The first piece in this series laid out the dumb statistical basis for expecting a renewed recession in the US within a year. Those dumb stats tell us that the low-risk bet must now be for recession – and this means that it’s a brave Street economist who in such uncertain times is prepared to spurn the low-risk bet. Among many other factors, markets right now are anticipating an avalanche of downgrades of economies and earnings which we can assume will follow.
But the second article didn’t allow the data to be dumb – when we got it to talk, it had a tale to tell. It turned out that the pattern of 1Q and 2Q growth was at worst explicable, and at best compatible with renewed growth. The slowdown was largely due to the impact of the supply-jam in the auto sector, the impact of higher fuel prices on fuel consumption, and the predictable (but larger than previously seen) fiscal drags which appear regularly as an economy exits recession. Absent these effects, personal consumption demand is holding up well, investment spending far better. The private sector, in other words, isn’t rolling over.
Nonetheless, we are moving into a period where the statistics very likely to point to a recession. It’s also very likely that the Street will also change its forecasts, and maybe its view, to get into line with the dumb power of the stats. Later, we must also factor in the impact of a falling stockmarket on household finances.
So is there a case for ‘no recession’ and if so, what is it?
To get a handle on the likely turning points in a business cycle, I look at what’s happening to the return on factors of production. If return on capital is rising, ceteris paribus, it’s unlikely that investment spending will fall. If return on labour (ie, real labour productivity) is rising, it is unlikely that unemployment will rise. If both investment spending and employment continue to grow, it takes something pretty extraordinary to overturn the business cycle. (Although, again, a complete and prolonged collapse in financial confidence might just do it).
Working out what’s probably happening to return on invested capital isn’t necessarily easy (and conventional economics has shockingly little useful to say about it). What I do is express the flow of GDP as an income from a stock of fixed capital. The trick then is to work out what’s happening to the stock of capital, and this I do by taking a 10yr straight line depreciation over all fixed investment spending. If GDP is rising faster than the stock of capital, then it’s a fair bet that the return on that capital is rising. This is an unconventional measure, which, as far as I know, I’m alone in calculating/using. However, it has two far more respectable relations. First, it is an attempt to replicate the ‘asset turns’ ratio (total revenues / total assets) used in the Dupont decomposition of return on equity. Second, it is kissing cousins with the ultra-respectable ICOR (Incremental Capital-Output Ratio) so beloved of the World Bank, IMF, OECD etc. Anyone who’s actually tried to work with ICOR will know its drawbacks: but if it helps, think of this measure as an ACOR (Average Capital Output Ratio).
Here’s how returns to the US factors of production look right now:
Even after the GDP revisions, return on capital is clearly rising fast, and is probably at its best since the turn of the century. Real labour productivity growth has come off slightly, but is still extraordinarily positive.
Has there ever been a situation in which these readings were so positive, and yet were interrupted by a recession?
Since 1980, we’ve had four recessions: each of them had been preceded by an inflection in this return on capital indicator. Contrariwise, we’ve not had a recession whilst this indicator was rising. It is extremely unlikely that this indicator is going to turn any time soon, since capital stock is still shrinking by around 0.32% YoY, whilst nominal GDP is growing (in 2Q) by 3.7%.
What about labour? Here the picture is not quite so clear-cut. The three last recessions happened after a period when real labour productivity inflected had downwards, and was sharply negative. Right now, labour productivity continues to grow very sharply, but it has inflected downwards. This negative inflection seems to have produced a period of labour market softness. However, this is very easy to exaggerate, and I believe it has been very much exaggerated by the seasonal adjustment process. Before seasonal adjustments, non-farm payrolls were growing by 0.9% YoY in June, down from a high of 1.1% in April. They grew 0.5% in May (in line with historic seasonal expectations) and grew 0.3% MoM in June (vs flat historic seasonal expectations). The rising unemployment ratio, of course, masks the continued growth in jobs.
If employment is still growing quite strongly, growth of wages has nonetheless slipped – and it’s this we see showing up flat personal income data.
Once again, a little historic perspective might be helpful: compensation of employees as % GDP hit a multi-decadal low, nearly 3SDs from historic average, at the end of 2010, it recovered somewhat in 1Q 2011, but was steady during 2Q 2011. Do we think this is proportion has the potential to go lower? If not, then there remains some potential for a positive surprise from wages (and demand).
Notice, once again, that over the last 30 years, recessions have tended to happen after a period when this indicator was rising, or at a high level relative to recent previous experience. At the moment, we seem a very long way from that.
My conclusion? The dumb stats point to the likelihood of recession. When you get the underlying data to talk, that doesn’t seem like a foregone conclusion. And when you look at what’s happening to returns to factors of production – even after all the revisions – a recession in the next year would be unprecedented in recent US economic history.
So it’s probably right to be brave at this point. Unfortunately, that leads us straight into the next collision. “The only thing we have to fear is fear itself.” Was FDR right?
No comments:
Post a Comment