Predicting Wine Quality with Regression Modeling

We were learning about linear regressions in Data & Decisions a couple of weeks ago, and the topic of wine came up in lecture. Back in 1990, a Princeton Economics professor named Orley Ashenfelter developed a simple linear equation to predict the quality of Bordeaux vintages. Obviously wine critics laughed him off - Robert Parker called it "ludicrous and absurd", but Ashenfelter ended up predicting correctly that 1989 and 1990 vintages were the "greatest of the century".

Interestingly, Ashenfelter was able to isolate the determining factors of wine quality to weather. He modeled age and weather (temperature, rainfall, etc) as independent variables and selected auction price as the dependent variable. He ended up with a pretty good model, which is shown below:

Wine quality = 12.145 / 0.00117 * Winter Rainfall + 0.0614 average growing season temp – 0.00386 harvest rainfall

Crazy right? The coolest part is that one can make the predictions long before the wine is ready to drink which as all sorts of implications in wine futures. Makes me wonder what the wine industry is doing nowadays with technology and data. Does anyone know?

How Much Does it Cost to Make Wine?

One of the things that I find interesting about the wine industry is that there is massive price ranges. One could buy a bottle for $7 or one could buy a bottle for $200. What’s often unclear to me is how much of that pricing is driven by true costs versus markups (supplier, wholesaler and retailer).

I set out to try and understand the cost structure of a bottle of wine. For this exercise, I split it the economics into two parts: costs to the supplier and costs imposed by the three tier system. For this simple exercise, I broke down a bottle of wine into its various component parts (grapes, oak, glass, etc.) and built a simple bottoms up model to price out a case and bottle. For this exercise, I assumed that the supplier is a premium supplier that buys grapes (instead of growing them). The first part of the model is below:

While this should not be surprising, grapes dominate the price of wine accounting for upwards of a third of the the cost structure. Winemaking operations are also a massive cost bucket which includes the cost of equipment, overhead, utilities, employees, etc. That said, there is seemingly a lot of flex with the rest of the cost structure. While I was conducting this research, pricing on some of the other items like label, closures and even oak varied quite significantly. That indicates that there is some flex for a supplier to eliminate costs should they want to.

Overall, I am left with the impression that winemaking is not exactly a cheap endeavor which I thought going in to this exercise.

I’m going to continue this series in which I track the costs imposed by the three tier system. In that example, I assume a markup for the supplier and track the costs of everything from freight to excise taxes. The idea is to see how much of the final price the consumer pays is imposed by forcing suppliers to use the three tiered system.

Choice Overload

The contrast between last week’s readings was super interesting. It set up an interesting discussion about whether the massive choices we face as consumers of wine contributes to or actually inhibits our purchase of wine.

On the one hand, we had the article, “When Choice is Demotivating: Can One Desire Too Much of a Good Thing?”. Published in the Journal of Personality and Social Psychology, a premier psychology journal, it was a serious breakthrough when published. I had even read it in undergrad. The article challenges the notion that more choice is always better by providing empirical evidence that having more choices is intrinsically less motivating than having fewer. They show that when purchasing items, such as gourmet jams and chocolate, people buy more when confronted with fewer (e.g. 6) choices versus more (e.g. 24 or 30) choices. This article argues that “choice overload” makes it harder for people to make decisions when there are too many options at hand.

On the other hand, the article, “Drowning in the Wine Lake: Does Choice Overload Exist in Wine Retail?”, takes the opposite point of view with respect to wine. This isn’t initially surprising, given that it was written in the American Association of Wine Economists. But it makes some very compelling points that both attack the original research and also carve out wine retail as a fundamentally different category from jam or chocolate. They argue, for example, that wine has a large number of intrinsic versus extrinsic characteristics and that as wine consumers become more knowledgable, they desire a much broader range of choices. This is supported by empirical research showing that wine retail did not have the same “choice overload” problem as the JPSP article. 

Personally, I find the strongest argument to be that the way wine is organized in stores puts it in a fundamentally different category than chocolates or jams. With wine, it can be organized into a hierarchy (e.g. by country, region, and price) that actually breaks down the choices quite a bit and usually leaves you with fewer than ten choices in the specific category you are searching for. The paper also raises the point that in order for this to be true, there has to be alignment between the consumer’s hierarchy and the one that the store uses. If the consumer is simply looking for “red wine for dinner tonight”, the country/region categorization might not be as helpful. This reminds me of our discussion of websites that sell wine, and how difficult it can be for them to create a categorization that works for mass audiences. Perhaps this is why they – along with retailers – also lean on “Manager’s Pick” selections to give a smaller, more curated set of options for consumers that respond better to that.

Interestingly, this Stanford GSB research review adds more nuance to the discussion: https://www.gsb.stanford.edu/insights/when-customers-equate-choice-quality. In particular, it explains that with products such as wine, people perceive brands with more variety as having higher quality, and are therefore more likely to purchase products from that brand. In part, this is because they see these brands as having more expertise. This adds another layer of tension to what brands and retail managers need to consider.

This conversation is obviously relevant to wine brands and retail outlets. The key question for brands is whether they should focus the wines offered under the same brand, or create more choice for consumers by offering many types. For retailers, they need to decide how many different types of wine to carry, and how to balance the opposing forces of choice overload with the “expert effect” associated with having a larger offering. Getting it right in the right context could have big implications on whether brands are successful, even outside of the typical characteristics (price, quality) that are the primary focus. 

Prime Property

Last week’s case on Chateau Pontet-Canet left me thinking about the exorbitant price of property in the prime wine regions of the world. The debate in class was about whether or not to sell the chateau. The arguments against selling the land were rooted in the family’s affinity to the industry and their love of winemaking. But what about the idea that the land itself is an investment worth holding onto? In light of our discussion, I thought it would be interesting to investigate the comparative prices of real estate in the prime wine regions of the world.

According to La SAFER (Société d'aménagement foncier et d'établissement rural), the agricultural governing body in France that tracks all vineyard transactions, over 9000 vineyards were sold in in France in 2017; close to a quarter of these were to foreign investors. Vineyard prices in France rose by 3.8%, with the average price per hectare of an AOP vineyard (outside of Champagne) standing at €67,500. Non AOP vineyards were far more reasonable at €13,500 per hectare, increasing by 2.2% from the previous year. Last year, Burgundy had the highest priced vineyards in France with most transactions sitting at around €5,000,000 per hectare and reaching a maximum of €12,000,000 per hectare. Champagne had the greatest concentration of expensive vineyards, with each hectare selling for upwards of €1,000,000.

While the price of acreage in the United States is yet to reach the levels of Burgundy and Champagne, Napa and Sonoma are certainly comparable with other AOP regions in France. Moreover, prices are increasing at a far greater rate than any other region in France. According to the California Chapter of the American Society of Farm Managers and Rural Appraisers, the price of acreage in Napa has increased by about 15% in the past year, and is only expected to rise.

Below I have made a table with the ranges for some of the prime wine regions in the world:

REGION	$/HECTARE (2.47 acres/hectare)
Alsace, France	Average: €124,000 Range: €50,000 - €300,000
Bordeaux, France	Average: €92,200 Range: Some appellations up to €2,000,000
Burgundy, France	Average: €5,000,000 Range: Some appellations up to €12,000,000
Champagne, France	Range: > €1,000,000
Rhone, France	Range: €14,000 - €1,000,000
Napa, USA	Range: $101,200 - $160,000
Sonoma, USA	Range: $28,349 - $101,200
Barolo, Piedmont, Italy	Range: €400,000 - €1,000,000
Cartizze, Italy	Range: Up to €1,800,000