In the last post, I shared a one-year projection for Napa Merlot prices. The model only worked when a variable was created to measure the effect of the movie Sideways on the price of Merlot. Years ago, a study was done to measure the effect that the movie had on retail Merlot prices. I thought I would take a crack at determining how much money that movie has cost Merlot farmers.
This is not meant to be any sort of definitive study of the issue. I have no interns and this is a zero-revenue project. So take everything with a grain of salt. The first step was to create a model that satisfactorily explained the prices of California Merlot, (in aggregate, not by county or district). The model, of course, had to include a measurement of the effect of Sideways, as in the Napa Merlot projection.
The model has a correlation above 99%. Below is an explanation of the variables, without specifics as to what they are or exactly how they work, as that is proprietary. For those of you who regularly read this blog, these explanations should be interesting, as they shed some light on how the models I create work:
Generic Variables: Generic variables are those that determine the “shape” of the model. They include things like intercepts (a base number from which to start), linear trends and cyclical factors. This model uses two generic variables with p-values that are less than .000000001 and .000006, indicating as high a probability of relevance as we could ask for.
Grape Market Variables: These are the variables related to winegrapes. They measure (or proxy for) supply of the grape being studied. Demand is usually accounted for by variables related to the national economy, not the grape market. In this model, only one grape market variable makes the cut, with a p-value of under .0000002. Again, looks like a great variable to be able to include.
Macroeconomic Variables: In most of my models, I use only previous year economic numbers, since I’m doing a projection into the future. For this one, I had the luxury of using both previous year numbers and same year numbers, and used two of the former and one of the latter. Typically, one macroeconomic variable is used in each model (and one model I can think of uses none.) It seems likely that many Merlot drinkers have been sensitive to the [wine price] to [bank account size] ratio. The p-values are great, below .000001, .000004 and .000002.
Wild Card Variables: These are variables that explain the effect of one-off events, like Sideways. Unsurprisingly, the model includes one of these, which explains the effect of Sideways. Though the p-value is not as miniscule as the others, it is still almost zero, at about .001, well within any criteria for acceptance.
This model has great correlation and more solid variables than in any other model I’ve designed. If only I could use it for projections! It’s such a strong model, because it’s retrospective. Anyways, I used the Sideways variable to look at how much Merlot prices were depressed by the movie and multiplied that number by the yields for that year to find the total loss to Merlot farmers. I then converted this loss to 2013 dollars to normalize them. The chart below shows the results. Note that nothing is included for 2004, as the movie came out during harvest, when prices were already set (yes, some people still likely got hurt, but I can’t achieve statistical significance for this).
Year Harvest(t) Loss in 2013 USD
2013 345,203 $(4,584,959.82) $(4,584,959.82)
2012 334,942 $(8,897,368.94) $(8,986,342.63)
2011 286,339 $(11,409,423.21) $(11,865,800.14)
2010 310,716 $(16,507,643.63) $(17,663,178.69)
2009 326,356 $(21,673,202.16) $(23,623,790.35)
2008 225,770 $(17,991,979.99) $(19,431,338.39)
2007 304,078 $(28,271,217.79) $(31,663,763.92)
2006 333,502 $(35,436,414.71) $(41,106,241.06)
2005 423,712 $(50,649,415.97) $(60,272,805.00)
Total: 2,890,618 $(195,421,626.23) $(219,198,220.02)
But the headline said the loss was $400M, you say! Yes, the next part I really could have used an intern for, but instead, I used some admittedly really rough assumptions and approximations. See, the loss in payments for harvested grapes would have been much greater, but farmers responded: they budded over, they replanted, they left fruit on the vine. Now, I have no way to measure the fruit left on the vine or the amount of land budded over to other varieties, but we can estimate the costs of replanting.
Since Sideways came out, the reported Merlot acreage in California has dropped by 7,650 acres. Some would have likely died anyways, some would have been replanted to other crops or turned over to non-agricultural use, and many acres would have simply been budded over, but to keep things easy, let’s assume it was all replanted to another grape. Replanting costs vary greatly by area and year, but one might estimate a cost of $15K over 3 years for the state. Additionally, we could assume about half that again in lost profits from Merlot grapes over those 3 years. Finally, we could adjust upward for inflation and get an easy-to-use estimate of $25K per acre. Do the math and you get a loss of just over $191 million. Add that to the first number and we get about $410 million in losses. And though the second number is only a back-of-the-napkin estimate, the fruit left un-harvested and the losses from 2004 of grapes sold on the spot market (which I could not quantify), would likely make up for any overestimate there.
But before we can blame Paul Giamatti for this, let’s take a quick look in the mirror. Did Giamatti’s character utter that drunken phrase because he hated how Merlots taste? No, what he was getting at was that he could not respect someone who picked a wine based on it being a Merlot, with no other information. Why? Because the industry had overplanted Merlot and was shoving sub-standard Merlots through their sales channels and into the consumers’ mouths. Merlot was great for growers, since it’s easy; great for marketers, because it was abundant and, therefore, worth pushing; and great for distributors, because it was turned into a fashionable or at least well-known wine that was easy to sell regardless of quality. For the same reasons it was great for the industry, it ceased being great for the consumer. Then, just at the moment that overplanting was about to destroy prices, Big Bad Paul came and made it much, much worse.