Macroscope
RU
Loading findings…

  1. ForecastsUnited States

    US: the first estimate had GDP falling at a 3.8% annual rate in late 2008. Today’s data: 8.5%

    −3.8%first estimate, January 2009
    −8.5%now

    US: quarterly GDP growth at an annual rate

    020072010

    Revisions go both ways: since 1992 a quarter’s growth was later raised 82 times and lowered 54. In the second quarter of 2020 the fall was first put at 32.9%, now 28.0%.

    Source: BEA (via the Philadelphia Fed), ONS, Destatis (via the Bundesbank), BLS: first estimates and latest dataOpen on the chart →
    More

    The US statistical agency (BEA) publishes its first estimate of a quarter’s GDP about a month after it ends, when data are still thin. It is then refined for years with tax records and annual surveys.

    Quarter by quarter, 2007–2009: blue, the first estimate of GDP growth; orange, today’s data (annual rates). The shading is the revision.

    First estimates: the Philadelphia Fed Real-Time Data Set; latest: BEA. Since 1992: earlier first estimates were of other concepts (GNP, fixed-weight prices). “Raised” and “lowered”: revisions above 0.05 percentage points; 138 quarters.

    A revision is not a deception: first estimates rest on partial data, and part of the old revisions is changes in the definition of GDP. But the first estimates understated the start of the 2008–09 crisis elsewhere too: the UK −0.5% → −1.5% (2008Q3, on the previous quarter); Germany +1.5% → +0.5% (2008Q1, on the previous quarter).

    A question for this picture

    If first estimates see a crisis half as deep while it unfolds, what should those who decide in that same month rely on?

More findings

All findings →