Yesterday we reported that over the past few weeks [12], something very disturbing has taken place at the Atlanta Fed Center for Quantitative Economic Research, which keeps a model, called GDPNow [13], that mimics the methods used by the BEA to estimate real GDP growth.
According to the AtlantaFed, "the GDPNow forecast is constructed by aggregating statistical model forecasts of 13 subcomponents that comprise GDP. Other private forecasters use similar approaches to “nowcast” GDP growth. However, these forecasts are not updated more than once a month or quarter, are not publicly available, or do not have forecasts of the subcomponents of GDP that add “color” to the top-line number. The Atlanta Fed GDPNow model fills these three voids."
In other words, what the AtlantaFed has done is recreate the bean-counting methodology used by the BEA, and all other forecastsers, however instead of using monthly data update cadence, it does so with data in real time.
This is a problem because as we showed yesterday [12], this most real-time model of GDP estimation, is showing something scary: a 1.2% GDP in Q1 compared to consensus estimates in the mid-2% range, and a tumble of more than 1% to just what this same model predicted Q1 GDP would be one month ago!
[14]
What's going on here, and how is it possible that there is such a massive disconnect between the one, arguably most accurate and updated Fed forecasting model and everyone else.
Courtesy of an excel spreadsheet called, logically enough, GDPTrackingModelDataAndForecasts.xlsx (link [15]), we can find the answer.
While we urge readers to play around with the model at their leisure, here is the bottom line: a snapshot of the weekly evolution of the summary tab, which shows precisely which line item is leading to the collapse in Q1 GDP, from 2.3% as of February 13 to half that as of March 2.
[16]
For those who are used seeing it on a cumulative basis, here is the other summary table that also lays out how the AtlantaFed reaches its shocking 1.2% GDP number:
[17]
What becomes immediately apparent is that while there has been a sharp deterioration across most sources of economic output including government spending, which is now expected to detract -0.8% from growth, Net Exports, and residential investment, it is the collapse in "Structures", aka non-residential investment, best known as the Capital expenditures spending or lack thereof by US shale companies on such items as offices and extraction structures, including oil and gas wells this is about to pummel US economic growth.
Incidentally, we previewed all of this in "The Next Victim Of Crashing Oil Prices: Housing [18]" and "Houston, You Have A Huge Problem: One-Sixth Of US Office Space Under Construction Is In This Texas City [19]."
Digging down into the underlying data, which feeds through the weekly updates of Atlanta Fed staffer Patrick Higgins (the cell comments are his), we have confirmation of what is about to "shock" everyone when the BEA reports its advance GDP report in two months time: it is all non-res structures, and especially petroleum and natural gas wells, which will (or rather already have) unleashed a full-blown investment shock for the US economy.
[20]
Of course, none of this should come as news: after all, just this past October, another Fed, this time the St. Loius Fed, had a must-read post on the correlation between oil pricess and business fixed investment in structures [21].
Most economists believe lower oil prices are positive for the economy: They lead to lower gasoline and diesel prices, which tend to reduce headline inflation, which increases consumer purchasing power. Lower oil prices also tend to reduce operating expenses for transportation firms, such as airlines, trucking, and delivery services. The sharp drop in crude oil prices since mid-June 2014 is generally expected to produce positive (if temporary) economic effects. Lower oil prices generally don’t benefit energy producers, but the vast majority of households, firms, and organizations are net consumers, not net producers; so, lower prices still tend to bring net benefits.
One way the effects of lower oil prices reveal themselves is through mining activity. (More precisely, “real private nonresidential fixed investment in mining exploration, shafts, and wells.”) In 2013, fixed private investment in mining activity was about 5 percent of total fixed private investment and only 0.8 percent of real GDP. Still, since the third quarter of 2009, mining activity has increased at a 17.1 percent annual rate—much faster than the 5.5 percent rate of gain in total fixed private investment.
As the graph [below], mining activity (which includes drilling) is positively correlated with crude oil prices. When oil prices rise, this activity increases and so does investment in it. When oil prices fall, this activity slows and investment in it falls.
How this graph was created: Search for mining investment to find the first series, then add “Crude oil prices WTI” for the second. Limit the sample to start in 1999.
We took the liberty of recreating the St Louis Fed blog graph which can be found here [21]. It is shown below.
So it looks like indeed the Atlanta Fed's GDPNow model is not only accurate but is predicting precisely what will happen in the coming months.
But wait, there's more.
As was widely reported earlier this morning, oil giant Exxon became the latest major to slash CapEx as a result of plunging oil prices, and it now plans to commit some 12% less to its E&P capital spending budget in 2015, a grand total of $34 billion, compared to the $38.5 billion spent last year. Add up across all the other majors, shale and other E&P companies and suddenly you have a spending collapse in the hundreds of billions.
What this means for GDP is that 1.2% growth in Q1 may just be the beginning as capital spending collapses across the board, especially since the Atlanta Fed model description says "as more monthly source data becomes available, the GDPNow forecast for a particular quarter evolves and generally becomes more accurate."
In the meantime, for everyone else confused by such concepts as spending on non-residential structures, and capex in general, or how various components of the GDP calculation actually add up across to the bottom line number, we urge everyone to open Atlanta Fed' spreadsheet GDPTrackingModelDataAndForecasts.xlsx [15] which may well be the "scariest" spreadsheet in the Fed's possession right now, but it is also the most informative and not burdened by non-GAAP, seasonal and various other propaganda adjustments, in order to understand why consensus is once again off by orders of magnitude from what the final Q1 GDP number will end up being, and why, no, it won't be the "weather's fault" this time.
Source: AtlantaFed [13]and GDPTrackingModelDataAndForecasts.xls [15]