Insurers can segment their customers into different homogeneous risk groups based on their risk profiles and characteristics, using clustering techniques. This can help insurers to understand and tailor their products, services, and pricing to each group, according to their behaviour and preferences. Insurers can also optimize their premium rates for each group and ultimately each customer based on the willingness to pay driven by their expected utility (i.e., based on their individual demand function). By using this method, the insurer can maximize customer satisfaction with risk-adjusted profitability. By implementing utility-based pricing clustering, insurers can achieve a more efficient and effective pricing strategy that can enhance their competitive advantage in the marketplace while boosting its risk-adjusted performance. Indeed, they can offer more personalized and differentiated prices that reflect the customer’s risk profile and preferences, while better meeting the insurer’s commercial and financial goals. This can lead to higher customer satisfaction and loyalty, lower anti-selection and moral hazard, and higher risk-adjusted profitability and growth for the insurer. One possible way to implement this approach is by using a Deep Neural Network (DNN) model, an AI-based model, that can learn from data and perform complex tasks. A DNN model can perform both clustering and pricing tasks simultaneously, by using a multi-task learning approach. However, a DNN model also faces some challenges and limitations, such as data availability, computational cost, overfitting or underfitting problems, ethical and regulatory issues. Moreover, a DNN model may have implications under IFRS 17 with regard to the onerousity test, which requires the expected present value of future cash flows, adjusted for non-financial risk and time value of money, to be higher than the premium charged. A DNN model can improve the quality and consistency of the IFRS 17 model and reduce the uncertainty and volatility of the results, if the DNN model pricing clusters match the IFRS 17 cohorts, and if the assumptions between the two are coherent. If so, it can also help insurers to identify and measure onerous contracts more accurately and reliably.
However, in practice, due to different modelling and assumptions between the DNN model and IFRS 17 model, the DNN model may create inconsistencies and discrepancies between the pricing decisions and the accounting results. It may also raise questions or concerns about the reliability or validity of the IFRS 17 model and the onerousity test. Therefore, insurers need to carefully evaluate the feasibility and suitability of having utility-based pricing clustering based on DNN model, as they need to consider their business objectives, but also their accounting implications. They need to ensure that their pricing decisions are consistent and coherent with their IFRS 17 model and their onerousity test, and that they can justify and explain their choices and results to their stakeholders. In conclusion, utility-based pricing clustering based on DNN model is a promising and innovative approach that can help insurers to segment their customers and optimize their premiums and retention rates, while maximizing customer satisfaction and risk-adjusted profitability, with the potential of having the latter significantly in excess of the weighted average cost of capital (WACC) of the allocated capital. However, this approach also poses some technical, ethical, and accounting challenges that need to be carefully addressed and resolved. Insurers need to ensure that their DNN model is aligned and consistent with their IFRS 17 model and their onerousity test, and that they can provide sufficient and transparent information and explanation to their stakeholders about their pricing decisions and accounting results. “Change is the law of life. And those who look only to the past or present are certain to miss the future.” – These wise words of John F. Kennedy were never so true!