In this paper, I propose a resource allocation (planning) process designed for an economy where increasing returns to scale prevail. The main feature of the process is to supplement the market mechanism by the central allocation of investment funds. It is well known that the unmodified market mechanism fails to solve optimal allocation problems, if non-decreasing returns to scale play a very important role in the economy. In order to meet the difficulty, Arrow and Hurwicz [1] introduced price speculation into a tatonnement process. Although locally convergent and informationally most efficient,2 their process is not entirely satisfactory especially from the motivational point of view. An alternative line of attack on optimization problems under non-decreasing returns may be found in rationalizing traditional planning methods of the central allocation of material quotas. This was first done by Kornai and Liptak [3] for the case of linear technologies. Later, in this Review, Heal [4] proposed a planning process based on essentially the same idea, but considerably generalized so as to cover non-convex technologies. In his process, each manager of a production process is instructed to report to the Central Planning Board (referred to as the CPB hereafter) the shadow prices of every input centrally allocated. In the light of this data, the CPB proposes a new allocation of inputs in which, by comparison with the previous one, resources have been shifted towards the uses where higher shadow prices are given. In this way, Heal has succeeded in defining a planning process with certain desirable properties, such as monotonicity and well-definedness3 in Malinvaud's sense [6], which the Arrow and Hurwicz process lacks. However, there are still some points to be improved on in this method: