John Hurliman

Since January I’ve been working on a final research paper for my undergraduate degree in economics, and it is available to download now. The paper explores property valuation in virtual worlds, and specifically focuses on land prices in Second Life. To collect the data I wrote a libsecondlife bot that periodically logs in, takes a sampling of the land for sale, and compares it against the previous snapshots to detect ownership changes and check if a sale took place. A hedonic regression is constructed to try and make sense of land valuation and a few interesting things were discovered, but nothing an experienced realtor in Second Life wouldn’t already have discovered through experience. In my opinion the most important thing discovered through the research is that the limitations of searching for land in Second life introduced a selectivity bias in the collected data, and does not deliver complete information on the market to land buyers. Maybe one of the third party new search technologies can address these limitations.

Abstract

This paper explores the fast growing market of property in online virtual worlds by researching land prices in the online world of Second Life. As computing power and access to broadband Internet increases, immersive three dimensional worlds are increasing in popularity. In Second Life, a currency system with a floating exchange rate to the US dollar is in place and user-created synthetic goods are bought and sold. The only constrained resource is virtual real estate, and the more you own the larger and more complex structures you can build. A strong correlation between property size and price can be clearly established, but is this the only determinant of cost? A closer look at the parcels being purchased shows that size may not be the only consideration. Some residents may prefer a tropical beachfront location over snowy cliffs, or a more heavily trafficked area as a location for a new business. A background on the methods used to collect research data from Second Life is presented, and a basic hedonic pricing model for property in Second Life is constructed.

Virtual Appraisals: A Hedonic Pricing Model of Property in Second Life
Research paper + Excel data, source code, and MySQL schema

The six top sell offers on the LindeX market right now. A linear regression shows a T-value of -18.889, well outside the significance test at 99.9%. An R-squared value of 98.9%; I wonder if this is temporary or a reoccurring pattern of sell volume in the market? Could it be Linden Labs using the new sell option in the terms of service? I’ll have to keep my eye open for this more often.