Affiliations: [a] Department of Economics, University of Pennsylvania,
Philadelphia, USA | [b] Economics Department, Daiwa Institute of Research,
Ltd., Tokyo, Japan
Correspondence:
[*]
Department of Economics, McNeil Bldg, Room 335,
University of Pennsylvania, 3718 Locust Walk, Philadelphia, PA 19104-6297, USA
Abstract: Through many comprehensive revisions of the National Income and Product Accounts
(NIPA) of the US, a significant discrepancy has persisted, namely that between
the estimate of the headline total now called Gross Domestic Product (GDP) from
the side of expenditures or from the side of imcome payments. This discrepancy
is not trivial (now in the neighborhood of -\$100^+ bn.); it is not random; it
is wrongly attributed exclusively to the income side estimates. There have been
interesting proposals for systematically allocating it among NIPA entries
according to some statistical rule. Students of the discrepancy, over the years,
have noted systematic variation with respect to international trade, inventory
investment, total output and other variables. In recent years, the income side
total has given a different estimate of the historical rate of change of output
per worker, obviously an extremely important statistic. In a fresh examination
of the discrepancy through 1996 (quarterly), we find, in this paper, suggestive
correlation with business earnings, itself a very important but difficult
magnitude to measure. If the NIPA data are to be used in an important way for
policy guidance, a more careful treatment based on economic and statistical
analysis is called for.
