The ideas of economists and political philosophers, both when they are right and when they are wrong, are more powerful than is commonly understood. Indeed the world is ruled by little else. Practical men, who believe themselves exempt from any intellectual influences, are usually the slaves of some defunct economist.
- John Maynard Keynes, 1936.
Recent work in growth economics combining endogeneous growth models with median voter political theory has produced the proposition that inequality can reduce economic growth in countries with democratic political systems. Empirical work based upon those models confirms that hypothesis across countries and through tame. This paper discusses and evaluates that literature, known as the "new political economy of growth."
Classical economists considered the relationship between the distribution of income across the social classes as a central determinant of the rate of accumulation; for example, David Ricardo  saw the distribution of the economic surplus (the social product less the wages of productive workers) between landlords and capitalists as the key to economic growth. Since landlords used their revenues for luxurious consumption, while capitalists used theirs for investment in productive capital, an increase in the ratio of land rents to profits would reduce the rate of accumulation and the increase in wealth. Karl Marx stressed the distribution of income flows going to labor and capital as determining the rate of accumulation and economic growth.(1)
That approach reappeared in the work of twentieth century heterodox economists such as Michael Kalecki , Nicholas Kaldor [1955-56], and Luigi Pasinetti ; all saw the distribution of income across social classes (factor shares) as an important influence over capital accumulation and growth because profits finance accumulation. Keynes  argued that distribution also influenced accumulation indirectly through its influence over aggregate demand for final goods and ultimately over profit expectations.
Until recently, however, orthodox neoclassical economists largely ignored the relationship between distribution and growth. The Solow  and Swan  approach to economic growth dominated growth theory for several decades; this approach abstracted from distribution across social classes and across individuals. A "representative agent" chose a savings rate; all savings were invested, and the rate of growth tended toward a steady state set by the exogeneously determined rate of technical progress. In the long run, the savings rate had no influence over the growth rate due to the assumption of diminishing marginal returns to capital. And since all agents were assumed to have the same savings rate, distributional shifts would not alter the savings (investment) rate.
At lower levels of abstraction in less formal models, neoclassical economists would observe that there was probably a trade-off between equality and growth: more equality would lead to less saving and investment and hence a lower growth rate (see any principles of economics text). In the long run, the benefits of economic growth would trickle down to the poor, and income distributions would move toward less inequality as societies became wealthier (the Kuznets hypothesis ).
Interest in the distribution-growth link has recently revived as the result of a new approach to economic growth theory: the family of "endogeneous" growth models that began to appear in the 1980s [Romer 1986]. The Solow-Swan models predict a convergence of growth rates among wealthy nations; as diminishing returns to physical capital set in, growth rates should converge to the exogeneous rate of technical progress (total factor productivity). Empirically this does not appear to happen; the wealthy, mature economies continue to grow at differential rates, and most of the poor countries do not catch up; growth rates diverge rather than converge [Barro 1991]. Romer and others argue that productivity and technical change are largely the result of intergenerational spillover effects of knowledge acquisition (investment in human capital), the benefits of which cannot be captured indefinitely by investing rent-seeking individuals and firms. Knowledge is endogeneously determined within the economic process since it is the result of investment in both human and physical capital (engineers, mechanics, and technicians learn by doing).
An implication is that income distribution can influence growth through its influence over the ability of individuals to access education and invest in physical capital. The effects of a change in inequality or of the degree of inequality at the beginning of a period of growth now became an object of theoretical...