[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the Institute & Faculty of Actuaries
.

Insurance: Striking the perfect balance

Price optimisation has become established in the UK and a small number of other markets as the best way to build or maintain a profitable business. The price should be set so that the costs of doing business are covered, and it should include a reasonable margin for profit. In personal lines insurance there has been a technological and informational arms race to find the perfect balance between inadequate and excessive premiums. Finding this balance forms the essence of price optimisation.

What is price optimisation?
The term ‘price optimisation’ can describe a whole range of techniques that combine information about expected claims experience and customer behaviour to calculate optimum premiums with regard to a particular business objective. It involves determining an optimum price for a service or product that incorporates an understanding of market, consumer and competitor behaviour. It uses as building blocks:
>> Information on how customer characteristics affect claims experience
>> Retention and conversion behaviour
>> Propensity to purchase additional cover and products.

These components are combined in a model that allows a company to investigate how different pricing strategies might affect the profitability of the company and the number of policies written. This model is then used to determine, on a policy-by-policy basis, the premium that delivers the company’s objectives. The actual methodology used to deliver this can vary but all the methods have a common aim of balancing profitability, practicality and retention (and new business) rates.

The final pricing strategy selected will depend just as much on the company’s objectives as the data and models used. The same models and data will produce radically different premiums if the target is to grow volume at 10% a year, rather than to maintain current volume and increase profitability. Similarly, models that consider cross-selling may lead to premiums different from those calculated by models that do not.

Why optimise?
Optimisation offers the possibility of achieving higher profitability compared to a straightforward cost-plus approach. The method allows potentially profitable segments of the market to be identified and targeted with attractive premium rates, for example, by accepting low profitability on a policy because of its future cross-selling potential. In particular, these methods enable a company to adjust premiums to allow for differing price sensitivities in different segments of the market.

The more information you hold on an individual, the more accurately you can predict their behaviour. This means that affinity groups and similar intermediaries have a key advantage. An intermediary has much lower uncertainty over costs; their cash flow involves a fixed net premium rather than an unknown claims cost, and they often have access to a wealth of customer information unavailable to the insurer.

Direct writers, although they have less information than some intermediaries, are also well placed to benefit from optimisation. The ability to change rates and monitor the resulting changes in consumer behaviour in close to real-time allows the construction of accurate models of customer behaviour. It seems likely that optimisation will continue its move into the mainstream for this area of business.

Other insurers may find themselves struggling to build accurate models, as a result of these information asymmetries. Smaller insurers may simply not have sufficient volume or flexibility to apply these methods. For larger companies that place a significant portion of their business through intermediaries, the problems will relate to the amount of data their partners are willing to give them. In the worst case, these companies may be forced back to a bare-bone, cost-plus model, producing a completely commoditised policy at the lowest cost possible, effectively ceding all the customer value to the intermediary.

Practical considerations
So, if optimisation is right for your company, what are the key areas to consider? Before you even start work there are a number of issues that need careful thought, including legal considerations, price promises, market and brand positioning.

In the European Union, the gender directive allows, subject to the local implementation, the use of gender as a rating factor if it is accompanied by published information on the underlying risk characteristics. While it seems unlikely that any one individual will challenge the price differential between two policies, it remains a possibility that a pressure group will require an insurance company to explain its practices. If the optimisation assumes different price elasticity for men and women, the resulting premium differential may be hard to justify on the basis of the published risk data.

Marketing material often includes price promises, such as “10% off if you buy online”. Where this is the case you will need to include a restriction in the optimisation model, as it is unlikely that the exact promised differential will hold equally across all policies. Additionally, there is often an implicit price promise in the brand positioning, so that it might be desirable for a ‘premium’ brand to have a higher cost even if the optimisation implies a lower value is appropriate. Finally, a company will often also have a market position in mind, wanting to achieve a certain level of performance in terms of competitiveness by segment.

It should be remembered that placing additional constraints on the optimisation model will dampen the improvement in the final position. Re-running versions of the model with some of the constraints removed highlights how these limitations affect the results. This sensitivity testing can show some of the constraints should be removed, and this may of course require changes to the corresponding marketing messages.

It is important that the customer is not subject to confusing price changes following minor policy changes; this is true at the quotation stage as well as mid-term. It is not unusual for policyholders to ask about different excess levels or additional drivers. This problem can be managed by optimising only once for each customer, then using a schedule of adjustments based, for example, on the risk relativities for these minor alterations. For larger changes, a change of address or a change of vehicle, it is probably reasonable to re-optimise to ensure that the quotation remains competitive.

Implementation
There are two main ways to implement optimisation: ‘back office’ and ‘point of sale’. The back-office approach involves conducting the optimisation at regular intervals, alongside the regular rate review. At this point, all the key models are updated, and the optimisations run over a suitable portfolio. The output from this is an individually tailored premium for every policy reviewed. This output can be used directly for renewal premiums, with the tailored premium being fed straight into the renewal systems. Alternatively, a model can be fitted to the optimised premium in order to derive a traditional rating structure approximating the true optimal premium. The approximation can be poor but the rating structure will give premiums that are likely to perform better against the key targets than premiums calculated under the cost-plus approach. The results of the model of optimised premium can be used to price all business, or it can be used just for new business, using the fully optimised premium for renewals.

The back-office approach is useful where a rating structure is still required, for example, where intermediaries do not have access to a live pricing system. Most companies will have some systems that assume the existence of a rating structure, and so this approach allows rapid implementation of price optimisation at relatively low cost.

The point-of-sale approach seeks to set a premium using the most up-to-date information available at the time the quotation is prepared. This potentially means that an identical risk is automatically offered different premiums on different days only because the system’s perception of the market is different. This makes the premium responsive to the market and the model will move quickly to take advantage of changes. The point-of-sale algorithm still requires regular calibration, however, and it will need careful monitoring.

Conclusion
Price optimisation is here to stay, and many companies have been using these techniques with great success. There remain considerable challenges for small to medium-sized companies, and it seems likely that some of these will decide that optimisation is not the right choice for them. Intermediaries are the surest winners where they have a strong relationship with, and good data about, the customer. For all who decide to follow the optimisation route, there are considerable technological and management hurdles, and careful thought and preparation is vital to ensure success.


James Tanser works in the Insurance and Financial Services practice at Watson Wyatt. He leads the non-life predictive modelling team and manages the development of Watson Wyatt’s predictive modelling software, Pretium.