April 12, 2002

Productivity Discrepancies

The rate of measured economic productivity growth in the United States accelerated from 1.5 percent per year over 1990 to 1995 to 2.5 percent per year over 1995 to 2000. Or did it? Official productivity statistics are fundamentally based on the "production" side of the national income and product accounts. But there is another, "income" side of the national income and product accounts that is also used to measure economic growth. The "production" and "income" sides should tell the same story: a few technical adjustments aside, they are supposed to be identical by design. Their difference is the "statistical discrepancy"--a measure of the extent to which the two sets of measurements of what is supposed to be the same thing disagree. And this statistical discrepancy grows large in the late 1990s.

Suppose you look not at the production but at the income side. You find that the acceleration of economic growth is considerably faster. The acceleration of economic growth from 1990-1995 to 1995-2000 is not 1.0 but 1.6 percentage points per year. That means that the "new economy" is more than half again as large by the income-side measurements as by the product-side measurements. That means that today the U.S. economy is some 3.6% bigger according to the income side than it is according to the product side.

This large statistical discrepancy is unprecedented. Before 1995 the income and product side measures tended to run remarkably in harness.

Which of the two measures gives the correct picture of what was going on in the second half of the 1990s? I don't know. Neither does anybody else that I know. Martin Baily, who taught me all this, recommends splitting the difference between the two measures. That seems to me to be a good thing to do.

Posted by DeLong at April 12, 2002 03:53 PM | TrackBack

Post a comment