|
|
|
|
Studies & Opinions |
94% Annual Domain
Name Price Appreciation
Alex
Tajirian
June 15, 2006
Abstract
Using a new statistical
methodology the annual price appreciation on a very large class
of domain names is estimated to be 94 percent. The paper outlines
the methodology and points out its advantages and areas for future
investigation.
Methodology
To estimate returns from domain
name price appreciation, one can adopt statistical techniques
developed to study the art and wine markets. Although these models
are also viable for studying domain names, they require the availability
of prices on multiple purchases of the same domain name over time.
Unfortunately, however, examining our database of over 4,500 sales
records since 2004, we are unable to find enough observations
to meaningfully implement such an estimation technique. Thus,
we developed a new robust methodology to overcome this major obstacle.
Our methodology is based on the following
steps:
-
Classify the domain names in the
dataset, described below, into groups with similar characteristics
using tree-regression techniques.[1]
-
Estimate the average quarterly price
for each group.
-
Calculate price appreciation for
the first quarter of 2006 over the first quarter of 2005
for each group, by taking the difference between the average
prices of these two quarters estimated in Step 2, above.
-
Calculate the weighted average of
price appreciation, where each groups weight is the total
sales value for the group in the initial period divided
by the sum of total sales values in the same period across
all groups.
-
Calculate the compounded annual
appreciation rate obtained in Step 4, above.
Superiority of Methodology
The following are the advantages
of our methodology:
-
Other
studies on domain name price appreciation neither make their
methodology nor sales prices public. Thus, making their
results highly suspicious.
-
The
grouping of similar domain names allows a researcher to
examine the performance of each group separately over time
and across classes, such as hyphenated and numeric.
-
The
model allows testing the statistical significance of inter-group
and over-all price appreciation over time.
-
The
model overcomes the problem of paucity of multiple buy-sell
data for the same domain name.
Dataset
Our database contains over 4,500 publicly available domain name
sale prices between January 2004 and May 2006. For this study,
2004 sales data where excluded. Also excluded are domain names
with the following characteristics: foreign, hyphenated, numeric,
none .com, containing single-character prefixes and suffixes,
and words ignored by Overture.com.
The resulting dataset has 2,050 .com
domain name sales records.
Concluding Remarks
We are currently investigating the
relative strength of brand and traffic value drivers in their
contribution to price appreciation.[2] DomainMart will publish the price appreciation
estimates quarterly.
|
|
|
|
