Distributional Forecasting
Pair every baseline forecast with explicit upside and downside risks
- Difficulty
- Advanced
- Time to result
- ~weeks to results
- Steps
- 6
- Confidence
- 98%
Distributional Forecasting replaces the false precision of a lone point estimate with a baseline and an explicit map of surrounding risks. Begin with the most defensible central outcome for a defined variable and horizon, then identify the forces capable of producing stronger or weaker results. Assess whether the distribution is balanced or tilted toward one side, and communicate that skew in plain language. Meyer gives the example of forecasting three percent growth while explaining that two percent is more probable than four percent. The method preserves the practical need for a forecast without concealing uncertainty, judgment, or model limitations. It also improves updating: when an assumption changes, analysts can see which scenario has become more likely rather than simply replacing one unexplained number with another.
Origin
Extracted from Masters in Business, where Michelle Meyer proposed formalizing the risk distribution surrounding Wall Street economic forecasts.
Core principles
- 01A point estimate is incomplete without its surrounding risks
- 02State the baseline without pretending it is certain
- 03Distinguish upside scenarios from downside scenarios
- 04Communicate skew as well as the central forecast
- 05Update judgments when data changes
How to run it
- 1
Define the decision horizon
Specify the variable, measurement period, and decision the forecast will inform. A distribution is only meaningful when every scenario refers to the same target.
Pro tip Write the forecast as a dated question before analyzing it.
Watch out Do not mix short-term data surprises with a medium-term forecast.
- 2
Build the baseline
Use available data, established relationships, and transparent judgment to identify the most likely central outcome.
Pro tip Document the few assumptions doing most of the analytical work.
Watch out Do not describe the baseline as guaranteed or objectively complete.
- 3
Map upside risks
Identify developments that could produce a materially better or higher outcome than the baseline. Explain the mechanism connecting each development to the forecast.
Pro tip Include both gradual improvements and discrete positive shocks.
Watch out A wish list is not an upside scenario unless the causal path is credible.
- 4
Map downside risks
Identify developments that could produce a materially worse or lower outcome. Consider both familiar historical risks and exceptional scenarios that conventional models may underweight.
Pro tip Ask what evidence you would expect to observe before each downside scenario materializes.
Watch out Do not dismiss an outcome solely because it is uncomfortable or historically rare.
- 5
Assess the skew
Compare the relative probability and impact of outcomes on each side of the baseline. State whether risks are balanced, tilted upward, or tilted downward.
Pro tip Use ranges or qualitative probability bands when numerical precision is unsupported.
Watch out Avoid invented probabilities that imply more knowledge than the evidence provides.
- 6
Communicate and update
Present the baseline, risk distribution, assumptions, and update triggers together. Revise the distribution as new information arrives.
Pro tip Explain what changed in the distribution before announcing a revised point estimate.
Watch out Do not let daily market volatility automatically redefine a medium-term view.
In the wild
An economist's baseline calls for three percent growth. Instead of publishing only that number, the economist explains that two percent growth is more probable than four percent growth, making the downside skew visible to clients.
→ Decision-makers receive both a usable central forecast and a clearer picture of asymmetric risk.
A company expects ten percent sales growth and builds its hiring plan around that baseline. It also models an upside case driven by a major contract and a downside case caused by customer churn. Because the downside is judged more probable than the upside, the company stages hiring in two rounds and defines revenue triggers for the second round.
→ The company retains the capacity to grow while limiting the cost of a forecast miss.
Common mistakes
Publishing only the point estimate
A central number without upside and downside context hides the uncertainty that matters to decisions.
Treating history as destiny
Historical relationships are useful inputs, but shocks and turning points can prevent a simple return to the past.
Reacting to every market move
Allowing daily volatility to dictate the forecast can obscure the underlying medium-term economic signal.
Is it for you?
Best for
It is best for economic, investment, operational, and strategic decisions where several outcomes remain plausible.
Not ideal for
It is not ideal when a precise deterministic answer is available or when probabilities would be fabricated without evidence.
From the transcript
“I always try to represent both sides of the argument by saying, you know, here's my baseline view, but here's what can go right, and…”
“I think it'd be very helpful to have at all times some sort of modal distribution.”
“rather than saying our view is for three percent growth, you say our view is the economy will realize three percent growth, but the risks…”
From the episode
Bank Of America's Michelle Meyer: Masters in Business (Audio)
Michelle Meyer