Structural Econometric Tests Of General Equilibrium Theory On Data From Large-Scale Experimental Financial Markets

We develop structural econometric tests of asset pricing theory for application to data from experimental financial markets. The tests differ from those used in the analysis of field data because they verify the consistency between prices and allocations, as opposed to merely testing whether only prices satisfy equilibrium restrictions. Our tests also differ from standard field tests because the experimentor controls (knows) the very parameters that need to be estimated in the field. The tests build on an extension of general equilibrium theory where agents’ choices need not be predicted perfectly, rendering the theory more realistic, without however destroying the main pricing results. The tests are implemented on two series of large-scale financial markets experiments. The first series is the certainty equivalent of the second one, with nonlinearity in payoffs replacing uncertainty. The goal of the tests is to resolve an apparent price-allocation paradox in the experiments, namely, why asset prices can satisfy theoretical restrictions (in our case, the CAPM) when allocations deviate substantially (portfolio separation fails). Most puzzling is the finding that allocations seem not to be any closer to CAPM predictions than when prices grossly violate CAPM restrictions. We find that our theory explains the paradox in both sets of experiments. That is, when end-of-period prices are such that the market portfolio is close to mean-variance efficient (the CAPM pricing prediction), the allocations and our theory explain why. When end-of-period prices make the market portfolio far from mean-variance optimal, our tests reject.