SEMIPARAMETRIC (DISTRIBUTION-FREE) TESTING OF THE EXPECTATIONS HYPOTHESIS IN A PARIMUTUEL GAMBLING MARKET

The expectations hypothesis maintains that a current forward or future s price should be an unbiased forecast of the expected future price. T his article tests the expectations hypothesis in the parimutuel gambli ng market for greyhound racing using parametric and semiparametric est imators. Parimutuel gambling markets are similar to speculative asset markets in many regards. Conventional maximum likelihood tests of asse t pricing require a priori specification of the statistical distributi on governing agents' expectations. Distributional misspecifications ma y bias conventional tests. The semiparametric estimators applied in th is article overcome these problems and, in addition, maintain consiste ncy under heteroscedasticity. The results reject the expectations hypo thesis.