Here's a mystery: if we know that macroeconomic forecasts are inaccurate, that equity analysts underperform naive baselines, and that political forecasters barely beat chance, why do organizations keep paying for forecasts?
The answer reveals something about how institutions actually work—and why the problem is not going away.
The Structural Reasons
First: Organizations need plans, and plans need numbers.
A corporation cannot write its budget without revenue forecasts. A government cannot write policy without GDP projections. An investor cannot allocate capital without some view of returns.
The numbers must come from somewhere. When an economist offers a forecast, it fills that need—whether or not the forecast is actually accurate.
The number matters more than its quality.
This is not because organizations are stupid. It is structural. Planning requires inputs. Forecasts are the standard input for planning. The fact that forecasts are bad is secondary to the organizational need for numerical inputs.
Second: Forecasters face no consequences for being wrong.
A macroeconomist's compensation is paid whether they forecast correctly or not. When they are wrong, there is no clawback. There is no penalty. There is, often, a promotion or a new position.
This is completely different from domains where expertise actually exists. An oncologist whose patients die when they should live faces consequences: loss of patients, loss of reputation, loss of licensure.
A macroeconomist whose forecasts are completely wrong faces consequences: none, usually. Maybe a mild reputational ding, but nothing structural.
When failure is not costly, confidence is not checked.
Third: The "I Was Almost Right" defense is unfalsifiable.
After a forecast misses, the forecaster says: "I was directionally correct—I predicted slowdown and we got slowdown, just at a different magnitude."
Or: "I identified the key risks, even if the timing was wrong."
Or: "That outcome was within my stated range of possibility."
Every one of these is unfalsifiable. Any actual outcome can be retrofitted into "almost right."
Imagine if a surgeon said the same thing: "I operated on the wrong leg, but I was directionally correct—I was moving toward surgical intervention." The surgeon would lose their license.
Imagine if an engineer said: "The bridge collapsed, but I was directionally correct about load capacity." The engineer would face liability.
But a forecaster says these things and faces nothing.
The unfalsifiability of the defense lets bad forecasters survive indefinitely. They can never be proven catastrophically wrong because "almost right" accommodates almost everything.
Fourth: The industry sells confidence, and confidence is what clients buy.
A financial advisor who says "I don't really know what the market will do" doesn't get clients. The advisor who says "I expect growth of 2.3% over the next three years" gets clients.
The second forecast is more likely to be wrong. It is also more likely to be purchased.
Organizations and individuals want confident numbers. They will pay for them. So forecasters produce them.
This creates a market dynamic: the worst forecasters are the most confident ones (because confidence is selected for), and the most confident ones are the ones selling forecasts.
Fifth: Institutions unconsciously need the narrative.
Beyond the explicit need for numbers, there is an implicit need for narrative. A story that makes the present understandable and the future predictable.
A forecaster who says "I don't know, the future is uncertain" provides no narrative comfort. A forecaster who says "Here's why the market will move, here's the sequence of events, here's the likely outcome" provides narrative.
Organizations buy the narrative as much as the number. The narrative makes leadership feel like they understand what is happening and what will happen. This is comforting, even if false.
Why This Problem Is Structural and Unfixable
The forecasting industry persists despite documented inaccuracy because:
- Organizations structurally need plans with numerical inputs.
- Forecasters face no structural penalty for being wrong.
- Forecasts are unfalsifiable once stated ("almost right" defense).
- Clients demand confidence, not accuracy.
- Organizations desire narrative comfort as much as accuracy.
None of these problems is going away. You cannot eliminate the need for organizational planning. You cannot eliminate the preference for confident predictions over humble uncertainty. You cannot create penalties for being wrong in Extremistan domains without destroying the institutions themselves.
The "scandal of prediction" (Taleb's phrase) is not a scandal that will be resolved. It is a structural feature of how organizations operating in Extremistan must operate.
What to Do About It
Knowing this, the rational response is not to demand better forecasts. It is to:
-
Recognize forecasts for what they are: narrative devices that fill a structural need, not predictions with meaningful accuracy.
-
Demand error bars: a forecast without a confidence interval is not information. Make forecasters specify the width of their uncertainty.
-
Measure against baselines: before hiring a forecaster, ask for their track record against naive baselines. If they can't show it, they're not a forecaster; they're a storyteller.
-
Plan for range, not point: instead of building strategy around a single forecast number, build strategy that works across a wide range of outcomes.
-
Reduce reliance on forecasts where possible: use rules, heuristics, and structural measures instead of predictions. "If leverage exceeds X, reduce it" is more robust than "leverage is safe at current levels based on forecasts."
-
Accept that some uncertainty cannot be reduced: the best response to irreducible uncertainty is not more sophistication but more humility—smaller positions, more redundancy, more flexibility.
The forecasting industry will not disappear. Organizations will continue paying for forecasts. Forecasters will continue making them.
But understanding why this persists despite documented inaccuracy lets you listen to forecasts with appropriate skepticism—and make better decisions as a result.