Differences in Relative Prices and International Differences in Growth Rates

Abstract

The new growth literature, following Paul M. Romer (1986) and Robert E. Lucas, Jr. (1988) has formulated models in which growth rates can differ systematically across countries. A large empirical literature has used cross-sectional techniques to examine the cause of differences in growth rates.' This paper examines whether this work has used data which have been constructed in such a way as to introduce inadvertently a spurious correlation between growth and income levels. This effect occurs because relative evolve because of different rates of technological progress in different industries. Many of the new growth models assume single goods or goods produced by the same technology, so this phenomenon has so far escaped analysis.. If technological progress is more rapid in some sectors than others, relative will change during the process of development. The change of relative has two effects on measured growth rates. First, the choice of base affects the growth rate; this phenomenon is sometimes called the effect. Second, using any base or weights introduces a spurious correlation between income and growth. Less developed countries would tend to have lower growth rates, not because they grew more slowly, but because of data construction techniques. For the sake of simplicity, these phenomena will be called the Gerschenkron (or the level) effect and the spurious-correlation effect.3 This paper examines the International Comparison Program (ICP) and the Penn World Tables (PWT), which are based on the ICP. The Penn World Tables are the data most widely used in empirical studies of differences in economic growth. Both the ICP and PWT use international prices to adjust for differences in the purchasing power of currencies. These most closely resemble Hungarian prices. Using such tends to produce time series with lower growth rates than those in the national accounts for countries less developed than Hungary; conversely, the growth rates would be higher for countries more developed than Hungary. However, the growth rates of poorer countries are not systematically lower in the Penn World Tables. The paper examines why the growth rates are not distorted and makes some specific recommendations for data construction and use. * Virginia Polytechnic Institute and State University, Blacksburg, VA 24061-0316. Part of this work was supported by a grant from the Institute of International Studies at Brown University and by a summer internship at the Division of International Finance at the Federal Reserve Board. I am grateful to Peter Garber, David Weil, Oded Galor, and anonymous referees for helpful comments. Alan Heston and Robert Summers have also helped me understand ICP and PWT methodology. Of course, I alone am responsible for any mistakes in this paper. IExamples include Steve Dowrick and Duc-Tho Nguyen (1989), Robert J. Barro (1991), J. Bradford De Long and Lawrence H. Summers (1991), and N. Gregory Mankiw et al. (1992). 2A series of papers by Christina D. Romer (1986a,b, 1989) has already emphasized the importance of data construction techniques in another context. 3There is a third effect, a real acceleration effect caused by shifting consumption patterns and the reallocation of resources between sectors. Growth accountants such as Angus Maddison (1987) adjust estimates of total factor productivity for this sort of shift. William J. Baumol et al. (1989) have also referred to this phenomenon. For an analysis in terms of a simple multisector Solow model, see Nuxoll (1992a).

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 123 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.