New wine-price index from The Economist

[RMann]

Would love to see the article and numbers behind this. However, my experiences with Wine bid is that they tended to be slightly under market. So if you buy from a mainstream store like Morrell and sell through Wine bid, that number may be as much as 15-20% below what you could have sold it for, which inevitably skews the results.

Since all the data is still drawn from Winebid, it’s equally subjected to the skewing that you suggest. That is, unless you have reason to believe that the Winebid market is significantly different now than 16 years ago. I think your point is valid if you’re talking about your total returns, but in terms of the % appreciation over time, it shouldn’t really matter.


Sure, there are a million quibbles you could make (storage, illiquidity, limits on scale of purchasing, insurance, etc), but the data itself is fairly interesting and speaks for itself. I also can’t read the full article, so it’s hard to engage too deeply.

Maybe to take away a lot of the ambiguity in terms of availability and storage costs, it’d be interesting to just use the costs associated with Winebid (paying for Winebid pro + storage costs) for the last 16 years and look to see how well you’d have done if 15 years ago you just literally bought all the red burg or other types of wines above different price thresholds.

Fair comments Dan! For what it’s worth, I can tell you that overall wine auction market (including WineBid) is in fact significantly different, larger and broader than 16 years ago. Based on the Wine Spectator data, online wine auctions, including Acker, Zachys, Sotheby’s- all great players in their own rights!, etc are now 10% of the total wine auction market and online is the fastest growing segment outside of the Asian auction market in the past 5 years (North American and European traditional/offline/live auctions are still growing but less rapidly). The new generation of Amazon/Instagram-trained wine buyers is more internet and data savvy, and more accustomed to on-demand instant gratification from their iPhones. Again based on the Wine Spectator data, WineBid is the largest online auction, so has the most comprehensive and representative data set available.

The other points that no one has mentioned is that just like any index, the returns are based on point of entry for each benchmark, which go up and down. So the IRR and comparison to benchmarks would be very different if you entered in 2003 vs 2007 vs 2009 vs 2017. Further, point of entry on wines could be influenced by individual vintage values, and point of entry on the S&P is influenced by elections, crises, wars, etc.

At the end of the day, the main point was more of a rhetorical and enjoyable piece for the broader audience- more to be enjoyed and contemplated while drinking wine than trading it (maybe while comforting themselves over today’s stock market declines…)

Happy tasting and sharing!

[John_S]

FWIW, on weighting the index, marketvalue is the standard for things like this. Marketvalue is what defines the size of auction market and is the most common with equities as well (like the S&P 500). Knowing Nathan I’m guessing he would posit that the bottle value * #bottles traded to be an appropriate weighting measure to create said index. That may well be how it was done and would at least be weighted by the value traded.

And Daniel is correct that by using Winebid as the sole source of data, any biases, would largely cancel out. The devil is always in the details but also just getting close is often good enough to do rough analysis and even draw rough conclusions. As i said I enjoyed seeing that graph and the data and can geek out too much which I am trying not to do.

[RMann]

FWIW, on weighting the index, marketvalue is the standard for things like this. Marketvalue is what defines the size of auction market and is the most common with equities as well (like the S&P 500). Knowing Nathan I’m guessing he would posit that the bottle value * #bottles traded to be an appropriate weighting measure to create said index. That may well be how it was done and would at least be weighted by the value traded.

And Daniel is correct that by using Winebid as the sole source of data, any biases, would largely cancel out. The devil is always in the details but also just getting close is often good enough to do rough analysis and even draw rough conclusions. As i said I enjoyed seeing that graph and the data and can geek out too much which I am trying not to do.

Great points. “Geeking out” is the phrase for the “deep knowledge and appreciation,” so cheers to all the geeks. Wine geeks, tech geeks, data geeks, investing geeks, everyone’s invited to geek out.

[Dan_Rosenheck]

John S: I did not weight by value. It’s basically an equal-weighted index of the most expensive XX wines in each region, with only very rare wines given a special down-weighting. There’s no analog in this dataset to a company’s market capitalization, because what trades on WineBid is a small and potentially unrepresentative sample of all trading in a wine, plus you don’t know how much has already been drunk.

[Josh_Grossman]

I wish it was not behind a paywall.

If you use Chrome, this and incognito browsing do away with most paywalls. I support some journalism but not all: Behind The Overlay - Chrome Web Store

[R_Frankel]

Fascinating analysis. Cheers to Dan for doing the work, and to Russ for opening the Winebid data.

I take this as rough market analysis, far more useful/interesting for a consumer than an investor.

[John_S]

Constructing these indices is super tough to do in any reasonably cohesive manner. Hats off to Dan and I know many assumptions and other data issues are tough to deal with when constructing a specific bucket of wines to track over time. The. You have to deal with some wines being added and some being removed. But best to start somewhere even when data is thin. The basic rank ordering of regions pretty much jibes with my intuition which is always where I start on looking as the data.