Retrading,production, and asset market performance

Prior studies have shown that traders quickly converge to the price–quantity equilibrium in markets for goods that are immediately consumed, but they produce speculative price bubbles in resalable asset markets. We present a stock-flow model of durable assets in which the existing stock of assets is subject to depreciation and producers may produce additional units of the asset. In our laboratory experiments inexperienced consumers who can resell their units disregard the consumption value of the assets and compete vigorously with producers, depressing prices and production. Consumers who have first participated in experiments without resale learn to heed their consumption values and, when they are given the option to resell, trade at equilibrium prices. Reproducibility is therefore the most natural and most effective treatment for suppression of bubbles in asset market experiments.Contrary to rational theory the first asset market experiments with complete common information on fundamental value exhibited unexpectedly strong tendencies to yield price bubbles (1). The results, however, were soon extended (2, 3), independently replicated (4–8), and have quite important consequences for improved understanding of the sources of instability in the economy (9–11).In contrast with these robust asset market findings, earlier experiments had established that repeated trade across time periods in static supply and demand experiments yielded efficient rapid convergence to rational competitive equilibrium outcomes under strictly private decentralized information (12). In the supply and demand experiments, however, trades were for immediate consumption, as with hamburgers and haircuts; as noted in ref. 13, items could not be retraded and individuals knew in advance that they were specialized as buyers or sellers and could not switch roles depending on price as in asset markets. [These features also characterize nondurable goods and services in the US national accounts and represent approximately 75% of private product. Instability in the US national accounts arises from the remaining 25% (11)].Motivated by the glaring contrast in these two kinds of markets, Dickhaut et al. (13) reformulated the traditional supply and demand environment by explicitly modeling two goods: “cash” as a means of payment and a “commodity” that had a heterogeneous end-of-period “dividend” consumption yield value. This formulation exactly parallels the asset trading environment, except that cash and commodity endowments have only a one-period life and dividends are not common. Individual subjects received diverse endowments, but in each of 10 periods a subject was endowed with the same amounts of cash and commodity, thus inducing pure static supply and demand conditions across 10 periods. This reformulated framework allowed the study of convergence in a 2 × 2 design, (retrade, no retrade) × (low cash, high cash). Convergence was markedly slower in retrade vs. no retrade because traders who could retrade had difficulty learning from market prices their optimal role as buyers or sellers. High cash exacerbated this difficulty relative to low cash, in line with previous findings (14).Haruvy et al. (ref. 15, p. 7) observe that because durable goods can be retraded they carry an implicit option value for resale at a higher or lower price depending on individual expectations. [Harrison and Kreps (16) also demonstrated this possibility in a theoretical model.] A perishable good has only consumptive utility value, if purchased, whereas a durable good will yield consumptive value if retained, but a resale value if not. Thus, durable goods can involve a speculative motive whereas nondurables cannot. It is this speculative motive that has driven price bubbles in studies of asset markets.Previous studies have modeled a durable asset whose life extends over the entire horizon of the experiment; in those studies—unlike house or automobile markets, for example—no new units were produced and existing units did not depreciate during the course of the experiment. [However, Haruvy et al. (15) study a securities market subject to external interventions to repurchase existing shares that reduce the asset stock, or issue new shares that increase the stock.] In this paper we model producers who at private unit cost may elect to sell newly produced units that add to the existing stock; also, we introduce an elementary “replacement demand” for asset units: The existing units “depreciate” with a constant probability, independent of age and price. Hence, there is a net addition (or loss) of asset units in the market depending on whether production is greater than or less than the depreciation rate. We set the stage for experimental studies of asset market trading characterized by dynamic stock-flow trajectories over time with the stock determined endogenously. [Abstract general stock-flow models were introduced by Clower (17) for capital asset pricing and Smith (18) for the theory of the firm.] In our new experiments we find that resale inhibits price discovery, which in turn distorts production and retards efficiency. However, whereas resale has generated price bubbles in asset markets without production, in our asset markets resale suppresses prices well below equilibrium. This is because many consumers fail to specialize as buyers, and instead compete with producers as sellers. When we rerun the resale experiments with subjects who are experienced as specialized buyers and sellers in the no-resale treatment prices and production converge to equilibrium and efficiency improves. Reproducibility is therefore the most natural and most effective treatment for suppression of bubbles in asset market experiments. [Hong et al. (19) develop a theoretical model in which release of locked up shares can suppress a price bubble following an initial public offering.]