Logo of Piisa


PIISA workshops gathered adaptation experts from all over Europe to discuss and exchange knowledge about climate adaptation insurance.

11th of June 2024

Incentivizing Green Roof adoption

PIISA project’s two webinar workshops in the end of April 2024 focused on knowledge exchange and cooperation across the communities and networks dedicated to climate adaptation insurance in Europe. The first session focused on the challenges, barriers, and opportunities we must reduce the insurance protection gap and accelerate adaptation and resilience building. The second session focused on actuarial risk modelling – state-of-the-art & innovation potential. This event was supported by Risk Kan Network.

During the first session, we presented the methodology and initial findings. Our approach includes a comprehensive literature review and a mapping of the European climate insurance market. We found that there are geographical variations in insurance penetration: southern and southeastern European countries (excluding Spain) exhibit lower insurance penetration rates against various hazards. In order to reduce the climate insurance protection gap, there are effective systems to be utilised:

  • Fee structures: Systems where premiums are partially or entirely based on fixed fees or flat rates tend to be more effective.
  • Role of public entities: The presence of Public-Private Partnership (PPP) entities or the state acting as the primary insurer enhances protection.
  • Mandatory insurance uptake: Implementing some form of mandatory requirement for natural catastrophe (NatCat) insurance significantly reduces the protection gap.
  • Avoiding ad hoc relief measures: The non-use of ex post disaster relief measures by the state promotes better insurance coverage.

We also had a constructive discussion session where participants could ask their questions. Below is a list of the main questions exchanged. Additionally, we conducted a survey and an open Q&A to assess the attendees' level of knowledge.

During the second session, we provided a review of the existing literature on climate risk insurance modeling, emphasizing the increasing relevance of such models in the face of rising losses attributable to climate change. These models are used to estimate risk for different perils in combination with simulating risk related parameters for insurance schemes such as premiums and deductibles. It was noted that not all existing insurance models are prepared for simulating the effects of climate change. Therefore, there are disparities between current practices and those essential for addressing climate related challenges. It was also emphasized that there is an overemphasis on flood insurance in the model applications, a bias toward developed nations, and insufficient coverage for non-agricultural commercial sectors.

Furthermore, less than half of the analyzed papers take a forward-looking approach by incorporating climate change scenarios, and an even smaller percentage consider socio-economic development scenarios. This limitation shows current methods require additional development for assessing the effects of future climate risk for insurance. Overall, future research is needed to address these gaps. We advocated for more refined models, expanded geographical and hazard coverage, enhanced inclusion of the commercial sector, and a forward-looking approach. The insights from these model advancements can help the insurance sector to proactively adapt to the challenges posed to insurance for climate risks due to climate change and socio-economic developments.

After the presentation of the task 1.2 results, we discussed the impact of increased premiums and the inherent uncertainty in risk-based premiums. Emphasis was placed on promoting awareness and stimulating Disaster Risk Reduction (DRR) efforts, while maintaining a balance with affordability considerations. Additionally, we highlighted the crucial role of government involvement in ensuring affordability and enforcing regulations.

Once again, we facilitated an interactive discussion during the session. The main questions posed by participants are detailed later in this document, and the results of our Q&A and survey can be found in the accompanying presentation.

You will find links to the session recordings below.

Session 1: Challenges, barriers, and opportunities to reduce insurance protection gap and accelerate adaptation and resilience building https://www.youtube.com/watch?v=fBDkdtjW_oU
Session 2: Actuarial risk modelling – state-of-the-art & innovation potential https://www.youtube.com/watch?v=o6BKuYsdK9o

Questions and answers

You will find answers to questions asked during the sessions below.

Session 1

Q: Why is it not possible to sell parametric insurance in Finland?
A: First, parametric instances do not necessarily connect with the damage. It is said in the insurance law that insurance, indemnity payment, is not allowed to make policyholder financial situation better than before the damage (prohibition of enrichment). We have 150 years history of risk management via damages; thus it will take time to change the public opinion. Parametric insurances are a shortcut from risk to indemnity payments without damage control or insurance claims and claims control. This shortcut has obvious advantages (no asymmetric information between policyholder and insurance company), but also challenges like basis risk.
Second, markets are conservative, especially insurance markets. Parametric mechanisms are rather new (10 years max) so it is not recognized yet as insurance product for some insurance regulatory bodies.

Q: For Stefano: I discovered in earlier discussions on the market penetration on wildfire cover that there are often misunderstandings about the definition. Do you mean wildfire coverage for property risk of forestry risk?
A: We have two focuses for wildfire pilots: at household level (property risk) and for forest stakeholder (protecting the natural assets)
Good that you focus on both. I just wanted to mention that the maps with the penetration could give the wrong impression as the question about the market penetration could have been interpreted in different ways by the people who have provided this information.

Q: For Stefano: Which AI powered search tool did you use in the review? Do you already know when the results of the review will be published?
A: The AI tool is Research Rabbit (https://www.researchrabbit.ai/). The results of the review will be published on the PIISA website, in the page dedicated to public deliverables (https://piisa-project.eu/deliverables).

Q: 1. Do you consider market penetration in terms of lives covered beyond just premium income?
2. Is there scope to expand the framework beyond Europe to developing/emerging countries?
A: There are many different ways to characterize penetration, and indeed several sources use different approaches.
We based our mapping mainly on the data provided by EIOPA, which is itself a combination of information gathered from many different sources.

Session 2

Q: What is the difference between flooding r, c or o? A: River, Coastal, Other flooding

Q: Did your study on models make a distinction between non-life direct insurance and reinsurance? A: The models focused on direct damages, so the focus was on non-life direct insurance. Models that focused on reinsurance solely were not found. However, reinsurance was often included as a component of the model as part of the larger system.

Q: Did you find also concerns about risk-based premiums in relation to spatial equity? A: Yes, some areas have a too high risk: people will not take an insurance in this area. Premiums are too expensive, and the complement is adverse selection

Q: For Max presentation, were models for parametric insurance also considered? A: Not every time, for example about agriculture insurance, no paper about parametric . It was indeed not included in the task 1.2. presented by Peter and Michiel.

Q: To M Ingles: If flooding does not cause the most damage cost, what hazard is most costly. And why do you think floods are overrepresented A: I think the point is more that the larger focus on modelling flood risk is not commensurate with the share of overall natural disaster damage caused by flooding.

Q: AXA Climate was working on the parametric insurance based on DLT. Is this also going to be part of PIISA? A: Parametric Insurance tries to avoid blockchain most of time, so we do not want black-boxes, so we are able to explain the modelling to our clients. There is a focus on parametric insurance in work package 3 of the PIISA project, in the forest and agriculture pilots.

Q: Were the NAT CAT models, that were reviewed, used for insurance purposes or were there other purposes as well for the models? A: The reviewed models were mostly employed to answer research questions about climate risk insurance.

Q: Do you know which models from your review the direct insurance companies actually use? A: The basis of models used in the insurance industry is mostly the same as these academic models, but the models used in the insurance industry are often not forward-looking models accounting for climate change and more focused on underwriting.

Q: Is there any empirical evidence for NbS approaches for Disaster Risk Reduction (DRR)? A: To our knowledge, there is a lot of empirical literature on the economic value of NbS and some empirical literature existing on the DRR/adaptation effectiveness of NbS. The difficulty for insurance purpose is to have access to a representative dataset in terms of geography and risk reduction so it becomes effective for underwriting. As specific use cases do not make the case for worldwide application unfortunately.