Over the weekend, we posted a rhetorical question, wondering about the sustainability of the so-called economic rebound, when we showed that contrary to Atlanta Fed Nowcasts and various other estimations, Goldman's own internal economic tracker, its Current Activity Indicator (CAI), had slowed to 1.3%, the lowest print in 7 years.
Today, Goldman updated its CAI tracker and found the following:
Several major economic indicators have recently disappointed, including both the May employment report and the ISM non-manufacturing survey. While highly valuable, even these individual indicators can be noisy from month-to-month. We therefore rely heavily on our current activity indicator (CAI), a composite measure of economic activity based on the correlations between a large number of high-frequency indicators. The CAI now stands at 1.2% in May, down from 2.2% in April, but with the 3-month moving average still at 1.7% versus 2.0% in December.
In other words, the June CAI just dropped once more, from May's 1.3% print, to 1.2%, the lowest economic "expansion" estimate since 2009. Perhaps not surprising is that this series has been declining in virtually a straight line since the end of QE3...
So where does this dramatic weakness not captured in Q2 GDP estimates comes from? Goldman explains:
We construct a CAI “heatmap” in two steps. First, starting from the list of 56 indicators entering the calculation of the CAI, we replace component-level indicators with their headline series (e.g. the CAI includes 10 components of nonfarm payrolls; for this exercise we just use overall employment). Second, we express each indicator in GDP-equivalent terms. We calculate GDP growth implied by the univariate relationships and allow the intercept to vary over time for each of the 31 indicators to reflect, for instance, changes in trend productivity growth.
Exhibit 1 shows current implied GDP growth rates by indicator, all expressed as 3-month moving averages. Housing sits on top, with single-family new home sales and single-family housing starts listed in the top-3. Michigan consumer expectations, real retail sales and real personal consumption expenditures also imply a solid 2-3% GDP growth rate, taken in isolation. However, several labor market indicators look soft, with total nonfarm payrolls, the payrolls diffusion index and household employment in the bottom quartile of current implied GDP growth rates
The CAI—expressed as a 3-month moving average—has declined from 2.0% in December to 1.7% in May, implying a deceleration in broad growth momentum. To analyze the source of this deceleration, the year-to-date changes in implied GDP growth rates are shown in Exhibit 2. Various manufacturing surveys have picked up, albeit from depressed levels. The five indicators that decelerated the most include labor market gauges (nonfarm payrolls, the diffusion index and household employment) as well as the ISM and Markit PMI services surveys.
The notable recent deceleration of services and labor market data reflects two broader types of rebalancing. First, US factories appear to be gradually recovering after the slowing in industrial activity which started in mid-2014. The gap between our non-industrial and industrial CAI— which peaked in October 2015 at 2.5 percentage points (pp)—has now narrowed to 0.8pp reflecting mostly a 1.2pp weakening of the non-industry CAI and also a 0.4pp improvement in the industry CAI. The modest recovery of US factories still appears tentative, given the weakness in some of the May surveys. Second, the gap between labor and non-labor market data now seems roughly closed. The gap between the labor and non-labor CAIs peaked in December 2015 at 1.5pp but recently actually turned negative (right panel of Exhibit 3). The 1.8pp change in this gap is mostly driven by a 1.4pp contribution from weaker labor market data, while the 0.4pp contribution from stronger non-labor market data is more modest. This rebalancing may reflect a normalization of measured productivity growth after a period of excessive weakness.
The upshot from our analysis is that both labor market as well as manufacturing data will be particularly important to watch in the coming weeks and months. Our take is that the job market is still making progress, but at a slower pace, and that reduced slack does not provide a firm basis for employment growth to slow to its 85k breakeven rate soon. We also expect the manufacturing sector to recover further, as the negative shocks associated with dollar appreciation and the shale collapse gradually fade.
In other words, forget not only a June rate hike, but certainly July and most likely September as well. As for the Fed hiking at he end of the year, well that may be up to the next president.