Elliot did a posting today (click here to read it) about how the IDNX domain index, run by Thies Lindenthal and using data from Sedo.com, announced that domain prices have doubled since 2006.
I looked at idnx.com and read Lindenthal’s academic paper and it all seems very well researched, but I wish he published an actual list of the domain prices that he used to come up with this doubling in value. As a long time domainer (since 1995), one thing I have learned is that it is very hard to estimate the value of a domain or compare one domain sale with another. On the other hand, that does not mean a site like IDNX can’t track the domain market as a whole. Some of the IDNX price estimate is based on repeat sales of the same domains over time, and that is great way to do it, but some of the estimate is based on hedonic regression analysis which I think is kind of like what estibot.com does (I love estibot).
The point is that there is some element of human input into those values, even if it was just to come up with the initial valuation formula, so that means for example that I might come up with a different value for IDNX if I were the one doing it. If I trust that Thies Lindenthal knows what he is doing with domains, that is fine, but until earlier today I had never heard of him or looked at his site, so that is why I wanted to check it all out. I was actually very impressed after reading everything.
Another issue I want to bring up is that depending on what type of domainer you are (either .com domains or alternate extensions), this “doubling” info can be a little misleading (I am not blaming IDNX, I think what they are doing is great). If you look more closely at the IDNX data, it looks like .com domains only went up around the same rate as inflation since 2006 (inflation went up 18% in the USA). Most of the IDNX price increase was due to the rise of non-.com extensions such as .info and .biz. It is hard to get the exact breakdown by domain extension from their site for the past 12 months, so I am just estimating some of this.
In no way am I trying to be negative about what IDNX is doing. They are a pioneer in this area, and seem to have thought everything out in great detail. They have spent years working on this, while I only spent around 15 minutes looking at their site, so I may be misreading some of their methods and data. I just wanted to put my thoughts out there for people to consider and also to elicit further comments.
Thanks for the great discussion. I enjoyed reading it. A few things might be worth noting:
– IDNX tracks the value of domains that are actually traded. This is not the same as the universe of all domains out there. Something like IBM.com will probably never be sold – and therefore never show up in the statistics.
– IDNX is mostly a repeat sales index, which is pretty good for tracking the market as a whole (the only hedonic factor going into the estimate is the TLD). This technique is different from e. g. Estibot’s (as far as I know). Estibot has a different goal, as they appraise individual domains. This is a very different game. Repeat sales won’t work here.
– About half of all domains in the IDNX sample (~300K sales in total) are COM domains. This is a not too bad representation of all domains out there. But you are right, if you were invested in COMs only, the overall market trend might deviate from the value of your portfolio.
Another question I just thought of is that how does a “big” sedo.com sale like DataCenter.com for $500,000 effect your index? Is that weighted or tracked any differently than a small sale (like for $1000)? I would think the market for super premium domains goes up and down differently at times than the market for low priced domains, so that would be something to look at in the future (if you are not already doing so).
Also, if a domain that sold for $1000 in 2006 sells for $10,000 today, I am not sure that really means much, it might just mean there are more buyers now so it is easier to get what the domain is really worth, and it was really worth $10,000 in 2006 but did not sell for that much. But, if DataCenter.com sold for $100,000 in 2006 (I am just making this up) and sold again for $500,000 in 2014 (that was a real sale), that seems to have much more significance in my opinion.
Good question. The estimation is on returns (second sale divided by first sale of same domain), so the “big money” transactions are not necessarily those where the return was highest. One could zoom in on e. g. transactions above a certain threshold – but I don’t do it right now. I want to track the market as broadly as possible.
Your are absolutely right about the role of liquidity. Sometimes there is simply not a suitable buyer for that exact domain. Again, IDNX is looking at the market in general and, on average, domain specific differences in liquidity cancel out.
I guess what I am trying to say by my previous comment, is that I don’t think the market for low value domains is very efficient I have sold a domain for $100 and then received an offer the next day for $1000 for it. I have bought a domain for $7500 and sold it 3 days later for $10,000 (tom somebody who emailed me because they saw I had just bought it). I bought a domain in February 2014 for $7,000 and had a bunch of offers as high as $12,000 for it in the past month, but I don’t think that represents the entire .com domain market going up 71%.
Also, I think domains that tend to get sold more than once in the span of a few years are many times domainers flipping them, so the initial price they paid may be low and then the 2nd time it is sold is a high price, and that would skew your stats since it would not represent a “real” price increase in the market.
It might turn out that the opposite is also true though, in that a company spends a big amount on buying a domain, goes out of business, and liquidates that same domain for a lot lower price a few years later.
My point is that I am not sure domains that have been sold multiple times are a good representation of the entire domain market.
Somebody told me (forgot who) that any domain that ever sold is not representative for the entire market… “normal” domains never sell. Which means that domains that sold twice might be an even more special segment of the market. IDNX alleviates this selection bias by also matching domains with the same SLD: Example.COM is compared with sales of Example.NET. The differences in TLDs are accounted for (that is the hedonic piece I mentioned earlier). So there are many domains that sold only once – but their SLD has sold multiple times.
I see now, that makes sense.
As I understand, many high-priced domain sales are not reported. So, accepting the limit of this index, I just use it a a general indication of the market.