The Pioneer Press gets confused between GDP per capita and GDP per worker

On Tuesday, we released our report The State of Minnesota’s Economy: 2018. In a generally fair review, Dave Orrick of the Pioneer Press wrote the following

Phelan cited data that has become popular with conservative economists, gross domestic product per worker. By that measure, Minnesota ranks 28th among the 50 states and Washington, D.C., and is well below the national average.

It’s in stark contrast to the figures cited by economists, including gross domestic product per capita. By that measure, Minnesota is indeed above the national average and ranked 15th.

There are two important things wrong with this.

GDP per worker is a standard measure of productivity, not a ‘conservative’ thing

First, GDP (or output) per worker isn’t a concoction of “conservative economists”, whoever they are. It is a standard measure of labor productivity used by economists of all stripes. Here is a rummage through my old undergrad and postgrad textbooks.

In his textbook Economics, David Colander of Middlebury College defines “labor productivity” as “the average output per worker” (p. 396). In the textbook Principles of Macroeconomics, Ben Bernanke of the Brookings Institution and Cornell’s Robert H. Frank define “average labor productivity” as “output per worker employed” (p. 101). Stephen D. Williamson of the University of Western Ontario defines “average labor productivity” and “aggregate productivity” as “the total quantity of output produced per worker” (p. 16) in his textbook Macroeconomics.

GDP per worker isn’t the only measure of productivity. In their textbook Economics, Richard Lipsey of Simon Fraser University and Alec Chrystal of the Cass Business School define productivity as “Output per unit of input employed” (p. 662). They define “labour productivity” as “Total output divided by the labour used in producing it, i.e. output per unit of labour” (p. 657). Paul Samuelson, late of MIT, and William Nordhaus of Yale write in their textbook Economics that “total output divided by labor inputs is labor productivity” (p. 671). Harvard’s N. Gregory Mankiw defines “productivity” as “the quantity of goods and services produced from each unit of labor input” (p. 13) in his textbook Principles of Macroeconomics. In his textbook Macroeconomics: Policy and Practice, Frederic Mishkin of Columbia defines “labor productivity” as “The amount of output produced per unit of labor” (p. G-6).

“Inputs” or “unit[s] of labour” could mean either a per worker of per hours worked measure. David Miles of Imperial College and Andrew Scott of the London Business School define “Labor productivity” as “Output divided by employment (either persons or hours worked)” (p. 591) in their textbook Macroeconomics: Understanding the Wealth of Nations. In their textbook Economics, Joseph Stiglitz of Columbia and John Driffill (a former professor of mine), currently at Yale, define “productivity” as “GDP per hour worked” (p. A15). Olivier Blanchard of MIT defines “productivity” as “Output per hour worked” (p. 28) in his textbook Macroeconomics.

We examine both measures in our report. We find that on a GDP per worker basis – that used by Colander, Bernanke, Frank, and Williamson – Minnesota’s workers lagged the national average by 7.8% in 2017. On a GDP per hour worked basis – that recommended by Stiglitz, Driffill, Miles, Scott, and Blanchard – Minnesota’s workers lagged the national average by 7.3% in the goods producing sector and by 6.9% in the services sector in 2017.

Have a look at these names; Bernanke, Frank, Samuelson, Nordhaus, Mankiw, Stiglitz. Are these guys “conservative economists” as opposed to regular “economists”?

GDP per worker and GDP per capita aren’t different measures of the same thing, they are measures of different things

Second, GDP per worker and GDP per capita are not different measures of the same thing. They are measures of different things. There is no mystery about the “stark contrast”. As David Weil of Brandeis University writes in his textbook Economic Growth (p. 143)

When we think about how productive a country is, it is often natural to focus on GDP per worker. However, when we ask how well off a country is, the more relevant measure is the amount of output that is available for every person in the economy – that is, GDP per capita. 

In their textbook Introduction to Economic Growth, Charles I. Jones of Stanford and Dietrich Vollrath of the University of Houston write (p. 6-7)

Perhaps per capita GDP is a more general measure of welfare in that it tells us how much output per person is available to be consumed, invested, or put to some other use. On the other hand, GDP per worker tells us more about the productivity of the labor force. In this sense, the first statistic can be thought of as a welfare measure, while the second is a productivity measure. This seems to be a reasonable way to interpret these statistics, but one can also make the case for using GDP per worker as a welfare measure. Persons not officially counted as being in the labor force may be engaged in “home production” or may work in the underground economy. Neither of these activities is included in GDP, and in this case measured output divided by measured labor input may prove more accurate for making welfare comparisons.

It is just not true that GDP (or output) per worker is a tool of “conservative economists” and that “economists” use GDP per capita instead. GDP per worker is a standard productivity measure used by economists who would run a mile from being labelled ‘conservative’. And the two measures are compliments, not substitutes, as economists say. One is a welfare measure, one is a productivity measure. It simply is not the case that you can pick which one you like depending on your ideological persuasion. You have to look at both to get the full picture. That is what we do in our report.

John Phelan is an economist at the Center of the American Experiment.