The following is a guest article written by Christopher Armstrong who comments on virtual economics over at ChrisArmstrong.org. The article discusses the recent works of economist Ed Castronova to see if real world economic laws carry over into the virtual world.
Recently Ed Castronova and the folks over at Indiana University have released the results of their study on the law of demand in virtual worlds in a paper entitled A Test of the Law of Demand in a Virtual World: Exploring the Petri Dish Approach to Social Science. The paper can be downloaded in its entirety from the SSRN.
Before we delve into the findings and details of the experiment, we must understand why it’s important to verify that this economic principle applies in virtual worlds. The law of supply and demand is essential to any economy. It predicts that in every economy, there is a point at which supply and demand intersect (the equilibrium price). This price shows the ideal market price of a good – the intersection of the supply price, the price asked by the producers of a good, and the quantity demanded by the consumers. When the price of a good is high, consumers will demand less of a good and suppliers will tend to lower the price in order to sell enough of the good to be profitable. Conversely, when the price is low consumers will demand more of a good and producers will tend to raise the price.
In any economy where goods are exchanged (including virtual economies), the law of demand is an essential aspect of the market. Every exchange is subject to this law, and thus it is a good candidate to “test the waters” of an economy. Thus, if the law of demand holds in an economy, further economic study can be justified. As Ed Castronova put it:
“Our test focused on the Law of Demand, which holds that as the price of a good
rises, all else equal, the quantity demanded of it will fall. There is perhaps no
principle of human behavior more universally accepted in all of economic theory.
We know of no introductory textbook that does not state the Law of Demand in its
earliest, and foundational, chapters. Thus a test of the Law of Demand is indeed
a stern one for virtual worlds. If the Law of Demand does not hold in a virtual
world, we may as well conclude that virtual worlds generate no economic
behavior of general interest.”
In order to test this economic principle, Castronova and his team set out to create a virtual world. The result of a year’s work was Arden, the world of William Shakespeare. It was developed on the Neverwinter Nights framework developed by BioWare. The team populated their virtual world with NPCs, monsters, and gold that would appear on the ground. This gold would be the currency for the game, and would be used to purchase the item that the team used to test the law of demand – healing potions. These potions allowed players to immediately heal themselves during combat.
Castronova recruited 43 of his students, all of whom volunteered for this experiment. They were split into two near-equal groups and placed into nearly identical virtual worlds – the only difference being that the price of the healing potions was twice as much in one world as in the other. Would the price differential influence the amount of potions purchased, thereby validating the law of demand?
“It also appears to be the case that the subjects were respondent to price
incentives, as the Law of Demand suggests. Subjects in the low-cost world
bought 27.14 potions on average, while those in the high-cost world bought only
15.43 Potions.”
Castronova concluded that yes, this particular virtual world adhered to the law of demand. He calculated that the world’s demand elasticity was -0.431, a perfectly reasonable number.
Lastly, several caveats to these findings were addressed, with the final conclusion of the paper being that virtual worlds can in fact be used as a ground for experimentation in the social sciences – a “petri dish”, as Castronova puts it. While I agree that this study is a giant step in the right direction for the further academic study of the economics of virtual worlds, I argue that it alone cannot be used to validate the academic merit of our industry.
The first question one may have when reading the study is the necessity of the healing potions. Castronova makes no mention in the paper of the importance of healing potions – are they vital to a character’s survival in a fight with monsters or are they simply an aid, an optional “upgrade”? Anyone familiar with MMORPGs or fantasy worlds knows that there are usually multiple ways of healing oneself, whether it be healing spells, potions, or other items acquired throughout the game. Without any knowledge of Castronova’s custom game world we have no way of knowing how necessary healing potions are to the avatars in the game. If the item is optional, is it possible that the results would be different had the item been created to be more of a necessity in the game?
Second, Castronova himself acknowledges that his students are not a prime sample. He states that his students are mostly male. Also, participants must have a somewhat recent computer in order to be able to run the game effectively. However, one possible point not addressed is the student’s familiarity with virtual worlds in general. The students are members of one of Castronova’s classes on virtual worlds, so surely they must be familiar with at least some aspects of virtual worlds. It’s possible that they could have a firmer grasp on how many healing potions they’d need to prepare for an encounter. All virtual worlds of course are different, but I submit that those with prior virtual worlds experience may have different purchasing patterns than those new to the science.
The last caveat I can identify is that the findings of this small-scale experiment may not accurately reflect the larger and broader field of virtual economics. The fact that economic agents in this particular world make economically rational decisions doesn’t necessarily hold true for players of other games. This is a necessary evil of this sort of research, though. Each world is bound to be different in many ways, all of which could have an effect on the purchasing patterns of its inhabitants. A truly accurate test of the decision-making of virtual worlds would be a simultaneous test of many different virtual worlds, with similar goods in each world being consumed. This type of large-scale experiment would be exponentially more difficult to conduct, so for now we must be content with the current research progresses.
Even with these possible caveats, Castronova appears to understand the importance of the study. He doesn’t feel that it necessarily validates the feasibility of the study of virtual economics, but he acknowledges that it’s further proof of the rational economic behavior of agents in virtual economies – a step in the right direction.
“Despite these clear limitations of the study, we argue that it is a reasonably solid
first step in a promising new direction. There are major methodological
advantages to addressing macro-scale social science questions using virtual
world petri dishes.”
…
“This at least is the conclusion toward which the research in this paper
points. Determining how far this kind of generality extends, however, is a task
that should occupy the research agenda in this area for many years.”
The results of the Arden experiment further reinforce the academic merit of the study of virtual worlds. Castronova and his team should be commended for attempting to justify our research. Perhaps our more “real-world” colleagues will be more inclined to support and assist with virtual worlds research in the future.