Random walk theory and the weak-form efficiency of the US art auction prices

We perform variance ratio tests based on non-parametric methods to detect the size of the randomwalk component of the USartauctionprices. The past 134 years of the USartprices exhibit large transitory component (72%) and based on this, the randomwalk hypothesis does not hold. However, possibly due to sparse data before 1935 or due to institutional changes of the art market after World War II, we detect structural breakpoints and find that the randomwalk hypothesis and the weak-formefficiency of the USart market cannot be rejected at least for the past 64 years.