ABSTRACT
This chapter describes convergence dynamics in double auctions for a stochastically lived asset. Rapid, reliable convergence to competitive equilibrium has been observed in most experimental double auction markets for commodities or single-period assets. Trading volume and allocative efficiency were often low during convergence. The chapter reports market experiments in which subjects’ trade stochastically lived assets that pay a dividend each period and live from period to period with a known probability. It finds that prices converge slowly to the discounted dividend value; experienced subjects traded at prices close to the discounted dividend value. Without "swingback" learning, as in earlier experimental double auctions for multi-period assets, convergence to discounted dividend value was very slow. The double auction does not fail completely at generating convergence to competitive equilibrium prices for stochastically lived assets—convergence occurs with experienced traders—but the auction mechanism performs much more slowly and erratically than in simpler settings.
