New theories of `endogenous growth' address the implications of public finance decisions concerning the level of public services which directly or indirectly enhance private sector productivity but take little account of their distributional effects. In Discussion Paper No. 565, Research Fellows Alberto Alesina and Dani Rodrik consider long- term growth in political equilibrium by relating endogenous growth theory to the literature on majority voting on tax rates.

For an economy consisting of only `capitalists' and `workers', in which the government spends revenue from capital taxation on productive public services and direct transfers to workers, they show that only a government that attaches no weight to workers' welfare will maximize economic growth, which is inversely related to the weight placed on workers' welfare, even when workers are no less patient than capitalists. If workers are less patient, the optimal policy is one of falling taxation from an initially high level with a corresponding rise in economic growth from a low base; but this path cannot be sustained since governments have an incentive to renege on earlier commitments. For an economy of individuals with a continuum of differing ownership shares in the capital stock, where public finance decisions are made by majority vote, they show that increased inequality in the distribution of capital is associated with lower economic growth.
Alesina and Rodrik infer that democratic countries with uneven wealth distributions will grow relatively slowly, while for less democratic countries the weights policy-makers attach to different classes' welfare will determine the growth rate: `pro- capitalist' governments should be more conducive to growth than those that are `pro-labour'. They test this relationship by regressing average per capita GNP growth over 1960-85 for 67 countries including 24 democracies on 1960 per capita income, the 1960 primary school enrolment ratio and an income distribution variable (as a proxy for wealth distribution) dated as close to 1960 as the available data allow.

Alesina and Rodrik's empirical results are consistent with their predictions. For all countries taken together, the coefficients on the income distribution variables have the expected signs and are statistically significant in some cases; for non-democracies, the coefficients on the income distribution variable are insignificant; and for democracies, the income distribution variables are statistically significant with the expected signs. The other independent variables are also significant throughout: 1960 income has a statistically significant negative coefficient in every regression, indicating `convergence'; while primary school enrolment has a significant positive effect on growth.

Distributive Politics and Economic Growth
Alberto Alesina and Dani Rodrik

Discussion Paper No. 565, June 1991 (IM)