Part 3: Results

The headlines

34
We analysed only the appropriations where we had high or medium confidence in their relevance to natural hazards. The main findings from our analysis of this data are as follows:

• We identified $11.4 billion of government expenditure related to natural hazards between 2009 and 2018, which is 1.36% ($11.4 billion out of $837.7 billion¹⁰) of the total government expenditure authorised through appropriations during these 10 years.
• We categorised $5.6 billion (49%) of the $11.4 billion expenditure identified as relating to hazard response and recovery, $2.8 billion (24%) was categorised as relating to risk reduction, and $3 billion (27%) was categorised as other civil defence.
• In 2011, after the Canterbury Earthquakes, we identified in the appropriation data a $2.5 billion expenditure related to natural hazards, the highest of the 10-year period.
• By types of hazard, “Earthquake” has the largest identifiable expenditure during the period, and then “Weather”, “Flood”, “Erosion”, “Tsunami”, and “Bushfire”. The Insurance Council of New Zealand has earthquakes as the most costly type of natural hazard in New Zealand.¹¹
• Risk reduction expenditure appears to follow a three-year cycle. We identified about $24 million for each year between 2012 and 2014. Between 2015 and 2017, this increased to about $600 million for each year, and then dropped back to about $24 million in 2018. This pattern is not consistent with the response and recovery expenditure, which has a more steady trend. 
• In the last 10 years, 21 out of the 39 government departments incurred or administered some expenditure related to natural hazards. This might suggest that natural hazards result in, or require expenditure across, most government departments.
• The third highest expenditure that we identified in the appropriation data was incurred or administered by the Ministry of Social Development. This does not mean that it incurred the third most expenditure, but it does indicate that the appropriation data for the Ministry includes a larger number of specific references to natural hazards and response or risk reduction expenditure. 
• Some government departments incurred or administered expenditure related to natural hazards consistently throughout the 10 years (for example, the Ministry of Social Development and the Ministry for Primary Industries), and some government departments appear to have incurred or administered more one-off identifiable expenditure (for example, the New Zealand Defence Force and the Ministry of Justice).
• The proportion of identifiable natural hazard expenditure relative to total government expenditure authorised through appropriations has been relatively stable since 2011 (about 1-2%). Identifiable expenditure related to natural hazards before 2011 was significantly lower (less than 0.5% of total government expenditure authorised through appropriations).

35
It is worth noting that our searches returned no matches for the following key words:

• Types of natural hazards: tornado, cyclone, volcano, eruption, snow, wind, drought, and rain.
• Natural hazard events in New Zealand: Gita, Fehi, Pam, and Lusi.

36
This might indicate that the expenditure related to these natural hazards was not significant enough to need a specific appropriation, or to reference within the scope statement of another appropriation. Expenditure associated with these hazards might have been included within a broader appropriation.

Detailed findings

37
The following tables and graphs summarise identifiable expenditure by votes, administering departments, types of hazards, and types of focus throughout the 10-year period (Figures 6 to 16). The “Total value” fields below are the total identifiable expenditure against appropriations categorised with confidence level high and medium. Amounts for low confidence level are presented where possible as additional information.

By confidence level and type of expenditure

Figure 6
Breakdown of identifiable expenditure, by type of expenditure and confidence level

The table shows a breakdown of expenditure by type of expenditure and confidence level. Of the $11.4 billion expenditure, we categorised $5.6 billion (49%) as relating to response and recovery, $2.8 billion (24%) as relating to risk reduction, and $3 billion (27%) was categorised as “other civil defence”.

Confidence level	Rounded total  $million	Response and recovery $million	Risk reduction $million	Other civil defence $million	Unclear $million
High	4,872	4,871.9	-	-	-
Medium	6,513	682.9	2,791.0	3,039.2	-
Total value*	11,385	5,554.8	2,791.0	3,039.2	-
Low	3,455	576.5	41.6	-	2,836.5

* Total values do not include low-confidence level amounts.
Source: Analysed from the enriched appropriation data.

By vote

Figure 7
Breakdown of identifiable expenditure, by vote, administering department, and confidence level

The table shows a breakdown of expenditure in $millions by vote, administering department, and confidence level. The intention is to show the votes that had the most identifiable natural hazard expenditure. A vote is a collection of appropriations. Note that a department can administer multiple votes. We identified $3.3 billion for Vote Finance (for the Treasury), which is the biggest expenditure of all votes.

			Confidence level
Vote	Administering department	Total value* $million	High $million	Medium $million	Low $million
Finance	The Treasury	3,273.7	62.2	3,211.5	69.1
Canterbury Earthquake Recovery	Department of the Prime Minister and Cabinet	3,151.5	3,151.5	-	-
Social Development	Ministry of Social Development	3,066.4	216.6	2,849.8	160.0
Canterbury Earthquake Recovery (old)	Department of the Prime Minister and Cabinet	567.4	567.4	-	-
Transport	Ministry of Transport	477.0	280.0	197.0	555.9
Tertiary Education	Ministry of Education	210.1	210.1	-	7.5
Prime Minister and Cabinet	Department of the Prime Minister and Cabinet	175.6	33.3	142.3	1.0
Emergency Management (old)	Department of Internal Affairs	134.6	89.2	45.4	-
Lands	Land Information New Zealand	102.2	102.2	-	-
Building and Housing	Ministry of Business, Innovation and Employment	73.9	70.9	3.0	0.8
Internal Affairs	Department of Internal Affairs	21.4	2.6	18.8	-
Defence Force	New Zealand Defence Force	16.2	16.2	-	-
Education	Ministry of Education	14.7	14.7	-	-
Conservation	Department of Conservation	14.5	14.5	-	35.0
Justice	Ministry of Justice	14.4	1.2	13.2	117.1
Agriculture, Biosecurity, Fisheries and Food Safety	Ministry for Primary Industries	13.1	-	13.1	257.4
Agriculture and Forestry (old)	Ministry for Primary Industries	13.0	-	13.0	126.3
Arts, Culture and Heritage	Ministry for Culture and Heritage	11.8	10.0	1.8	140.1
Business, Science and Innovation	Ministry of Business, Innovation and Employment	10.2	10.2	-	198.1
Revenue	Inland Revenue Department	9.4	9.4	-	0.1
Foreign Affairs and Trade	Ministry of Foreign Affairs and Trade	5.5	5.5	-	-
Primary Industries and Food Safety	Ministry for Primary Industries	3.9	-	3.9	105.8
Statistics	Statistics New Zealand	1.7	1.7	-	-
Labour Market	Ministry of Business, Innovation and Employment	1.1	1.1	-	-
Defence	Ministry of Defence	1.1	1.1	-	-
Parliamentary Service	Parliamentary Service	0.2	0.2	-	-
Ombudsmen	Office of the Ombudsman	0.1	0.1	-	-
Health	Ministry of Health	0.3	-	0.3	910.1
Environment	Ministry for the Environment	-	-	-	359.3
Science and Innovation	Ministry of Business, Innovation and Employment	-	-	-	347.9
Climate Change	Ministry for the Environment	-	-	-	48.2
Māori Development	Te Puni Kōkiri	-	-	-	10.2
Energy	Ministry of Business, Innovation and Employment	-	-	-	4.7

* Total values do not include low-confidence-level amounts.
Source: Analysed from the enriched appropriation data.

By administering department

Figure 8
Breakdown of identifiable expenditure, by administering department and confidence level

The table shows a breakdown of expenditure in $millions by administering department and confidence level. The intention is to show the administering departments that had the most identifiable natural hazard expenditure. Three government departments – the Department of the Prime Minister and Cabinet, the Treasury, and the Ministry of Social Development – each had more than $3 billion expenditure related to natural hazards, while the other government departments each had less than $500 million expenditure related to natural hazards that we could identify.

		Confidence level
Administering departments	Total value* $million	High $million	Medium $million	Low $million
Department of the Prime Minister and Cabinet	3,894.5	3,752.2	142.3	1.0
The Treasury	3,273.7	62.2	3,211.5	69.1
Ministry of Social Development	3,066.4	216.6	2,849.8	160.0
Ministry of Transport	477.0	280.0	197.0	555.9
Ministry of Education	224.8	224.8	-	7.5
Department of Internal Affairs	156.2	91.9	64.3	-
Land Information New Zealand	102.2	102.2	-	-
Ministry of Business, Innovation and Employment	79.8	76.8	3.0	551.6
Ministry for Primary Industries	30.0	-	30.0	489.5
New Zealand Defence Force	16.2	16.2	-	-
Ministry of Justice	14.5	14.5	-	35.0
Department of Conservation	14.4	1.2	13.2	117.1
Ministry for Culture and Heritage	11.8	10.0	1.8	140.1
Inland Revenue Department	9.4	9.4	-	0.1
Ministry of Foreign Affairs and Trade	5.5	5.5	-	-
Ministry of Economic Development (old)	5.4	5.4	-	-
Statistics New Zealand	1.7	1.7	-	-
Ministry of Defence	1.1	1.1	-	-
Ministry of Health	0.3	-	0.3	910.1
Parliamentary Service	0.2	0.2	-	-
Office of the Ombudsman	0.1	0.1	-	-
Ministry for the Environment	-	-	-	407.4
Te Puni Kōkiri	-	-	-	10.2
Rounded total	11,385	4,872	6,513	3,455

* Total values do not include low-confidence-level amounts.
Source: Analysed from the enriched appropriation data.

Figure 9
Proportion of identifiable expenditure related to natural hazards, by administering department

Across the 10-year period, the Department of the Prime Minister and Cabinet administered the most expenditure related to natural hazards. The top three government departments (the Department of the Prime Minister and Cabinet, the Treasury, and the Ministry of Social Development) are responsible for administering 90% of the identifiable expenditure related to natural hazards.

Source: Analysed from the enriched appropriation data.

By administering department and year

Figure 10
Breakdown of identifiable expenditure, by administering department and financial year

The table shows a breakdown of expenditure in $millions by administering department and year. The Department of the Prime Minister and Cabinet administered the most expenditure related to natural hazards in 2011 ($1 billion) and then the expenditure slowly decreased. Some government departments administered expenditure related to natural hazards consistently throughout the 10 years (for example, the Ministry of Social Development and the Ministry for Primary Industries), and some government departments appear to have more one-off identifiable expenditure (for example, the New Zealand Defence Force and the Ministry of Justice).

Administering department	2009 $m	2010 $m	2011 $m	2012 $m	2013 $m	2014 $m	2015 $m	2016 $m	2017 $m	2018 $m
Department of the Prime Minister and Cabinet	-	-	1,045.6	741.7	871.3	494.5	362.7	241.0	79.3	58.5
The Treasury	-	-	878.7	-	-	-	666.0	712.1	823.4	193.5
Ministry of Social Development	227.3	270.4	471.6	263.9	271.3	270.8	276.9	289.6	369.9	354.7
Ministry of Transport	17.5	18.6	18.6	63.6	44.8	50.8	30.6	36.2	32.9	163.4
Ministry of Education	-	-	14.7	0.1	-	10.0	100.0	85.0	15.0	-
Department of Internal Affairs	10.5	10.2	21.8	93.0	10.4	8.4	-	-	0.5	1.3
Land Information New Zealand	-	-	-	-	0.2	-	-	28.2	42.3	31.6
Ministry of Business, Innovation and Employment	-	-	8.1	21.6	15.4	14.4	6.5	3.0	6.2	4.4
Ministry for Primary Industries	3.5	4.2	3.3	2.1	2.2	1.7	2.1	2.1	5.8	3.1
New Zealand Defence Force	-	-	-	-	-	-	-	-	16.2	-
Ministry of Justice	-	-	14.5	-	-	-	-	-	-	-
Department of Conservation	0.2	0.9	7.1	6.2	-	0.1	-	-	-	-
Ministry for Culture and Heritage	-	1.5	2.3	4.9	0.5	2.4	-	-	-	0.3
Inland Revenue Department	-	-	3.2	6.2	-	-	-	-	-	-
Ministry of Foreign Affairs and Trade	0.5	-	5.0	-	-	-	-	-	-	-
Ministry of Economic Development (old)	-	-	3.4	2.0	-	-	-	-	-	-
Statistics New Zealand	-	-	1.7	-	-	-	-	-	-	-
Ministry of Defence	-	-	-	-	-	-	-	-	1.1	-
Ministry of Health	-	-	-	0.3	-	-	-	-	-	-
Parliamentary Service	-	-	0.2	0.0	-	-	-	-	-	-
Office of the Ombudsman	-	-	0.1	-	-	-	-	-	-	-

Note: The cumulative totals across years may not match the totals in Figure 8 due to rounding.
Source: Analysed from the enriched appropriation data.

By type of expenditure and year

Figure 11
Breakdown of identifiable expenditure, by type of expenditure and year

The table shows a breakdown of expenditure in $millions by type of expenditure and year. Risk reduction expenditure appears to follow a three-year cycle. We identified about $24 million for each year between 2012 and 2014. Between 2015 and 2017, this increased to about $600 million for each year, and then dropped back to about $24 million in 2018. This pattern is not consistent with the response and recovery expenditure, which has a more steady trend.

Year ended 30 June	Risk reduction $million	Response and recovery $million	Other civil defence $million	Total natural hazards expenditure $million	Percentage of total appropriated expenditure
2009	17.7	4.0	237.9	259.6	0.27%
2010	20.9	4.2	280.6	305.7	0.42%
2011	862.8	1,352.0	285.0	2,499.8	3.11%
2012	26.8	913.6	281.7	1,205.4	1.55%
2013	20.8	914	282	1,216	1.54%
2014	21.7	550.6	280.8	853.1	1.06%
2015	686.7	468.4	289.6	1,444.7	1.74%
2016	467.3	624.7	305.2	1,397.2	1.62%
2017	640.0	340.7	411.8	1,393.5	1.56%
2018	26.3	393.9	390.5	810.7	0.88%
Rounded total	2,791	5,555	3,039	11,385	1.36%

Note: Percentage of total government department expenditure is the proportion of the total appropriated expenditure according to the appropriation data for that year.
Source: Analysed from the enriched appropriation data.

Figure 12
Proportion of identifiable expenditure, by type of expenditure

Of the total $11.4 billion expenditure, we categorised $5.6 billion (49%) as relating to hazard response, $2.8 billion (24%) as relating to risk reduction, and $3 billion (27%) as other civil defence.

Source: Analysed from the enriched appropriation data.

Figure 13
Trend of identifiable expenditure, by type of expenditure and year

The bar graph below is a visual way to show the information in Figure 11. In 2011, after the Canterbury Earthquakes, we identified $2.5 billion expenditure related to natural hazards, which is the highest of the 10-year period.

Source: Analysed from the enriched appropriation data.

By type of hazard and hazard event

Figure 14
Breakdown of identifiable expenditure, by type of hazard and hazard event

The table below shows a breakdown of expenditure in $millions by type of hazards and hazard events. By type of hazards, “Earthquake” has the largest identifiable expenditure during the period, and then “Weather”, “Flood”, “Erosion”, “Tsunami”, and “Bushfire”.

Hazard event	Earthquake $m	Weather $m	Flood $m	Erosion $m	Earthquake, tsunami $m	Bushfire $m	Unknown $m	Total value $m
2009 Victoria bushfires	-	-	-	-	-	0.5	-	0.5
2011 Queensland floods	-	-	4.0	-	-	-	-	4.0
2010/11 Canterbury earthquakes	5,498.6	-	-	-	-	-	-	5,498.6
2011 Great East Japan earthquake	-	-	-	-	1.0	-	-	1.0
2016 Kaikōura earthquake	38.1	-	-	-	-	-	-	38.1
2017 Edgecumbe flood	-	-	0.3	-	-	-	-	0.3
Total for events specified	5,536.7	-	4.3	-	1.0	0.5	-	5,542.5
Unknown	2,567.4	197.0	20.9	18.0	-	-	3,039.2	5,842.5
Rounded total	8,104	197	25	18	1	1	3,039	11,385

Source: Analysed from the enriched appropriation data.

By type of hazard and year

Figure 15
Breakdown of identifiable expenditure, by type of hazard and year

Across the 10-year period, 2011 was the first year to show a significant expenditure related to natural hazards of which types of hazards are not in the category of “unknown”. From 2011, “earthquake” remains the hazard type with the largest identifiable expenditure during the period.

Source: Analysed from the enriched appropriation data.

Figure 16
Breakdown of identifiable expenditure, by type of hazard and year

The table below shows a breakdown of expenditure in millions by type of hazard and year. Expenditure related to earthquake has been significant since 2011, while the expenditure related to erosion and weather is consistent over the years at about $3 million and $20 million respectively.

Year ended 30 June	Earthquake $m	Bushfire $m	Earthquake, tsunami $m	Erosion $m	Flood $m	Weather $m	Unknown $m	Total value $m
2009	-	0.5	-	-	3.7	17.5	237.9	259.6
2010	1.5	-	-	-	5.0	18.6	280.6	305.7
2011	2,182.0	-	1.0	3.3	10.0	18.6	285.0	2,499.9
2012	902.6	-	-	2.1	6.2	18.6	276.0	1,205.5
2013	913.6	-	-	2.2	-	18.6	281.7	1,216.1
2014	550.6	-	-	1.7	-	20.0	280.8	853.1
2015	1,134.4	-	-	2.1	-	18.6	289.6	1,444.7
2016	1,068.7	-	-	2.1	-	21.2	305.2	1,397.2
2017	956.6	-	-	2.0	0.2	21.9	411.8	1,392.5
2018	394.1	-	-	2.6	0.2	23.4	390.5	810.8
Rounded total	8,104	1	1	18	25	197	3,039	11,385

Source: Analysed from the enriched appropriation data.

Limitations of our analysis

38
We know expenditure related to natural hazards will be greater than what we were able to identify from this data set for several reasons. These include:

• Our analysis was limited to only the information provided in the appropriation data set. This was sourced from Estimates data on the Treasury’s website. We note that appropriation scope statements are required to be short and do not include performance information or results from the appropriation expenditure. As such, important supporting information and detail will not be derived from analysis of appropriation scope statements alone.
• Our methodology means that it is likely that we have mainly identified expenditure that is directly related to natural hazards. There are many other indirect costs that the Government incurs to manage and respond to the risks associated with natural hazards that we are unlikely to have identified in this analysis (for example, costs to the health sector associated with injury, or economic costs associated with loss of employment and tax revenue to the Crown). Where expenditure related to natural hazards as been incurred through an appropriation where the appropriation title or scope statement is broad or general, this expenditure might not have been identified as being hazard-related.
• Our methodology used "fuzzy" search techniques – that is, searches were not limited to the exact search term. Searches will pick up partial words in a search term and whole phrases that contain a search term. After each search, we also reviewed a sample of results to refine our key words and search criteria. This allowed us to minimise the risk of false positives and false negatives. Some risk remains, but we consider it unlikely to have significantly affected our analysis.
• We did not find specific mention of some natural hazards that we know happened in the last 10 years from the appropriation data (for example, Gita, Fehi, and Pam). Further research, such as examination of historical Cabinet papers or wider analysis of budget documents or other accountability documents, might identify expenditure related to these events.
• There are likely to be other relationships that we did not identify because of the limitations of our methodology. We have not carried out wider engagement with subject matter experts (for example, officials within the Treasury or the National Emergency Management Agency), which we consider would aid the development of this work. This document is exploratory work that we intend to use internally as we scope our planned work on risk and resilience. We welcome further feedback and might carry out further work that could include wider engagement with subject matter experts to improve the analysis.
• We acknowledge that large areas of expenditure and liabilities associated with the 2010/11 Canterbury Earthquakes (in particular, Earthquake Commission insurance claims¹²) have not been fully captured in this analysis. Expenditure we have identified in the appropriation data amounts to $5.5 billion. In 2017, the Government's financial statements had already identified $15 billion of expenditure.¹³
• We acknowledge that there is an international standard on risk management (ISO 31000¹⁴), and that the 4Rs framework is more widely used in New Zealand. We might align any future work more closely with these.

39
We could do further work to address the limitations set out above, such as: 

• further review and analysis of the appropriation data where we have a low level of confidence to determine whether it should have been included – for example: 
  • a similar search and analysis of Cabinet minutes in the past 10 years might identify relationships not visible in this data set – for example, if expenditure has been incurred under broad appropriations that were not picked up by key word searches; and
  • more in-depth analysis of government department annual reports might provide more information about the activities of government departments and insight as to whether expenditure related to natural hazards has occurred; and
• analysing information from other out-of-cycle funding mechanisms (such as between budget contingencies) and baseline updates might identify additional relevant expenditure.

40
The initial scope of our analysis was limited to expenditure information from the appropriation data. As well as the limitations already acknowledged for expenditure, we are aware that councils invest heavily in risk reduction and treatment, such as flooding and stormwater prevention. We might carry out additional work to examine their long-term plans and financial statements.

Quality assurance

41
The method we used went through several iterations of development and testing, and each iteration was internally reviewed by the Senior Data Analyst and the Director, Data and Analytics at the Office of the Auditor-General.

42
We had all the scripts developed to do searches and categorisations internally peer-reviewed to ensure that they reflected the criteria in the approaches. 

43
Search and categorisation rules metadata have been captured for every successful search result and categorisation. This includes the rule and rule type that was used. 

44
A sample of all search and categorisation results was manually reviewed for each key word or group of key words. This review was used to refine our search criteria and to minimise the risks of false negatives and false positives. Searches with no results and refinements to searches were also documented for transparency.

45
A random sample of 50 appropriations that were not in the search results were checked to confirm we had not missed any in our search criteria. The sample check did not find any more relevant appropriations.

46
A random sample of 10 appropriations were checked from our appropriation data set against relevant public organisations’ annual reports to confirm that the information in the appropriation data set was consistent with the audited annual financial information. The sample check showed that the appropriation data always has the same information that is in the annual reports.

---

10: This figure is the total actual expenditure from the appropriation data for the 10-year period. 

11: Insurance Council of New Zealand. Cost of Natural Disasters at www.icnz.org.nz.

12: The Treasury (2014), Financial Statements of the Government of New Zealand for the Year Ended 30 June 2014, Wellington, at www.treasury.govt.nz.

13: The Treasury (2017), Financial Statements of the Government of New Zealand for the Year Ended 30 June 2017, Wellington, at www.treasury.govt.nz.

14: ISO 31000 – Risk Management at www.iso.org.