Introduction

The GDyn model presented in Chapter 2 inherits from the standard GTAP model its specification of the regional household demand system and, in particular, the treatment of saving. As in the standard GTAP model, regional households in GDyn spend their income according to a Cobb-Douglas per capita utility function specified over three sources of utility: private consumption, government consumption, and real saving. Because the model is not forward-looking but is recursively dynamic, the utility function is static – it represents utility from present but not future consumption. The practice of including saving in the static utility function derives from Howe (1975) and allows regional households to value saving in a temporal settings. Because of the Cobb-Douglas functional form, the average propensity to save is fixed, and saving is a fixed proportion of income in each region.

There are several unwelcome implications of this assumption. Because propensities to save are fixed and incomes are rising over time, countries in which saving substantially exceeds investment, like Japan, accumulate unrealistically large stocks of foreign assets. If like China they also exhibit high rates of growth in income, at the end of long-run GDyn simulations that span many decades, such countries may end up owning a large part of the wealth of the whole world. Although such outcomes cannot altogether be ruled out a priori, they are very strong predictions resting on a very weak empirical basis.