On Manufacturing Employment and Structural Change

by Reihan Salam

Last week, David Henderson lamented the fact that the candidates for the Republican presidential nomination were lamenting the fact that manufacturing employment has declined, and he offered the following framework for thinking about the supposed problem:

It’s unlikely that we’ll have more manufacturing jobs. The story of economic progress is one of doing more with less. This has been especially true in U.S. manufacturing, where the value of manufacturing, in inflation-adjusted terms, reached an all-time high before our current recession. And while that has happened, we’ve had fewer and fewer people becoming more and more productive. Moreover, every major manufacturing economy, including China’s, has lost manufacturing jobs.

Henderson linked to a column by Tim Worstall on the same broad theme:

Think back 200 years, when 80% of the people had to work on the land to produce the surplus food to feed the other 20% who did everything that was not agriculture.

More recently, since I was a child, the increasing mechanisation of manufacturing (this is the same statement as the increasing productivity of labour in manufacturing) has meant that output from America’s manufacturing sector continues to grow while the number of people needed to do the manufacturing falls precipitately. This frees up that labour to move into services.

We couldn’t have 13 million people working in health care if all of them were still working in factories or down on the farm. And no, there’s no real evidence that service jobs pay less than manufacturing. The average (median) salary for a registered nurse was $63,000 back in 2008. That just isn’t chump change and there’s 2.6 million people doing that job. [Emphasis added]

I enthusiastically agree with Henderson and Worstall that there is nothing intrinsically wrong with shrinking manufacturing employment. As manufacturing grows more productive, it is entirely reasonable that we will employ fewer people in manufacturing. What concerns me, hence the highlighting in the passage above, is that the shift from tradable manufacturing to nontradable services has meant that a large number of workers are shifting from a highly productive sector to a markedly less productive sector, yet “there’s no real evidence that service jobs pay less than manufacturing.”

If we buy the Spence-Hlatshwayo thesis, this is a problem. Maggie McMillan writes:

When economists talk about structural transformation, they typically have in mind developing countries and the dual economy models à la W. Arthur Lewis that emphasize productivity differentials between broad sectors of the economy, such as agriculture and manufacturing. They don’t usually think about countries like the United States where this type of transformation has already taken place. But the figure below indicates that it is something we should be thinking about. The horizontal axis shows that between 1998 and 2007, the share of the labor force in manufacturing fell by around 3%. The share of the labor force in services (cspsgs) increased by a little more than two percent. The problem with this is that labor productivity in services is lower than economywide productivity (vertical axis) so this sectoral shift in employment lowers economywide productivity. Note that these changes took place before the Great Recession (the picture looks much worse for the period 1998 to 2009).

The cost of this transformation has not been well understood. For example, one often hears that the loss of jobs in manufacturing is no big deal because productivity in manufacturing is increasing and this will drive growth. But this argument ignores the economywide effects of labor reallocation. Another argument that is often heard is that we don’t need to worry about losing jobs in manufacturing because jobs in professional and business services are growing. But average labor productivity in professional and business services (not shown separately) is lower than average labor productivity in manufacturing. Using the March version of the Current Population Survey which follows workers over time, Ebenstein et al ( http://pluto.huji.ac.il/~ebenstein/) show that the majority of workers who leave manufacturing end up in the service sector where their wages are between 3 and 11 percent lower. [Emphasis added]

To revisit David Henderson’s point,

The story of economic progress is one of doing more with less. This has been especially true in U.S. manufacturing, where the value of manufacturing, in inflation-adjusted terms, reached an all-time high before our current recession. And while that has happened, we’ve had fewer and fewer people becoming more and more productive.

This is true — in the manufacturing sector. In the service sector, to overgeneralize for a moment, we’ve had more and more people not becoming more and more productive. In K-12 education, we have arguably had more and more people becoming less and less productive. In elite knowledge-intensive services, we’ve had a very small number of people becoming more and more productive.

The fixation on manufacturing employment as such is, I’ll happily agree, pretty silly. But I don’t think it is unreasonable to fret about the fact that productivity gains in the service sector are hard to come by, and that wage gains in the absence of productivity gains are difficult to sustain.

This, incidentally, is the underlying dynamic behind the battles over public sector compensation. Public workers use their political muscle to secure as much compensation as they can from legislators until taxpayers resist. I don’t think that taxpayers would be quite as exercised if improvements in the quality of public services were commensurate with compensation gains and the total tax burden.

All of this is to say that anxiety about manufacturing output could be a proxy for anxiety about the structural transformation McMillan describes, which, incidentally, is a prime reason why many leading IPE scholars like Barry Eichengreen believe that Chinese growth is may well run into a wall in the medium-term future:

Slowdowns are also more likely in countries where the manufacturing sector’s share of employment exceeds 20%, since it then becomes necessary to shift workers into services, where productivity growth is slower. This, too, is now China’s situation, reflecting past success in expanding its manufacturing base.

So the really interesting questions aren’t about manufacturing employment but rather about services innovation. Though I’m not convinced by Richard Florida’s proposed strategy for “upgrading” service sector jobs, I do think he’s hit upon the right problem:

Against the backdrop of a massive decline in once high-paying blue collar manufacturing jobs which is eerily similar to the decline of agricultural jobs a century or so ago, this third transformation is creating not one overall, but two distinct categories of jobs and employment.

The first category includes millions of the best jobs America has ever seen: high-pay, high-skill jobs in knowledge-based professional and creative fields. Almost a third of American workers now have these kinds of jobs, which pay more than double most manufacturing jobs and which have been rather impervious to unemployment. When unemployment among production workers climbed to more than 15 percent and surged above 20 percent for construction workers, unemployment among professional, technical and creative workers never got much above five percent.

But the second category, which comprises such routine service work as personal care assistants and home health care aids, retail sales clerks, and food preparers — is not so good. In fact, the pay for these jobs is roughly half that of manufacturing jobs. The result is as simple as it is tragic: a startling bifurcation of the job market and an increasingly unequal and divided society. Once we see this, it becomes clear that neither of the two most commonly cited prescriptions — the counter-cyclical approach to job creation by boosting investment and demand, or the path of educating more people for higher-paying knowledge-based jobs — can work.

Implicit in Florida’s argument is a call for more consolidation of the service sector:

[G]iant companies like Zappos and Whole Foods are only a small part of the picture. Most service firms are smaller, mom-and-pop operations. To bring them into the 21st century, the administration should develop strategies to help these smaller firms learn the advantages of seeing workers as sources of innovation and productivity gains. This could be a modest, low cost public-private partnership, involving universities, community colleges, and industry groups, modeled perhaps along the lines of the old Agricultural and Manufacturing Extension programs. The administration should also consider using incentives to encourage companies to upgrade service jobs, which would have the added benefit of improving the overall productivity of the highly fragmented service sector — the last great frontier of inefficiency in advanced economies — lifting the productivity of the economy overall, while boosting wages and lifting consumer demand.

I say implicit because Florida seems to think that mom-and-pops can mimic well-capitalized, national- and international-scale firms like Zappos and Whole Foods with the aid of federal programs and partnerships, which seems unlikely to me.

Rather, I believe that the upgrading of the service sector would flow from the following:

(1) further consolidation of retail, hospitality, food services, etc.;

(2) geographical concentration of workers in dense regions, thus facilitating the diffusion of skills and better matching willing workers with willing employers;

(3) rapid productivity and wealth gains at the top and near-top — i.e., an acceleration of the Plutonomy dynamic — that will raise the time-cost of household production and raise wages for workers providing in-person services.

But I think that short-sighted egalitarianism might undermine (3) while short-sighted building restrictions and transportation policies might undermine (2), which will weaken the drive for (1).

I should specify that with (3) I have in mind a number of interrelated phenomena: more spectacular fortunes, yes, but also growth in the ranks of the mass upper middle class.

The Agenda

NRO’s domestic-policy blog, by Reihan Salam.