Outgrace
OutgraceThe Leadership Magazine, home
StrategyExplainer

Scenario planning or forecasting: how to choose between them

A point forecast tells you what is likely. A scenario set tells you what would hurt, and how early you would know. Here is how to choose between them and build three stories with signposts.

A point forecast gives one number for a future date. A scenario set gives several internally consistent stories about how the future might unfold, with no claim that any one of them is most likely. Use the first when a single expected value drives the decision. Use the second when a bad outcome in one direction would be costly, slow to undo, and invisible in the central case.

The confusion is expensive in both directions. Boards ask for a forecast when they need to see what a shock would do, and they receive a tidy number that hides the shock. Other teams build elaborate scenario exercises for choices that needed only a sensible base case and a clear view of what being wrong would cost. In our experience the mistake is rarely the method itself. It is choosing a method before anyone has written down the decision it is meant to inform.

How it works

A forecast answers the question of what will most likely happen. The European Central Bank's Survey of Professional Forecasters, which the ECB has run since 1999, shows the form well. It collects expectations for euro area inflation, real GDP growth and unemployment at several horizons, from the current year to the longer term. According to the ECB, those expectations are reported as point forecasts and also as probability distributions.

A probability distribution is still a forecast, but it is an honest one about its own limits. The ECB says such distributions give a quantitative assessment of risk and uncertainty. They tell you how likely each outcome is within a range the forecaster has already chosen. Scenario work starts from a different assumption: the range itself may be wrong, and the useful question is which combinations of events would hurt you. A scenario is a coherent account of several drivers moving together. It is not a band drawn around a single variable.

Stress testing is close to scenario work, though its purpose is narrower. The Bank of England describes its exercises as a check on whether individual firms and whole sectors could cope with the worst. The Bank Capital Stress Test applies to the largest and most systemically important UK banks and building societies. System-wide exploratory scenarios trace an adverse shock across the whole financial system or a particular market. A separate Cyber & Operational Resilience Stress Test looks at operational disruption, such as cyber attacks.

The Bank's page adds a practical point for everyone else. Banks and building societies outside the Bank Capital Stress Test can use the published scenarios for their own stress testing. Life and general insurers are also stress tested, to assess the resilience of the insurance sector. The page does not say how closely any published scenario fits a particular firm's exposures. That fit is a judgement the firm must make and document for itself, and it is the step most often skipped.

Year the ECB's Survey of Professional Forecasters began, per the ECB
1999
Provisional year of the survey's next release, 5 February 2027, per the ECB
2027

Who it affects

The most direct audience is UK banks and building societies. The largest and most systemically important sit inside the Bank Capital Stress Test, and the rest work from the published scenarios. UK life and general insurers face stress testing aimed at sector resilience. This guide's audience also includes UK central counterparties, but the Bank's stress-testing page does not describe any test aimed at them. Confirm their position with the relevant supervisor before relying on it.

Anyone planning against the euro area economy has a different use for the survey. Its inflation, growth and unemployment expectations, together with their probability distributions, show how much uncertainty professional forecasters attach to the outlook. A treasurer or planning director can read the spread as well as the central estimate. The ECB lists its next release as provisional for 5 February 2027, so a planning calendar should allow for that date.

Global banking supervision is the third group. According to the Bank for International Settlements, the Basel Committee on Banking Supervision aims to deepen understanding of key supervisory issues and raise the quality of banking supervision worldwide. The BIS also compiles statistics with central banks. Those statistics inform analysis of monetary and financial stability, international spillovers and global liquidity, and they are useful for any leader whose funding depends on conditions beyond one country's borders.

What leaders get wrong

Five errors come up again and again. In our experience each begins with treating the two methods as if they answer the same question, which they do not. The list below is the set of checks we would run on any planning paper before it reaches a board. If a paper fails one of them, the fix is usually small, but it is worth making before the numbers are discussed rather than after.

  • They ask a point forecast to answer a tail-risk question. A central case says nothing about how bad the bad case would be, and a single number will never show it.
  • They write three scenarios that differ only in size. If the drivers do not change from one story to the next, the set is a sensitivity table with better prose.
  • They attach probabilities to scenarios without grounds. A percentage on a scenario looks like a forecast. Unless the team can defend the figure, it misleads the people who read it.
  • They build scenarios that name no decision. A story that changes nothing about what the company does on Monday is an exercise, not a planning tool.
  • They leave out signposts. Without observable early warnings, nobody can tell which story is unfolding, and the set goes back on the shelf.

What to do about it

Start with the decision, not the model. Write down what you will commit to, what it would cost to reverse, and what would make you change course. If one expected value drives the choice, such as a price you can revisit next quarter, a point forecast with a sensible range is usually enough. If the choice is hard to reverse and one direction of surprise would hurt, use a scenario set. Put this test on one page before any modelling begins, and let it decide which tool the team uses.

None of the sources we checked describes how to build a scenario set. The Bank of England and ECB pages explain what their exercises are for and how their survey data is organised. They do not describe how a scenario is constructed, and the BIS pages we reviewed do not cover it either. What follows is our judgement, not a sourced standard. Anyone who has built scenarios before should test it against their own experience before using it for a material decision.

Build three scenarios from two or three drivers that matter to the decision and that you cannot control. Make each story internally consistent: the drivers in a given scenario should move together for a reason you can state in one sentence. Give each scenario a name that describes the situation rather than the outcome. Then, for each one, write down the decision it would force, what that decision would cost if it turned out wrong, and the earliest date by which you would know the story was unfolding.

Take a hypothetical manufacturer with three plants that is considering a fourth. Its drivers are input costs, export orders and the reliability of one main supplier. The base story is steady demand, with input costs inside the band set by its supply contracts. The squeeze story has input costs rising for several quarters while export orders fall. The disruption story is a supplier or plant outage, with a cyber incident as one possible trigger. The fourth plant is a sensible commitment in the first story, a dangerous one in the second, and something to delay in the third.

A signpost is an observable indicator with a threshold written down in advance. Set the threshold before you need it, because otherwise the argument about what the number means will happen after the event, when everyone already has a preferred story. For the three stories above, a usable set would look like the list below. Each signpost is a number the finance team can report and the board can check without interpreting a narrative.

  • Steady: order intake from the two largest customers stays at or above the level of the last full year, and supplier quotes move inside the contract band.
  • Squeeze: export orders fall for two consecutive quarters, and the main input price passes the level written into the supply contract.
  • Disruption: a key supplier's delivery times lengthen beyond the agreed limit, or the business continuity plan's activation trigger is met.

Review the signposts at a fixed interval, for example at every quarterly board meeting. A signpost that has fired should trigger a decision, not a discussion about whether it counts. If none has fired, the steady story stands by default, and the fourth plant can proceed with the safeguard the disruption story requires, such as a second qualified supplier. Record each review in the board pack so that the reasoning can be checked later, by the board and by the auditors.

Avoid giving the scenarios probabilities unless you can defend them. The ECB's distributions are a quantified statement of uncertainty from a survey that has run since 1999, and that history gives them a backing that a planning team's scenarios seldom have. A sensible alternative is to rank the scenarios by the damage each would do and by how soon you would see it coming. For most boards that ranking is more useful than a percentage, and it does not pretend to know more than the team does.

Newsletter

The Briefing

One email a week: the leaders we profiled, the ideas worth stealing, and nothing else.

Weekly, free, one click to leave. Read our privacy policy.