Transcript Slide 1
Price optimisation for personal lines insurance Richard Brookes 26 June 2013 Price optimisation Basic principle How do we calculate profit? – Conventional solution is as a constant proportion of cost (profit margin), but – By varying the profit margin for different customer segments we can take advantage of how they react to different price levels/changes – X% of cost • Profit Cost (risk, expenses etc) This can improve the average profit margin by around 3% of cost whilst retaining the same business volume © Taylor Fry Pty Ltd 2 Price optimisation Optimisation set-up • Maximise – • Average profit margin By varying – Individual policy premiums • Subject to – A global constraint of the number of policies in force, and – Individual profit margin constraints for each policy, say the interval [-$50, $50] around a “technical” profit margin • To do this we need a relationship between policy price and the number of risks in force © Taylor Fry Pty Ltd 3 Price optimisation Demand model • Logistic regression model of renewal rate – Policy characteristics just before renewal notice is sent out • Tenure, socio-demographic information • Behavioural indicators – Premium related predictors • Premium increase since last renewal • Premium in relation to competitor premia © Taylor Fry Pty Ltd 4 Price optimisation Individual demand curves Renewal curves for two policies 99% Renewal rate (%) 98% Compe titor price 97% Compe titor price 96% 95% Compe titor price 94% 93% 92% -30% -20% -10% 0% 10% 20% 30% Price change from current price ($) Policy 1 • Current price Policy 2 Combine the objective function, constraints, demand model and an optimisation algorithm © Taylor Fry Pty Ltd 5 Price optimisation Portfolio results Insurance profit vs Renewal rate 25,000,000 Insurance profit ($) 20,000,000 15,000,000 $4M (30%) Current 10,000,000 1½% 5,000,000 0 87% 88% 89% 90% 91% 92% Renewal rate © Taylor Fry Pty Ltd 6 Price optimisation Distribution of price adjustments Frequency of price adjustment Tend to be less elastic 80% Proportion of policies 70% These policies move to a competitor price or a point of slope change in the demand function 60% 50% 40% Tend to be more elastic 30% 20% 10% 0% -50 -40 -30 -20 -10 0 10 20 30 40 50 Price adjustment • Caution required - this can lead to a deterioration in the portfolio over time © Taylor Fry Pty Ltd 7 Price optimisation Optimisation cycle Data collection Demand modelling Ongoing data collection: Statistical models predicting how renewal and strike rates will change in response to price changes • Renewal rates and quote strike rates • Price flexing • Competitor rates Projection and optimisation Projections of portfolio volume given price changes Optimal price changes to maximise profit at given portfolio volumes • Customer characteristics © Taylor Fry Pty Ltd 8 Price optimisation The leading edge • The best basic optimisation uses – – – • Price testing and/or competitor rate deconstructions Hold out segments to assess ongoing effectiveness Accurate, up to date demand and risk cost models • Monitoring and recalibration of these models is important • Demand models must address slope and level Leading edge optimisation extends to: – – Real time optimisation of new business quotes Taking into account extra dimensions of behaviour (see diagram to the right) © Taylor Fry Pty Ltd Optimising over the full expected lifetime of each customer i.e. multi-year optimisation Optimisation taking into account of the multiple brands offered to each customer Optimisation taking into account each customer’s multiple product holdings 9