The debate about how to raise UK economic growth begs a question: is rapid growth possible even with the best feasible policies?
My chart shows growth in real GDP per head over 30-year periods since 1750. Trend growth of much above 1% per year is unusual, seen mostly only in the post-1945 period. From a long-run perspective, therefore, the stagnation of recent years are a reversion to the norm.
Why was the post-1945 period so unusual?
On one view, it was because WWII created a large backlog of civilian investment opportunities as well as a need to repair war-damaged assets, not least of them housing. This implies that growth in that period was indeed one-off, something that cannot be repeated.
On another view, though, it was because post-war insitutions facilitated growth. The credible promise to maintain full employment gave companies the confidence to invest because they knew that demand would be strong enough to justify it. Public ownership and high income tax rates reduced incentives and opportunties for rent-seeking. And strong unions and rising wages incentivized companies to invest in labour-saving technologies and productivity improvements.
Personally, I find this debate of historical rather than practical interest. The political obstacles to recreating post-war institutions are now so huge that doing so is a mere fantasy.
It's not just social democracy that faces strong opposition, though. So too do centrist attempts to raise growth. Incumbent companies and monopolists don't want increased competition; lawyers and accountants don't want tax simplification; landlords and financiers don't want taxes to shift from incomes to capital or land; Brexiteers don't want us to rejoin the single market; and nimbys don't want laxer planning controls. For that matter, many voters and workers also don't want the disruption and uncertainty that must accompany creative destruction. A stagnant economy in which zombie firms preserve jobs and in which we face less threat from foreign competition or new technology is perfectly tolerable for many; they spent the 2010s consistently voting for it.
Not that there's anything unusual in such opposition. Joel Mokyr has written (pdf):
Technological progress in a given society is by and large a temporary and vulnerable process, with many powerful enemies whose vested interest in the status quo or aversion to change of any kind continuously threaten it. The net result is that changes in technology, the mainspring of economic progress, have been rare and that stasis or change at very slow rates has been the rule rather than the exception.
Barriers to rapid growth, however, are not only political. There are also economic ones.
Often in thinking of long-term growth it helps to distinguish between the level of potential output and the speed at which an economy approaches it. There are barriers to both.
One is simply that even big investments yield only moderate increases in output. Conventional production functions put the elasticity of output with respect to capital at around 0.3, meaning a 10% rise in the capital stock yields a 3% rise in potential output. With the ONS estimating that the UK's net capital stock excluding housing is £3.5 trillion (yes, I know), this implies we need £350bn of extra investment to get just a 3% rise in potential GDP. To get that in a year requires more than doubling business investment*.
Why do we get so little bang from our buck from investment? One answer is at the end of my road. The canal there has been unused since the 1840s when traffic switched to the newly opened railway. Which is how investment works: it renders earlier projects useless. When Lidl invests in a new store it reduces Tesco's sales; Amazon's expansion has cut the sales of bricks and mortar retailers; investment in wind turbines is meant to displace gas-fired power stations. And so on. Investment doesn't add much to output but rather shifts production from one capital to another.
This is not the only reason, though. The railways and internet were both huge transformative technologies. Both, however, fuelled investments in projects that simply failed. Lots of new capital; not so much extra output. And of course, it's not just shiny new technologies that have this problem. Any high street sees shops open and close within a few months.
A lot of capital spending is wasted. This is partly because investment is driven by sentiment rather than a rational view of a project's prospects. It's because the future is unknowable. The "dark forces of time and ignorance" mean that some investments will always fail.
Even potentially productive investments, however, don't yield big immediate rises in aggregate output. 19th century factories tell us one reason why. For years, electrification of these did little to raise productivity. Until someone realized that electricity meant that machines didn't have to be clustered close to a driveshaft but could instead be arranged linearly - which allowed for more efficient workflow and the development of the production line. To be really useful, technological change often requires organizational change. And that can take many years.
Simple maths tells us another reason. New dynamic industries are small. Even if they grow very quickly, therefore, they'll add little to overall GDP. If an industry accounts for 1% of GDP and trebles in size over ten years it will add only 0.2 percentage points per year to GDP growth. This is one reason why the early years of the industrial revolution did not greatly raise GDP; there just weren't that many steam engines.
There's another reason for that low growth, though. The UK saw financial crises in 1772, 1796, 1810, 1815, 1825, 1837, 1847, 1857, 1866 and 1890. Such panics retarded capital spending by creating uncertainty and restricting credit. It's no accident that the period of history to have seen the fastest growth - the 1940s to early 70s - was also one which did not see a crisis.
Crises, though, aren't the only reason for low investment. Another is that, except in periods of irrational exuberance, companies hold back on capital spending either in the hope of benefitting from future better, cheaper technology or from fear that if they invest today their rivals will use those technologies to undercut them. William Nordhaus famously showed that producers capture "only a minuscule fraction of the social returns from technological advances". That companies have now learned this is, I suspect, one reason for our recent low growth.
There's a further problem. At times of near-full employment, extra production of capital goods requires extra workers - workers who must move from other industries. But people are slow to move. As Abhijit Banerjee and Esther Duflo say in Good Economics for Hard Times, economies are "sticky." Most people prefer to stay in jobs they know; employers prefer relevant experience; it takes time for people to discover opportunities in different industries; to decide to make a radical career change; and to retrain. All of which delays the movement necessary for growth.
It doesn't just retard investment, however. It also slows down productivity growth.
To see the problem here, consider how much your personal productivity has risen in the last few years? I don't mean in the sense of being able to do the job more comfortably with less stress, but in the sense of doing more or the same in fewer paid hours. If you've been in your job for more than say ten years, the answer is: not much. Ater a while, the productivity gains from learning are small. We've two pieces of evidence for this. One comes from the effect of experience and tenure upon wages; this is generally small (pdf) for older people staying in the same job. The other comes from Jonathan Haskell and colleagues, who have shown that most productivity (pdf) gains don't come from companies simply upping their game.
Instead, they come from inefficient places shutting and more efficient ones opening or expanding - creative destruction. But sticky economies and low labour mobility retard this process too.
We shouldn't expect AI to change any of this. If it is anything like every other technical innovation in history it will destroy some jobs and create others (and if it doesn't, that'll be a failure of policy). But this requires labour mobility, which is slow. It might also, as with electricity in factories, require organizational changes of a sort we cannot now foresee. Companies will be slow to make these.
And there's a downside to AI: it can increase the efficiency not just of productive workers but also of the enemies of genuine productivity. It will enable nimbys and their lawyers to write more letters to planning committees; it'll allow patent trolls to become even more litigious; and in facilitating more cybercrime it'll divert even more of our brightest people away from production towards cyber security. And even if it does raise productivity in the short-term, it could cut it in the longer-term. In writing essays for students, AI is depriving people of useful future skills. And in replacing some graduate trainees AI raises productivity today, but where will the experienced accountants and lawyers come from in 20 years' time?**
As if all this were not enough, there's one other problem. Every significant classical economist - Smith, Ricardo, Mill, Marx - thought that even the meagre (capitalist) growth of their time would eventually cease as diminishing returns won the race against technical progress. They would not be surprised that we are now talking about stagnation; what would surprise them is just that it's taken so long.
The problem here is not simply that returns on capital have fallen, thereby diminishing the motive and means for investment - though this does seem to have happened. It's also that we've plucked the long-hanging fruit, the easiest ideas. As Robert Gordon pointed out, the really great, life-changing innovations are long behind us. And the efficiency of R&D spending is declining around the world.
Now, none of this is to say we shouldn't try to increase productivity. We certainly should, but it's unlikely that small efforts will triumph in the face of big obstables; a big wall needs a big sledgehammer. It'd nice to have a government that at least tried to prove me wrong. Nor, of course, is it to deny the possibility of a short-term cyclical upswing.
What it does mean is that we must think about a plan B: what if we cannot grow our way out of the fiscal corner described so well by Giles?
This is a dispiriting thought. John Stuart Mill thought that the stationary state would be "a very considerable improvement on our present condition" which would give us "as much scope as ever for all kinds of mental culture, and moral and social progress". Recent experience, however, suggests he was wrong; low growth breeds not social progress but its opposite - intolerance and racism. Which poses the question: how might we avoid stagnation having this effect?
* Perhaps a lot more. Much of this investment is replacing worn-out assets rather than pure additions to the capital stock.
** My concern here is mitigated by the fact that a lot of what these people do is not to increase aggregate output but to cream off rents.