View all publications

3. Costs of New Zealand floods

3.1 Overall Cost

At least 93 major freshwater floods have struck New Zealand since 1858. Cost estimates, via Insurance Council data, begin in 1968 with the storm in which the Wahine was sank (see Appendix B ).

Table 4 shows aggregated (ICNZ and EQC) data for flood costs in New Zealand (BTE, 2001).

Table 4 Aggregated Data for Flood Costs in New Zealand

Insurance costs based on ICNZ and EQC data, rounded to nearest $5m.

Year Cost of Floods Year Cost of Floods
1976 40 1988 25
1977 0 1989 0
1978 40 1990 5
1979 0 1991 5
1980 40 1992 0
1981 30 1993 10
1982 0 1994 10
1983 5 1995 25
1984 110 1996 0
1985 10 1997 10
1986 35 1998 15
1987 0    

Source: BTE, 2001

BTE (2001) notes significant concerns with the data, most notably:

  • Uncertainties regarding the nature of the EQC data: "It was impossible to determine whether the data provided by the EQC represented total insurance cost, total residential insurance cost or the insurance cost for residents who held insurance policies with the Commission."
  • Data gaps-e.g. no floods are recorded prior to 1976-indicating events not recorded or a lack of significant insurance coverage where the event occurred.
  • Incomplete information on the inflation-adjustment procedure used by ICNZ and EQC. BTE (2001) combined the ICNZ and EQC datasets by estimating an inflation-adjustment method, noting that their choice might or might not be appropriate.

Further evidence, as shown in Appendix B, supports BTE's cautions. Some events in Appendix B are listed in years where BTE's (2001) combination of ICNZ and EQC data yielded zero losses. Similarly, the 1988 figures in Table 4 do not include the Cyclone Bola insurance payout listed in Appendix B . These disparities highlight the challenge in separating storm costs from flood costs.

The data available are incomplete in that the insurance costs dominate the values given. Furthermore, to expand BTE's (2001) critique of the EQC data available, the insurance cost does not indicate what items are included. Insured business interruption costs, life and medical insurance payouts, and insured agricultural losses cost might or might not be included in the ICNZ values. Given that some of the ICNZ events in Appendix B (those labelled "floods assumed") are identified by location only, not by disaster type, detailed research into ICNZ's archives would be necessary to fully answer the questions regarding their data.

EM-DAT (2004) is the only other source providing cost estimates for many flood events, but those values are suspect. For example, for the 14 January 2002 Canterbury flood, EM-DAT (2004) reports that 300 people were affected for a total cost of USD500. A cost of less than NZD5 per person seems unlikely, particularly with an insurance payout of NZD21.5 million not including EQC claims (ICNZ, 2004).

To explore some of these costing issues in detail, specific events were sought to try to establish the breakdown in costs (refer section 3.2).

3.1.1 Flood-loss trends

Ericksen (1986) evaluated Earthquake and War Damage Commission (now EQC) and insurance claims from the early 1950s through to 1984 to attempt to establish the trend in flood losses over the period. The data from his analysis, with losses measured in 1984 dollars, is replicated in Table 5 below.

Ericksen hypothesised from discussions with EQC that the contribution of the Disaster Fund to total insured losses may have been declining over the period. However, as he notes, the percentage of losses covered by EQC shown in Table 5 is highly variable, and not immediately suggestive of a declining trend. Furthermore, he describes a number of deficiencies in the data sources employed, and as a consequence is unable to identify with confidence a trend in the level of flood losses.

Table 5 Total flood damage, 1957-84

View total flood damage, 1957-84 (large table).

Ericksen concludes by saying:

[Between 1968 and 1984] about $0.25 billion (1984 dollars) was paid out by the insurance industry for 9 major floods, including nearly $50 million from the Disaster Fund of the Earthquake and War Damage Commission. On this basis, the cost to the nation for direct losses may have been over $1 billion, and perhaps $1.5 billion overall - an average annual loss of around $90 million (1984 dollars). The losses seemed to be increasing throughout the 16 year period. (p. 264)

Ericksen's valuation in 1984 dollars of $1.5 billion translates to over $2.1 billion in today's prices. Similarly, Ericksen's annual cost estimate is equivalent to around $128 million in today's prices.

By comparison, BTE (2001) estimate that floods in Australia cost A$10.4 billion (in 1999 dollars) between 1967 and 1999. The average annual cost is therefore a little more than A$452 million, which in today's terms equates to around A$510 million per annum (or around NZ$580 million per annum) [Strictly speaking, each event’s loss should be converted to the New Zealand dollar at the rate prevailing at the time that the event took place. The conversion here is thus indicative only.].

BTE (2001) also estimate the cost of natural disasters, including floods, in New Zealand. They find that the total cost of all natural disasters in New Zealand from 1962 to 1998 is approximately NZ$1.2 billion, with an approximate annual average cost since 1980 of NZ$43 million. They state, however, that "the severe limitations of the data mean that this is likely to be a significant underestimate".

It is difficult to assess the accuracy of Ericksen's estimate of annual average New Zealand flood losses. If we assume though that BTE's estimate of Australian losses is robust, then the relative sizes of the two countries, as proxied by relative populations or economy sizes, then Ericksen's estimate would at least appear to be in the right order of magnitude.

Table 6 presents an amalgam of Ericksen's (1986) data (covering 1957-1984) and BTE's (2001) data (covering 1976-1998), all expressed in 2004 dollars.

Table 6 New Zealand flood losses, 1957-1998

2004 NZ$

Year NZ$ Year NZ$

1957

23,864,180

1978

57,927,777

1958

40,886,133

1979

0

1959

0

1980

12,712,496

1960

8,277,112

1981

20,096,308

1961

4,487,754

1982

0

1962

0

1983

8,910,742

1963

0

1984

132,596,703

1964

0

1985

11,036,889

1965

0

1986

38,629,113

1966

0

1987

0

1967

0

1988

27,592,223

1968

191,899,448

1989

0

1969

0

1990

5,518,445

1970

0

1991

5,518,445

1971

0

1992

0

1972

0

1993

11,036,889

1973

0

1994

11,036,889

1974

0

1995

27,592,223

1975

78,312,343

1996

0

1976

57,347,466

1997

11,036,889

1977

0

1998

16,555,334

Source: Ericksen (1986), BTE (2001), SNZ

The absence of any discernible trend in these values is highlighted in Figure 4. However, it is also evident that there are clear gaps in the data series. As noted by BTE (2001):

Another problem with the data was the large gap in the number of events recorded in the Insurance Council's database between 1968 and 1975. Considering the frequency of events that occurred after this period, it is highly unlikely that no disasters occurred during this time. For example, flooding in New Zealand is a significant problem with an impact almost every year. However, the first record of a flood in the database is 1976. Some possible reasons why the data lacks records for this period are the lack of media attention, the availability of residential flood insurance or flooding occurring in non-residential areas.

Another particularly notable exclusion is Cyclone Bola (1988). This is almost certainly recorded in Insurance Council data as a severe storm rather than a flood event. This illustrates the problems associated with pigeon-holing events with varying impacts into a single category.

Figure 4 New Zealand flood losses 1958-1998

Thumbnail of image. See figure at its full size.

3.2 Distribution of losses

This section draws on two detailed flood event analyses to illustrate the possible distribution of flood losses. Given the discussion thus far regarding the paucity of New Zealand flood loss assessments, it is probably unsurprising that in-depth analyses of the impacts of individual events are rare indeed. Ericksen (1986) combines the effects of the Nelson and New Plymouth floods of 1970 and 1971 into a single event to illustrate the disbursement of flood losses. This is re-presented here. This section also draws on the analysis by Walton et al (2004) of the Waikato Weather Bomb event of June 2002.

3.2.1 Nelson and New Plymouth, 1970 and 1971

The following table shows the estimates presented in Ericksen (1986), which were in turn based largely on Howard (1973).

Table 7 Flood losses for Nelson and New Plymouth, 1970 and 1971

NZ$ million, 2004 dollars

Direct losses    
Central government works and services    
Roading 28.0  
Railways 3.7  
Bulk power supply 0.7  
Flood control and drainage works 11.4  
Sub-total   43.8
Local government works and services    
Roading 4.0  
Flood control and drainage 3.9  
Water, sewage and telecommunications 2.7  
Sub-total   10.6

Private sector

   
Farm land 0.5  
Disaster Fund payouts 5.7  
Insurance industry payouts 17.1  
Uninsured property 11.4  
Sub-total   34.7
Total direct losses   89.1

Indirect losses

   

Income and production(1)

13.3  

Total flood losses

  102.5

Notes: (1) Based on the view that the value of indirect losses probably conservatively equates to around 15% of direct losses (Ericksen, 1986, p. 82)

Source: Adapted from Ericksen (1986), Howard (1973), SNZ

Briefly, Nelson and New Plymouth experienced severe flooding in August 1970 and February 1971, respectively. In Nelson losses were caused by a 50 year flood, while New Plymouth was struck by a 100 year flood event.

Notably, this flood event is absent from the Insurance Council and EQC data presented in Table 4 and Table 6. Further, the values of losses from the Nelson and New Plymouth events show that losses backed by private insurance and Disaster Fund claims represent around 65% of total direct losses and just 22% of the estimated total flood losses.

3.2.2 The Waikato Weather Bomb, 2002

The following table, drawn from Walton et al (2004), shows similar analysis of the Waikato Weather Bomb event of June 2002.

Table 8 Flood losses for the Waikato Weather Bomb, 2002

NZ$ million, 2004 dollars

Direct costs  
Insured losses 8.0
Uninsured losses 2.1
Response agency costs 3.1
Total direct costs 13.2
Indirect costs  
Business disruption losses 0.0
Insurance excess payments 0.5
Total indirect costs 0.5
Total costs 13.7

Source: GNS/NZIER

EQC payouts for the Weather Bomb were slightly less than $1 million, and are additional to costs listed above. The Weather Bomb event provides a useful contrast to the Nelson/New Plymouth floods in terms of flood loss distribution. Of the $14.7 million total cost of the Weather Bomb, $9.0 million, or over 60%, was borne by private insurance or EQC. This compares to 22% for the much larger Nelson/New Plymouth events. Further, the percentage of indirect costs to total costs for the Weather Bomb flood was much lower (at less than 4%) than that for the Nelson/New Plymouth flooding (at 15%). Although this latter difference reflects in part the differing methodologies used to determine the indirect losses, it seems plausible that the greater the extent of the flooding, the greater the likely incidence of business disruption and lifeline breakages i.e. as a general rule, the greater the extent of flooding, the greater the indirect effects.

3.2.3 Generalised loss distribution

Ericksen (1986) assumes that the loss distribution of the Nelson/New Plymouth flooding holds for nine major floods from 1968 to 1984 to provide an indication of the magnitude and direction of loss disbursement. From this generalisation exercise, Ericksen concludes that:

Setting aside the human misery and anguish caused by flood disasters, the key point to emerge...is that most of the direct flood losses are ultimately borne by agencies beyond the stricken community, perhaps as much as 85 percent. Only about 15 percent of direct property losses are borne by the territorial local government.

The Weather Bomb event is a good example of the distribution of loss from flooding. Table 8 shows that of the $13.7 million total cost, $8 million, or 58% of the total, was borne by insurers. Further, although response agency costs are listed as being $3.1 million, it is not known how much of this was offset by central government disaster funding. Of the direct costs, therefore, it is only certain that $2.1 million, or around 16% of the total direct costs, were incurred locally.

By comparison, a study of the Queensland January 1998 floods, found that roughly 50% of losses arising from the flood were borne by the local community [J.W. Handmer and O. Percovich (2002)]. Of the $A245.1 million total loss, $A69.4 million was covered by private insurance, and $A$52.6 million by the Natural Disaster Relief Arrangements fund.

The Insurance Council's own estimates are that between 25% and 40% of property is under- or uninsured. The converse suggests that anywhere between 60% and 75% of all direct losses are covered by insurance policies, and are thus not incurred within the region of the event. Indirect tangible losses are more likely to be met locally; business disruption losses are typically not well insured against.

Ericksen (op cit) also notes that many of the resources required to restore a flood-affected community come from outside the region. In effect, communities externalise many of the costs of its floodplains development. Ericksen argues, however, that much of the flood loss is caused by the "locational decisions" of central government agencies, and thus "agencies from the outside, including central government, must share the blame for creating community flood hazards and consequent losses along with the local community."

3.3 Limitations

These events' indicativeness of what to expect from climate change is unknown. Indicativeness must also be considered of the storm system, the resultant flooding, and the costs caused by the flooding-one being normal does not necessarily imply that the other two would be normal and vice versa. Finally, in dealing with extreme events, describing how "normal" or "expected" they would be is somewhat incongruous. The short data set published for New Zealand adds to the challenge: effectively 150 years of observations, although proper examination of Māori history could significantly expand that, but only approximately 50 years of systematic observations. A systematic investigation of Māori history to extract and describe flood events over the centuries of settlement would be a useful future project.

For the February 2004 floods, NIWA (2004) writes:

Traditionally, February is a settled month for the North Island, dominated by anticyclones (but with a small risk of a subtropical low bringing heavy rain). It is usually the warmest and driest time of year for many parts of the North Island. But February 2004 was one for the record books - the monthly rainfall was four to six times typical February amounts from the Waikato to Wellington, and also in the Wairarapa. It was very cold, with record-low February temperatures recorded in some inland and southern parts of the South Island. And consistently strong westerly winds affected the country - it was the windiest month over the North Island since monitoring started in 1941.

Thus, the inclusion of the February 2004 floods as a "normal" or indicative extreme event could be questioned. Care must be taken before reaching such a conclusion. The weather around North Island in February 2004 was unprecedented compared to past the Februarys for which data exist, but was merely extreme compared to winter months (NIWA, 2004). Economic impacts depend on the month in which an extreme event strikes and the effects on tourism and agriculture were likely exacerbated by (a) the event striking in February and (b) the lack of expectation of, and preparation for, a major flood occurring in February. Classifying the floods as abnormal and unprecedented for New Zealand, though, might be less accurate than classifying the economic impacts as abnormal and unprecedented for New Zealand.

Table 9 illustrates the situation for four locations. February 2004 rainfalls which were 3-4 times the normal February rainfall are far less extreme when compared to winter month rainfalls. If climate change increases winter rainfall by 10-20%, as is feasible in these locations, the February 2004 rainfall levels could become relatively common in winter months even if they never again recur in February.

Table 9 Monthly Rainfall Statistics for some North Island Locations

Location Mean February Rainfall (mm) February 2004 Rainfall (mm) Highest Mean Winter Month Rainfall (mm)
Hamilton 85 332 126
Kaitaia 82 309 166
Taupo 77 181 109
Wellington 72 291 147

Source: NIWA, 2004

Given the above discussion, the February 2004 floods should not be dismissed as an outrageously unusual event, but the economic impacts are likely to be higher than what would occur in most circumstances given the present state of vulnerability. Thus, the February 2004 floods are accepted as establishing an upper limit to the economic impacts which would be expected from a storm-related freshwater flood in New Zealand [Initial estimates by MAF suggest that the cost of the February 2004 floods to farmers will touch $180 million and the Government pledged $130 million in relief funding. A more comprehensive analysis has been commissioned by Horizons Regional Council, and is due for release soon.].