First print

The number a market trades on, and the least complete version of it. A statistical agency publishes to a fixed calendar, so the release date arrives whether or not every survey response and administrative return has landed. The output is a considered estimate assembled from what was available, and the agency says so: the word preliminary or advance is usually in the title, and the coverage rate is stated in the release.

Nothing is concealed, and the trade off is deliberate. An agency could wait until every input had arrived and publish one figure that never changed, at the cost of publishing it long after the period it describes had ceased to be decision relevant. Timeliness is chosen over finality, and the revision that follows is the price of that choice rather than evidence of an error.

The consequence for anyone reading a screen is that the figure with the most market impact is the one with the widest margin around it. Later estimates are more accurate and move prices less, because by then the market has already positioned on the first. A comparison drawn between a first print and a fully revised figure from an earlier period is a comparison between two different things, which is the commonest way a data series is misread.