In the book titled Thinking, Fast and Slow, Daniel Kahneman introduces two ways of thinking:

• System 1: Slow, deliberate and analytical, requiring effort but allowing for deeper reasoning.
• System 2: Fast, intuitive and automatic, but prone to biases.

Great value can be derived from applying this framework to insurance pricing. The process of insurance pricing should ideally adopt both System 1 and System 2 thinking at different stages in the process. The challenge lies in determining when and where to apply each system of thought to generate optimal value.

System 1: Going slow

Risk Models and Rating Structures Building a strong foundation for insurance pricing requires thoughtful analysis and deep reasoning, like that defined by System 1. Developing a risk model requires you to understand the data, uncover key relationships and experiment with different rating structures to find out what is truly driving your claims experience. It is a slow and deliberate process, that necessitates deep investment in order to derive real value.

The risk premium is the cost of goods sold in insurance. Before factoring in expenses, commissions and profit margins, insurers must first ensure that they have a reliable estimate of the cost of their risk. If this step is done incorrectly, it can cost a company millions. Risk modelling requires a great deal of scrutiny because it forms the basis of the office premium, which is the final price that reaches policyholders. Accurately calculating this is where the most time and brainpower should be spent.

The rate at which new data becomes available in risk modelling is also very slow. Claims data takes time to mature, often taking three months to give a full picture of the losses. As a result, new risk models can only be built once or twice a year if done correctly. Therefore, the consequences of a poorly structured model are far-reaching and difficult to unwind, making it critical to get things right the first time. Unlike other industries where feedback loops are immediate, insurers must make pricing decisions with data that inherently lags behind real-world changes. This makes careful iteration and testing all the more important.

System 2: Thinking fast

Price Optimisation and Tactical Adjustments Once a rating structure is established, the focus often needs to shift to System 2 thinking, where rates are adjusted to maximise profitability or market share. Price optimisation models take existing risk assessments and fine-tune premiums based on competitive positioning, conversion rates and business objectives. These adjustments are based on real-time data, and do not require a deep re-evaluation of the rating structure.

Pricing teams typically receive a high volume of closing data daily, which matures within a few days. New business volumes, renewal retention rates and shifts in competitive position allow insurers to react quickly. The rate of information change in these scenarios is much higher and decision making is better facilitated by System 2 type thinking. Small pricing adjustments can easily be reverted, and errors will not be as far-reaching as in the case of a poorly modelled risk structure.

Because the cost of being wrong is much lower during price optimisation and adjustments, this is the area where automation makes the most sense. Many insurers use black-box machine learning models for price optimisation, as the impact of errors is smaller and models can be retrained frequently. Unlike risk modelling, where incorrect assumptions can lead to significant financial consequences, a suboptimal decision in pricing can be quickly reversed, affecting only the quotes between the two revisions.

Balancing Speed and Thoughtfulness The challenge of pricing comes in accurately balancing these two approaches. System 2 style optimisations can be valuable, but if they are not anchored in a strong System 1 thought foundation, they can lead to unintended consequences. Some of these consequences include overfitting short-term trends, mispricing emerging risks or making opaque pricing decisions that are difficult to justify later on.

The best pricing strategies are dynamic and have been tailored to fit the nuances found in each stage of the process. They should allow underwriters and actuaries to experiment thoughtfully with rating structures while still responding effectively to market conditions.

In the end, sustainable competitive advantage in pricing does not come from reacting quickly alone. It comes from thinking deeply first.