Investment, Exports and Growth

Investment, Exports and Growth Vinod Dubey THE very interesting analysis by Patnaik and Chandrasekhar (P and C) (EPW, January 6) provokes some reactions which might be of wider interest. At the very start, since cross-country analysis is sensitive to the countries included in the analysis and the time period, it would have been useful if the authors had said a few words about why they selected the particular 25 countries and the period. Surely GDP, GDI and export data are available in World Bank sources for more countries than the 25 selected and perhaps for later years than 1988. As it is, one is tempted to ask why Tanzania is included and its neighbour Kenya is not; why Ghana is included and neighbouring Cote d' Ivoire excluded. One is also struck by the exclusion of the Maghreb countries Algeria, Morocco and Tunisia. One wonders how sensitive are the findings to the broadening (or narrowing) of the list of countries? Using data in the World Tables 1994 (World Bank) and restricting the exercise to developing countries (a) with a GDP in 1992 of 5 billion 1987 US dollars or more, plus Paraguay which is below that threshold, (b) for which data are available in the World Tables for the period 1972-92, and (c) excluding the transition economies of eastern Europe and the island economics of Hong Kong and Singapore, I computed GDP growth rates and average investment to GDP ratios for 44 countries. These included 23 countries in the P and C study (with the exception of China for which the World Tables do not include investment data and Hong Kong) as well as 21 other countries (Table 1). The regression equation for the 44 countries for the period 1972-92 of the GDP growth rate G on the investment ratio I/Y was estimated to be G = 0.45+0.18 (I/Y). The estimated O/C coefficient is thus almost half the 0.32 estimated from the P and C sample. The R2 also drops to 0.245, which shows the I/Y ratio to explain only 25 per cent of variations in growth rates. This contrasts with an R2 of 0.72 in the P and C sample. It appears that the findings about investment and growth are quite sensitive to sample composition and time period.