2. There are other considerations, such as a more in-depth treatment of variable expenses, investment income, cash flows, claims development, return on capital and fixed expenses.
3. How we can build these different rating components into a pricing system.
4. Most statistical analyses are compromised in order to fit into a rigid table design from which the premiums are calculate. The situation is further exaberated when profit margins especially when loads and discounts given for a variety of reasons.
SOLUTION
1. There should be a more flexible pricing engine that calculates the premium in several stages. This is shown schematically below.
2. This is component pricing. The quotation engine separately calculates each individual element of the premium, and the total premium only calculated at the end.
3. The profit optimization process is the most interesting part of a component pricing system as it is subject to corporate goal constraints. Whilst the other parts of component pricing can be driven from a series of look-up tables, profit loadings are best implemented by an optimization algorithm.
ALTERNATIVE USE
1. One way to cater for diffent corporate goals is to define a utility function that explicitly maps a trade-off between volumes and profitability. The utility function effectively adjusts the expected profit curve so that a new maximum can be derived.
2. Another factor which the insurer might want to build into the equation is profits made by crossselling other products. This can be done by either reducing the breakeven premium Po by a fixed amount for each policy, or by building a cross-sell model to reflect different cross-sell propensities for different customers.
3. Another alternative use of this process is to reverse engineer the whole process, and use the expected claims costs and demand behaviour to identify the most valuable customers under a set of restricted premium rates. These could bc ranked in order of value and scored in terms of customer value. It may be then possible to use marketing means to attract those more desirable customers.
4. The techniques can also be used to effectively cost the consequences of particular pricing action taken, both in terms of expected claims cost and the volumes of business written. The models developed will be based on each individual customer's characteristics and will move dynamically with changing market activity, providing early feedback on pricing decisions taken.
CONTROLLING THE PROCESS
1. The actual experience of each component can be monitored against the expected experience. For example, the expected risk premium can be compared against the actual claims experience; the expected conversion and renewal rates can be compared against actual conversion and renewal rates.
2. a profit forecast can easily be produced for each month's' written business. This will give an early indication to management of reduced levels of profitability; furthermore, the cause of the reduction in profitability can easily be identified. By monitoring each element of the price separately, different people or departments can be made accountable for achieving certain goals.
THOUGHTS
1. By estimating both the risk premium and demand functions for individual customers, these could be combined and used to set optimal profit loads tailored to the behaviour of each individual customer. Alternatively, the profit load can be estimated for any given premium. This concept can be extended to cover longer time horizons to create life-time pricing models. Alternatively corporate goals can also be built into the optimization algorithms.
(Source: Karl P Murphy, Michael J Brockman, Peter K W Lee)
