Economics

Virtual worlds (the more persistent ones, anyway) have economies. Economists study economies. Virtual-world economies are therefore legitimate objects of study for economists.

This doesn’t mean that many economists do study virtual worlds, however. Those who are involved in the area are more often than not players of virtual worlds who happen to be economists in real life, rather than economists coming to virtual worlds from the outside. That said, academic economists are becoming increasingly aware of virtual worlds, and we can expect more of them to become interested in the topic as time goes on.

I said “the topic” there, but there are actually three main areas of importance:

• Using virtual worlds to simulate a real-world economy for experimental purposes.

• Figuring out how the in-world economies of virtual worlds work (or fail to work).

• Examining the interaction between virtual-world economies and the real-world economy.

The first of these is not perhaps as widespread as might be expected. I have only one paper in my collection that addresses it, by Thomas Grimm and Johann Mitlöhner ¹. Their approach was to take a regular textual world and add commands specific to performing monetary transactions. The first of these, “deal,” proposed a contract between two individuals (the buyer and the seller), concerning one object. Negotiation could then follow, using an “offer” command to change the terms of the proposed contract (price, quantity, delivery date, and so on), plus “accept” and “withdraw” commands to terminate the deal one way or the other. The system (that is, the virtual world engine) would then either execute or delete the contract, as appropriate. None of the contract commands accrued any cost in themselves (it was a lawyer-free environment).

Grimm and Mitlöhner’s purpose was to create artificial agents that exhibited different “lifelike” behaviors, and see just how well these worked in practice in an environment shared with each other and with real people. They were using a virtual world to simulate theories that are hard to test in the real one. Economic modeling by computer is not a new idea, of course; virtual worlds offer the additional presence of humans, which makes the model they present more plausible ².

The problem is, virtual world economies don’t seem to work the same way as the real one. When they’re constructed to work like the real world works, they don’t. Money quickly becomes either worthless or (less often) impossible to acquire; yet trade still happens. This is fascinating stuff, made all the more interesting because although it’s not too hard to see why the design for a particular virtual-world economy might fail, it’s much harder to come up with a design for one that wouldn’t.

It should be noted that there are several different types of economy available in virtual worlds. Some, such as TinyMUD’s tokens, are little more than permissions to interact with the virtual world: They limit the amount of building that can be done, but there’s not a lot else for which they can be used. The designers of the early graphical world Habitat characterized this as a “socialist economy” ³, based as it was on the equitable distribution of limited real-world computer resources. Their own solution (which was, almost inevitably, inflationary) involved having in-world money that could be spent on in-world goods; again, though, it was still basically one-on-one between character and database—a point well made in a 1995 talk ⁴ to economists by Julian Dibbell. Economies only really get going when players trade between each other, so as to obtain things on which they place different values. This brings us to the second topic of interest to economists: the internal economics of virtual worlds.

The first report on the economics of an individual world remains the archetype: Zachary Booth Simpson’s The In-game Economics of Ultima Online ⁵. This seminal work describes the similarities and differences between Ultima Online’s economy and real-world economies, the thought that went into its design, and the places where the expectation did not match the eventuality.

There were five reasons that the designers of UO wanted their world to have an economy, universal across all large-scale virtual worlds:

• To ration power, so that not everyone can have every piece of desirable kit.

• To support specialization (and therefore diversity), so that players concentrate on finding their own individual niche because of the limits on what they can own.

• To encourage interaction, for trade and work.

• To provide goals.

• To support economic role-playing, for people who like to be artisans, merchants, and so on.

These were the aims. Simpson identified the following areas where the designers had to make unconventional ⁶ decisions because of UO’s virtual nature:

• Balancing fun and realism. The real world’s way of doing it may not be fun.

• What to do with characters when players log off.

• How to handle muling by players with more than one character.

• Hoarding of objects by players creates database load and must be discouraged.

• Newbies need to be given help so they can advance quickly, but this must not be exploitable by experienced players through their mules.

• What to do when bugs allow people to cheat.

The original solution to all these criteria was a closed economy where resources, money, raw materials, goods, and character inventory were all interlinked through player and system actions to create a single economic flow. Within a year of launch, it had switched to faucet/drain. Why was this?

Simpson argues that it was a failure of the original design to consider microeconomic aspects of the world, concentrating as it did on the macro economy. There was severe over-production of goods, because players created many items solely so that their creation skills would rise. This led to a fall in prices that players were not willing to accept. If a player makes a ton of identical shirts and takes them to some NPC vend or that already has a ton of these shirts and has been unable to sell them, the player should expect that the vendor refuses to buy any more; unfortunately, the player instead regards it as a bug—the virtual world encouraged them to create clothes, so it ought to obey its obligations and buy them.

A second problem was that the NPC shopkeeper economy didn’t work. NPC shopkeepers set their prices using a hugely complicated formula based on many different factors. The problem was, players wanted to sell things to NPC vendors but didn’t want to buy anything from them. This meant that vendors operating to a positive cash flow paradigm would cease buying when they ran out of money. Those objects they sold that players could not get elsewhere were subject to a pricing scheme that groups of players could combine to exploit once they figured it out (which of course they did—buying low in one place then teleporting to where they could sell high, for example).

A third problem was hoarding. In a closed economy where the number of resources is constant, new resources are only added to the system when old resources are expended. For example, if a sword breaks, then the metal it is made up of is returned to the resource pool and a miner somewhere will dig up some ore that replaces it. This keeps down inflation (which is its intention), but unfortunately it means that whenever people hold on to objects, rather than using them, there is a resource shortage. Scarcity then encourages players to hold onto the resource even more (because who knows when they may next find some?). The solution of increasing the pool didn’t work—people kept on hoarding and resources kept on drying up. Only by switching to a faucet (constant input of resources, irrespective of use) did that problem go away, but at the expense of server load (because people still hoarded objects they couldn’t or wouldn’t sell). The faucet needed sufficient drains to match it, but these weren’t present. Had these been there at the start, people would have accepted them; because they weren’t, though, players regarded attempts to impose new drains (for example, taxes) as taking things away from them, which was deeply unpopular.

The final reason that the UO economy failed was because of counterfeiting. Loopholes existed that enabled players to clone money, and because of this, they could create as much as they liked until the programmers could fix the bug. This took much longer to tackle than it should have.

Simpson finishes his report with some suggestions as to how future virtual worlds might be able to address these issues, while noting that actually the failure of UO’s economy was not as bad as it would have been for a real-world economy—players still had fun. I won’t list all his conclusions here, because there are too many of them; besides, if you’re at all interested in any of this, you would do better to read the original report. I’ll list his most important recommendations, however, for comparison with what other people propose:

• Create enforceable contracts. If people agree to pay for work, or to perform work for pay, then neither should be able to rip off the other without serious consequences.

• Create auctions. These define price very efficiently, and are fun social events.

• Allow player-controlled vendors to buy as well as sell.

• Deepen production paths. If there are transitional steps to producing goods, newbies can have an apprenticelike intermediate role to play in production instead of creating finished goods themselves.

• Make characters purchase skill improvements, rather than improve them through use.

As a result of Simpson’s paper, later large-scale virtual worlds have invariably gone for an open model rather than a closed one. However, that isn’t to say that a closed one can never work. Although in Chapter 4 I suggested that no one has yet created a successful closed model economy for virtual worlds, that’s actually only true for the big, graphical worlds. In smaller-scale worlds, with fewer player management issues, a closed economy can be implemented and run successfully. Geoff Wong describes ⁷ how by adhering to a strict monetarist Loans Standard policy, the designers of Shattered World ⁸ have created an economy in which prices have remained relatively stable since the early 1990s. The key to this is ensuring that all manufacturers, shops, and banks in the world are owned by PCs, rather than by NPCs. Players specify the prices for absolutely everything—nothing is set by the system itself except the amount of money that the individual banks can loan.

The success of the Scattered World economic system is evidenced by its stable prices. It does have periods of boom and bust (related to changes in the size of the player base), but has never suffered the kind of collapse that befell UO. Its main problem is that to keep the money supply under control, newbies can’t be given a grubstake when they enter—they have either to persuade a banker to loan them the money or perform menial tasks for other players who don’t want to do the tasks themselves. Only when a newbie has demonstrated that they’re likely to become a regular player will bankers deign to lend them any money. This is fine for a small-scale virtual world, but it’s easy to see how one with thousands of newbies couldn’t expect to keep them for long.

There are other problems, too. In a comparison ⁹ of open and closed virtual-world economies, Dan Hastings notes the following difficulties:

• There is a huge amount of work involved in tracking the creation and destruction of economically significant items. Pretty well everything must be audited so as to ensure that the overall money supply remains under control. This can lead to very serious overheads in implementation ¹⁰.

• Virtual worlds where not all characters are PCs must simulate the behavior of their implied NPCs. The virtual world must somehow attempt to model the fact that PCs are “really” only a small fraction of the total population, even though they drive its economy. This is very hard to get right for a closed economy.

• Closed economies remove some of the designer’s freedom. If you create a city with a “poor” area, under a closed economy it’s very difficult to prevent this from becoming a “rich” area if players decide to make it such. This means any premade quests or whatever that assume it’s a poor area will become out of date. Deliberate, artistic placement of economic entities is also hard to sustain.

It may be possible to combine the best of open and closed systems by maintaining a fixed pool of total wealth and using taxation as a mechanism for managing it. This is the approach that has been taken by Star Wars Galaxies, for which excellent documents ¹¹ exist describing the virtual world’s economy itself and the rationale behind its creation. It features a central bank, to control and monitor the cash flow into and out of the virtual economy, ensuring that the amount provided by the faucet is the same as that recovered by taxation. The plan is to keep taxation ¹² constant and vary the incoming wealth (amortized across all distribution mechanisms) to match it; this avoids the tax-level changes that players so hate. There is some leeway in the system, in that calculations are performed weekly rather than continually; these calculations also account for net increases in the number of players joining the virtual world since they were last run. The hope is that this economy will:

• Discourage hoarding, because objects decay.

• Increase immediate spending, because money devalues quicker than goods.

• Create equal opportunities, because newbies don’t have to play catch-up with players who have been around for months.

In general, as long as a taxation system is imposed from the beginning, the players will accept it; they would not feel that anything was being taken from them as a punishment for their collective success. Personally, I believe that this is probably the right way to go in the long term, but of course it greatly depends on its actual implementation; it may have unforeseen consequences on gameplay. Nevertheless, having taxation as a weapon right from the start should make solving most of the unexpected problems that do arise a great deal easier.

Part of the issue with virtual-world economies is that they touch on so many aspects of design. Certain things have to be in place for the economy to function, and some of these are directly at odds with what would otherwise be preferred for gameplay reasons. For example, if there’s a severe recession in the real world, then there’s not a lot you can do about it; in a virtual world, you instead can go play some other world where there isn’t a recession. Periodic recessions may be required to cool down overheated economies, but they’re no fun for the players caught up in them.

This is particularly problematical for players who enjoy the economic side of virtual life. Primarily, this means gatherers and crafters rather than traders (although it will mean traders too in any world where people can’t teleport about at will, buying stuff at source). In a celebrated attack on the hack-and-slay disposition of many virtual world designers, Sie Ming wrote an article ¹³—I Want to Bake Bread ¹⁴—pointing out some of the things that gatherers/crafters liked to do that weren’t being addressed by existing virtual worlds. This was followed up a few months later by a second article ¹⁵—I Want to Forge Swords—describing mechanisms by which designers might tackle these issues. It’s essentially an argument for an integrated economy, whereby there is a symbiotic relationship between those who create objects and those who consume them (that is, supply and demand). Although some of the suggestions are too specialized for many virtual worlds ¹⁶, most are eminently sensible and should not be difficult to incorporate to some degree in any reasonably open-ended new design. That the sentiments the articles express have not become widespread is due mainly to three main factors: A lack of understanding by well-meaning designers who want to be nice to crafters, but somehow just can’t get “it’s all about combat” ideas out of their heads—designers who subscribe to the overall sentiments, but lack the time and budget to do them justice; designers who perfectly understand the issues, but are creating a virtual world that really isn’t about crafting.

Crafting is an area where virtual-world economic activity doesn’t match the real-world version it hopes to model. Real-world economies can grow through extension of manufacturing, improved production methods, and better distribution. In virtual worlds, all these have to be programmed in and are therefore inviolate. In the real world, I can figure out a new way to make toothbrushes that takes less time or creates better toothbrushes; in a virtual world, I can only do this if the designers have anticipated me.

So far, I’ve described how those players of virtual worlds who have an interest in the subject have undertaken economics research concerning virtual worlds. There are, however, economists who come to virtual worlds as economists rather than as players. Their starting point is almost invariably the economic set-up of the real world, in which light they look at the economies of virtual worlds.

This isn’t a bad way to go about things, as long as you realize that there are differences between real-world and virtual-world economies. Although virtual worlds strive for realism when this helps immersion, nevertheless there are situations where the virtual cannot work the same as the real. Given that economies touch on so many aspects of virtual worlds, it would be surprising if they weren’t affected by some of these.

In an underrated introduction ¹⁷ to the topic, John Beezer sets about assessing the differences between real-world and virtual-world economies. He lists the conditions that must be satisfied in order to create a virtual economy recognizably close enough to a real (mercantile) economy that it could be studied as such. The conditions are categorized under three headings:

• Technical. To do with the hardware and software.

• Natural. Facts of real life, which are so fundamental to real-world economies that they must be transferred to virtual worlds.

• Cultural. Related to the way that people interact within the virtual world.

Of these, the most interesting ones are the natural ones. Beezer makes some very uncomfortable points here. For example, a major economic motivator in real life is fear of dying. Indeed, an “economic good” can be legitimately defined as “any product or service that prolongs or protects life”—it’s that fundamental.

Death must be permanent, and also sufficiently unpleasant that players will take it seriously ¹⁸. Characters must be able to kill one another, so that those characters with economic power can be prevented from abusing it (by armed insurrection). Virtual worlds without permanent death are therefore not going to work like the real world, right from the start.

Similarly, a lot of real-world economic activity is concerned with the provision of food and shelter. People need to eat a range of foods, too, not just a certain number of calories worth. Virtual worlds where the necessities aren’t necessary will also not work like the real world.

Thirdly, the way the real world’s economy works is moderated by transaction costs. Resources are not spread out evenly: Some crops grow better in one place than they do elsewhere; some minerals occur in one type of rock but not another; some animals prefer a particular climate. The further away you are from where something is produced, the more it costs to get it. Teleporting should therefore require effort, and teleporting lots of things should be proportionately harder than teleporting just a few of them. Virtual worlds with instant teleportation in them also aren’t going to have an economy like the real world’s.

Beezer’s points are all attainable: You could create a virtual world economy as he describes, and it would indeed work like the real one ¹⁹. The thing is, most virtual worlds do not adhere to these principles, and therefore their economies are not going to make sense if designed along real-world lines. The reason they don’t adhere to the principles is because if they did then their players would migrate to some other virtual world where “life” was easier—ironically, this is basically an economic argument.

The problem thus becomes one of creating a virtual world economy that looks to the players like a real economy but that behind the scenes is not. To this end, some of the ways in which virtual worlds differ from the real one can be used to their advantage. In the real world, for example, central banks aim to keep the money supply in equilibrium with the supply of goods; they do this by tweaking interest rates and by printing banknotes. The virtual world There.com, on the other hand, plans on doing something that the real world can’t—increasing the number of goods created so as to be in equilibrium with the money supply ²⁰. A second weapon they have is to maintain an exchange rate between their internal currency and the real-world dollar, so that the value of the newbie grubstake (and potentially other fixed-amount items) can be kept reasonably constant even if there is rampant inflation. It should be interesting to see how all this pans out.

There.com is a commercial virtual world that explicitly acknowledges the fact that its internal economy is linked to that of the real world. This has been something that has been evident for some time, ever since Ultima Online characters and equipment started being offered for sale on eBay ²¹. Some virtual worlds encourage it as part of their business model; others condone it; most condemn it. Basically, though, it’s going to happen with any large-scale commercial virtual world. It’s just a case of whether this is a black-market, gray-market or white-market economy ²².

And so we come to perhaps the second most famous ²³ of all writings about virtual worlds: Edward Castronova’s Virtual Worlds: A First-Hand Account of Market and Society on the Cyberian Frontier ²⁴. From its somewhat dry title, you might guess that the paper itself hasn’t been read by millions of people. Its conclusions, however, made national newspapers across the globe.

Castronova looks at Norrath (the name of EverQuest’s world) as if it were a real country. He does it absolutely straight, discarding objections that it’s “just a game”: If it’s important to people, it’s important to study it. Choosing Norrath on the grounds that it was the biggest virtual world in terms of subscriptions and revenues ²⁵, he spent four months playing and collecting data. This was supplemented by a survey that attracted 3,619 respondents over a two-day period; the results of this survey (weighted so as to account for the activities of less involved players who didn’t hear about it) form the basis of Castronova’s analysis.

Although this study is all very absorbing, particularly the sections concerning the different markets for commerce within Norrath, it’s what Castronova further did with his data that hit the headlines. He noted that there was considerable trade between Norrath and the real world, and decided to collect further information concerning currency auctions so as to estimate its volume. He found that in May 2001 one Norrathian platinum piece sold for $0.0133 (that is, about one and a third cents), although by September it had dropped somewhat to $0.0098. He then looked at auctions of characters, to see if they were over- or under-valued according to this exchange rate. He found that a $1,000 character typically had more than $1,000 of wealth (in terms of platinum piece prices), so they were undervalued. It then occurred to him to figure out how much a character was typically worth per level, and use data from his user survey about the rate at which characters gain levels to calculate how much wealth Norrath generates as a whole.

He discovered that each level of a character is worth about $13 at auction. Across the whole of Norrath, players were adding around $15,000 of levels per hour. This leads to a gross national product (GNP) of about $135,000,000. Dividing this by the number of players to get the per capita GNP gives a figure of $2,266—roughly the same as that of Russia, and the 77 th richest in the world. This is what attracted the attention of journalists. Virtual worlds may be inhabited by people you wouldn’t want your daughter to marry (not that she’d ever meet one), but ye gods! They comprise the 77 th richest country in the world! Maybe there’s something important going on here that merits significant attention?

As pleased as I am whenever virtual worlds are being taken seriously, I nevertheless feel slightly uneasy here. GNP per capita is used as an indicator for the social and economic well-being of a country; it’s more to do with standards of living than absolute wealth. Although it is indeed frequently used to measure the relative wealth of countries, it’s very sensitive to population figures; low-population countries that are faring well will rate better than high-population counties that aren’t, even though the higher-population countries have as countries far more wealth than the lower-population ones. Norrath may equate with Russia in GNP per capita, but I don’t see Norrath financing a space program.

Castronova followed up his findings with a second paper ²⁶ six months later that considered whether virtual economies might grow to the extent that they have an impact on real economies. From a virtual world designer’s point of view, this is the more interesting paper, mainly because of the many philosophical points it raises. I personally believe that most of these can be addressed by an understanding that virtual world developers rule by divine right, but I appreciate how this may not be to many players’ liking.

Castronova ends this second paper with a list of areas where virtual-world economies don’t fit the standard models of real-world economies:

• In the real world, governments shouldn’t fix prices. In virtual worlds, developers can create and destroy goods effortlessly, therefore they don’t care about excess supply or demand. Some price-fixing could work in them.

• People in the real world would rather not work. People in virtual worlds would rather not not work—they get bored with nothing to do.

• In the real world, economic growth is good. In virtual worlds, increases in per capita wealth make it easier to accomplish tasks and therefore lower the challenge that they present. This means the world is less interesting, so growth can be bad.

• Real-world economics treats populations as fixed and assumes that their tastes and (initial) abilities are fixed. In virtual worlds, people can choose (through player class selection) to exhibit a subset of their abilities. They also can choose (in the context of the world) to be more than one person, and when to be alive.

Virtual world developers who employ real-world economists to design their virtual economies would do well to ensure these individuals have read On Virtual Economies first.

Castronova’s work suggests that people can make a living from virtual worlds. This is borne out by the facts, too: In an April 2001 discussion ²⁷ among Ultima Online eBay sellers, it transpired that some of them were taking in nearly $20,000 a month. On the whole, it was much less than this, of course: High sellers are often resellers, who (to use a real estate analogy) invest hard cash in acquiring run-down “properties” that they then spend considerable time and effort “refurbishing” to sell at a profit.

As it happens, many big deals in UO involve what would be called real estate if it weren’t actually virtual estate. People buy and sell houses the whole time. For an article ²⁸ he was writing, Julian Dibbell tracked the sale of a three-story tower from seller to middle-man to buyer—just one such transaction of the thousands that are undertaken daily in the real world for imaginary items.

When you can sell the virtual, how real does that make it?

Well, it makes it real enough for real companies to want a presence in virtual worlds. After all, when you’ve run out of real-world territories to reach, where else can you go? The Sims Online started the trend by integrating McDonald’s kiosks into their virtual world; although intended mainly as a marketing ploy (it certainly garnered a lot of publicity), this also had the interesting effect of making the virtual seem more real (and, perhaps in time, the real seem more virtual). It may be that the most important thing that economists have brought to virtual worlds is the realization that they are part of Reality.

Whether this is a good thing or a bad thing remains to be seen, of course. If you bring the real into the virtual, you can’t pick and choose what you get. Virtual McDonald’s means virtual anti-McDonald’s protesters ²⁹.

This links rather nicely (although, as will immediately become apparent, not exactly convincingly) to our next topic: politics.