MMO-ECON :: Economics, Statistics, and Rationality of MMOs

First Look at Measuring a Server’s GDP

Author: Roboticus

Now that my Auction House data exporter is working, I wanted to give a few glimpses of what sort of information is being captured.  Currently, I am only working with data from one server (Stormreaver) at one time (11/9/2008), but expect future analyses that look across time and across different servers.

First let me present an aggregate of all items listed on the Auction House, sorted by type of item. Note that all figures are represented in Gold, with the decimal places representing remainder Silver and Copper.

Nothing surprising here, except that we can now see the sum of all buyouts on all items in the Auction House is 320,943 Gold, 48 Silver, and 3 Copper at this particular point in time.

I expect that the calculation of this value, while uninteresting by itself, will become very useful for identifying changes in a server’s economy over time, tracking inflation, and comparing one server’s economy to another.  And more to the point of todays article, finding this figure is the first step towards measuring a server’s GDP.

In addition to overall GPD, I expect to be able to find sector-by-sector values.  The following table is a derivative of the table above, but breaks down the items into subgroups.

While that may have been more information than most people need on the issue, I wanted to show the level of granularity that total-market scans can unveil.

Even just in presenting these numbers, I already can see the many difficulties that must be addressed before a firm methodology for establishing server GDP.  In these numbers, I found two outliers that more than doubled the overall value of the economy.  These items were simple Copper Bullets, selling for 180,000 Gold — an item that normally sells for 15 Silver.  A clear method for identifying these outliers must be found, preferably one that does not require looking at data sets one by one. Other factors that need to be identified is how to determine what items successfully sell, which items are re-listings from previous scans, and what time frame to use for measuring GDP.

But before we get on to that, I wanted to share some more types of information that can be gleaned from these data.  Below is a table that compares total value of buyouts for goods, sorted by the level of quality of the item.  For the non-WoW players here, Common items are generally items found by gathering professions and drops off of minor mobs while Epic items are those that drop off the most difficult raid bosses and from high end crafting professions.

The surprising find here is that each of the four gradients are relatively near each other in sum of buyout values.  I would have expected that epic items would have represented a great majority of the Auction House’s value.

If you have any questions, or want more details about these data, please feel free to leave a comment and I will get back to you!

Tags: data collection, GDP

This entry was posted on Wednesday, November 12th, 2008 at 4:16 pm and is filed under Econometrics. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.