PARADIGMS OF ECONOMIC DECENTRALISATION 51 which the demise of Keynesianism itself would belie Keynes9 optimism; it also relates to whether socialisation of ownership is even a sufficient condition for socialisation of investment. A world in which autonomous state-owned units relate to one another through the market and face "hard budget constraints", i.e. a world conforming to the vision of "market socialism", would also be a demand-constrained system, and indeed reproduce many of the features of the capitalist market economy, as Maurice Dobb had argued long ago. This is hardly surprising: both the conditions mentioned above which make a system demand-constrained in Kornai^s sense would be satisfied in a decentralised economic regime that mirrors "classical capitalism", even if the means of production are state-owned. Keynesianism was not the only example of the "decentralisation-produces-suboptimal-equilibrium" argument. Ragnar Nurkse had produced a very similar argument in support of his case for a big push to investment to achieve "balanced growth" in an underdeveloped economy. Even though if all production units undertook an expansion in their scale of operation, they would ipso facto enlarge their collective market, each production unit is deterred from enlarging its operations because such enlargement would not create an equivalent market for its own product. As a'result, in the absence of collusion, the scale of production as well as the size of the market remains low. The economy is trapped in a suboptimal equilibrium because of decentralised decision-making. Nurkse's "big push" required therefore an intervention in decentralised decision-making. The reason for my going back to this old debate is not just to provide an intellectual backdrop to our discussion. It is because even as decentralisation in the sense of this conference ("government-level decentralisation") is finding adherents, there is a simultaneous effort by the votaries of "liberalisation" to promote decentralisation in the classical sense ("firm-level decentralisation"), against which Keynes and Nurkse, not to mention the Marxist tradition, had argued so forcefully. Not only do these two kinds of decentralisation have to be distinguished strictly (I shall come to this later), but the case for supporting "government-level decentralisation" must be distinguished from the "liberalisation" argument which in my view is wrong inter alia precisely for the reasons just mentioned, namely that it seeks to institutionalise the "irrationality" flowing from supposedly "rational" decision-making at the micro-unit level. This is no idle theoretical claim. Indeed the "irrationality" already