1

Enterprise Risk Management for P&C Insurance Companies

Shaun S. Wang

Robert T. Faber

2006-9-29 2

Agenda

ERM Research Project

Summary of Research Findings

Underwriting Psychology

Our Proposed Actions

2006-9-29 3

ERM Research Project

Co-sponsored by Casualty Actuarial Society Risk Management Section (SOA & CAS) ERM Institute International, Ltd

Researcher Team Shaun Wang (actuary, scholar) Bob Faber (executive, senior underwriter) Assisted by Project Oversight Team chaired

by John Kollar (ISO)

2006-9-29 4

Objectives & Time line

Objectives: Propose a fresh

ERM Theory that is applicable to all sectors

Make ERM operational for P&C insurers

Time Line Feb 16, started June 15, exposure

draft August 1, completed

2006-9-29 5

The Concept of Risk Dynamics

The big Universe consists of many projects (risks and opportunities), external players (customers, competitors), external forces (financial, regulatory)

An enterprise sits within the big universe selected projects, internal players & internal

forces “risk dynamics” refers to the interactions of

forces and players within and without the enterprise.

2006-9-29 6

We define ERM as

studies of the system of risk dynamics of the enterprise, including interactions among internal and external risk dynamics, and how players’ actions (including the risk management practices) can influence the behaviors of the risk dynamics, with the ultimate goal of improving the performance and resiliency of the system.

2006-9-29 7

Element of ERM Framework

1) Analyze the business model

2) Define the scope of business operations,

3) Identify operating constraints by regulators and rating agencies,

4) Measure sensitivity to external and internal forces,

5) Develop business or risk strategies to interact with the various forces

6) Monitor the dynamics

2006-9-29 8

P&C Insurer Risk Dynamics -- Players

Stock Analyst

Pricing

Claims &Reserving

I.T., H.R.Accounting

Investment

Marketing &Underwriting

Board ofDirectors

CEO; CFO;CIO; CRO

Competitor

RatingAgency

Regulator

2006-9-29 9

Net Effect of Diversification Benefit/Penalty It depends on the nature of business model,

& how you conduct the business Personal Lines: diversification essential for

managing catastrophe exposure Commercial Lines: we observed dramatic

differences in financial performance among companies with different underwriting/pricing practices

2006-9-29 10

Empirical Studies

A sample of 29 insurance companies:1) Small Companies (14 companies)

2) Jumbo Regional (7 companies)

3) Large National (8 companies) Focused on WC and GL

use gross loss ratios use gross loss triangles

2006-9-29 11

Mean and Deviation of Loss Ratio in relation to Size of Written Premium - WC

Workers's Compensation: Small Companies, Jumbo Regional, Large National

0%

20%

40%

60%

80%

100%

SC

: lo

gP

=1

0.4

SC

: lo

gP

=1

1.4

SC

: lo

gP

=1

1.5

SC

: lo

gP

=1

1.6

SC

: lo

gP

=1

1.8

SC

: lo

gP

=1

1.9

SC

: lo

gP

=1

2.1

SC

: lo

gP

=1

2.7

SC

: lo

gP

=1

2.8

SC

: lo

gP

=1

2.9

SC

: lo

gP

=1

3.3

SC

: lo

gP

=1

3.9

JR

: lo

gP

=1

2.1

JR

: lo

gP

=1

2.3

JR

: lo

gP

=1

4.3

JR

: lo

gP

=1

4.5

JR

: lo

gP

=1

4.6

JR

: lo

gP

=1

4.9

JR

: lo

gP

=1

5.0

LN

: lo

gP

=1

5.0

LN

: lo

gP

=1

6.0

LN

: lo

gP

=1

6.2

LN

: lo

gP

=1

6.4

LN

: lo

gP

=1

6.5

LN

: lo

gP

=1

6.8

LN

: lo

gP

=1

7.1

LN

: lo

gP

=1

7.6

Type of Company & Logarithm of Written Premium during 1985 - 2003

Deviation of Loss Ratio

Mean Loss Ratio

Linear (Deviation of Loss Ratio)

Linear (Mean Loss Ratio)

2006-9-29 12

Mean and Deviation of Loss Ratio in relation to Size of Written Premium - GL

General Liability: Average and Stdev of Loss Ratio By CompanySmall Companies, Jumbo Regional, and Large National

0%

20%

40%

60%

80%

100%

120%

SC

: lo

gP

=9

.93

SC

: lo

gP

=1

0.0

SC

: lo

gP

=1

0.2

SC

: lo

gP

=1

1.0

SC

: lo

gP

=1

1.4

SC

: lo

gP

=1

1.5

SC

: lo

gP

=1

1.6

SC

: lo

gP

=1

1.6

SC

: lo

gP

=1

1.7

SC

: lo

gP

=1

1.8

SC

: lo

gP

=1

2.1

SC

: lo

gP

=1

2.3

JR

: lo

gP

=1

2.8

JR

: lo

gP

=1

3.0

JR

: lo

gP

=1

3.1

JR

: lo

gP

=1

3.4

JR

: lo

gP

=1

4.1

JR

: lo

gP

=1

4.7

JR

: lo

gP

=1

4.8

LN

: lo

gP

=1

5.0

LN

: lo

gP

=1

5.0

LN

: lo

gP

=1

5.1

LN

: lo

gP

=1

5.3

LN

: lo

gP

=1

5.6

LN

: lo

gP

=1

5.6

LN

: lo

gP

=1

6.2

LN

: lo

gP

=1

6.4

Type of Company & Logarithm of Written Premium (1985-2003)

Deviation of Loss Ratio

Mean Loss Ratio

Linear (Mean Loss Ratio)

Linear (Deviation of Loss Ratio)

2006-9-29 13

Summary Result by Company Type

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

Small Companies Jumbo Regional Large National

Workers Compensation: Loss Ratios By Company Type

Average (MeanLoss Ratio)

Average(Deviationof Loss Ratio)

2006-9-29 14

Loss Reserve Practices

For a selected company and 1997 accident year, we observe booked loss ratios at 12/31/1997, 12/31/1998, up to 12/31/2004.

Loss Development for Accident Year 1997 =

{Estimated Incurred Loss as of 12/31/2004 } – {Initial Incurred Loss Ratio of12/31/1997 }

2006-9-29 15

Workers Compensation Loss Development

Worker's Compensation Reserve Development: Difference in Updated Loss Ratio and Initial Loss Ratio

-0.30

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15

0.20

0.25

1992 1993 1994 1995 1996 1997 1998 1999 2000

Large Companies

Small Companies

Jumbo Regionals

2006-9-29 16

General Liability Loss Development

Other Liability (Occurrence) Reserve Development:Difference in Updated Loss Ratio and Initial Loss Ratio

-0.30

-0.20

-0.10

0.00

0.10

0.20

0.30

0.40

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Large Companies

Small Companies

Jumbo Regionals

2006-9-29 17

Back Testing of the Chain-Ladder Development Method For each company and a given line of

business (WC or GL), we use the loss triangle data up to the end of year 2000 as input.

Applied the Chain-Ladder method to project future development to end of 2004.

Compared the projected losses with the actual observed losses by the end of 2004.

2006-9-29 18

Large National Companies

Li berty Mutual WC Loss Devel opment Actual vs. Proj ected Losses as of 12/ 2004

0

1, 000, 000

2, 000, 000

3, 000, 000

4, 000, 000

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

Acci dent Year

Actual Loss Proj ected Loss

Conti nental WC Loss Devel opment Actual vs. Proj ected Losses as of 12/ 2004

0

400, 000

800, 000

1, 200, 000

1, 600, 000

2, 000, 000

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

Acci dent Year

Actual Loss Proj ected Loss

2006-9-29 19

Small & Regional

Central WC Loss Devel opment Actual vs. Proj ected as of

12/ 31/ 2004

0

5, 000

10, 000

15, 000

20, 000

25, 000

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

Acci dent Year

Actual Loss

Proj ected Loss

West Bend WC Loss Devel opment Actual vs. Proj ected as of 12/ 31/ 2004

010, 00020, 00030, 00040, 00050, 00060, 00070, 00080, 000

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

Acci dent Year

Actual Loss Proj ected Loss

2006-9-29 20

Number of Insurer Upgrades vs. Downgrades, 1993 to 2003 (Source: Standard & Poor’s)

0

2

4

6

8

10

12

14

16

18

20

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003

Upgrades

Downgrades

2006-9-29 21

Leading Indicators vs Lagging Indicators Reserves can only be a lagging indicator.

-at 12 months, are almost never in an adverse position to precipitate big change

-24-36 months at best for true indications Rate change indications on renewal are adequate only if:

-constant policy form (SIRs, deductibles, limits, etc) dominates the book of business and-policies are rated on a true exposure basis, and-new business constitutes a small percentage of premium.

By 1997, most companies were rate monitoring and were still surprised by the depth of unmeasured rate decreases for 1998-1999.

2006-9-29 22

Underwriting Psychology

Underwriting psychology is more important to understand than underwriting philosophy.

Most underwriters think frequency more than severity. It is more tangible.

Most underwriters have not been in the position long enough to live with their tail. The adverse development always belongs to others.

Especially in larger companies, underwriting is not a career path, management is.

It is much easier to cut the rate on new business than on renewals. (You didn’t screw up the rating last year)

Underwriters learn the unwritten rule early: No matter what management says, those with the bigger books of business get rewarded. Premium is measured by underwriter, loss ratio and development is not.

2006-9-29 23

Price Monitoring

True price monitoring can only take place when anchored firmly on a rate for the exposure base, attaching at first dollar.

The goal of price monitoring is to project the loss ratio range. This can only occur if the proper rate level per exposure unit is known.

Failing that goal, projecting the direction and magnitude of the change still has value.

What has a higher priority with company management: expense ratio or price monitors

2006-9-29 24

Terms and Conditions

Experience rating by its very nature, using valued claims, can have a market or insured driven bias.

Deductible and Self Insured Retention valuations have to take place exactly in accordance with price monitoring techniques.

Changes in coverage have to be quantified, if only by estimate.

2006-9-29 25

Quantification of Risk

Risk level increases significantly as the attachment point rises. Deductibles and SIRs remove the more predictable part of the risk.

Risk knowledge decreases as risk size increases. It is harder to establish exposure bases and there are more variables to balance.

Balance of knowledge shifts to the buyer as size increases, increasing risk.

Risk level increases with the inexperience of the underwriter on the book. (Note we did not say inexperience of the underwriter)

2006-9-29 26

Monitoring Process

Establish base rates per exposure units to be measured against. This does not fully recognize risk quality, but creates a hard base line.

Measure renewals creating a history. Force identical quantification of all terms variation

(composite rates, deductibles, SIRs, etc) against these base rates.

Measure new business against the same standards Even when using experience rating as a methodology,

measure the resulting rate against the same basis. Take tracking of rate level change to desk level.

2006-9-29 27

Correcting the Decision-Making Process Effective price monitoring has to occur at the

same location as effective underwriting; the underwriter’s desk.

Price monitoring needs to become a real part of the evaluation process for rating agencies and analysts. (Remember the grading chart)

Senior management and analyst emphasis on top line premium growth and expense control is often at odds with effective price monitoring.

2006-9-29 28

ERM Implementation

First line of defense starts at the desk-underwriter

Track exposure data Track pricing data Integrate pricing /

underwriting/ claims / reserving

Pricing

UnderwritingClaims

Reserving

ERMFor P&C

Companies

29

Contact

1. Shaun Wang, swang@ermii.org, 678-524-9222

2. Bob Faber, robert.faber@risklighthouse.com, 847-428-2533