Lies, damned lies, and PMIs

Via FTAlphaville comes a point I’ve been making repeatedly myself:

The purchasing managers’ indices are among the world’s most closely watched economic indicators, moving markets and provoking policymakers to act. And with good reason.

The polls — most of which are compiled by data firm IHS Markit — offer an early sign of how well, or badly, the economy is faring. While GDP data are only available with a lag of, at best, four weeks after the end of each quarter, the PMIs offer a near-instantaneous insight into how well activity has fared over the past month. Over time, they also tend to mirror GDP.

Today marked the release of the final readings for manufacturing activity for May. And we have some concerns about how they’re being interpreted.

A month-on-month indicator

First, we want to explain how the PMIs work.

The surveys ask purchasing managers whether activity over the current month has declined, improved, or stayed about the same compared with the previous month. Note, it does not ask them whether activity has declined, improved, or stayed about the same compared with average levels of activity.

These responses are then tallied up to produce a net figure where 50 signals no change in activity from month to month. Therefore, anything below 50 is a contraction, and anything above 50 an expansion.

Here’s how IHS Markit puts it in the footnotes of each of its surveys:

Survey responses are collected in the second half of each month and indicate the direction of change compared to the previous month. A diffusion index is calculated for each survey variable. The index is the sum of the percentage of ‘higher’ responses and half the percentage of ‘unchanged’ responses. The indices vary between 0 and 100, with a reading above 50 indicating an overall increase compared to the previous month, and below 50 an overall decrease. The indices are then seasonally adjusted.

Was May really worse than April?

For much of April advanced economies shut down due to the pandemic. Production lines across Europe that had ground to a halt in March — either because of local restrictions or disrupted supply chains — were still more or less closed.

Towards the end of the month and into May, however, many factories began to reopen. In Germany, for instance, Volkswagen restarted production at its Wolfsburg plant on April 27. In harder-hit Italy, manufacturers began to return to work in early May.

So it is rather surprising, then, that today’s PMIs suggest that manufacturing activity continued to decline between April and May.

The Italian headline reading at 45.4 is still below the crucial 50 level that separates an expansion in activity from a contraction — which is remarkable given that it fell all the way to 31.1 in April. The German reading came in at 36.6, above the April figure of 34.5 but still deep below 50, suggesting month-on-month conditions worsened.

Are markets aware of what the surveys actually mean?

Given how awful conditions were in April, one would expect any reading below 50 for May to be seized upon as an unmitigated disaster.

Not so, judging from the market reaction. Many of the major European indices are closed today, but the FTSE 100 is up by just over 1 per cent at time of writing. (In the UK, the headline reading was 40.7 for May).

Let’s see how European equities react tomorrow. But we suspect the same problem that we saw in interpreting the data for China, where a pretty average headline reading of 52 in was taken as evidence of a v-shaped recovery, will persist.

What does IHS Markit think is happening here?

Perhaps part of the problem is that people are forgetting that 50 marks the crucial boundary between expansion and contraction and are viewing a higher number as a good thing in and of itself — even though a figure below 50 following a month in which all-time lows were hit in many jurisdictions suggests conditions remain abysmal.

Others might interpret the PMI as a gauge of where conditions are compared with more normal times. But that’s not what the survey records at all. So unless respondents are en masse answering the wrong question, this would appear unlikely factor for the May readings.

So what is going on here? One feature of the current crisis that IHS Markit’s chief business economist Chris Williamson thinks might be distorting the headline readings is the degree to which global supply chains have been disrupted. Usually delivery times shorten when demand is weak, with this “positive” offering a boost to headline readings when output is falling. But that’s not happening in this instance, where we have both a supply and demand shock happening at the same time. Via Williamson:

You’ve got five subcomponents that make up the headline reading, including output, orders and supplier delivery times. Usually these five components all move together: when output falls, supply and delivery times get shorter (dragging on the PMI) because there is less demand for components, and vice versa when output quickens. But because of the pandemic supply these delivery times are getting longer, so that is distorting the headline PMI figures again in May.

So you have to look at the output sub-index more at the moment. In the case of China, this did bounce to over 61 before adjusting for seasonality in March — up from 18.0 in February at the height of the lockdown and has remained in growth territory in April and May.

Italy, which also locked down relatively early, has seen a similar, though less marked, rebound into growth territory in month on month terms (again, looking at the data before seasonally adjustment).

Here are the charts for output index for China and three European economies (note that the Italian output figure is slightly above 50, in contrast to the headline reading):

However, Williamson is still surprised at how bad the European readings for May actually are:

If anything this adds to the gloom. You would have expected a big rebound in output based on what we saw in China, but we’ve not seen that in Europe yet. When you look at the response patterns, those European purchasing managers that reported lower production in April, around half of them reported a further decline in May. That’s not what we hoped to see.

Whereas China eased its lockdown earlier and appears to have benefited from an initial boost from reviving supply chains, Europe has not been so lucky. This initial supply chain boost is showing signs of fading as the global economic outlook has deteriorated, meaning in Europe we are seeing a big drop in orders for many types of product, notably corporate demand for business equipment and machinery, so output just isn’t picking up as factories reopen. Those that stayed open are often reporting a weakening production trend.

Williamson says official data will almost certainly report a rebound in GDP in May, simply because those figures will focus on the change from zero production in many factories in Europe for much of April to at least some in May. However, official monthly are notoriously volatile and tend not to track the monthly PMIs, which bear a closer resemblance to the quarterly figures.

Here, for instance, are the relevant data points for the UK:

The picture remains bleak, then:

The PMIs are telling us that you may well see a rebound in the official production data in May, but that rise will mislead on what’s in fact a much worse underlying picture of falling demand. What the responses reflect is just how far off full capacity we still are. Companies are looking at a diminishing forward pipeline of work: backlogs of work are generally still falling as new order inflows are dropping month by month. Many economists may be putting too much weight on a few large factories and construction sites reopening in May, while underestimating the second-round longer-term impact of the pandemic on demand.

The PMIs’ trajectories might look v-shaped. But in terms of convincing us that we’re in the throes of a v-shaped recovery, they do anything but.

David Llewellyn-Smith
Latest posts by David Llewellyn-Smith (see all)

Comments are hidden for Membership Subscribers only.