22.3 IRM Parameter Development

Potential challenges on parameters that need to be developed

• Data quality

• Unique characteristics of the company or LoB

• Differing risk attitudes

Expert opinion is important in developing parameters

• Especially for segments with small volume and low quality data

• This will increase the time to implement

Correlation assessment is important and challenging

• Lack of data

• High political sensitivity

• Spans multiple business units

• Significant impact on the overall risk profile and capital allocation

Validation will be difficult as there are no model to compare

• Validate by reviewing a series of tests involving complementary variables, e.g.:

  • Compare extreme outcomes from the model to actual extreme outcomes

  • Compare variation of loss ratios to historical variation

  • Model accurately handle correlation

  • Capital allocation make sense? (long tail and cat prone risk have more capital)

• Gain comfort with the reasonableness of the model through the use of many metrics

22.3.1 Recommendtaions

Modeling software

• How much is pre-built from purchased software

• What needs to be built in-house

• Skills of the team should align with what they need to build

  • e.g. Buy an ESG unless your team have finance and economics experts with time series modeling

Parameter development

• Need to develop a systematic way to capture the expert opinion (product expertise) from the following departments:

• U/w, planning, claims, planning and actuarial

Correlations

• Needs to be owned at a high level (C-suites) as it crosses LoBs and have significant impact on the allocated capital

• IRM team will provide recommendations on correlation assumptions

Validation

• Validate and test over extended period

• Provide training so that interested parties all have a basic understanding of the model