“Persistent Inequality ”

Dilip Mookherjee and Debraj Ray
IED Discussion Paper 108, September 2000

The neoclassical growth model and its reformulations in the context of intergenerational mobility predict that in the absence of persistent random shocks, the market mechanism promotes a tendency towards convergence of incomes across different agents, families or countries. In contrast, a recent literature based on capital market imperfections generates opposite predictions about the persistence of inequality and the permanent impact of one-shot redistributive policies. These models are set in a wide variety of contexts including labor markets, occupational choice, and human capital. The features they typically share in common are assumptions concerning indivisibilities in investment, occupation or locational choices, and savings behavior not based on dynamic optimization of long run utility.

Mookherjee and Ray’s paper provides a general setting that encompasses several different modeling approaches, as well as extending it in new directions. Their model allows a flexible formulation of the degree of divisibility of investment choices, varying from perfect indivisibilities to perfect divisibility. They allow for pecuniary externalities in the returns to and the costs of human capital accumulation, thus allowing investment thresholds to arise endogenously

from the operation of the price mechanism. In particular, both returns and costs of investments depend on relative prices. Under weak conditions (e.g., the existence of at least two professions that are distinct in terms of training cost and returns) they show that long run inequality is inevitable, despite savings motivated by long-run optimization, a convex technology, and absence of any random shocks.

However, the conclusions concerning history dependence of the long-run wealth distribution (i.e., multiplicity of long-run steady states) do depend on the divisibility of investments. Greater occupational heterogeneity restricts the multiplicity of steady states, with a unique steady state in the case of perfect divisibility. An important implication of these results concerns the impact of one-shot redistributive policies: these have long-run impact only if there is significant indivisibility in investments.

The authors subsequently study the question of convergence to steady states and comparative dynamic effects of redistributive policies in a two occupation context. Finally, they enlarge the model to encompass multiple forms of (financial or physical) capital, and show that it is possible for ownership of human and nonhuman capital to be negatively correlated in a manner that eliminates inequality.