How We Count Counts: Diane Coyle on The Rise and Fall of American Growth
Diane Coyle has reviewed Robert Gordon’s new book (out late January), The Rise and Fall of American Growth: The U.S. Standard of Living since the Civil War.
Gordon’s central argument will be familiar to readers of his work. In his view, the main technological and productivity enhancing innovations that drove American growth in the early to mid 20th century – electricity, internal combustion engine, running water, indoor toilets, communications, TV, chemicals, petroleum – could only happen once, have run their course, and the prospects of future growth look uninspiring. For Gordon, it is foreseeable that the rapid progress made over the past 250 years will turn out to be a unique episode in human history.
Coyle zeros in on the two main mechanisms to which Gordon attributes the slowing of growth. The first is that future innovation will be slower or its effects less important. Coyle finds this argument less convincing.
“What I find odd about Gordon’s argument is his insistence that there is a kind of competition between the good old days of ‘great innovations’ and today’s innovations – which are necessarily different.
One issue is the extent to which he ignores all but a limited range of digital innovation; low carbon energy, automated vehicles, new materials such as graphene, gene-based medicine etc. don’t feature. The book claims more recent innovations are occurring mainly in entertainment, communication and information technologies, and presents these as simply less important (while making great play of the importance of radio, telephone and TV earlier).”
While I have yet to read the book, Gordon makes several similar arguments in an NBER working paper. There he gives a few examples of his view of more recent technological innovations as compared to the Great Inventions of the mid-20th century.
“…more familiar was the rapid development of the web and ecommerce after 1995, a process largely completed by 2005. Many one-time-only conversions occurred, for instance from card catalogues in wooden cabinets to flat screens in the world’s libraries and the replacement of punch-hole paper catalogues with flat-screen electronic ordering systems in the world’s auto dealers and wholesalers.”
In other words, the benefits of the computer revolution were one time boosts, not lasting increases in labor productivity. Gordon then invokes Solow’s famous sentence that “we [could] see the computers everywhere except in the productivity statistics.” When the effects do show up, Gordon says, they fade out by 2004 and labor productivity flat lines.
Solow’s interpretation (~26 mins into the interview) of where the productivity gains went is different, and more consistent with Coyle’s deeper point. In short, the statistics themselves doesn’t capture the full gains from innovation:
“And when that happened, it happened in an interesting way. It turned out when there were first clear indications, maybe 8 or 10 years later, of improvements in productivity on a national scale that could be traced to computers statistically, it turned out a large part of those gains came not in the use of the computer, but in the production of computers. Because the cost of an item of computing machinery was falling like a stone, and the quality was at the same time, the capacity at the same time was improving. And people were buying a lot of computers, so this was not a trivial industry….You got big productivity gains in the production of computers and whatnot. But you could also begin to see productivity improvements on a national scale that traced to the use of computers”.
Coyle’s central criticism is not just on the interpretation of the data, but on an interesting switch in Gordon’s argument:
“… throughout the first two parts of the book, Gordon repeatedly explains why it is not possible to evaluate the impact of inventions through the GDP and price statistics, and therefore through the total factor productivity figures based on them – and then uses the real GDP figures to downplay modern innovation.”
Coyle’s understanding of the use and abuse of GDP figures leads her to the fundamental point:
“While the very long run of real GDP figures (the ‘hockey stick of history’) does portray the explosion of living standards under market capitalism, one needs a much richer picture of the qualitative change brought about by innovation and variety. This must include the social consequences too – and the book touches on these, from the rise of the suburbs to the transformation of the social lives of women.”
To understand Coyle’s insights more deeply, her discussion with Russ Roberts gives a fascinating discussion of GDP (no, really!).
In my view, it seems to come down to differing views about where Moore’s Law – is taking us. The exponentially increasing computational power with increasing product quality at decreasing prices – has never happened at such a sustained pace before. The technological Great Inventions that Gordon sees as fundamental to driving sustained growth of the past all were bursts of innovation followed by a substantial time period where entrepreneurs figured out how to effectively commodify and deliver that technology to the broader economy and society. What is so interesting about the pattern of exponential technological progress is that price/performance gains have not slowed, even as some bits of these gains have just shown signs of commodification – Uber, 3D printing, biosynthesis of living tissue, etc.
There are good reasons to think that in the past we have failed to capture all the gains from innovation in measures of total factor productivity and labor productivity, as Gordon rightly points out. But if this is true, it seems strange to me to look at the current patterns of technological progress and not see the potential for these innovations to lead to sustained growth and increases in human well-being.
This is, of course, conditional on the political economy in which innovation takes place. The second cause for low future growth for Gordon concerns headwinds slowing down whatever innovation-driven growth there might be. Here I look forward to reading the relative weights Gordon assigns to factors such as demography, education, inequality, globalization, energy/environment, and consumer and government debt. In particular, I hope to read Gordon’s own take (and others) on how the political economy environment could change the magnitude or sign of these headwinds.
The review is worth a read in advance of what will likely prove to be an important book in the debate on development and growth.