Insurance and the Credic Crisis: Impact and Ten Consequences for risk Management and Supervision

Insurance and the Credit Crisis: Impact and
Ten Consequences for Risk Management and
Supervision
Martin Eling und Hato Schmeiser
Preprint Series: 2009-13
Fakult¨at f¨ur Mathematik und Wirtschaftswissenschaften
UNIVERSIT ¨
AT ULM

---

Insurance and the Credit Crisis:
Impact and Ten Consequences for Risk Management and Supervision

Martin Elinga*, Hato Schmeiserb

aInstitute of Insurance Science, Ulm University, Helmholtzstr. 22, 89081 Ulm, Germany
bInstitute of Insurance Economics, University of St. Gallen, Kirchlistr. 2, 9010 St. Gallen, Switzerland

_________________________________________________________________________________________

Abstract

Although the insurance industry is less affected than the banking industry, the credit crisis has revealed room for
improvement in its risk management and supervision. Based on this observation, we formulate ten consequences
for risk management and insurance regulation. Many of these reflect current discussions in academia and prac-
tice, but we also add a number of new ideas that have not yet been the focus of discussion. Among these are
specific aspects of agency and portfolio theory, a concept for a controlled runoff for insolvent insurers, new
principles in stress testing, improved communication aspects, market discipline, and accountability. Another
contribution of this article is to embed the current practitioners’ discussion in the recent academic literature, for
example, with regard to the regulation of financial conglomerates.

JEL classification: G01; G15; G20; G22
Keywords: Credit crisis; integrated risk management; principles-based supervision; financial conglomerates;
regulatory arbitrage

_________________________________________________________________________________________

* Corresponding author. Tel.: +49 731 5031183; fax: +49 731 5031188
E-mail addresses: martin.eling@uni-ulm.de (Martin Eling), hato.schmeiser@unisg.ch (Hato Schmeiser)

---

1. Introduction

In this paper, we address the credit crisis from the perspective of the insurance industry. Our
aim is to highlight the impact the crisis had on insurance companies and to derive conse-
quences for risk management and insurance regulation. The prudent and conservative busi-
ness policies that most insurers engage in have proven the industry to be quite resistant
throughout the crisis. We thus believe that the insurance industry is generally in a strong posi-
tion that will improve in the coming years, especially compared to other financial services
providers. However, not all insurance market participants have followed a prudent strategy
(e.g., AIG or Yamato Life). Hence, the crisis has revealed several deficiencies in the fields of
risk management and supervision.

Based on these observations, the aim of this article is to formulate ten consequences for risk
management and supervision. Many of these consequences reflect current discussions in aca-
demia (see, e.g., Felton/Reinhardt, 2008; Schanz, 2009) and practice (see, e.g., CRO Forum,
2009; CEA, 2008), but we also integrate a number of new ideas that have not yet been the
focus of discussion regarding the credit crisis. Among these are some basic lessons from
agency theory and portfolio theory, the consideration of a controlled runoff for insolvent in-
surers, new principles in stress testing, and improving communication, as well as aspects of
market discipline, and accountability––especially in respect to rating agencies. Another con-
tribution of this article is to embed practitioners’ discussion in academic literature, for exam-
ple, with regard to management compensation or the regulation of financial conglomerates.
Most consequences we discuss are applicable not only to the insurance industry, but also to
other sectors of the financial services market. We think that one of the most fundamental les-
sons to be learned from the credit crisis is that financial services should take place in an inte-
grated marketplace that combines integrated risk management and supervision. The separate
regulation of banking, insurance, and other financial services providers can always create op-
tions for regulatory arbitrage, which was one of the roots of the credit crisis.

Table 1 provides an overview of the impact the financial crisis had on insurance companies
and the ten consequences we see arising from the current crisis. The ten lessons are not easily
separated into those more applicable for risk management and those applicable for supervi-
sion; instead, we see supervision as an element of integrated risk management. We thus be-
lieve that Solvency II and the Swiss Solvency Test (SST)––both of which aim at integrating
supervision and risk management––are steps in the right direction toward risk-based capital
standards.
1

---

Impact of the credit crisis on insurance companies
-
Crisis emerged in the banking industry (Lehman, Fannie Mae, Freddy Mac, among others)
-
Insurers less affected than banking, but nevertheless serious impact (AIG, Yamato Life)
-
Impact on assets: losses on stock and bond markets; increase in credit risk
-
Impact on liabilities: insurance in credit markets, reinsurance, D&O, E&O, reduced demand for insurance
-
Crisis revealed deficiencies in risk management and supervision
Ten consequences for risk management and supervision
1)
We need to strengthen risk management and supervision
2)
We need to take care of model risk and nonlinearities
3)
We need easy to use and understandable risk management
4)
Take heed of the lessons from agency theory—the right incentives are needed
5)
Take heed of the lessons from portfolio theory—risk, return, and diversification
6)
Principles instead of rules-based regulation—Solvency II and SST are steps in the right direction
7)
A concept for a controlled runoff in the insurance industry is needed
8)
Financial conglomerates need to be supervised at the group level
9)
No regulatory arbitrage in financial services markets
10) Transparency, market discipline, and accountability are needed
Table 1: Impact of the crisis and consequences for risk management and supervision

The remainder of this article is structured as follows. In Section 2, we present a short over-
view of the emergence of the crisis and its impact on insurance companies. In Section 3, con-
sequences for future risk management and supervision of insurance companies are derived.
We conclude in Section 4.

2. Impact of the credit crisis on insurance companies

Due to differences in business models, insurance companies are less affected by the credit
crisis than is the banking industry. Insurance companies are generally not at risk of a bank run
given that, for example, in non-life insurance, payments are linked to claim events. In addi-
tion, insurers are funded in advance. In life insurance, surrendering a contract has disadvan-
tages, such as lapse costs, so that the policyholder has a limited incentive to terminate the con-
tract. Furthermore, most insurers do not have significant exposure to mortgage-backed securi-
ties and other forms of securitization and thus have not been directly affected by the credit
crunch that was at the root of the current financial crisis (see, e.g., CEA, 2008). Underwriting
risk comprises a high proportion of an insurer’s overall risk. The liability portfolio is diversi-
fied and in many lines of business largely uncorrelated with the asset side (and, hence, to the
capital market in general). Again, this is an important difference from the banking industry,
where the portfolio of outstanding loans is highly correlated with general economic factors
(see Pan European Insurance Forum, 2009).

Nevertheless, the insurance industry has suffered substantially in the recent crisis, on both the
asset and the liability side. Insurers are among the largest institutional investors on the capital
2

---

market and thus negative development regarding asset value is almost unavoidable. On the
liability side, insurers can be adversely affected through insurance in the credit market, by
directors and officers (D&O) as well as errors and omissions (E&O) insurance, or by a rein-
surers’ default. Furthermore, in a situation of economic downturn, insurers will suffer a de-
cline in demand for insurance products (see, e.g., Grace/Hotchkiss, 1995; Chen/Wong/Lee,
1999).

Figure 1 shows the Dow Jones 30 index for the years 2005 to 2009 along with some of the
most frequently mentioned events of the financial crisis. The lower part of the figure empha-
sizes events affecting the insurance industry.

09/08
03/08
Fannie Mae;
Bear
10/07
Freddie Mac;
Stearns
Merrill
Lehman;
Lynch
Merrill Lynch;
Industry-
12/07
Morgan Stanley;
wide
Citi-
Goldman Sachs
events
09/07
group
…
Northern
Rock
09/08
Events in
11/07
10/08
AIG
insurance
Swiss Re
Yamato
markets …
…
Phase 1 (until 12/05):
Phase 2 (01/06-08/07):
Phase 3 (09/07-08/08):
Phase 4 (f rom 09/08 on):
Maximum inf lation in
First warning
First hits and
Big hits and govern-
US housing prices
signs
depreciation
ment bailouts

Figure 1: Dow Jones 30 index and main events of the financial crisis

The financial crises can divided into four phases. The first phase was a time of low interest
rates and increasing U.S. housing prices (reaching its maximum in 2005). Warning signs then
appeared in Phase 2 (2006 until August 2007), e.g., with a flat and then inverse yield curve.
The subprime crisis in U.S. housing then started in the summer of 2007 (see Reinhart/Rogoff,
2008). One of the first visible events in respect to the financial crisis was the bank run on
Northern Rock in September 2007 and the consequent support from the Bank of England (be-
ginning of phase 3). At that time, many market participants in the banking and insurance in-
dustry reported large write-downs due to mortgage defaults or related problems in credit mar-
kets. Among these were Merrill Lynch (with a loss of $8 billion), Citigroup ($18 billion), and
Swiss Re ($1.1 billion; Swiss Re is only one of many insurers to suffer write-downs, but it
was the first large write-down in the insurance sector and is thus mentioned).
3

---

In March 2008, the Federal Reserve Bank of New York provided an emergency loan to Bear
Stearns in order to avert a sudden collapse of the company. The fourth phase of “big hits” and
government bail-outs began in September 2008 with the federal takeover of Fannie Mae and
Freddie Mac (September 7), the bankruptcy of Lehman Brothers (September 15), and Federal
Reserve support of the American International Group AIG (September 16). Furthermore,
Merrill Lynch was sold to the Bank of America (September 14) and Morgan Stanley as well
as Goldman Sachs changed their status from investment banks to traditional bank holding
companies (September 21). Among the subsequent events were the Royal Bank of Scotland
announcing the biggest corporate losses in U.K. history (January 2009) and AIG reporting the
biggest corporate losses—almost $62 billion—in U.S. history (March 2009).

The three most often reported events of the crisis for the insurance industry are the govern-
ment bail-out of the American International Group (AIG), the write-downs at Swiss Re (due
to reinsurance in credit portfolios), and the insolvency of Yamato Life Insurance (due to se-
vere risk management failures in asset management). All three events have different characte-
ristics and illustrate that insurers’ balance sheets were affected by different aspects of the cri-
sis. These three cases show on the one hand that an adverse scenario can include a combina-
tion of negative developments on both the asset side and the liability side. On the other hand,
however, the different nature of these three events also reveals that they had only a limited
systematic impact at the global industry level, which occurred on the asset side of the balance
sheet: almost all insurers were hit by negative developments regarding asset values within the
capital market. Only some insurers were directly affected from investments in structured cre-
dit products, but most felt an indirect impact from the losses in many investments during the
recent capital market plunge. That these effects on asset management can produce a threaten-
ing economic situation is illustrated by the Japanese life insurer Yamato Life Insurance: this
company experienced losses in the subprime area and losses due to a high investment in
stocks. From the underwriting side, however, no specific problems have been reported.

One advantage of the continental European insurance industry in this context is that tradition-
ally its asset allocation is conservative and it invests a relatively low portion of assets in
stocks. Therefore, these insurers were not too adversely affected by the stock market plunge
when, for example, there was a reduction of 30–50% in 2008 for all major stock indices, in-
cluding the Dow Jones 30, the Nikkei, the FTSE 100, and the DAX 30. It appears that insurers
had learned a valuable lesson from their bad experience with the stock market plunge at the
beginning of this century. However, a main difference between the current capital market
4

---

plunge and other stock market plunges, especially in 2002, is that these days it is not only
stock markets that are negatively affected; there are also adverse reactions in bond markets
and a massive increase in credit risk for products and institutions previously considered safe.
An example is the default of Lehman Brothers, in which a number of insurers were deeply
involved (e.g., the German health insurer Landeskrankenhilfe with an asset volume of around
4 billion Euro had invested 200 million Euro at Lehman Brothers; see Fromme/Krüger, 2009).
Some insurers (e.g., the U.S.-based Aflac) were also engaged in hybrid capital and other sub-
ordinate debt issued by banks, resulting in large write-downs (see Mai/Bayer, 2009).

The liability side of the insurance industry has also been affected by the financial crises, but
less severely, with effects largely dependent on the insurer’s line of business. If the insurer is
engaged in credit markets, then it could suffer a negative impact due to the sudden increase in
credit risk, which is what happened at Swiss Re with a depreciation of $1.1 billion in Novem-
ber 2007. The loss resulted from two credit default swaps (CDS) designed to provide protec-
tion for a client against a fall in the value of a portfolio of mortgage-backed securities (see
Swiss Re, 2007). Insurance companies such as AIG, MBIA, and Ambac first suffered ratings
downgrades when mortgage defaults increased their potential exposure to CDS losses. AIG
had CDSs insuring $440 billion of mortgage-backed securities (see Harrington/Moses, 2009;
Baranoff/Sager, 2009). Thus, following the subprime crisis, AIG had depreciations of $11
billion on its credit portfolio in the fourth quarter of 2008 and a quarterly loss of $5.3 billion,
finally resulting in the government bailout (for more details on the AIG case, see Sjostrom,
2009). In addition to these impacts on the insurance and reinsurance sector, there are also
worries with D&O insurance. Many U.S. insurers have already begun to set up reserves for
potential claims following the financial crisis (see Fromme, 2008). Another aspect that is es-
pecially relevant for life insurers is that the uncertainty surrounding the macroeconomic envi-
ronment and interest rates pose difficulties for providing investment guarantees and hence
may lead to the necessity of redesigning life insurance products (see AM Best, 2009).

Overall, European insurers are not significantly exposed to credit risks and thus have not been
directly affected by the credit crunch that was at the root of the financial crisis. However, as
some of the largest institutional investors, they have suffered from the dramatic write-downs
of financial assets. Moreover, the insurance industry could certainly be affected by group con-
tagion effects, by an increase in D&O claims, and by a fall in the sale of insurance products
due to the economic slowdown (see CEA, 2008, p. 4).





5

---

3. Ten consequences for risk management and supervision

Issues related to supervision and corporate governance have often been deemed causes of the
crisis. These issues include pro-cyclicality and similar behavior due to regulatory rules, regu-
latory arbitrage, inappropriate accounting rules based on historical acquisition costs, lack of
transparency, and inadequate management decisions, probably driven by wrong incentives.
While insurance regulation has already been the subject of reform in Europe (Solvency II,
Swiss Solvency Test), the ongoing financial market crisis has focused even more attention on
risk management and regulation in financial services, both in academia and practice. In this
section we derive the ten main consequences that we see arising from the crisis.

1) We need to strengthen risk management and supervision

Identifying, measuring, and valuing risk is at the core of the insurer business model and
should not be delegated to a third party. Although there is evidence that rating agencies are
relatively successful in identifying financial distress compared to regulators (see Pottiers and
Sommer, 2002), the financial crisis has made clear that relying heavily on ratings can be mis-
leading and dangerous. Insurers and regulators should thus be aware of substituting a rating
for their own due diligence as rating agencies’ methodologies are not really transparent. In
contrast to Solvency I, ratings are essential in the SST and under Solvency II, e.g., for deriv-
ing the credit risk of the insurer’s bond portfolio and for determining the default risk of rein-
surance exposure and regulators need to review these rules (see, e.g., Eling/Gatzert/
Schmeiser, 2008).

In light of the challenging capital and insurance market environment, strong enterprise risk
management (ERM) is a crucial element in maintaining financial strength and ensuring a safe
insurance industry. Risk management must be proactive, independent, and have sufficient
power and authority. Independence is important because of the possibility of conflicts of in-
terest, including those between the underwriting sector, the sales department, and risk manag-
ers. Risk management must play a leading role in each insurance company, which could be
accomplished by transferring the concept of “responsible actuary” (“verantwortlicher Ak-
tuar”; implemented in Germany, Austria, and Switzerland) or “appointed actuary” (in the
United Kingdom, Belgium, and the Netherlands; see Daykin, 1999) to that of an “appointed
risk manager.” By law, the responsible actuary has a predefined function, responsibility, inde-
pendence, and reporting requirements with regard to the board. Defining a corresponding
“appointed risk manager” as a specific function with a wider role in an insurance company
might help to more clearly organize what risk management is and what it should do. The “ap-
6

---

pointed risk manager” could also be a contact person for the regulator in order to ensure that
regulatory rules are embedded within an integrated risk management scheme. Those with ac-
tuarial skills are well suited for such a position because of their training in analyzing various
forms of risk, and their ability to judge the potential for upside gain, as well as downside loss,
associated with these forms of risk (see D’Arcy, 2005). However, such a position could also
be filled by nonmathematicians with a business or economics background. Note that in some
countries the function of risk management is one of the duties of the “appointed actuary.” In
such a case we either need to separate the tasks of the “appointed actuary” from those of the
“appointed risk manager” or combine both jobs into one position enjoying greater power and
authority. We believe that splitting this large and important task into two positions will work
best: the “appointed actuary” being responsible for adequate premium and reserves calcula-
tion amongst others, and the “appointed risk manager” being responsible for risk modeling,
risk management, and implementing the results in an integrated risk management process.
Clearly defined responsibilities, along with close collaboration between these two important
functions, are two prerequisites for successful risk management.

2) We need to take care of model risk and nonlinearities

One of the greatest pitfalls of risk models and solvency approaches is model risk. For in-
stance, there is always the possibility that the underlying risk distributions have been wrongly
specified. This can occur when there is not a sufficient number of historical observations
available (a smaller data set, ceteris paribus, increases the probability of a misspecification).
Moreover, the underlying distribution might not be stable over time and, hence, probability
distributions observable in the past provide very little information about the future. In addition
to misspecifications as to the “true” probability distributions, the chosen stochastic model
itself might be inappropriate.

To guard against too much faith being placed in a specific risk model/solvency approach and
its assumptions, we believe that it is important to vary the implicit model parameters in some
specified range, similar to what is done in stress testing. By doing this, risk managers and reg-
ulators can obtain a much better understanding of the sensitivity of specific results of the sol-
vency model and provide additional information regarding an insurer’s main sources of risk.
A first step in this direction––one that has less to do with model risk, and more to do with the
economic environment––has been taken in the scenario testing concept given of the SST. In-
surance is a risky business, making it necessary to think in terms of confidence intervals ra-
ther than in terms of expected values.
7

---

The results of risk models and the quality of decisions based thereon depend on an appropriate
modeling of the stochastic behavior of assets and liabilities. In this context, mapping nonli-
near dependencies is a point of concern (see Eling/Toplek, 2009). Many risk models and most
practitioners still focus on linear correlation even though the literature suggests that solely
considering linear correlation is inappropriate when modeling dependence structures between
heavy-tailed and skewed risks, which are frequent in the insurance context (see, e.g., Em-
brechts/McNeil/Straumann, 2002). These risks are especially relevant in case of extreme
events, such as the September 11, 2001, terrorist attacks that resulted in large losses for insur-
ance companies both from their underwriting business and the related capital market plunge
(see, e.g., Achleitner/Biebel/Wichels, 2002). We believe that the current financial crisis is
another example of a situation in which some insurers have sustained large losses from their
investments, e.g., in mortgage-backed securities, as well as from insuring structured credit
products such as collateralized debt obligations. These complex interactions may not be ade-
quately captured by linear dependencies (Ashby/Sharma/McDonnell, 2003) and hence the
current crisis emphasizes the relevance of modeling nonlinear dependencies.

Another question in the modeling context is what risks should be considered. The most dan-
gerous risks, of course, are those that are unforeseen. However, most models focus on market
risk. There are models for credit and underwriting risk, but the credit crisis has shown that we
do not have sufficiently good models to handle liquidity risk (see, e.g., Rudolph, 2008). We
thus need to develop new models for liquidity risk management and we need to take into con-
sideration new risk sources that have not yet been the focus of discussion. In addition, we
need to remember that one of the main assumptions of many pricing and risk management
models is a liquid market. If a liquid market does not exist (anymore), the use of such models
is highly questionable.

3) We need easy to use and understandable risk management

The interaction between risk models, the risk management process, and managerial decisions
can be improved. The best risk models are useless if the results are not understood by the
people who make decisions based on them. A serious problem in this context is the communi-
cation gap between risk managers and decision makers on the executive board. Risk managers
and actuaries develop and implement risk models and it is likely that most of them are aware
of the underlying assumptions and limitations of the model when interpreting its results.
However, the executive board may not have the same degree of competence in this particular
area or the time to develop it. They thus require easy to use and understandable statistics.

However, due to the inherent problems of models as discussed above, regardless of how well
presented, their results should not be the sole basis for management decisions. Model results
8

---