11_Brehm Flashcards

1
Q

Enterprise Risk Management

A

Enterprise Risk Management is the process of systematically and comprehensively identifying critical risks, quantifying their impacts, and implementing integrated strategies to maximize enterprise value.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Key Aspects of ERM

A
  • An effective ERM program should be a regular process, not a one-time project
  • Should consider risks on an enterprise-wide basis
  • Focus on risks that represent a material impact to the value of the firm
  • Risk can be positive or negative; it’s the fact that actual outcomes vary from expected
  • Risks must be quantified where possible, including correlations among risks
  • Create strategies to avoid, mitigate, or exploit risk factors
  • Evaluate risk management strategies for risk/return to maximize firm value
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

Types of Insurance Company Risk Factors

A

Insurance Hazard Risk - Risk assumed by insurer
* Underwriting (non-cat) – in force
* Accumulation/cat-risk – in force
* Reserve Deterioration <— from past exposures

Financial Risk - Risk to asset portfolio due to volatility in interest rates, foreign exchange rates, equity prices, credit quality, liquidity

Operational Risks - Risks in the operation/execution of the company; the actions the company takes

Strategic Risks – The risks of strategic choices the company takes –> The risk of choosing the wrong plan

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

Enterprise Risk Management Process
(different phases of risk management process)

A

Diagnose - High-level risk assessment of risk factors that pose a potentially serious threat to the firm
-> General environment risks
-> Industry risks
-> Firm-specific risks

Analyze - Quantify risks with probability distributions of potential outcomes. Include correlations.

Implement Risk Management (forms of traditional risk management) - Avoidance, reduction, mitigation, elimination/transfer, or retain/assume risks

Monitor - Monitor results vs. expectation, update plans

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Enterprise Risk Modeling
Helps with the Following Strategic Decisions

A
  • Determining capital needed to support risk or maintain rating
  • Identifying sources of significant risk
  • Deciding on reinsurance strategies
  • Planning growth
  • Managing asset mix
  • Valuing companies for M&A
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

Most Important Elements for Model Quality

A
  • Model reflects relative importance of different risks to business decisions
  • Modelers have a deep knowledge of the risk fundamentals
  • Model incorporates the dependencies between different risks
  • Modelers have a trusted relationship with senior management
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

“Essential Elements” of an Enterprise Risk Model

A
  • Underwriting risk
  • Reserving risk
  • Asset risk
  • Dependencies (correlation)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Underwriting Risk

A

Loss Frequency and Severity Distributions
Used to quantify loss potential

Pricing Risk
Underwriting cycle, risk of unnoticed underpricing until losses accumulate, resulting in a reserve deficiency

Parameter Risk
Risks from mis-estimated parameters, imperfect model form, unmodeled risks
(Estimation, projection event, systemic risk)

Catastrophe Modeling Uncertainty
Uncertainty in 3rd party CAT models (e.g. probability of events/loss)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Parameter Risk

A

Estimation Risk
Risk that the form and parameters of the frequency/severity distributions don’t reflect the “true” form and parameters

Projection Risk
* Changes over time and uncertainty in the projection of changes
* Trends in frequency/severity from time of data to current/future periods
* Development of losses to ultimate

Event Risk – Events outside company control that impact frequency/severity trends (e.g. class action suits, asbestos, new cause of loss)

Systematic Risk – Nondiversifying, impacting many policies (e.g. inflation)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Reserving Risk

A
  • Risk that reserves develop differently than expected
  • Reserve uncertainty impacts the amount of required capital and time that capital must be held
  • A model can understate both a reserve estimate and reserve uncertainty
    -> We need a model of reserve uncertainty in an enterprise risk model
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Asset Risk and its key aspect

A
  • Model asset risk by generating probabilistic scenarios based on historical patterns and testing the insurer’s strategy against the scenarios, taking into account the scenario likelihood
  • Model should account for:
    o Bonds
    o Equities
    o Foreign exchange/interest rates

Asset risk refers to variability in asset values. Important asset classes to model include bonds, equities, real estate and exchange rates. A key aspect of asset modeling is modeling scenarios consistent with historical patterns.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Sources of Dependencies

A
  • Simultaneous impact of macroeconomic conditions on multiple risks
    o E.g. Inflation impacts both underwriting losses and loss reserve development
  • Dependencies that impact multiple lines of business
    o E.g. UW cycles, loss trends, reserve developments, catastrophes (event risk)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

Modeling Dependency

A
  • Use copulas to incorporate dependency if there’s higher correlation in the tail
  • Correlation through a multivariate normal distribution has low tail dependency
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

Why Default Avoidance isn’t the Most Important Reference Point to Set Capital

A
  • Default avoidance is about protecting current policyholders
  • To protect shareholders, should avoid significant partial losses of capital that could damage franchise value
    -> which would impact customer base, agency relationships, reputation
  • Loss that would cause a significant ratings downgrade is a more meaningful reference point
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Meaningful Reference Points for Setting Capital (Besides Default)

A
  • Maintaining enough capital to avoid ratings downgrade below a certain level
  • Maintaining enough capital to service renewal business
    -> If writing renewal business requires 80% of capital, an appropriate reference point is a 20% loss of capital
  • Maintaining enough capital so insurer thrives after a catastrophe (not just survives)
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Challenge When Using Extreme Reference Points to Set Capital

A
  • Model is least reliable in the extreme tail (e.g. exhausting all capital)
    -> Little data far in the tail
    -> Results are sensitive to assumptions about the distribution
    -> Far tail is poorly understood
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Deterministic Project Analysis

A
  • Uses a single deterministic forecast to estimate present value or IRR
  • Uncertainty is handled judgmentally by decision makers
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

Risk Analysis
And why it is better than deterministic project analysis

A

DFA
* Uses forecasted distributions of the critical inputs in a Monte Carlo simulation to calculate a distribution of present value.
* The risk judgment is intuitively applied by decision makers.

Better because it directly incorporate the uncertainty of the critical variables in the model

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

Certainty Equivalent

A

Certainty Equivalent
Similar to risk analysis, but quantifies the risk judgment with a corporate risk preference or utility function for consistency.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

Internal Risk Model:
Corporate Risk Tolerance

A

Helps answer the question: How much risk (standard deviation) are we willing to tolerate?
Corporate risk tolerance depends on:
* Ability and willingness to tolerate volatility
* Company size
* Financial resources
These considerations determine the impact a loss has on the company.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

Internal Risk Model:
Cost of Capital Allocated

A

Cost of risk capital is allocated to the individual risk sources (e.g. Line of business).
RORAC = Risk-AdjustedCapital ⋅ HurdleRate

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

Internal Risk Model:
Cost-Benefit Analysis for Risk Mitigation

A

Any risk-mitigation effort with a positive EVA is worth doing. This means the reduction of capital cost is greater than the cost of risk-mitigation.

EVA = NPV Return − Cost of Capital
Cost of Capital = RiskCapital ⋅HurdleRate

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
23
Q

How is economic capital measured and what’s the advantages of Economic Capital

A

Measured with VaR at a remote probability level.

  • It’s a unifying measure for all risks across the enterprise
  • More meaningful to management than risk-based capital or capital adequacy ratios
  • Forces a firm to quantify and aggregate risks into a probability distribution
  • Provides a framework for setting risk levels for the organization and individual business units

所以model challenge是ERM model aren’t reliable at such remote probabilities because of the approximation, assumptions and lack of data in the tail

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
24
Q

Moment-Based Measures:
Advantages and Disadvantages

A

Advantages
Reflects all losses in the distribution

Disadvantages
Standard Deviation treats favorable distributions the same as unfavorable ones

Exponential moment -> responds more to larger losses

Semi-standard deviation -> only uses unfavorable deviations (but this won’t highlight the tail risk enough, better for LR or pricing adequacy)

同理,VaR and TVaR wouldn’t be suitable for LR or pricing adequacy because they are more focusing on tail events
For lines not as extreme as CAT, like excess casualty, can use semi-SD

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
25
V@R
V@R Percentile of the probability distribution Pro - Emphasizes large losses Con - Only looks at one point in the distribution
26
TV@R
TV@Rp% = E[Loss | Loss > V@Rp%] Expected loss given a loss is equal or greater than a specified probability level Pro - Reflects losses that exceed V@R Con - Losses are reflected linearly in the tail
27
XTV@R
XTV@Rp% = TV@Rp% −Mean The mean loss might be funded by other means (such as premium). Capital is needed for losses above the mean.
28
Expected Policyholder Deficit (EPD)
EPD = (TV@Rp%-V@Rp% )⋅(1− p%) For EPD, the probability level is set so that capital is V@R at that level
29
Value of Default Option
The market value of the remaining default risk after reinsurance. This would be estimated based on options pricing.
30
Probability Transforms
* Shifts the probability distribution towards larger losses, then computes risk measure Examples: * Expected loss with transformed probabilities * Minimum martingale transform and minimum entropy martingale transform, Wang transform * Weighted risk measures: WV@R, WTV@R, WXTV@R [TVaR is linear in the tail (a loss twice as large is considered twice as bad). Under transformed probabilities, TVaR becomes WTVaR, this is not linear in the tail and consider a loss that is twice as large to be more than twice as bad]
31
Generalized Moments
Expected values that aren’t simply powers of the variable. Example: Blurred V@R, which adds weight to losses around the percentile (not just looking at the loss percentile)
32
Customer Reaction to Capital Levels
Some insurance customers are concerned about capital levels, reflecting the “quality of the insurance guarantee.” * Rating increases can slowly increase growth * Rating decreases can result in a rapid decline in business (Rating is agency rating)
33
Purpose of Capital Allocation
Capital allocation shows the contribution of each business unit to the overall risk. Capital allocation can be used for calculating risk-adjusted profitability and setting capacity controls by line of business.
34
Proportional Capital Allocation
Proportional Capital Allocation Allocates total risk down to business units. First calculate the risk measure for each business unit. Then allocate the overall risk measure proportionally.
35
Allocating Capital: Risk Decomposition
Risk Decomposition Decomposition uses co-measures to calculate the contribution of each business unit to the overall risk measure. -> e.g. Co-TV@R It is desirable for a risk decomposition method to be marginal All marginal decompositions are also co-measures. Common risk measures that can be expressed as marginal decompositions are TVaR and standard deviation
36
Allocating Capital: Marginal Allocation
Marginal Allocation Marginal Allocation measures the change to the total risk measure for the company if there is a small change in a business unit’s volume. * The change to the total risk measure is assigned to the business unit.
37
Marginal Allocation Advantages
Advantages * Marginal allocation also produces co-measures * Marginal attributions sum to the total risk measure * Leads to consistent strategic implications o Ex. - Growing a business with an above-average profit-to-risk ratio increases the company’s profit-to-risk ratio
38
Risk-Adjusted Profitability
If the risk measure is proportional to the market value of risk, then a higher risk-adjusted profitability ratio means a business unit is more profitable relative to its risk. risk-adjusted profitability ratio = Profit_j/Risk Measure_j
39
Allocating the Cost of Capital
Allocating the cost of capital to business units sets a minimum profit target for each unit. * If a business unit’s profit exceeds the minimum profit target, this excess is value added to the firm. The cost of capital reflects the risk that a business unit has the right to access the insurer’s entire capital.
40
Disadvantages of Allocating Capital Compared to Allocating the Cost of Capital
Disadvantages of allocating capital * It’s arbitrary because different risk measures allocate capital differently. * It’s artificial because each business unit has access to the entire company’s capital.
41
Regulatory and Rating agency capital Adequacy Models: Leverage Ratios
* Leverage ratios generally compare a ratio to surplus (e.g. Net Written Premium-to-Surplus) to a threshold for testing capital adequacy. -> 12 IRIS ratios [4+ ratios does not fall within their corresponding reasonable ranges are considered to be at risk and warrant regulatory scrutiny Advantages * Easy to calculate and monitor Disadvantages * Doesn’t distinguish between lines of business * Ignores risks other than underwriting risks
42
Regulatory and Rating agency capital Adequacy Models: Risk-Based Capital Models
Risk-based capital models combine multiple aspects of risk into a single number. Risk aspects included * Invested Asset Risk * Credit Risk * Premium Risk * Reserve Risk * Accumulation/CAT Risk (not in US or S&P models) * Covariance Adjustment (in some models like US and Best)
43
Reasoning for Different Factors Between Different Risk-Based Capital Models
Different Model Uses * Rating agency models (AM Best, S&P) focus on long-term viability * Regulatory models (like US RBC) focus on 1-year solvency, so these models will have relatively lower factors Covariance Adjustment * Models with a covariance adjustment (like AM Best) will have relatively higher factors than models without it (like S&P) o The covariance adjustment will reflect independence between risks so that the overall capital is less than the sum of individual risks
44
Regulatory and Rating agency capital Adequacy Models: Scenario Testing
An insurer may do its own risk assessment using scenario testing or stochastic modeling, which would be reviewed by a regulator. Scenario Testing Requires * A 1-5 year financial projection model * Probability distributions for sources of uncertainty * Correlations between risks * Management responses to adverse financial results
45
Asset-Liability Management
The analysis and management of the asset portfolio, reflecting current liabilities, future cash flows, future premium flows, and the existing asset and liability portfolios. The goal of ALM is to help the insurer make better risk-return decisions.
46
Asset-Liability Matching
Asset-Liability Matching Setting an investment portfolio to have the same duration as the duration of the liability portfolio to protect the firm against changing interest rates.
47
Additional Risks and Actions that Asset-Liability Management Considers
Beyond interest rate risk, ALM considers: * Inflation risk * Credit risk * Market risk * Equities and reinsurance as methods for hedging
48
Layers of Complexity in Asset-Liability Management
* Analysis of the asset portfolio in isolation (risk vs. return) * Adding fixed liabilities to the analysis o Reinvestment risk if asset duration is shorter o Risk of selling depressed assets if asset duration is longer * Adding variability to the amount/timing of liability cash flows * Adding variable underwriting cash flows * Adding tax considerations and equity investments -> With real-world complexity, a true enterprise-risk model is needed. all above are increasing complexity
49
How does a company’s choice of risk-return metrics impact the optimal investment strategy?
Statutory Accounting Metrics Since bonds are amortized and liabilities aren’t discounted, this approach shows little hedging from duration-matching. GAAP Accounting Metrics Bonds are valued at market value, but this also shows little hedging from duration-matching. True Economics Metrics Duration-matching lowers interest rate risk, but including cash flows complicates the analysis.
50
Asset-Liability Modeling Steps
1) Model assets, existing liabilities and current business operation 2) Define risk metrics - income or balance sheet-based, accounting basis 3) Define return metrics - income or balance sheet-based 4) Set the analysis time horizon - single or multi-year 5) Include model constraints – e.g. regulatory constraints 6) Run the model with different investment, underwriting, and reinsurance strategies, calculating risk-return metrics 7) Plot an efficient frontier based on the different portfolios 8) Test the effects of different reinsurance structures 9) Review simulations where portfolios performed poorly - hedging strategies or new policies may reduce downside risks
51
Naïve Approach to Measuring Reinsurance Value
Comparing ceded premium (cost of reinsurance) to reinsurance recoveries and ceding commissions (benefit) over many years typically shows a negative net benefit. Reinsurers expect to make a profit, so a simple cost-benefit analysis is a poor way to assess reinsurance value.
52
Measuring Reinsurance Value: Paradigm 1
Reinsurance Provides Stability * Protects surplus from adverse results * Improves predictability of earnings and growth * Improves customer confidence that they’ll recover insured losses Ceded Premium - Recoveries is a better cost measure under this paradigm. (Cost of this stability) But still requires significant judgment to evaluate the stability benefit
53
Measuring Reinsurance Value: Paradigm 2
Reinsurance is a Substitute for Risk Capital * Increased stability lowers the required risk capital ROE cost of reinsurance = ReinsuranceCost/Capital Freed If the ROE cost of reinsurance is less than the company’s target return, getting reinsurance is a good deal.
54
Measuring Reinsurance Value: Paradigm 3
Reinsurance Adds Value Ideally, we could measure the value of reinsurance by the incremental increase in market value to the company. There’s no clear way to measure the marginal impact of reinsurance on firm value
55
Disadvantage of the Quantifying Stability paradigm for Measuring Reinsurance Value
Disadvantage Significant judgment is needed to evaluate the benefit of stability against the net cost of reinsurance for different programs.
56
Reviewing Probability Distributions of Financial Measures for Reinsurance Options
Look for: * Which option protects best against the worst losses * Options that don’t limit profitable scenarios too much * Compare which options produce better results in different areas of the probability distribution
57
Box/Space Needle View for Comparing Reinsurance Options
Shows the probability in different ranges. Compare programs based on: * Which program protects from the most unfavorable scenarios * Which sacrifices profitable good years
58
Cost-Benefit Diagram for Reviewing Reinsurance Options
X-axis Cost of reinsurance (Ceded Premium - Expected Recovery) Y-axis Loss amount (Net Premium - Net Loss) * Plot multiple probability levels (e.g. 1-in-10, 1-in-100, 1-in-250) Look for: * Which options perform best at different probability levels * If a program is more costly but has worse loss outcomes at each probability level, it’s inefficient and shouldn’t be considered.
59
Efficient Frontier for Reviewing Reinsurance Options
Efficient Frontier Graphs the risk-return of different reinsurance programs at various probability levels. * Programs that are below the efficient frontier are inefficient.
60
Describe two classes of models for measuring the change in capital due to reinsurance.
Theoretical Models Required capital is calculated with risk measures using an enterprise risk model. * Advantage -> Required capital is more consistent with management’s views about risk Practical Models Required capital is calculated using rating agency or regulatory models (e.g. 175% of BCAR) * Advantage -> Easier to implement * Disadvantage -> Uses proxies for risk (e.g. premium) to measure capital instead of modeling risks directly
61
Measuring the Marginal ROE Cost for Buying More Reinsurance
* The cost of buying more reinsurance is NPV(Ceded Premium – Ceded Loss) compared to the current program (negative) * This should release capital (negative) Marginal ROE(cost) = delta(Ceded Prem - Ceded Loss)/delta(Capital) If the marginal ROE (cost) is less than the cost of capital, buying reinsurance is a good deal.
62
Measuring the Marginal ROE Benefit for Buying Less Reinsurance
* The benefit of buying less reinsurance is NPV(Ceded Premium - Ceded Loss) compared to the current program (positive) * This should consume capital (positive) Marginal ROE(benefit) = delta(Ceded Prem - Ceded Loss)/delta(Capital) If the marginal ROE (benefit) is greater than the cost of capital, buying less reinsurance is a good deal.
63
Why Accumulated Risk Must be Considered When Evaluating the Value of Reinsurance for Long-Tail Lines
* Marginal ROEs for measuring reinsurance value are based on capital for a single year. * Long-tail lines have loss reserves and accumulated risk that must be supported by capital for multiple years.
64
As-If Loss Reserves
As-If Loss Reserves Loss reserves that would exist at the beginning of a new accident year if that business had been written in a steady state for all prior years. We can approximate the present value of required capital for an accident year over time by modeling the capital needed for both: * The current year for the current accident year * The as-if loss reserves
65
Advantages of Using As-If Loss Reserves
Advantages * This approach can measure the impact that correlated risk factors have on accumulated risk * It can fully measure the impact of reinsurance by applying reinsurance to the accident year and as-if reserves
66
Considerations When Implementing an ERM: Staffing and Scope
Staffing/Scope * Organization Chart (modeling team reporting, …) * Functions Represented (actuarial, finance, …) * Resource Commitment (mix of skill set, full vs. part-time, …) * Critical Roles and Responsibilities (control of inputs/outputs, …) * Purpose * Scope
67
Staffing and Scope Recommendations
Staffing/Scope Recommendations * Reporting Relationship - Team leader should have a reputation of fairness and balance * Resource Commitment - The team should have a full-time commitment to the implementation * Inputs and Outputs - Should be controlled similarly to the general ledger * Initial Scope - Prospective underwriting period, variation around plan
68
Considerations When Implementing an ERM: Parameter Development
Parameter Development * Modeling Software (software capabilities, integration with systems, …) * Developing Input Parameters (data driven, requires expert opinion, …) * Correlations (corporate-level ownership of correlations) * Validation and Testing
69
Parameter Development Recommendations
Parameter Development Recommendations * Modeling Software - Capabilities of the modeling team should determine how much is pre-built and how much the team builds * Parameter Development - Include expert opinion from underwriting, claims, planning and actuarial * Correlation - Modeling team recommends assumptions, which are owned at the corporate level (CRO/CEO/CUO) * Validation - Validate and test over an extended period
70
Considerations When Implementing an ERM: Implementation
Implementation * Priority Setting (priority/timeline driven from the top, …) * Interest and Impact (visibility and interest, communication plans, …) * Pilot Test * Education Process
71
Implementation Recommendations
Implementation Recommendations * Priority setting - Top management should set the priority for implementation * Communications - Regular communication to broad audiences * Pilot Testing - Do pilot testing to prepare stakeholders for the magnitude of the change * Education - Bring leadership to a base level of understanding about the model
72
Considerations When Implementing an ERM: Integration and Maintenance
Integration and Maintenance * Cycle (integrating into corporate calendar for decision-making, …) * Updating (frequency/magnitude of updates, …) * Controls (control of inputs/outputs, control of analytical templates, …)
73
Integration and Maintenance Recommendations
Integration and Maintenance Recommendations * Cycle - Integrate into the corporate calendar (at least for planning) * Updating - Major updates to inputs no more frequently than semiannually * Controls - Maintain centralized control of inputs, outputs and templates
74
Impact of Parameter Risk on Small vs. Large Companies
Small Company A small insurer already has significant uncertainty, so the added impact of parameter risk isn’t too large. Large Company Without parameter risk, the loss ratio modeled for a large insurer is unrealistically stable. Parameter risk isn’t diversified away with more insureds, so it significantly increases uncertainty for a large insurer.
75
Three Aspects of Parameter Risk
Projection Risk The risk in projecting past trends into the future. Estimation Risk The risk that parameter estimates used are not the “true” parameter estimates for the underlying process. Model Risk Uncertainty about which fitted distribution form is the “correct” distribution (e.g. Gamma vs. Weibull).
76
Projection Risk Models: Simple Trend Model
A trend line is fit to historical loss costs to project future severities. Disadvantages * Loss cost data is based on historical claims that haven’t settled, which adds uncertainty. * Assumes a single constant trend for historical data that will continue into the future.
77
Projection Risk Models: Severity Trend and Inflation Model
Models the trend as the sum of general inflation and “superimposed inflation.” Advantage * An ERM model can reflect the dependency between trend and inflation.
78
Projection Risk Models: Time Series Model
Advantages * Reflects more uncertainty than a simple trend model * Models trend as a time series as opposed to assuming a single, constant trend Disadvantages * A substantial number of data points is needed * If the data is limited, the model will understate the uncertainty
79
How to Include Parameter Estimation Risk in an Enterprise Risk Model
1) Parameter estimates are calculated using the MLE 2) Fit a joint lognormal (or normal) distribution to the covariance matrix from the MLE 3) In the ERM model, each simulation draws a random sample from the joint lognormal distribution to be the parameters for the loss distribution -> This approach adds variability to the parameters used in the model.
80
How to Include Model Risk (Part of Parameter Risk) in an Enterprise Risk Model
1) Assign probabilities of being “correct” to each of the better-fitting distributions. 2) For each simulation: a) Select a distribution from the set of distributions. b) Select the parameters from the lognormal distribution of parameters (to add estimation risk). c) Use this fixed distribution for all losses in the simulated scenario.
81
Why Incorporating Dependency in an Enterprise Risk Model is Important
Each individual model may be realistic, but if dependencies between different lines and risks are unrealistic, then the ERM model in aggregate will be unrealistic.
82
The Difference Between Correlation and Dependency and Why Dependencies May be Better Modeled with a Copula
Correlation uses a single value (the correlation coefficient) and doesn’t differentiate between different levels of dependency in different parts of the distribution. Dependency between different risks and lines may be higher in the tail than the rest of the distribution. -> This is better modeled with a copula.
83
How do copulas reflect dependency?
Copulas are a way to aggregate different distributions into a combined multivariate distribution. Different copulas reflect different levels of dependency at different parts of the distribution.
84
Frank Copula
* Light-tailed copula * R(1) = 0
85
Gumbel Copula
* Heavier tail than Frank * Asymmetric with more weight in the right tail *R(1) > 0
86
Heavy Right Tail (HRT) Copula
* Low correlation in left tail, heavy right tail * R(1) > 0
87
Normal Copula
* Heavier tail than the Frank copula * Lighter tail than Gumbel or HRT * R(1) = 0 * Can be used for more than two variables
88
t-Copula
* Multivariate copula, can be used for more than two variables * Has an additional parameter for the heaviness of the tail
89
Describing Copulas: Left and Right Tail Concentration Functions
The right tail concentration function, R(z), shows the probability that losses from one distribution will be large given that losses from another distribution are large. R(z) = Pr(U > z|V > z) L(z) = Pr(U < z|V < z)
90
Upper Tail Dependence Coefficient
Upper tail dependence coefficient The limit of R(z) as z -> 1 If this limit is greater than zero, there is high tail dependency. There still can be significant dependence near the limits, so it’s useful to look at R(z) when z is near 1.
91
Describing Copulas: Joint Distributions
The joint probability distribution of two distributions combined with a copula shows areas with high joint probability densities. This can be used to compare how different copulas behave, especially in the tail.
92
Fitting Copulas to Data
Left and Right-Tail concentration functions Look near the limit of R(z) as z -> 1 J Function Compare the fitted J to the empirical J function to evaluate which copula fits the data best. Chi Function Compare the fitted Chi function to the empirical Chi function.
93
Operational Risk
Operational Risk The risk of loss during the execution of the company due to: Inadequate or failed internal processes, people and systems External events Includes legal risk, but excludes strategic and reputational risk
94
List of Types of Operational Risk
Types of Operational Risk * Internal Fraud * External Fraud * Employment Practices and Workplace Safety * Clients, Products and Business Practices * Damage to Physical Assets * Business Disruption and System Failures * Execution, Delivery and Process Management
95
Types of Operational Risk: Internal Fraud
Internal Fraud Acts involving people within the company intended to defraud, misappropriate property or avoid regulations, the law, or company policy (excluding diversity/discrimination). Examples include: * Employee theft * Claim falsification
96
Types of Operational Risk: External Fraud
External Fraud Acts by a third party intended to defraud, misappropriate property or avoid the law. Examples include: * Claims fraud * Falsifying application information * Damage from computer hacking
97
Types of Operational Risk: Employment Practices and Workplace Safety
Employment Practices and Workplace Safety Acts inconsistent with employment, health, or safety laws/agreements, or acts resulting in the payment of personal injury claims, or claims relating to diversity/discrimination. Examples include: * Workers Compensation claims * Violation of employee health/safety rules * Organized labor activities * Discrimination claims * General liability (e.g. customer slip and fall at a branch office)
98
Types of Operational Risk: Clients, Products and Business Practices
Clients, Products and Business Practices Unintentional or negligent failure to meet a professional obligation to specific clients or from the nature or design of a product. Examples include: * Client privacy * Bad faith claims * Red-lining
99
Types of Operational Risk: Damage to Physical Assets
Damage to Physical Assets Loss or damage to the company's physical assets from natural disaster or other events (such as terrorism/vandalism). This excludes policyholder losses (which is insurance hazard risk). Examples include: * Damage to own office * Damage to own auto fleet
100
Types of Operational Risk: Business Disruption and System Failures
Business Disruption and System Failures Disruption of business or system failures. Examples include: * Processing center downtime * Hardware/software failures * Telecommunication problems * Utility outages
101
Types of Operational Risk: Execution, Delivery and Process Management
Execution, Delivery and Process Management Failed transaction processing or process management, and relations with trade counterparties and vendors. Examples include: * Policy processing * Claim payment errors * Data entry errors
102
Issues with a Bridging Model
Bridging Model * Ultimate loss ratios from mature years are trended forward to set the new plan loss ratio. * Ultimate loss ratios for immature prior years are calculated with the BF method using ELRs set to the initial plan loss ratio for each prior year. Problem with a Bridging Model Prior Year ultimate loss ratios become highly correlated. If the older prior year loss ratios were too optimistic, the plan loss ratios would be set too low. This causes many of the prior year ultimate loss ratios to be too low. -> Eventually, older years deteriorate, causing many AYs to deteriorate.
103
Possible Explanations if an Insurer Failed Due to Reserve Deterioration from a Bridging Model
* The model could not have accurately forecasted the loss ratio (or reserves). o Operational Risk: Process and system failure (or inherent uncertainty and no operational risk if all competitors were wrong) * The model could have accurately forecasted the loss ratio (or reserves), but was improperly used. o Operational Risk: People failure * The model did accurately forecast the loss ratio, but the indications were unpopular and ignored. o Operational Risk: Process and governance failure
104
Operational Risk and Underwriting Cycle Management
Underwriting Cycle Management The management of underwriting capacity as market pricing fluctuates due to the underwriting cycle. Assess different cycle management strategies based on: * Stability * Availability * Reliability * Affordability
105
Naïve Underwriting Cycle Management and its Possible Results
Naïve Cycle Management Writing business at inadequate prices during a soft market in order to maintain market share. Possible Results: * Insurer may be downgraded, causing policyholders to switch insurers o Failure of stability/availability * Insurer may go insolvent with partial recoveries on claims for policyholders o Failure of reliability/affordability
106
Effective Underwriting Cycle Management
Effective Cycle Management Disciplined Underwriting: * Soft Market - Decrease Written Premium volume (don’t write inadequately priced business) * Hard Market - Increase Written Premium volume To implement, must overhaul underwriting decision processes, including: * Planning * Underwriting * Objective Setting * Incentive bonuses
107
How an Insurer Can Implement Effective Underwriting Cycle Management
Intellectual Property * Retain and continue to develop top talent through soft markets * Maintain a presence in the core marketing channels * Maintain consistent investment in systems, models and databases Underwriter Incentives * Incentive plans should be flexible so underwriting decisions are aligned with corporate objectives, which change with market conditions. Example: If prices drop too low, underwriters may need to stop writing new business, but their bonuses & employment shouldn’t be at risk. Market Overreaction * When prices fall too low in a soft market, the insurer should maintain underwriting discipline and write less. * When the market hardens and prices overcorrect, the insurer will have the capacity to write significantly more business profitably. Owner Education Some financial figures may diverge from those of peer companies, such as: * Premium Volume - Will drop during soft markets. * Overhead expense ratio - Maintaining intellectual property during a premium decline will cause the expense ratio to rise.
108
Goals of Agency Theory
Agency theory is important because there are potential operational risks when management and owner interests are misaligned. Goals of Agency Theory * How to align management and owner interests * Understanding the impacts when management and owner interests are different
109
Risks Due to Different Management Incentive Compensation Plans
Incentive - Percentage of the increase in market cap after n years Result: Management may take too much risk to increase firm value because: * High upside benefit (big payoff if there’s a major increase in firm value) * Low downside risk (“gambling with someone else’s money”) Incentive - Stock grants or stock options Result: Management may become too risk-averse because: * Shareholders may have diversified portfolios * Management has a large portion of their wealth tied to the company
110
Examples of How to Manage Different Types of Operational Risks
* Pension Funding Issues - Quantify this risk with models that include financial risk and firm demographics * IT Failure Risk - Monitor and control with contingency planning. The remaining risk could be quantified and funded * Other HR Risks - Identification and control of these risks is most important * Reputational Risk - Identification and control of these risks is most important * Lawsuits - Monitoring of this risk is most important
111
Control Self Assessment
Control Self Assessment A process to examine and assess the effectiveness of internal controls with a goal of providing reasonable assurance that all business objectives will be met. Primary objectives of internal controls are to ensure: - reliability and integrity of information - compliance with policies, plans, procedures, laws, regulations and contracts - safeguarding of assets - the economical and efficient use of resources - accomplishment of established objectives and goals for operations or programs
112
Primary Objectives of Internal Controls
Primary objectives of internal controls * Reliability and integrity of information * Compliance with policies, plans, procedures, laws, regulations, and contracts * Safeguarding of assets * Economical and efficient use of resources * Accomplishment of established objectives and goals for operations or programs
113
Key Risk Indicators
Key Risk Indicators Measures used to monitor the activity and status of controls in a business area for a given operational risk category. KRIs are leading indicators of risk and can be measured frequently to flag attention if a threshold is hit.
114
Examples of Key Risk Indicators
Examples of Insurer Key Risk Indicators: * Production - Hit ratios, retention ratios, pricing levels, rate per unit of exposure * Internal Controls - Audit results, audit frequency * Staffing - Employee turnover, training budget, premium per employee, policies per employee * Claims - Frequency, severity, new classes of loss
115
Six Sigma
Six Sigma Six Sigma can help firms identify and eliminate process issues such as inefficiencies, errors, overlaps and gaps in communication. Examples of insurer processes to improve with six sigma: * Underwriting - Exposure data verification/capture, classification/hazard selection * Claims - Coverage verification, ALAE, use of outside counsel to set initial case reserves * Reinsurance - Treaty claim reporting, coverage verification, reinsurance recoverables
116
Operational Risk Modeling
Operational Risk Modeling Steps 1) Identify exposure base (e.g. payroll, head count, policy count, premium volume) 2) Measure the exposure level 3) Estimate the loss potential (frequency and severity) per unit of exposure 4) Create loss frequency and severity distributions by combining #2 and #3 5) Estimate the impact of risk mitigation, process improvement or risk transfer on the frequency/severity distributions
117
Strategic Risk
Strategic Risk Strategic Risk is the risk to the company from making the right or wrong strategic decisions, not making a decision or not recognizing that a strategic decision needs to be made. It’s the risk of choosing the wrong plan given market conditions.
118
Types of Strategic Risk
Industry (Very high) - Capital intensiveness, overcapacity, commoditization, deregulation, cycle volatility eg: Underwriting Cycle, Insurance as a commodity Project (High) - Failure of R&D, IT, business development or M&A eg: Value-destroying M&A, underinvesting in R&D and IT Stagnation ( High) - Flat or declining volume, price decline, weak pipeline eg: Response to changes in the underwriting cycle Brand (Moderate) - Erosion or collapse eg: Reputation loss through bad press or class action lawsuits Competitor (Moderate) - Global rivals, gainers, unique competitors eg: Predatory pricing from competitors, Entrance into new markets with inadequate expertise or systems Customer (Moderate) - Priority shift, power, concentration eg: This is an issue with large commercial insurance business Technology (Low)- Technology shift, patents, obsolescence eg: Data Management, Innovations in distribution over the internet
119
Steps for Scenario Planning to Manage Strategic Risk
1) Define the scope of the analysis - Time frame, business segments, ... 2) Identify the major stakeholders - Customers, competitors, employees, shareholders 3) Identify basic trends 4) Identify key uncertainties 5) Construct initial scenario themes - Combine key elements to create scenarios 6) Check for consistency and plausibility - Scenarios should be realistic and internally consistent 7) Develop learning scenarios - Identify themes that are strategically relevant (e.g. “soft market” scenario) 8) Identify research needs - Better understand the trends and uncertainties 9) Develop quantitative models 10) Evolve toward decisions scenarios - Scenarios used to test different strategies
120
Advantages & Disadvantages of Scenario Planning for Insurers
Advantages 1. Company thinks through responses beforehand – Can select the best strategic response to different market conditions 2. Reduces organizational inertia – Provides more flexibility compared to trying to “make plan” even if market conditions change Disadvantage: The insurer is going to have to design and implement an approach to monitor the market conditions, which could be difficult
121
Underwriting Cycle Definition
Underwriting Cycle “The recurring pattern of increases and decreases in insurance prices and profits.”
122
Characteristic Stages of the Underwriting Cycle
Emergence * For a new line, there is a cycle between soft and hard markets: o Price wars between competitors -> price corrections when weak competitors leave -> period of profitability -> new competition to restart the cycle * Competition drives pricing dynamics Control * Stabilization as rating bureaus and state DOIs regulate prices. * Statistical data lags drive pricing dynamics Breakdown * Controls breakdown due to technological and societal changes * Competition and data lags drive pricing dynamic Reorganization * A new “version” of the line of business or marketplace emerges, returning to the emergence phase. * Competition drives pricing dynamics
123
Theories of What Drives the Underwriting Cycle
Institutional Factors 体制因素 The cycle is driven by time lags involved in pricing. Time lags due to reporting and regulatory delays also contribute to the underwriting cycle. (Immature losses for estimation, and delay with regulatory approval) Competition The cycle is driven by competition between insurers toward lower rates - to unprofitable levels for market share leading to crisis and price correction Underwriting strategies fluctuate between growth and price maintenance. (Not all competitors have the same view of the future. Inexperienced firms may have poorer loss forecasts than mature firms. As a result, inexperienced firms may drop prices based on poor forecasts. This eventually pushes the market toward lower rates) Supply and Demand, Capacity Constraints and Shocks Shocks to capital (catastrophes, …) reduce capacity, causing price increases. Economic Linkages Insurance profitability and cost of capital are driven by macroeconomics. – An insurer’s investment income impacts its profitability – The cost of capital is based on the state of the economy – Expected losses are influenced by economic factors such as inflation/ GNP growth/ unemployment
124
Potential Predictor Variables for Modeling the Underwriting Cycle
* Previous time period profitability * Internal financial variables (e.g. reserves, investment income, catastrophe losses, capital flows) * Regulatory and rating variables (rating upgrades/downgrades) * Reinsurance sector financials * Macroeconomic variables (inflation, unemployment, GNP) * Financial market variables (interest rates, equity returns)
125
Methods to model the Underwriting Cycle:
Soft Approaches Soft approaches focus on gathering data and doing competitor intelligence, which analysts look at to gauge the state of the underwriting cycle. eg. Scenarios/ Delphi Method/ Competitor Analysis Technical Modeling The underwriting cycle is modeled as a time series such as an autoregressive model. The model can then create forecasts for the future, which can be used in an ERM model or for strategic planning. Econometric Modeling (Behavioral Modeling) Balances soft approaches, which seek to understand the structure of the underwriting cycle, and technical modeling, which seeks to create a statistical model. It models changes in supply and demand of insurance and how those changes influence equilibrium price of insurance
126
Components of an Econometric Model: Supply and Demand Curves
Supply and Demand Curves * Shocks to capital (e.g. catastrophes) decrease supply (and potentially decreases demand if quality decreases) * Increased competition increases supply (and potentially increases demand if insurance quality rises) * Supply curve could be modeled with a minimum and maximum price.
127
Components of an Econometric Model: Capital Flows
Capital Flows * High expected profits result in capital inflows (new competitors, capital infusions) * Low profits result in capital outflows (firms exiting the market) * Normal dividends result in a steady outflow of capital
128
Components of an Econometric Model: Assembling the Model - Underwriting Cycle Relationships
Assembling the Econometric Model * Economic factors influence supply and demand curves * Capital flows impact supply and demand curves * Current level of prices and losses influence supply * Price and quantity of insurance is modeled at the intersection of supply and demand * Premiums and losses affect capital * Profitability impacts capital flows
129
primary cause of insolvency root reason for insurer failure
Although deficient reserves are normally cited as the primary cause of insolvency the root reason for insurer failure is the accumulation of too much exposure for the supporting asset base.
130
Compare the prediction intervals constructed using a simple trend model with those constructed using a time series.
In the simple trend model, the prediction intervals widen with time due to the uncertainty in the estimated trend rate. In the time series model, the prediction intervals widen with time as well, but the effect is more pronounced and the prediction intervals are wider. This is due to the additional uncertainty of the auto-regressive process
131
When building a model, various rules and metrics are used to select the best model form. However, the selected form may still be wrong. Describe a process to overcome this problem.
-Assign probabilities of being right to all of the better-fitting distributions. These probabilities can be based on the Hannan-Quinn Information Criteria (HQIC) metric or a Bayesian analysis -Use a simulation model to select a distribution from the better-fitting distributions -Select the parameters from the joint log-normal distribution of parameters for the selected distribution -Simulate a loss scenario using the parameterized distribution -Start the process over again with the next scenario
132
For large datasets, the parameter distributions in the MLE procedure are multivariate normal. Briefly describe two problems that may arise when this normality assumption is used for a small dataset.
1) The standard deviations of the parameters can be high enough to produce negative parameter values with significant probability 2) The distribution of the parameters may be heavy-tailed (the bi-variate normal is not heavy-tailed)
133
Briefly describe how estimation risk is assessed using maximum likelihood estimation (MLE).
To assess estimation risk, we use the covariance matrix that results from the standard MLE procedure (based on second partial derivatives of the parameters), but we assume the parameters follow a joint log-normal distribution with that covariance matrix (works for both large and small datasets)
134
Briefly describe a situation in which estimating parameters using maximum likelihood estimation (MLE) is difficult.
The best-fitting parameters can be difficult to determine if the likelihood "surface" is very flat near the maximum. When the surface is flat near the maximum, a wide range of parameter sets have almost the same likelihood. Thus, the set that maximizes the likelihood might not be any better than one that has slightly smaller likelihood
135
Rank the 3 models below in terms of increasing acceptance by the insurance industry: DFA, ERM, Cat model Explain one reason explaining the level of acceptance of each of the models mentioned
DFA< ERM< Cat model DFA: difficult to implement as it requires an internal champion (that most insurers lacked) in order to work with the different silos of the insurer. There was insufficient pressure to move to DFA to overcome this obstacle ERM: this is also difficult to implement as it requires working with the different parts of the insurer. However in this case there is increasing pressure to use it, one reason being because the ERM program will be assessed by the rating agencies Cat models: high pressure to use these as insurers suffered huge losses from Hurricane Andrew
136
List 2 examples of risks that are difficult to incorporate in an Enterprise Risk model. Briefly describe 2 options how to handle these risks
including operational risks such as IT hardware & software exposures, pension funding inadequacy, loss of key executives, fraud, etc... One option is to model operational risks in bulk using judgment. The issue is that this will have a high degree of uncertainty * A more reasonable alternative is to just use the model for risks that it can quantify reasonably well, and recognize that there are operational risks that are not handled well by the model that need some other management method
137
An insurer is setting its capital need based on the TVaR at the 99%. Describe 3 issues with this Suggest an alternative that would be better, and explain why this is preferable
It is difficult to model scenarios at this level of confidence Tail events are poorly understood There is little available data to help derive the tail TVaR at 80%: this is far enough in the tail to still generate significant issues for the insurer. However, because this scenario is further away from the tail, it should be easier for the insurer to model (and should be understood better by the modelers).
138
Common approaches for setting capital requirements
⋄ Holding enough capital to continue to service renewals (since renewals tend to be more profitable) ⋄ Holding enough capital so that the insurer not only survives a major cat but thrives in its aftermath ⋄ Holding enough capital so that the probability of default is remote. However, it is very conservative and mainly protects the policyholder ⋄ Holding enough capital to maximize the insurer’s franchise value. Franchise value includes an insurer’s balance sheet, customer base, agency relationships, reputation, etc. Maximizing franchise value protects both the policyholder and the shareholder
139
For a public firm, why a certainty equivalent approach with a corporate risk preference be undesirable. And what’s the issues with this reason
Diversified investors only care about systematic risk since firm specific risk is diversified away in a diversified portfolio. Therefore, firm managers should also be indifferent to firm-specific risk in order to maximize shareholder value. Thus unnecessary. Its assumptions are unlikely to apply in practice. For example: * It is difficult for management to determine which risk is firm specific versus systematic * The risk-adjusted rate is often used to be a proxy for the risk level. However this rate mainly reflects risks that will arise in the future, but the insurer is also concerned about risks that may arise before that * The market based data is so noisy and so it would be difficult for management to conduct a proper cost-benefit analysis
140
considerations when implementing an (Enterprise Risk Model) Internal Risk Model
1 Startup: Staffing & Scope 2 IRM Parameter Development 3 Implementation 4 Integration & Maintenance
141
elements are needed to make decisions from the output of the (Enterprise Risk Model) Internal Risk Model
1. Corporate Risk Tolerance 2. Cost of Capital Allocated 3. Cost-Benefit Analysis for Risk Mitigation
142
Briefly describe 2 methods to generate social inflation from the company data
1. Calculate total trend from the claims data, and then net out inflation to generate the social inflation 2. Use general inflation metrics to adjust the data, and then model the residual superimposed inflation
143
Tools can be used to manage operational risk
1. Control Self assessment A process to examine and assess the effectiveness of internal controls to provide reasonable assurance that all business objectives will be met 2. Key Risk Indicator Measures used to monitor the activity and status of controls in a business area for a given operational risk category. KRIs are leading indicators of risk and can be measured frequently to flag attention if a threshold is hit 3. Six sigma can help firms identify and eliminate process issues such as inefficiencies, errors, overlaps and gaps in communication
144
key component of operational risk
plan loss ratio determination The authors believe that "insufficient reserves" are a symptom of poor company analyses/managerial decisions They are a lagging indicator of either: – insufficient reserving initially (due to optimistic plan loss ratios); or – premature reserve releases
145
issue with this planning: Loss Ratio 65% Rate Change 8% Premium $400M
The issue is that it is static, having just one inflexible plan, and may not be appropriate if there are changes in the market. For example if the rate level drops, it may make more sense to reduce the premium goal accordingly.
146
Explain why Gron supply curve may apply instead of a "traditional" shape
There is a minimum price which reflects the expected losses and marginal costs of writing the business. It would not make sense to sell policies below this price. * Up to a certain quantity level, all business will sell at this minimum price * After this level, the industry has reached capacity and will therefore need to obtain more capital in order to write more business. It will need to charge for the incremental cost of capital, causing the supply curve to slow upward * Towards the right of the curve, the premium itself is providing enough margin to support this cost of capital, and so it reaches a maximum level
147
3 paradigms of measuring reinsurance value
1) reinsurance provides stability 2) reinsurance frees up capital 3) reinsurance adds market value to the firm
148
Explain the difficulty in determining the cross-class correlation component of the internal model
1. Correlation requires lots of data to estimate, and is thus difficult to quantify 2. The correlation may change across the distribution, which requires a complex method such as copulas model 3. Correlation of 2 distributions can be done with a wide array of copulas. The options are limited when combining many classes.
149
Delphi method
Experts are given background information and asked for their opinions in a questionnaire. The answers are aggregated and then summaries are given back to the participants. Based on the summaries, participants can change their answers are articulate their reasons for disagreeing. This process is repeated until consensus is reached
150
Briefly describe one feature that econometric modeling of the UW cycle has in common with “soft” approaches and one feature that it has in common with technical models
1. Similarity to soft approach - econometric modeling includes the recognition of human factors impacting the UW cycle 2. Similarity to technical approach - econometric modeling requires mathematical rigor
151
An insurance company has decided to manage the underwriting cycle by reducing market share when pricing is soft and expanding market share when pricing is hard Justify an asset management strategy that could reduce the company’s earnings volatility. And describe a risk that would increase if this strategy were implemented
During soft market, invest more in high yielding assets such as equity and high-yield corporate bonds during soft market. Because during soft market, company is taking on less insurance risk by reducing market share, so it makes sense to take on more asset risk, and extra investment income would help offset the reduction in UW income However, asset risk would increase during soft market. Equities are riskier than bonds. There is a risk that market price would decline after you invest more heavily in equities
152
2 potential shortcomings of a traditional unilateral planning approach in which one version of “the plan” is set
1. When actual deviate from the overly optimistic plan, managers are reluctant to deviate from the plan numbers. This results in booked numbers that are unrealistic for far too long 2. Underwriters do whatever is necessary to meet their plan numbers (even if what they are doing is bad for the company)
153
2 potential negative consequences of an optimistic plan loss ratio to the company’s financial results
Using an optimistic planned loss ratio as an ELR can lead to: 1. Reserve deficiencies 2. Inadequate rates
154
Why is it difficult to separate operational risk from underwriting risk when explaining the impact of an optimistic plan loss ratio on the company’s financial results in retrospect
It’s difficult to determine if the forecasting model could not accurately predict the loss ratio (underwriting risk) or if the model was simply not used/implemented appropriately (operational risk)
155
Why parameter risk is a key source of uncertainty in enterprise risk modeling
Because if an estimated parameter does not accurately reflect the true underlying parameter, the projected results may be significantly understated or overstated
156
Why graphs of right tail concentration function can often be misleading and recommend a solution to this problem
A copula can have significant right tail dependence even if R(1) = 0. This is due to the fact that the function decreases rapidly as z approaches 1. A solution is to look at the function at values slightly below 1 and assess the strength of the dependence at those values
157
Define and explain the right tail and left tail concentration function that may be used to characterize a copula
The right tail concentration function calculates the probability density of a copula in the right tail. R(z) = P(U>z, V>z)/(1-z) = [1 - 2z + C(z,z)]/(1-z) L(z) = P(U
158
Capital must be sufficient to:
1. Support growth 2. Sustain current underwriting 3. Satisfy regulators, rating agencies, and shareholders 4. Provide for adverse reserve changes 5. Provide for declines in assets
159
Good enterprise risk model and weak enterprise risk model
1. The model shows the balance between risk and reward from different strategies (such as changing the asset mix or reinsurance program) 2. The model recognizes and reflects its own imperfection, including parameter uncertainty, simplistic assumptions or poor data quality And the 4 model qualities Models without the characteristic above often exaggerates certain aspects of risk while underestimating others. This can lead to overly aggressive or overly cautious corporate decision.
160
Different methods the company can use to allocate capital
Proportional allocation Calculate risk measure for the insurer and each business unit separately. Allocate the total risk measure for the insurer proportionally using the individual risk measure Marginal decomposition Calculate the overall risk measure for the insurer. Then calculate the marginal co-measures for each business unit. The marginal co-measures sum to the company risk measure. Marginal decomposition is better since it reflects how the risk from each business unit impacts the total risk profile as opposed to looking at them in isolation
161
Regulatory and Rating Agency Capital Adequacy Models
Purpose: to evaluate capital adequacy 1. Leverage ratios 2. Risk-Based Capital Models 3. Scenario Testing
162
Briefly discuss a risk of holding short-term assets and a risk of holding long-term assets, relative to the liability duration
Short-term: Reinvestment risk: if interest rate drop, bonds will be reinvested at lower interest rates and investment income may be too low to cover the liabilities Long term: If interest rate increase, the value of the bonds fall and they may need to be liquidated at depressed prices to fund liabilities
163
Considerations that should be included in the asset-liability analysis to make it more realistic and how they might impact the asset-liability matching
1. Insurance liabilities are variable in amount and timing, making duration matching impossible 2. The company can pay claims from premium cash flows, so an enterprise-wide model is needed 3. Tax considerations will also impact the optimal investment strategy
164
Discuss how the accumulated risk of loss reserves can be incorporated into the model for long tail line
Model the capital absorbed in the current year using the combination of current accident year and as-if reserves. The as-if reserves are the loss reserves that would exist at the start of an accident year if the business currently being written had been written in a steady state for all prior years.
165
Briefly discuss the impact of accumulated risk on the distribution of underwriting loss and how this will affect the required capital for the insurer
Adding accumulated risk spreads out the distribution and results in a fatter tail. This will cause TVaR to be significantly higher, resulting in a higher required capital
166
Describe how an efficient frontier can be used to select an insurance portfolio
An efficient frontier plot shows risk on the y-axis and reward (or return in this case) on the x-axis. The curve graphs the portfolios that maximize return for a given risk level. If the current portfolio has the same return as one of the efficient portfolios but more risk, then it is sub-optimal. In order to select one of the efficient portfolios, firms must decide how much risk they are willing to tolerate and how much reward they are willing to give up for a reduction in risk 左上角的points更好,但是要consider company preferences and budget constraints,有多少additional cost 才能lower the probability of making plan
167
2 required conditions for a marginal decomposition
1. Works when business units can change volume in a homogeneous fashion 2. Works when the risk measure is scalable. This means that multiplying the random variable by a factor multiplies the risk measure by the same factor Rho(aY) = a*rho(Y) Also known as homogenous of degree 1
168
Difference between static scenarios and stochastic scenarios
Static scenarios are pre-defined scenarios (defined by the firm). Stochastic scenarios are generated through a stochastic process
169
Why using combined ratio to compare reinsurance programs is misleading
If a reinsurance has slightly stronger underwriting results but far more ceded premium, then the combined ratio will actually be worse (even though the absolute underwriting income is better) (对于insurer来说,ceded premium多了,所以net premium变少了,所以combined ratio变少了)
170
To project future losses, an actuary fit a trend line to historical data. Using standard statistical procedures, the actuary placed prediction intervals around the projected losses. Explain why these prediction intervals may be too narrow
Historical data is often based on estimates of past claims which have not yet settled. In the projection period, the projection uncertainty is a combination of the uncertainty in each historical point AND the uncertainty in the fitted trend line. Thus, the actuary’s prediction intervals may be too narrow due to the missing uncertainty associated with the historical data
171
Primary causes of P&C company impairment
1. Deficient loss reserves 2. Underpricing 3. Rapid growth 4. Alleged fraud 5. overstated asset 6. Catastrophe 7. Reinsurance failure
172
Definitions of strategic risk focus on either strategic risk taking or strategic risks
Strategic risk-taking refers to intentional risk-taking as an essential part of a company’s strategic execution. Strategic risks are unintentional risks that occur as a result of strategy planning or execution
173
How scenarios and Delphi method feed off each other
A Delphi process can create a set of scenarios and scenarios can form the input to a Delphi assessment about the likelihood of each scenario
174
4 questions that must be answered before an econometric model can be built
1) how do economic factors (interest rate, inflation, cost of capital) influence the supply and demand curves? 2) how does capital influence the supply and demand curve? 3) how does the supply demand curve jointly determine price and quantity? 4) how does profitability affect external capital flows?
175
Once an econometric model is built, describe how it can be used to create an empirical distribution of possible future market equilibrium prices
An econometric model is comprised of various components such as supply curves, demand curves, capital flows, inflation, etc. Each component influences the equilibrium price in specific ways. Various component changes can be simulated to create an empirical distribution of possible future equilibrium prices.
176
3 styles of modeling UW cycle, how do they vary by 3 dimensions
Dimension 1 - data quantity, variety and complexity: soft > behavioral > technical Dimension 2 - recognition of human factors: soft > behavioral > technical Dimension 3 - mathematical formalism and rigor: technical > behavioral > soft
177
4 sources of competitor intelligence that could be used to inform an UW cycle model
Sources of information include 1. Customer surveys 2. Trade publication 3. News scanning 4. Rate filling