Column
Data Science and AI

The Price of Loyalty: Rethinking Optimisation in Insurance Pricing

Claim your CPD points

Recent headlines in Australia have brought attention to insurance pricing practices. A 14% annual increase in premiums, highlighted by the Australian Council of Trade Unions (ACTU) , as the sharpest rise in the Consumer Price Index, has prompted public concern about the affordability of insurance during a period of economic pressure.

In parallel, the private health insurance sector has come under scrutiny for practices such as “ product phoenixing ”, where policies are closed and relaunched under new names, often with higher premiums.

These examples have intensified wider concerns about fairness and transparency in insurance pricing. This includes growing attention to strategies such as the loyalty penalty and price optimisation, where premiums may vary not only by risk but also by customer behaviour, such as the likelihood of switching providers.

In the context of insurance, an industry built on risk pooling and financial protection, behaviour-based pricing practices raise important questions about their impact on consumer outcomes, particularly on trust, transparency and fairness.

Ongoing debate continues over whether and to what extent, these pricing practices align with the traditional expectations and responsibilities of insurers. In this column, we take a closer look at this evolving issue.

Understanding price discrimination in insurance

In economics, charging different prices for identical products is known as price discrimination, which is a neutral term. When applied to insurance, this concept requires a specific clarification: “identical products” is less straightforward since insured risks differ.

For instance, two auto insurance policies that offer the same coverage and benefits may be priced differently due to differences in expected losses, based on factors like the driver's history, vehicle type, or location. This approach, known as “risk-based pricing”, is a standard and widely accepted practice in insurance, consistent with the principle of actuarial fairness.

However, insurers may also offer different prices for customers with identical risk profiles. This could be based on behavioural factors, such as the likelihood of switching providers or renewing a policy. This form of pricing, often referred to as “non-risk-based price discrimination” or “price optimisation”, reflects differences not in risk, but in consumer behaviour or perceived willingness to pay.

Price discrimination is typically classified into three degrees:

  • First-degree price discrimination or perfect price discrimination, occurs when a seller charges each consumer exactly their maximum willingness to pay — a theoretical ideal that is rarely achievable due to informational constraints. 
  • Second-degree discrimination involves pricing based on the quantity or version of the product purchased, such as offering discounts for bulk buying or tiered service levels.

    Third-degree discrimination occurs when prices vary across identifiable groups with different elasticities of demand — such as discounts for students, seniors, or members of particular professions.

In the insurance market, similar patterns of price discrimination exist. In some jurisdictions, insurers commonly offer discounted rates to attract new customers, while long-standing policyholders may face higher premiums despite having the same risk profile. Research has documented that these loyalty penalties are often driven by price optimisation techniques.

Why insurance is different

Price discrimination is a widely accepted strategy in many markets, where it can enhance efficiency by aligning prices with consumers’ willingness to pay. In industries like software or airlines, this can expand access and match pricing to demand.

However, insurance operates under a distinct set of conditions that make the implications of price discrimination more complex, see, for example, Thomas (2012). Below, I summarise several features that differentiate the insurance market from other industries and shape the economic and ethical considerations surrounding pricing practices.

  • Information imbalance: Insurance relies on the doctrine of utmost good faith, which requires consumers to disclose relevant personal information truthfully. In return, insurers were traditionally expected to use that information primarily for assessing risk. The use of behavioural data to adjust premiums, such as estimating the likelihood that a customer will renew, raises concerns about whether this exchange remains equitable.
  • Complexity and opacity: Even under traditional risk-based pricing, insurance products can be challenging for consumers to understand or judge fairness, particularly in terms of how various rating factors influence premiums. The addition of behaviour-based pricing — where factors like customer retention likelihood or shopping habits are considered — adds another layer of complexity. This can further reduce transparency and make it more difficult for consumers to interpret or anticipate how their premiums are determined. Moreover, most insurance customers are unaware that their premiums may be influenced by behavioural factors such as loyalty, online activity, or inertia. This lack of transparency limits consumers’ ability to evaluate options, make informed decisions, or respond effectively to pricing strategies that are not explicitly disclosed.
  • Limited market expansion: One economic justification for price discrimination is that it can enable more consumers to access a product, especially those who are price-sensitive. This rationale is less applicable in insurance, where participation is often mandatory, or where — even when not legally required — most people purchase coverage voluntarily, and where marginal costs are significant. Moreover, because insurance is inherently policyholder-specific — with products tailored to individual risk profiles — even offering different prices does not meaningfully expand the market in the way it can for non-customised goods or services. As a result, price discrimination is less likely to expand market coverage than in other industries with low marginal costs.
  • Consumer utility is difficult to verify: In many markets, price differences are often attributed to variations in consumer preferences, such as willingness to pay more for convenience, brand reputation, or enhanced service. In insurance, however, consumers typically interact with the product only when a claim is made, and switching providers can involve time, effort, and uncertainty. As a result, it is challenging to determine whether differences in premiums reflect genuine perceived value or are instead driven by limited information, time constraints, or financial literacy. This makes it more difficult to assess whether price variation is aligned with actual utility. In addition, pricing strategies that offer introductory discounts followed by price increases over time often generate consumer disutility, as they compel policyholders to engage in repeated comparison shopping or switching to avoid higher costs — a process that is time-consuming, stressful, and particularly disadvantageous for those with limited financial literacy or digital access.
  • Proxy discrimination concern: Non-risk-based behavioural factors — such as digital engagement, shopping habits, or customer inertia — can serve as proxies for socioeconomic disadvantage. Vulnerable groups, including older Australians and those with lower incomes, limited digital access, lower financial literacy, may face systematically higher premiums not because of higher risk but because they are less likely to engage or switch providers. As a result, behaviour-based pricing can unintentionally reinforce existing inequities and lead to proxy discrimination despite appearing neutral on the surface.
Global responses

Regulatory responses to price discrimination in insurance vary across jurisdictions.

In the United States, concerns about fairness in personal insurance date back to the 19th century, when industry associations sought to limit practices that allowed wealthier or more influential clients to secure lower premiums not justified by their level of risk. At the time, the focus was on preventing what was termed “unfairly discriminatory rates”, instances where pricing diverged from actuarial principles.

More recently, several U.S. states have taken steps to restrict behaviour-based pricing practices.

Since 2015, approximately 20 states have introduced bans on the use of price optimisation and demand modelling in personal insurance lines. These measures reflect growing regulatory attention to the potential for such practices to disadvantage consumers. Anecdotal reports from industry professionals suggest that price optimisation is increasingly viewed with caution in some North American markets.

In 2022, the UK Financial Conduct Authority banned “price walking” — where renewing customers are charged more than new ones for equivalent coverage — citing concerns that the practice harmed consumers through a lack of pricing transparency, penalised loyalty and undermined trust and fairness in the insurance market.

However, the FCA did not prohibit all forms of optimisation, acknowledging that some uses — such as improving price efficiency or stimulating competition — may benefit consumers under appropriate safeguards.

In the European Union, the European Insurance and Occupational Pensions Authority (EIOPA) issued a recommendation in March 2023 advocating for a similar ban on price walking in insurance renewals when not justified by changes in risk.

In Australia, regulatory action has been more limited to date. However, rising consumer concern, public commentary from groups such as the Australian Council of Trade Unions (ACTU), and an ongoing federal investigation into health insurance pricing indicate increased scrutiny of pricing fairness in the insurance sector.

Time to reassess?

The use of price optimisation in insurance, where premiums are influenced not only by risk but also by non-risk-based consumer behaviour, has raised concerns about transparency, fairness, and consumer trust. However, some industry observers caution that the issue is complex. Even in the absence of formal price optimisation, subjectivity can enter pricing decisions.

For example, risk-based pricing and expense loadings often involve judgment calls that may implicitly reflect underwriters’ views on customer behaviour, retention likelihood, or risk tolerance. Insurers have long used simpler pricing strategies — such as targeting younger customers through discounts or digital channels — to improve competitiveness, strategies that might inadvertently fall within the scope of strict price optimisation bans. These nuances highlight the importance of ongoing discussion and critical examination of the principles that should underpin fair and responsible insurance pricing.

Miller and Moulder (2024) highlight several practical limitations of price optimisation in insurance, including model uncertainty, misestimation of customer elasticity, and declining performance over time due to shifts in the customer base. Additionally, as shown in Huang and Shimao (2025), the extent and impact of price optimisation are influenced by market competitiveness, with regulatory effects becoming less pronounced in more competitive environments.

As the industry continues to evolve, regular evaluation of pricing practices is necessary to ensure alignment with principles of transparency, consumer protection, and the broader social function of insurance. Defining appropriate boundaries for the use of behaviour-based pricing is likely to require coordinated action by both regulators and industry stakeholders.

References

Thomas, G. R. (2012). Non-risk price discrimination in insurance: market outcomes and public policy. The Geneva Papers on Risk and Insurance-Issues and Practice, 37, 27-46. Available at https://www.jstor.org/stable/41953166

Huang, F., & Shimao, H. (2025). Welfare Implications of Fair and Accountable Insurance Pricing. UNSW Business School Research Paper Forthcoming. Available at SSRN: https://ssrn.com/abstract=4225159

Miller, H., & Moulder, T. (2024). Optimal illusion – A look at the practical limitations of price optimisation. Paper presented at the Actuaries Institute 2024 All-Actuaries Summit, Sydney, Australia. Available at: https://actuaries.logicaldoc.cloud/download-ticket?ticketId=5b86fb11-f32b-4826-bdf7-c0774d66a961



About the authors
Dr Fei Huang
Dr. Fei Huang is an Associate Professor in the School of Risk and Actuaries Studies at UNSW Business School. Her research focuses on responsible data-driven decision-making, including fair and non-discriminatory insurance pricing, interpretable machine learning, mortality modelling and customer relationship management. Specifically, she examines ways to make insurance equitable, affordable and sustainable in the contexts of AI and climate change. For more information, please refer to her profile webpage https://www.unsw.edu.au/staff/fei-huang.