The cost of luxury wines

In the previous post (Luxury wines and the relationship of quality to price) I discussed the concept that luxury wines do not fit into the "normal" price structure. There are thus two independent concepts of "price" for wines, and these have different relationships to quality. Indeed, luxury wines are more of a commodity than a drink — they are traded as investments rather than drunk as a beverage.

The dataset used in that post had only had a few luxury wines, and so I could not examine how price varies among the luxury wines, only among the "normal" wines. In order to look at the luxury wines I need a bigger list of such wines — that is, we need to have a list of the world's most expensive wines.

One possible source of that is the list provided by the Wine-Searcher site of The World's Top 50 Most Expensive Wines. Here, I will use the list updated on 1 June 2016.

The web site describes the list this way:

The list is based on the average price of a standard bottle (750 ml). To be included in the list, a wine must have at least 10 offers spanning at least four vintages, two of which must have been in the last 10 years.

The data are shown in the graph (click it for the full image).

Of the 50 wines in the list, 39 are from France, with 34 being from Burgundy (shown in red in the graph). Of the remaining five French wines, two are from Pomerol (Bordeaux) and three are from Champagne. Of the 11 non-French wines, nine are from Germany (mostly trockenbeerenauslese, the sweetest type of wine), one is from the USA (a Napa Valley cabernet sauvignon), and one from the Ukraine (a port style).

That Burgundy wines dominate this list is well know, as is the most likely cause. Unlike most vineyard regions (notably those of Bordeaux), the vineyards of Burgundy are intricately subdivided, based on the local terroir (soil and climate conditions), with many named climats (or lieux dits). This means that each wine is slightly different, and they are also in limited supply. Basic economic theory indicates that under these circumstances the prices can become outrageous. In general, Burgundy wines will never be good value for money.

Of particular interest for this post, though, is that the prices clearly fit an exponential (or log-linear) model. The correlation coefficient for the rank order of the wines with the logarithm of the price is –0.915 (the coefficient of determination is 84%), which is a very good fit of the data to this simple economic model.

So, we now know that the prices of both "ordinary" wines and "luxury" wines fit the same type of economic model (exponential), but the details of that model vary dramatically between these wines. That is, for both wine types the price increases in a multiplicative manner, and the multiplier itself also increases. (See the post The relationship of wine quality to price.) However, the "base" price is very different between these wine types, and the multiplier is also different.

You and I may never be able to buy the luxury wines, but at least we know that the rich must also think their wine prices go up pretty steeply — US$ 13,000 is a lot to ask for a bottle of flavoured alcohol!

Luxury wines and the relationship of quality to price

In an earlier post I looked at a specific example of the general relationship that seems to exist between the price of a wine and its perceived quality (The relationship of wine quality to price). This turns out to fit what is called an exponential model, so I then used this model to look at how we might make a rational choice about buying wine (Choosing value-for-money wines).

There is one important point that was not covered in these previous blog posts. This is the issue of wines that score very high on the quality scale — their price often seems to have little to do with the prices of other wines.

The data set that I used in those previous posts did not have any very high-scoring wines. This made it very useful as a basic example of the application of the exponential economics model to the relationship between the quality and price of wines. However, we need a different example if we are to look at high-scoring wines.

Luxury goods

The principal issue with high-scoring wines is that it introduces the concept of what are called "luxury goods". These goods, whether they are alcoholic beverages, or clothes, or meals, or accommodation, or transport, or anything else that you can think of, have a price structure that is completely divorced from the price of other goods, except than that the luxury goods will always coast a lot more money than you and I can afford.

An obvious example is flying first class on a plane. You arrive at your destination at the same time as everyone else on the plane, having departed at the same time as well, but you pay 3-4 times as much money. Obviously, you are not paying for the transportation itself. Instead, you are paying for a more luxurious experience while travelling.

So, for luxury goods we pay for what is called "brand prestige". For wine, this is also called "label drinking" — you can actually pay more for the label on the bottle than you do for the bottle's contents. James Bond (secret agent 007) is the best-known example of the sort of person who specifies labels rather than contents when ordering a drink.

This means that there are two relationships between quality and price when we are looking at wines — the first divides the wines into two groups (luxury versus the rest), and the second then describes the relationship between quality and price within each of those two groups separately.

I will illustrate these concepts with a specific example. As in the previous post (The relationship of wine quality to price), I need to pick a particular grape variety and geographical region, as the relationship of quality to price can vary between varietals and regions.

An example

The example data set I will use is the en primeur prices for some of the red wines of Bordeaux for the 2004 vintage. These are the initial prices set by the wineries, before the wines are bottled (also called "futures" or "subscription" prices). I will compare these prices to the average quality scores of several of the media critics who sampled the wines before bottling.

It is important to pick a vintage before 2005, because that was the year when the modern world of outrageous Bordeaux red-wine prices began. This was hailed as the best vintage for 30 years or more; and the quality : price ratio changed with that vintage, never to return to its old existence. I will discuss this in a future blog post.

The en primeur assessment scores are "barrel scores", since they are produced while the wine is still aging in the wooden barrels. They indicate the potential quality of the wine. A "bottle score" will be produced later, when the wines are finally bottled, after another year or so in the barrels. The latter scores may increase or decrease relative to the barrel scores.

It is worthwhile to use a composite quality score from several critics. These wine assessors have different ratings scales, and each one uses their scale in a different way compared to the others (see, for example, Who tosses around the 90 point ratings more casually?). Furthermore, these experts notoriously do not always agree with each other over particular wines, although in general there is usually a consensus opinion about the relative qualities of the world's wines. For Bordeaux wines, these issues have been discussed by Robert Ashton, by Domenic & Arnie Cicchetti, and by Jean-Marie Cardebat, Jean-Marc Figuet & Emmanuel Paroissien (see the Reference list at the bottom of the post).

So, the quality : price relationship can sometimes differ between assessors; and it is worthwhile to avoid this difference if possible.

Fortunately, the issue of combining assessor scores has been studied for Bordeaux red wines by Jean-Marie Cardebat & Emmanuel Paroissien (see the Reference list). So, I can simply use their "affine" mathematical formulae to convert each assessor's scores to the same scale, which is the one used by Robert Parker. A composite score can then be calculated as the average of these transformed scores. (Note: this is quite different from averaging the scores of a panel of people tasting together.)

The data

I sourced the data from Bolomey Wijnimport's bordOverview web site. This interesting site has compiled the en primeur prices for the main Bordeaux regions for every vintage since 2004. It has also compiled the corresponding quality scores from a range of commentators around the world, measured on whatever scale each assessor has used.

The wines used in my data set come from the so-called Left Bank of the Bordeaux region, which includes the Médoc, Haut-Médoc and Pessac-Léognan (Graves) sub-regions. At the time I downloaded the 2004 data (September 2007) there were quality scores from up to seven different assessors. I used all of the wines for which there were scores from at least 3 of the 7 assessors, plus a stated price (in euros). This provided a data set of 106 wines, as shown in the first graph.

Each wine is represented by a single point in the scatterplot, located according to its quality score (horizontally) and price (vertically). The blue points represent the "normal" wines, with the five luxury wines in red.

The points are scattered a good deal, indicating a lot of variation in wine price for any given quality score. This presumably reflects the myriad of different influences on wine quality and wine price, which can act independently of each other. Nevertheless, the points are much less scattered than we saw for the chardonnay data in the earlier blog post (The relationship of wine quality to price). This is a result of averaging the scores from several assessors, which reduces the spread of the points horizontally across the graph.

The luxury wines in this dataset are the five premiers crus classé (first growth) châteaux (in order of decreasing score: Latour, Margaux, Lafite-Rothschild, Haut-Brion, and Mouton-Rothschild). There are also two wannabe luxury wines in the data set, whose points are well above the main group of non-luxury wines. These are Château La Mission Haut-Brion and Château Palmer — La Mission Haut-Brion has consistently had high prices, as indicated by The Liv-ex Bordeaux Classification, although Palmer is usually somewhat cheaper.

As I did for the chardonnay data in the earlier blog post (The relationship of wine quality to price), I can fit the exponential (log-linear) economic model to the non-luxury wines. This produces the summary curve shown in the second graph. In this case, 64% of the variation in wine price is related to wine quality, which is much greater that we saw for the chardonnay data. This is thus a very good-fitting model.

Since there are only five luxury wines in this dataset, almost any model will fit the data. (However, see the next post: The cost of luxury wines.) So, I have simply shown a dashed straight line on the graph for these wines. Note that the two wannabe luxury wines do not really fit either of the two lines on the graph.

Since the model fit is good, we can now consider the matter of choosing a non-luxury wine based on this dataset, using the ideas presented in my post: Choosing value-for-money wines.

For wines of < US$ 20 (=€18), which are the ones below the pink line on the third graph, the wines have quality scores in the range 75–85. This is not very good value for money, at least on a global scale — I can easily get better-quality wines for this price. It might, however, be a good quality : price range for Bordeaux wines!

For wines of US$ 20–30 (=€18–27), which are between the pink and green lines on the graph, the wines have scores in the 80–88 range. This is still very poor value compared to the chardonnay dataset, with wines from the USA. Indeed, a score of 88, which I have suggested would indicate a good-quality wine, means paying a price of at least €29 (US$ 32).

Needless to say, I did not buy any of the 2004 Bordeaux en primeur red wines.

---

References

Robert H. Ashton (2013) Is there consensus among wine quality ratings of prominent critics? An empirical analysis of red Bordeaux, 2004–2010. Journal of Wine Economics 8:225-234.

Jean-Marie Cardebat, Jean-Marc Figuet, Emmanuel Paroissien (2014) Expert opinion and bordeaux wine prices: an attempt to correct biases in subjective judgments. Journal of Wine Economics 9:282-303.

Jean-Marie Cardebat, Emmanuel Paroissien (2015) Reducing quality uncertainty for Bordeaux en primeur wines: a uniform wine score. American Association of Wine Economists Working Paper No. 180.

Domenic V. Cicchetti, Arnie F. Cicchetti (2013) As wine experts disagree, consumers’ taste buds flourish: how two experts rate the 2004 Bordeaux vintage. Journal of Wine Research 24:311-317.

Dom Cicchetti, Arnie Cicchetti (2014) Two enological titans rate the 2009 Bordeaux wines. Wine Economics and Policy 3:28-36.

Literature on Bordeaux Wine Prices

Orley Ashenfelter (2008) Predicting the quality and prices of Bordeaux wine. Economic Journal 118:F174–F184.

Jean-Marie Cardebat, Jean-Marc Figuet (2004) What explains Bordeaux wine prices? Applied Economics Letters 11:293-296.

Pierre Combris, Sébastien Lecocq, Michael Visser (1997) Estimation of a hedonic price equation for Bordeaux wine: does quality matter? Economic Journal 107:390-402.

Elroy Dimson, Peter L. Rousseauc, Christophe Spaenjers (2015) The price of wine. Journal of Financial Economics 118:431-449.

Pierre Dubois, Céline Nauges (2005) Identifying the effect of unobserved quality and experts’ reviews in the pricing of experience goods: empirical application on Bordeaux wine. Working Paper, University of Toulouse.

Philippe Février, William Roos, Michael Visser (2005) The buyer’s option in multi-unit ascending auctions: the case of wine auctions at Drouot. Journal of Economics & Management Strategy 14:813-847.

Héla Hadj Ali, Céline Nauges (2007) The pricing of experience goods: the example of en primeur wine. American Journal of Agricultural Economics 89:91-103.

Héla Hadj Ali, Sébastien Lecocq, Michael Visser (2008) The impact of gurus: Parker grades and en primeur wine prices. Economic Journal 118:F158-F173.

Ira Horowitz, Larry Lockshin (2002) What price quality? An investigation into the prediction of wine-quality ratings. Journal of Wine Research 13:7-22.

Gregory V. Jones, Karl-Heinz Storchmann (2001) Wine market prices and investment under uncertainty: an econometric model for Bordeaux crus classés. Agricultural Economics 26:115-133.

Stuart Landon, Constance E. Smith (1998) Quality expectations, reputation, and price. Southern Economic Journal 64:628-647.

Philippe Mahenc, Valérie Meunier (2006) Early sales of Bordeaux grands crus. Journal of Wine Economics 1:57-74.

Philippe Masset, Jean-Philippe Weisskopf, Mathieu Cossutta (2015) Wine tasters, ratings, and en primeur prices. Journal of Wine Economics 1:75-107.

Tim Noparumpa, Burak Kazaz, Scott Webster (2015) Wine futures and advance selling under quality uncertainty. Manufacturing & Service Operations Management 17:411-426.

Barolo 1947

I guess this wouldn't be a wine blog without the occasional photo of a wine bottle.

This one is a Barolo Tenuta Canubio, from Dottore Giacomo Damilano, vintage 1947. It was purchased on eBay, which we have long been able to do here in the European Union (but only very recently in the North America).

It was, sadly, past its prime, but was still very drinkable. The wine was 69 years old, and seemed like an old-age pensioner, sitting on the verandah in the sun, thinking that it has had a good life, but is not as frisky as it once was.

The colour had mostly disappeared into the sediment, of which there was an inordinate amount. It was very acidic at first, but started to open up about 2 hours after opening.

Colour: pale amber. The wine cleared from the fine sediment with time.
Aroma: plum, wood, leather, cinnamon, prune, spice cake, vanilla, turpentine.
Flavour: honey, lemon, nut, apricot, bitter almond.
Finish: very short.

[Tasting notes by Susanne Stenlund.]

When critics disagree

Many people have presented themselves as assessors of wine quality, whether they be wine professionals, competition judges, or interested amateurs (with a blog). In general, it is usually accepted that wine assessments are highly correlated, so that high-scoring wines for one assessor are also high-scoring wines for other assessors. This is a topic that I will address in a future post.

In the meantime, it is interesting to consider situations where the assessors disagree. Some of these disagreements are legendary, while others are more prosaic. They usually involve a single vintage of a single wine — for example, the well-known disagreement between Jancis Robinson and Robert Parker over the 2003 Château Pavie (Bordeaux).

However, disagreements can be larger than this, and the example I am presenting here is one of the more thorough ones that I have encountered.

The wine in question is Palari Faro, from Sicily. The Palari winery was at the forefront of the renaissance of Sicilian wine-making in the 1990s; and their top wine, Faro, is considered by several commentators to be the best Sicilian red wine.

A few years ago (June 2013, to be precise) I was considering buying a bottle or two, and I wondered which vintage I should buy. So, I looked at what the online commentators had to suggest. I found two sites that seemed to be in thorough disagreement: that of Walter Speller and that of Doctor Wine.

Their scores for the 14 vintages from 1996 to 2009, inclusive, are shown in the first graph. Each point represents one vintage, located based on the two assessments that they each received.

The graph shows almost no relationship between the two sets of scores — they neither increase in synchrony nor decrease in contradiction. Mathematically, the correlation coefficient is –0.043, which is as close to zero as makes no odds. (For the cognoscenti, the coefficient of determination is a miserable 0.2%.)

Since the two assessors used different scales (20-point versus 100-point), the next graph adds a line that shows the nominal relationship between the two scales. If the two critics agreed exactly, then the points would all lie nearly on the line.

This shows that only 2 of the 14 vintages have been given comparable scores by the assessors (ie. they lie near the line). Indeed, the Doctor Wine scores often seem to be somewhat higher than the Speller scores (ie. they deviate from the line further to the right than to the left). Only one wine is rated highly by both assessors (in the top-right of the graph) — it turns out to be the 1998 vintage.

I do not attempt to explain this situation. However, you can imagine that I did not find it helpful for choosing a wine to buy. The only other vintage scores readily available for this wine are from the Wine Enthusiast and CellarTracker websites, but their scores cover only the second half of the time period. During that time, both of these score sets are moderately correlated with the Doctor Wine scores, rather than with the Speller scores.

Finally, it is worth noting that neither of the two web sites linked above seems to have their Faro reviews available any more. I happened to download them at a unique time in internet history, apparently. Most of the scores are still available from the Wine-Searcher website, with their Speller scores coming from the Jancis Robinson website.

Global wine imports

In a previous post I presented the 2015 data on Wine exports for different countries. This leads to an obvious question, about who is importing this exported wine.

The first graph here shows the top 7 countries according to the 2015 data recently provided by the International Organisation of Vine and Wine (OIV), in terms of volume of wine imported.

Most of these countries are not big wine-exporting countries, but both the USA and France apparently both export and import large amounts of wine.

To work out what is going on here, we can consider the monetary value of the wine. After all, it makes a difference whether one is importing cheap wine or expensive wine. This is shown in the next graph.

It turns out that the USA is spending much more money than other countries on the wine that it imports, while France does not even make it into the list of the top 6 countries by price. So, France is mainly importing cheap wine and exporting expensive wine. Apparently, the situation is that the French (i) export their own expensive wines, and (ii) import cheap wine in bulk (mainly from Spain), bottle it, and then re-export it (with a French label).

This situation becomes clearer if we combine the two data sets, and thus calculate who spends the most money per volume of imports. This is shown in the third graph.

Clearly, the USA is importing more expensive wine than is anyone else. Since the USA is also a big producer and exporter of wine, there is clearly some delicate balance going on. The US exports wine at an average of 332 euros per hectolitre and imports wine at an average of 441 euros per hectolitre, so that there is a net trade deficit. Presumably the cheapest wine is staying within the country, in general, to create this difference in average prices.

This contrasts with Germany, for example, which exports wine at an average of 265 euros per hectolitre and imports wine at an average of 163 euros per hectolitre. Thus, the Germans are generally importing cheap wine, which presumably reflects a desire for wine that cannot be produced within the country (in this case, inexpensive red wine).

Canada, China and the UK also cannot meet their wine needs from domestic production alone. In the case of Canada and the UK this is for climatic reasons, as it also is in Germany. They therefore import both cheap and expensive wines. China, on the other hand, is rapidly expanding its vineyard area and wine production, which covers many suitable climatic zones. Presumably, its imports will decrease in the future.