Forecasting Art Prices with Bayesian Models

In this paper several potential art price determinants are considered. Forexample, stock market indices, other commodity prices, exchange rates, GDP,disposable income, consumption, interest rates, etc. The analysis is based onquarterly data starting in 1998 and ending in 2015. The methodology is based onBMA (Bayesian Model Averaging) and DMA (Dynamic Model Averaging), whichis applicable in case of the uncertainty about the suitable predictors. Prices ofvarious type of art goods are analysed. The results suggest that art market isquite a complex one and even in case of including many predictors it is hard tomodel. However, it is found that DMA outperforms BMA.