Choosing value-for-money wines

How to find good quality wine at a reasonable price is a popular topic for wine bloggers. Unfortunately, most of the advice is rather vague, usually doing nothing more than telling us why increasing price must follow from increasing quality. The advice is thus often of little practical value, no matter how well-intentioned or informative it is. Perhaps we can do a bit better here by looking at some specific data.

In my previous blog post I presented a specific dataset, which we can use here. It is representative of many other data sets that I have looked at, and therefore it can act as a typical example.

The data set came from Richard Jennings at the RJ on Wine blog, and involves 230 US chardonnays representative of what can be found on grocery store shelves in late 2015. That is, they represent the wines that average people buy, as opposed to those bought by people with a special interest in wine.

The data are shown in the first graph here. Each point in this graph represents a wine, located according to its quality score (horizontally) and price (vertically). However, quite a few wines are super-imposed in the graph, because they have the same score and price. Note, also, that the quality assessment assesses only the sensory qualities of the wine; other definitions of "quality" may not apply.

For this data set we can clearly see the relationship between quality and price, which is what is known as an exponential relationship. The line represents the "expected" value for money — at any given quality score we can expect to pay the amount of money shown by the line, on average. To put it another way, for wines near the line we are paying for what we get, and we are getting what we pay for. It is also sometimes interpreted as being the price that we would be expected to pay if both the producer and the consumer had perfect knowledge about wine quality.

The red dot at the bottom of the graph is the best-value wine in this data set. That is, it is the point that is furthest directly below the line. Since the line represents the "expected" price for a wine of any given quality, the further below the line then the better "bang for the buck" is the wine. This particular wine happens to be the 2013 ForestVille Chardonnay; but it is only the best value if you can get it US$ 6 or less.

The orange points represent some of the other wines that are good value for money, as these are also a long way below the line. That is, they are the cheapest wines for each quality level, or the highest-quality wines at each price point, depending on which was you want to look at it. We could do a lot worse than buy these wines, if we happen to be looking for value-for-money U.S. chardonnay.

The second graph highlights those wines that are are very poor value for money. The further above the line, then the worse is the value for money. The red dots at the top of the graph are all further from the line than is the red dot at the bottom. So, there are 25 wines that are worse value than the best wine is good value! This is, unfortunately, to be expected in economics — there are more factors that tend to increase prices than there are factors that decrease them.

Given some knowledge about the quantitative relationship between wine quality and price (as represented by the graph), there are two basic strategies for optimizing the buying of wine. In economics, the optimal decision is usually taken to mean maximizing the quality : price ratio (QPR) — that is, getting the most for your money. [Note that the novice wine buyer usually does not have any knowledge about the relationship between wine quality and price. Acquiring some of this knowledge is often the first step away from being a novice, because better informed buyers are more likely to identify bargains.]

First, we could set a maximum price and then choose the highest quality wines below that price. Second, we could set a minimum quality level and then choose the cheapest wines above that level. Note that we cannot optimize both price and quality simultaneously — we need to choose one of these two features and then optimize the other one.

These are not the only possible ways to buy wine, of course, but all of others involve subjective decisions for each individual person.

Let's look at these two strategies in relation to the data set presented here. To the graph we can add a few more lines of our own, indicating possible quality or price categories. For example, the third graph adds two lines representing particular price points ($12 and $18).

These are not arbitrarily chosen points. In my experience you will rarely find good wines below the pink price line, and you usually need to buy wines above the green line to get any really good stuff. In other currencies, these price points are currently:

US$ 12 = € 11 = £  8 = SEK 100
US$ 18 = € 16 = £12 = SEK 150

For our example data, you will note that below the pink line no wines exceed 87 quality points. Furthermore, no wines below the green price line exceed 91 quality points. So, these can be taken as a reasonable quality expectations for these price points. That is, if I choose my maximum price then I am also (indirectly) setting my maximum expected wine quality. My optimal buying strategy must be to get as close to that maximum as I can.

If you want to try the second strategy, instead, then the final graph adds a line representing a particular quality point (87.5). This is also not an arbitrary choice — I have discovered over the years that my preferred wines tend to score 88 points or more.

You can see from the graph that I will therefore have to pay at least $12 for these wines (or SEK 100, actually, since I live in Sweden), and that I can expect to pay $19 (SEK 160) on average. So, any time I find an 88-scoring wine below $19 then I am getting a bargain; and my optimal buying strategy must be to reduce my costs as much as I can. In this example, I would therefore be interested in those wines with a score of at least 88 and a price less than (say) $18 — there are 11 wines that fit into this group, from which to choose.

My experience seems to agree with a number of other commentators on the internet. As just one example, the sommelier Jörn Kleinhans, interviewed at Business Insider (How to get the best quality wine at any price point), suggests:

$10 — the vast majority of wine is awful at this level
$20 — this is where quality wine begins, but there is great variation
$30 — wines become reliably high-quality around this price.

Similar suggestion have also been made at Food & Wine (Taste vs. price: how to find wine value). You will note that in the graphs above there are two wines at the $20 price that do not have good quality scores, which emphasizes the erratic nature of wines at this price point.

I rarely enter into discussions about wine beyond these price ranges, because that is out of my financial league. I almost never go beyond SEK 350 (US$ 45), and even then the wine would need to be more than 20 years old to tempt me.

That leaves us with the question as to where higher-priced wines fit into the quality : price relationship, which is a topic that I will discuss in a later post.