1. The document discusses using price elasticity in insurance pricing models to optimize prices and profits while staying within regulatory guidelines. 2. It describes how European insurers have benefited from explicitly incorporating price elasticity, while US insurers commonly make "intuition-led" pricing decisions that are implicitly motivated by elasticity concepts. 3. The author argues that insurance pricing could be improved by taking a more scientific approach to incorporating price elasticity data, lifetime customer value, and other factors beyond pure loss costs.

1. ®
Squeezing Price Elasticity into the
Pricing Matrix
By
Deepak Ramanathan & Ed Combs
Fractal Analytics Inc.
Presented at
Auto Insurance Report National Conference 2011
Confidential | Copyright © Fractal 2011 1

2. ®
Key points to be covered in the next 25 mins
1. European insurers have realized substantial benefits by using
price elasticity in their pricing models
2. In the US, though European approach is prohibited, ‘intuition led’
changes motivated by price elasticity occur
3. By being a bit more scientific while incorporating elasticity we
can improve performance
– Without changing the rating structure
– Without introducing new variables in the ROC
– While maintaining ‘loss cost’ as the most important component of pricing
4. Using price elasticity, we can optimize prices while staying
within the allowable band of loss cost indicated relativities
Confidential | Copyright © Fractal 2011 2

3. ®
Elasticity based pricing & optimization is a well known
concept in the insurance industry
It is about reaching the efficiency frontier
More volume /
same profit
Volume (GWP)
More profit /
same volume
Loss cost based pricing
Profit (1 – loss ratio)
Confidential | Copyright © Fractal 2011 3

4. ®
Why is Elasticity Based Pricing important?
10%
7.5%
Premium
Lost Opportunity
5.3%
Mth 1 Mth 6
Proposed Realized
Elasticity Based Pricing has huge potential to improve both
top-line and bottom-line
Confidential | Copyright © Fractal 2011 4

5. ®
European insurers leverage elasticity in pricing
Selective Discounts Frequent Rate Changes
Discounts are offered at point of sale Prices are changed rapidly, some times
purely based on elasticity multiple times in a day
Price Testing
Insurers experiment with prices in the
marketplace to create data for elasticity
UK market Regulations make it easy for insurers to leverage price elasticity
Confidential | Copyright © Fractal 2011 5

6. ®
In the US, elasticity had not been widely used in the
past because of…
Regulatory constraints Data hurdles
Two households with the same rating
characteristics cannot be charged Lack of good data, inability to price test
different rates
The wait is over
 We can capture part of the gain within the regulatory framework
 People like us are already doing this
Confidential | Copyright © Fractal 2011 6

7. ®
…But ‘intuition led’ Elasticity Based Pricing is
common
 Aren’t factors frequently revised after meeting the sales team?
 Isn’t actuarially justified discount such as persistency discount
overridden?
 Isn’t rate capping used frequently to avoid disruption?
What is the rationale for such "intuition led", "common sense led”
decisions?
What if we could make these decisions more data driven?
We often override pure loss cost in favor of more revenue
Confidential | Copyright © Fractal 2011 7

8. ®
Scientific Elasticity Based Pricing requires three
essential components
Loss Cost Modeling
Price Elasticity
Cost of doing business
 Measures customer risk PRICING  Measures customers’
 Helps in segmenting reaction to price changes
customers based on risk  Helps realize different profit
attributes margin depending on price
sensitivity
Future revenue potential /
Lifetime value (LTV)
Index for customer loyalty and
cross-sell potential
This helps insurers go beyond ‘cost plus’ pricing model and incorporate
key customer characteristics
Confidential | Copyright © Fractal 2011 8

9. ®
…And requires a lead time of up to a year
Components of traditional pricing Additional components needed
workbench
• Policy & Quote data • Policy level elasticity data
• Factors used in pricing  Estimated renewal / conversion
• Factors Relativities  Estimated renewal premium
 Current • Policy level LTV data
 Proposed  Estimated survival
 Indicated  Estimated future cross-sell
• Pricing engine • Elasticity & LTV measurement at various
• Competitor data price changes
Key Insights
• It takes 4 to 6 months to build elasticity & LTV capabilities
• It takes an additional 6 months to run a pilot and validate results
Confidential | Copyright © Fractal 2011 9

10. ®
We can optimize prices by varying a limited number
of rating factors using elasticity & LTV
Age Limits Territory
35 – 40 years 30/50 Jacksonville
Upper Confidence Upper Confidence Upper Confidence
Level Level Level
Selected Relativity
Model Relativity Model Relativity Selected Relativity=
Model Relativity
Selected Relativity
Lower Confidence Lower Confidence Lower Confidence
Level Level Level
Elasticity & LTV provide new insights about the customer . This can help us
select “better” relativities
Confidential | Copyright © Fractal 2011 10

11. ®
Tests show that this approach works
Case Study: Simulation results from a large P & C insurer in the US
US Regulatory UK Market Scenario
Parameter
Scenario (-10% to +10%)
Average premium
0% (by design) 0% (by design)
change
Number of policies +0.9% +3.7%
Written premium +3.5% +9.7%
Loss ratio -1.0% -2.3%
Better retention & better top line growth, while remaining risk neutral
Confidential | Copyright © Fractal 2011 11

12. ®
The wait is over…Don’t get left behind
 We can capture part of the gain within the regulatory
constraints
 Our estimate suggests companies that account for ~ 35% of
the market share:
 Either have this capability
 Or in the advanced stages of developing it
 We expect this number to grow in the next year
 Other benefits include better forecasting and objective
decision making
Confidential | Copyright © Fractal 2011 12

13. ®
Though challenges exist, this is no rocket science
Defining price elasticity No price testing
The standard definition has to be Due to regulations, price testing cannot
modified to include rate avoidance, etc. be used to estimate elasticity
Non-linear nature Balancing elasticity and LTV with loss cost
Elasticity LTV
Elasticity Based Pricing is highly non- Elasticity and LTV can be first computed
linear with multiple local optima at a segment level & then at a policy level
Confidential | Copyright © Fractal 2011 13

14. ®
In Summary…
• Optimization techniques have evolved to incorporate elasticity
• Due to limited regulations and potential upside, European
companies have been early adopters
• US insurers are realizing the value of elasticity led optimization
• While the accrued benefits may not be as high as in the European
scenario, there is money to be made within regulatory constraints
• It takes a year to build this capability and go to market with it
Confidential | Copyright © Fractal 2011 14

15. ®
Thank You
Deepak Ramanathan Ed Combs
Insurance Director – Fractal Analytics Insurance Advisor – Fractal Analytics
deepakr@fractalanalytics.com ed@fractalanalytics.com
323-719-4165 ed@combsconsults.com
818 706-3467
Fractal Analytics Inc.
www. fractalanalytics.com
Confidential | Copyright © Fractal 2011 15

16. ®
Nothing
here
Confidential | Copyright © Fractal 2011 16