Thanks to the international stock-market bloodbath that welcomed in the new year, I’m once again being bombarded with ‘recession looming’ graphs. But is that justified?
Let’s start with some random claims I have seen in recent days. If the high-yield market (corporate bonds with a low credit rating) posts a negative annual return, recession is never far behind. Or, if US corporate earnings fall for three quarters on the trot, recession is imminent. But I also liked this one: if the expansion has been going for five years and unemployment falls below 6%, this is quickly followed by recession (see graph).
Many of these claims are incorrect. In 1994 and 2002 high yield posted negative returns but there was no sign of a recession. As for the US earnings, although US corporate earnings are under some pressure on aggregate, this is primarily due to developments in the oil and mining sectors. Excluding these two sectors, which are relatively small in the US economy, US earnings actually remained remarkably stable. And in terms of the second graph: I’ve rarely seen such a blatant example of data mining. If I’ve counted correctly, there are four bars that ‘accurately’ predicted a recession, but I’m looking at 11 recessions. 4 out of 11: flipping a coin works better, I would say. On closer inspection, the blue lines in the chart only show when an expansion period has been going on for five years: the 6% unemployment doesn’t play any role at all in this graph. In short: if an expansion persists for more than five years, the odds of a recession increase. The fact that it sometimes takes another three years for that recession to actually materialize is not mentioned.
Back to the first graph
But I digress: back to the first graph. This graph originated from a presentation by Jeff Gundlach, CEO of Double line. It shows the development of producer confidence in the processing industry (ISM Manufacturing) and the US GDP. US producer confidence in the industrial sector has been under severe pressure for some time now, thanks to a strong dollar and oil-price developments. That last one in particular is exerting a heavy toll on the US shale gas sector, which is already in recession according to the classical definition (two quarters of negative growth in a row). Nobody would deny that they’ve got it bad in that part of the US economy. However, that doesn’t mean the entire economy is collapsing.
But Gundlach’s graph does suggest that. His statement during the presentation was “If manufacturing isn’t a good indicator of the US economy, then you need to explain this”.
Well, historically there is a strong link between producer confidence and the underlying economic cycle, but the graph presented by Gundlach raises two questions: why doesn’t this graph go back further than 2008, and why are we looking at nominal and not real GDP?
The second part of that question remains guesswork for me, but I think I can answer the first. If we look back further in time, it’s clear that we have witnessed severe declines in producer confidence that were not followed by economic recession more often. The graph below shows real GDP (shaded area), the ISM manufacturing (black line) and the ISM non-manufacturing (green line). That last one is producer confidence in the services sector.
So? In 1985, 1994 and 1998, producer confidence declined strongly without leading to recession. It even fell quite a bit below current levels, while the real growth rate clearly remained at a higher level.
But I didn’t include the green line for fun: at the moment we are facing a striking divergence between confidence in the processing industry and in the services sector. But why we should look at the black line (with a weight of less than 20% of the total US economy) as opposed to the green line (with a weight of 80% or more) is of course a good question.