Japan's 3Q GDP: Not Quite The Disaster It Seems

Although no-one anticipated the 1.6% fall in Japan's quarterly real seasonally adjusted and annualized GDP preliminary estimate, beyond the volatility, the underlying picture remains surprisingly still intact.

Let's start with the obvious: no-one knows how to forecast Japan's quarterly real annualized GDP results. The last consensus was 2.2%, my own pin-the-tail-on-the-donkey effort was 2%, and the result was minus 1.6%. Not one of the 28 economists contributing to the Bloomberg consensus forecast a contraction, and I'm not gloating: I've spent year trying and rejecting various ways to make this forecast, and, except upon request, no longer make the attempt.

In any case, it's not clear that the quarterly seasonally adjusted annualized 'real' GDP growth is any longer the most important measure for Japan:

1. with population declining by around 0.2% a year, arguably what matters more is the longer-term trend in GDP and GDP per capita. After today's estimates, 'real' GDP is growing around 1% pa on a 12m basis, and so, roughly 1.2% real GDP per capita.
2. Given Japan's history with deflation, and its overhang of public debt, nominal GDP is surely just as important as 'real' GDP. Now, nominal GDP was disappointing, rising only 0.8% yoy in 3Q (whilst the deflator rose to 1.9%), which cut the 12m nominal growth to 1.9% (with a deflator of 0.8%).

These perspective make Japan's 3Q GDP performance already seem rather less disastrous than today's headlines proclaim. Clearly April's tax rise introduced volatility not just into Japan's GDP numbers, but also into various aspects of Japanese economic behaviour. But if you're prepared to look beyond that volatility (and that's a big if – most people aren't), the picture looks to be improving in several important ways.

First, the capital cycle is probably still intact: non-residential investment rose by 3.9% yoy, almost holding historic seasonal trends, and by my estimates (depreciating all nominal non-residential investment spending over 10 years), Japan's capital stock is now finally rising for the first time since 1Q09. More, despite the volatility of the last two quarters, nominal GDP expressed as a return on capital stock is still rising, and is now approaching pre-crisis levels, which, incidentally, were the best since the bubble years. Beyond the tax-generated volatility, one would expect the historically high and rising return on capital indicator to perpetuate the investment cycle.  

Second, despite the poor headline GDP numbers, output per worker, when deflated by changes in capital stock per worker, continues to rise, which underpins continued employment gains. During 3Q, employment rose 0.7% yoy, and in the 12m to September total output per worker rose 1.3% when deflated by changes to capital per worker. Continued productivity gains should underpin continued employment growth, just as rising asset turns/ROC can be expected to underpin the capital cycle, despite the current volatility.

Third, compensation rose by 2.6% yoy in 3Q and was up 1.6% on a 12ma: this may not sound much, but these are the highest rises in so far this century. It pushed compensation as percentage of GDP to 51.8% in the 12m to September: this is not a record, but it is a full standard deviation higher than the average this century. And it is rising. What is more, the rise in compensation now exactly matches the rise in nominal private consumption, which also rose 1.6% in the 12m to September.

Using the Kaleckian idea of disaggregation elements of profits (Investment; Consumption minus Wage; Net government spending), it seems likely that profits inched up in the 12m to September, but with the pace slowing to a crawl. Those profits are underpinned by increased investment spending and, to a lesser extent, consumption minus wages, whilst they are being eroded by fractional fiscal tightening and the growing trade deficit.  Above all, however, the fact is that the source of Japan's profits are now more obviously aligned with likely sources of sustainable growth than they have probably ever been. The picture is almost classically 'normal'.

  

  

Finally, to return to the problem: the reason why it's rare to forecast Japan's quarterly GDP with any accuracy or certainty is that there appears to be no stable relationship between what is reported on a monthly basis, and what tumbles out of the quarterly national accounts.  For that reasons it is worth understanding what that monthly data is telling us. Here are my monthly momentum indicators for the industrial economy, for domestic demand, and for monetary conditions, showing the 6m trendline, which expresses how many standard deviations away from seasonalized trends the data currently is running:

These tell what I think is a coherent and plausible story: the industrial momentum trendline clearly shows the pre-tax acceleration which peaks in March and subsequently rapidly subsides. It shows a similar trajectory for domestic demand, which rallies up to February 2014 as purchases are brought forward to pre-empt the tax rise, only to slump proportionately afterwards. In both cases, there is a hint of stabilization visible by September (when the industrial and domestic demand momentum indicators end).  But already there is some response in monetary conditions, with a modest expansion opening up from July onwards, and still developing. (And since this indicator ends in October, it has yet to reflect the impact of Bank of Japan's expansion of QE, or the full extent of the Yen's depreciation). 

That hint of stabilization in industrial conditions and domestic demand shows up more clearly when one includes the monthly noise as well as the 6m signal line: 

In the case of industrial momentum, September produced the most positive set of data since April, with industrial production up 2.9% mom sa, which was enough to raise capacity utilization rates to 99.9 (0.4SDs above long-term average), whilst exports rose 6.9% yoy (in yen terms), which was enough to reverse the run-up in inventory/shipment ratio seen since April. 

Aggregate domestic demand indicators also produced the most positive deviation against seasonal trends in September since May's dead-cat bounce.  Employment rose 0.7% yoy, on a monthly movt which was 0.8SDs higher than historic seasonal trends, and average monthly cash earnings also rose 0.7% yoy, which was 0.2SDs above trend.  Retail sales rose 2.3% yoy, which was the highest since March, on a monthly movt which was 0.6SDs above trend, and vehicle sales fell only 5.5% yoy against a tough base of comparison and on a movt which was a full SD above trend.  Set against this, however, was a 45.2% yoy slump in private construction orders, which was 0.9SDs below trend. Overall, however, the continuing strength of labour markets coupled with renewed industrial momentum suggests there may be more to the strength of September's gains than the dead-cat bounce of May.