RDP 2025-09: Forecasts of Period-average Exchange Rates: Insights from Real-time Daily Data Appendix D: Real-time Forecast Accuracy for Other Exchange Rates

Martin McCarthy and Stephen Snudden

December 2025

D.1 No-change forecasts

Figure D1: Distribution of Forecast Performance of End-of-month No-change versus Month-average No-change Forecasts
Figure D1: Distribution of Forecast Performance of End-of-month No-change versus Month-average No-change Forecasts

Notes: Plot shows quantiles for 83 countries. RMSFE ratios less than 1 (indicated by the dashed line) improve upon the month-average no-change. Success ratios greater than 0.5 (indicated by the dashed line) are improvements upon random chance.

Sources: Authors' calculations; Eikon; International Monetary Fund.

D.2 Model-based forecasts

Table D1: Median Performance of Forecasts for Month-average REER
Forecast Model inputs Horizon (months)
1 3 6 12 24 36
RMSFE ratio
Recursive Month-average 1.00 0.99 0.99 1.00 1.01 1.02
Recursive End-of-month 0.96 0.98 0.98 1.01 1.02 1.01
Recursive Bottom-up 0.95 0.97 0.99 1.03 1.04 1.04
Direct Month-average 1.00 0.98 0.99 0.99 1.03 1.09
Direct End-of-month 0.96 0.97 0.98 0.99 1.04 1.10
Direct UMIDAS 0.97 0.97 0.98 0.99 1.03 1.09
Success ratio
Recursive Month-average 0.51 0.51 0.51 0.53 0.56 0.54
Recursive End-of-month 0.68 0.57 0.54 0.54 0.57 0.55
Recursive Bottom-up 0.67 0.57 0.53 0.52 0.56 0.53
Direct Month-average 0.51 0.51 0.52 0.53 0.56 0.54
Direct End-of-month 0.68 0.57 0.54 0.53 0.56 0.53
Direct UMIDAS 0.69 0.58 0.54 0.54 0.56 0.53
Notes: Reports the median result across countries relative to the month-average no-change forecast. RMSFE ratios less than 1 improve upon the month-average no-change. Success ratios greater than 0.5 are improvements upon random chance.
Table D2: Median Performance of Forecasts for Month-average Bilateral NER
Forecast Model inputs Horizon (months)
1 3 6 12 24 36
RMSFE ratio
Recursive Month-average 1.00 1.01 1.01 1.02 1.05 1.07
Recursive End-of-month 0.77 0.95 0.99 1.02 1.03 1.07
Recursive Bottom-up 0.76 0.95 0.98 1.01 1.02 1.06
Direct Month-average 1.00 1.01 1.01 1.02 1.08 1.32
Direct End-of-month 0.77 0.95 0.98 1.00 1.07 1.31
Direct UMIDAS 0.76 0.95 0.98 1.00 1.07 1.31
Success ratio
Recursive Month-average 0.51 0.45 0.47 0.50 0.57 0.57
Recursive End-of-month 0.70 0.56 0.53 0.51 0.56 0.58
Recursive Bottom-up 0.69 0.57 0.52 0.51 0.52 0.55
Direct Month-average 0.51 0.45 0.48 0.51 0.57 0.56
Direct End-of-month 0.69 0.57 0.52 0.53 0.57 0.57
Direct UMIDAS 0.69 0.58 0.53 0.53 0.58 0.57
Notes: Reports the median result across countries relative to the month-average no-change forecast. RMSFE ratios less than 1 improve upon the month-average no-change. Success ratios greater than 0.5 are improvements upon random chance.
Table D3: Median Performance of Forecasts for Month-average NEER
Forecast Model inputs Horizon (months)
1 3 6 12 24 36
RMSFE ratio
Recursive Month-average 1.00 1.01 1.02 1.05 1.12 1.15
Recursive End-of-month 0.95 0.99 1.01 1.04 1.10 1.11
Recursive Bottom-up 0.95 0.99 1.04 1.07 1.13 1.14
Direct Month-average 1.00 1.01 1.03 1.06 1.14 1.20
Direct End-of-month 0.95 0.99 1.01 1.06 1.14 1.21
Direct UMIDAS 0.95 0.98 1.01 1.06 1.14 1.19
Success ratio
Recursive Month-average 0.50 0.49 0.49 0.50 0.51 0.49
Recursive End-of-month 0.69 0.57 0.53 0.53 0.51 0.48
Recursive Bottom-up 0.70 0.56 0.52 0.52 0.50 0.48
Direct Month-average 0.50 0.50 0.50 0.52 0.51 0.47
Direct End-of-month 0.69 0.57 0.53 0.52 0.51 0.47
Direct UMIDAS 0.71 0.57 0.53 0.52 0.51 0.48
Notes: Reports the median result across countries relative to the month-average no-change forecast. RMSFE ratios less than 1 improve upon the month-average no-change. Success ratios greater than 0.5 are improvements upon random chance.
Figure D2: Accuracy of One-month-ahead Forecasts for Bilateral RER Relative to End-of-month No-change Benchmark
Figure D2: Accuracy of One-month-ahead Forecasts for Bilateral RER Relative to End-of-month No-change Benchmark

Notes: The box and whisker plots show the distribution of RMSFE ratios and SRs across countries. ‘End-of-month’ uses the end-of-month forecast as the forecast of the monthly average. ‘Recursive Bottom-up’ ex post averages daily forecasts. ‘Direct UMIDAS’ forecasts use the end-of-month observation only. Outliers have been omitted. RMSFE ratios less than 1 (indicated by the dashed line) improve upon the month-average no-change. Success ratios greater than 0.5 (indicated by the dashed line) are improvements upon random chance.

Sources: Authors' calculations; Eikon; International Monetary Fund; World Bank.

Figure D3: Accuracy of One-month-ahead Forecasts for Bilateral NER Relative to End-of-month No-change Benchmark
Figure D3: Accuracy of One-month-ahead Forecasts for Bilateral NER Relative to End-of-month No-change Benchmark

Notes: The box and whisker plots show the distribution of RMSFE ratios and SRs across countries. ‘End-of-month’ uses the end-of-month forecast as the forecast of the monthly average. ‘Recursive Bottom-up’ ex post averages daily forecasts. ‘Direct UMIDAS’ forecasts use the end-of-month observation only. Outliers have been omitted. RMSFE ratios less than 1 (indicated by the dashed line) improve upon the month-average no-change. Success ratios greater than 0.5 (indicated by the dashed line) are improvements upon random chance.

Sources: Authors' calculations; Eikon; International Monetary Fund.

Figure D4: Accuracy of One-month-ahead Forecasts for NEER Relative to End-of-month No-change Benchmark
Figure D4: Accuracy of One-month-ahead Forecasts for NEER Relative to End-of-month No-change Benchmark

Notes: The box and whisker plots show the distribution of RMSFE ratios and SRs across countries. ‘End-of-month’ uses the end-of-month forecast as the forecast of the monthly average. ‘Recursive Bottom-up’ ex post averages daily forecasts. ‘Direct UMIDAS’ forecasts use the end-of-month observation only. Outliers have been omitted. RMSFE ratios less than 1 (indicated by the dashed line) improve upon the month-average no-change. Success ratios greater than 0.5 (indicated by the dashed line) are improvements upon random chance.

Sources: Authors' calculations; Eikon; International Monetary Fund.