Research Letter 23


In retrospect it seems simple, even obvious, but it was not until the 17th century that the idea was formulated:

Fear of harm [is] proportional not merely to the gravity of the harm, but also to the probability of the event” (see Bernstein 1998, p.71).

This is the idea behind ‘Fire Threat Analysis’ (and ‘Standards of Fire Cover’ and a good deal more). Today we would be inclined to restate the idea as:

Threat is proportional not merely to the potential loss but also to the probability of the event causing the loss’.

An example of the sense of the idea can be seen by comparing two areas:

The potential total damage bill, or threat, in both of our areas is here deemed to be the same (i.e. ‘potential loss’ times ‘probability’ is the same).

What if one of the components of threat (i.e. ‘probability of an event’, ‘potential loss in the event’), or just the output (the‘threat’), was used as the single criterion of assessment? What outcomes might we expect?

Firstly, let’s take probability of fire as the only criterion . Remember that in Area 1, there are lots of fires but in Area 2 there are few, so, immediately, we focus on Area 1. There we find that fires are caused predominantly by human carelessness so we mount a campaign to reduce fire-starts there. [Notice that I have introduced ‘cause of fire’ into the discussion, an explanatory variable.]

Secondly, let’s use potential loss per event as our criterion. Immediately, we would focus on Area 2 even though fire events there are rare. On investigation we find that the losses there are due to damage to expensive houses so we might investigate the need for building codes to specify the installation of fire-protection systems in higher value properties there. [Notice the introduction of ‘individual property value’ into the discussion.]

Thirdly, let’s look at threat as the only criterion. Remember that ‘threat’ is the same everywhere in our example. We cannot be drawn to focus on either area but, recognizing the level of threat, distribute new resources for fire suppression evenly across both areas.

Thus knowing the components of the ‘threat analysis’, and the threat itself, could have benefits according to the resources available for mitigating the ‘threat’. All three values are useful.

Threat could be estimated from the historical probability of a fire event and from actual figures for historical losses but this is often not done. Why? Some of the reasons for this may be that:

Having calculated a ‘threat’ we can make comparisons from area to area, as above, and from time to time. Once we try to do this in the real world, we immediately come across questions of equivalence of seasons (due to weather differences for example), areas (due to differences in slopes, aspects, suppression forces, etc.) and fire starts. Some people factor in such differences so they become part of the calculation of threat rather than part of its explanation.

‘Potential for loss’ is related to what values are at stake. The values (economic, social, environmental) can be scored subjectively to create a single number which then is said to represent the ‘potential for loss’. To be even-handed in creating such a value is very difficult because of different perceptions of value held by different people. By using groups of people to determine relative values some of these problems can be overcome but, perhaps better, is the creating a ‘threat index’ for each identified value. Thus there would be a series of ‘threat’ indexes displayed together with their threat-components. It would then be clear to the user what the identified values were and whether their particular ‘value’ was included or overlooked in the analysis. The user could weight the different ‘threats’ according to their own set of values.

Whether we talk of ‘harm’, ‘injury’, ‘damage’ or ‘loss’, and ‘risk’, probability’ or ‘chance’, and ‘threat’ or ‘hazard’, the “simple” idea behind threat analysis is a valuable one.


Bernstein, P. (1998). Against the Gods. The Remarkable Story of Risk. John Wiley and Sons, New York.

A. Malcolm Gill,

November 2000