US Labour Participation: Andrew Smithers and Lessons from Japan

The analytical debate about the fall in the US labour participation ratio essentially revolves around the degree to which potential pensioners are deferring or taking retirement in response to the impact of the great financial crisis. It's a genuinely important argument because on it turns estimates of likely potential growth rates and the size of the output gap (if any). Since these are key inputs into most monetary policy decision models, the stakes could hardly be higher.

Andrew Smithers' piece in the FT today  is the latest contribution. As I read it, the key message is that the fall in the participation ratio over the last few years paints an inaccurate picture of labour market slackness thanks to the way the US calculates the participation ratio. In most countries, the calculation is approximately employed + unemployed as a proportion of those of working aged – ie, aged 15-64. But in the US, the calculation is employed + unemployed as a proportion of all those aged over 15 years. So, naturally enough, as more people move into retirement, the US count of labour participation will fall. However, if you take the more accepted 15-64 participation ratio, you'll find that it has been rising quite steeply over the last few years.  In fact, the US labour participation ratio is not around 63% as claimed, but rather is slightly above 69%.  Smithers conclusion? There is less slack in labour markets than you might think, which means a) that the potential growth rate is commensurately lower and b) wage inflation will emerge sooner than you think. From which we must conclude that the Fed's room for sustained easing is less than we think – so sell bonds. 

Instinctively, or perhaps even ideologically, I'm sceptical, because: 

1. I'm convinced there must be a large cyclical element in the recent decline in the US participation ratio, which might be glossed as a large number of people reassessing what they want out of their lives in the wake of the financial crash;  
2. I see no good reason to hold on to the notion that people in non-labour-intensive employment will want to retire aged 65.  Take Mr Smithers, for example: I don't know how old he is, but if he joined SG Warburg in 1962, then it's a safe bet that he's a distinguished living exemplar of this point. The view of the inevitability, or desirability, of retirement aged 65 should be recognized as an extremely rare exception in any culture at any time, and one which there is no good a priori reason to expect to endure. If you doubt me, read Dickens.
3. I believe the essence of economics and economic growth is improvisation. This is why Goodhart's law is so crucial to any working economist, and more generally why one should always be sceptical of 'the economic data' (since by definition, it is measuring activities which are already superseded, or are in the process of being superseded).  In the case of demographics and labour markets,  we'll probably discover that the 'inevitability of demographics' turns out to be neither particularly inevitable nor specifically predictable.  Or put it another way: old age isn't going to be what it used to be.

Those are my prejudices, and, of course, they may be wrong.  But let's look at the experience of the country with the longest history of responding to an aging population:  Japan.  It is worth running Mr Smithers' charts for Japan over the long term to see how various measurements of labour participation have changed over time. 

First, let's have a look at the underlying demographic change – in particular, the rise of those aged 65+. This has been dramatic, rising steadily from 8.9% in 1980 to 25% in 2013. This compares to an US rate estimated at 13.7% in 2012. Not only is Japan's total elderly population greater, it also grew more rapidly, averaging a rise of around 0.6% of the population per year during the last 20 years, which is approximately double the rate at which the proportion of similarly-aged Americans is rising. 

Given this far more dramatic demographic aging, it surely should be expected that we would see commensurately more dramatic twists and turns in Japan's labour participation rates, on both bases of calculation, than we are seeing in the US currently. If, that is, the current fluctuations are indeed a product primarily of demographic change. 

The chart below measures changes in the labour participation ratio on both the 15-65 measure and the 15 and above measure. 

As one would expect, the shape is similar to what one sees currently in the US:  labour participation rises on the Aged 15+ count, but has fallen on the Aged 15-65 count. But it is not the similarities which are important, but the differences.  First, the decline in the Aged 15+ count is far less dramatic than one sees currently in the US, even though the aging was more rapid. The peak of participation on this count was in 1992, at 63.9% when the those aged 15+ accounted for only 13% of the population.  Since then it has declined by only 4.6 percentage points to 59.3%, even though the proportion of the population aged 15+ jumped 12 percentage points to 25%.  By contrast, the US participation rate fell from 63.4% to a low of 58.2%  - a fall of 5.2 percentage points -  between 2007 and 2010.  

Both the size of the fall, and its speed, are unlike anything seen in Japan.  It's too much, and much too fast.

The second point is that Japan's experience places a huge question-mark over whether it is reasonable to accept 65 as a relevant age to start counting retirees.  After all, the participation rate calculated against a labour force of those aged 15-65 has now risen to no less than 83% currently in Japan, whilst on Mr Smithers' calculations,  on the same base of calculation, the US has yet to hit 70%.  The opportunity for extended working lives clearly has been grasped strongly in Japan, and if the US is anything like similar, there is simply no reason to believe that current levels of participation are anywhere near a ceiling. 

Finally,  Mr Smithers' conclusions about the  possible impact on labour market pressures of a higher-than-advertised participation ratio also finds no support from Japan's experience. At the same time as corporate Japan has evidently employed ever greater numbers of 'retired' workers, so it has persistently reported an oversupply of labour, and wage inflation has been almost completely absent. 

This is not to say that Mr Smithers will ultimately be wrong: he has long argued that the upper limit on Japanese labour numbers provides a tight constraint on Japan's GDP growth, since it simply is not possible to substitute capital for labour without accepting ever-diminishing rates of return.  And so far, the overall evidence from Japan bears him out. 

But in the case of the US right now, the evidence he provides is not strong enough to support strong conclusions. Demographics probably aren't the overwhelming factor in the recent decline in US labour participation ratios; there's no reason to believe that the 'real participation rate' is approaching a ceiling; and there's every reason to believe that improvisation will trump demographics in US labour markets for the foreseeable future.