Coming tomorrow morning – the revisions to GDP over the last five years incorporating lots of new source data may change substantially our understanding of the evolution of economic activity (discussion by Claudia Sahm) (in contrast, last year’s annual revisions didn’t really change the outlook).
Here’s what we know about GDP, GDO, GDI and GDP+, as well as nowcasts of GDP.
Figure 1: GDP (blue), GDI (green), GDO (orange), GDP+ (red), GDPNow (sky blue square), NY Fed nowcast (pink triangle), all in billions Ch.2012$ SAAR. Level of GDP+ based upon 2019Q4 GDP level. Source: BEA 2023Q2 2nd release, Philadelphia Fed, Atlanta Fed (9/27), and NY Fed (9/22), and author’s calculations.
If GDP is revised down, the nowcasts can be thought of being pulled down (while the growth rate from 2023Q2 to 2023Q3 is unchanged, as the GDP nowcasts are couched in growth rates).
Will GDP likely be revised down? As documented by CEA (2015) and Nalewaik and Braun (BPEA, 2011), this is a likely outcome, with GDP reverting more to GDO than otherwise. For the most recent quarters, GDI growth is more apt to be growth rate of GDP post-revision.
BEA (2022) reports the size of revisions from advance to final vintages.
Source: BEA (2022).
Interestingly, pre-revision growth rates do not appear to be biased, relative to final estimates. However, the RMSE is 1% (MAE is 1.2%). So, even without taking a stand on direction, one can say big changes are not unlikely (which is why the NBER Business Cycle Dating Committee does not put primary reliance on GDP — see 2001 recession).