Producitivty growth: the long-term

If today's GDP figures are to be believed - which is a big if - productivity has slumped. The ONS says GDP fell 0.3% in the last two years whilst total hours worked rose 2.2%. This implies a drop in GDP per hour of 2.4%. What's more, productivity is 3.3% lower than it was five years ago.

My chart, taken from the Bank of England's database*, puts this into historic context. It shows that, unless productivity recovers significantly in the rest of this year (which is possible), we'll see the first fall in peacetime five-year productivity since 1932, and only the second such drop since the late 1880s.

Now, I'm prepared to believe that productivity isn't really this bad: maybe GDP has been under-recorded and employment over-recorded; there's a lot of noise in the latter.

However, it's unlikely that data revisions will overturn the big message of my chart - that productivity growth has been trending downwards since the late 60s. As Noah Smith says, something big happened in the 70s.

To some extent, this something is simple mean reversion. If we split annual peacetime productivity growth into distinct periods we get:
1856-1913 = 1.3%
1920-1939 = 1.7%
1947-1973 = 3.1%
1973-2011 = 1.9%

What stands out here is the fast productivity growth during post-war period. The last 40 years of weaker growth look normal compared to the pre-war period.

This poses the question: why did productivity grow so well in the golden age?

Part of the story is that post-war reconstruction permitted a productivity catch-up. The new equipment firms bought after the war embodied the technological gains made in earlier years, but these gains natually eventually slowed down; Verdoorn's law predicts that rapid growth for any reason (eg growth of our export markets) will be associated with rapid productivity growth; and low commodity prices helped spur productivity too.

On this view, the golden age was a happy period which simply cannot be reattained.

This is undoubtedly part of the story. But is it the whole story?What if social democracy itself contributed to that productivity growth, for example:

- The belief that Keynesian policies would ensure full employment, along with a framework of statist planning, gave firms the confidence to invest, and better capital equimpment means more productive workers.

- Full employment and strong trades unions meant that managers could not make big profits merely by depressing wages, and so had to look for genuine efficiency gains.

- Workplaces in which managers consult with unions, and so gain cooperation, find it easier to grow productivity than dictatorial workplaces;industrial relations weren't entirely conflictual in the 60s, remember.

What lends this view some credence is the sluggish productivity growth of the pre-1914 period. This suggests that free markets on their own do not unleash rapid productivity improvements.

If this line of thinking is anywhere near-correct, then a serious recovery in productivity requires big political change.

* I'm measuring labour input by average weekly hours multiplied by employment. Note that the working week has almost halved since the 1850s, from almost 60 hours to just over 31.

Comments

[Keith]

This may be right. I would add however that it may also be due to the application of scientific progress. The advances in knowledge in the middle twentieth century may have been relatively easy to apply at an industry level. Since the exhaustion of the inventions of that period progress has been slower. We have also seen the financisation of the economy; high returns to finance seem to have done little for those not in that sector of the economy. Bain capitals off shoring does nothing for the productivity of American workers.

[steko mann]

why can your voice not be heard on newsnight, instead of the usual freemarket apologists?

[rogerh]

One big change since the 1980s is that the best equipment can be situated anywhere in the world - concrete slab+electricity+workers+transport is all that is needed. The clever designers can also be anywhere in the world. The effect is to drastically cut our productivity growth - it goes elsewhere.

Efficiency improvements run into diminishing returns and provoke the 'do it different' reaction. Now it is easier to 'do it different'. Cooperation was relevant when there was no easy alternative.

Human intelligence seems rather limited and no feasible education policy seems likely to produce a big increase in clever people who's work can fund the rest of us.

Hard to see what political change would do any good - protectionism - communism? I fear the only realistic route is to cut costs and split society in a much more explicit way - perhaps create company towns or even regions where restrictions on
housing, wage rates and employment law are drastically loosened, think Samsungville or Hyundai-shire. The comfy jobs
elsewhere can be competed for by the educated classes and those newly incentivised by the comparatively less comfy
conditions in the new corporate townships.

[Zorblog]

Chris, I don't believe you are always right, but you do think of questions other economists don't address.
This is certainly the biggest flaw of the new-classical and new-Keynesian frameworks: they prevent economists from taking the broader view, they ignore the fundamental questions.

[Luis Enrique]

I would benefit from studying the seventies - I have this impression of constant industrial action and chronically inefficient state-run enterprises - British Leyland - which is hard to square with this golden age of productivity that this view of the data gives us.

zorblog I think that's a bit unfair, plenty of mainstream economists think about long-run, or deeper, determinants of productivity. New Keynesian models are for short-run business cycle problems.

[Account Deleted]

Rogerh is onto something when he talks about productivity growth being offshored, but it is misleading to group equipment and designers together. Our problem has been the attempt to retain the latter in lieu of the former ("high-tech jobs for all").

The main way you achieve productivity growth is by producing the same output with fewer people. Thus manufacturing and production has seen encormous increases due to automation and supply-chain improvements such as containerisation (fewer dockers).

People-centric processes (services and admin) tend to produce lower productivity growth. This is not just a matter of surplus labour value. Parkinson's Law also applies: thus time saved through email is wasted reading pointless emails.

As the natural rate of demand for white-collar roles has gone down, we have seen the creation of supernumerary jobs, which in turn lowers productivity: corporate accountants, HR, BPR, marketing etc.

That might appear self-defeating, but we should remember that profitability has been high enough (in part due to the one-off boost of globalisation), to preserve white-collar jobs as blue-collar ones have evaporated. This preservation is necessary to ensure sufficient electoral support for the economic order.

The UK's problem is that an industrial strategy based on high-tech, design and the creative industries is delusional, for two reasons.

First, the high-wage jobs we think we'll keep as the low-wage jobs are offshored are themselves now being offshored, as wages in China and elsewhere rise and as they reap the benefits of massive investment in education.

Second, an economy our size focused on services (with structurally lower productivity growth) needs either a falling exchange rate or a series of bubbles to provide earnings growth in lieu of productivity. A smaller country can escape this death spiral if focused on high-end services that command a premium, but the UK is simply too big.

[Account Deleted]

@Luis, the impression given in the 70s by the media was that nationalised industries were rubbish because they were nationalised and that inflation was caused by trades union militancy.

Some (not all) nationalised industries were rubbish because the state took over failing private enterprises that had suffered from decades of under-investment and fragmentation, notably coal mining, the railways and car manufacturing.

The problem with the strategy was that government then didn't invest enough in modernisation, left rationalisation till too late, and diverted too many resources to "white heat" poster-boys like Concorde and nuclear power. They were also reluctant to purge the old management and baulked at workers' control, which led to fractious labour relations.

The overall level of strikes was heavily influenced by inflation, which averaged 13% a year, peaking at 25% in 1975 - i.e. inflation led to militancy, not the other way round. This in turn led to wage repression, i.e. the stagnation in median earnings that has continued to this day. This is the "something big" that the linked article by Noah Smith refers to.

[Luke]

Is lower productivity growth in services inevitable? Partly I'm wondering whether technology can help in services. I think there's some evidence that it took time for electricity to have much effect on manufacturing as it took time for people to learn how to use it - could something similar apply to IT while dinosaurs like me lumber on?

And could wage growth shut down or diminish some services that have limited capacity for productivity improvement? Eg if people got fed up with paying barbers and gave themselves a No 2, leaving barbers to find something else to do? Same goes for a lot of service industries - I'm guessing that not that much of the fund management industry needs to be actively managed.

Apologies if I have ignored some well known and accepted economic theory/model.

[Account Deleted]

@Luke, the problem is not services per se but people, i.e. labour. Services have a disproportionately large/costly people factor, hence below-average productivity growth.

Personal services (like barbering) could, in theory, be automated, however the suspicion is that people want the human touch and auxiliary benefits like having a chat. Even substituting cheaper people can be counter-productive, where this degrades auxiliary value, as offshored call centres have shown.

[Luke]

A to E, I can see the high labour problem, but as long as people value the chat, aren't they getting value? At some point, they say "much as I value what mr Samson says [shameless plug for my local barber despite him never giving me a discount for being follically challenged], I'm going to trim my own." I'm getting close to that point -discounts for students with long hair - what about me? Isn't that the creative destruction we read about? And on anecdotal evidence, I suspect more men are going for Nos 1-3. Chris can probably produce an elegant model to prove or disprove this theory.

[Account Deleted]

@Luke, be careful with the barber microfoundation - that way madness lies: http://en.wikipedia.org/wiki/Barber_paradox

[Luke]

So you fully haired people have time to get haircuts, wash your luxuriant locks AND read 19th century logicians' writings on barbers? My point stands - you could usefully be doing something - anything- else.

[Greg]

As an industry becomes more productive its share of employment falls, due to diminishing marginal utility.

No matter how efficiently cars are made, for example, most people won't want more than two each. They'll buy holidays in Ibiza instead of more cars. So after an initial phase, the car industry declines in importance as an employer in proportion to its productivity growth.

As demand moves from one set of industries to the next, productivity growth speeds up or slows down according to the potential (at the the then-current state of technology) for productivity growth inherent in the newly important industries.

Why did productivity grow so fast in the "golden age"? The reasons Chris gives are important but secondary.

Productivity growth was high because, at the time, people demanded the products of industries with (also at the time) the potential for high productivity growth - demanded so much of them that those industries were a large part of employment for a couple of decades.

Nowadays, people don't have much in the way of unsatisfied needs (in Maslow's sense), and the economy is accordingly diversified. High-productivity industries are, necessarily, negligible employers.

Employment is dominated by low-productivity activities, which have low productivity after many years because their potential productivity growth rate is small. So aggregate productivity growth is slow.

What happens next? Under most imaginable scenarios, productivity growth stays low, with or without political change -- unless a meaningful universal basic income is introduced. (Simply driving a lot of people out of the workforce merely depresses demand, investment, and productivity.)

[Account Deleted]

The basic income point is key. In a mature economy such as ours, underlying technological productivity growth will remain high, but this will be offset by the counter-movement of white-collar job creation, producing net sluggish growth.

Faster productivity growth could be achieved (i.e. the underlying rate revealed), if we started to gradually shorten the working week while maintaining output. The reduction in wages could be offset by the introduction of a basic income, leaving net income the same but hours shorter. The larger profit would fund the basic income.

Over time, real productivity gains could be remitted in two ways. Some of the gain would be remitted as cash, to maintain parity with price inflation, but some would be remitted as shorter time. Living standards would still rise due to long-term commodity deflation - i.e. goods becoming cheaper in real and (in some cases) absolute terms.

In summary, we are artifically depressing productivity in order to maintain working hours, though this process is biased to privilege white-collar over blue-collar jobs. The surplus value of labour is being split between capital and the privileged electoral bloc.

A basic income stands little chance of implementation ("free money for chavs") because it would lead to greater redistribution not of money but of labour time. The world we are making is increasingly a clientelistic one, in which work is token by which we extract rent.

[lokalizator gps]

I can learn lot of data about productivity growth,stumbling and mumbling.Thank you for your sharing.