In this issue:

  • Damage Index 1999


Director: Professor Russell Blong

NHRC is kindly sponsored by:
Swiss Re
QBE Insurance
Benfield Greig Australia
Guy Carpenter & Co



There are numerous types of damage scales with various attributes, qualities, difficulties, and/or advantages.  The scale described here is constructed in such a way that damage to buildings at various locations and from various events can be compared and/or summed.  The Damage Index is based on buildings damaged or destroyed - it doesn't include building contents, cars, machinery, aircraft or crops.  Australian examples are used here but in principle the scale can be used anywhere.  

Buildings, Costs and Replacement Ratios

An average house is assigned a construction cost per square metre of $800 and a Cost Ratio per square metre of 1.00.  Other building types are assigned a Cost Ratio per m2 based on published values for Sydney in 1999 [Rawlinsons, 1999].  The important aspect is not the actual cost per m2 of various types of buildings, but the costs in comparison with a house.  Construction costs for several building types and Cost Ratios are listed in Table 1.  While these relativities refer to Sydney, they provide reasonable proxies for anywhere in Australia and are likely to be reasonably accurate for many other countries.

The Australian $800 /m2 cost for the average house used in Table 1 is a compromise.  Double brick houses are more expensive, weatherboard or timber houses rather cheaper.  The chosen value will produce overestimates for project homes, holiday units, and 2-storey houses, but underestimates for elevated homes, houses with more than 1 bathroom, and split-level houses.

Replacement Ratio [RR] is calculated as:

RR = [Cost Ratio x Area]/House Median Area

Obviously, buildings such as hospitals, hotels, and shopping centres vary enormously in scale.  Where better information about cost ratios or size than that presented here is available, it should be used.  Considerable judgement can be used where additional information is available about specific buildings.

The Replacement Ratios in Table 1 are the most important estimates - these estimates indicate how the costs of replacing a building compare with the cost of replacing a median-sized house.  In effect, RR is the multiple required to express the cost of a building as a number of houses.

Central Damage Values

Numerous studies have set out classes of damage to buildings resulting from natural hazards.  The five damage classes recognised here are each described by a Central Damage Value and a range of values.  The values chosen represent a synthesis from previous studies.  As Central Damage Values [CDV] are used here for calculations there is little need to define the ranges tightly [top of Table 2].

Table 2 defines each CDV for a selection of natural hazards.  In this table each building is categorised according to the worst category of damage sustained.  Values intermediate between the Central Values can be used where deemed appropriate.  Judgement should be used.  The Central Damage Values could be varied from peril to peril if required.

Damage Estimation

From the descriptions above, the damage to an individual building of a specific type is

Damage [HE] = RR x CDV

where RR is the Replacement Ratio [Table 1] and CDV is the Central Damage Ratio [Table 2].  Damage is expressed in House Equivalents [HE].  Thus, a motel in a country town with RR = 7.0 [from Table 1], suffering Heavy damage in an tropical cyclone [CDV = 0.4 in Tables 2]  has Damage [HE]  =  7 x 0.4 = 2.8

More generally,

Damage [HE] = No of Buildings x RR x CDV

For example, as the result of a flood, 40 houses have suffered Moderate damage, 60 houses Severe damage and 10 houses have totally collapsed.  In addition a Motel has suffered Severe damage.  Thus,

Damage [HE]  = [40 x 1.0 x 0.1] + [60 x 1.0 x 0.75] + [10 x 1.0 x 1.0] + [1 x 7.0 x .75]
                       =  4 + 45 + 10 + 5.25
                       =  64.25 HE

The Damage Index [DI]

Using the methodology described above, any damage to buildings can be expressed as damage equivalent to the number of houses totally destroyed.  For a major city severely damaged in an earthquake, damage might reach 1 million HE.

Such large numbers make comparisons different.  It is also recognised that informed estimates have been multiplied by approximations in arriving at HE values.  Here the number of HE is converted to a Damage Index by taking log2 HE.

Log2 HE provides a convenient range of values.  For example,

                 Log2 32 HE     =  5
            Log2 1,024 HE     = 10
        Log2 1,048,576 HE  = 20

The Damage Index for 20 HE = log10 20/log10 2 = 4.32.  Thus, Damage Index, DI = 4.3

Using Log2, the Damage Index would run from 1-20; in the Australian context [and most others], values >10 would rarely be used.  The use of log2 means that total damage must be >1.0 HE before DI>0.0.

DI values can be calculated for a Location, or an Event.  They can also be summed through time, to provide an estimate of the total DI, for example, for all tropical cyclones at a place in a specified period, or to estimate the sum of all damage from one or more perils at a place or in an area in a given period.

The above examples of damage imply that considerable information is available concerning the damage produced by an event.  Frequently, only limited information is available and there are no specific data about the number of buildings involved.  The generic Damage Index described below uses such poor quality data, for Australian situations.

The 1996 Australian Census shows that for all dwelling types the average occupancy per dwelling is 2.65 persons.  Using median floor areas for the most important dwelling types, the weighted mean dwelling size is close to 160m2, or 60 m2/ person.  The Gross Lettable Area in shopping centres in NSW in 1998/1999 was just under 4 million m2, or 0.68m2/ person.  As this total excludes many neighbourhood shops, a first [generous] approximation is to double this area to 1.4m2/person.  An average Cost Ratio[Table 1] for shopping centres is about 2.5.  The equivalent Cost Ratios for office, industrial and public buildings are 2.5, 1.0 and 2.5 respectively.  If we assume, in the absence of better information, that the areas per person of office, industrial and public building space are equivalent to that for retail space, the estimate of all non-dwelling space per person at about 12 m2.

This estimate of 12 m2 per person of non-dwelling space can be combined with the 60 m2 of dwelling space to provide a total of about 72 m2. This is equivalent to 0.4 HE, taking into account all categories of buildings, and providing an estimate for Australian conditions.

The equation HE = 0.4 x Population x Central Damage Value provides an overestimate of damage for underserviced suburbs and an underestimate for commercial areas.  However, these variations are unlikely to be very serious; the estimates above suggest that such generic estimates of HE will generally lie within 20% of actual values.  Using the equation Log2 HE = 0.4 x Population x CDV, an approximate DI can be estimated.

Some Results

In the last two years the Natural Hazards Research Centre has completed the first phase of a major project for the Insurance Foundation, a division of the Insurance Council of Australia.  This PerilAUS database contains more than 4,800 records of hazardous events in Australia this century.  Damage Indices have been constructed for 1191 events all events where sufficient information is judged to be available.

Figure 1 shows the frequency distribution for the DI.  All but 8% of the DI have values <7.  The 22 events with Damage Indices of 9 or more [equivalent to the complete destruction of just over 500 houses] include 6 tropical cyclones, 5 floods, 5 severe thunderstorms, 4 bushfires, and 2 earthquakes, emphasising that Australia is not dominated by the influence of just one natural peril. Queensland and NSW appear most frequently in the list.  Western Australia and the ACT appear not at all, although a DI has been estimated for more than 120 events in WA and for 7 events in the ACT.


Figure 2 indicates the number of events per year for which a Damage Index has been calculated.  The progressive increase in the number toward the present day reflects the improved record of recent events, that it is easier to calculate a DI for recent events, that the record of recent events is better preserved, and the increasing population of buildings at risk in Australia.  Which of these influences is most important is difficult to judge at present.


Figure 3 indicates the total Damage Index for all affected locations where a DI can be calculated for all events this century.  NSW and Queensland compete for poll position, with the ACT right at the back of the grid.  The DI for all Australia for the period is 15.3, equal to the complete destruction of nearly 42,000 House Equivalents.  Together, NSW and Queensland represent 56% of the Total DI, with the Northern Territory and Victoria contributing another 32%.  These values are uncorrected for the area or population of the State or Territory.

On a peril-by-peril basis, the Total DI for affected locations where a DI can be calculated is greatest for tropical cyclones, with floods and bushfires in second place, followed by gusts, hailstorms and earthquakes.  These estimates are summarised in Figure 4. 


Even within Australia, there are significant variations in the cost of construction from State to State.  Rawlinsons' [1999] guide indicates that construction costs on Cape York in Far North Queensland are almost twice those at Brisbane in southeast Queensland.  The DI method removes some of the problem of estimating damage in terms of variations in cost from place to place.

This advantage appears to be retained where inter-country comparisons are made - though further investigation is required.

Obviously, there can be considerable problems with the quality of information about damage, the areal extent of damage, and the degree of damage to individual buildings.  While inaccuracies in estimation affect the actual DI estimate, this is also common where $ values are used.  For example, there are considerable variations in the costs of repairs from one tradesman to another.  Similarly, where large numbers of buildings are damaged in the one event, post-disaster inflation of re-construction costs can amount to as much as 30% of the pre-disaster value.  Expressing building damage in terms of Central Damage Values and as House Equivalents largely eliminates these problems.

Building values change markedly through time.  While we can expect some drift in the value of a House Equivalent with variations in building style and size through periods of decades or more, this drift is much less serious than that imposed by the relentless increase in building values and construction costs.  For example, taking the Melbourne Building Price Index [BPI] as 100 in early 1967, the BPI had reached 1000 by mid-1998.  In the same period the Melbourne Consumer Price Index increased from 100 to 750, and the two curves are far from parallel [Rawlinsons, 1999].

The Damage Index considers only building damage with other important forms of damage excluded.  We would not expect necessarily a close parallel between DI values and records of insured losses.  For example, in the April 1999 hailstorm in Sydney about 38% of the insured damage bill was for motor vehicles and aviation hulls.  Neither of these losses are considered in the Damage Index.

Although new DI information is still being added to PerilAUS, preliminary conclusions about damage to buildings by natural hazards in Australia this century are possible:

* Tropical cyclones, bushfires, thunderstorms, floods and earthquakes all figure importantly in the short list of the most damaging events [DI =9].  Queensland and NSW appear most frequently in this list.

* New South Wales and Queensland have the highest cumulative DI for all events [56%] for which a DI has been calculated, with the Northern Territory and Victoria contributing an additional 31% of the total.

* Tropical cyclones, floods and bushfires dominate the list of natural perils for which DI has been calculated, contributing 73% of the cumulative total.

Past experience suggests that some of these tentative conclusions could easily be altered by a single event - but will it be a flood, cyclone, bushfire, hailstorm or earthquake?  And in which State or Territory?


Rawlinsons, 1999, Australian Construction Handbook, 1999, 17th edition, Rawlhouse Publishing Pty, Perth, 906p.

For further information contact Russell Blong
Telephone: +61-2-9850 9683.  Fax: +61-2-9850 9394

NHQ Homepage
NHRC Homepage