Replying to Critic on GDP and Intermediate Inputs

Includes: AGG, DIA, SPY
by: Michael Mandel

Karl Smith writes "In Which I Disagree With Almost Every Word Mike Mandel Says." It’s a long post (though not nearly as long as mine), and I just wanted to reply to two points.

First, Karl says:

The only way to get GDP wrong is either to miscount the number of goods and services sold in the U.S. or to misestimate the price index of final goods – not intermediate goods.

Um, no. This statement is simply wrong.

Gross domestic product = exports + gross domestic purchases – imports.

Imports, as it turns, out, include a lot of imported intermediate inputs (according to this piece in the February 2011 Survey of Current Business, "BEA estimates that about 40 percent of imported commodities are used as intermediate inputs by businesses"). So that getting the price index wrong for imported intermediate inputs slides right into GDP.

More fundamentally, remember that GDP is a value-added measure. However, the fundamental unit of observation for the BEA each quarter is revenues/shipments for various industries, which is a gross output measure. Then the BEA has the herculean task – which I never fully appreciated before – of figuring out how much of each industry’s revenues is final product, and how much is intermediate input. A simple example: Each month the revenues of law firms are reported each quarter by the Census Bureau. Part of those revenues are final product (personal consumption of legal services), and part are intermediate inputs (legal services to business). Taking the real growth rate of observed revenues as given, any error in estimating the real growth rate of intermediate inputs of legal services will translate directly into an error in estimating the real growth rate of personal consumption of legal services.

This sort of error does not wash out in final GDP. Consider the related question of whether R&D should be treated as business investment or as intermediate inputs. Currently, R&D is treated as an intermediate input, but the BEA has calculated that treating R&D as investment would boost real GDP growth.

Second, Karl says:

An improvement in the terms of trade, which is what Mandel is identifying, is a productivity improvement for U.S. workers. Its not based on U.S. innovation, but it does lead to more output per U.S. worker.

The question is an important one: Is a "terms-of-trade" productivity improvement equivalent to a "domestic" productivity improvement? Here I’m going to cheat: The short answer is that I’m about to finish a theoretical paper showing the specific sense in which they are not equivalent. But you will have to wait a couple of weeks for that one … I kind of overdid the last post.