Message in a Bottle: Forecasting Wine Prices

Abstract

Fine wine and alcohol prices can be predicted, but the accuracy of the prediction depends on the chosen forecasting horizon. In our study, we analyse the fine wine indices, as well as the retail and wholesale alcohol prices in the US from January 1992 to March 2022. We use comprehensive datasets that include economic, survey, and financial variables to identify the most relevant determinants of price. Using various combination and dimension reduction techniques, we develop a range of models to forecast prices across different selling levels and alcohol categories. Our research shows that fine wine prices can be predicted for twelve-month horizons, while retail and wholesale alcohol prices can be predicted for horizons ranging from one month to two years. To improve the accuracy of these predictions, we recommend incorporating consumer survey data and macroeconomic factors, such as international factors and mature market equity risk factors.

Keywords: Wine Prices, Forecasting, Agricultural Commodities, Big data

JEL Classification: Q10, Q14, G12, C53, C55