Abstract
This paper shows that aggregate investment expenditure shares on traded and nontraded goods are very similar in rich and poor countries. Furthermore, the two expenditure shares have remained close to constant over time, with the average nontraded expenditure share varying between 0.54-0.60 over the 1960-2002 period. Combined with the fact that the relative price of nontraded goods correlates positively with income and exhibits large differences across space and time, our findings suggest that investment can be modeled using the Cobb-Douglas aggregator. The results of this paper offer a new restriction for the two-sector growth model. We apply the restriction to a study by Hsieh and Klenow (2003), which argues that differences in relative productivity between traded and nontraded sectors, ie., the Balassa-Samuelson effect, is the main cause of higher PPP-adjusted investment rates in rich countries. With the restriction imposed on the model, only around 25 percent of the differences in PPP-adjusted investment rates between rich and poor counties can be attributed to the Balassa-Samuelson effect.