Three minimal market institutions: Theory and Experiments

Markets are a society`s means to exchange goods. Our modern economy has developed numerous ways to interact - from stock exchanges over supermarkets to e-bay. In this project we define three minimal market institutions, examine their theoretical properties, and compare them to the outcomes observed in laboratory experiments with human subjects and in computer simulations. Minimal market institutions are so stripped down of details that it is not possible to simplify them any further without infringing on the basic phenomenon to be considered. In the early 1950ies Martin Shubik has defined several possible minimal market institutions, mostly distinguished by the strategy set (the number of choices to be made, see Shubik 1959, 1973). In this study we explore their properties in laboratory experiments and compare the results to theoretical predictions. The three basic price formation mechanisms we look at are listed here by the nature of the strategy sets in a single market for each trader: 1. The sell-all model (strategy set dimension 1) 2. The buy-sell model (strategy set dimension 2) 3. The simultaneous double auction model (strategy set dimension 4) Some of the experiments deal with credit limitations, others with monopolistic and oligopolistic power. We have theoretical predictions for prices, efficiency, and other market variables in non-cooperative and competitive equilibria (CE). As the number of players increases non-cooperative equilibria should converge to the respective CE. It is one of the main research questions whether this really happens. Since general equilibrium theory abstracts away from the market mechanism, it makes no predictions about how the paths of convergence to the CE may differ across market mechanisms. However, the pre-test we ran showed marked differences in efficiency between the three market forms. CE allows no role for money or credit, but we expect them to play a prominent role. In contrast to most market experiments conducted in open or partial equilibrium settings, we report on closed settings that include feedbacks. In our pre-test we find that the simplest market form, the sell-all model, generates significantly higher efficiency than the other two market forms. In this project we plan to expand our benchmark treatments presented above into four directions: in one series of experiments we vary the number of players on each market side from 2 to 10 to explore whether theoretical predictions (higher efficiency with more players) emerge and whether this happens in all three market settings. In the second series of experiments we will always have 10 players with 100 units of good A, but the number of players with good B will vary from 1 (monopoly) to 10. Here the focus is the impact of monopoly and oligopoly. In the third series of experiments the focus is on the impact of money holdings. We will vary the money endowment given to each participant. They suggest that up to a certain level this should influence prices, but above this level it should not. In the last series we explore the influence of different incentive structures.

How should markets be set up to work efficiently? How should incentives be set? Can an economy work efficiently without a central bank? Under which conditions will inflation or deflation occur? Why do people accept - essentially worthless - fiat money and how it its value supported or justified? In this project these and similar questions were explored with laboratory markets. We found that for the simultaneous trade in several goods a double auction is less efficient than other market forms, notable call markets set up as sell-all or buy-sell markets. We attribute this to the high complexity and the time pressure in double auction markets. In another experiment we explored whether an economy can function with money printed and issued by individuals. With a perfect clearinghouse and no possibility to renege on promises to deliver goods this is the case. However, in a second set of treatments, where agents have the option of not delivering on their promises, a high enough penalty for non- delivery is necessary to ensure an efficient market; a lower penalty leads to inefficient, even collapsing, markets due to moral hazard. In a subsequent experimental series we explore whether different levels of default penalties can be employed to select from multiple equilibria. Real economies may have multiple equilibria, a fact that is usually ignored in macroeconomic models. We think default and bankruptcy laws are required to prevent strategic default, and these laws can also serve to provide the conditions for uniqueness. Our data show that the choice of default penalty takes the economy to the neighbourhood of the chosen equilibrium. The theory and evidence together reinforce the idea that accounting, bankruptcy and possibly other aspects of social mechanisms play an important role in resolving the otherwise mathematically intractable challenges associated with multiplicity of equilibria in closed economies.