EML.2 Broschüre 16.07.

1999 10:21 Uhr Seite 1

ERC Frankona

EML or PML
Does It Make
A Difference?
Practical Use And Calculation
Of Loss Potential Estimates
In Fire Insurance
EML.2 Broschüre 16.07.1999 10:21 Uhr Seite 2

Contents

Page

Foreword 3

1 Historical Development 4

2 Consequences of Incorrect 5
Loss Estimates

3 General Approach to Loss 6

Estimates

4 Estimated Maximum Loss 7

4.1 Definition
4.2 Assumptions and Assessment
Criteria
4.3 Fire Loss of Profits

5 Probable Maximum Loss 10

5.1 Definition
5.2 Assumptions and Assessment
Criteria
5.3 Fire Loss of Profits

6 Example of Loss Estimates 12

6.1 Estimated Maximum Loss
6.2 Probable Maximum Loss

7 Other Commonly Used Concepts 14

7.1 Normal Loss Expectancy
7.2 Maximum Foreseeable Loss
7.3 Catastrophe Loss
7.4 Maximum Amount Subject
7.5 Maximum Credible Loss
7.6 Possible Maximum Loss,
Maximum Possible Loss

8 Overview of Various Concepts 15

9 Final Remarks 16

Appendix – LIRMA Definition 17

of a “Single Risk”

References 18
EML.2 Broschüre 16.07.1999 10:22 Uhr Seite 3

Foreword

rrespective of the definition and method used to assess loss potential, the liability

I assumed by insurers and reinsurers in any given risk is ultimately limited by the
share accepted in the risk’s total sum insured. Nevertheless, it is imperative for
insurers and reinsurers to have a common understanding of the various definitions
and criteria applied in determining EML/PML to allow them to assess accurately the
financial consequences of loss occurrences. This is particularly important for losses
exceeding the previously calculated loss potential.
Despite the relevance of loss potential estimation for Property underwriting,
no common standard has yet been adopted by the insurance industry as a whole.
Underwriters employ a variety of definitions of and methods of determining Estimated
Maximum Losses (EMLs) and Probable Maximum Losses (PMLs). In broad terms, the
diverse definitions result from differences in the ways in which various levels of fire
protection and fire-fighting features are taken into consideration. The meaning
of abbreviations used may also vary widely, which illustrates the inherent danger in
assuming that similar or identical abbreviations and terminology are identical.
Misinterpretations may prove to be dangerous and costly. For example, the most
commonly confused abbreviation, “PML”, may mean either “probable” or “possible”
maximum loss.
This brochure provides a general introduction to loss potential estimates in Fire
insurance, i.e. the purpose and practical use of loss potential estimates and the basis of
practical, and particularly of technical, calculations of EML and PML. It focuses exclu-
sively on single-risk exposure. Loss estimations at the portfolio level, i.e. cumulative
losses, are not considered in detail. Within this context, however, we may be permitted
to emphasise how important it is that underwriters always make the appropriate
analysis (e.g. earthquake, windstorm, flood, etc.).
An overview of terms currently in use in various countries has also been included.
There may be deviations peculiar to certain companies and insurance clients as a result
of their specific calculation assumptions, but such deviations have not been dealt with
in this publication.
The objective of this brochure is to establish a common understanding of the
various approaches to Fire loss assessment.

3
EML.2 Broschüre 16.07.1999 10:22 Uhr Seite 4
1 Historical Development
I
nitially, all Property underwriting and reinsurance
was done on the basis of the total sum insured per
policy. However, as industries grew and individual
policies were extended to cover more plants, insurers
found that they were being compelled to cede too much
of their business in the form of reinsurance and/or
coinsurance.
Insurers found that they could increase their net
retention without endangering their company’s share
capital by basing their underwriting and net retention
first on the “highest sum insured per location”, which is
now expressed as “Maximum Unit at Risk” (MUR), and
somewhat later, on the “maximum sum insured, subject
to one fire at each location”, which is now expressed as
“Amount Subject” (AS).
Underwriting policy evolved concurrently with tech-
nical developments in the building sector, i.e. the use of
fire walls and fireproof materials in accordance with
strict building regulations, the introduction of automatic
fire-extinguishing installations and fire alarms, etc., and
last but not least due to generally favourable loss experi-
ence. Many insurers determined their net retentions on
the basis of “maximum loss under normally unfavourable
circumstances per location” or even “per fire section”.
Experience showed, however, that this was an extremely
dangerous approach because too many fires spread to
adjacent sections, resulting in losses considerably exceed-
ing the estimates on which the net retention had been
based.
Following the crisis in the fifties, underwriting policy
was again placed on a more conservative footing, and
loss potential estimates were redefined. The terms most
frequently used to express the maximum loss which
might be caused by a single fire (or other insured peril)
were – and still are – the following:
PML: Probable Maximum Loss
PML: Possible Maximum Loss
MPL: Maximum Probable Loss
MPL: Maximum Possible Loss
MCL: Maximum Credible Loss
MFL: Maximum Foreseeable Loss
EML: Estimated Maximum Loss.
4
The various terms gave rise to the major problem,
however, that each and every insurer could have used
different philosophies and definitions of what they
thought to be a proper loss estimate. The definitions are
all subjective and open to different interpretations. Also,
the letter P in MPL and PML caused a great deal of
confusion because some companies used it to mean
“possible”, while others used it to mean “probable”.
In the early seventies this led to a proposal by the
Reinsurance Offices’ Association (ROA) in London that a
standard definition of EML be used for reinsurance
purposes. The ROA has since been superseded by the
London Institute Insurance and Reinsurance Manage-
ment Association (LIRMA).
Despite the obvious advantages of having a single,
commonly accepted definition of maximum loss estima-
tion, the LIRMA definition has unfortunately never been
generally accepted by the market as it deserved to be.
This is because, even today, companies prefer to use
their own definitions of EML, PML, MFL, etc. when
assessing risks and deciding the size of their net reten-
tions.
Since the LIRMA definition is the only common defini-
tion of maximum loss potential available, we will initially
focus on this definition. Before examining this definition
in detail, however, we would like to discuss briefly the
general approach to loss estimation and the conse-
quences of actual losses which exceed loss estimates.
EML.2 Broschüre 16.07.1999 10:22 Uhr Seite 5
2 Consequences of Incorrect
Loss Estimates
A devastating explosion in the 1970s at a chemical plant
in Flixborough caused enormous damage. Here a view of the
tank farm in flames.
E
stimated Maximum Losses (EMLs) are produced
for the use of the underwriter to assist him
or her with deciding the size and extent of line
to be written and the level of reinsurance re-
quired. They are useful in demonstrating the relationship
between the level of premium being obtained and the
likely extent of loss. In addition, the purpose of EMLs is
to allow insurers to optimise their net retentions and
thus to keep as much premium as possible for their own
account. In other words, the purpose is to decide how
large a monetary loss the company is prepared to bear
for its own account, set against its own financial
strength, or possibly pass on to its reinsurance pro-
gramme. By writing a share on EML basis, an insurer
can write more of the risk, but the consequences of
EML failure might be damaging.
A reinsurance programme based on EML is not prob-
lematic in itself. Irrespective of how EMLs are defined,
however, their calculation is based on a human evalua-
tion of the loss potential entailed in specific risks. This
means that it will always be possible for estimates to vary.
Fortunately, the frequency of inaccurate calculations in
general should be low, but when they do occur, they are
often of serious magnitude. If the “true“ maximum loss
is incorrectly assessed and then actually occurs, both the
ceding company and its reinsurer may have to pay a
considerably greater amount than was assumed when
the treaty or facultative offer was accepted.
One of the most spectacular errors involving an
inaccurate EML calculation was the Flixborough
disaster. Despite the fact that the incident took place
more than 20 years ago, it continues to be a good
example because the damage was so extensive that
most insurers and reinsurers still remember what
happened on 1 June 1974.
The chemical plant was insured against material
damage due to fire or explosion up to an amount of
USD 43 million. Leakage led to a devastating explosion
which destroyed nearly 80% of the plant. The EML,
however, had been estimated at only 20%. The balance
of 60% amounted to nearly USD 26 million. In addition
to the Property damage, the insured was also protected
by a Loss of Profits (LOP) policy with a limit of USD 9 mil-
lion, for which the EML had been assessed at 60%.
However, the LOP loss was 66.5%. Although a small
difference in percentage, the balance amounted
to USD 1 million. All in all, the insurers and reinsurers
had to pay nearly USD 27 million more than had been
anticipated when they accepted shares of the risk.
It is obvious that such errors could lead to solvency
problems for the parties involved. Insurance brokers,
underwriters and production companies have developed
a number of methods to help quantify such catastrophic
EMLs. For example, engineering-based models1 and
empirical approaches based on loss history are now
favoured in energy insurance.
1 Sedgwick Energy Ltd., Vapour Cloud Explosion Methodology, March 1994.
5
EML.2 Broschüre 16.07.1999 10:23 Uhr Seite 6

3 General Approach to
Loss Estimates

G
enerally, the Estimated Maximum Loss (EML)
or Probable Maximum Loss (PML) are estimated
by dividing the risk into complexes. A com-
plex may consist of one or more buildings or
rooms or structures which themselves contain structural
boundaries or separations. They need not be completely
separated from neighbouring buildings or structures.
Caution should be exercised in defining complexes
because experience has shown that structural separation
in the conventional sense is no longer entirely effective
in the event of a loss. For example, today’s rapid techno-
logical advancement has greatly increased fire loads and
the danger of explosion.
It is necessary to identify the complex with the
greatest exposure. Additionally, it should be considered
that a fire can also spread to other complexes. The pos-
sibility that a fire may spread beyond the complex in which
it starts is suggested by the following risk characteristics
or events2: As with this major fire which caused damage
● Risk of explosion amounting to more than USD 5 million, it is often difficult
● Risk of consequential damage resulting from corrosive to prove whether arson is involved.
gases or vapours
● Risks created by the neighbourhood
● Cases of simultaneous arson in several separate
complexes ● Site interviews should be conducted with members
● Disaster-like effects of external factors connected of senior management.
neither directly nor indirectly with the risk insured, ● Loss estimates should be updated regularly.
e.g. plane crash. When assessing the EML or PML, the actual policy word-
General practice guidelines may be summarised as ing has to be considered too, e.g. coverage for residual
follows3: values, compliance with new building regulations, re-
● Any loss estimation, and particularly EML/PML moval of debris. In this context it is worth mentioning that
calculations, should be accompanied by definitions of in some countries the buildings will be insured at the
the terms used. new replacement value without a fixed maximum sum
● Calculations should be accompanied by a description insured.
of the assumptions made and the method used. EMLs or PMLs are expressed as a percentage of the
● Assessments should consider a range of factors. total sum insured or in an absolute amount.
Business Interruption assessments should take into In order to clarify this issue, the following definitions
account any interdependencies which may exist and and assumptions may be taken as a possible guideline in
the length of the indemnity period. assessing the risk term of EMLs and PMLs.
● Maximum loss assessments should be based on
fire/explosion potential unless another peril such as
earthquake, flood, windstorm or aircraft damage
represents a greater exposure. In that case, the peril
representing a greater exposure should also be
evaluated in more detail.

2 Recommendations by the former German Association of Property

Insurers, Cologne, on the Determination of the Probable Maximum Loss
with Fire and FLOP Insurance. Published in the Manual for Industrial
Fire and Fire Loss of Profits Insurance, Reports.
3 Insurance Institute of London, EML Assessments: London Market
Practice, September 1995.

6
EML.2 Broschüre 16.07.1999 10:23 Uhr Seite 7
4 Estimated Maximum Loss
T
he only definition generally recognised
today is the one originally proposed by the
ROA (which has now been superseded by
LIRMA)4:
4.1 Definition
Estimated Maximum Loss (Fire and explosion,
material damage only)
An estimate of the monetary loss which could be sus-
tained by insurers on a single risk as a result of a single
fire or explosion considered to be within the realms of
probability. The estimate ignores such remote coincidences
and catastrophes as may be possibilities but which still
remain unlikely.
As is evident, this definition contains a number of
subjective words which make it possible to take into con-
sideration local experience and factors which may vary
from country to country or even from place to place.
However, the LIRMA definition goes on to state what
is meant by a “single risk“5 and to list the parameters
which should or should not be taken into account when
estimating the monetary loss, i.e. the definition attempts
to specify what should be regarded as being “within the
realms of probability“ and what should be considered to
be “remote and unlikely coincidences or catastrophes“.
Most of the insurance industry uses the LIRMA
definition as a basis for defining both a single risk and
an EML.
4.2
Assumptions and Assessment Criteria
Estimated Maximum Losses (EMLs) are based on a
number of assumptions: firstly, that automatic fire-alarm
and extinguishing systems (e. g. sprinklers, carbon
dioxide and foam systems, etc.) are not in service, or
they are non-existent. Secondly, it is also assumed that
competent private and public assistance is delayed, and
that the fire doors between at least two adjacent fire
areas are open. Fire doors that are held open magnetic-
ally do not close. Thirdly, it should be assumed that full,
sealed firewalls will hold, unless an exceptionally high
fire load or evident risk of explosion makes it seem justi-
fied to assume otherwise. The separation walls, fire
doors included, are constructed to withstand fire for at
least two hours.
It is also assumed that the risks are exposed to nor-
mal circumstances. Abnormal circumstances, which are
not taken into account:
● Plane crashes or objects falling from the sky
● Explosion in a vessel due to vapour
● Arson and sabotage.
A standard method for calculating EMLs should
consider the following specific risk factors6:
● The dimensions and shape of the area potentially
exposed to a single fire or explosion
● The construction of roof, walls and floors
● The presence of combustible linings on walls, roofs,
ceilings and partitions
● The nature, distribution and combustibility of contents
(fire load)
● The use of hazardous processes and substances and
the degree to which they are separated
● The susceptibility of the contents to damage by smoke,
heat and water
● The risk of explosion (where liquefied petroleum gas,
flammable goods and dust are stored/used)
● Any hazards arising from gases or corrosive materials
● Any concentrations of value within a small area
● The standards of management and housekeeping
● The clearance between buildings, with consideration
given to design and materials
● The distance to any stocks such as timber, goods,
cistern park or liquefied petroleum gas tank
● Any special geographical or meteorological features
such as wind conditions.
4 Reinsurance Offices’ Association, Definition of Estimated Maximum Loss,
London, 1974.
5 LIRMA Definition of a “Single Risk”, see Appendix.
6 LIRMA Definition of Estimated Maximum Loss.
7
EML.2 Broschüre 16.07.1999 10:23 Uhr Seite 8
Fire or faults in control rooms can also disrupt operations
in other sections of a plant.
Factors which should not be included when assessing
an EML7:
● Any horizontal separations within a building
● Fire-resisting doors
● The absence of normal sources of ignition
● The presence or planned installation of fire-detection,
prevention or extinguishing systems, including sprink-
lers, and the adequacy of fire-brigade services.
The existence of such protective installations may merit a
higher net retention than that calculated for a similar,
unprotected type of risk, but it should be clearly under-
stood that such protective systems may be rendered in-
operative and should therefore be disregarded when cal-
culating an EML.
In certain territories, however, it is customary to
make allowance for the presence of overall sprinkler
protection, horizontal separations and fire-resisting
doors in calculating non-LIRMA EMLs. In such cases, the
amount by which the EML has been amended should be
clearly stated, as should also the fact that such protec-
tive facilities have been taken into account.
Regardless of how EMLs are defined, the practical
calculation of the maximum Property loss, the maximum
Business Interruption loss or the maximum combined
loss can be divided into three steps:
7 LIRMA Definition of Estimated Maximum Loss.
8
1. Identify the potential risk areas among the single
risks at the plant. When defining fire areas, also
consider any production bottlenecks or specialised
machinery. Business Interruption losses often exceed
Property damage (PD) losses.
2. Identify the loss scenario entailing the maximum
monetary loss in the selected areas (PD and FLOP),
taking into consideration the continuity of combust-
ibles, hazards, building construction, unimpaired
protection and business interruption. Estimate the
extent of fire spread and damage which would result
from a major fire, given the fire area and factors
assumed.
3. Calculate the PD and FLOP losses for the selected
areas, based on the identified loss scenarios.
All calculations should be prepared in accordance with
the selected definition of EML (i.e. if the LIRMA defini-
tion is used), giving due thought to the list of applicable
factors and, of course, the actual sums insured and
coverage provided.
4.3 Fire Loss of Profits
The obvious weakness of the LIRMA definition is that it
says nothing about Fire Loss of Profits (FLOP), which is
now often the most important factor in estimating loss
potential. However, since the maximum Business Inter-
ruption loss is usually triggered by physical loss or
damage, the LIRMA definition can easily be amended in
such a way that it could also be applied to combined
EMLs for Property damage and FLOP. Combined loss
estimates should be prepared in all cases where both
Property damage and Business Interruption are being
underwritten.
In addition to the principles applied to Fire EMLs,
underwriters assessing Fire Loss of Profits EMLs or
EML.2 Broschüre 16.07.1999 10:24 Uhr Seite 9
Carelessness during welding work caused highly flammable plastics to ignite
and led to the terrible disaster at Dusseldorf Airport in April 1996.
combined EMLs should also take into consideration the
following general aspects:
● Bottlenecks, e.g. key machinery
● Rebuilding or reinstatement period
● Computer dependency, e.g. process computer
● Contingency plans, which should be fully documented,
continually updated and, where feasible, validated
through testing
● Market situation, recovery of market share
● Dependencies upon suppliers and/or utility companies,
e.g. dependence on public utilities
● Interdependencies, e.g. between locations or premises
of an insured who owns several operating companies
and many sites
● Just-in-time deliveries, e.g. of raw materials, compon-
ents, parts or finished products
● Off-site exposures from adjacent sites owned by
other insureds.
In addition to these general aspects, it is obviously
necessary to consider a number of specific questions
which vary, depending on the type of company involved,
for example whether it is a company that operates
on a seasonal basis, whether rental facilities are available
and whether it makes extensive use of automation
and/or robotics, central warehousing or central
computing facilities.
Since it is not unusual for an insured to own several
operating companies and a number of sites, Business
Interruption losses are not always premises-specific.
Also, in cases where a single event may affect more than
one insured, underwriting should be aware of the com-
plexity involved in the accumulation of risk.
Other factors which should be taken into considera-
tion include the presence of special risk parameters,
e.g. polychlorinated biphenyl’s (PCBs) and/or asbestos
roofing, which may be noted on the PD survey, meaning
that access to the site may be restricted following
damage and thus result in a longer interruption of
business. The effects of delays in receiving planning per-
mission are important: for example, authorities could
close a site, impose restrictions on operation or require a
redesign in order to bring the plant technically up to date.
Furthermore, failure to keep off-site copies of specifi-
cations and/or drawings for tooling and computer
systems records should be considered. Attention should
also be paid to workforce availability and construction/
fabrication market conditions. Factors delaying recovery
may include requirements for a unique asset, a unique
location, critical off-site dependencies or key types of
employees.
Assuming a uniform period of liability of 12 months
for each fire area insured, the EML can be calculated on
the basis of the maximum estimated loss of capacity
during this period and the corresponding values involved.
If shorter indemnity periods have been fixed for salaries,
wages and commissions, the EML may be reduced.
It should be noted that the possibilities of reducing a
loss are not as great in cases where indemnity periods
are shorter.
In some cases loss limits are introduced in original
policies, as for example with large FLOP risks in the
German market. The maximum indemnity payable in the
event of a loss is determined by the loss limit which
corresponds to the EML.
9
EML.2 Broschüre 16.07.1999 10:25 Uhr Seite 10

5 Probable Maximum Loss

A
fter the term EML, the second most commonly
used term is Probable Maximum Loss (PML).
The basic concept is nearly the same as in EML
assessment. The following section describes
the definition of PML and the assumptions made in
calculating it.
The concept of PML attempts to quantify the conse-
quences of a major fire or explosion – regardless of how
the fire or explosion started – when the risk’s primary
fire-protection system is impaired. The key to under-
standing the concept of PML is the word “probable”.
Recognising that even the best fire-protection system is
subject to the ravages of Murphy’s Law

“If something can go wrong,

it will, and always at
the most inopportune time.”,

a probable or likely hypothetical impairment of the

risk’s primary fire-protection system is assumed. A loss
scenario is then prepared based on the construction,
occupancy, protection and exposure of the risk.

5.1 Definition
The PML is defined as the largest estimated loss arising
from a single event which was assessed with due care,
taking into account all the elements of the risk8.
In order to estimate the PML for a risk, it is necessary
to define a scenario in which a major fire occurs under
reasonably adverse conditions. The term “reasonably
adverse conditions” is assumed to mean that the fire
occurs in the worst possible place at the worst possible
time. An example would be a fire occurring in the middle
of a warehouse in the early hours of the morning at a
time when the sprinkler system has been shut off for
repairs.
The cause of the fire, which completely destroyed this plant
producing the raw materials for paint, is still unknown. The resultant
property damage came to more than USD 60 million.
5.2 Assumptions and Assessment Criteria
The underlying assumption is that all fire-protection
systems and measures, for example automatic sprinkler
systems and fire alarms, are ineffective.
The assumptions made for estimating the PML can
be summarised as follows:
● Water spray is ineffective against radiant heat
● Water spray is effective in preventing steel from
failure
● The sprinkler systems are properly designed but will
fail to control fires in areas containing high concen-
trations of plastics

8 K. Gerathewohl, Reinsurance Principles and Practice, 1980.

10
 EML.2 Broschüre 16.07.1999 10:26 Uhr Seite 11

PMLs for non-sprinklered buildings of combustible

construction (frame, joisted masonry) will usually
approach 100% in amount subject fire divisions. Excep-
tions would include long or widespread buildings in
situations where good fire-department response is
expected and plentiful water supplies are available.
Damage estimates for non-combustible construction will
be a function of the building size, combustion load and
fire-department response.

5.3 Fire Loss of Profits

To develop a Business Interruption PML, it is necessary to
prepare an analysis using the full annual risk values. The
PML will be limited only by the policy limit or the BI sub-
limit where such a limit exists in the client’s policy.
A number of factors should be considered in deter-
mining the BI resulting from a PML-type fire. In addition
to Fire factors, underwriters should take into account the
following Fire Loss of Profits (FLOP) factors:
● Operations and/or production facilities which can
cause total BI
● Operations and/or production facilities which can
cause partial BI
● Department(s) whose failure has the most critical
effect(s) on profits.
When assessing a FLOP PML, analysts should not reduce
the PML because buffer stocks or redundant capacity
exist, nor should they include any unusual changes in the
company’s financial situation or entrepreneurial risk.

ant

● Tied walls will remain standing

● The fire brigade is unable to control or extinguish
a fire
● The fire brigade is effective in controlling damage
from salvage and clean-up procedures
● Highly sensitive facilities and stocks will sustain ex-
tensive damage from water and/or smoke
● Offices are considered as an ordinary hazard
● Combustion safeguards on ovens, furnaces and other
fired vessels are inoperable.
The factors that are not taken into consideration are
basically the same as those described in the EML section.

11
EML.2 Broschüre 16.07.1999 10:26 Uhr Seite 12

6 Example of Loss Estimates

T
he facility in question is a food processing plant
with extensive processing installations whose Plant values Amount in USD
fire/explosion hazard is very high in relation to
the fire load. In view of the structure type and Total sum insured (TSI) 20,000,000
explosion hazard, loss events are sure to spread to other The values break down as follows:
areas of the plant and buildings, in particular to the
areas Storage I, Production I and Production II (silo ex- Production I 5,000,000
plosion). The building and plant-framing construction Production II 5,000,000
consists primarily of fireproof components. The facility’s Storage I 3,000,000
exposure does not seem to be increased by neighbour-
ing plants. It is possible for unauthorised persons, e.g. Storage II 5,000,000
saboteurs or arsonists, to enter the premises over the Other Facilities 2,000,000
unsecured, standard wire-mesh fence.

Storage I (Silos) Raw Materials 50 metres Administration

Gate-
house
25 metres

50 metres

Production I Foodstuff

Production II Special Products

25 metres Storage II Products

Plant site plan

12
EML.2 Broschüre 16.07.1999 10:26 Uhr Seite 13
6.1 Estimated Maximum Loss
This denotes the largest loss that can result from a single
incident in the plant. It assumes that the initial incident
is so extensive that the active protection systems are
rendered inoperative, and only the passive protection
facilities, such as spacing and fire proofing, are effective.
A scenario with a silo explosion and fire in Storage
area I has been presumed. The distance of 25 metres
between Storage I and Production I is assumed to be
inadequate in the event of a silo explosion. However,
the administration building is adequately separated
(each 50 metres). The fire will spread to Production I and II,
and completely destroy the whole production area.
However, Storage II can be rescued even assuming the
private/public fire fighting is delayed.
The EML for Fire Property – as a percentage of the
total sum insured – is calculated as follows:
USD 13,000,000 x 100 = 65%
USD 20,000,000
6.2 Probable Maximum Loss
This denotes the largest estimated loss resulting from a
single event in the plant under reasonably adverse condi-
tions. The underlying assumptions here are that all pos-
sible and/or required fire-safety measures are ineffective,
and that no effective fire-fighting occurs. Damage is
limited only by adequate separation.
On the one hand, the underwriter should take into
account the heavy fire load, particularly in Production
area I. On the other hand, the Production I and Produc-
tion II buildings are not adequately separated by a fire-
wall. The given scenario is nearly the same as in the EML
calculation. The difference is that the fire-fighting is
completely ineffective. The fire spreads to Storage II and
destroys it completely.
The PML for Fire Property – as a percentage of the
total sum insured – is calculated as follows:
USD 18,000,000 x 100 = 90%
USD 20,000,000
13
EML.2 Broschüre 16.07.1999 10:26 Uhr Seite 14
7 Other Commonly Used
Concepts
F
inally, it should be mentioned that most companies
now use two or even three different concepts of
loss estimation. Common concepts for calcu-
lating Property damage or Business Interruption
losses include:
7.1 Normal Loss Expectancy
The Normal Loss Expectancy (NLE) is an estimate of the
largest loss – excluding a catastrophe loss – which is to
be expected at a given site, assuming that all available
protective systems and measures function properly. The
NLE for a risk will typically be the loss resulting from a
significant fire or explosion which was extinguished or
controlled and will take into consideration the adequacy
of the risk’s fire protection systems.9
7.2 Maximum Foreseeable Loss
The Maximum Foreseeable Loss (MFL) is an estimate of
the largest loss – excluding a catastrophe loss – which is
to be expected at a given site, assuming that the primary
protection systems are either impaired or activated only
after a delay.9
The MFL for a risk is expressed in the same manner
as the PML. The MFL concept quantifies the under-
writer’s greatest fear of “everything that can go wrong,
doing just that.” Expressed in other terms, the MFL for a
risk represents the values for the amount subject fire
divisions. It is assumed that fire damage will be limited
only by the structural features of the buildings involved,
i.e. blank, free-standing fire walls or adequate distances
between buildings.
7.3 Catastrophe Loss9
The Catastrophe Loss (CL) is an estimate of the loss
which will occur at a site as a result of the following
types of events:
● Explosions resulting from massive releases of
flammable vapours or gases, commonly known as
vapour cloud explosions
14
● Tank or vessel failures resulting in the combustion
of flammable liquids over large areas of the plant,
including boiling liquid expanding vapour cloud
explosions (BLEVEs)
● Pressure ruptures of process equipment resulting in
shrapnel damage
● The escape of a large quantity of toxic gas which,
while not necessarily causing significant property
damage, may result in significant business interrup-
tion due to a shutdown pending investigation.
These events do not include sabotage, natural disasters
such as hurricanes or earthquakes, or falling aircraft.
7.4 Maximum Amount Subject
The Maximum Amount Subject (MAS) is defined as
the largest percentage of the total sum insured values
(PD and BI) at the location which is subject to fire,
including attendant heat, smoke or water damage result-
ing from the fire, or efforts to extinguish it, plus the cost
of debris removal if such is included in the treaty. The
only factor which has a limiting effect on PD amount
subject is adequate separation. For BI, the MAS is 100%
for 12 months, unless the surveying engineer feels that
larger values are likely, in which case an explanation
should be provided.
7.5 Maximum Credible Loss
The Maximum Credible Loss (MCL) is defined as the
largest monetary loss that can credibly result from a
single major occurrence of an insured peril at a specific
risk. It is assumed that none of the safeguards are
effective.
7.6 Possible Maximum Loss,
Maximum Possible Loss
The Possible Maximum Loss (PML) or Maximum Possible
Loss (MPL) are defined as the monetary loss which may
occur in extraordinary coincidences of the most disad-
vantageous circumstances with the effect of preventing
or impeding fire-fighting measures, so that the fire con-
tinues to burn until it has exhausted the supply of com-
bustible material or is stopped by impassable obstacles.
9 Chemical Industries Association, Guidelines on Business Interruption
Risk Assessment in the Chemical Industry, August 1992.
EML.2 Broschüre 16.07.1999 10:26 Uhr Seite 15

8 Overview of Various
Concepts

The table below provides a summary of commonly used

terms and the main assumptions which are made.

Term Abbre- Scenario Main Assumptions

viation

Normal Loss Expectancy NLE Normal fire/explosion ● All fire protection systems, including
sprinkler systems, are effective
● Private/public fire-fighting is effective

Maximum Foreseeable Loss
MFL Worst-case fire/explosion ● Primary protection systems are either
impaired or activated only after a delay
● Private/public fire-fighting is delayed
● Damage is limited by adequate
separation of structures

Estimated Maximum Loss
EML Worst-case fire/explosion ● Fire protection systems, including
automatic fire alarm and sprinkler
systems, are not effective
● Private/public fire-fighting is delayed
● Damage is limited by adequate
separation of structures

Probable Maximum Loss PML Worst-case fire/explosion ● No fire protection systems are effective
Maximum Credible Loss MCL ● No effective fire-fighting
Maximum Amount Subject MAS ● Damage is limited only by adequate
separation of structures

Possible Maximum Loss /
Maximum Possible Loss
Catastrophe Loss
PML Worst-case fire/explosion ● No fire protection systems are effective
MPL ● No effective fire-fighting
● Damage is not limited by separation
CL ● Vapour cloud explosion
● BLEVEs
● Natural hazards

Overview of Commonly Used Terms and Main Assumptions

15
EML.2 Broschüre 16.07.1999 10:26 Uhr Seite 16

9 Final Remarks

I
t is clear that the insurance industry needs to de- Even though a variety of terms are used in the insur-
velop standardised methods of estimating loss poten- ance market, EML is the term generally used between
tials. There should be two basic concepts. The first insurers and reinsurers. A wider set of definitions is
should deal with normal losses in which at least found in facultative business, i.e. MFL and PML.
some of the available protection systems are operational ERC Frankona prefers the LIRMA definition of EML. Due
and effective. This concept might also take into account to the nature of the insurance marketplace, however, we
the availability of a rigorous engineering and loss-preven- are compelled to adapt our definition to match that of
tion programme. The second concept should be a worst- our clients. Quite often, we find the concepts MAS, MFL
case estimate. Both concepts should be broad enough to or PML. This compels us to make loss estimates which
deal with PD, LOP and other perils such as earthquake, may be either too conservative or not conservative
windstorm and flood. We can only support this enough for the specific risk.
approach, while emphasising at the same time how
important it is to document any loss estimation advised
with a clear definition.
It should be emphasised that, despite the availability
of any definitions, loss estimates remain subjective esti-
mates which still depend on the experience and skill of
the person making the assessment. In view of the pos-
sibly serious consequences of too optimistic loss esti-
mates, it is advisable to use a conservative method and
to have estimates prepared only by highly experienced
people possessing the requisite technical knowledge.

16
EML.2 Broschüre 16.07.1999 10:26 Uhr Seite 17

Appendix

LIRMA Definition of a “Single Risk” These separations are considered to be adequate to

“A single risk”, for this purpose, means any building or prevent the spread of fire. However, where an inherent
range of buildings not separated either: explosion hazard exists, a perfect party wall (see c)
a from other buildings by an open space (with no tall above) cannot necessarily be considered adequate
grass, timber or any other combustible element) of separation. Moreover, where highly explosive materials
15 metres where there are no opposing openings are used, even greater distances are required to provide
and where the other buildings have external walls of adequate separation.
bricks, stone or concrete (reinforced or otherwise) Where stocks are insured on a “floating” or “declara-
and are roofed with slates, tiles, concrete (reinforced tion” basis, the amount to be taken into account is the
or otherwise) or metal sheeting loss limit on stocks in the “single risk” under considera-
b from other buildings by an open space (with no com- tion or, where there is no loss limit, the total sum insured
bustible element) of 25 metres where there are on stocks. Variation of this rule is allowed only in cases
opposing openings or where the other buildings are where the maximum value of stocks in the “single risk” is
of a construction inferior to that defined in a) above clearly limited by the capacity of the building or buildings,
c from adjacent property by a perfect party wall of in which case the amount to be taken into account shall
brick, stone or concrete (reinforced or otherwise) of not be less than that related to the maximum capacity.
at least 21 centimetres thickness, entirely without Prevailing climatic conditions, e.g. strong winds or
opening and extending at least 37 centimetres above extreme cold, which could have a material effect on fire
the roof of both sides. Where one roof is of concrete, spread must also be taken into account in defining
the party wall need not extend above it. “single risk”.
In buildings separated in accordance with a) or b), an
open-sided covered way, not used for any trade process Used by kind permission of LIRMA
or for storage, would not be considered as forming com-
munication.
In the case of plants in the open (refineries, chemical
factories and the like) or of tank storage, the minimum
open space between units or groups of tanks for the
purpose shall be 25 metres. In addition, storage tanks
must be adequately bonded by container walls sufficient
to hold at least 110% of the maximum capacity of the
tanks.
Wherever there is sloping ground coupled with a risk
which, in the event of fire, could produce burning liquid,
regard should be taken of the fact when assessing “a
single risk”, notwithstanding the existence of separations
as defined above.

17
EML.2 Broschüre 16.07.1999 10:26 Uhr Seite 18

References

Sedgwick Energy Ltd., Vapour Cloud Explosion

Methodology, March 1994.

Recommendations by the former German Association

of Property Insurers, Cologne, on the Determination of
the Probable Maximum Loss with Fire and FLOP Insurance.
Published in the Manual for Industrial Fire and Fire Loss
of Profits Insurance, Reports.

Insurance Institute of London, EML Assessments:

London Market Practice, September 1995.

Reinsurance Offices’ Association (now LIRMA),

Definition of Estimated Maximum Loss, London, 1974.

K. Gerathewohl, Reinsurance Principles and Practice, 1980.

Chemical Industries Association, Guidelines on Business

Interruption Risk Assessment in the Chemical Industry,
August 1992.

Abbreviations

AS Amount Subject
BI Business Interruption
BLEVEs Boiling Liquid Expanding Vapour Cloud Explosions
CL Catastrophe Loss
EML Estimated Maximum Loss
FLOP Fire Loss of Profits
LIRMA London Institute Insurance and Reinsurance
Management Association
LOP Loss of Profits
MAS Maximum Amount Subject
MCL Maximum Credible Loss
MFL Maximum Foreseeable Loss
MPL Maximum Possible Loss
MPL Maximum Probable Loss
MUR Maximum Unit at Risk
NLE Normal Loss Expectancy
PD Property Damage
PML Possible Maximum Loss
PML Probable Maximum Loss
ROA Reinsurance Offices’ Association

18
EML.2 Broschüre 16.07.1999 10:26 Uhr Seite 19

Publisher
ERC Frankona
Rückversicherungs-AG
Munich

Editor-in-Chief
Frank Reichelt
Property Manager
ERC Frankona
Rückversicherungs-AG
Munich

Authors
Kurt Bjørlig
Dr Erika Penzenstadler

Final Editing
Dr Erika Penzenstadler

Editorial Address
ERC Frankona
Rückversicherungs-AG
Maria-Theresia-Strasse 35
D-81675 Munich, Germany
Telephone: +49 (0)89 92 28-0
Telefax: +49 (0)89 92 28-395

Production
EGE-Verlagsservice
Munich

Design
Stahl Grafikbüro
Munich

Printed by
Universal Druck GmbH
Munich

Illustration
Erik Liebermann

Photos
The Image Bank (Title, P. 3 & 8)
dpa (P. 5, 6, 9 & 10)

Copyright 1997 by ERC Frankona.

Text reprinting allowed provided
source quoted and complimentary
copy supplied.

Our experts will be happy

to answer any queries!
Tel. +49 (0)89 9228-716

19