GDP nowcast
Also called nowcast, GDPNow, nowcasting
A model-based estimate of the current quarter's GDP growth, updated as new data are released and made before the official figure is out. The Atlanta Fed's GDPNow is a well-known example; it is not an official Fed forecast.

On this page· 5 sections
Gross domestic product is released with a delay. A nowcast fills the gap: a model-based estimate of growth for a quarter that is still going on, before the official figure exists.
How it works
The best-known US example is the Atlanta Fed’s GDPNow, described by the bank as “a running estimate of real GDP growth based on available economic data for the current measured quarter”[1]. It uses a method similar to the one the Bureau of Economic Analysis (BEA) uses for the official figure[1].
As reports come out from the Bureau of Labor Statistics, the Census Bureau, the Institute for Supply Management and the Treasury, they are fed into the model and the estimate updates[1]. Results come from calculation alone: no subjective adjustments are made once the model is set[1].
GDPNow is not the Atlanta Fed’s official forecast, and it is not endorsed by the Fed’s president, the Federal Reserve System or the FOMC[1].
Simpler rules exist too. In the euro area, the ECB describes a PMI-based tracker rule: quarterly real GDP growth is 10% of the quarterly average of the composite output PMI minus 50. The ECB notes that this rule-of-thumb nowcast well in the pre-COVID period[2]. The ECB’s staff models also use bridge equations, which link monthly data to quarterly GDP[2].
Example
Take the ECB rule with made-up data. Say the quarterly average of the euro area composite output PMI is 52.0.
- Subtract 50: 52.0 − 50 = 2.0.
- Take 10%: 0.10 × 2.0 = 0.2.
The rule gives quarterly real GDP growth of 0.2%. A reading of 50.0 gives 0, because 50 is the no-change level of the PMI[2].
Watch out
A nowcast moves with every release. On 1 July 2022 GDPNow’s estimate for the second quarter fell to −2.1% from −1.0% the day before, after the ISM and construction reports[1]. That is the model reacting to inputs, not a new view.
Accuracy changes over time. The Atlanta Fed reported an absolute forecast error of 0.5 percent over the first 5½ years the tool was online, before the pandemic. It has been less accurate since[1].
The ECB notes that a PMI-based tool lost some predictive capacity during pandemic lockdowns, then recovered[2].
Sources
- Central bankFederal Reserve Bank of AtlantaPulling Back the Curtain on GDPNow (2022)Checked November 9, 2026
- Central bankEuropean Central BankIs the PMI a reliable indicator for nowcasting euro area real GDP? (Economic Bulletin 1/2024, G. de Bondt and L. Saiz)Checked November 9, 2026
