Macro Nowcasting
Official GDP and inflation figures arrive weeks or months after the period they describe, so nowcasting builds a running, constantly updated estimate of the current quarter from faster, partial data — trading the true value for a much shorter lag.
Prerequisites: Economic Data Releases and Market Reaction
GDP for a given quarter is not published until roughly a month after the quarter ends, and even then it gets revised twice more over the following two months. A trader positioning ahead of the next GDP print has no choice but to guess — the question is whether that guess is a hunch or a disciplined estimate built from data that arrives faster. Nowcasting is the practice of building that faster estimate, updating it continuously as new partial information — employment reports, retail sales, industrial production, even credit card spending data — arrives throughout the quarter.
A nowcast is a running estimate of a slow-moving official statistic, built by combining many faster, partial indicators and updated every time one of them is released. It trades precision for speed: it's rougher than the eventual official number, but available weeks or months earlier.
How the estimate updates
A nowcasting model — the Atlanta Fed's GDPNow and the New York Fed's Nowcast are the best-known public examples — starts each quarter with a rough prior, often based on recent trend growth, and revises it every time a relevant data release lands: a strong jobs report nudges the estimate up, a weak retail sales number nudges it down, each release weighted by how informative it has historically been about GDP. By the time the quarter ends, dozens of updates have narrowed the estimate considerably, well before the official BEA release confirms or contradicts it.
Worked example
At the start of Q2, a GDP nowcast starts at 1.8% annualized growth, based on the trailing trend. A strong jobs report (payrolls up 275,000 versus 180,000 expected) pushes the estimate to 2.3%. Two weeks later, a weak retail sales report (-0.3% versus +0.2% expected) pulls it back to 1.9%. Industrial production comes in roughly as expected and leaves it unchanged. By the end of the quarter, after a dozen such updates, the nowcast sits at 2.1% — and when the official first estimate is released a month later at 2.0%, the nowcast has tracked it within 0.1 percentage points, using only data available in real time.
What this means in practice
Macro traders use nowcasts to position ahead of the official release rather than react to it — if the nowcast has been running well above consensus forecasts for the official print, a trader might expect the actual GDP release to beat expectations and position accordingly in rate or equity-index futures before the number lands.
A nowcast is a statistical average of past relationships between fast data and GDP — it can be badly wrong exactly when it matters most, in a regime break like the onset of a recession or a supply shock, because the historical weights it learned no longer describe how the economy is behaving.
Related concepts
Practice in interviews
Further reading
- Federal Reserve Bank of New York, 'Nowcasting Report' methodology
- Federal Reserve Bank of Atlanta, GDPNow methodology