Last quarter (third quarter), a huge initial discrepancy between the Atlanta Fed GDPNow Model and the FRBNY Nowcast Model ultimately resolved in favor of Nowcast.

For the third quarter, GDPNow tracked as high as 3.8% with the Nowcast lower nearly every step of the way until convergence, much lower near the end.

Once again, at the start of 4th quarter estimates, GDPNow is substantially higher. Will things play out the same way?

Nowcast November 18, 2016 Highlights

  • The FRBNY Staff Nowcast stands at 2.4% for 2016 Q4.
  • The current nowcast is 0.8 percentage point higher than two weeks ago, when the previous report was published.
  • The improvement was driven by large positive news from retail sales and housing starts.

Nowcast 4th Quarter 2016 GDP Estimate


Nowcast 4th Quarter 2016 GDP Detail



  • Retail sales added 0.249 percentage points to the model.
  • Housing starts added 0.472 percentage points to the model.
  • I match the calculation for retail sales. For housing starts, I come up with (25.5 – 4.85) * 0.023 = 0.475 percentage points. I suspected a rounding error on housing starts but they actually rose by 25.52% which would also yield 0.475 with rounding.
  • A discrepancy of .003 is totally meaningless of course. I was just testing the calculation.

Predicting the Outcome in Advance

If we knew weights and forecasts in advance, we could predict the effect each economic report would have on the Nowcast with a very high degree of reliability.

The problem, I believe, is that each data point affects the model prediction of the next data point, so advance numbers would be subject to change. I believe such changes would be minimal, however.

GDPNow Latest Forecast: 3.6 Percent — November 17, 2016


Final GDPNow for 3rd Quarter


Last quarter started with GDPNow nearly a percentage point higher and here we go again, with an even larger discrepancy. It will be interesting to see if the numbers converge and in which direction.

Mike “Mish” Shedlock