Lavorgna (SMBC Nikko) and Millar (Barclays) say look to GDI for a better estimate of economic activity. Both argue that GDI might better signal an incipient recession (USAToday):

Figure 1: GDP (bold black), GDO (bold light blue), GDP+ (red), GDI (blue), all in bn.Ch.2017$ SAAR. GDP+ level uses GDP+ growth rates iterated on 2019Q4 GDP. Source: BEA 2023Q3 2nd release, Philadelphia Fed, and author’s calculations.

GDI is indeed far below GDP (Zerohedge asserts that GDP will be revised down toward GDI). It’s true that GDI has shrunk by 0.2% y/y; on the other hand, q/q annualized growth was +1.5%. In any case, My reading of the CEA (2015) and Nalewaik and Braun (BPEA, 2011), is that GDO — the arithmetic average of GDP and GDI — is a more accurate measure of final GDP, with GDP reverting more to GDO than otherwise. GDO growth is indeed more modest than GDP growth (3.3% vs. 5.2%, q/q SAAR).

GDP+, published by the Philadelphia Fed, incorporates information regarding the revisions to obtain a more precise estimate of final GDP. One problem with comparing this series is that only growth rates are provided. I iterate the GDP+ growth rates on the 2019Q4 level of GDP (when GDP and GDO matched). The implied level of GDP+ is close to the same level as GDO in 2023Q3, but with yet again more modest growth.

As Jacobs et al. (JBES, 2022) point out (see Jim’s post on GDP++), the statistical approach used in GDP+ makes the latent factor estimate of GDP less volatile than the observed. GDP++ does not incorporate this requirement. In the 2021Q3-22Q2 period, GDP++ growth matches GDO.

From my standpoint, I’d put more weight on GDO than on strictly GDI (or GDP, for that matter). That would imply economic activity is 1.25% lower than indicated by the expenditure side measure (GDP).

More By This Author: