Efficiency Tests in the Art Market Using Cointegration and the Error Correction Model

Abstract

I aim to understand art prices and returns from a market efficiency point of view and we seek to understand whether the art market exhibits predictability or randomness. I tested the weak-form efficiency random walk using ADF and found that most of my time-series data are non-stationary; hence, they all possess randomness. I proceeded and tested my art data using cointegration and the Error Correction Model (ECM) by comparing them to non-risky and risky asset as well as an economic indicator. To be market-efficient, I cannot reject the null hypothesis and accept the alternative hypotheses of cointegrated data. Thus, the results are mixed - art market behaviour can be partly forecasted. I can predict the outcome of art using gold as a benchmark. Bonds are not useful predictors for the art market. Oil & gas as well as the GDP are good predictors of the general art market. The implications of market efficient are mixed. Art can be used as hedge when bundled with bonds. Portfolio diversification seems to be less favourable for the art market. The way to profit is to look at the returns of oil & gas, assuming other people do not possess this information. My study contradicted the academic belief of market efficiency in favour of practitioners. Subjected to scrutiny, my mixed results do not suggest fully abandoning the notion of efficiency in the art market.