English weather and Rhine wine quality: An ordered probit model

Abstract

This paper analyses the quality of Rhine wine vintages from Schloss Johannisberg over the last 300 years. It draws on vintage lists and transforms the verbal qualitative assessments into five quality ranks. These ranks are related to temperature and precipitation data, using an ordered probit model. Since reliable instrumental weather data for the Rhine region do not exist for the time before 1826, we utilised the English Manley temperature series (beginning in 1659) and precipitation data for Kew Gardens (available from 1697 on). In addition, we used index data for Germany from the historical climate data bank Historische Klimadatenbank Deutschland of the University of Heidelberg. The results show that English weather is a good proxy variable for the actual weather conditions in the Johannisberg vineyards. While Frankfurt weather data yield a better goodness-to-fit, the data cover only half of the observation period. The models show that the biggest marginal temperature effects occur in the months May and September, i.e. during the blossoming and ripening periods. A 1°C temperature increase for the entire growing season will increase the probability of harvesting a top vintage from approximately 20% to more than 50%. Therefore, the model suggests that moderate global warming is likely to improve the quality of Rhine wines.