Created 8/13/1997
Go to Brad DeLong's Home Page

How Do We Measure Economic Growth?

Getting an unambiguously "correct" measure of overall economic growth--a single number to describe a complex, multi-dimensional process of change and progress--is inherently impossible: as impossible as accurately and without distortion drawing a two-dimensional flat map of the globe.

Simon Kuznets knew this, and when he began the measuring-economic-growth business he laid down certain rules of the game:

What are the implications of these--necessary and explicit--assumptions?

Begin thinking about it by moving from our real world into one of economists' two-dimensional worlds, one with "high tech" and "low tech" goods, and with a "past" and a "today":

Following Kuznets, we measure economic growth by valuing past production at today's prices--the equivalent of stuffing output in the past into a time machine, bringing it forward to the present, and then selling it.

One big problem with Kuznets' concept: it is psychologically false. Tell someone that real productivity in the past was $X per worker per year, and they think of what they could buy today with $X per year--the blue dot in the figure above. But that bundle of goods and services could not be produced in the past. Instead, the past was at the black "past production" dot--a bundle of goods and services that (from the point of view of today's consumer) was vastly inferior to what we could (and would choose) to buy at today's prices.

Kuznets' valuation thus misses a big part of economic growth: the red arrow in the figure above Our welfare today is greatly enhanced by the invention of new goods and new types of goods, and by changes in relative prices and thus opportunities that have no impact on Kuznets' estimates of long-run economic growth--but a big impact on human welfare.

If we know what utility functions are, then we can begin to quantify (or at least measure gaps) between (i) welfare in the past, (ii) welfare at today's prices and production possibilities with Kuznets' estimate of real production in the past, and (iii) welfare in the present. But we don't know what utility functions are. And so Kuznets ducked this issue not because he thought it unimportant but because he was after a concept that he could measure.

If we do know utility functions, then we can form a better measure of real income in the past: what income in the present would have to be to make a consumer indifferent between living today and living in the past. But this measure will not be independent of today's prices. And what if utility functions change over time?



Created 8/13/1997
Go to Brad DeLong's Home Page

Professor of Economics J. Bradford DeLong, 601 Evans, #3880
University of California at Berkeley
Berkeley, CA 94720-3880
(510) 643-4027 phone (510) 642-6615 fax