whereRi,t andtempi,t representthevintageratinginpointsandtheaveragegrowing season temperature in ◦C for vintage t in region i. To account for quality improve- ments that are independent of climatic changes we introduced a trend variable trend for each region i. The trend variable begins with the value one in 1950 and continues in one-unit steps (i.e., taking on the value 50 in 1999). The equation constant and marginal effects of each variable are given by α and β where a positive value for β1 indicates better ratings over time (independent of climate), which could potentially be explained by improvements in production technologies or a time correlated bias of wine critics, i.e., “score inflation”. The final term in the equation represents the stochastic error εi,t . Equation (1) assumes a linear relationship between growing season temperatures and wine quality.
Ashenfelter et al. (1995) and Jones and Storchmann (2001) used the same linear relationship and found a positive correlation between temperature and prices for