Courtesy of Mish
On Friday, April 28, the BEA will release its preliminary estimate for first quarter GDP.
Prior to that release, here is a compilation of six estimates from ZeroHedge, GDPNow, Nowcast, ISM, Markit, and me.
GDPNow Forecast: 0.5 Percent — April 18, 2017
FRBNY Nowcast: 2.8 Percent — April 21, 2017
Model Flaws
- Nowcast uses no hard auto data: This is a serious error. Autos account for 20% of retail sales and fleet sales are also very important.
- Nowcast has an incorrect reliance on unemployment rate: People dropping out of the labor force and actual employment rising can both move the number in the same direction. Both things cannot mean the same thing.
- ISM vs PMI: Both reports measure the same thing, yet those reports signal very different things. At least one of them is wrong. GDPNow and Nowcast both rely on ISM even though the PMI reports have been more accurate, at least recently.
- The GDPNow and Nowcast models both suffer from an inability to think. The weather provides a nice example. In December, the weather was unusually cold, causing Industrial Production numbers to soar (heat and electric production), for the entire upcoming quarter. I estimated in advance, January would take away those numbers. My assertion played out, at least for GDPNow. I still cannot account for Nowcast.
ISM vs PMI
I discussed the difference between ISM and Markit’s PMI estimates recently, for both manufacturing and non-manufacturing (services).