Projections distorted if historical premium doesn't reflect past rate changes
Develop premium to ultimate levels when premium is still changing
Project to expected future premium level
Includes premium trend and changes in mix of business
Calendar Year Aggregation and Accident Year Aggregation of premium
All premium transactions within calendar period
Regardless of policy effective date
Represented graphically by squares
Policy Year Aggregation of premium
All premium transactions on policies with effective dates during the year
Represented graphically by parallelograms
Need 24 calendar months to complete (assuming annual policies)
Reason that calendar year more often used
Describe Written Premium
1. Dollar amounts charged by an insurer for policies written during a specific time period
2. Calendar Year Written Premium includes:
Premium for policies with effective dates during the calendar period
Mid-term adjustments during calendar year regardless of policy effective date
Describe Policy Year Written Premium
All premium and adjustments for policies with effective dates in policy period
What is an Earned Premium?
Amount of the policy premiums that have been exposed to risk during a specified time period
Premium for coverage that has been provided
Represents portion of total premium that insurer can retain if policy is canceled
Describe a Policy Year
Earned premium assign to year policy is effective
When policy expires, PY Written = PY Earned
Unlike CY, premium for one policy cannot be earned in two dierent policy years
Policy year premium is not fixed at end of PY for audited LOBs
Will continue to develop after end of PY period
Describe an Unearned Premium
Portion of written premium for which coverage has not been provided
Amount of premium company has not yet earned
Insured is due back upon policy cancellation
Written Premium = Earned Premium + Unearned Premium
Describe an Inforce Premium
Full-term premium of all policies in effect at a specific point in time
Best estimate of the company's mix of business as of a given date
Current Rate Level Adjustments to Premium
Adjustments made to historical premium to account for rate changes
Failure to adjust will incorrectly project future premium
Extension of Exposures Adjustments to Premium
Method to bring rates to current level by re-rating each policy using current rates
Advantage: Most accurate method to bring rates to current level
Disadvantage: Need detailed data
Parallelogram Method (a.k.a. geometric method) Adjustment to Premium
Adjusts the aggregated historical premium by an average factor to get premium on-level
Assumes exposure is uniformly distributed over time
Parallelogram Method factors that affect application of method
Policy term
Calendar year versus policy year
Whether rate change affects policies midterm or only on effective date of policies
What are the Standard Calculations for the Parallelogram Method?
Group policies into rate level groups according to timing of each rate change
Calculate the portion of earned premium corresponding to each rate level group
Calculate the cumulative relative rate level for each group
Calculate the average relative rate level for each year
Calculate the on-level factor = current relative rate level / average relative rate level
Apply on-level factor to earned premium for appropriate year
What are some of the problems with the Parallelogram Method
Problem 1 - Assumes policies written uniformly throughout the year
Reasonable for some lines but inappropriate for others
Problem 2 - Generally applied at aggregate level using overall average changes
Adjusted premium will likely not be acceptable for classification analysis
What is Premium Development needed for?
At time of analysis, ultimate premium for experience period may be unknown
Incomplete year of data
Premium audits
Premium Trend
Distributional changes causing average premium level to change
Rating Characteristic: Example - Homeowners Amount of Insurance
Premium Level Changes: Example - Insurer chooses to move policyholders to a higher deductible
Shifts in Mix of Business: Example - Company purchases another insurer's book of business
What are the steps to adjust for premium trend?
Determine how to measure any changes that have occurred
Determine if distributional shifts were one-time events or continuous
Judgmentally include any expected future shifts
Written vs. Earned Premium to select premium trend
Argument for Earned
Used in most other parts of ratemaking analysis
Argument for Written
Allows for capture of more recent data because premium not earned until well after it is written
Using earned would postpone recognition of effects of most recent changes
Trends in written will eventually emerge in earned
Selection of trend
1. Often use quarterly average written premium
More responsive than annual average written
2. Adjust data for rate changes and other one-time effects
If not done, selected trend will account for change, which will then adjust for it twice
3. Compare quarterly average written to prior year quarterly average
4. May fit exponential or linear trend to data
Describe One-Step Trending
Applies a single annual trend factor across the entire experience period and into the future policy period
Each year's earned premium in the experience period is trended separately to same point in time
One Step Trending Period
1. Trend period from the average experience period written date to the average effective period written date
2. Average written date is half the policy term earlier than the average earned date
3. Average written date for future policy period is effective date of new rates plus half of the period in which the rates are projected to be effective
Items that can affect the length of the trending period
Policy term
Historical premium is policy year rather than calendar year
Length of time rates are expected to be in effect
Difficulties in applying one-step trending
Changes in average premium vary signicantly year-by-year
Historical changes are significantly different than expected future changes
What are the advantages of Two-Step Trending?
1. Single annual premium trend is not appropriate for each year in experience period
2. Trend in historical period is significantly different that what is expected in future
What is Step 1 in Two-Step Trending?
Trend all experience period data to the latest trend data point
Trend premium to the midpoint of the latest trend data point in the series
May use latest available quarter average written
May use latest annual average written
What is Step 2 in Two-Step Trending?
Trend from the latest data trend point to the average effective period written date