The Contest Part I

1
The Contest Part I

• CAS Seminar on Ratemaking
• March 9-10, 2000
• Session SPE-47
• MIGHTY DUCK ACTUARIAL CONSULTANTS
• Jerome E. Tuttle, FCAS St. Paul Re
• Stephen J. Talley, ACAS St. Paul Re

2
The Contest

• We decided to
• Concentrate on Standards
• Borrow from Loss Reserves, especially
• AAA Note on Reserve Opinions
• Berquist Sherman, PCAS 1977
• Be innovative, especially in cover letter and
  narrative
• Have fun!

3
Cover Letter

• Qualifications of rate filer
• In my opinion these rates
• Meet requirements of Ins. Laws of Texas, incl.
  articles 5.01,...,21.81
• Are adequate, not excessive, not unfairly
  discriminatory
• Are calculated based on accepted ratemaking
  standards principles, including CAS Statement
  of Ratemaking Principles ASB Standards 9, 12,
  13, 23, 29, 30 textbooks published papers
• Peer review of rate filing
• Disk with all exhibits as spreadsheets
• Picture of duck as letterhead

4
Background Research

• We read CAS principles, relevant ASB Standards
• We read relevant insurance laws of Texas
• We browsed 20 years of CAS papers for potentially
  useful ones, and we referred to them
• We talked to actuaries, agents, claims folks,
  marketing folks, underwriters, as a
  reasonability check on our assumptions
• We questioned the accuracy of the data, and did
  not use data we were unsure of

5
Unusual Things We Did

• Triangulated CY Earned Premium
• Calcd Incd Indication Paid Indication
• Tempered a trend factor, justified in part from
• People we spoke with
• Published paper in another line
• Decided certain expenses should not be passed on
  to insureds
• Allocated expenses between fixed variable,
  between salary-based non salary-based

6
Effect of 9/95 Tort Reform

• We interpret Texas Code Article 5.14 as follows
• A single loss reduction percentage of 11.4,
  regardless of specific years in the ratemaking
  database.
• We spoke with people who feel the 11.4 is OK
• But this single percentage implies A/Y 96 losses
  should be reduced by 11.4, even though it
  already reflects tort reform savings
• We complied with law, feel it conflicts with
  sound practice disclosed our objection

7
Actuarial Adjustments
8
Actuarial Adjustments

• Premium On-Leveling
• Traditional parallelogram method impacted by
  changing level of exposures.
• Use the discrete approximation offered by Frank
  Karlinski in his discussion of Miller and Davis
  paper, A Refined Model for Premium Adjustment.
• Given Written Premiums by Calendar Quarter,
  calculate Earned Premiums by Calendar Year.
• Given Rate Level by Calendar Quarter, calculate
  weighted average Rate Level by Calendar Year
  using Earned Premium as weights.

9
Actuarial Adjustments
10
Actuarial Adjustments

• Frequency Trend
• Frequency Paid Counts / Earned Exposures.
• Distorted when level of exposures is changing.
• Compare Assigned Risk Frequency Trend with
  Voluntary Market Frequency Trend.
• Adjust Assigned Risk Data to account for changing
  level of exposures.

11
Assigned Risk Data
12
Voluntary Market Data
13
Actuarial Adjustments

• Frequency Trend (BI)
• Lag exposures using paid loss pattern. Try to
  match paid counts with the exposures that
  generated counts.
• 158,839 exposures written in 1st quarter of 1993.
• 40 earned between 1st quarter 93 and 1st
  quarter 94.
• 45 earned between 1st quarter 94 and 1st
  quarter 95.
• 15 earned between 1st quarter 95 and 1st
  quarter 96.
• 7,942 exposures earned in 1st quarter of 93
  from exposures written in 1st quarter of 93.
  (158,839 x 0.125 x 0.4)
• For each calendar quarter, sum earned exposures
  from all exposure-writing quarters.

14
Adjusted Assigned Risk Data
15
Actuarial Adjustments

• Frequency Trend
• Unadjusted Assigned Risk Data indication 18.5 ,
  22.8
• Voluntary Market Data indication 0.1 , -3.7
• Adjusted Assigned Risk Data indication -5.6 ,
  -24.0
• Select -6.5 through 12/31/97.

16
Actuarial Adjustments

• Frequency Trend
• Considerations
• Using Paid Loss development pattern to estimate
  Paid Count development pattern.
• Effect of Tort Reform in later data points.
• Quality of Assigned Risk insureds worsening
  over time.

17
Actuarial Adjustments

• Frequency Trend
• PD not as distorted by changing exposure level.
• PIP and UM Assigned Risk Data indications not
  credible.
• Used voluntary Market Data indications for PIP
  and UM.

18
Actuarial Adjustments

• Severity Trend
• Distortion of changing exposure levels not as
  severe as for frequency.
• For BI PD, indications from voluntary and
  assigned risk data are similar.
• PIP UM not credible - used voluntary indication.