IMF WEO Forecast Tracker

How accurate are IMF forecasts for each country? This table scores one-year-ahead forecasts against actual outcomes. A high % Over means the IMF consistently overestimated; the Bias z-score shows how unusual that pattern is relative to other countries. Volatile economies naturally have higher errors (MAE), so the bias score helps distinguish genuine forecast bias from inherent unpredictability. Click a column to sort, or click a row to view that country's chart.

% Over: share of one-year-ahead forecasts that exceeded the actual outcome. Bias: z-score of a country's mean forecast error relative to all countries (positive = systematic overforecast). MAE: mean absolute error in percentage points. Download full scores (CSV)

• Greece — Government Debt: The IMF's own Independent Evaluation Office criticized the austerity-era debt forecasts. The dot cloud shows how projections kept underestimating Greek debt as the crisis deepened. In the scores table above, Greece's GDP growth shows 60% of forecasts were too high since 2010.
• Japan — CPI Inflation: For years, forecasts predicted inflation would return to 2%—and it never did. The cloud fans out above the flatline, until post-COVID inflation finally arrived and caught forecasters going the other way. Over the full sample, 64% of one-year-ahead inflation forecasts were too high.
• Ethiopia — Real GDP Growth: Ethiopia's state-led development strategy produced consistently strong growth that exceeded IMF forecasts. Since 2010, only 29% of one-year-ahead forecasts were too high—one of the lowest rates among emerging economies (bias score: −1.07).
• Turkey — Real GDP Growth: The IMF has been consistently too pessimistic on Turkish growth. Since 2010, only 13% of one-year-ahead forecasts were too high, giving Turkey one of the strongest negative bias scores (−1.38) among major economies.

The World Economic Outlook is the IMF's flagship forecast publication, covering GDP growth, inflation, fiscal balances, current account balances, and more for nearly every country. It has been published twice yearly (April and October) since 1980.

Each edition contains forecasts extending 5–6 years forward. As the forecast horizon shrinks, estimates generally converge toward the actual outcome—but not always. The IMF's Independent Evaluation Office found systematic optimism bias during recessions. Research by Loungani et al. (2018) showed the “record of failure to predict recessions is virtually unblemished.” A 2025 IMF working paper found forecasts overreact to good news while responding more cautiously to bad news.

By overlaying all editions, you can see these patterns directly—the cloud of dots reveals where forecasts converged or diverged, and how the latest estimates compare to historical predictions.

• IEO: IMF Forecasts — Process, Quality, and Country Perspectives (2014)
• An, Jalles & Loungani: How Well Do Economists Forecast Recessions? (2018)
• Celasun et al.: An Evaluation of WEO Growth Forecasts 2004–17 (2021)
• Evidence of Asymmetry in WEO Forecasts (2025)
• Hadzi-Vaskov et al.: Patterns in IMF Growth Forecast Revisions (2021)
• IMF API Guide — Working with WEO Data
• Source code on GitHub