Data Ethics in Insurance Underwriting

The way in which data and analytics are used in underwriting has ethical implications on several levels…

  • how the structure of data and design of algorithms can influence underwriting decisions;
  • their influence on the products and services on offer;
  • how they frame the insurer's relationship with consumers;
  • how they are changing the nature of insurance itself.

These ethical implications are often very significant. We often hear about how much good will come out of ‘this or that’ use of data and analytics. What we need to address as well are the ethical concerns that are being raised.

The insurance sector is facing a series of challenges around data ethics in underwriting. Here you can find out more about the nature of those challenges, and about how your firm can organise its thinking and response.


The Role of Algorithms in Delivering Ethical Outcomes 

As the influence of analytics on underwriting decisions continues to grow, so does the need for those decisions to produce outcomes that fall within the ethical expectations of the firm, its customers and the regulator. It’s a complex task - data scientists working in insurance have to contend with layers of complicatedness. Yet trust in insurance rests upon them designing analytics that deliver fair outcomes. These articles explore some of the challenges they face, and how they might address them…

How Good are your Firm's Algorithms? Can You Prove It? which I weigh a four phase approach to assessing the trustworthiness of claims about the performance of an algorithm.

5 Sources of Bias in Algorithms which I set out a framework for addressing bias at different stages of an algorithm's development. 

Why Underwriters need to Use Predictive Analytics with Care which, back in 2013, I outline the ethical issues associated with what has since become known as application fraud.

Should There be Compulsory Ethical Training for Insurance Algorithms? which I ponder if the compulsory ethics training that insurance people in some countries have to take, will soon be required of underwriting algorithms.


What Consumers Think…

Three recent surveys have identify serious concerns amongst consumers about some of the digital trends in underwriting.

Most relevant for UK insurers was the ABI survey published in February 2020, in which serious issues around consent, secondary use of data, data trading and value were identified.

In February 2021, research from Stanford University in the US found clear concerns amongst auto policyholders about telematics data. For example, many people understood insurers using data about how a car was being driven, but many were unhappy with the use of data about where or when a person was driving. The author concluded that people were evidently very capable of assigning social and ethical meaning to the use of personal data in market settings. Data was not just a neutral market commodity.

And in that same February, research by Deloittes found that... “internet-connected offerings, such as motor telematics and home insurance linked to home sensors, cause apprehension among many customers; they do not feel comfortable sharing data from car and home sensors with insurers”.

Insurers need therefore to understand that consumers are judging how they acquire and use their both their personal and secondary data. In response, underwriters must bring ethical parameters into the design and use of their decision making systems. 


Influence on Products and Services

There's no doubt that the data and analytics being used by insurers are generating new types of products and the scope for new types of services. And there's no doubt that these come with ethical questions attached. 

Telematics is the iconic development, but in this article, I outline why it is much more influential than simply a source of pricing data... Telematics will be much more influential than people think 

Insurers know that their data is valuable, and not only to themselves. The market for data is huge and high quality data (which insurers get from policyholders) is sought after. The issue here is destructive demand and 'ethical fading, which I explain in this article...  Destructive Demand - is it happening in underwriting now?


Ethics and Underwriting Data  

Insurers gather data for underwriting in order to enhance portfolio performance. Yet the act of gathering certain types of data does raise ethical questions. As the sector expands its data horizons, so the urgency of those ethical questions increases too. In these articles, I explore three sources of data that are attracting the most questions.

Why insurers need to review their use of credit data  ...explores a data source with direct relevance to key regulatory concerns around vulnerability and access to insurance. 

One of the most controversial sources of data is that of our emotions and mental health. In this article from 2019, I look at just how reputationally explosive it is...  How do you feel about insurers tracking and analysing your emotions? 


Independent Insight 

I help insurers think about complex issues from a different perspective.

Underwriters are facing some complex ethical challenges, around developments in pricing, products and services. They use my insight and lateral thinking to gain an extra level of understanding about such challenges.

Example - A leading UK insurer wanted to better understand the arguments about discrimination in insurance and so brought me in to provide independent insight. This helped them to conduct further in-house analysis and monitor ongoing developments.

Get in touch if you'd like to have an informal chat.


Accessing Digital Medical Records

Access to a patient's digital medical records is an important goal for life and health underwriters. In this article from 2018, I look at four key steps that insurers need to consider for this to be achieved. They are significant steps, but ones that, if done right, could secure public trust...  Access to Medical Records is Controversial but Achievable   

And in 2020, I followed this up with a more in-depth of the issues involved. You can download this free piece of analytic (no email required) by clicking on the button below...