ASOP.12

Reading: Actuarial Standard Of Practice No. 12 – Risk Classification

Synopsis: Guidance on selecting risk characteristics plus establishing and testing risk classification systems

Study Tips

If you're reading the source material, Section 3 is the key part. Refer to Section 2 to verify definitions. Although the syllabus doesn't say it, you're probably safe skipping the appendices.

Estimated study time: 30 minutes (not including subsequent review time)

BattleTable

Based on past exams, the main things you need to know (in rough order of importance) are:

• Considerations when selecting risk characteristics
• Considerations when establishing risk classes
• Adverse selection and testing risk classes

reference	part (a)	part (b)	part (c)	part (d)
E (2018.Spring #1)
E (2018.Spring #1)
E (2018.Spring #1)
E (2018.Spring #1)
E (2018.Fall #4)
E (2017.Fall #1)
E (2016.Fall #3)
E (2013.Fall #1)

In Plain English!

This standard only applies to actuaries providing professional services involving the design, review, or change of risk classification systems. Pricing actuaries should be very familiar! Other areas where this standard may apply involve expert testimony, regulatory and/or legislative activities.

Indirect impacts on risk classification systems don't excuse you from following this ASOP when there is a likelihood of a material impact.

Laws/Statutes/Regulations should be followed over this ASOP. However, you should document why you were required to deviate from the ASOP.

Issues and Recommended Practices when Selecting Risk Characteristics

• Relationship of risk characteristics and expected outcomes
• Objectively measure
• Causality
• Equitable
• Practical
• Applicable law
• Industry/Business Practices

Let's go into more detail on each of these.

Relationship of Risk Characteristics and Expected Outcomes

Should choose risk characteristics that you can demonstrate a relationship between the characteristic and outcome such as expected pure premium. For instance, if you're exploring water claim frequency then the number of dogs probably isn't a good characteristic to capture but the age of plumbing likely is. Should consider the interdependence of risk characteristics and adjust if interactions are material (more on this in [GLM.ModelBuilding]). For instance, amount of Coverage A and square feet of living area are two risk characteristics that are likely interdependent - bigger homes tend to be more expensive.

Objectively Measure

Alice is reviewing a rate filing. It proposes applying a surcharge of 10% for "Owning a fast car". This is great - Alice's drag car in Miami isn't fast! Instead, risk characteristics should be clearly defined, not easy to manipulate, and readily verifiable. Alice suggests a 10% surcharge when owning a car capable of 0-60mph in under 1 second, and looks elsewhere to insure her drag car.

Causality

It's not necessary for the actuary to establish cause and effect before using the risk characteristic.

Equitable

Differences in rates should reflect material differences in the expected cost for risk characteristics.

Practical

The time, money and effort required to evaluate the risk should be reasonable.

Applicable Law

Proposed risk characteristics need to compliant with the law. Various laws may limit or exclude the use of certain variables. For instance, race and religion are almost always not allowed. The use of credit is limited or prohibited in some states.

Industry/Business Practices

Does the proposed variable follow industry standards or practices? Are there any business limitations?

Considerations when Establishing Risk Classes

A risk classification system assigns each risk to a risk class based on observing/measuring its risk characteristics.

Here are the key things to check when designing risk classes (insert mnemonic).

• Intended use
• Actuarial considerations
  • Risk of adverse selection
  • Credibility
  • Practicality
• Reasonableness of results

In more detail:

Intended Use

Will the risk classification help solve the problem? For instance, it may be sufficient to classify Homeowners risks via Special, Tenant or Condominium Owners Forms for pricing but if you're setting loss reserves then this classification might need to be refined to analyze by the amount of insurance.

Adverse Selection

Is the variation of expected outcomes too large within a risk class? If so, you need to improve the homogeneity of the class.

Credibility

Risk classes should be large enough to have believable statistics. It's important to balance this with homogeneity.

Practicality

Can you place a risk in the correct class accurately and in a timely and affordable manner?

Reasonableness of Results

Look at the risk classes for consistent patterns in rates/values/factors (or a lack of). Should avoid illogical rating such as 0 dogs = 1.000, 1 dog = 1.500, 2+ dogs = 1.100 for Homeowner Liability.

Testing the Risk Classification System

You should assess the potential effects of adverse selection resulting from the design or implementation of the risk classification system. Consider the long-term viability of using different risk classes, quantifying the impacts where possible and determining if there is a need to change from an existing classification.

Other Applicable ASOPs

23: Data Quality - use when relying on data or information supplied by others.

41: Actuarial Communications - document assumptions and methodologies used. Retain appropriate documentation.

Disclosures

Need to disclose when there are significant limitations due to laws, business and industry practices (among others). Disclose the reliance on the work of others such as other actuaries or data scientists. Disclose when you followed legally prescribed methods and/or deviated from the guidance in this ASOP and/or traditional methods found in other ASOPs.

Legal Test for Risk Classification

The legal test for risk classification has remained unchanged for several decades. Risk classification is allowed so long as it is "based on sound actuarial principles" and "related to actual or reasonably anticipated experience".