Will our children really not know economic growth?

February 2nd, 2016

Summers writes a review of Robert Gordon’s The Rise and Fall of American Growth: The US standard of living since the Civil War in the February issue of Prospect Magazine.

Robert Gordon’s The Rise and Fall of American Growth is an extraordinary work of economic scholarship. At a time when too much of the economics profession prioritises theorising about small issues, Gordon provides new data bearing on what may be the most important economic question of all—what will economic growth be like over the next couple of generations? Moreover, this is one of the rare economics books that is on the one hand deeply analytical, with over 100 figures and tables, and on the other a pleasure to read: it is chock full of anecdotes about everything from flying out of Chicago’s O’Hare airport in the 1970s to the spread of radio in the 1920s to the travails of pharmaceutical research. Pick any random page and you will learn something interesting about American life.

It has been said that the further forward you want to forecast, the further back you have to look. Gordon’s interest in this volume transcends the business cycle and even momentous fluctuations like the Great Recession that followed the 2008 financial crisis. His interest is in what kind of growth in living standards average people entering the labour force are likely to see during their working lives.

Gordon lines up with Silicon Valley entrepreneur Peter Thiel, who has said that, “We wanted flying cars, instead we got 140 characters.” Despite all the hype coming out of Silicon Valley, Gordon believes we are in a period of modest progress. Whereas many observers worry that because of technology there will no longer be work for an increasing share of able-bodied adults, he thinks there will be work for all, but very little increase in productivity. Disturbingly, his reading of history and his assessment of a variety of factors in the current environment that he calls “headwinds” lead him to the judgement that, “headwinds are sufficiently strong to leave virtually no room for growth over the next 25 years in median disposable real income per person.”

The strongest part of Gordon’s book is his evocation of the remarkable 50-year period between 1920 and 1970. My grandmother was born around the turn of the century and died in the mid-1970s. Reading Gordon’s chapters, I was reminded that over her adult lifetime she saw the flush toilet, electricity for lighting and central heating go from being luxuries enjoyed by a quarter or less of the population to becoming universal. She saw radio come into being and then be supplanted by black-and-white and ultimately colour television. She saw air-conditioning, washing machines, dryers and refrigerators go from non-existent to universal. Over her lifetime, transportation went from meaning walking, riding a horse or taking some kind of train to being primarily based on cars and aeroplanes. When she had my mother in 1922, infant mortality was 75 per 1000 and large families could expect to suffer an infant or childhood death. When her youngest grandchild was born in the 1960s, infant mortality was below 20 per 1000 and life expectancy had risen by more than a decade.

It is striking to contrast the changes during my grandmother’s lifetime with those during mine. I have seen the microwave become universal in American kitchens. Automotive air-conditioning has gone from common to universal.

A much wider range of TV programmes are now available and with a much sharper picture. There is a wider array of healthy foods. And, of course, I carry a smartphone that keeps me more connected to information sources, friends and colleagues than was imaginable 50 years ago.

But whereas my grandmother would have been at sea if returned to her girlhood home, I would miss relatively little if suddenly placed in the home I grew up in. It takes longer and is less comfortable to fly from Boston to Washington or London than it was 40 years ago. There are more highways now but much more traffic congestion as well. Life expectancy has continued to increase, though at about half the pace it did during my grandmother’s day. But the most important transformation—child death being an extraordinary event—had already happened by the time I was born.

Gordon’s most compelling argument (although this is not how he puts it) is that the greatest generation is in some ways also the luckiest generation. He provides an onslaught of statistical evidence and carefully considered anecdote in support of the idea that the 1920-1970 period stands out as a period of extraordinary progress in the annals of economic history. While Gordon provides rich discussions of the impact of the Depression, the Second World War and much else, his basic explanation for the remarkable character of the 1920-1970 period is simple. Certain kinds of progress can happen only once. Living in a controlled climate, having access to indoor plumbing, largely eliminating child mortality, controlling infectious disease, and being able to communicate immediately in the absence of physical presence are all examples of transformations that can be built on and improved but seem hard to replicate.

Of Gordon’s many charts and graphs the one that impressed me most shows the growth in officially measured total factor productivity by decade from 1900 to the present, pictured above. It steadily escalates from 1900 to 1950 when productivity grew at almost 3.5 per cent and then unsteadily declines to the point where it has averaged well under 1 per cent in the generation since 1990.

Many rightly wonder about mis-measurement of productivity as new products become available and quality improves. Gordon is compelling in arguing that productivity growth is indeed significantly underestimated. He is also more persuasive than I expected in arguing that, if anything, this understatement was greater decades ago than it has been recently. In part this is because there were more of these transformational changes that are inherently hard to assimilate in standard frameworks. In part it is because the statisticians do a much better job than they once did of taking account of quality change.

Gordon’s aspirations go far beyond writing history. He also seeks to describe the current epoch and to forecast the future.  Gordon confronts and largely rejects the views of those he calls the techno-optimists like MIT scholars Erik Brynjolfsson and Andrew McAfee and most of Silicon Valley where Marc Andreessen’s view that “software is eating the world” is widely shared. This optimistic view points to the internet and developments like driverless cars, robotics, 3D printing, and artificial intelligence to argue that we are in the midst of a vast economic transformation in which much of the work currently being done by people will be mechanised leading, presumably, to huge gains in output per worker while raising profound questions of job availability and inequity.

Gordon acknowledges a “third industrial revolution” built around software and IT but argues that it is much less significant than the “second industrial revolution” of the mid-century because its impacts are largely confined to telecommunications and entertainment. He further argues that it may already be largely played out, and that in any event concerns of technology replacing jobs are not new; workers displaced by new technologies tend to find new jobs, often in sectors created by those new technologies.

Gordon is right in pointing to the huge disjunction between the techno-optimist narrative and the productivity statistics. It is hard to see how technology could be displacing huge numbers of workers without raising measured productivity. And if its effects are so pervasive as to lead to large shifts in the distribution of income with innovators capturing huge rents, why do we not see more evidence of increased output?

I do not have compelling answers to these questions. Yet it is hard to shake the sense that something unusual and important and job threatening is going on. The share of men between the age of 25 and 54 who are out of work has risen from about 5 per cent in the 1960s to about 15 per cent today as it appears that many who lose jobs in recessions never come back. Perhaps, as Gordon seems to suggest, this is a sociological phenomenon or a reflection of increasing problems in education. But I suspect technology along with trade have played an important role.

Economic historians like Paul David have long noted that it took surprisingly long for the benefits of electricity to show up in the economic statistics. Creative destruction takes time. Think about a canonical major innovation like the supermarket. Initially when it is first introduced all the other shops remain in business with reduced volumes. Measured productivity—defined as total sales per retail worker—go down as employment in retailing rises and total sales remain roughly constant. Only with the passage of time and the closing of traditional shops will measured retail productivity increase. If this story is playing out in many different sectors on different rhythms, overall productivity growth could be relatively slow even as there is substantial job losses in sectors further along in the adjustment process.

The question of how to square developments that are large enough to have a major impact on wage and employment patterns with the paucity of measured productivity growth looms for future research. But it is not only the expectation of slow total factor productivity growth that informs Gordon’s pessimistic belief that median incomes will be stagnant over the next generation. He points to reduced growth in average educational attainment, rising inequality, an ageing population, a growing national debt and breakdowns in family structure as headwinds that will further slow growth.

I wish that I could convincingly rebut his claims. While there is room for argument—for example, with his analysis of fiscal policy (which gives too little weight to the substantial reduction in debt carrying costs) or his confidence that inequality will continue to rise—his broad point seems likely valid.

While as already noted, I find Gordon persuasive in his claim that the slowdown in productivity growth is not a figment of mis-measurement, the fact that measured median incomes will be stagnant does not mean that most people will not see rising standards of living over time. Incomes rise as people get further into their careers. And quality improvements and new products are improving life in ways that do not show up in economic statistics, though possibly less so than in the past. So it would be a mistake to regard our children as condemned to economic stasis even before considering Gordon’s various ideas for accelerating growth.

Economic frustration is a central challenge of our time. It is surely intimately connected with political dysfunction, loss of faith in institutions and much else. Like most things, it is best viewed with historical perspective. There is no better way to get that perspective than by reading Robert Gordon’s landmark work.

Click here for the Prospect article.