enhanced governance frameworks for assumption management - soa · pdf filesession 121 pd,...
TRANSCRIPT
Session 121 PD, Enhanced Governance Frameworks for Assumption Management
Moderator:
Paula M. Hodges, FSA, MAAA
Presenters: Paula M. Hodges, FSA, MAAA Jason E. Kehrberg, FSA, MAAA
Daniel Burt Moulden, ASA, CERA, MAAA
SOA Antitrust Disclaimer SOA Presentation Disclaimer
Jason Kehrberg, PolySystems, Inc.
121PD: Enhanced Governance Frameworksfor Assumption Management
10/25/2016
Agenda• Polling questions• Factors driving increased assumption governance• PBR implications• Key components and leading practices
2
Polling questions
3
What is your employer's primary line of business?
4
A. Life insuranceB. AnnuitiesC. Health insuranceD. ConsultingE. SoftwareF. GovernmentG. Non-profitH. RetiredI. Other
Polling question 1
47%
18%
10%12%
4%3%
0% 0%
7%
54%
14%
8%
14%
2% 2% 1% 1%
4%
A B C D E F G H I
2017 2012
Which of the following topics are you most interested in?
5
Polling question 2
A. Tools for managing multiple sets of assumptions across multiple products, frameworks and functional areas
B. Governance processes for setting and approving assumptions
C. Requirements for assumption governance under PBR/VM-20
D. Setting prudent estimate assumptions for PBR/VM-20
E. What to do when credible experience does not exist
31%34%
10%
6%
18%
A B C D E
Does your company have a documented assumption governance/management process?
6
Polling question 3
A. Yes, in place for 3 or more yearsB. Yes, in place for less than 3 yearsC. No, plan to developD. No, not consideringE. Not sure or not applicable
24%27%
35%
7% 7%
19%22%
18%21% 20%
A B C D E
2017 2012
Does your company have an assumption governance/management committee in place?
7
Polling question 4
A. Yes, in place for 3 or more yearsB. Yes, in place for less than 3 yearsC. No, plan to developD. No, not consideringE. Not sure or not applicable
17%
33%
26%
14%
10%
A B C D E
Does your company have processes in place to ensure best estimate assumptions are kept consistent across multiple products, frameworks and functional areas?
8
Polling question 5
A. YesB. NoC. Not sure or not applicable
33%39%
28%
63%
37%
0%
A B C
2017 2012
Where in your company are actuarial experience studies conducted?
9
Polling question 6
A. Multiple departmentsB. Corporate Actuarial / Valuation /
ModelingC. Special / central experience study
departmentD. ALM / ERME. ProductF. OtherG. Not sure or not applicable
28%24% 24%
1%
15%
1%6%
20%
37%
20%
1%
7%3%
12%
A B C D E F G
2017 2012
Where in your company are actuarial assumptions set?
10
Polling question 7
A. Multiple departmentsB. Corporate Actuarial / Valuation /
ModelingC. Special / central experience study
departmentD. ALM / ERME. ProductF. OtherG. Not sure or not applicable
43%
31%
13%
0%
6%3%
6%
A B C D E F G
Does your company use the same resources to conduct experience studies and set assumptions?
11
Polling question 8
A. AlwaysB. UsuallyC. SometimesD. NeverE. Not sure or not applicable 19%
36%
29%
11%
5%
A B C D E
Where do official actuarial assumptions “live” at your company?
12
Polling question 9
A. Centralized database (e.g. SQL)B. Centralized Excel filesC. Centralized actuarial software production
setup(s)D. Decentralized Excel files in many locationsE. Decentralized setups of actuarial softwareF. Semi-centralized combination of aboveG. OtherH. Not sure or not applicable
6%
10%
6%
20%
12%
29%
3%
14%
A B C D E F G H
Does your company conduct sensitivities as part of the assumption setting process, e.g. when setting margins?
13
Polling question 10
A. AlwaysB. UsuallyC. SometimesD. NeverE. Not sure or not applicable
14%
26%
30%
14%
17%
A B C D E
Does you company utilize predictive analytics in the assumption setting process?
14
Polling question 11
A. YesB. No, plan to developC. No, not consideringD. Not sure or not applicable
12%
29%31%
28%
A B C D
What is your primary source of experience data on which to set mortality assumptions?
15
Polling question 12
A. Company dataB. Industry dataC. Consultant dataD. Reinsurer dataE. OtherF. Not sure or not applicable
73%
3% 2%5% 3%
14%
82%
11%
5%2% 0% 0%
A B C D E F
2017 2012
What sources does your company use in setting the assumption for mortality improvement?
16
Polling question 13
A. General population mortalityB. Insured population mortalityC. Both general and insured
population mortalityD. SOA mortality improvement
studyE. All of the aboveF. OtherG. Not sure or not applicable
11%
14%
6%
13%
18%
6%
31%
24%
31%
0%
27%
0%
18%
0%
A B C D E F G
2017 2012
How often do you update/review your best estimate mortality assumptions?
17
Polling question 14
A. More often than annuallyB. AnnuallyC. Every 2 or 3 yearsD. Every 4 or more yearsE. OtherF. Not sure or not applicable
2%
68%
10%7%
0%
13%
7%
62%
19%
5%
0%
7%
A B C D E F
2017 2012
How often do you update/review your best estimate lapse assumptions?
18
Polling question 15
A. More often than annuallyB. AnnuallyC. Every 2 or 3 yearsD. Every 4 or more yearsE. OtherF. Not sure or not applicable
0%
69%
12%
3%1%
14%
A B C D E F
How often do you update/review your best estimate expense assumptions?
19
Polling question 16
A. More often than annuallyB. AnnuallyC. Every 2 or 3 yearsD. Every 4 or more yearsE. OtherF. Not sure or not applicable
2%
79%
6%
1% 0%
12%
A B C D E F
Paula Hodges121 – Enhanced Governance Frameworks for Assumption ManagementTuesday October 25, 2016 3:45 – 5:00 pm
Background / History
21
Assumption Evolution
22
PricingForecast
Asset Adequacy GAAP
Illustration Actuary
Capital Mgmt
Assumptions required for each areaDeveloped independently
Raised concerns
Consistency
Transparency
Replication of Results
Desired State
Assumption Development Timing
23
GAAP
CFT
Pricing
Mortality Studies
Lapse Studies
Premium /PHBehavior
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Assumption Management Challenges
24
Competition with other projects
Small staff devoted to development and validation
of assumptions
Pricing assumptions developed; Lag time before implementation for
inforce applications
Volatility •How much does the assumption change over time?
Credibility •How much company/industry data is available?
Sensitivity •Change in results vs. change in assumption
Considerations for frequency of review
Structure and Process
25
Effective Governance Framework
26
Policies & Procedures Training Resources
(Budget, Staff)
Audits &
AssessmentsSanctions
Assumption Approval Options
27
Asset Approval Team
Chief Pricing Actuary
Appointed Actuary
Chief Investment Officer Risk Officer
Liability Approval Team
Chief Pricing Actuary
Appointed Actuary
One path for asset assumption approval; One path for liability assumption approval.
Asset & Capital Markets vs. Liability
Assumption Approval Options
28
Pension Approval
Pension Chief Actuary
Appointed Actuary
Chief Investment Officer Risk Officer
Life Approval
Life Chief Actuary Appointed Actuary
Chief Investment Officer Risk Officer
Annuity Approval
Annuity Chief Actuary
Appointed Actuary
Chief Investment Officer Risk Officer
Health Approval
Health Chief Actuary
Appointed Actuary
Chief Investment Officer Risk Officer
One path for assumption approval – different approvers by LOB
by Lines of Business
Assumption Approval Options
29
Pricing Approval
Pricing Actuary*
Appointed Actuary
CFO Head of LOB
Inforce Approval
Appointed Actuary*
Pricing Actuary
CFO CRO
Pricing process includes LOB approval, Inforce approval includes CRO
* - Committee Chair
Pricing vs. Inforce
30
Assumption Governance Process
• Assumption team – varies – GAAP, CFT, Forecast, etc.
Area
Experience Study
Model Developers or Model Users
Assumption Development Team*
Model Review Team
Assumption Review Team
Identify a/e discrepancy Request
assumption review
Change in model
Assess Impacts
Peer review
Prep for MRT / ART
Review / Approve Change
Review / Approve Change
Document&
Implement
Document&
Implement
Identify a/e discrepancy
Scheduled review
Develop Assumption
Recommendation
Assumption Calendar
Activities
Communication & Documentation
31
Consistency• Create templates for approval recommendations• Official minutes at each meeting – have approved at next meeting• Always invite stakeholders – ensure transparency
• Experience team• Modelers
Governance• Understand the hierarchy within the company• Responsibilities under PBR
Documentation Process• Paper trail for the approval process – comments, simplifications, etc. • Document the process from data sources, experience study process, assumption approval, and
implementation• Assumption repository for approved assumptions
Challenges
32
Challenges
33
Assumption approval processDocumentation• Approval process: Look into GRC tools (Governance, Risk Management, Compliance)• Maintain recommendation documents, meeting notes, approver questions / responses, official approval
documentation
Small Co vs. Large CoCompany Size• Small companies – Approver may be intimately involved in assumption development – no “arms –length approval”• Large companies – Approver is too far removed – no subject matter expertise to challenge the recommendation
Assumption Inventory and ScheduleInventory / Schedule• Maintaining various dimensions of assumptions: best estimate, PAD, short-term view, long-term view• Frequency of updates, securing resources to meet that schedule• Timing of inforce implementation vs. pricing assumption
Implementation challengesModeling • Model implementation timing varies by application• Assumption simplifications – are these changes to the “assumptions”• Modeling methods, such as compression – are these “assumptions”?
Key Take-Aways:
• Strive toward continuous improvement• Automation will become increasingly necessary
• Make friends with your IT partners!
• Tooling and processes: • Documentation trail for the approvals• Repository of assumption values
• You can never over-communicate
34
Daniel Moulden121 – Enhanced Governance Frameworks for Assumption ManagementTuesday October 25, 2016 3:45 – 5:00 pm
Governance Process Infrastructures
36
• In addition to a robust governance framework, a solid infrastructure for model data, assumptions and other information provide advantages while supporting the governance process
• Using databases and custom applications is one method to build that infrastructure
37
First Step - Assess your situation: • Determine the models and assumptions within the scope.
• Required resources for programming, data storage, model conversion, etc. There is potentially a lot of work involved depending on where your organization is in the development of governance.
• Available tools for creating and incorporating adequate infrastructure to support your governance process.
38
• Application – a published application for all users that can be updated regularly and provide a common interface.
• Database - A centralized database preferably supported by IT and backed up daily. Set user security based on access needed (e.g. write access for model owners/users, read only for everyone else). (SQL Server used in the following examples).
39
Life Pricing
• Example: Current state of models - disconnected and not organized
40
Health Pricing
Re-Pricing Entire Portfolio for a new
distribution line
Revised Life Pricing
Updated Health Pricing
Special Pricing for that one product that got shelved
Valuation Reserves
Valuation Projections
Scenario Testing
Cash Flow TestingStrategic Planning
That one model we used for that one project – I
think Tina made it? Total Company Models
New Business Profitability
FAS 142
C3 Phase 1
That one model we used for that one project – I
think Bob made it?
Next Steps
• Create a Model Inventory
• Create a centralized database
• Create an application (start simple, refine)
• Designate authorized users (security and control)
• Begin building your infrastructure
41
• Helpful Tips• Use a robust and stable modeling tool (in house or third
party) for as many models as possible. May require some conversion.
• Create systematic approaches for handling those models and their assumptions (ad hoc and manual methods are not reliably consistent)
• Systematic and standard processes are easier to automate
42
• Model Inventory - Gather essential information for each model in the inventory
• Organize this data so it can be stored in a database
43
Model InfoNames
DescriptionsPurpose/Scope
CompanyOwner/Users
Time period(s)
Model Locations(Network Path)
• Assign model ID codes to each unique model• Create an outline of integrated models
44
Model ID Code Model InfoName
DescriptionsPurpose/Scope
CompanyOwner/Users
Time period(s)Model Location (Network Path)
• Add all of this to database tables with model ID code as a primary key (unique identifier) – then join them together
45
Model ID Code Model Name Company Purpose Scope OwnerTime Period
A_GAAP A_Valuation A Valuation and Projection All Jackie 2016 08
A_Stat_Life A_Valuation A Valuation Life Wilson 2016 Q2
A_Stat_Other A_Valuation A Valuation Other Kelly 2016 Q2
B_Pricing Pricing Co B B Pricing All Bill2014 -2016
B_NewBusinessProfit Profit Co B B Projection All Bill 2016
• Add a model ID code field to every table in the database so tables can be joined in queries
• Every aspect of all the models are now stored in the same place and easily distinguishable
46
Model ID Code
Projection Results
Assumption Mapping
Model Coding
• Code the application to move model coding, parameters and assumptions to and from database
47
Model Locations (Network Path)
Application Database
• Code the application to capture user activity and add records to log files for documentation
• Allow authorized users to add and modify model information, coding and assumptions
48
User Activity (load models, import assumptions from other models, etc.)
Application Activity Logs in Database
• Triggers - In the database, triggers on each table can control changes and capture user name and date/time changes were made
Update and Insert – Executes automatically when any field is inserted or updated (can be used to record who makes changes and when –also, to rollback changes by unauthorized users)
Delete – Executes automatically when any record is deleted (can be used to archive deleted records so data is not lost)
• This is invaluable information for documentation and auditing purposes
49
• Integrate existing processes/data to incorporate into any model.
50
Models
ApplicationDatabase
Study Results
Scenario Test Data
New Plan Specifications
• Ultimately, everything is connected
51
ModelsApplication
Database
Study Results
Scenario Data
New Plan Specifications
Model Output
Model Inventory
Model Inputs
Model History
Assumption Mapping
User Activity
• A living model inventory is a direct result
52
Model ID Code Model Name Company / LOB
Owner Last Updated
A_GAAP A_Valuation A / ALL Jackie 2016/08/31
A_Stat_Life A_Valuation A / Life Wilson 2016/06/30
A_Stat_Other A_Valuation A / Other Kelly 2016/06/30
B_Pricing Profit Co B B / Pricing Bill 2015/11/23
• Quickly isolate models that might need to be updated
53
Model ID Code Model Name Company / LOB
Owner Last Updated
A_GAAP A_Valuation A / ALL Jackie 2016/08/31
A_Stat_Life A_Valuation A / Life Wilson 2016/06/30
A_Stat_Other A_Valuation A / Other Kelly 2016/06/30
B_Pricing Profit Co B B / Pricing Bill 2015/11/23
• Compare assumptions across models for any captured level of detail
54
File Assumption Model ID Code Plan Code
SWA_Lapse_2016 Lapse A_GAAP SWA_737
SWA_Lapse_2016 Lapse A_Projection SWA_737
SPR_Lapse_2016 Lapse B_GAAP SPR_A324
SPR_Lapse_2016 Lapse B_Projection SPR_A324
SWA_Lapse_2016 Lapse Pricing A SWA_737
SWA_Lapse_2016 Lapse Pricing B SPR_727
• Quickly identify possible coding errors
55
File Assumption Model ID Code Plan Code
SWA_Lapse_2016 Lapse A_GAAP SWA_737
SWA_Lapse_2016 Lapse A_Projection SWA_737
SPR_Lapse_2016 Lapse B_GAAP SPR_A324
SPR_Lapse_2016 Lapse B_Projection SPR_A324
SWA_Lapse_2016 Lapse Pricing A SWA_737
SWA_Lapse_2016 Lapse Pricing B SPR_727
• Compare assumptions mapping consistency within a model across time periods
56
Plan code Assumption Model ID Code File
SWA_737 Lapse A_GAAP 201609 SWA_Lapse_2016
SWA_737 Lapse A_GAAP 201606 SWA_Lapse_2016
SWA_737 Lapse A_GAAP 201603 SWA_Lapse_2015
SWA_737 Lapse A_GAAP 201512 SWA_Lapse_2015
• Verify assumptions were updated in a model after being approved by the governance committee
57
Plan code Assumption Model ID Code File
SWA_737 Lapse A_GAAP 201609 SWA_Lapse_2016
SWA_737 Lapse A_GAAP 201606 SWA_Lapse_2016
SWA_737 Lapse A_GAAP 201603 SWA_Lapse_2015
SWA_737 Lapse A_GAAP 201512 SWA_Lapse_2015
• Load all models and assumptions to the centralized database
• Create snapshots of all model data (coding and assumptions) for change analysis and documentation.
• Quickly replicate results, change naming conventions, search for coding errors, compare models to one another
58
• Reduced time and effort finding assumptions and determining how plans are modeled.
• Allow direct importing of assumptions to/from any model – activity log records user, date-time, assumptions and models.
• Much, much more….
59
• Build a solid infrastructure to support your governance framework.
• Combine your model coding, assumptions, model output, documentation, etc. by taking a systematic approach to the design of the information flow and the necessary controls of your governance framework.
• Evolve the methods to suit your business needs. All changes can be universal or targeted as needed so that years down the road, you can use these same tools.
• Move toward a more unified platform within your organization.
60
Factors driving increased assumption governance
61
Assumptions Governance
Background
• The application of formal governance guidance and oversight to the development and use of assumptionsDefinition
• Scope • Policies and procedures• Roles and responsibilities • Controls and monitoring
Key elements
• Efficient, controlled development of appropriate assumptions that are consistent across the company and well documented
Key goals
62
Drivers causing companies to consider assumption governance• Regulatory developments
• PBR/VM-20 prudent estimate assumption requirements• PBR/VM-31 disclosure requirements• PBR/VM-G governance requirements• ORSA, Solvency II “three lines of defense” requirements
• Increased scrutiny• Many stakeholders have increasingly high expectations
regarding the frequency and thoroughness of reviews• Audit firms paying closer attention due to PCAOB practice
alert 11• Appropriate review and approval of assumptions by management• Supporting evidence for assumptions that require significant
judgement
63
Drivers causing companies to consider assumption governance (continued)• Business practice
• Most companies have some sort of limited assumption governance in place for DAC and other older regulations
• Positive experiences leading some companies to consider expanding use of assumption governance as part of good business practice
• For example, in new areas of modeling such as predictive analytics
• P/C industry frequently uses for pricing and underwriting, as well as marketing, customer retention, fraud detection
• Life insurance companies increasingly these advanced statistical algorithms to predict chosen target variables
• E.g. for setting mortality and policyholder behavior assumptions, and informing sales, underwriting and customer retention programs
64
Key components and leading practices
65
Key components of an assumption governance document• Assumptions in scope• Roles and responsibilities• Policies and procedures
• Selecting data sources• Conducting experience studies• Setting/updating assumptions• Review and approval• Documenting assumptions• Storing assumptions• Communicating assumptions
• Controls and Monitoring• Data controls• Process controls• Monitoring assumption use• Monitoring experience
• Resolution Procedures• Exception criteria• Escalation paths
• Appendices• Assumption classification scheme• Assumption inventory• Documentation templates
66
Characteristics of leading practices• Formalized assumption governance document
• Specify scope, roles, controls, resolution procedures, asset classifications, documentation requirements, etc.
• Processes for assumption updates, reviews, approvals, and storage• May have processes for experience studies and setting assumptions
• Buy in and mandate from senior executives• Centralized oversight from board or board committee• Assumption governance committee for review and approval
• Senior executives representing key stakeholder groups• Centralized database for storage and documentation
• Appropriate IT involvement for controls, automation, etc.• Feedback loop for monitoring experience and use of assumptions• Assumption review schedules that balance the need to stay
current with the effort to analyze and update assumptions• Standardized documentation requirements, with templates
67
Company size considerations for selecting data sources and other resources
Own Judgement Consultant Reinsurer Industry Own
Experience
68
Company Size
Small Medium Large
Cons
ider
atio
ns
• Little company experience- Heavy reliance on
external sources and judgement
• May outsource steps in the assumption setting process- Determination of best
source- Adjustment to fit specific
risk profile
• Moderate company experience- Supplement product
specific experience with related experience or external sources
• Use internal resources to set most assumptions- May still require significant
judgment
• Extensive company experience- Adequate credibility at
granular level- May still need to
supplement with external sources
• Some judgment still necessary- When experience does not
exist or is ambiguous
Assumption categorization examples• Need clear definition of what constitutes an assumption and
the various assumption categories important to your company• Functional area – product, corporate, ALM, modeling, planning, risk• Purpose – pricing, valuation, PBR), CFT, economic capital, EV• Accounting – GAAP, statutory, economic• Materiality – immaterial, moderately material, key risk driver• Product type – traditional life, UL, annuities• Granularity – very granular detail, high level trends• Structure – deterministic-static, deterministic-dynamic, stochastic• Margin – best estimate, padded, prudent estimate, stressed, random• Risk type – insurance, economic/market• Risk – mortality, lapse, expenses, interest rates, defaults, reinvestments• Update frequency – quarterly, annual, every three years• Source – prescribed, industry standard, reinsurer, company experience
69
Trend towards more centralized assumption management
Decentralized Centralized
Company Size • Smaller • Larger
Governance • Informal and less developed • Formal and more developed
Approval • Individuals • By committee
IT • Not that involved • Key role
Storage • Assumptions often stored in multiple spreadsheets or the models themselves
• Assumptions often stored in a single database or centralized location
Assumptions • Assumptions often developed and used in a single area of the company (more siloed)
• Better support for assumptions that are to be used consistently in multiple areas throughout the company
70
Documentation is a key component of assumption governance• Assumption review/approval documentation – formalized, audit
friendly, with decision recorded and templates if possible• Assumption characteristics and classification• Inputs – data sources, subject matter experts, etc.• Rationale (e.g., sensitivities) for best estimates, margins and final
assumption chosen, including justification for any divergent exceptions• Impact quantification (e.g., attribution analyses)• Clear formal signoff of final answer, even if no change recommended• Instructions for implementation, use, and monitoring
• Other relevant pieces of documentation• Official assumption governance document• Assumption inventory, with assumption update calendar and triggers• Experience study documentation• Assumption classification scheme
71
PBR implications
77
Assumption governance is key for PBR since it allows companies to use their own assumptions
78
• Lists requirements for developing and using assumptions• Guidance on prudent estimate assumptions which equal
anticipated experience plus a margin for uncertainty
VM-20Life PBR
• Contains documentation requirements for assumptions• Focus is on methods and rationale used
VM-31 PBR
Actuarial Report
• Clarifies PBR governance responsibilities, including assumptions• Stems from recognition that determination of prudent estimate
assumptions are the responsibility of the company
VM-G Governance
VM-20 requirements related to assumptions
• Prescribed, stochastic, and prudent estimate assumptions• Anticipated experience assumptions based on own
experience data to extent relevant and credible• Relevant data? (ASOP 23 on data quality)• Credible data? (ASOP 25 on credibility procedures)• Supplement with industry data (prescribed for mortality)• Consistent with company risk assessment practices
• Margins to reflect uncertainty• Due to adverse deviations, estimation error, modeling limitations• For each material risk factor not prescribed or stochastic• Must increase modeled reserve in proportion to uncertainty• Prescribed for mortality
79
VM-31 requirements related to assumptions
• No actuarial opinion component (unlike VM-30 AOM)• In line with leading assumption documentation practices• Overview summarizes methods used and decision rationale
• Method, and experience used vs. actuarial judgment• Rationale behind significant changes from prior year• Monitoring – experience used, update schedule, triggers• Inconsistencies with company risk management practices
• Life section adds additional details for each assumption• Mortality, PH behavior, expenses, assets, reinsurance, NGE, etc.• Impact of margins on DR, individually and in aggregate• Use of sensitivity tests • Limitations, approximations, simplifications
80
VM-G requirements related to assumptions• Oversee relevant resources, policies and procedures• Review senior management reports on infrastructure and risks, and
work with them to resolve control issuesBoard
• Ensure adequate resources and processes set to validate experience data, validate assumptions, control material risks
• Review assumptions (consistent with risk practices?)• Review and address material issues
Senior Management
• Oversee application of assumptions in the calculation of PBR• Verify compliant assumptions and related methods/controls• Report to board and management on processes and results• Prepare VM-31 PBR Actuarial Report• Disclose significant unresolved issues to regulators and auditors
Qualified Actuaries
81
82