de comptes, les engagements sont comptabilisés à la valeur de marché des
placements en représentation. Mais des provisions complémentaires peuvent être constituées en cas
de garantie plancher.

25
La Provision pour Sinistre à Payer : PSAP

La PSAP constitue le pendant de la PM en Vie et représente, dans le bilan d’un assureur non-vie,
plus de la moitié du passif.

La PSAP correspond à la valeur estimative des dépenses, tant internes qu’externes, nécessaires
au règlement des au règlement de tous les sinistres survenus et non payés.

En FR GAAP, elle doit être calculée de façon prudente.

L’entreprise peut choisir une méthode dossier à dossier (évaluation par le gestionnaire du montant
restant à payer et estimation de la charge de tardifs) ou des méthodes statistiques (fondées sur
l’évolution historique de la sinistralité par branche d’activité) pour évaluer la provision.

26
La Provision pour Participation aux Excédents

Via les clauses de participation aux bénéfices, les compagnies s’engagent à reverser une part des
bénéfices qu’elles réalisent chaque année.

Les compagnies peuvent décider de différer dans le temps (sur 8 ans), le versement de ces
bénéfices attribués aux assurés et d’en lisser la distribution.

Elles peuvent ainsi doter la PPE.

Il s’agit donc d’une réserve de lissage qui permet de piloter la façon dont la compagnie souhaite
distribuer les produits financiers qui appartiennent aux assurés sur plusieurs années.

Les reprises de réserve ne peuvent être utilisées que pour verser une rémunération au-delà du taux
minimum garanti des contrats.

Buffer important économiquement : valorisation des fonds

propres, sensibilité des fonds propres aux facteurs

27
La Réserve de Capitalisation

L’objectif de cette provision est d’empêcher les entreprises d’extérioriser et de distribuer les plus-
values obligataires en bas de cycle des taux d’intérêt et d’appauvrir en termes de rendement comptable
le stock.

Réserve de lissage, elle vise donc à amortir les effets de mouvements de taux d’intérêt et favorise une
gestion Buy & Hold.

En cas de cession d’un titre dans un contexte de baisse des taux, la réserve est dotée à hauteur de la
plus-value réalisée. Dans le cas inverse c’est une reprise de réserve qui vient compenser la moins-value
(dans la limite de la réserve).

Depuis 2010, un impôt est applicable aux plus ou moins-values obligataires et donc s’applique avant
dotation/reprise de la réserve.

Buffer important économiquement : valorisation des fonds

propres, sensibilité des fonds propres aux facteurs

28
La Provision pour risque d’exigibilité : PRE

Cetteprovision technique a pour fonction de permettre à l’entreprise de faire face à ses

engagements en cas de moins-value latente des actifs non amortissables.

Elleest calculée de façon globale sur la poche des actifs classés en R332-20 (actions, fonds,
etc…).

La moins-value globale de la poche est provisionnée.

Un amortissement du provisionnement est possible : par tiers ou depuis 2009 sur une durée, de
maximum 8 ans, égale à la duration du passif.

La valorisation en coût historique de l’actif est respectée puisque la valorisation des titres à l’actif de
change pas.

Mécanisme procyclique fort: valorisation des fonds propres,

sensibilité des fonds propres aux facteurs

29
La Provision pour risque de Dépréciation
Durable : PDD

Contrairement à la PRE, cette provision est une provision financière (comptabilisée à l’actif, en
déduction de la valeur comptable).

La PDD se calcule ligne à ligne.

Elle a pour but de venir compenser un risque de ne pas recouvrer une moins-value durable sur
un titre constatée depuis une période prolongée ou de provisionner un risque de défaut avéré sur
un titre obligataire.

L’écart entre la valeur de réalisation et la valeur de recouvrabilité doit être provisionné.

Les seuils généralement utilisés sont une baisse de –X% constatée depuis plus de Y mois.

30
La Provision pour Aléa Financier : PAF

La PAF a pour but de compenser une baisse du taux de rendement de l’actif.

Constituée dès lors que le montant total des intérêts techniques et du montant minimum de
participations aux bénéfices rapporté aux PM est supérieur à 80% du taux de rendement de l’actif.

L e montant de la provision est alors égal à la différence entre les PM de l’inventaire et les PM
recalculées avec un taux d’actualisation égal à 80% du taux de rendement comptable des actifs.

Pour mémoire, le taux de rendement comptable de actifs correspond au rapport des produits nets
comptables des placements augmentés des plus-values comptables sur cession, nettes de moins-
values, nettes d’amortissement sur le montant moyen des placements.

La PAF constitue une marge de prudence, mais elle n’est pas prospective.

31
La Provision Globale de Gestion : PGG

La PGG vise à compenser une mauvaise tarification des contrats et garantir la capacité de
l’assureur à gérer les contrats.

Elle est dotée à hauteur de l’ensemble des charges futures non couvertes par des
chargements sur primes ou par des prélèvements sur contrats.

Elle se calcule de façon prospective via la projection de comptes de résultat. Les marges
négatives sont ensuite actualisée avec un taux prudent.

32
La provision pour Garantie Plancher

Les contrats en Unités de Compte (UC) peuvent comporter des garanties en cas de décès ou en cas de
vie consistant à garantir un montant minimum du capital au bénéficiaire.

Hors les UC sont comptabilisés à la valeur de marché des supports en représentations.

Une provision doit être constituée afin de s’assurer que l’assureur sera en mesure de délivrer les
garanties offertes, même en situation adverse des marchés financiers.

Deux méthodes sont envisageables :

 une méthode déterministe : provisionnement des capitaux sous risque

 une méthode stochastique, dite « méthode des puts » correspondant à de l’évaluation d’option

33
Lien avec Solvabilité II

La comptabilisation au coût historique dans le bilan FR GAAP conduit à la mise en place de provisions
venant compenser les limites de la méthode : sous-tarification, risques de marché durable, etc …

Les méthodes de provisionnement sont le plus souvent déterministes.

Solvency II introduit la notion de valeur de marché à l’actif et de Best estimate+ Risk margin au passif.

En conséquence, dans un bilan Solvency II, tous les postes « provisions techniques » disparaissent pour
laisser place au Best Estimate (BE).

Le BE correspond à l’estimation des flux de trésoreries futurs des contrats d’assurance (primes
périodiques, sinistres, intérêts et PB versés, frais) pondérés par leur probabilité d’occurrence et actualisés, et
donc couvre le champ de chaque provision existant jusqu’à présent.

L’assureur va désormais se placer dans un environnement prospectif, le plus souvent stochastique, afin
d’afficher au bilan une vision économique des ses engagements selon les paramètres préconisés par
Solvency II (e.g. courbe d’actualisation, reconnaissance des primes futures selon le principe de « time
boundary »).

34
Plan

• Introduction

• La nature des passifs d’assurance

• Les grands équilibres bilanciels d’une compagnie d’assurance

• Outils « classiques » de la gestion actif-passif

• Valorisation des actifs financiers et des passifs d’assurance

35
Basic traditional techniques

As discussed previously, one of the key ALM objectives is the adequacy between the assets and
the liabilities.

Traditional approach : Deterministic scenario

A traditional ALM tool consist in performing a single deterministic scenario and analyzing the asset and
liability cash-flows. The scenarios are essentially used in order to assess/manage the interest rate risk.

In this context, 2 useful tools based on the expected cash flows under “base” or stressed scenario are :
• Liquidity gap

• Duration gap

Although the stochastic techniques are widely used in Life business, the deterministic scenario analysis is
still a valuable tool in the following cases:
• Communication tool for the decision makers

• Non life business

• Stress-scenarios and reverse scenarios analysis in order to complement the Economic capital calculations (VaR
99.5%, ES 99%), see later.

36
Basic tools : duration (1)

Duration (“modified duration”) (see [3] for details) :

Duration is a simple tool largely used in asset management, and by extension in ALM (in banking and in insurance sector).
Assume i is the compounded, while δ is the continuous interest rate; e = 1 + i
δ

(1) D ( P ) = − (1 / P )( dP / di ) = − d ln( P ) / di

Where P is the price of the instrument (e.g. a bond).

Note that the definition is very general, and it is valid for cash flow dependent on the interest rate, including or excluding
embedded interest rate options, etc. It is also applicable if P is valued through stochastic valuation techniques.

For a small ∆ i ; the duration is approximated through :

(P ( i + ∆ i ) − P ( i ) )/ P ( i )

For that reason, the duration is often interpreted intuitively as the increase( resp. decrease) of the market price (in bps)
following a parallel decrease (increase) of interest rate by 1 bps.

If the change of the interest rates is expressed in bps per year, the unit of the duration is « years ».

See [3],[10] for details.

37
Basic tools : duration (2)

Duration viewed as a weighted average time to maturity :

Assume the cash flows do not depend on the interest rates and interest rate curve is flat.
n
The present value is given by P = ∑ CF k /( 1 + i ) k
k =1

 n

Therefore, D ( P ) =  − 1 /( 1 + i ) ∑ k /( 1 + i ) k  / P
 k =1 

This duration is called “modified” because of the term 1/(1+ i) .

Sometimes a “non modified” version is used (i.e. w/o the term above) corresponding to the original concept introduced by
Macaulay (1938) :

D Macaulay ( P ) = D ( P )( 1 + i )

The Macaulay duration and the modified duration are identical in case of a continuous compounding.

The “weighted average time to maturity” definition cannot be applied straightforwardly to “stochastic” cash flows. As a
consequence, in this case the definition (1) is used.

38
Basic tools : duration (3)

•For investment instruments (or liabilities) with fixed cash flows, the modified duration can be computed analytically. For
example :
n
• For 0 coupon bond : P = F /( 1 + i ) D ( P ) = n /( 1 + i ) D Macaulay ( P ) = n (intuitive)

• For a perpetual bond P = coupon / i D ( P ) = 1/ i D Macaulay ( P ) = (1 + i ) / i (it is not infinity)

• The modified duration (resp. Macaulay duration) of a portfolio of assets is given by the weighted average (to the
market value) of the modified durations (resp. Macaulay durations) of each asset composing the portfolio :

n n
D ( P ) = ∑ D ( Pk ) ω k = Pk / ∑ P j = Pk / P
k =1 j =1

•Advantages of the duration concept : overall an intuitive tool, easy to implement.

•Drawbacks : first order term in the Taylor's expansion, captures adequately a significant part of the interest
rate risk only if the instrument is simple (e.g. bonds).

39
Basic tools : convexity (1)

Definition (Convexity): C = (1/P)(d 2 P / dr 2 )

The convexity is expressed in “periods squared”.

Back to the Taylor's expansion :

∆ P/P = -D ∆ r + 0 . 5 C (∆ r )
2

In other words, the duration captures the first order effect and the convexity captures the second order effect regarding the
sensitivity of the value to changes in the interest rates.

D ( P ) = n ( n + 1) /( 1 + i )
•The convexity of a 0 –coupon bond is

•The convexity is an important concept in ALM (see next chapters).

See [3],[10] for details.

40
 Basic tools : convexity (2)
Example : convexity of a bond

Duration and convexity of a bond

(maturity 20 years,coupon 5%)

210,0
•The inclusion of the convexity (second term of Taylor’s
190,0
expansion) leads to a better estimate of the price
170,0
sensitivity to the interest rates.
150,0
Price

130,0

110,0 •Later on, we will discuss the liability convexity which is

90,0 usually larger than the asset convexity.
70,0

50,0
0,00% 1,00% 2,00% 3,00% 4,00% 5,00% 6,00%
Yield

Bond price Duration price estimate Convexity price estimate

350,0

300,0
Some stylized facts about duration and convexity :
250,0

Shorter maturities, Bond price Maturity 5;

Bond price
200,0 coupon 2%
Bond price Maturity 20;

Higher coupons, coupon 2%
150,0 Bond price Maturity 20;

Higher yields, coupon 10%

100,0
have a shorter duration and a lower convexity.
50,0

0,0
0,00% 2,00% 4,00% 6,00% 8,00%
Yield

41
Basic ALM adequacy tools: the duration gap based on
a single economic scenario

The duration gap is defined as

Duration gap= Duration Liability – Duration Asset

Similarly to the bond duration, the duration gap measures the 1st order sensitivity of the economic worth.

Usually a modified duration is applied.

A “weighted” duration gap is often used because the liability amount does not take into account the shareholders’ equity

Duration gap= Duration Liability *(Market value Liab/Market value Assets – Duration assets)

Key ideas :
Cash flows

• A positive duration gap means that the economic worth will 16,0
decrease if the interest rates decrease and vice versa. 14,0

12,0
• The duration gap is only a 1st order proxy of the insurer’s D=8.21
10,0
interest rate risk. This measure should be considered with

Cash flows
Portfolio 1
8,0
caution, especially if derived from a single scenario because of Portfolio 2

the liability convexity (esp. in life business) and the model 6,0
D=8.21
uncertainties. 4,0

2,0

• A duration gap close to 0 does not imply that the asset and the 0,0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
liability cash flows match perfectly. Two portfolios with exactly Time
the same duration of 8.21 years are shown in this example.

42
Les gaps de trésorerie (liquidity gap) (1/2)

Le rôle du gestionnaire ALM est de s’assurer d’une bonne adéquation des flux issus du portefeuille
d’actifs (tombée de coupons, remboursements, dividendes) avec les flux issus du portefeuille des
assurés (échéances, rachats, décès, rentes, rentes incapacité-invalidté, paiement de sinistres en non
vie).

Le niveau d’adéquation visé dépendra de l’appétence au risque de l’entreprise.

Les trésoriers appellent « gap de liquidité» la différence entre les flux actifs et passifs pour des
intervalles de temps définis. Cette mesure donne une idée sur les risques de liquidité et de taux
auxquels une compagnie est exposée.

La plupart des flux ne sont pas déterministes, il est nécessaire de faire des hypothèses pour étudier
les projections des flux (taux de rachats, taus de croissance des dividendes,…). Une fois ces
hypothèses fixées, les flux peuvent être projetés et comparés.

Des stress scénarii de liquidité restent un outil important malgré le développement des techniques de
projections stochastiques car celle-ci tiennent modélisent rarement les risques de liquidité.

43
 Les gaps de trésorerie (liquidity gap) (2/2)

Gaps de trésorerie sur 30 ans

600000
Ce type d’indicateur permet
d’orienter les
500000 investissements d’aboutir à
un bon adossement des flux
400000 Flux actif- flux passif et éviter un risque de
liquidité.
Solde de trésorerie en Euros

300000
En revanche, il est
200000 nécessaire d’étudier
plusieurs scénarii
100000 économiques afin de capter
les interdépendances.
0

-100000

-200000

-300000
2013

2014

2015

2016

2017

2018

2019

2020

2021

2022

2023

2024

2025

2026

2027

2028

2029

2030

2031

2032

2033

2034

2035

2036

2037

2038

2039

2040

2041

2042

2043
44
Le ratio de financement (1/2)

Le ratio de financement est défini comme le rapport de l’actif en valeur de marché et de la valeur
actuelle des engagements de l’assureur au taux du marché.

Le ratio de financement doit être supérieur à 100% (corresponds à des « own funds » au sens de
Solvency II positifs).

Traditionnellement, le ratio de financement à calculé dans un référenciel déterministe.

Dans une perspective ALM, il est intéressant de mesurer la sensibilité du ratio aux différents facteurs
de risques, et notamment aux niveaux des taux.

45
Le ratio de financement (2/2)

Evolution du ratio de financement

Revalorisation minimale à Max(TMG, Taux marché)

110.0%
Ratio de financement au
31/12/12
107.5%

Niveau de taux maximum

105.0%
supportable par rapport aux
fonds propres disponibles
Ratio de Financement

102.5%

100.0%

97.5%

95.0%

Taux de TMG moyen du

portefeuille
92.5% Niveau des taux 10 ans au
31/12/12

90.0%
0.00% 0.50% 1.00% 1.50% 2.00% 2.50% 3.00% 3.50% 4.00% 4.50% 5.00% 5.50% 6.00% 6.50% 7.00%
Niveau des taux d'intérêt

Exemple théorique largement inspiré de [9].

Cet exemple traduit dans une certaine mesure la
convexité des passifs (voir chapitres suivants).

46
Plan

• Introduction

• La nature des passifs d’assurance

• Les grands équilibres bilanciels d’une compagnie d’assurance

• Outils « classiques » de la gestion actif-passif

• Valorisation des actifs financiers et des passifs d’assurance

47
Recall : « market consistent » valuation principles
•The purpose of this chapter is not to present extensively the “market consistent” valuation and the “risk-neutral” techniques.

• However, these methods play a key role in the assessment of the available economic worth. Therefore, it is crucial to
understand the underlying assumptions, the limitations and the areas of applicability. The presentation is based on [10].

• In theory, there are 3 methods to achieve the “market consistency” and to represent the investors’ risk aversion:

T=0 T=t p,1 − p :"real world" or " physical"probabilities

up
p S1
S0 < e − rt E P [S ] = e − rt ( pS1up + (1 − p ) S1down ) because of the risk aversion
Asset price S0
1− p S1down

Risk neutral
Deflators Utility functions
valuation

Discount at a state -dependent rates d

Discount at the risk free rate r and adapt “Measure” the utility of the cash
the probabilities (« risk neutral ») flows :

S0 = S0 = e − rt E P [U ( S ) ]
S0 = e − rt E Q [S ] = − d up t
− rt up down
=e pS1up + e − d down t (1 − p ) S1down
= e ( qS 1 + (1 − q ) S 1 )

where Q is the “risk neutral”

probability

Active academic research is

Widely used for MCEV Only of theoretical
on-going.
purposes and valuation interest
internal models 48
« Risk neutral » valuation : why does it work ?
•A fundamental result : under some assumptions, the “risk neutral” valuation (i.e. under probability Q) provides
the market value of an option.

“No arbitrage” principle

Consider the following instrument :
T=0 T=t
up
S −K
1
It is a call
price = ?
0 T=t

•Let us consider the following portfolio at t=0 :  S up − S down 

S1up −  1 up 1 (S1up − K ) = S1down
• Short of k options, where k = S1up − S1down ; Long of 1 underlying asset  (S1 − K ) 
up
(S − K )
1
S1down

• The strategy leads to a deterministic pay-off at t=[Link] return from t=0 to t=1 must be the risk-free rate,
because of the “no arbitrage” assumption.
S1up − K
S0 − k price(call) = e S − rt down price(call) = up down (S0 −e−rt S1down)
1 (S1 − S1 )

S0ert − S1down
Let q =
S1up − S1down price(call) = q e−rt (S1up − K )
ert − d
If up
S = uS0 S
1
down
1 = dS0 q=
u −d
price(call) = q e−rt (uS0 − K )

Cet exemple n’est qu’un rappel; le cours ISFA « Modèles financiers et analyse de risque
dynamique en assurance » fournit une présentation exhaustive du sujet.
49
« Risk neutral » valuation : why does it work (2)

“Risk neutral” :
price(call) = e−rt EQ [CFt ]
 S0 − S down up  S0 − S1down 
S0 = e E [S ] = e
− rt Q − rt
( qS up
+ (1 − q ) Sdown
) price(call) = e  up down (S1 − K ) + 01 − up down 
−rt 1

1 1
 S1 − S1  S1 − S 
 1 
S0 e rt − S1down = qS1up − qS1down
−rt
 S0 − S1down up 
S0 e rt − S1down price(call) = e  up down (S1 − K )
q=
S1up − S1down  S1 − S1 

price(call) = q e−rt (S1up − K )

ert − d
If up
S =uS0 S
1
down
1 = dS0 q=
u −d
price(call) = q e−rt (uS0 − K )

The price based on the « no arbitrage » approach and the price based on the « risk neutral » valuation approach
are identical.

Moreover, if there is no arbitrage opportunities and the market is complete, the “risk neutral probability” is
unique.

Attention : risk –neutral probabilities are relevant only for valuation, not for risk measurement (ES, VaR) (see
next chapters).
For risk measurement, physical (or “real world”) probabilities are relevant.

50
« Risk neutral valuation » : approche intuitive

• Le fait qu’une valorisation en probabilité “risque neutre” permet d’obtenir le prix des instruments financiers
peut paraître paradoxal.

• Or, la valorisation “risque neutre” des garanties du passif est un outil très important en assurance vie, et
utilisée dans le cadre de la MCEV et de la construction du bilan SII (voir pages suivantes).

•Dès lors, il est important de comprendre intuitivement l’approche afin de pouvoir communiquer au sein de
l’entreprise et auprès des décideurs.

Les agents économiques sont averses au risque, pourquoi la valorisation sous la probabilité « risque
neutre » permet-elle d’obtenir le prix de marché?

-> Nous valorisons l’option financière en fonction du prix du sous-jacent (c’est-à-dire en termes « relatifs »).
Par conséquent, l’aversion au risque des agents n’a pas d’importance car elle est déjà intégrée au prix de marché du
sous-jacent.

51
 The well-known « nested simulations » issue

In contrast to MCEV which deals with the valuation at t=0, the solvency requirements under Solvency II
are based on the distribution at t=1 (unless a standard formula is used; in this case the VaR of the
distribution at t=1 is approximated through standardized shocks at t=0) .

Valuation concept (e.g. MCEV) Solvency requirement : based on

the distribution at t=1

T= 0 T= 1 T= 2 ….

T= 0 T= 1 T= 2 ….

Value= probability weighted

and discounted average of
future CF

Under risk-neutral m simulations n simulations

probability
Under physical Under risk-neutral probability
probability

52
Recall : valuation of liability guarantees : ESG 's

In Life business, the complex nature of the liabilities require a stochastic valuation approach.

Therefore, an ESG (economic scenario generator) is needed. For the purpose of this lecture, we recall
some basic principles .

Overall, the stochastic processes may be inspired by the economic theory, statistical approaches or a
mixture.

Example of ESG architecture –example from [11]:

Voir les liens avec les cours ISFA dans
“Bibliographie”.

53
Market consistent ESG’s: equity

Voir les liens avec les cours ISFA dans

“Bibliographie”.
Equity returns (see [3], [10])

Loosely speaking, a Markov process is a process the conditional probability distribution of future states of
the process depends only upon the present state,

In other words, the prices change randomly and independently from the situation at t-1, t-2, because the
exogenous data (e.g. news about the economic perspectives) are flowing randomly and are immediately
taken into account by all investors.

dS t / S t = µ dt + σ dW t

Lognormal returns (consequence of the geometric Brownian motion assumption) are consistent with this
idea.
[(
ln( S T / S 0 ) = φ µ − σ 2
)
/2 T , σ T ]

However, there is statistical evidence that equity returns are not always lognormally distributed with
constant volatility (example : extreme movements). Consequently, other models have been developed.

54
Market consistent ESG’s: equity (2)

Introduce stochastic volatility

The basic idea is that markets are exposed to different degree of uncertainty over time. Periods with high
uncertainty are leading to high volatility (in our simplistic example, investors are “over reacting “ to any
good or bad news ) and vice versa.

• GARCH

•Regime-switching (Hardy): eg S (t + 1) / S t is switching between 2 “regimes” of volatility with

transitional probabilities p:
p(1,2)

2
LogN (µ1, σ21 )(regime 1) LogN (µ 2, σ 2)(regime 2)
p(2,1)

•Heston

•Merton

etc…

It is important to keep in mind that for risk measurement purposes, the model quality regarding the
tails is key, while for valuation purposes a slightly lower quality might be acceptable.

55
Market consistent ESG’s: Interest rates (1)

Difference between interest rate and equities stochastic processes

•Because of the basic features of instruments traded (bonds), geometric brownian motion is not
appropriate for interest rate modeling.

• In theory, nominal interest* rates are expected to show “mean reversion” at least because :
• On the long run, interest rate cannot increase boundlessly, because very high interest rates will
jeopardize economic activity, resulting ultimately in a decrease of the interest rates.

• Despite some phenomena on the very short term, interest rates can not
Black swan
! undergoing ! decrease below 0 because the economic agents have the option to
withdraw the investment in convert it into paper money.
• On the one hand, nominal interest are linked to the real interest rates (because the central bank tend to
control the inflation). On the other hand, the long term Real interest rate depends on the economic growth
(and depends ultimately on the technical progress and demographic growth which are limited).

•This theoretical concept is supported to some extent by statistical evidence.

•In addition, interest rates have a term structure.

•Given the insurers’ typical strategic asset allocation, interest rate modeling is crucial.

*) Basically Nominal interest rates= real interest rates + inflation 56

Market consistent ESG’s: Interest rates (2)
Interest rates : stochastic processes (an overview)

Short rate models Forward rate LIBOR and

models swap market:

Equilibrium -type « No arbitrage » -type

Ho Lee:
dr t = a ( b − r t ) dt + σ ( r t ) γ dW t
dr t = θ ( t ) dt + σ t dW t Heath-Jarrow- Voir cours ISFA.
Mean reversion to b with « speed of Morton
adjustment» a. (voir cours ISFA)
Hull White :
If γ=0.5 -> Cox –Ingersoll Ross, if
γ=0 -> Vasicek dr ( t ) = [θ ( t ) − a ( t ) r ( t ) ]dt + σ ( t ) dW t

θ and α constant – the Vasicek model

θ has t dependence – the Hull-White model
θ and α also time-dependent – the extended
Vasicek model
Pro’s : simple and reflecting basic
features Pro’s : easy to match the term structure
Con’s : difficulties to match with (actually it is θ(t) : an input of the model).
the observed term structure Therefore appropriate for market
consistent valuation.
Other issues (eg Hull- White takes into account mean
negative interest rate reverting
with Vasicek possible). Con’s : In general, no closed formulas …

*)Basically Nominal intrest rates= real interest rates + inflation 57

Zoom to « no arbitrage models » (1)
« Flattening »

•Basic idea : instead of modeling short

rates, start with the initial curve term
structure and model the deviations.

•Historical examples of such deviations :

«Steepening »

«Inverted »
« Flatening »

Source : Bloomberg

58
Zoom to « no arbitrage models » (2)

•The extended 2 factor Black –Karasinski is widely used in MCEV field :used to model nominal interest rates for both
Market Consistent and Real World calibrations. Provides an exact fit to the initial yield curve and short term interest rates
are lognormally distributed, therefore the rates are positive.

•The number of factors reflects the number of noise sources. It allows the model to capture more complex behavior of the
interest rates.

short rate d ln(r (t )) =α1 (ln(m(t ) −ln(r (t )))dt +σ1dW1 W1 et W2 are 2 independent Brownian motions
long rate d ln(m(t )) =α1 (µ −ln(m(t )))dt +σ 2dW2

It can be shown that the short rate, conditionally on t=0, has a lognormal distribution. Therefore the short rate is always
positive.
Black swan
! undergoing !

Drawbacks : No closed formulas for 0-coupons bonds. For pricing a binomial tree approach is used.

Advantages : Can be used not only for valuation, but also for real-world projection. Asymmetry captured.

•Note that some companies are switching to the Libor model.

59
Zoom to inflation modelling

• The inflation can be derived from the difference between nominal and real interest rates

I (t +1) = I (t )ernominal −0.5( rreal (t+1)−rreal (t ))

• Real interest rates could be modeled though a 2 factor Vasicek for example :

( )
dr1 (t ) =α1 r2 (t ) − r1 (t )) dt +σ1dW1
are 2 independent Brownian motions
dr2 (t ) =α2 (µ − r2 (t )))dt +σ 2dW2

Drawback : note that with this process negative short rate could occur.

Advantages : analytical solution for the price of 0-coupon bond.

60
Modeling of interest rates : some practical problems

•Even though the issue with the negative rates is specific to the Vasicek model, most of the interest
rates models can lead to scenarios with extremely high interest rates.

Example :
Extreme interest rates might leads to numerical issues in the ALM projection engines due to the very low discounting
factor.

exp(-100%*40 years)= 0.00000000000000000004 and exp(-1000%*40 years) is considered as 0 -> numerical issues
in the code.

•Practical solution for valuation (easy but no necessarily market consistent) : cap the rates to some
extreme but still plausible rate (eg 50% or 100% per year).

61
Modeling of interest rates : calibration of the volatility

•The implicit volatility is derived from the observed market prices for swaptions

Example as of 31.12.12
(extract from
the MCEV report of an
insurance company).

62
Spreads

•Key question Pour plus de détails, se référer aux

rappels présentés en cours

Voir les liens avec les cours ISFA dans

•Spread risk, Migration risk, Default Risk
“Bibliographie”.

63
Correlations

Pour plus de détails, se référer aux

•Key question rappels présentés en cours

Voir les liens avec les cours ISFA dans

“Bibliographie”.
•Modeling tool : copulas , but difficult to calibrate.

•Often a Gaussian copula is used but may not be appropriate….

64
Real estate
Pour plus de détails, se référer aux
rappels présentés en cours
• Usually, a Brownian motion is assumed

by default.
Voir les liens avec les cours ISFA dans
“Bibliographie”.

• Issues with representing long cycles.

• Issues with data, since the markets are not liquid and not comparable.

65
Life insurance : valuation through closed formula

In the late nineties, the academic research focused on the valuation of life insurance liabilities
through closed formulae.

Thisapproach require many simplifications, but gives insight into the key life insurance
mechanisms.

For educational reasons, we will briefly discuss one of these papers. Other papers developed
further this idea in the nineties ([18], [19], [20] )

At present, insurers apply Monte Carlo techniques (see next chapters).

The Monte Carlo techniques are also used for MCEV (Market Consistent Embedded Value)
computations, the framework which replaced around 2005-2008 the TEV (Traditional Embedded
Value).

66
 Some models with closed formulae : an example
Briys and de Varenne (1997) :
Balance sheet at t=0 Liabilities :
Assets Liabilities and Equity L* : minimum guaranteed liability
Assets A0 Equity E0 = (1−α ) A0 δ : participation coefficient
Liabilities L0 = α A0 α : proportion of Cash flows paid at time T (in fact the model is single
initial assets period).
Total A0 A0 financed by equity

•Shareholder’s payoff : -Insolvent company ET = 0 si AT < L*T

-Only the guaranteed rate is paid L*T

ET = AT − L*T si L*T ≤ AT <
α
-Profit sharing is higher than the L*T
guaranteed rate ET = (1−δα ) AT − (1−δ ) L*T si ≤ AT
α

The equity and the liability are viewed as contingent claims -> can be solved using option pricing :
 L*   L* 
ET = max[0, AT − L*T ]−δα max 0, AT − T  LT = L*T P(t ,T ) − Put E ( AT , L*T ) +δαC E  At , T 
 α
 α 
European Call Option European Call Option The SH
Risk free part Profit sharing option
maturity T maturity T can default
*
Strike L*T Strike LT (option)
α
Interpretation :
shareholders have a
limited liability in case of
Interpretation
:Potential profit sharing
default 67
Illustration with a simplified « saving » product

Based on the same idea, let us a very simple product :

A « saving » product (simplified view of a GMAB-like product, lapses not allowed, ..)
 Profit sharing rate=85%, Minimum guaranteed rate=2.5%, Premium=1 000€, Maturity=5 years, risk free rate 5Y=3%

 Asset allocation : 20 % equity, 80% (zero-coupons) « risk free » bonds, maturity 5 years

 Pay-off t=5
5
(
Pay − offt =5 = Premium(1+ 2.5%) + Premiummax 85%Value_assetst =5 − (1+ 2.5%)5 ;0 )

Pay − off t =5 = Premium (1 + 2.5% )

5
+ (
Premium max 85%Value_a ssets t =5 − (1 + 2.5%) 5 ;0 )

⇒ ZCB maturity 5 Y, nominal 1 ( ( (

Premium max 85% 20%St =5 + 80% 1 + riskfree _ rate5Y ) )− (1 + 2.5%) ;0)
5 5

w1 = Premium (1 + 2.5% )
5
⇒

85% 25% Premium max St =5 −
(
(1 + 2.5%)5 − 85% 80% 1 + riskfree _ rate5Y
5
)
;0 
 85% 20% 
 

It is a call with maturity 5 Y, strike =2.14, underlying asset 100% equity

w 2 = 85% 20% Premium

68
 « Path dependency » and the role of the accounting
rules

In fact, the product described above is not the French « Euro » saving product.

The « profit sharing » rules (règle de « PB mini » ou « contractuelle ») introduces a « path dependency ».

Mathematical reserves (if neglecting the loadings) of a product with a 2.5% minimum guaranteed rate (MGR):

MR0 = 1 000€ MRt-1 MRt = MRt-1*(1+max(85%*acc_returnt, 2.5%))

t=0

At the maturity date, and if neglecting the loadings, the PH will receive the [Link] at the end of 5Y:

5
MR 5 = MR 0 ∏ (1 + max(85%
t =1
* accounting return t ,2.5%))

Each year, the policyholder benefits from an option on the « accounting return » of the invested assets.

The « accounting return » is not identical to the economic return and increases the complexity of the model.

Overall, the role of the French GAAP is significant because the profit sharing rules (regulatory or contractual) are based
on the “accounting” French GAAP return.

The valuation of this instrument is not possible through closed formulas in all cases…

69
Liability valuation : TEV (Traditional Embedded Value)
approach
•EV measures the value that the shareholders own in an insurance enterprise (ie “consolidated value of the
shareholders’ interests in the covered”).
•From the late nineties to 2004/2005, what we call currently the TEV approach (called during this period simply EV)
was widely used by Life insurers.
Main characteristics :
• New Business is not considered
•“Top-down “ approach : 1 deterministic projection; the cash flows are discounted using interests rates
above the risk-free rate
•An explicit “Cost of capital” component

CoC TEV=
PVFP (cost of capital)
(present value +ANAV
of future profit TEV
(Traditional +PVFP
Embedded value)
ANAV
(adjusted net -CoC
asset value)

Value in force (VIF)

70
 TEV components : ANAV

Generic Economic Balance sheet :

1 SH Equity (BV) + 1
Asset
(Book value) Actuarial
reserves + 2 3 4
«Hidden »
3 actuarial reserves
-Potential profit sharing of 2) 3) 4)
2 URGL’s
4 Other « hidden » res. -Elimination of items included in PVFP
-Taxes
=ANAV

* CFO Forum documents

71
TEV components : PVFP
• Valuation based on 1 unique deterministic projection, discounted using an interest rat above the risk free rate :

rTEV = rriskfree + beta * riskpremium

• The lapse rates are deterministic and do not depend on the market conditions.
• Often, the Asset-Liability interactions are modeled in a simplistic way.
• No explicit assessment of the Cost of Financial options& Guaranties and the Cost of Non Headgeable Risks.
(mortality, longevity, lapses, etc.., see next page). These cost ares assessed implicitly through the discounting.

Cash flows (CF’s) =

Discounting using rTEV + Premium
CF1
PVFP= CF2 +Interest income
CF3
…. –Benefits
–Increase [Link]
–Alloc. to participation
1 projection of the liability cash-flows
–Tax

n
CFi
PVFP = ∑
i =1 (1 + rTEV )

72
TEV components : CoC

•CoC represent the cost of holding a capital.

Difference between the RDR (risk discount rate) and the

return of the invested RqC= cost
n RqCi −1 * (rTEV − rriskfree * (1 − tax))
CoC = ∑ Sometimes, instead of the risk free rate, the expected
i =1 (1 + rTEV )i investment return of assets backing the capital is used.

where RqC is the amount of required capital assumed at time i.

•Usually, the Solvency I required capital was used as a basis, or alternatively 120 % or 150 % of this
amount.

73
TEV : main disadvantages

RDR (Risk discount rate)

• The risk premium of the RDR represents a compensation for the uncertainty of the future cash flows. Although the
concept is theoretically relevant, it raises practical challenges.
• Obviously, the risk premium depends on the underlying risk. As a consequence, it should depend on the company, on
the portfolio, even on the LOB (line of business).
• During the TEV-era (from the late nineties to 2005-2008), Life insurers released TEV reports w/o really answering the
question how exactly the RDR was derived. In fact, the rate was often based on expert judgment and on the market
practice justified by the observed equity premiums.
•In general, the RDR was only adjusted from one year to another for the changes of the risk free rate.

Projected investment return

•Under TEV, increasing the risk of the asset allocation is leading to a higher expected return and higher TEV, because
usually the RDR is not adjusted.
•As a consequence, the underlying valuation of the liabilities is not necessarily “market consistent”.

Cost of Financial and Non Financial Guarantees

•If the guarantees are not activated within the deterministic projection, the cost is not explicitly captured . The RDR
implicitly captures the cost, but in an inaccurate way (see next slide).

74
Valuation in Life Business: why a stochastic approach is
needed ?

• As already highlighted, life insurers offer various guarantees to the policyholders.

• Most of these guarantees can be analyzed using the tool of the financial theory. These guarantees are asymmetric :
• the policyholder has the right but not he obligation to surrender or to switch between Euro and UL;
• he is benefiting from the profit sharing mechanism, if the reinvestment return increases, and form the
Minimum Guaranteed rat (MGR) in case of market shock.

• The TEV does not capture adequately the cost, as illustrated below :

Vision TEV (illustrative)

Vision réaliste : nécessité d’un modèle
stochastique
En pratique, les Pertes potentielles
O&G n’ont aucune pour la compagnie.
incidence sur la TEV Rachats potentiels.
(ne se “déclenchent” Baisse de la PVFP.
Investment return jamais) .
=4.5%

Profit sharing (90%)=

=3.8%

TMG=3%

Time

75
“Market consistent liability valuation” (1)

•The market consistent valuation techniques in general, and especially the Monte Carlo computational
methods, are relatively recent in the insurance sector.

•Although many decision makers are aware of the principles, it is worth discussing some potential
misunderstandings/ erroneous interpretations (for the further details, see [15])

• The MCEV answers one specific question : « From sharehoders’ point of view, what is the value of the
company, including the cost financial options & guarantees on the liability side, given the market prices of
the financial instruments observed.».
•In particular, MCEV does not treat the question of the required economic capital.

• The MCEV approach does not necessarily imply Monte Carlo simulations. As seen previously, for some
guarantees a closed formula approach is possible( eg GMAB).

76
“Market consistent liability valuation” (2)

The“risk neutral valuation” implies that the average asset yield of the stochastic scenarios is the risk-free rate.
Sometimes, it is suggested that the guarantees (eg. minimum guaranteed rates) have to be “adjusted” (ie lowered).

In fact, the only difference between “real world” and “risk-neutral” valuation is the probability of the different scenarios.
The guarantees are identical in the “real” as in the “risk neutral” world (similarly, we do not adjust the strike of the B&S
formula).
A “risk neutral scenario” concept does not exist, only the scenario probabilities are “risk neutral”.

It was also argued that MCEV is not consistent with the idea of “mean reverting” diffusion process of the assets.
In reality, the existence of a “mean-reverting” phenomenon does not have any impact on the valuation of the liabilities
(in contrast with the capital requirements).

One of the key characteristics of a market-consistent valuation is that it reproduces market prices. So, in valuing an
equity or equity option, the possible outcomes are weighted by the probability of occurrence and by a discount factor
which implicitly reflects investors’ aversion to a loss.

77
Simple illustrative example (1)

Assume a theoretical product with the following cash flow payments under a given scenario i:
n
Liability i = Math .res 0 ∏ (1 + max( MGR ,1 + 90 % I ( t ) ) MGR=minimum guaranteed rate
t =1

Assume the loadings cover exactly the internal expenses : both are neglected

We model the investment return as follows : dI ( t ) = rdt + σ dW t

Liability CEV = Math .res t = 0 (1 + 90 % r ) n /( 1 + r ) n

Theoretical liability value under CEV :

Assuming that a single premium of 100 is invested at t=0, we perform 10 000 Monte Carlo simulations with r=3%,
sigma =4%, n=20.

 10000 
Liability MCEV =  ∑ Liability i  /10000
 i =1 

78
Simple illustrative example (2)

Sigma =4% with 90 % profit sharing Sigma =3% with 90 % profit sharing Sigma =3% w/o profit sharing

MGR CEV TVOG MCEV MGR CEV TVOG MCEV MGR CEV TVOG MCEV

0,0% 5,7 -8,9 -3,2 0,0% 5,7 -4,2 1,4 0,0% 44,6 0,0 44,6
1,0% 5,7 -14,6 -9,0 1,0% 5,7 -8,4 -2,8 1,0% 32,4 0,0 32,5
2,0% 5,7 -22,7 -17,0 2,0% 5,7 -15,1 -9,5 2,0% 17,7 0,0 17,7
3,0% 0,0 -28,1 -28,1 3,0% 0,0 -19,8 -19,8 3,0% 0,0 0,0 0,0
4,0% -21,3 -22,3 -43,6 4,0% -21,3 -13,5 -34,8
4,0% -21,3 0,0 -21,3

Some remarks :

These simple tests show why a stochastic risk neutral valuation is needed : under CEV, the MGR cost is nil unless
r>MGR.

A 0% guarantee has also a TVOG

The TVOG arises from the asymmetrical combination of the Minimum guaranteed rate and the Profit-sharing
mechanism.

The higher the asset volatility, the larger the TVOG effect is. It is important to understand that this occurs because we
are performing a valuation of guarantees.

Naturally,this mechanism does not imply the company should not invest in risky assets, but only that we cannot take
into account any future performance above the risk free rate before its actual materialization.

79
“Market Consistent Embedded Value “ : standard
introduced by the CFO Forum
• CFO Forum : founded in 2002, with representation from major European insurers.

• Members (as of Feb 2013)

• Developed an unified standard for Life insurance. There is an intermediary version called EEV (European
Embedded Value), but this presentation will focus on the latest MCEV guidelines.

• Most recent MCEV guidelines release : October 2009

80
MCEV : principles [see 16]

Principle 1: Market Consistent Embedded Value (MCEV) is a measure of the consolidated value of shareholders’ interests in the
covered business. Group Market Consistent Embedded Value (Group MCEV) is a measure of the consolidated value of shareholders’
interests in covered and non-covered business.

Principle 2: The business covered by the MCEVM should be clearly identified and disclosed.

Principle 3: MCEV represents the present value of shareholders’ interests in the earnings distributable from assets allocated to the
covered business after sufficient allowance for the aggregate risks in the covered business. The allowance for risk should be calibrated
to match the market price for risk where reliably observable. The MCEV consists of the following components:
• Free surplus allocated to the covered business
• Required capital; and
• Value of in-force covered business (VIF).
The value of future new business is excluded from the MCEV.

Principle 4: The free surplus is the market value of any assets allocated to, but not required
to support, the in-force covered business at the valuation date.

Principle 5: Required capital is the market value of assets, attributed to the covered business over and above that required to back
liabilities for covered business, whose distribution to shareholders is restricted.

81
MCEV : principles (2)

Principle 6: The value of in-force covered business (VIF) consists of the following components:
Present value of future profits (where profits are post taxation shareholder cash flows from the in-force
covered business and the assets backing the associated liabilities) (PVFP)
Time value of financial options and guarantees as defined in Principle 7
Frictional costs of required capital as defined in Principle 8
Cost of residual non hedgeable risks as defined in Principle 9

Principle 7: Allowance must be made in the MCEV for the potential impact on future shareholder cash flows of all financial options
and guarantees within the in-force covered business. The allowance for the time value of financial options and guarantees must be
based on stochastic techniques using methods and assumptions consistent with the underlying embedded value. All projected cash
flows should be valued using economic assumptions such that they are valued in line with the price of similar cash flows that are
traded in the capital markets.

Principle 8: An allowance should be made for the frictional costs of required capital for covered business. The allowance is
independent of the allowance for non hedgeable risks.

Principle 9: An allowance should be made for the cost of non hedgeable risks not already allowed for in the time value of options and
guarantees or the PVFP. This allowance should include the impact of non hedgeable non financial risks and non hedgeable financial
[Link] appropriate method of determining the allowance for the cost of residual non hedgeable risks should be applied and
sufficient disclosures provided to enable a comparison to a cost of capital methodology.

Principle 10: New business is defined as that arising from the sale of new contracts and in some cases increases to existing contracts
during the reporting period. The value of new business includes the value of expected renewals on those new contracts and expected
future contractual alterations to those new contracts.
The MCEV should only reflect in-force business, which excludes future new business. The value of new business should reflect the
additional value to shareholders created through the activity of writing new business.

82
MCEV : principles (3)

Principle 11: The assessment of appropriate assumptions for future experience should have regard to past, current and expected
future experience and to any other relevant data. The assumptions should be best estimate and entity specific rather than being based
on the assumptions a market participant would use. Changes in future experience should be allowed for in the VIF when sufficient
evidence exists. The assumptions should be actively reviewed.

Principle 12: Economic assumptions must be internally consistent and should be determined such that projected cash flows are valued
in line with the prices of similar cash flows that are traded on the capital market. No smoothing of market or account balance values
or unrealised gains is permitted.

Principle 13: VIF should be discounted using discount rates consistent with those that would be used to value such cash flows in the
capital markets.

Principle 14: The reference rate is a proxy for a risk free rate appropriate to the currency, term and liquidity of the liability cash
flows.
Where the liabilities are liquid the reference rate should, wherever possible, be the swap yield curve
appropriate to the currency of the cash flows.
Where the liabilities are not liquid the reference rate should be the swap yield curve with the inclusion of a
liquidity premium, where appropriate.

Principle 15: Stochastic models and the associated parameters should be appropriate for the covered business being valued, internally
consistent and, where appropriate, based on the most recent market data. Volatility assumptions should, wherever possible, be based
on those implied from derivative prices rather than the historical observed volatilities of the underlying instruments.

83
MCEV : principles (4)

Principle 16: For participating business the method must make assumptions about future bonus rates and the determination of profit
allocation between policyholders and shareholders. These assumptions should be made on a basis consistent with the projection
assumptions, established company practice and local market practice.

Principle 17: MCEV results should be disclosed at consolidated group level using a business classification consistent with the primary
statements, with clear description of what business is covered by MCEVM and what is not. Except where they are not considered
material, compliance with the MCEV Principles is compulsory and should be explicitly disclosed.

84
MCEV : overall perception, financial and sovereign debt
crisis 2008-2012

• In 2006-2007, the financial community welcomed the MCEV was by and considered as a progress in
understanding the life insurance business model and the value creation.

•In December 2008, the CFO forum published a press release supporting a review of the interest rate
curve and implicit volatilities used because of the turbulent financial markets.

•MCEV as of 31.12.08 released by the insurance companies were hardly comparable. Many companies
introduced a liquidity premium and averaged (smoothed) implicit volatilities based on different
methodologies.

• In December 2009, the CFO changed the Principle N14, allowing the use of a liquidity premium.

•In addition, there are on -going academic debates on the nature and the rationale of the liquidity
premium. Most of the “pro’s and con’s” presented are essentially similar to the arguments regarding the
Solvency II CCP (countercyclical premium).
•See [XX].

85
 Market Consistent Embedded value (MCEV) :
components

1 unique Frictional costs

deterministic (Certainty
(FC)
Equivalent
projection Value))
Time value of options
and guarantees
(TVOG)

Net asset Cost of non

Fonds hedgeable risks
Value (CNHR)
propres nets
Others

MCEV

«CNHR»
« Frictional costs » Market consistent approaches value the
Has a different meaning from the “financial” O&G (MGR, profit-sharing…).
CoC under TEV. « TVOG » Other guarantees are « non hedgeable » (e.g.
« FC » is only the tax effect on the Corresponds to the difference mortality, longevity, deviation of the expense
investment return of assets between: ratio).
backing the Shareholder equity  Certainty Equivalent Value The cost of theses guarantees is estimated
and the investment fees. In a (deterministic scenario) through a CoC approach.
world w/o taxes and investment
 PVFP
costs, the effect is nil.

Le passage consiste à supprimer la mesure simpliste du coût du risque dans la TEV (actualisation à un taux > taux sans
risque, Cost of capital), pour la remplacer par une mesure plus fine du coût des O&G financières (TVOG), tout en tenant
compte des risques démographiques et des coûts résiduels (FC).
86
 Focus on Life business

The Life Best estimate requires usually a 2 Characteristics of the products (loadings,
complex valuation model even for commissions, financial guarantees,
Financial
companies applying the standard guaranteed rates, insurance risks).
guarantees, profit
formula. sharing etc..
3 8
7
1 6
5
4
3
2
Economic Cost of 1

conditions Asset 0
O&G Interactions modeled
1 2 3 4 5 6 7 8 9 10
allocation
Valuation and risk profile
assessed through
Lapses, stochastic scenarios
« arbitrage » (switch
Euro/UL, conversion
into annuity…)
Expenses Biometric risks (mortality,
5 Strategic Asset longevity)
allocation 4

Policyholder’s
behavior
The management acts to various extent on 1) 2) 3)
This general scheme is valid for many life saving or 4) and 5) with a ”feedback loops”.
retirement products on major European markets.
The asset allocation is one of these decision levers.

As a consequence, the risk-return balance of a given asset class cannot be fully assessed w/o taking
into account the overall balance sheet strength and the interdependencies between assets and
liabilities.
87
MCEV : Value of New Business (1)

 Although not directly related to ALM,( but important for the ORSA process and the steering of the company, see
next chapter), the Value of New Business is an import concept within the MCEV framework.

Value of New Business (VnB) is defined as follows [ see]:

Principle 10: New business is defined as that arising from the sale of new contracts and in some cases increases to
existing contracts during the reporting period.

The value of new business includes the value of expected renewals on those new contracts and expected future
contractual alterations to those new contracts.

The MCEV should only reflect in-force business, which excludes future new business. The value of new business
should reflect the additional value to shareholders created through the activity of writing new business.

The Appraisal value is defined as :

Appraisal value = MCEV + multiple (eg 3) x VnB

The appraisal value is used in corporate finance and M&A deals, the multiple depends on the perception of the
potential for future New Business growth.

88
MCEV : Appraisal value

During the crisis, the implicit multiples were low, the market capitalizations were even below the MCEV (as of
31.12.10, Deloitte study, see [17] :

89
VnB : 2 different theoretical views

Answers to different questions :

« Stand-alone » approach : the value of the NB as if the contracts are underwritten by a fictive company
with an « empty » balancesheet but with indicators (eg expense rates) are derived from the « going concern »
principle.

“Marginal” approach : the change in MCEV induced by the underwriting of the New Business. In practice, it
is the difference between 1) a valuation of the stock w/o NB of the year and 2) a valuation with NB.

Essentially
acquisition « Stand alone » VnB
expenses

Assets Liab. NB

•Only investment of the NB, no or limited URGLS’

(depending on the investment date and the
approach (“point of sale” or “end of period”)
•All buffers at nil (exemple Fonds de PB, Réserve
de capitalisation)

90
VnB methodologies : pro’s and con’s

“Stand alone” “Marginal”

• Intuitive • Usually consistent with the decision

Avantages taking process (« marginally » regarding
• Relatively simple from a given context).
operational point of view (no cross
effects between In force business • Natural component of the MCEV
and New Business) Movement analysis (also valid for SII
own funds).

• Might be inappropriate regarding • Less intuitive, the analysis of the

Drawbacks the decision making within a given changes for marginal VnB is
context (the company’s decisions are challenging.
usually « marginal »).
• Insights into the potential cross- effects
•Does not fit the New business step between stock and New Business
in the Movement analysis : an
additional bridge is needed

91
ANNEXES
 Bibliographie

[1] Briys E, de Varenne F, (1998) On the Risk of Life Insurance Liabilities: Debunking Some Common Pitfalls, The
Wharton Financial Institutions Center

[2] Crouhy,M., Galai, D., Mark, R., The essentials of Risk management, McGraw-Hill

[3] Jorion P (2009) , Financial Risk Manager Handbook, 5th Edition, Wiley

[4] Casulalty Actuarial Society (2003) Overview of Enterprise Risk Management

[5] Frigo M, Anderson J (2011), ERM : Practical Approaches for Getting Started, COSO

[6] Godfarb R (2007), ERM practices and Rating agencies, Contingencies Sep/Oct 2007

[7] International Association Of Insurance Supervisors (2006), Standard On Asset-liability Management

[8] Deloitte (2011), Asset & Liability Management for Insurance Companies,

[9] Piermay P, Mathoulin P, Cohen A « Gestion Actif-Passif d’une compagnie d’assurance ou d’un investisseur

institutionnel »,

[10] Hull J (2011) Options, Futures and Other Derivatives: Global Edition, Pearson

[11] Insurance Economic Scenario Generator Modelling Suite , Barry and Hibbert, 2008

[12] Derman, E. Model Risk, Quantitative Strategies Research Notes, Godman Sachs, April 1996

[13] Marshall, C., Siegel, M., 1997, Value-at-Risk: Implementing a risk measurement standard, Journal of Derivatives 4,

[14] Vision on Own Risk and Solvency Assessment (ORSA) : Good Practice, ORSA Working Group of the Dutch
Association of Insurers (“Verbond van Verzekeraars”) , February 2012, release 2.0

93
 Bibliographie (2)

[15] Tower Perrin Tillinghast ,”Market-consistent embedded value : dispelling the myths”, 2005,

[16] CFO Forum “Market Consistent Embedded Value”,Principles, 2009

[17] Deloitte , “MCEV : a turning point”, 2011

[18] Generali presentation, “Principles and models for the Embedded Value calculation”, 2012, Trieste February 2012

[19] Grosen A, Jørgensen P L, (2000) Fair valuation of life insurance liabilities: The impact of interest rate guarantees,
surrender options, and bonus policies, Insurance: Mathematics and Economics 26 (2000) 37–57

[20] Bacinello A.R (2001) Fair pricing of life insurance participating policies with a minimum interest rate guaranteed,
ASTIN Bulletin,

[21] Denuit M., Robert C., Actuariat des assurances de personnes, Economica, 2007

94
Bibliographie : lien avec les autres cours ISFA

Cours ISFA approfondissant certains sujets mentionnés:

[Link] « Modèles financiers et analyses de risque dynamiques en assurance », Support de cours 2012-
2013 ,support de cours

F .Planchet, A. Kamega « Construire un générateur de scénarios économiques en assurance » Version

2.3 , Octobre 2012 , support de cours

O. Nteukam « Modèles financiers en assurance : agrégation des risques financiers en assurance » ,
support de cours

F. Planchet, « Construction d'un modèle d'actifs pour une société d'assurance : approche théorique et
aspects pratiques » Version 1.3, Modélisations avancées en assurance, Décembre 2010, support de cours

F. Planchet, « Solvabilité et valeurs extrêmes : Enjeux en termes de modélisation » Version 2.2 , Janvier
2013, support de cours

P. Thérond, "Introduction aux normes IFRS et à Solvabilité 2" ,cours ISFA

95
Bibliographie : références complémentaires

Loisel, S., Milhaud X.,” From deterministic to stochastic surrender risk models: impact of correlation crises on
economic capital” July 7, 2010

Planchet F, Guibert Q , Juillard M (2012) Measuring uncertainty of solvency coverage ratio in ORSA for non-life
insurance, European Actuarial Journal , vol.2. issue 2

Consultation Paper On the Proposal for Guidelines on Own Risk and Solvency Assessment, EIOPA, October 2011

Final Report on Public Consultation No. 11/008 On the Proposal for Guidelines on ORSA, EIOPA, 9 July 2012

96