CK Forecasting

Forecasting centre

Growth, inflation and unemployment for 29 economies this year, next year and the year after, each with its range and its track record. We publish a forecast only where it beat a no-change forecast in a backtest on the IMF’s past editions.

Baseline forecast CK’s central expectation from what is known now, not a scenario. For “what if” paths, such as a closed strait or a tariff, use the simulator: its results are illustrative scenarios and are never scored or shown as forecasts.

Track record

Mean absolute error of each forecast made from past October editions, pooled across countries, in percentage points. Lower is better. For next year, CK’s errors were 30–44% smaller than naive’s and within 7% of the IMF’s, whose forecast is their main input: a transparent, scored forecast rather than a better one than the IMF’s.

Real GDP growth

Pooled backtest errors of real gdp growth forecasts
Ahead	CK	Naive	IMF	In range
1 yr	0.96	3.02	0.92	76%
2 yr	1.69	3.04	1.69	83%
3 yr	1.97	3.00	2.01	73%

Next year: errors 44% smaller than naive’s, about the same as the IMF’s; 369 forecasts across 29 economies.

Inflation

Pooled backtest errors of inflation forecasts
Ahead	CK	Naive	IMF	In range
1 yr	0.30	1.73	0.30	80%
2 yr	1.55	2.39	1.46	80%
3 yr	2.25	2.78	2.11	68%

Next year: errors 35% smaller than naive’s, 7% larger than the IMF’s; 357 forecasts across 29 economies.

Unemployment rate

Pooled backtest errors of unemployment rate forecasts
Ahead	CK	Naive	IMF	In range
1 yr	0.36	0.61	0.35	76%
2 yr	0.68	0.96	0.64	80%
3 yr	0.94	1.19	0.94	73%

Next year: errors 30% smaller than naive’s, 6% larger than the IMF’s; 329 forecasts across 26 economies.

The forecasts

CK forecasts for 2026 (this year) and 2027 (next year), in percent; each country’s page has 2028, the 80% ranges, the IMF’s figures and the scorecards. A dash is a forecast we withhold.

CK forecasts by country for 2026 and 2027
Economy	GDP growth		Inflation		Unemployment
Economy	2026	2027	2026	2027	2026	2027
United States	2.3	2.3	3.6	2.8	4.2	4.0
Mexico	1.5	1.8	3.8	3.7	2.8	2.8
Canada	1.1	1.7	2.8	2.4	6.5	6.4
China	4.6	4.1	1.0	– (Withheld at this horizon: it didn’t pass the backtest)	– (Withheld: too few past editions to score it)	– (Withheld: too few past editions to score it)
Taiwan	10.3	4.3	2.0	1.7	3.4	3.4
Vietnam	8.3	7.5	4.7	5.0	– (Withheld: in the backtest it was no better than assuming no change)	– (Withheld: in the backtest it was no better than assuming no change)
Germany	1.3	1.1	2.7	3.0	4.0	3.7
Japan	0.7	0.7	2.0	2.5	2.5	2.5
South Korea	3.3	2.8	2.7	2.3	2.8	– (Withheld at this horizon: it didn’t pass the backtest)
United Kingdom	1.3	1.2	3.3	2.8	5.3	5.3
Thailand	2.1	2.2	2.1	1.7	– (Withheld: in the backtest it was no better than assuming no change)	– (Withheld: in the backtest it was no better than assuming no change)
India	7.0	7.0	5.0	4.3	– (Not covered by the IMF)	– (Not covered by the IMF)
Switzerland	1.9	1.2	0.7	0.8	3.0	3.0
Netherlands	1.5	1.3	2.9	2.8	4.0	4.1
Italy	0.8	0.6	3.3	2.7	5.6	5.8
France	0.4	0.7	2.4	2.1	– (Withheld: in the backtest it was no better than assuming no change)	– (Withheld: in the backtest it was no better than assuming no change)
Malaysia	5.2	4.6	2.2	1.9	3.0	3.0
Ireland	−0.7	3.8	3.4	2.7	5.0	5.0
Brazil	2.2	1.9	4.4	4.3	6.2	6.7
Singapore	5.0	3.1	2.3	2.1	2.1	2.1
Belgium	0.8	1.1	3.8	2.8	6.2	6.1
Australia	1.9	1.7	3.9	3.1	4.4	4.7
Argentina	2.1	3.4	– (Withheld: too few past editions to score it)	– (Withheld: too few past editions to score it)	7.6	7.1
Indonesia	5.1	5.1	3.3	2.9	4.7	4.7
Russia	1.0	1.0	6.1	5.2	2.2	2.4
Saudi Arabia	−2.9	7.7	1.9	2.2	– (Withheld: too few past editions to score it)	– (Withheld: too few past editions to score it)
South Africa	1.0	1.3	4.4	4.0	– (Withheld: in the backtest it was no better than assuming no change)	– (Withheld: in the backtest it was no better than assuming no change)
Türkiye	2.8	3.7	31.2	26.5	8.2	8.3
European Union	– (Withheld: too few past editions to score it)	– (Withheld: too few past editions to score it)	– (Withheld: too few past editions to score it)	– (Withheld: too few past editions to score it)	– (Not covered by the IMF)	– (Not covered by the IMF)

How these are made

A weighted average of two statistical models (ARIMA and exponential smoothing) and the IMF’s own forecast, with weights learned from past editions; the IMF’s forecast carries most of the weight, so ours stay close to it. The range is the spread of past errors. The methodology

Data vintage

IMF World Economic Outlook, October 2026, published 2026-10-08. Reported to 2025; forecasts for 2026–2028. Method ck-ensemble-v1.

What isn’t here

Market or policy-rate calls, and scenarios. A forecast is withheld, with its scorecard still public, where the backtest didn’t beat naive or was too short to judge.