Orley Ashenfelter, an economist at Princeton, wanted to guess the prices that different vintages of Bordeaux wine would have. This prediction would be most useful at the time of picking, so that investors can buy the young wine and allow it to come of age. In his own words:
The goal in this paper is to study how the price of mature wines may be predicted from data available when the grapes are picked, and then to explore the effect that this has on the initial and final prices of the wines.
For those of you not so au-fait with wine, prices vary a lot. At auction in 1991, a dozen bottles from Lafite vineyard were bought for:
- $649 for a 1964 vintage
- $190 for a 1965 vintage
- $1274 for a 1966 vintage
Wines from the same location can vary by a factor of 10 between different years. Before Ashenfelter’s paper, people predicted wine quality by experts, who tasted the wine and then guessed how good it would be in future. Ashenfelter’s great achievement was to bring some simple science to this otherwise untapped field (no pun intended).
He started by using the things that were “common knowledge”: in particular that weather affects quality and thus selling price. He checked this by looking at the historical data:
In general, high quality vintages for Bordeaux wines correspond to the years in which August and September are dry, the growing season is warm, and the previous winter has been wet.
Ashenfelter showed that 80% of price variation could be down to weather, and the remaining 20% down to age. With the given inputs, the model he built was:
log(Price) = Constant + 0.238 x Age + 0.616 x Average growing season temperature (April-September) -0.00386 x August rainfall + 0.001173 x Prior rainfall (October-March)
As it turned out, this simple model was better at guessing quality than the “wine expert”: a success for science against pure intuition. The smart part of his approach was getting insight in to the things people felt mattered (weather) and checking that wisdom. Here, he showed that yes it is quite appropriate to use weather and age to model wine prices.
Through the age variable, it also gives an average 2-3% annual return on investment  (note this is pre-2008 so is unlikely to behave like this today).
Should I buy wine? Quite possibly, as long as I don’t drink it all.