To some extent all pu’erh moves together in the market. When the pu’erh bust hit in 2007, teas across the board were hit. Raw tea, ripe tea, Dayi, Changtai, Lao Banzhang, Yiwu.. Of course this is only part of the picture. In the bust’s case, not everything was hit equally and teas definitely didn’t all recover at the same rate.. Some teas rebounded in a year or two, far more quickly than others. Others slowly creeped back up in five or six years and some teas and brands have never recovered.
- A more recent example mentioned by shah, is a controversy (documented in English by Lew Perin) over the potential of aflatoxins (bad for you!) in pu’erh. While it’s yet to be determined how much of an impact (if any) this will have on the pu’erh market it is the sort of scandal that has the potential to impact pretty much the entire pu’erh market..
- My spreadsheets/work is viewable here.
Concept: Similar Teas Move in Similar Patterns
One naive assumption is the idea that all teas (and their batches) are independent of one another and don’t move up and down in the market together. Under this guise, a tea could go up in price because someone gave it a positive review but that it wouldn’t have any impact on any other teas. This is flawed thinking and teas that are similar to one another tend to move in tandem and are likely affected by the same market forces. If 2012 7542 201 goes down 25% in price, there’s an extremely strong chance that 2012 7542 202 and also 2011 7542 will fall in price. It’s less likely that 2003 8582 goes down and even less likely that a 2017 Xiaguan production would go down in price. Things like odd batches can add some noisiness to this but shouldn’t distract from the overall pattern. A 2005 8582 504 batch sells for considerably more (~2x) than the 501 or 502 batch and correlates less than the other 2005 8582 batches.
- A political example of positive correlation would be that some states tend to vote similarly to other states with similar demographics. For instance, there is a higher correlation in the way Wisconsin and Minnesota tend to vote than New York and Alabama (two states with very different demographics).
Similar Teas Tend to Move Up & Down in Price in Conjunction. 1.0 indicates a a high correlation, -1.0 indicates a negative correlation, and 0.0 indicates no correlation.
In order to get some idea about which teas tend to move in the market in tandem, I used Donghe’s pricing data twice a year since the latter half of 2013 and compared seventeen different teas. The data is limited due to the timeframe and because I didn’t want to spend an outrageous amount of time crunching numbers in a spreadsheet.
I calculated the correlation between the pricing data. From here we can see what teas correlated most closely with one another.. It’s an intentionally mish-mashed mix of teas, but I tried to include a few special productions as well as cover a range of common recipes.
- 2011 Jindayi
- 2003 Jindayi
- 2012 Bulang Kongque
- 2006 Bulang Kongque
- 2012 Longteng Shengshi
- 2008 Da Jingdian
- 2006 Jinse Yunxiang
- 2012 7572
- 2008 7572
- 2006 7572
- 2012 7542
- 2008 7542
- 2006 7542
- 2004 7542 (Red)
- 2006 8582
- 2004 8582
Correlation Between Teas Supports the Idea that 2004 is Probably a Landmark Year
They just don’t make em like they used to. A popular belief amongst pu’erh nerds is that Menghai Tea Factory productions shifted for the worst around 2004. This purportedly comes from significant management changes and a need to raise their supply (consequently lowering quality) to meet the rising pu’erh hype..
In the previous article we observed that earlier versions of recipes (7542, 7572, 8582) were all more expensive, with big spikes occurring in the early to mid 2000s. Is this simply a matter of the younger versions needing more time or is something else at work here?
Most similar teas in correlation factor for 2012, 2008, 2006, and Red 2004 7542 Productions
|Tea||Most Similar Tea||2nd Most Similar Tea||3rd Most Similar Tea|
|2012 7542||2008 7542 (.97)||2012 Longteng Shengshi (.96)||2006 7542 (.92)|
|2008 7542||2012 7542 (.97)||2012 Longteng Shengshi (.92)||2006 7542 (.92)|
|2006 7542||2008 Da Jingdian (.97)||2012 Longteng Shengshi (.94)||
2012 7542/2006 7542 (.92)
|2004 7542 (Red)||2003 Jin Dayi (.95)||2006 Bulang Kongque (.9)||2011 Jin Dayi (.85)|
This can be taken as a piece of quantitative evidence that Dayi productions have shifted. They’ve made these recipes for many years, but 2004 and older are acting differently in the market than more recent productions. Are the 2004 and earlier recipes more likely to behave like some of the acclaimed special productions (Jin Dayi or Kongque series) as the 2004 7542 correlation seems to suggest? There’s a few surface-level reasons to think so, such as the total production run. Both special productions and pre-2005 numbered recipes were produced at a lower volume than more modern renditions of numbered recipes. Still I’m pretty hesitant to draw that conclusion armed with just this data.. We don’t know the actual production sizes and as we’ll look at below, it’s important not to extrapolate and assume too much from just our correlation matrix.
We also have one example of a special production made in both 2003 and 2011, the Jin Dayi. Despite being made in two fairly different years of pu’erh and the 2003 version being 10x the price of 2011, their prices are highly correlated (.97).
8582 & False Positives
It’s important to note that similar teas in our correlation matrix are one piece of data in the larger equation. Ideally, we’d have far more detailed pricing data and a larger matrix of teas. You can end up with some pretty weird similarities when working from a fairly limited dataset and timeframe like we are.
Most similar teas in correlation coefficient for Pricing. 2006 and 2008 8582
|Tea||Most Similar Tea||2nd Most Similar Tea||
3rd Most Similar Tea
|2006 8582||2008 7542 (.76)||2006 7542 (.76)||
2008 Da Jingdian, 2004 7542 (Red) (.71)
|2004 8582||2006 7572 (.95)||2011 Jin Dayi (.94)||2008 7572 (.93)|
This also represents a potential limitation in our data. Odds are if we had a more extended timeframe with more teas, and more data points different teas would show up as the most similar. Correlation can help to show a piece of the puzzle, but not the whole thing.
Special Productions have Greater Variance but Potentially Greater Reward
Another byproduct of a more modern era of big factory pu’erh production is the special production. There’s strong evidence that the numbered recipes have changed considerably. But what about special productions? How have they fared?
Intuitively, special productions should have a lot of variance. Most of them are intended to be premium teas and the production is often a smaller run, produced less frequently. This gives them more potential upside to grow than something like 2012 7542, but also subjects them to a less stable market. A simple eye test of the data Donghe’s data supports this.
Top productions (by price) from 2013-2017 have fluctuated a lot in a Short Period of Time (blue highlights signify over 50% change)
There are significant waves of price movement both upwards and downwards for tea, even though none of these teas really even qualify as semi-aged. This is some evidence of a product which high variance.
If you are hunting for the next Dayi tea that will sell for four digits and are concerned with the relatively stagnant market of modern numbered recipes, this is probably a better bet (a high risk, high reward one) than the numbered recipes that are produced annually or semi-annually. For instance, the 2011 Jin Dayi tracked the prices of older teas more closely than other young raw pu’erh.
Buying special productions also comes with accepting the high risk in the form of fakes, price fluctuations, etc. 2010 and earlier 7542, 8582, and 7572 are all ~$20 still, but there are a number of special productions that are well above this. The barrier to entry is higher — a 2017 Jin Dayi already sells for well over $100/cake. This seems like a lot for a young factory tea, but the starting jian price on Donghe was actually 20% higher!
- Similar teas go up and down in price in a similar fashion.
- 2004 was probably a landmark year and correlation data supports this.
- Special productions are a more variant, higher risk, higher reward bet than your higher-production recipes.