Working Paper: NBER ID: w24895
Authors: Jonathan H. Wright
Abstract: In the 2018 comprehensive update of the national income and product accounts, the Bureau of Economic Analysis released not seasonally adjusted data, and modified its seasonal adjustment procedures. I find some indication of residual seasonality in the seasonally adjusted data as published before this update. The evidence for residual seasonality is weaker in the seasonally adjusted data after the update. I also directly seasonally adjusted the aggregate not seasonally adjusted data, and this entirely avoids residual seasonality. The average absolute difference between my seasonally adjusted real GDP data and the current official published version is 1.1 percentage points in quarter-over-quarter annualized growth rates.
Keywords: No keywords provided
JEL Codes: C32; E01
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| BEA's 2018 seasonal adjustment update (E01) | presence of residual seasonality in NIPA data (E01) |
| new adjustment methodology (C82) | weaker indication of residual seasonality (C22) |
| prior indication of residual seasonality (C22) | significant negative first-quarter effect in real GDP growth rates (E20) |
| directly seasonally adjusting aggregate not seasonally adjusted data (C43) | residual seasonality is entirely avoided (C22) |
| residual seasonality post-revision (C22) | some components still show residual seasonality (C22) |