There is a commonly held belief that US manufacturing leads the rest of the economy, so it is surely a worry that factory output has been flat since late 2014. And yet the broad economy kept growing—with GDP up 2% YoY in 4Q15, consumption up 2.7% YoY, and home construction by almost 10%. One explanation for this apparent decoupling is the US’s shift to a more service-intensive “knowledge economy” which has rendered metal bashing and more generally, making stuff, a statistical sideshow. Another possibility is that the broad economy is about to follow the manufacturing sector’s lead by stalling out, or worse, contracting in a proper recession.
As it turns out, we think there is a simpler explanation.
The common belief that US manufacturing has predictive power stems from the Conference Board’s Leading Economic index which includes new factory orders and a measure of hours worked in the sector. Our own Charles Gave uses manufacturing data in his US recession indicator (see Gauging The Chances Of A US Recession). Indeed, the fact that monthly manufacturing data comes out well ahead of quarterly GDP releases means that in one sense it offers an early read on the situation.
However, that is not the same thing as leading the rest of the economy—as shown in the chart below. US manufacturing activity turns out to be a coincident indicator, and a volatile one at that. The ebbs and flows in production are highly correlated with US GDP (0.9), yet manufacturing can for years at a time grow at a faster or slower rate than GDP.
Manufacturing output is also far more volatile than GDP, which gives it the (false) appearance of leading the broad economy. So, the coincidence of a modest economic slowdown with a material fall in manufacturing activity is perfectly normal and does not point to a structural decoupling, at least not yet. US manufacturing and GDP could both get weaker from here, yet by the same token they could both get stronger.
What about the ISM manufacturing PMI, which is commonly considered a leading indicator of the sector due to its survey-based nature? Since ISM respondents are asked to only report on information for that specific month, then any apple-to-apple comparison with the YoY real GDP growth rate requires a 12 month smoothing of the PMI reading. The chart below shows that the ISM is also a coincident indicator, and not quite as reliable as manufacturing production. Its main advantage is that it is released earlier than actual production numbers.
Housing as a leading indicator
In seeking a genuine leading indicator of US growth, the one standout sector is residential construction which tends to lead by three to six months. This probably stems from US housing being mostly financed with debt, and so highly sensitive to changes in interest rate expectations. When the Federal Reserve cuts rates and the long end of the curve falls, as is the norm in recessions, the housing sector tends to benefit first. To be sure, during the post-2008 cycle, housing was slow to recover as it had been at the epicentre of the preceding boom and bust. Yet by 2012, normal cyclical service was restored; housing struggled in 2H13 after the “taper tantrum” caused mortgage rates to bolt higher, but picked up nicely once bond markets calmed down in 2014. Today, construction growth is robust which reflects low interest rates and decently good housing affordability (see US Housing: From Great To Good).
On the flip-side, as the cycle inevitably progresses inflation should move higher, sparking the Fed to hike interest rates. Mortgage rates will rise and at the point that financial conditions become too tight, housing is likely to be the first sector offering a clear signal of problems ahead.
In this sense, housing offers a better leading indicator than manufacturing (as it actually leads rather than just being released earlier). Also, housing affordability can be monitored as a leading indicator of our “leading” sector (see the second chart below). The drawback is that home construction is more volatile than manufacturing—so the signal throws up noise.
Hence we are sticking with our preferred recession indicator: the spread between the return on capital and the cost of that capital. This approach keeps a starring role for interest rates and shows that while the US outlook has worsened, the ROIC-COC spread is still significantly positive and so recession is not likely (See Returns On Capital Are Deteriorating). Manufacturing weakness aside, so long as housing affordability remains decent and key ROIC-COC spread remains in positive territory, we do not expect a recession to materialize—and will continue to capitalize on these volatile equity markets by buying the dips.