Wednesday, May 26, 2004
Steam versus computers
Posted by David Smith at 09:42 PM
Category: Independently-submitted research

This piece, based on an article in the April Economic Journal, summarises research by Professor Nick Crafts comparing the economic impact of steam power and computer power.

Society seems to be getting better at exploiting new ‘general purpose’ technologies like computers, electricity and steam. New research by Professor Nick Crafts, published in the latest Economic Journal, shows how long it took for steam power to have a big impact on economic growth. This contrasts with the rapid impact of information and communications technologies (ICT), which are transforming our lives and providing a powerful stimulus to economic growth as they raise productivity across the economy.

The impact of steam power following James Watt’s famous invention of 1769 is often supposed to have been similar to that of ICT. Professor Crafts’ research shows that the impact of steam on economic growth during the Industrial Revolution of 200 years ago was, in fact, trivial. Indeed, its main contribution was not delivered until long afterwards, in the second half of the nineteenth century.

He estimates that the contribution of steam power to growth in labour productivity was about 0.02 percentage points per year until 1830 rising to about 0.4 percentage points per year in the third quarter of the nineteenth century and sustained at about 0.3 percentage points from 1870 to 1910.

What lies behind these surprising results? First, steam’s impact on transport did not begin until the advent of railways starting with the Liverpool and Manchester Railway in 1830. Steam ships were not economically significant until the 1850s.

Second, very few stationary steam engines were in use long after Watt’s breakthrough. At the First Census of Production in 1907, there were about 9.7 million steam horsepower in use in Britain but as of 1830 only 165,000.

Steam power was less important for British industry than waterpower until about 1830. Widespread adoption of steam was long-delayed because it was too expensive for many uses until the development of high-pressure steam engines drastically reduced the consumption of coal after 1850.

For example, this made a massive difference to the viability of the steamship over extended distances because it solved the ‘Concorde problem’ of being unable to carry an economic payload – in this case because of the space taken up by coal to be burnt on the journey.

How does this compare with ICT today? Obviously, the range of applications of computers has also grown over time – the internet only happened in the 1990s but the microprocessor dates back to 1971. And the price of ICT equipment has fallen remorselessly over time eventually encouraging widespread adoption of the technology.

Where there is a dramatic difference is in the rates of price decline in the two cases. William Nordhaus estimates that the costs of a million standardised computer operations per second fell to a millionth of its 1950 level by 1990 and to a billionth of its 1950 level by 2000.

In contrast, Professor Crafts estimates the annual cost of a steam horsepower had fallen to about 40% of its 1760 level by 1830 and to about one-eighth of its 1760 costs by 1910. There really was no equivalent of Moore’s Law in the nineteenth century.

It is probably not surprising therefore that ICT has had a much greater impact on productivity growth sooner than did steam. Already by the 1980s, this was about 0.7 percentage points per year in the United States rising by the late 1990s to 1.8 percentage points per year according to Stephen Oliner and Daniel Sichel.

We also know from the work of Paul David that electricity had its maximum effect on productivity growth in the 1920s, about 40 years after commercial generation began.

So the message seems to be that society is getting steadily better at rapidly exploiting new general purpose technologies.

Notes: ‘Steam as a General Purpose Technology’ by Nicholas Crafts is published in the April 2004 issue of the Economic Journal.

Crafts is Professor of Economic History at the London School of Economics.