Earth data for managing insurance claims and underwriting
An overview of applications
I’ve been fascinated by how earth observation data can be used operationally. How do we move away from simply reporting and measuring a parameter of interest, and integrate it into workflows to improve overall efficiency?
For years, uptake in this respect has been inhibited by a lack of tools that make this type of data accessible. Sure, the data is out there, but that fundamentally provides no real value unless it’s integrated into existing workflows in a meaningful way.
The first industry I’ve been trying to understand is the insurance sector. The value that earth observation data can provide is clear. However, uptake has been slow or nonexistent. I’m only just starting to see companies that are aware of this and trying to tackle this problem.
So in this article, I’m reviewing a paper on how remote sensing data can be used across the insurance sector.
This paper is called “The Potential and Uptake of Remote Sensing in Insurance” and can be found here. It’s an older paper, but still very relevant.
First, it defines two types of risk:
idiosyncratic risk: the situation where an individual’s exposure does not concur with that of their neighbours
covariate / systemic risk: a single shock affects neighbouring individuals at the same time
Then, we need to understand two ways of classifying insurance:
claim-based insurance compensates the insured based on a claim of loss (this is the most commonly assumed definition of insurance)
index insurance in contrast compensates the insured customer based on an index related to such losses rather than the claimed loss.
If you’re confused by the two, here are a couple of examples.
Let’s dive into the applications of remote sensing across different insurance services.
Flood risk assessment: the spatial variation of flood risk can vary over just a few decimeters, as it’s dependent on the asset in question and the relevant elevation. By harnessing digital elevation models from earth observation imagery, this precision could be achieved, providing a more accurate assessment of risk.
In London, airborne imagery was used to model fine-scale accuracy across suburban neighbourhoods
The US has developed a consistent flood risk database which is developed using earth observation imagery, specifically LiDAR. It has a 15 cm vertical accuracy. This database is used across the US to assess flood risk.
Handling of claims
To verify the extent of damage for a claim, on-site assessments are required. For insurance policies of small monetary value, the on-site assessment can be a costly exercise to repeat across numerous properties. Earth observation data can be used to reduce the need for on-site assessments by assessing the extent of damage from high-resolution imagery. This can save a significant number of hours and can allow for faster claims processing.
Earth observation can be used to investigate suspicious claims and confirm that the claimant has been impacted by an insured risk.
Crop damage assessment
Earth observation imagery can be used to assess the impact of hail on cropland. According to one assessment, this could result in a cost-savings role by reducing the number of areas that require in-field verification. Keep in mind that this article is older. With the onset of newer, better instruments, it seems very likely that EO can save time and money for claims processing for this application.
It can reduce fraudulent claims of crop insurance (ex. verify that the crop for which insurance is being claimed is what was planted)
Forest risk damage assessments: This is an obvious one. EO data can be used to verify the extent of forest fire spread. By co-locating this with known insured assets, claims can be processed quickly and without the need for on-site verification.
In addition to tracking the fire, EO data can also be used to determine if claimants have complied with the stipulations required in their contracts. For instance, high-resolution imagery can be used to assess if fire breaks were installed, and if they’ve been maintained. Based on the temporal resolution of the sensor being used, the conditions of the firebreak can be assessed to determine compliance.
Index insurance
Now, let’s talk about how earth observation data can be used to analyze index insurance claims.
Index insurance claims are interesting. Let’s take a look at one example of this. You can get insurance if there is a forage scarcity across a rangeland for your livestock. More simply food, if there’s not enough greenery for your livestock to consume, it will have an obvious impact on the mortality of your livestock.
There is a strong and obvious connection between NDVI (an index between -1 and 1 that measures greenness), precipitation and forage. By harnessing EO data, we can assess the quality of the rangeland and whether the land has reached a scarcity suitable for reimbursement.
This type of correlation can be applied to other applications of index insurance. EO data can provide an affordable, reliable and impartial set of data in which to assess different types of claims.
Another example of EO being used is for the Index-Based Livestock Insurance. This index provides insurance against drought-related livestock mortality. NDVI data is fitted against household-level mortality data, and if the predicted livestock mortality falls below 15%, a payout is made.
The paper concludes with an overview of adoption, highlighting the slow uptake globally.
If you’re how to integrate earth data into your claims and underwriting workflow, shoot me an email at priyapatel@brightgrow.ca