The puzzle is this. Manufacturing industry in the US is far more productive than anywhere else in the world. It turns out more goods per worker-hour than even the other industrial titans like Germany and Japan. Thus British workers in industry produce only about two-thirds as much as their American equivalents and even German workers only three-quarters as much.
For much of the past 40 years, productivity in other nations has grown faster than in the US, so there has been some catch-up. But at the current rates of growth it would take up to 60 years for Britain to catch up with US productivity levels and up to 50 years for Germany to match them. Why is America's lead so dominant?
Of course, there are some obvious potential explanations. The principal candidates are investment in capital equipment, since more of it and newer machinery based on the most advanced technology should provide an obvious advantage; a more highly skilled workforce that is innately more efficient; and higher expenditure on research and development. Together, these go part way towards accounting for America's overwhelming productivity lead.
A recent working paper from the National Institute of Economic and Social Research* explores just how far they go by comparing productivity levels between the US, UK and Germany, industry by industry, and considering the level of capital employed, R&D spending and skill levels in each case.
The international comparison turns out to vary quite widely by industry. In the UK the best relative productivity performance is delivered by rubber and plastics, which almost matches the US. But instrument engineering, electrical engineering, chemicals and motor vehicles achieve only half the American output per employee. In Germany mechanical engineering and wood products actually do slightly better than their American equivalents and rubber and plastics, clothing and metal products do almost as well.
The researchers, Geoff Mason and Mary O'Mahony, look at the three candidate explanations industry by industry, and find them lacking. For example, the amount of physical capital per worker is unquestionably higher in the US than in Britain, but investment in Germany has grown at a much faster pace than in the US, so it does not lag nearly so far behind.
In fact, Germany has the newest machinery, embodying the most recent technology, while the age of the machines in use in the US and UK is comparable. And the pattern of high levels of capital or newer machinery does not correspond well to the pattern of productivity by industry.
The role of R&D fares better as a solution to the productivity puzzle. According to the most recent figures - and they are not without their measurement problems - the US has an overwhelming advantage in accumulated spending on R&D. In the industries where Germany had a lead in R&D, like metals and mechanical engineering, its productivity is close to US levels. Britain has particularly low R&D in the car industry, perhaps reflecting the fact that foreign-owned multinationals carry out most of this spending in their home country.
However, the role of skills is a less useful explanation. Both the US and UK have a higher proportion of workers with low skills, and although America has more workers with a very high skill level than either of the other two countries, Germany has an overwhelming lead in craft and intermediate skills.
All in all, the paper concludes that these three factors leave a large part of the productivity gap between the US and the others unaccounted for. In only four industries - computers, cars, instrument engineering and transport equipment - do the conventional explanations actually explain the difference.
This confirms previous research, and there has been no shortage of speculation among economists about what might then account for the gap. The difficulty of measuring "true" innovation and the organisation of production (mass production versus craft-based techniques) have been two favourites. The management consultancy firm McKinsey has favoured the notion that the regulatory and competitive environment is decisive, with early and extensive deregulation the source of the US advantage.
Mr Mason and Ms O'Mahony put forward an alternative based on NIESR comparisons of individual plants in various industries in the three countries. These strongly suggest that the key is the importance of economies of scale.
For example, in biscuit manufacture doubled output boosted productivity by around 16 per cent. More than half of the US output was produced in plants that ran at least some production lines uninterrupted for a week or longer. Only a quarter of the British plants did this, while the German biscuit industry was geared towards the production of expensive biscuits in small batches.
In precision engineering, all of the US productivity advantage was down to bigger batch sizes. The aggregate figures simply do not include any measure of the scale of production, which is why they overlook this explanation.
This result has some profound implications. Not least, it confirms the growing importance in modern economics of the phenomenon of increasing returns to scale - or the fact that bigger is better. This was a commonplace in the profession in the 1950s, and went out of favour because it was too hard to express mathematically in formal economic models. It is now enjoying a new lease of life thanks to the importance of increasing returns in high-technology industries like biotech and computing - and to the fact that economists are better at maths than they used to be.
Some economists started concentrating on the implications of a world in which increasing returns are all-pervasive about a decade ago, but it is a safe bet that almost all new developments in economic theory will be based on this world view.
And it could change the conventional wisdom dramatically. To take just one example, acknowledging the importance of economies of scale in industry tilts the cost-benefit analysis of Economic and Monetary Union in Europe decisively in favour of the single currency. For many people have assumed that the potential economic costs caused by the loss of the ability to devalue the currency are on a par with the somewhat nebulous benefits of "completing" the single European market.
But what if creating a single economy bigger in size than the US could lead to a 20 or 30 per cent gain in productivity levels? That is the possibility held out by this solution to the productivity puzzle.