While measuring the benefits from liberalisation is obviously important, from a theoretical as well as from a political point of view, one may question the validity of the presently widespread Walrassian approach. The latter does not take account of the fact that supply is just as must (and perhaps more) risk as (average) price responsive. Now, it turns out that if (contrary to the strong versions of the "rational expectation hypothesis"), producers do not take their decisions on the basis of equilibrium prices only, then the market, instead of equilibrium, can generate very harmful seemingly random fluctuations. These considerations are incorporated into a partial equilibrium model of the world sugar industry, and a GTAP style general equilibrium model of the world economy. In both cases, liberalisation, preventing price stabilisation policies, results in more instability, and a decrease in welfare. Such an outcome is specific of agriculture and other low price elasticity commodities.