Live calls and the scored record

Forecasts

Every inflation and growth forecast we publish, and how each one scored against the outcome and against the Wall Street consensus for the same month. Forecasts are published before the release, never revised afterwards, and no scheduled month is omitted.
Current calls next publication windows
US PCE — July 2026 not yet published
Headline
publishes ahead of the BEA release, 26 Aug 2026
Last call: Jun 2026, scored below PCE series →
US Real GDP — 2026Q3 not yet published
Annualised, SAAR
publishes ahead of the BEA advance release, 29 Oct 2026
Last call: 2026Q2, scored below GDP series →
US CPI — July 2026 not yet published
Headline
publishes ahead of the BLS release, 12 Aug 2026
Last call: Jun 2026, scored below CPI series →
Record at a glance
Inflation forecasts scored
26
CPI and PCE, headline and core
Our mean abs. error
0.104
percentage points, MoM
Consensus mean abs. error
0.096
same 26 forecasts
Across the full inflation record we are marginally behind the consensus. We lead it on core CPI and trail it on headline CPI, headline PCE and core PCE. Amber marks the smaller error in each row.
US CPI 6 prints · Jan–Jun 2026
Reference monthHeadline CPI, MoM %Core CPI, MoM %
MERCons.ActualOur errCons. errMERCons.ActualOur errCons. err
Jan 20260.330.300.170.160.130.210.300.300.090.00
Feb 20260.260.300.270.010.030.260.300.220.040.08
Mar 20260.721.000.870.150.130.240.300.200.040.10
Apr 20260.590.600.640.050.040.310.500.380.070.12
May 20260.500.500.470.030.030.300.300.210.090.09
Jun 2026−0.02−0.10−0.420.400.320.200.20−0.020.220.22
Mean absolute error0.1330.1130.0920.102
US PCE 7 prints · Dec 2025–Jun 2026
Reference monthHeadline PCE, MoM %Core PCE, MoM %
MERCons.ActualOur errCons. errMERCons.ActualOur errCons. err
Dec 20250.250.300.360.110.060.200.300.360.160.06
Jan 20260.190.200.280.090.080.210.200.360.150.16
Feb 20260.200.200.380.180.180.200.200.370.170.17
Mar 20260.600.600.660.060.060.300.300.290.010.01
Apr 20260.500.500.400.100.100.300.300.240.060.06
May 20260.390.400.450.060.050.200.200.320.120.12
Jun 2026−0.10−0.10−0.110.010.010.200.200.130.070.07
Mean absolute error0.0870.0770.1060.093
US Real GDP 13 quarters · 2023Q2–2026Q2
QuarterMERGDPNowSPFNY FedBEA advanceMER errGDPNow errSPF errNY Fed err
2023Q21.82.41.01.12.40.60.01.41.3
2023Q33.65.41.92.64.91.30.53.02.3
2023Q42.92.41.32.43.30.40.92.00.9
2024Q11.52.72.12.21.60.11.10.50.6
2024Q21.92.62.12.02.80.90.20.70.8
2024Q32.72.81.92.92.80.10.00.90.1
2024Q42.72.32.22.62.30.40.00.10.3
2025Q1−0.2−2.72.52.6−0.30.12.42.82.9
2025Q25.42.91.51.73.02.40.11.51.3
2025Q32.53.51.32.44.31.80.83.01.9
2025Q42.34.21.12.71.40.92.80.31.3
2026Q12.21.22.62.42.00.20.80.60.4
2026Q21.11.52.12.81.50.40.00.61.3
2026Q3nowcast not yet publishedBEA advance 2026-10-29
Mean absolute error0.740.751.351.19
Scored on the BEA advance estimate, the number each forecaster was predicting. Amber marks the closest call in each quarter.
Aggregate 13 quarters
ForecasternMAE (pp) RMSE (pp)Median err
Macro Edge Research130.741.010.40
Atlanta Fed GDPNow130.751.160.51
NY Fed Staff Nowcast131.191.441.30
Philadelphia Fed SPF, median131.351.680.91
Scored against each quarter’s BEA advance release. 2023Q2–2025Q4 are model reconstructions at each quarter’s historical data cut-off; 2026Q1 onward are published nowcasts. Atlanta Fed GDPNow’s mean absolute error is within 0.01 pp of ours over these 13 quarters — not a distinguishable margin, and it should be read as level rather than as a lead. RMSE separates them further, by 0.15 pp, because our misses cluster tighter even when the averages match.
Producer prices released record · not a forecast
PPI carries no scorecard here because we do not forecast it. BEA builds several core PCE components from PPI rather than CPI: hospital and physician services, health insurance, airfares and portfolio management. See the full PPI series, the production pipeline and the PCE-linked components →
And what the market and households expect NEXT, which is the only forward-looking series on the site: TIPS breakevens, the 5y5y forward the Fed cites when it calls expectations anchored, and the household survey that runs persistently above both. See inflation expectations →
How this table is built

Scored against the first release. Every forecast is compared with the statistical agency’s initial published estimate, not the latest revision, because that is the number the consensus was set against and the number the market traded.

Consensus as captured at our publication time. The comparison figure is the median Wall Street economist forecast available when we published, so both columns reflect the same information set. Consensus can move in the days before a release; once the number is out, neither column is restated.

Interval calibration, how often the outcome landed inside our published 68% and 95% bands, will be reported here once enough intervals have accumulated to make the figure meaningful.

Forecasts are estimates, not statements of fact, and are inherently uncertain. Past forecast accuracy is not indicative of future accuracy. See the full disclaimer.

© 2026 Macro Edge Research · All rights reserved · Disclaimer