Skip to main content
main content, press tab to continue

Identifying and targeting unfair discrimination in insurance

By Michael Chen | March 6, 2023

To meet the needs of more stakeholders and achieve ESG goals, actuaries and insurers need to understand and address the potential for unfair discrimination.
Insurance Consulting and Technology|ESG and Sustainability

The increasing importance of environmental, social and governance (ESG) means more consumers, regulators and other stakeholders expect greater transparency and accountability from insurers. In particular, more people want to understand how insurers’ activities either contribute to or help alleviate social inequities.

Insurers and actuaries able to identify and target unfair discrimination not only enable their organizations to keep pace with shifting consumer and regulatory expectations but also promote fairness more broadly. Such steps also help insurers manage governance and associated reputational and brand risks, particularly important in scenarios where actuarial principles compete.

This insight explains the challenges insurers and actuaries face, offering some practical considerations for navigating these complexities.

What’s the difference between unfair discrimination and disparate impact?

Under insurance law, unfair discrimination occurs when similar risks are treated differently. Historically, this has been the focus of conversations on fairness in insurance.

More recently, stakeholders such as consumer groups and insurance regulators have incorporated a variety of terms into debates, specifically:

  • Disparate impact, where seemingly neutral practices disproportionately impact a protected class
  • Proxy variables, which are measurable variables able to take the place of ones that are difficult to measure
  • Proxy discrimination, which is the intentional use of a neutral factor for the purpose of discriminating against a consumer

Balancing fairness and accuracy

The Casualty Actuarial Society's Statement of Principles Regarding Property and Casualty Insurance Ratemaking describes four principles that apply when determining and reviewing the accuracy and fairness of property & casualty insurance rates:

  1. A rate is an estimate of the expected value of future costs.
  2. A rate accounts for all costs associated with the transfer of risk.
  3. A rate provides for the costs associated with an individual risk transfer.
  4. A rate is reasonable and not excessive, inadequate or unfairly discriminatory if it is an actuarially sound estimate of the expected value of all future costs associated with an individual risk transfer.

While these principles appear straightforward, stakeholders are increasingly concerned about insurers’ and insurance regulators’ success in promoting fairness.

Understanding proxy discrimination and proxy variables

Modeling and machine learning can give rise to proxy variables that could generate disparate impact, with algorithms unintentionally reinforcing inherent bias and systemic discrimination against protected classes.

A simplified example of discussing proxy variables is looking at urban versus rural areas in rating. An urban area may have a higher concentration of cars in a smaller area, leading to a higher frequency of claims. An urban area may also have a disproportionate population of a protected class when compared with the population of a more rural area. This illustrates the importance of actuaries and insurers taking steps to understand such relationships, in this example, that between claim frequency, the urban area variable and the protected class.

Actuaries should be aware that efforts to address an apparent disparate impact may come at the cost of accuracy and therefore a trading-off of two guiding principles for actuaries. This could, for example, result in rates that don’t reflect all costs associated with the transfer of risk but could prove fairer to a protected class.

How to address unfair discrimination and disparate impact

Insurers can identify and minimize disparate impact by addressing it through pricing, marketing or claim settlement models. For example, an insurer might consider prohibited characteristics as control variables when developing a model but omit these characteristics when the model is deployed.

While there is no single right or wrong answer here, what is clear is the need for insurers and actuaries to prioritize transparency and understand the inputs going into models, whatever these may be.

Actuaries should be able to quantify the give-and-take at play. If you are choosing to be fairer by not using a given variable, for example, you need to understand the impact on accuracy in a way that is auditable.

Working with partners using models that address disparate impact can also help. Some insurance rating software, for example, includes fairness algorithms and enables insurers to combine data with the latest analytical machine learning approaches to understand how risk factors and customer characteristics affect cost and customer behavior in a way that’s auditable and transparent. 

Tackling unfair discrimination and the ESG agenda

As regulators and wider stakeholders draw more attention to issues of unfair discrimination, actuaries and insurers have a role to play in advancing fairness while protecting institutions from emerging governance and reputational risks.

Measures to identify and target discrimination and understand disparate impact also make good business sense by serving broader social and governance objectives, supporting business resilience in the longer term. This can also help differentiate organizations from competitors, building brand and attracting talent into the future.


Director/Pricing and Radar Subject Matter Expert
Insurance Consulting and Technology
email Email

Contact us