Linking ERM With Strategy Formation

Introduction

A strategic plan is a choice made before its results are known. A proposed expansion, a new product, or a change in distribution may create value under some conditions and disappoint under others. ERM helps decision-makers bring those uncertainties into the discussion while strategy is being formed, rather than treating risk as a review conducted after the main choices have been made.

The purpose is not to make uncertainty disappear or to reduce strategy to a calculation. It is to clarify the range of plausible outcomes, the assumptions behind them, and the trade-offs among strategic possibilities. That information can help leaders shape objectives and plans that are coherent with the organization’s circumstances and its capacity to bear risk. This is aligned with the CAS Exam 9 (Risk Management for Actuaries) syllabus.

Definition: Linking ERM with strategy formation means using an enterprise-wide view of uncertainty and potential outcomes to inform the development and comparison of strategic objectives and plans, before those objectives and plans are finalized. The analysis considers both the value a strategy may create and the uncertainty surrounding that value.

What ERM Adds Before a Strategy Is Chosen

Strategic discussions often begin with an attractive objective: grow in a market, improve service, or increase long-term value. ERM contributes by asking what important conditions could affect that objective and how outcomes might differ across plausible conditions. This is an input to strategy formation, not a substitute for management judgment or strategic expertise.

A useful analysis can include both quantitative and qualitative information. Quantitative scenario estimates can illustrate how alternative conditions affect a common outcome measure. Qualitative analysis can identify uncertainty that is difficult to represent credibly with a number, such as execution complexity, dependence on a key partner, or the possibility that customer behavior changes. In either case, the analysis should make assumptions visible rather than imply that uncertain estimates are precise facts.

A disciplined comparison usually asks:

What strategic possibilities are being compared, and what is the planning horizon?
What outcomes matter for the organization’s strategic objective?
What plausible scenarios could materially change those outcomes?
What assumptions support the scenario estimates and their likelihoods?
How do the alternatives compare not only in expected outcome but also in the pattern of outcomes?

These questions belong early enough to affect the formation of objectives and plans. If a material uncertainty is considered only after the strategic direction has been fixed, the organization may have less flexibility to alter the plan or to consider a different objective.

Scenario-Based Comparison

For a simple quantitative comparison, suppose each strategic option has a finite set of possible scenarios. Let $G_j$ be the outcome for option $j$ over a specified planning horizon, measured on a consistent basis. For scenario $s$, let $g_{js}$ be the outcome and $p_s$ its probability. The scenarios must be mutually exclusive and collectively exhaustive: exactly one scenario occurs, and the probabilities sum to one.

Result: Under a discrete scenario model with $p_s \geq 0$ and $\sum_s p_s=1$, the expected outcome for option $j$ is the probability-weighted average of its scenario outcomes:

E[G_j]=\sum_s p_s g_{js}.

This is an average across the modeled possibilities, not a promise about the outcome that will actually occur. It is also only as credible as the scenarios, outcome estimates, and likelihood assumptions used to calculate it. When options are compared, their outcomes should use the same horizon and measurement basis. If the modeled amounts occur at different times, they must first be placed on a consistent valuation basis; the examples below assume a common basis and do not apply discounting.

Why the Probability-Weighted Formula Holds

The formula follows from writing the outcome as a sum over scenarios. Let $I_s$ equal 1 if scenario $s$ occurs and 0 otherwise. Since exactly one scenario occurs, the outcome for option $j$ can be written as $G_j=\sum_s g_{js}I_s$. Taking expectations and using the additivity of expectation gives:

E[G_j] =E\left[\sum_s g_{js}I_s\right] =\sum_s g_{js}E[I_s] =\sum_s g_{js}p_s.

The step $E[I_s]=p_s$ holds because $I_s$ is an indicator: it is 1 with probability $p_s$ and 0 otherwise. This derivation requires no assumption that scenario outcomes are normally distributed. It does require a well-defined set of scenarios and probabilities that describe the modeled possibilities. If the scenarios omit a material possibility or their probabilities do not sum to one, the weighted average is not a valid expectation for the stated model.

Worked Example: Comparing Two Expansion Options

A hypothetical insurer is considering two expansion options. For each option, the listed outcome is net value in millions of dollars over the same planning horizon, measured on a common valuation basis. No interest-rate or mortality model is used; the scenario outcomes and probabilities are the full discrete model. Assume the three scenarios are mutually exclusive and exhaustive, with probabilities 0.30, 0.50, and 0.20. Compare expected outcomes, the chance of reaching at least $5 million, and the chance of a negative outcome.

Method. First calculate the probability-weighted outcome for each option. Then separately add probabilities for scenarios meeting the target or producing a negative result. Keeping these questions separate avoids treating the average as a full description of the outcomes.

Option A. Its scenario outcomes are $12 million, $6 million, and \(-$6 million. Thus:

E[G_A]=(0.30)(12)+(0.50)(6)+(0.20)(-6) =3.6+3.0-1.2 =5.4\text{ million dollars}.

The chance of reaching at least $5 million is $0.30+0.50=0.80$, because the first two outcomes meet the target. The chance of a negative outcome is $0.20$.

Option B. Its scenario outcomes are $9 million, $7 million, and $3 million:

E[G_B]=(0.30)(9)+(0.50)(7)+(0.20)(3) =2.7+3.5+0.6 =6.8\text{ million dollars}.

All three outcomes are at least $5 million, so the chance of reaching the target is $0.30+0.50+0.20=1.00$ under this model. None is negative, so the modeled chance of a negative outcome is 0.

Comparison and check. Option B’s expected outcome exceeds Option A’s by $6.8-5.4=1.4$ million dollars. The target probabilities are 1.00 for B and 0.80 for A. The probabilities in each calculation use all the relevant scenarios and sum to 1. The comparison is conditional on the stated scenarios and assumptions; it does not guarantee B’s outcome in practice.

Common mistake avoided: the weighted sum for A is $3.6+3.0-1.2=5.4$, not 4.2. The difference between the expected outcomes is therefore $1.4$, not 2.6.

Expected Outcome Does Not Tell the Whole Story

An expected outcome is useful for summarizing a probability model, but it can conceal differences in the possible results. Two options can have the same expected value and very different downside scenarios. Conversely, one option can have a higher expected value while also having a meaningful chance of an outcome that would challenge the organization’s plans.

For that reason, strategic analysis should identify the particular outcome features that matter to the decision. These might include the probability of falling below a stated objective, the scale of a plausible shortfall, or whether the plan depends heavily on one favorable condition. The appropriate comparison depends on the organization’s objectives and context; no single summary statistic answers every strategic question.

Worked Example: Testing Sensitivity to Scenario Likelihoods

Use the same hypothetical expansion options and the same scenario outcomes as in the prior example. A planning team now considers a less favorable outlook: the scenario probabilities are 0.20, 0.30, and 0.50, respectively. The outcomes remain on the same horizon and valuation basis, and the scenarios remain mutually exclusive and exhaustive. Recalculate the expected outcomes and explain what changed.

Method. Keep each option’s outcomes attached to the corresponding scenario, replace the probabilities with the revised assumptions, and recompute the weighted sums. This is a sensitivity analysis, not a claim that the revised probabilities are known with certainty.

Option A. The outcomes remain $12 million, $6 million, and \(-$6 million:

E[G_A]=(0.20)(12)+(0.30)(6)+(0.50)(-6) =2.4+1.8-3.0 =1.2\text{ million dollars}.

Option B. The outcomes remain $9 million, $7 million, and $3 million:

E[G_B]=(0.20)(9)+(0.30)(7)+(0.50)(3) =1.8+2.1+1.5 =5.4\text{ million dollars}.

Option B’s expected outcome is higher by $5.4-1.2=4.2$ million dollars under these revised probabilities. Option A has a 0.50 chance of a negative outcome because the third scenario now has probability 0.50. Option B has no negative outcome in the stated scenario set.

Check and interpretation. The revised probabilities sum to $0.20+0.30+0.50=1.00$. The expected outcomes are $1.2$ and $5.4$, not $1.8$ and $5.6$. The altered likelihood assumptions make the comparison more favorable to B, but sensitivity analysis should prompt examination of why the probabilities changed and how defensible they are.

Common mistake avoided: do not change the scenario order or pair a probability with the wrong outcome. Also, changing likelihoods does not change the scenario outcomes themselves unless the model explicitly says it does.

Compare an Objective with Its Probability of Achievement

A strategic objective may be expressed as a threshold, such as achieving at least a specified outcome over a defined period. The probability of meeting that threshold is a different quantity from the expected outcome. It can show how often the modeled result meets the objective, while the expected value summarizes the weighted average of all modeled results.

Worked Example: Expected Outcome Below a Target Despite Frequent Success

A hypothetical service initiative has three possible net outcomes over a common planning horizon: $14 million with probability 0.30, $8 million with probability 0.40, and \(-$2 million with probability 0.30. The values are stated on the same valuation basis, so no discounting is applied. There is no mortality assumption or separate interest-rate model. Calculate the expected outcome and the probability of meeting an objective of at least $8 million.

Method. Use the scenario-weighted average for the expected outcome. For the objective probability, include each scenario whose outcome is at least $8 million.

E[G]=(0.30)(14)+(0.40)(8)+(0.30)(-2) =4.2+3.2-0.6 =6.8\text{ million dollars}.

The outcomes meeting or exceeding $8 million are $14 million and $8 million, so:

P(G\geq 8)=0.30+0.40=0.70.

The expected outcome is $6.8 million, below the $8 million objective, while the model assigns a 0.70 probability to meeting or exceeding that objective. There is no contradiction: the $2 million loss outcome pulls down the average, even though it does not occur in the two scenarios that meet the threshold.

Check and common mistake avoided: the weighted terms total $4.2+3.2-0.6=6.8$, not 6.2. For the target probability, include outcomes equal to the threshold: the $8 million scenario counts because the objective is “at least $8 million.”

Using the Analysis in Strategy Formation

Scenario comparisons are most useful when they inform a real strategic discussion. ERM can help leaders make assumptions explicit, identify which uncertainties could materially change the comparison, and consider whether an objective or plan remains sensible across more than one plausible future. If an option looks attractive only under one narrow set of assumptions, that dependence is relevant before the organization commits to the strategy.

The analysis also supports constructive challenge. Decision-makers can ask whether the modeled scenarios are sufficiently broad, whether likelihood estimates have a sound basis, and whether important qualitative considerations have been left out. Where evidence is weak, it is better to state that limitation than to present a precise probability as though it were a fact.

The result should inform—not dictate—the choice. Strategy formation also involves purpose, capabilities, stakeholder needs, competitive conditions, and judgment. A higher expected outcome does not automatically make an option preferable if its downside is incompatible with the organization’s circumstances. Similarly, an option with a lower expected value may still merit consideration for reasons that the quantitative model does not capture. ERM brings those trade-offs into view; it does not choose the organization’s priorities for it.

Common Exam Traps

Confusing an expected outcome with a guaranteed result. An expectation is a probability-weighted average under assumptions; it does not say which scenario will occur.
Using probabilities that do not cover the full model. Confirm that the scenarios are mutually exclusive and exhaustive and that their probabilities sum to 1.
Comparing unlike quantities. Options need a common horizon and measurement basis for their outcomes to be directly comparable.
Reporting only the average. State relevant outcome probabilities or thresholds when they matter to the strategic objective.
Treating scenario probabilities as certain facts. They are assumptions in the analysis and may need sensitivity testing or qualitative qualification.

Key takeaway: ERM informs strategy formation by making uncertainty and potential outcomes visible before objectives and plans are fixed. A probability-weighted outcome is one useful summary, but a sound comparison also considers the scenarios, assumptions, and objective-relevant chances of different results.

Check Your Understanding

For each problem, assume the scenarios are mutually exclusive and exhaustive and the stated outcomes share a common planning horizon and valuation basis. These are hypothetical strategic analyses; no interest-rate or mortality model is required.

A proposed distribution change has outcomes of $10 million, $4 million, and $-$5 million, with probabilities 0.25, 0.50, and 0.25. Calculate its expected outcome and the probability of a negative outcome. (Answer: Expected outcome \(=3.25$ million dollars; probability of a negative outcome $=0.25$.)
Two hypothetical options have scenario probabilities 0.20, 0.50, and 0.30. Option D produces $8 million, $5 million, and $1 million; Option E produces $11 million, $4 million, and $-$4 million. Calculate both expected outcomes, identify the higher one and the difference, and state E’s probability of a negative outcome. (Answer: \(E[G_D]=4.4$ million dollars; $E[G_E]=3.0$ million dollars; D is higher by $1.4$ million dollars; E’s negative-outcome probability is $0.30$.)
A hypothetical project has outcomes $16 million, $7 million, and $-$8 million with probabilities 0.20, 0.50, and 0.30. Find the expected outcome and the probability of meeting an objective of at least $7 million. (Answer: Expected outcome \(=4.3$ million dollars; probability of meeting the objective $=0.70$.)
A plan’s outcomes are $10 million, $5 million, and $-$5 million. Initially, their probabilities are 0.40, 0.40, and 0.20; in a revised scenario analysis they are 0.25, 0.35, and 0.40. Calculate both expected outcomes, the decline in expectation, and the change in negative-outcome probability. (Answer: Initial expected outcome \(=5.0$ million dollars; revised expected outcome $=2.25$ million dollars; decline $=2.75$ million dollars; negative-outcome probability increases from $0.20$ to $0.40$.)
Option K pays $15 million if a favorable scenario occurs and loses $5 million otherwise. Option L has a certain outcome of $6 million. Find the favorable-scenario probability that makes their expected outcomes equal. If K’s favorable-scenario probability is 0.60, which has the higher expected outcome and by how much? (Answer: Indifference probability $=0.55$; at probability $0.60$, $E[G_K]=7$ million dollars and K is higher by $1$ million dollar.)

What You'll Learn