Captives
and Solvency ii
Captive
Insurance and Captive Reinsurance Companies after the Solvency ii
Directive
from the Solvency ii
Association, the largest Association of Solvency ii Professionals
in the world
PROPORTIONALITY - LEVEL 2
MEASURES
CEIOPS-CP-45/09, 2 July 2009,
Consultation Paper No. 45
Draft CEIOPS’
Advice for Level 2 Implementing Measures on Solvency II: Technical
Provisions – Article 85 h, Simplified
methods and techniques to calculate technical
provisions
Annex A: Gross-to-net techniques: the
story so far
A.1 This annex contains an analysis of the
gross-to-net techniques (“proxies”) developed in the Report on
Proxies elaborated by CEIOPS/Groupe Consultatif Coordination
Group as well as the gross-to-net techniques which were tested
(based on the recommendations contained in this report) in the
QIS4 exercise.
A.2 This description of gross-to-net techniques
has been included purely for informational purposes; it is
intended to provide an overview on the range and technical
specificities of such methods developed so far.
CEIOPS considers
that further technical work is necessary before it could
be decided whether it would be appropriate to include (some of)
these techniques in Level 2.
A.1 The Report on Proxies
A.3
Issues related to the use of Gross-to-Net proxies
are discussed in
some detail in the “Report on Proxies” elaborated by CEIOPS/Groupe Consultatif Coordination Group.
At the outset
Gross-to-Net proxies are defined as proxies that “transform a
gross of reinsurance estimate into a net estimate” and as such
used in combination with proxies for stipulating the technical
provisions gross of reinsurance.
Moreover, it is stated in
the report that “Gross-to-net proxies are used to convert best
estimates of claims or premium provisions into best estimates net
of reinsurance, in cases where there is not enough (technically
feasible) data to directly derive net estimates.”
A.4 The
report on proxies contains a list of 10-12 Gross-to-Net proxies
that have been considered by the national proxy expert groups.
A majority of the considered Gross-to-Net proxies is based on
accounting data (in a broad sense), including:
(1) Historic
accounting figures.
(2) Gross and net cumulated cash-flows
(paid claims) per accident (or underwriting) year.
(3) Gross
and net provisions for reported but not settled (RBNS)
claims (also referred to as case reserves) per accident (or
underwriting) year.
A.5 The considered proxies based on
accounting figures include also combinations of case (2) and (3),
e.g. proxies where cumulated paid claims and RBNS-provisions,
both gross and net of reinsurance, are applied when calculating
the IBNR-provisions net of reinsurance.
A.6 The application
criteria for the various Gross-to-Net proxies referred to in the
report are not always explained. However, some comments
regarding these criteria may be given:
• With respect to the
Gross-to-Net proxies based on historic accounting figures only it
is in general not possible to distinguish between the individual
accident years.
However, it may be possible to
distinguish between insurance classes or lines of business
(depending on e.g. the
reporting requirements in force).
•
Therefore, in order for these proxies to lead to reasonable results,
it would be necessary to assume that the reinsurance programme
and probably also the composition of the portfolio is stable over
time.
• On the other hand,
Gross-to-Net proxies based on
cumulated claims payments or provisions for RBNS claims (or both)
can be stipulated for individual lines of business as well as for
individual accident years (for a given line of business).
In these cases it is not
necessary to presuppose that the reinsurance programme is stable
over time.
• The considered
Gross-to-Net proxies are first and foremost designed for
calculating provisions for claims outstanding (“post claims”) net
of reinsurance – whether these calculations distinguish between
RBNS provisions and IBNR-provisions or not.
However, some of the
considered proxies may be used when calculating the
premium provisions (“pre claims”) net
of reinsurance, although the degree of accuracy/precision may be
less in these cases, cf. also the alternative proxies (i.e. case
(4) and (5) referred to below) and sub-section A.2.
• It is tacitly assumed that the accounting
figures referred to in cases (1)–(3) above are
undertaking-specific and as such must be available for
undertakings that want to apply these Gross-to-Net
proxies.
However, it should be possible to use market data (e.g.
risk statistics for the overall market) – in combination with
some basic
characteristics of (simplified) reinsurance treaties –
in order to establish Gross-to-Net proxies for individual lines
of business and individual accident years (for a given line of
business), cf. e.g. case (5) referred to below.
A.7 The list
of Gross-to-Net proxies referred to in the report on
proxies comprises also two proxies that go beyond the application
of accounting data:
(4) The first of these alternative proxies
applies the premium model for the line of business in question
(based on e.g. separate estimation of claim frequencies and claim
severities) in order to derive the percentage of the expected
claims costs being reinsured and uses this information as a basis
for stipulating the Gross-to-Net proxy.
(5) The other
alternative proxy is using available market data (per line of
business) regarding the (empirical) distribution of single
claim amounts to establish ratios between:
i. the expected
value of a (random) single claim net of reinsurance and
ii.
the expected value of a (random) single claim gross
of reinsurance for a prescribed set of excess points of a
simplified (pure) excessof- loss treaty.
These ratios are then
used in combination with e.g. suitable interpolation- techniques
to stipulate Gross-to-Net proxies for the
following cases:
i.
excess-of-loss covers only,
ii. combinations of proportional
reinsurance covers and excess-of- loss covers.
A.8 These
alternative Gross-to-Net proxies could be applied for the
individual lines of business as well as for the individual
accident years (for a given line of business).
A.2 The QIS4
Technical Specifications
A.9 With respect to QIS4, the report on
proxies proposed to test only two different designs of the
Gross-to-Net proxies, both of them based on accounting data (in a
broad sense):
• one based on the provisions for RBNS claims
(“case reserves”) and
• one based on cumulated cash flows (i.e.
cumulated claims payments).
These testing proposals were
incorporated into the Technical Specifications (TS) without
further changes.
A.10 This choice to narrow down the range of
Gross-to-Net techniques for the purposes of QIS4 was made in
order to keep the technical specifications sufficiently simple
and practical.
A.11 The main aspects of these testing proposals
are summarised below.
Gross-to-Net-proxy based on provisions for
RBNS-claims (“case reserves”)
A.12 This proxy uses a ratio of net
over gross provisions of an available portfolio A in order to
estimate the net provisions of another portfolio B (NPB) based on
the observable gross provisions of portfolio B (GPB).
In other
words, the Gross-to-Net proxy (GN) is stipulated as
GN =
NPA/GPA,
where NPA and GPA represents the net and gross
provisions of portfolio A, respectively.
Then this proxy is
applied to calculate the net provisions for portfolio B as
follows:
NPB = GN × GPB
A.13 However, it is not clear
from the QIS4 TS whether the purpose of this proxy is to
calculate the overall net provisions for claims outstanding
or only the net provisions for RBNS-claims
A.14 The following
criteria should be fulfilled in order to apply this proxy:
• The
benchmark portfolio (A) should be similar to the portfolio (B)
for which the proxy is used, cf. the principle of substance over
form.
• The ratio (GN) should be established by means of credible
and sustainable data.
This requires a data set exceeding at least
two years.
A.15 With respect to the
properties of this proxy the QIS4 TS state that
“ceded
reinsurance varies with the size, the financial soundness and the
risk aversion of a company, so that particular care is required
when applying a ratio of net over gross from another benchmark
portfolio.
Such an approach
should therefore only be used in cases where the
benchmark portfolio is known to have a very similar nature as the
own portfolio.
Even if this is the case, however, the cession
percentage for non-proportional reinsurance will heavily depend
on the actual occurrence of large losses, and therefore be very
volatile.”
Gross-to-Net-proxy based on cumulated paid claims
(cumulated cash flows)
A.16 This proxy derives an estimate of net
provisions for claims outstanding by using the gross provisions
for claims outstanding in combination with an estimate of the
impact of the reinsurance covers for the individual
accident years.
A.17 With respect to the rationale for using
this proxy, it is noticed that for past accident years the
reinsurance structure for an individual year is known and will
(likely) not change retroactively.
Accordingly, a comparison of
net over gross cumulated cash flows per line of business in the
past –differentiated by accident year – may be used to derive an
estimate of the impact of proportional and non-proportional
reinsurance for the individual
accident year
(i.e. a Gross-to-Net
proxy for the individual accident year).
A.18 For each line of
business the Gross-to-Net proxies for the accident years not
finally developed (GNi) are stipulated as follows:
GNi =
ANet,i,n–i/AGross,i,n–i
where AGross,i,n–i and ANet,i,n–i
represent the cumulated paid claims gross and net of reinsurance,
respectively, and n is the latest accident year with observed
values of these cash-flows.
A.19 These proxies are then used to
calculate the net provisions for claims outstanding for the
individual accident years, that is
PCONet,i = GNi
×
PCOGross,i
where PCOGross,i
and PCONet,i represent the gross
and net provisions for claims outstanding for accident year i,
respectively.
A.20 In order to apply this proxy both gross and
net cumulated paid claims (gross and net cash flows) per accident
year need to be available for each line of business.
A.21 The
QIS4 TS briefly explain some of the properties of this proxy:
•
For newer accident years and especially the last accident year
(where i=n) the stipulated proxy might be a bit too high due to
the fact that the IBNR claims are likely to constitute a large
part of the provisions for claims outstanding.
Accordingly, the
stipulated proxy is likely to lead to an overestimation of the
net provisions in these cases.
• The Gross-to-Net proxies
referred to above concern the provisions for claims outstanding.
For the premium provisions, i.e. the provisions for (covered but
not incurred) claims related to the current accident (business)
year (where i=n+1), a Gross-to-Net proxy can be stipulated by
using the (anticipated) proportional part of the reinsurance
cover for this year.
This will be a conservative approach for the
ceding (re)insurance undertaking, since the impact of the
non proportional
reinsurance for the current accident (business)
year is not taken into account.
A.3 The QIS4 Results
A.22
The use of Gross-to-Net proxies in QIS4 is summarised as follows
in CEIOPS’ QIS4-report (see the sub-section 7.2.5 on
simplifications and proxies):
“Concerning reinsurance, only
few undertakings were able to determine amounts relating to
reinsurance recoverables (or net figures) by applying actuarial
reserving techniques based on reinsured or net triangular
claims data.
Instead, many participants used triangle analysis
techniques only for the calculation of best estimates gross of
reinsurance, and derived the reinsurer’s part of gross provisions
by applying one of the two Gross-to- Net proxies.
The wide use of
Gross-to-Net proxies underlines that it is difficult for the
undertakings to get data net of reinsurance.
However, some
undertakings remarked that an application of this proxy may
lead to poor results in the case of excess loss covers, where the
risk mitigating effect of the reinsurance cover would be
underestimated.
It was also remarked that the use of both types
of Gross-to-Net proxies described in the specifications on the
same portfolio sometimes resulted in materially different
valuations.
A similar situation could be observed with regard to
the determination of premium provisions, where only a few
participants were capable of carrying out an actuarial
projection of future cash flows arising from future claim events.
…”
A.23 Some further comments are given regarding the
participating undertakings’ experience with the Gross-to-Net
proxies stipulated for QIS4-purposes (see sub-section 7.3.3 on
best estimates in non-life insurance):
“The gross-to-net
proxy was used by some undertakings as net claims data triangles
are unsuitable for immediate application of actuarial reserving
techniques since they often contain irregularities.
Undertakings
within one country commented that it is difficult to
use actuarial techniques to calculate the best estimate
reinsurance provision taking into account all contractual
details.
…
More guidance should be developed concerning the
valuation of reinsurer’s shares in technical provisions.
To avoid
over-reliance on very simple techniques such as the Gross-to-Net
Proxy, guidance on other more sophisticated
actuarial techniques
which would be better aligned with the true risk mitigating
effect of reinsurance covers should be sought.”
A.24 As a general
summary regarding the experiences from QIS4, it may be stated
that the need for Gross-to-Net proxies has been confirmed, cf.
the statement that many insurance undertakings have problems with
determining the cash flows related to reinsurance
recoverables.
A.25 On the other hand, the experience from QIS4
highlights the need to introduce clear admissibility criteria for
the use of such Gross-to-Net techniques in order to ensure that
the valuation of technical provisions net of reinsurance will
lead to consistent results across different undertakings and
markets.
Also, it seems necessary to develop actuarial guidance on
a range of techniques for determining net provisions to avoid an
overreliance on a few proxy techniques.
A.26 In this context
it should also be noticed that the problems of identifying the
cash flows related to reinsurance arrangements seem to apply to
all kinds of (non-life) insurance undertakings (i.e. independent
of their size) – a fact that should be taken into account when
deciding on the scope of Gross-to-Net techniques for Solvency II
purposes.
PROPORTIONALITY - LEVEL 2
MEASURES
CEIOPS-CP-45/09, 2 July 2009,
Consultation Paper No. 45
1.
Introduction
2.
Advice - Proportionality
3.
Proportionality Assessment – A three step process
4.
Simplified Methods
5.
Reinsurance Recoverables
6.
Annex A: Gross-to-net Techniques
New:
Solvency ii and Captives, CEIOPS Level 2 measures
|
