Document Properties
- Type of Publication: Memorandum
- Date: July 2013
- Audiences: P&C
Purpose
This memorandum sets out a revised proposal regarding the methodology for the measurement of earthquake exposure and earthquake financial resource formula. In addition, the memorandum presents a proposed revision to the annual Earthquake Exposure Data Form.
Background
In February 2013, OSFI published Guideline B-9 (Earthquake Exposure Sound Practices), which expects companies to account for risk that may result from exposures to more than one region. As outlined in the cover letter to Guideline B-9, OSFI agreed to continue discussions with the industry to finalize the methodology to measure earthquake exposure.
Earthquake Probable Maximum Losses (PMLs) have historically been based on the larger of the British Columbia (BC) or Quebec (QC) PMLs. The historical approach understates the PML for insurers with significant exposures in both earthquake zones. OSFI has been conducting consultations and revising relevant guidance to ensure that earthquake PML is being appropriately measured and accounted for in its guidelines and expectations.
To continue these discussions, OSFI formed a working group that included the British Columbia Financial Institutions Commission (FICOM), Québec’s Autorité des marchés financiers (AMF) and the Insurance Bureau of Canada (IBC). This working group was tasked to find appropriate alternative earthquake PML measures. The working group agreed that the financial resource requirements for earthquake exposures should take into account a company’s actual level of earthquake exposure, whether in one or more regions.
The working group used the following evaluation criteria when considering alternative measures:
The proposed measure would be at least as large as current one.
Financial resources for a company with country-wide exposure would reflect their higher exposures compared to a regional company.
The methodology would be simple enough to be auditable.
The proposed measure for a Canada-wide risk would be less than the sum of the required regions, effectively recognizing diversification assuming the earthquake risks from multiple regions are independent.
Proposal
The formula arrived at by the working group is as follows:
Country-wide PML500 = (East Canada PML5001.5 + West Canada PML5001.5)1/1.5
The formula was selected because it fulfills the working group’s mandate and meets or exceeds all evaluation criteria outlined above. The parameter 1.5 is chosen from a range of one to two as being appropriate at the current time.
It builds on the current, well-understood regional PML measure (Eastern and Western), expanding to account for risk exposure across Canada, and provides continuity with the current approach.
It is a sub-additive, one of characteristics of a coherent risk measure.
It is based on transparent mathematical formula and well-known risk measures.
It does not require any changes to modelling software or estimation technology.
It is simple to interpret and explain to senior management and Boards of Directors.
It is an internationally recognised and tested approach.
OSFI will continue to monitor earthquake models, and, if models change significantly, will consider changes to this formula at that time. Draft wording for the Minimum Capital Test (MCT) Guideline with the formula is attached in the appendix A.
Analysis
OSFI used earthquake filing data to study the overall impact of the formula. Detailed analysis is attached as Appendix C.
The overall PML regulatory requirement for the year 2022 will increase by $2.8 billion or 8.9% from the current $31.3 billion to $34.1 billion. The changes have the largest impact on the national writers where their BC and QC exposure are similar. Foreign companies, companies without personal property or large-sized companies (i.e. those with equity greater than $500 million) are more likely to have a greater than 15% increase in PML.
More than half of the affected companies already have financial resources related to earthquake risk at a level higher than the proposed 2022 regulatory requirement. Therefore, only $1.5 billion out of $2.8 billion will require additional financial resources. Further, the annual cost of an additional reinsurance requirement of $1.4 billion is likely to be less than $50 million.
The additional requirements, like the current requirements, will be fully phased in by 2022. Given the phase in period and a marketplace where more than half of the companies already meet the proposed requirement, OSFI expects the impact of the additional reinsurance cost will be manageable on an industry-wide basis.
Consultation Period
All federally regulated P&C companies are required to file an Earthquake Exposure Data Form with OSFI annually. Appendix B - the proposed Earthquake Exposure Data Form – is provided for your comments.
OSFI anticipates both the revised MCT guideline (Appendix A) and Earthquake Exposure Data Form (Appendix B) to become effective January 1, 2015. OSFI would appreciate receiving comments on both the proposed changes on or before August 31, 2013.
Questions concerning this memorandum should be addressed to your OSFI Relationship Manager or to Kun Zhang, Actuarial Division, by email at kun.zhang@osfi-bsif.gc.ca.
Questions with respect to the draft MCT guideline section may also be addressed to Judith Roberge, Capital Division, by email at Judith.Roberge@osfi-bsif.gc.ca.
Appendix A: Wording for the Minimum Capital Test (MCT) Guideline
4.6. Catastrophes
4.6.1 Earthquake Risk Exposure
Insurers must refer to OSFI’s Guideline B-9 (Earthquake Exposure Sound Practices) for details on OSFI’s expectations relating to P&C insurers’ earthquake exposure risk management. The MCT Guideline outlines the framework for quantifying the earthquake risk exposure for regulatory capital purposes and assessing insurers’ capacity and financial preparedness to meet contractual obligations that may arise from a major earthquake.
Earthquake Reserves Formula:
Earthquake Reserves = (EPR + ERC) × 1.25,
The amount of earthquake reserves includes Earthquake Premium Reserve (EPR) and Earthquake Reserve Component (ERC) and is added to total capital requirements for the purposes of the MCT/BAAT as capital/margin required at the target level.
* Please note, under the proposed MCT Guideline, capital required will be calculated at a pre-determined target level of confidence (CTE99%). A gross-up multiplier of 1.25 was used to bring current minimum capital requirements to the supervisory target level when data was not sufficient to use the CTE99% approach to determine the supervisory target capital required. This multiplier was derived using the relationship between the 120% and 150% taking into consideration the 20% capital requirement for operational risk that is currently implicitly reflected in the 150% supervisory target capital requirements. Total capital/margin required at target will be divided by 1.5 to arrive at minimum capital requirements that would feed into the MCT ratio and an explicit operational risk charge will be applied. Refer to Discussion Paper on OSFI’s Proposed Changes to the Regulatory Capital Framework for Federally Regulated Property and Casualty Insurers for further details.
where
ERC = {Earthquake Risk Exposure (Section 4.6.1.1)} − {Financial Resources (Section 4.6.1.2)}
4.6.1.1 Measurement of Earthquake Risk Exposure
The earthquake Probable Maximum Loss (PML) is the threshold dollar value of losses beyond which losses caused by a major earthquake are unlikely. Gross PML, which is the PML amount after deductibles but before catastrophic and other reinsurance protection, is used for calculating earthquake exposure for regulatory purposes. In this section, PML refers to a dollar amount that includes adjustments for data quality, non-modelled exposures and model uncertainty as outlined in Guideline B-9 (Earthquake Exposure Sound Practices).
Model approach
P&C insurers with material exposure to earthquake risk are required to use models to estimate their PML. Earthquake models include models licensed from various commercial vendors and maintained in-house or run by third parties on behalf of the insurer or can be an internal estimation technique or model developed by the insurer to OSFI’s satisfaction.
OSFI requires that P&C insurers continue to progress to a 500 year PML from earthquakes by 2022. Consequently, OSFI expects an insurer to meet a test of financial preparedness for a 500 year return period country-wide earthquake event by no later than the end of fiscal year 2022.
Country-wide PML500 = (East Canada PML5001.5 + West Canada PML5001.5)1/1.5
where
- East Canada PML500
is a one in 500 year Eastern Canada event, which represents the 99.8th percentile of the exceeding probability curve plus appropriate adjustments for data quality, model uncertainty, non-modelled business etc., using exceeding probability curves based only on earthquake exposure in Eastern Canada.
- West Canada PML500
is a one in 500 year Western Canada event, which represents the 99.8th percentile of the exceeding probability curve plus appropriate adjustments for data quality, model uncertainty, non-modelled business etc., using exceeding probability curves based only on earthquake exposure in Western Canada.
Recognizing the impact resulting from the new country-wide PML500 requirement, insurers may continue to phase-in their increased earthquake risk exposure until 2022 using the following formula:
Country-wide PML (Year) = Country-wide PML500 × (Year − 2014)/8 + MAX [East Canada PML420, West Canada PML420] × (2022 − Year)/8,
where
- Year
is the current reporting year (subject to a maximum of 2022)
- East Canada PML420
is a one in 420 year Eastern Canada event, which represents the 99.76th percentile of the exceeding probability curve plus appropriate adjustments for data quality, model uncertainty, non-modelled business etc., using exceeding probability curves based only on earthquake exposure in Eastern Canada.
- West Canada PML420
is a one in 420 year Western Canada event, which represents the 99.76th percentile of the exceeding probability curve plus appropriate adjustments for data quality, model uncertainty, non-modelled business etc., using exceeding probability curves based only on earthquake exposure in Western Canada.
Standard approach
P&C insurers should use the following standard formula for calculating their PML if;
- The insurer does not use an earthquake model for calculating its PML, or
- An earthquake risk exposure estimation technique is not to OSFI’s satisfaction
Country-wide PML = Maximum (East Canada PTIV − applicable policyholder deductibles, West Canada PTIV − applicable policyholders’ deductibles)
where
- PTIV
is the property total insured value for earthquake exposure, which includes building, contents, outbuilding, additional living expenses and business interruption.
4.6.1.2 Financial Resources
An insurer must have adequate financial resources to cover its earthquake exposure calculated in section 4.6.1.1. Financial resources that can be used to support the insurer’s earthquake exposure include:
- capital & surplus;
- earthquake reserves:
- earthquake premium reserve,
- earthquake reserve component;
- reinsurance coverage; and
- capital market financing.
Capital and Surplus
Insurers can count up to a maximum of 10% of capital and surplus as part of their financial resources to cover their earthquake risk exposure. This maximum limit is subject to supervisory discretion and can be lowered to an amount less than 10% of capital and surplus.
For Canadian P&C insurers, capital and surplus corresponds to a maximum of 10% of total equity as reported in the most recent P&C-1 return.
A Canadian branch of a foreign P&C insurer may use up to 10% of its worldwide capital and surplus to cover its earthquake exposure; however, it must be able to demonstrate that after an event, at least 10% of the worldwide capital and surplus is still available to meet its obligations to Canadian policyholders. The amount of worldwide capital and surplus corresponds to the Canadian dollars amount reported in the most recent P&C-2 return.
Earthquake Reserves
Earthquake Premium Reserve (EPR) is the voluntary accumulation of earthquake premiums. This amount must not exceed the country-wide PML500.
In the case of catastrophic reinsurance coverage not specific to earthquake risk, an allocation of the premium amount must be made. Companies should be able to demonstrate the reasonableness of their ratemaking.
Any earthquake premium contributed to the EPR must remain in the EPR unless there is a material decrease in the exposure.
Should an earthquake occur and trigger claims, companies would establish an unpaid claims provision as well as a provision for claims adjustment expenses. The EPR component would be reduced by an amount equal to the claims reserves.
Any reduction in EPR should be brought back into unappropriated surplus immediately.
The EPR is a component of the reserves amount reported on the balance sheet.
Earthquake Reserve Component (ERC) is an additional component used to cover an insurer’s earthquake exposure not covered by the other financial resources. The ERC must always be greater than or equal to 0 and is equal to:
ERC = {Country-wide PML500 × (Year − 2014)/8 + MAX [East Canada PML420, West Canada PML420] × (2022 − Year)/8} − capital and surplus − reinsurance coverage − capital market financing − EPR
Should an earthquake occur and trigger claims, companies would establish an unpaid claims provision as well as a provision for claims adjustment expenses. The ERC component would be reduced after the EPR, by an amount equal to the claims reserves.
Any reduction in ERC should be brought back into unappropriated surplus immediately.
The ERC is a component of the reserves amount reported on the balance sheet.
Reinsurance Coverage
Estimated reinsurance coverage available should be based on reinsurance policies in force on the day immediately following the end of the financial reporting period (e.g., policies in force on July 1 for MCT calculations as of June 30).
Capital Market Financing
Prior approval from OSFI is required before these instruments can be recognized as a financial resource in the calculation of the earthquake risk formula.
Refer to Guideline B-9 (Earthquake Exposure Sound Practices) for additional information.
Appendix B: OSFI Earthquake Filing Form
Name of Company(s): Line to fill out ______________________________________
Date (January 1, 201X): Line to fill out _______________________________________
1. Materiality
1.1. Check the appropriate response.
No exposure to earthquake risk to this company.
If your answer is a., please disregard the remainder of the form.
Non-material exposure to earthquake risk to this company and we will not be using a model to estimate our earthquake exposure, therefore we will use the standard approach formula outlined in the MCT Guideline.
If your answer is b., please complete section 1 and 2.
For all other companies using an EQ model, please complete the remainder of the form.
1.2. Describe how materiality was determined for earthquake risk.
1.3. If your answer is b, complete the following table:
Date of data
|
Region
|
Total Earthquake Exposed Insured Value
|
Policyholders’ Deductibles
|
|
East Canada
|
|
|
|
West Canada
|
|
|
2. Elements of the Reserving Formula
2.1. Complete the following table based on the requirement in MCT guideline.
|
Max (East, West) PML420
|
Country-Wide PML(Year)
|
Country-Wide PML500
|
PML (Total, all perils)
|
|
|
|
Catastrophe treaty collectable
|
Shaded cell
|
|
|
Other reinsurance collectable
|
Shaded cell
|
|
|
Retention
|
Shaded cell
|
|
|
10% of Capital & Surplus
|
|
Capital market financing
|
|
Voluntary EPR
|
|
Required ERC
|
|
Earthquake Reserves
|
|
Annual renewal date for - catastrophe program |
|
3. Model selection
3.1. If you have determined that earthquake risk is material for your company, you are required to use a model (external or internal) to determine your earthquake exposure. What models are used and who operates them for your company? Check as many as apply.
Model Information
|
Operated By
|
Provider
|
Model Name
|
Model Version
|
Internal Staff
|
Outsourced
|
Reinsurance Broker
|
Others
|
RMS
|
|
|
|
|
|
AIR
|
|
|
|
|
|
EQE
|
|
|
|
|
|
Internal Model (please describe in 3.2)
|
|
Shaded cell
|
Shaded cell
|
3.2. Internal models can vary significantly in sophistication (and simple approaches may be appropriate for some insurers). Please briefly describe your internal estimation technique or model approach.
4. Non-modeled perils and model operations
4.1. Indicate whether the following risks are considered in modeling the PML. If yes, describe the assumptions used in the loss estimation.
Risk
|
Included in PML
|
If yes, briefly describe the assumptions
|
Yes/No
|
Exposure growth
|
|
|
Business interruption
|
|
|
Claims handling expenses
|
|
|
Adequacy of insurance to value
|
|
|
Guaranteed replacement cost
|
|
|
Debris removal
|
|
|
Increased seismicity after a large event
|
|
|
Blanket coverage
|
|
|
Coverage extensions or clauses
|
|
|
4.2. Supplemental perils and model options:
|
Included in PML
|
Yes/No
|
Tsunami
|
|
Secondary uncertainty
|
|
Time dependency
|
|
4.3. Other adjustments information:
|
Yes/No
|
Dollar Amount ('000s)
|
Adjustments for exposure data quality:
|
|
|
Adjustments for model deficiencies-severity:
|
|
|
Adjustments for model deficiencies-frequency:
|
|
|
Adjustments for demand surge:
|
|
|
Other factors for prudence:
|
|
|
5. Data quality control
5.1. Describe the company’s quality control processes around data collection and entry including materiality standards.
5.2. Describe the company’s review processes independent of those responsible for data collection and data quality (e.g. internal or external review).
6. Model results for Primary Insurer
6.1. Complete the following table with the range of modeled output
|
Personal Property
|
Commercial Property
|
Auto
|
Total
|
West Canada
|
Shake (1)
|
Fire (2)
|
Other (3)
|
Shake (4)
|
Fire (5)
|
Other (6)
|
All Perils (7)
|
Sum (1) to (7)
|
TIV ($ Millions)
|
|
|
|
|
|
|
n/a
|
n/a
|
PML 250 ('000s)
|
|
|
|
|
|
|
|
|
PML 500 ('000s)
|
|
|
|
|
|
|
|
|
|
Personal Property
|
Commercial Property
|
Auto
|
Total
|
East Canada
|
Shake (1)
|
Fire (2)
|
Other (3)
|
Shake (4)
|
Fire (5)
|
Other (6)
|
All Perils (7)
|
Sum (1) to (7)
|
TIV ($ Millions)
|
|
|
|
|
|
|
n/a
|
n/a
|
PML 250 ('000s)
|
|
|
|
|
|
|
|
|
PML 500 ('000s)
|
|
|
|
|
|
|
|
|
|
Personal Property
|
Commercial Property
|
Auto
|
Total
|
Worldwide
|
Shake (1)
|
Fire (2)
|
Other (3)
|
Shake (4)
|
Fire (5)
|
Other (6)
|
All Perils (7)
|
Sum (1) to (7)
|
TIV ($ Millions)
|
|
|
|
|
|
|
n/a
|
n/a
|
PML 500 ('000s)
|
|
|
|
|
|
|
|
|
6.2. Complete the following table in regard to Average Annual Loss (AAL) as produced by your model:
Top 3 forward sorting area (FSA) in terms of AAL
|
AAL (‘000s)
|
1.
|
|
2.
|
|
3.
|
|
Overall AAL
|
|
6.3. Divide the worldwide PML at the reserving level for the current year-end into the following categories:
World-wide PML
|
('000s)
|
Primary risks
|
|
Assumed risks
|
|
6.4. Divide the reinsurance coverage used to cover the PML for the end of relevant period of exposure into the following categories:
Reinsurance coverage
|
('000s)
|
Canadian companies
|
|
Foreign companies (branches)
|
|
Lloyd's
|
|
Non-registered affiliates
|
|
Other non-registered
|
|
Total reinsurance coverage
|
|
6.5. Complete the following table as to the level of detail to which data is coded
|
% of TIV
|
Personal Property
|
Comm. Property
|
Geocoding
|
|
|
Exact street address
|
|
|
6-digit postal code
|
|
|
FSA
|
|
|
Other
|
|
|
7. Model results for Reinsurer
7.1. Complete the following table with the range of modeled output
|
Excess of Loss
|
Proportional
|
Facultative
|
Total sum (1) to (6)
|
West Canada
|
Catastrophe (1)
|
Per Risk (2)
|
Proportional (3)
|
Other (4)
|
XOL (5)
|
Proportional (6)
|
TIV ($ Millions)
|
|
|
|
|
|
|
|
PML 250 ('000s)
|
|
|
|
|
|
|
|
PML 500 ('000s)
|
|
|
|
|
|
|
|
|
Excess of Loss
|
Proportional
|
Facultative
|
Total sum (1) to (6)
|
East Canada
|
Catastrophe (1)
|
Per Risk (2)
|
Proportional (3)
|
Other (4)
|
XOL (5)
|
Proportional (6)
|
TIV ($ Millions)
|
|
|
|
|
|
|
|
PML 250 ('000s)
|
|
|
|
|
|
|
|
PML 500 ('000s)
|
|
|
|
|
|
|
|
|
Excess of Loss
|
Proportional
|
Facultative
|
Total sum (1) to (6)
|
Worldwide
|
Catastrophe (1)
|
Per Risk (2)
|
Proportional (3)
|
Other (4)
|
XOL (5)
|
Proportional (6)
|
TIV ($ Millions)
|
|
|
|
|
|
|
|
PML 500 ('000s)
|
|
|
|
|
|
|
|
7.2. Complete the following table in regard to Average Annual Loss (AAL) as produced by your model.
Top 3 cedents in terms of AAL
|
AAL (‘000s)
|
1.
|
|
2.
|
|
3.
|
|
Overall AAL
|
|
7.3. Complete the following table in regard to your catastrophe-excess of loss coverage (Cat XOL).
Top 3 cedents in terms of Cat XOL
|
Limit Provided
|
1.
|
|
2.
|
|
3.
|
|
Overall catastrophe excess of loss
|
|
7.4. Divide the reinsurance coverage used to cover the PML for the current year-end into the following categories:
Reinsurance Coverage
|
('000s)
|
Canadian companies
|
|
Foreign companies (branches)
|
|
Lloyd's
|
|
Non-registered affiliates
|
|
Other non-registered
|
|
Total reinsurance coverage
|
|
Appendix C – Data study
The proposed approximation formula to the country-wide (CW) 500 year PML is based on an adjustment to the well-known “square root of sum of squares”. The square root of sum of squares is appropriate for normal loss distribution. However, for the earthquake peril with heavy-tailed loss distribution, the parameter 1.5 is chosen from a range of one to two as being appropriate at the current time. OSFI studied the proposed impact using data from 2012 Earthquake Data Filings and Appointed Actuarial Reports.
1. Summary of the companies used in analysis:
There were 82 companies used in the analysis.
There are 10 companies that do not have earthquake (EQ) exposure but share reinsurance with other affiliated companies of the same group, so they are included within the capital adequacy study.
There are 57 companies excluded due to a lack of EQ exposure in BC or QC.
There are 18 (out of 167) companies excluded due to EQ data error in their filings.
Table1: Summary of Data (*Net earned premium is from all lines combined.)
$ in Million |
BC Net Earned Premium |
QC Net Earned Premium |
Total Equity |
# of companies |
companies used in analysis |
2,742 |
4,047 |
29,216 |
82 |
companies no EQ exposure but share reinsurance |
90 |
42 |
878 |
10 |
companies excluded due to lack of EQ exposure |
172 |
136 |
7,027 |
57 |
companies excluded due to EQ data error |
224 |
704 |
3,361 |
18 |
Total |
3,228 |
4,929 |
40,482 |
167 |
companies excluded due to EQ data error (% of total) |
7% |
14% |
8% |
11% |
2. Overall PML Movement
The overall PML regulatory requirement for the year 2022 will increase by $2.8 billion (1.5 + 1.3) or 8.9% from the current $31.3 billion to $34.1 billion.
More than half of the affected companies already have financial resources related to earthquake risk at a level higher than the proposed 2022 regulatory requirement. Therefore only $1.5 billion out of $2.8 billion will require additional financial resources.
Graph 1: the overall movement of PML
Table2: The Overall Movement of PML ($ in Million)
Count of Companies |
BC PML500 |
QC PML500 |
2012 Current PML |
2022 Max
(BC500, QC500) |
2022 CW PML500 |
82 |
30,974 |
7,882 |
26,668 |
31,338 |
34,103 |
Incremental increase |
|
|
|
4,670 |
2,765 |
Cumulative increase |
|
|
|
4,670 |
7,435 |
3. Impact of CW PML500 to Max PML500
Impact = CW PML500 / Max PML500 − 1
Roughly 90% of the market (measured by exposure to earthquake) will have increases in minimum requirements of less than 15%.
Foreign companies, companies without personal property or large-sized companies (i.e. those with equity greater than $500 million) are more likely to have a greater than 15% increase in PML.
Graph2: Impact in PML by Characteristics of Companies
4. Financial Resource Adequacy (group level) (in ‘000)
Financial Resource Available = Reinsurance Collectable + 10% Capital and Surplus + Earthquake Reserves
If the Catastrophe limit and deductible are available in Appointed Actuary Report, these data points were used to calculate financial resource available. Otherwise, the reinsurance collectable from EQ filing is used.
If the company did not have EQ exposure, its capital was still included in financial resource available for the overall insurance group. There are 10 such companies in this category.
Additional reinsurance required is $1.4 billion. The additional annual industry aggregate reinsurance cost is estimated to likely be less than $50 million (assuming rate on line (ROL) is 3%).
Table3: Financial Resource Adequacy & Reinsurance Cost ($ in Million)
Meet 2022 CW PML Requirement
|
2022 Max(BC500, QC500)
|
2022 CW PML 500
|
Finance Resource Available
|
Equity
|
Addi. Reins. Required
|
Reins. Cost 3% ROL
|
Reins. Cost To Equity Ratio
|
no
|
14,395
|
15,856
|
13,746
|
9,156
|
1,393
|
41.8
|
0.46%
|
yes
|
16,943
|
18,247
|
37,786
|
20,938
|
|
|
|
Total
|
31,338
|
34,103
|
51,532
|
30,093
|
1,393
|
41.8
|
0.14%
|
Graph3: Financial Adequacy by Characteristics of Companies