How extreme events reframe our thinking on weather-related perils in New Zealand

New Zealand was hit by two severe weather events just two weeks apart in early 2023, highlighting the growing importance of weather-related perils in this country. For each event separately – the Auckland floods end of January and Tropical Cyclone Gabrielle in mid-February – industry loss estimates are within the NZD 2bn loss range. The New Zealand Ministry of Foreign Affairs considers the damage by Gabrielle as the costliest non-earthquake related natural disaster on record.

The record level of rainfall, which fell over Auckland in the typically-dry late January period, was a result of moisture from a former tropical depression being channeled between a low front in the northwest of New Zealand and a blocking high front to the east of New Zealand. This narrow corridor of concentrated moisture, an atmospheric river, conveyed tropical moisture and record rainfall over the most populous part of New Zealand.

Auckland was only starting to recover from the impacts of the Auckland event when Cyclone Gabrielle hit the North Island of New Zealand in mid February. Whilst we have seen decaying Tropical Cyclones make their way to New Zealand in the past (e.g. Bola 1988), the storms often weaken as they encounter colder ocean waters. Gabrielle, however, was supported by warmer-than-average waters, under the influence of a third consecutive summer of La Niña conditions. Again, record rainfall lead to significant damage to the North Island.

The scale of these events has resulted in record insured weather losses, and an enormous task of recovery. The events also prompt a re-think of weather as a risk driver in New Zealand. Here we offer three perspectives.

1. Severe events are occurring more frequently than indicated by past experience
With new data points at hand, the historical return periods between large events shorten. This new information shows, in hindsight, an optimism with respect to our view of the return period of past events. In Australia, we have seen frequent hail, flood, and bushfire events over the past decade, which have repeatedly reset our expectations of how frequently natural catastrophe events occur. Across the globe, we had similar learnings with severe convective storms in France. Before 2022, storm Ela (2014) was commonly considered to have a return period of 20-50 years. Yet, Qiara and Maya (2022) each exceeded the Ela loss, forcing the industry to reconsider event return periods of this size in France. Ela, Qiara and Maya all exceeded the EUR 600-700 mn threshold in less than a decade, which resulted in the industry adopting a much shorter return period for a loss of this size.

2. Historical analogues can be helpful to understand today's risk view
New Zealand's Cyclone Giselle (1967) was, arguably, a storm capable of producing as-if losses of equivalent scale to those observed in Gabrielle. The last couple of decades have been relatively quiet in New Zealand with respect to large weather losses, however, looking further back Giselle provides an analogue for the recent large weather losses in New Zealand. Similar observations have been made in Japan, where the 2018/19 typhoon risk level led the industry to acknowledge that the recent events were similar to typhoon activity observed in the 1950s and 1960s – the conclusion being that the more recent losses were not unprecedented.

3. New data points can result in step change shifts of risk views
For New Zealand, the early 2023 weather event losses exceed the previously observed insured losses by a significant margin. There is uncertainty in return period estimations at hand, as we have experienced a relatively quiet gap in large weather events for a few decades. However, Bola and the Invercargill Floods in the 1980s, just as much as Giselle, are key industry benchmarks. Appropriate observation periods also matter when using the past to inform the future – it is important to consider longer-term historical analogues alongside emerging shorter-term trends.

Both Gabrielle and the 2023 Auckland Floods exceeded any earlier event, which demands a reconsideration of event return period assumptions by the industry. The event-loss level of around NZD2bn has now been exceeded twice in one year alone. Studying event physics of the older events, and debiasing historical losses from the 1960s and 1980s, it appears Giselle would be capable of generating comparable losses to Gabrielle. Further, Bola would also generate meaningful as-if losses.

By making use of the new data points, as well as appropriately indexing historical events, it can be estimated that the return period of an NZD 1bn loss is considerably less than 20 years. Similarly, with three historical events in 55 years likely reaching the NZD 2bn loss range, we can no longer consider events of this loss size to be rare and we must also reframe our views of the loss potential of truly extreme weather losses in New Zealand.

As a country with pronounced earthquake exposure and weather losses long seen as secondary, the recent events must change our perspective of overall risk drivers in New Zealand, as weather losses are now rapidly closing the gap to earthquake on both loss severity and frequency expectations.

Takng the above three factors into account, in our view:

• Weather related losses in New Zealand are on the rise, reflecting similar trends observed worldwide.
• A meaningful analysis of historic losses requires consideration of both longer observation periods to include historical analogues from the past, and shorter-term observation periods to reflect recent trends.  
• The 2023 weather losses are the highest on record in New Zealand, and whilst the return period for these loss levels remain uncertain, the recent events provide robust benchmarks for estimating that such losses should be expected to happen more frequently.
• Insurance risk assessment and pricing for weather related perils in New Zealand calls for a meaningful correction to support revised risk views and maintain long-term sustainability of the insurance offering for these perils.

This year's events highlight that the weather risk in New Zealand has been underestimated and brings more clarity in terms of what the market needs to consider.

While the loss trend can be partially explained by macro trends such as economic growth, inflation, urbanisation, climate change, and an increase in built-up land area, along with vulnerability changes and social inflation trends, it is crucial to carefully account for such trends when quantitatively evaluating historical losses. The current inflationary environment contributes another overlay to consider in assessments of weather loss potential.

An adequate risk view, leading to the right protection at the right price, is one key to enable resilient societies. It means accounting for long-term historical experience and ongoing physical and socioeconomic trends. When assessing weather related risks in New Zealand we should ask ourselves: "what's the next 'unprecedented' loss we could reasonably predict to happen?" These are considerations we owe resilient societies in New Zealand and anywhere in the world where we assess risks.