Revisiting Professional Forecasters’ Past Performance and the Outlook for 2026

At times, it feels as though economists make forecasts only because it’s a job requirement rather than because our crystal balls reveal signs of growth or decline. The U.S. economy has consistently surprised to the upside, first strongly rebounding from the COVID-19 pandemic and then showing resiliency in the face of high inflation and rising interest rates. Economic and political uncertainty was also an ongoing theme of 2025. Despite these challenges, the forecasts we described in a December 2024 blog post turned out to be quite accurate based on the data available today. (Some data have been delayed because of the government shutdown).

The first row of the table below reports the Blue Chip projections for the current year (2025) for four key economic indicators: real gross domestic product (GDP) growth, consumer price index (CPI) inflation, the unemployment rate, and the 10-year Treasury yield. These projections can be viewed as very reliable estimates of what the official statistics will show, as much of the data for 2025 are already available to forecasters.

One year ago, Blue Chip forecasters predicted real GDP growth of 2.1% for 2025, with recent estimates indicating growth will likely be around 1.9%. The government shutdown likely will shift real GDP growth from late 2025 to early 2026. Forecasts for the unemployment rate and 10-year Treasury yields were quite accurate, with actual values of 4.3% for both indicators, near the “consensus”—the average among all forecasters. The consensus predicted a greater deceleration in inflation than has materialized, however; recent estimates indicate CPI inflation of 2.8% in 2025, which is right at the upper end of estimates made one year ago.

Forecasters now expect GDP growth to continue at a similar rate in 2026, with both the unemployment rate and CPI inflation to tick up. Interest rates are forecast to be slightly lower than in 2025. Despite forecasters' accuracy in 2025, their historical track record indicates these economic forecasts are rarely fully correct. So how likely will economic performance in 2026 be in line with professional forecasts? And if they are wrong, as they have been in the past, how wrong will they be?

Forecasts of 2026 Economic Activity

We look at the one-year-ahead projections reported in the December 2025 Blue Chip survey of about 50 professional economic forecasters. Each individual forecaster’s projections are reported, but it is the consensus that garners the most attention.

The middle three rows of the table below report the forecasts for 2026. The consensus outlook for real GDP growth in 2026 is 1.9%, but it seems that forecasters have considerable disagreement on the outlook for the year ahead, as reflected in the following:

• The average of the top 10 GDP growth rate forecasts is 2.5%, while the average of the bottom 10 forecasts is 1.2%.
• Some forecasters see the unemployment rate rising while others predict it will fall.
• Similarly, some forecasters predict CPI inflation will accelerate, while others predict it will slow.

How Likely Will 2026 Unfold as Expected?

The bottom three rows of the table report on forecast performance over the period 1993 to 2024. The first measure is the percentage of years in which the actual data fell within the range of the bottom 10 and top 10 average forecasts; actual values of real GDP growth, unemployment, and interest rates were within this range of forecasts less than half the time. Forecasts of CPI inflation have been a little more accurate (56%). Based on historical performance, it is essentially a coin toss as to whether observed real GDP growth in 2026 will fall within the range of 1.2% to 2.5%.

This may seem like very poor performance, but the mean absolute forecast error (MAFE) suggests otherwise. The MAFE looks at the historical performance of the forecasts to predict the range in which the actual values should fall. The MAFE for the real GDP growth forecast is 1.0 percentage point, which suggests that if the consensus forecast of real GDP growth is incorrect, the actual real GDP growth value will typically fall within a band of 1% over and 1% under the consensus forecast.

So how is this good? It’s important to note that real GDP growth typically deviates from its annual average by 1.75 percentage points. For example, if the economy were to increase 2.0% in a given year, actual growth in the following year would likely be between 0.25% and 3.75% based on historical volatility. In contrast, an MAFE of 1.0 percentage point and the current consensus forecast of 1.9% indicate 2026 growth will likely be in a range of 0.9% to 2.9%, which is a narrower range of probable outcomes. This eliminates a fair bit of uncertainty. The consensus forecast and MAFE also tighten the range of likely outcomes for other variables in a similar manner.

The final measure, mean forecast error (MFE), indicates whether the consensus forecast is biased—that is, whether it has historically missed in a certain direction. For most indicators, the MFE is near zero, indicating little to no forecast bias (over the past several decades) toward either overshooting or undershooting. However, the MFE on the 10-year Treasury yield forecast is -0.4 percentage points, indicating that forecasters have typically predicted higher interest rates by an average of 40 basis points.

The 2026 Economic Outlook

In summary, the consensus outlook for economic growth in 2026 is very similar to the outlook one year ago, although this projection is much more uncertain given the range of forecasts. A similar story unfolds for the unemployment rate, inflation and interest rates. The lack of government statistics about the economy over the past few months could be one reason for the greater degree of disagreement among forecasters. However, forecasters express disagreement about other key drivers of economic growth. For example, on the impact of artificial intelligence (AI) on the economy, 57% of forecasters think AI is already boosting productivity while 43% think it is not. The truth may lie somewhere in the middle, which is why the average forecast tends to be more accurate than individual forecasters over many years.