The Sensitivity of International Comparisons of Capital Stock Measures to Different "Real" Exchange Rates

Gross national investment flows can be discounted and accumulated to form measurements of national capital stocks for cross-national comparisons of capital abundance. The formation of these capital stock figures can depend substantially on the method by which investment measured in different currencies is translated into real investment figures that are comparable across time and across countries. Three different methods for forming time-series of real investment figures are considered here. One of these methods translates foreign currencies into dollars in each year using the current exchange rate and then divides by the U.S. deflator for gross investment to form a series on real investment. The second method accumulates real investment in the local currency deflated by the home country price deflator for gross national investment and then translates into dollars using the base year (1966) exchange rate. The third measure of capital substitutes the purchasing power parity rates from Robert Summers and Alan Heston (1984) in place of observed exchange rates. These three capital stock comparisons are contrasted and used here in a latent variable model of trade. The conclusion from this data analysis is that although the measures of capital stock can differ substantially for some countries, these differences matter very little when capital stock is treated as one variable in explaining the composition of trade. The data do suggest some slight preference for the third measure of capital based on the purchasing power parity ex