Adair Lewis grasps the nettle of the great unknown and analyses statistics submitted for large loss fires that lack some detail.

THIS MONTH’S statistical review looks at the problem of reports where the building category and other details have not been assigned, together with other peculiarities affecting the database. In the seven years since the FPA started to collect statistics from large loss fires, there have been 514 submissions in which the category of the property has not been included in the data by the loss adjusters who submit the details.

It is not at all clear why there are such omissions. In some cases, the loss adjuster may not have received sufficient information from the insurer to complete the entry, although the insurer should know the information from their portfolio records. In other cases, the building may not have been occupied and thus the type of business being undertaken in the premises may not have been apparent. It will not simply be a matter of the fire being so serious that the type of building could not be determined on site.

In many cases the barest of details seem to have been submitted; just enough to allow an entry to be made, eg the figures show that 83% of the unassigned submissions did not record the cause of the fire either, although the proportions of accidental and deliberate fires in those where it was included were in much the same ratio as for the database as a whole.

Looking at the time of day, the information was omitted in nearly 95% of the cases. Of the instances where the details were provided, the proportions of fires in the various time frames were generally in the same proportions as in the database as a whole, though with some slight deviations in the morning and evening figures.

The figures for the impedances experienced by the fire and rescue services at the incidents supports the view that unassigned fires are not particularly large or problematic. Only three of the 514 are recorded as involving acetylene cylinders and in only four cases were problems experienced with water supplies.

The aspect that does seem to be complete is data relating to financial loss – the records show that unassigned fires account for 17.6% of all monetary losses. It is no doubt for this reason that the submission has been made, as the unassigned category accounts for a whopping £528,537,976 of losses during the period of the survey, which equates to over £75 million a year. In view of this large figure, it would have been good to have learnt more details of the incidents.

Thankfully, there are some positives: although there are a large number of reports where the building category has not been assigned, there are probably few instances where the financial figures have been omitted – indeed, it is probably the extent of the loss that has encouraged the loss adjuster to send in the details. Also, while other details may have been omitted and, looking at the large loss fires as a whole, the absolute figures may not be 100% accurate, the proportions and ratios of numbers are sufficiently accurate to give a realistic overall picture of the problems associated with large fire losses.


Statistical significance

Another factor that is unrelated to the problem above and that occurs in statistical surveys is the size of the groups of data. While there is a danger that large groups, such as unassigned categories, can influence a survey, there is the opposite danger that small groups may be swallowed up and not be noticed at all, even though they may be particularly significant.

Fires in public administration buildings are a case in point. This category covers police, fire and ambulance stations, government offices, security premises, prisons and the like. A search of the category during the seven years of this survey revealed just five records: one in a government office, one in an ‘other’ government building, one in a police station and, ironically, two in fire stations. Three occurred between midnight and 06:00hrs and the other two, inevitably, were unassigned. However, one of these (in a government office rather than a fire station, many will be relieved to note) resulted in a loss of Åí9.25 million. This one fire distorts the statistics to the extent that the mean cost of a fire in a public building is greater than for any other category. One major fire can thus have a significant impact on the statistics as a whole.

Unoccupied buildings

Mention was made above of unoccupied buildings and many businesses, especially in the current economic climate, have unoccupied premises. Hopefully they will be used again in the future or find an alternative use or occupier.

RISCAuthority publishes BDM10: Code of Practice for the Protection of Empty Buildings – Fire Safety and Security, which sets out good practice for the custody of such premises.

Key points include:

  • consider organising temporary use of otherwise unoccupied buildings – especially in the case of small retail premises whereby an occupier, eg a charity, may welcome the use of the property for a nominal rent and its presence may avoid problems associated with the security of void buildings and ensure it remains warm with services in working order
  • minimise potential sources of ignition and the combustible fire load, but maintain power supplies for fire and intruder detection systems
  • maintain the provisions for perimeter security, especially by denying access to flat roofs
  • in some instances, eg in enclosed shopping malls with a 24 hour security presence, it may be acceptable to leave glazing unprotected but in most cases the shutter should be lowered and secured – where impossible or impractical to install a shutter, boarding up may be necessary though this tends to be unsightly, advertises the vacant premises and encourages fly posting, graffiti and possibly even deliberate fire raising

Adair Lewis is technical consultant at the FPA

These statistics are based on information supplied by loss adjusters to the FPA on a voluntary basis and not all insurers conducting business in the UK contribute to this dataset. They represent only sums paid out where the total loss is in excess of £100k and are deficient of losses under £100K, deductibles, underinsurance, uninsured, self-insured and captively insured components, which may be significant. In a year, total losses captured typically account for 50% of the ABI declared annual fire loss figure – which is similarly deficient of the same components (except the £100k threshold).

Share →