Of all the intricate problems which beset the path of those who endeavour to evolve a dynamic theory of economic equilibrium, none is more formidable than that presented by the existence of different velocities of adjustment of different data. Adopting Mr Kaldor's terminology we may define as 'velocity of adjustment' 'the time required for a full quantitative adjustment to take place (either on the supply-side or on the demand-side) corresponding to a given price-change, i.e. the time elapsing between the establishment of a certain price and the full quantitative adjustment of that price.' 1 Now, it goes without saying that, where the velocity of adjustment of some datum is so small that, while its adjustment is taking place, other data are liable to change, equilibrium becomes hopeless and only 'perfect foresight' seems capable of saving the situation. It has been pointed out in recent discussions, 2 however, that 'perfect foresight' cannot be regarded as a notion suitable for a starting-point of a science which (besides the drawing of so-called 'indifference-curves') is concerned with the explanation of human actions. The question therefore arises whether a satisfactory theory of dynamic equilibrium, i.e. a theory describing the movement from one equilibrium to another as a process in time, can be based upon conditions embracing imperfect foresight.
If we confine ourselves to the study of cases, where data are constant, but velocities of adjustment differ, it becomes clear that even with imperfect foresight a stable equilibrium may be reached, if foresight is inversely proportionate to velocity of adjustment, i.e. where those have the greatest foresight whose adjustment takes the longest time. If, e.g., demand reacts instantaneously to price-