A computational theory of willingness to exchange

A new model of exchange is presented following Marr's conception of a "computational theory". The model combines assumptions from perceptual theory and economic theory to develop a highly generalised formal model. The approach departs from previous models by focussing not on how ownership alters preferences, but instead on difficulties inherent in the process of exchange in real markets. Agents treat their own perceptual uncertainty when valuing a potential exchange item as a signal regarding the variability of potential bids and offers. The analysis shows how optimising agents, with no aversion to risk or loss, will produce an endowment effect of variable degree, in line with empirical findings. The model implies that the endowment effect is not a laboratory finding that may not occur in real markets, but rather a market phenomenon that may not occur in the laboratory.