Forecasters get more overconfident the further into the future they look
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Aïda Amer/Axios
Economic forecasters are pretty accurate when they project where things are headed in the near term — but are overconfident in their ability to see around the corner to what the world will look like a year or two out.
Why it matters: This finding, from a New York Fed piece out Tuesday based on decades of professional forecasters' predictions, has philosophical implications for anyone who finds themselves needing to predict the future.
- Even among the people paid to offer rigorous, methodical economic projections, it is easy to be overconfident by underestimating how much underlying conditions can change over long time periods.
Zoom in: Economist Marco Del Negro analyzed the Philadelphia Fed's Survey of Professional Forecasters from 1982 to 2022. He looked at the forecasters' confidence level in their outlook for growth and inflation — as measured by the odds they placed on economic results significantly different from their forecast.
- In other words, he looked not just at how accurate or inaccurate the forecasters were but at whether they had appropriately large error bands around their forecast.
- It is a test not of whether the forecasters were right or wrong but of whether they were overconfident.
State of play: For time periods of less than a year, the forecasters were actually underconfident. That is, they had wider error bands around their projections than were ultimately justified by events.
- But for periods of one to two years, they were significantly overconfident, putting too-low odds on outcomes significantly diverging from their forecast.
Between the lines: This finding jibes with a concept we've been thinking about in other realms, whether it's forecasting financial markets, geopolitics or anything else.
- Over short time frames, it is possible to have a pretty good handle on the range of possibilities. Over longer time frames, the world can evolve in ways that are more radically different from what our puny brains can imagine.
- In the political realm, for example, it is one thing to place odds on either former President Trump or Kamala Harris winning the November election. But if you think you can reasonably predict what the political landscape will look like in 2028, you're probably wrong (ask anyone who in 2015 was certain the next president would be either Jeb Bush or Hillary Clinton).
Yes, but: This thinking can result in a nobody-knows-nothin' kind of nihilism. The reality is that in many pursuits, you have to forecast something to have some baseline.
- Companies have to project what demand will look like years down the road when deciding to build a factory. Investors must decide how to allocate their retirement portfolios.
The bottom line: The lesson of this finding isn't that we must never write down what we expect to happen far in the future. It's that we must do so from a position of intellectual modesty, and be ready to update assumptions as the facts change.
