Actuarial Notes

Business Strategy

Study guides: Exam PCPA, Exam 9, Exam CFE-101, Exam CFE-201, Exam CP-311

Business Strategy is the set of choices by which an insurer pursues its goals: which lines, markets and customers to write, how much risk to retain, and how much capital to hold and where to deploy it. Strategic management is the process of setting, executing and monitoring those choices. ERM ties risk to strategy by putting every option on a common risk-adjusted footing. In a modelling project, the business goal decides what the model must predict and how its success is judged.

\text{RAROC} = \frac{E[\text{Profit}]}{\text{Risk Capital}}

\text{EVA} = E[\text{Profit}] - r_h \times \text{Risk Capital}

  • Risk capital comes from the ERM model through a risk measure, such as VaR or TVaR at a chosen level (Risk Capital). r_h is the hurdle rate, the cost of capital. A strategy adds value when RAROC exceeds r_h, or equivalently when EVA (economic value added) is positive (Risk-Adjusted Performance). Ranking by RAROC and by EVA can disagree: the ratio can favour a smaller business, while EVA measures the value actually added.
  • How risk measures and models shape strategy (Exam 9). The ERM model turns strategic questions into comparable distributions of results. Those questions include capital adequacy, allocating capital to lines and units, reinsurance (Reinsurance Strategy), asset mix and planned growth. Plotting the alternatives by risk and return traces an efficient frontier, and the insurer picks a point on it within its risk appetite.
  • The choice of risk measure matters. Changing the measure, its confidence level or the allocation method changes each unit's capital, and so its measured return and which lines look worth growing.
  • Strategic management and levels of strategy (DISC-RM). The Strategic Management Process sets a vision, mission and goals, analyzes the environment, formulates, implements and evaluates. Strategy is set at more than one level: corporate strategy chooses which businesses to be in, and business strategy chooses how each one competes, for example by cost leadership or differentiation.
  • Strategic risk. The risk of choosing the wrong strategy, or executing it badly, is a risk category in its own right (Risk Taxonomies, Business Risk).
  • Business goals in a model (PCPA). The goal sets the target (loss cost, retention, litigation), the unit of analysis and the success metric. The GLM monograph names lack of stakeholder alignment on goals and outcomes as one of the most common reasons modelling projects fail or fall behind. Results are reported as business outcomes, such as premium movements, expected loss ratio and retention (Actuarial Communication).

Choosing a Reinsurance Strategy by EVA The hurdle rate is 12\%. Expected profit and risk capital (from the ERM model, in $ millions) under three options are:

  • A, no reinsurance: profit 60, capital 400
  • B, a 40% quota share: profit 45, capital 250
  • C, catastrophe excess of loss: profit 52, capital 300

Rank the options by RAROC and by EVA.

Answer

  • RAROC: A 60/400 = 15.0\%, B 45/250 = 18.0\%, C 52/300 = 17.3\%.
  • EVA: A 60 - 0.12(400) = 12, B 45 - 0.12(250) = 15, C 52 - 0.12(300) = 16.

All three clear the hurdle. B has the best ratio, but C adds the most value: it gives up less profit to the reinsurer than the quota share while still releasing \$100 million of capital. If the choice were made on RAROC alone, the insurer would shrink more than it should.

The ranking is only as good as the capital figures. Under a different risk measure or confidence level, the three capital amounts change, and so can the ranking.

Same Data, Different Business Goal A commercial property insurer asks for two models from the same submissions data. Underwriting wants to decide which new submissions get an on-site inspection, and it can inspect 15% of them. Pricing wants new rating relativities. How should the two designs differ?

Answer

  • Target. Underwriting needs the probability of a large loss, or of a loss ratio above a threshold. Pricing needs the expected pure premium (frequency and severity, or Tweedie) with an exposure offset.
  • Success metric. Underwriting needs the share of large losses captured in the top 15% of scores, because that is the inspection budget. Pricing needs lift over the current plan: quantile plots, a double lift chart and a loss ratio chart.
  • Constraints. The underwriting model may use any submission information available when the decision is made, subject to underwriting rules. Pricing variables must be permissible, filable and programmable.
  • The message to the business. For underwriting: "inspecting the top 15% finds about X% of the large losses." For pricing: which customers' premiums move, by how much, and the expected change in loss ratio.

A single model built without settling these questions would serve neither goal well.