The neoclassical theory of capital accumulation and growth under certainty for both positive and optimal savings functions has received extensive study in the literature for almost two decades. However, the study of capital accumulation under uncertainty began much later and these analyses for the most part confined themselves to linear technologies. In his pioneering work, Phelps [19] and, later, Levhari and Srinivisan [10] ,Hahn [5] and Leland [23], examine the optimal consumption-saving decision under uncertainty with a given linear production technology. Hakansson [6], Leland [9] and Samuelson [21] in discrete time and Merton [12, 13] in continuous time, along with a host of other authors, have studied the combined consumption-saving-portfolio problem where the production functions are linear, but where there is a choice among alternative technologies. There have been a few notable exceptions to this concentration on linear technologies. In a seminal paper, Mirrlees [17] tackled the stochastic Ramsey problem in a continuoustime neoclassical one-sector model subject to uncertainty about technical progress. Later, in [18], he expanded his analysis to other types of technologies. Mirman [16] for positive savings functions and Brock and Mirman [1] for optimal savings functions, using a discrete-time, neoclassical one-sector model, proved the existence, uniqueness and stability of a steady-state (or asymptotic) distribution for the capital-labour ratio. These steadystate distributions are the natural generalizations under uncertainty to the golden-age/ golden-rule levels of the capital-labour ratio as deduced in the certainty case. While these papers are important contributions with respect to existence and uniqueness, they have little to say about the specific structure of these asymptotic distributions or about the biases (in an expected-value sense) induced by assuming a certainty model when, in fact, outcomes are uncertain. The basic model used in this paper is a one-sector neoclassical growth model of the Solow-type where the dynamics of the capital-labour ratio can be described by a diffusiontype stochastic process. The particular source of uncertainty chosen is the population size although the analysis would be equally applicable to technological or other sources of uncertainties. The first part of the paper analyses the stochastic processes and asymptotic distributions for various economic variables, for an exogeneously given savings function, and deduces a number of first-moment relationships which will obtain in the steady-state. In addition, the special case of a Cobb-Douglas production function with a constant savings function is examined in detail and the steady-state distributions for
Read full abstract