LEI DECONSTRUCTED – UPDATE
By Gaius Marius of Decline and Fall of Western Civilization
I haven’t run this analysis in a year’s time, but it’s worth updating now. On twitter I’ve become vocal in the belief that the united states is headed into a severe double dip that will end all reasonable debate as to whether we are or are not in a depression. The theoretical basis for the notion emerges from the work of richard koo, whose framework of a balance sheet recession leaves little room for expansion given the rolloff of fiscal stimulus spending. I’ve further been amazed and disturbed by the continuing development of steve keen’s stock-flow macro model, which has made enormous strides in the last year.
But, while theory is fine for hypothesis, there is no substitute for evidence. unfortunately there’s been enough to convince me.
I last mentioned here the consumer metrics institute‘s GDP predictor in march. It has since marked out a horrifying path of collapse since peaking in october of last year (coincidentally about the time I last updated this study). CMI has helpfully overlaid the continuing contraction in their real-time transaction data atop the 2008-9 recession in this chart, to frightening effect. This is not theoretical exposition; a large decline in real transactions of consumer goods is behind this movement. I see this as confirmation of thesis.
A second confirmation emerges from a deconstruction of the ECRI’s leading indicators on the methodology previously described. With a broken credit mechanism — total loans and leases is still declining, as it has been for two years now — and unprecedented federal reserve bank intervention in funding markets, I feel a handful of the LEI components — yield curve, m2, stocks and the consumer expectations that largely follow stocks — probably represent broken signals better designed for standard post-war recessions than the depression we’re experiencing.
Removing their contribution to the LEI reveals the weakness of the stimulus-fueled inventory cycle “recovery” of 2009-10. Moreover, for much of this year, the “real” component aggregate was demonstrating outright contraction, with the level peaking in april 2010 — a sharp warning of an impending return to recession, even as the ECRI LEI as reported set a new high in the last reading of the series. This of course mirrors the contraction reported in CMI data.
A number of regional manufacturing indeces (philly fed, empire state) has also shown a sharp slowdown in the last two months, with inventories again starting to build — these being coindcident-to-lagging indications of economic activity, it would seem likely that contraction in consumer activity is starting to feed back into the manufacturing chain as CMI has for some time expected. House prices have continued to decline after a tax-credit-induced respite and jobs data has also disappointed, showing little or no recovery in the labor market. I expect these data series will now begin to show more severe deterioration through the end of the year — following the CMI data and the rate of federal stimulus spending down — and to remain very weak until the united states government is compelled to return to accelerating deficits in order to offset the debt deflation of the private sector.
Interesting proposal not sure it’s ok to equate the ECRI Weekly Leading Index with the Conference Board’s LEI. I picked up a few nuggets of info from a Q&A with an ECRI researcher in New York. While it is true their origins are related, Geoffrey Moore created the original LEI in the 1960, and the Weekly Leading Index in the 1980s, he presumably did not use the same inputs. Also, since the Conference Board took over the LEI maintenance in the 1990s they’ve changed inputs, including adding the yield curve, which Moore did not use.
It turns out that the ECRI WLI is almost 1:1 correlated with equities which calls into question whether it is remotely useful and a leading indicator of
anything….
I would be very interested in hearing opinions on the efficacy of the Consumer Metrics Institute tracker. I’ve been surprised that their online consumer discretionary purchase tracking declines have not shown in the broader retail sales tracking data. I understand that they track different series, but given the duration and severity of the decline, the gap has been surprising. Opinions would be appreciated from anyone with insight.
I suspect there are two major contributors. First, the most sighted retail numbers are percent gains/losses for same store sales. The retail industry has been on the whole closing poor performing stores. So if you want to shop at GAP and the store closest to you shuts down, you drive an additional 15 minutes to one that is still open. Same store sales of the one left open goes up eventhough total sales may have declined. Second, the middle class has taken the brunt of the economic hit so their sales transactons (they reprsent the majority of sales transactions) have declined as they have lost their jobs (U6 back over 17% and perhaps even gone on food stamps (new record).