Growth Scare - What Killed US GDP?

Fear and greed: today, the sharpest fear - fear that takes down the S&P five percentage points in a single day, fear that recoils from the double-dip recession that today seems certainly upon us.

The markets have already done their own forecast revisions, leaving the economists as usual limping behind, blinking and sniffing as we try to understand the path. But as we tinker with our models, the truth is there is a biggish problem: there’s virtually nothing in the high-frequency monthly or weekly data which should have led anyone to anticipate such a collapse in growth during 1Q. That monthly data is statistical history right now, and ought to be safely dead and buried in the databanks. But now, like some bewhiskered Victorian detectives, we need to exhume it to see if, belatedly, it holds clues as to “What Killed 1Q.”

The odds don’t initially seem good. For years whilst in Asia I kept a small momentum model for US domestic demand, tracking employment, wages, retail sales, auto sales and construction orders. For all of these, I compared monthly movements against seasonalised historic patterns, and normalised the ‘error’. The results showed how many standard deviations this data was above or below what you’d normally expect. It’s a crude model, and not one I’d rely on now, but it has rarely been as abominably out of kilter with US GDP growth as it was over the last 12 months to June 2011:

(PS. There are loads of ways I could present this as a more attractive visual fit, either by a bit of quiet smoothing on both indexes, or just presenting the straightforward YoY GDP, rather than the annualized QoQ. But it would be particularly perverse to fiddle the visuals in a piece about how to read the underlying data. Actually, you’re looking at an R score of 61 over 60 observations.)

But when one looks at the personal consumption expenditure portion of GDP, things turn out not to be out of kilter with the underlying data after all:

This tells us immediately that whatever it is that afflicted the recent revisions, it wasn’t a shock about the underlying strength of demand in the US economy.  It’s something else.

In fact, the fall in consumption expenditure in 2Q is not so difficult to trace:  auto sales were down 22.7%  annualized, and it seems reasonable to accept this as a an enforced constriction of consumption with its origins in the disruption of Japanese supply systems post March 11. Gasoline sales were down 6.7% annualized, presumably as consumers adjusted their schedules to higher gasoline prices. Once again, it needs no great leap of imagination to expect this adjustment to have a lagged effect on the rest of consumption. Exclude those two exogenous impacts, and the rest of consumption expenditure was growing by 1.8% annualized in 1Q11, slowing to 1.4% annualized in 2Q11.

And that is about in line with the monthly surveys of personal income and spending, in which spending was running at an annualized 1.2% in 2Q11. It’s not a great result, but, crucially, it is an explicable result, and one which does not necessarily commit us to the early onset of a new recession.  There’s no surprise that the US economy slowed in 2Q – but equally it probably was a slowdown, not a stall.

What about ‘the rest’– the c29% of the US economy which is not accounted for by personal consumption? Here the news for 2Q is surprisingly good: ‘the rest’ grew by an annualized 4.2%, recouping almost all the ground lost in 4Q10 (when it contracted 0.5%) and 1Q11 (when it contracted an annualized 3.7%). Above all, the fluctuations of the last nine months look ‘normal’ – and particularly normal as the economy exits a recession. Look at the chart below, and in particular compare what happened in ‘the rest’ over the last nine months with what happened to it in 2002-2003.

When we look at the detail, things clarify quite quickly: gross capital formation is doing well, rising 7.1% annualized in 2Q after 3.8% in 1Q , with matters looking pretty similar even once you take out the inventory cycle (1.2% in 1Q followed by 5.8% in 2Q). Net exports were up 10.2% annualized in 1Q and down 16.5% in 2Q, but the net figure is so small (around 3% of GDP) that these fluctuations make little impact on the overall GDP total.

 

Which leaves government consumption, which at 18.9% of GDP, or 64% of ‘the rest’, is inevitably the dominant factor.  And the annualized growth trajectories over the last three quarters stand like this: -2.8% in 4Q10, -5.9% in 1Q11, and -1.1% in 2Q11.   I don’t want to antagonize either side of the US debate about the trajectory of government debt but, folks, this is good old-fashioned fiscal drag. As you come out of a recession, tax receipts begin to rise, social security claims begin to fall and before you know it, the pace of the build-up in the national debt begins to moderate, and net government consumption begins to fall. It happened in 2002-03, and it’s happening again now – albeit on the more exaggerated scale with which we are now in other respects familiar.

Yes, it’s snuffling works for dullards, this poking around in details. If you’re still hanging on in there at the end of it all, you deserve the conclusion. Which is simple enough: the presumption of a double dip is today exaggerated hugely by fear. There are shadows across the US economic x-ray, no doubt. But we know what they are, and, to me at least, they don’t look like The Big One.