That makes the spread between Nowcast and GDPNow 1.9 percentage points. However, that straight-up comparison is invalid for reasons discussed below.
First, let’s investigate Nowcast changes.
March 17, 2017: Highlights
- The FRBNY Staff Nowcast stands at 2.8% for 2017:Q1 and 2.5% for 2017:Q2.
- This week’s news had a negative impact on the nowcast, which declined 0.4 percentage point for Q1 and 0.5 percentage point for Q2.
- Retail sales, housing and construction, and manufacturing data all had negative contributions for both quarters.
Nowcast 1st Quarter Detail
- Retail Sales: -0.133
- Building Permits: -0.140
- Industrial Production Index: -0.087
- Capacity Utilization: -0.093
- Data Revisions: +0.112
The net effect of the above five items is -0.341. The net change from March 10 to March 17 was -0.360.
I thought the report was on the weak side, not only because of the low headline number but also because of the decline in auto sales. Econoday suggested averaging January and February, and that is a reasonable approach.
January was revised from +0.4% to +0.6% making the two-month average about +0.35%, a respectable total. Perhaps that accounts for the +0.112 the Nowcast assigns to “Data Revisions“.
Overall, the standout item in the report was the drop in auto sales for two consecutive months. Auto sales fell 0.2 percent in February and 1.3 percent in January.
Housing Starts and Building Permits
Starts for February were up 3% but permits were down 6.2%.
In my estimation, this should have added to 1st quarter GDP but subtracted from future GDP. Permits will translate into starts sometime down the road, not immediately.
Industrial Production and Capacity Utilization
I commented on IP-CU on March 17 in Industrial Production Flat, Manufacturing Jumps: Another Weather-Related Phenomenon?
Click on the link for a series of 9 charts.
The manufacturing component took a big jump but in perhaps another weather-related related event, industrial production was flat.
The rise in manufacturing was arguably a good sign but the overall number and year-over-year comparisons were not encouraging.
- Index: +0.31%
- Manufacturing: +1.39%
- Motor Vehicles and Parts: +2.62%
- Crude: -1.96%
The strongest component on a y-o-y basis is motor vehicles, but there are strong retail sales signs that suggest the best days for auto sales are in the rear view mirror.
GDPNow did not comment on Friday’s Industrial Production numbers. This is the way things appeared on Thursday, March 16.
Did the Spread Shrink?
As noted above, the spread is now down to 1.9 percentage points as of Friday. But GDPNow did not comment on Friday’s IP-CU report. The next release for both GDPNow and Nowcast is Friday, March 24.
Had GDPNow updated its forecast on Friday, it’s likely, but not guaranteed that its 1st quarter GDP estimate would have declined as well.
We cannot say for sure, but my guess is that GDPNow would have further declined by 0.2 or 0.3 percentage points. It’s possible the spread widened.
Let’s assume the IP-CU impact on GDPNow was -0.3 percentage points.
That would keep the spread between the two estimates above 2 percentage points, meaning they both could still be wrong, even with a margin of error of +- 1 percentage point.
For further discussion, please see Rogue Nowcasts: 67% Chance Both are Wrong.
As noted previously, Models Don’t Think.
However, I like the reports because I can make weather-related adjustments, auto-related adjustments, and housing-related adjusts, etc. as I see fit.
Over time, one can learn what is likely to matter and what doesn’t, even if the models cannot make such adjustments.
Mike “Mish” Shedlock