The Washington Post has just published an article by Brad Plumer on The Economics of Video Games, based on (amongst other things) an interview that I gave him on my work at Valve. Click here for the Post’s site or read on…

The economics of video games

Inflation can be a headache for any central banker. But it takes a certain type of economist to know what to do when a belligerent spaceship fleet attacks an interstellar trading post, causing mineral prices to surge across the galaxy.

Eyjólfur Guðmundsson is just that economist. Working for the Icelandic company CCP Games, he oversees the virtual economy of the massively multiplayer video game Eve Online. Within this world, players build their own spaceships and traverse a galaxy of 7,500 star systems. They buy and sell raw materials, creating their own fluctuating markets. They speculate on commodities. They form trade coalitions and banks.

It’s a sprawling economy, with more than 400,000 players participating in its virtual market — more people, in fact, than live in Iceland. Inflation, deflation and even recessions can occur. Which is why, from his office in Reyjkjavik, Guðmundsson leads a team of eight analysts poring over reams of data to make sure everything in Eve Online is running smoothly. His job bears more than a passing resemblance to that of Ben Bernanke, who oversees the U.S. economy from the Federal Reserve.

“For all intents and purposes, this is an economy that has activity equal to a small country in real life,” Guðmundsson says. “There’s nothing ‘virtual’ about this world.”

Nowadays, many massively multiplayer online video games have become so complex that game companies are turning to economists for help. Without oversight, the games’ economies can go badly awry — as when a gambling ban triggered a virtual bank run in the online world of Second Life in 2007, with one bank alone costing players $750,000 in real-life money.

But there’s a flip side, too. Just as video game designers are in dire need of economic advice, many academic economists are keen on studying video games. A virtual world, after all, allows economists to study concepts that rarely occur in real life, such as full-reserve banking, a popular libertarian alternative to the current banking system that cropped up in Eve Online. The data is richer. And it’s easier to run economy-wide experiments in a video game — experiments that, for obvious reasons, can’t be run on countries.

That ability to experiment on a massive scale, academics say, could revolutionize economics.

“Economic theory has come to a dead end — the last real breakthroughs were in the 1960s,” says Yanis Varoufakis, a Greek economist recently hired by the video-game company Valve. “But that’s not because we stopped being clever. We came up against a hard barrier. The future is going to be in experimentation and simulation — and video game communities give us a chance to do all that.”

At least, that’s the dream. The reality, as always, is more complicated. Game companies are often wary of meddling economists trying to conduct experiments that suck the fun out of their virtual worlds. And some academics scoff at the notion that there’s anything to learn from universes filled with warlocks and starfleets. Game companies and economists may need each other. Now if only they could learn to share the controller.

Running a virtual economy

In June, Varoufakis announced on his blog that he had been hired as an in-house economist by Valve, the maker of the popular Half-Life games. Varoufakis wasn’t an obscure number-cruncher. From his perch at the University of Athens, he had become famous for his trenchant analyses of Greece’s debt woes and the euro crisis.

It was clear why Valve was interested. The company oversees a network of games such as Team Fortress 2 that run on its online gaming platform, called Steam.

Valve wanted to link different Steam games together so players could trade virtual items. As Gabe Newell, the chief executive of Valve, explained in an e-mail to Varoufakis: “We are discussing an issue of linking economies in two virtual environments (creating a shared currency), and wrestling with some of the thornier problems of balance of payments.”

Whom better to ask, Newell figured, than an expert on the difficulties that Germany and Greece faced after joining the euro?

To date, only two companies — CCP and Valve — have gone so far as to hire in-house economists. But several academics who study virtual worlds say they have consulted with game designers.

“If you’re creating a game with 100,000 users, with things that they can buy and sell, you need an economist just to help you tweak that system so that it doesn’t spin out of control,” says Robert Bloomfield, an economist who studies virtual worlds at Cornell’s Johnson School of Management.

In Eve Online, Guðmundsson oversees an economy that can fluctuate wildly — he says it expanded 42 percent between February 2011 and February 2012, then contracted 15 percent by the summer. His team will periodically have to address imbalances in the money supply. For instance, they can curb inflation by introducing a new type of weapon, say, to absorb virtual currency — not unlike the way a central bank might sell bonds to shrink the money supply. (In theory, Eve Online’s currency has real-world value — the highest-level spaceships, the Titans, are worth the equivalent of $5,000 to $8,000.)

Guðmundsson, who gave up a tenure track position at the University of Akureyri to join CCP in 2007, says that managing the virtual economy can be surprisingly difficult: “It’s so much more compelling and much more interesting and much more in-depth than I ever imagined.”

Periodically, for instance, CCP’s game designers will introduce a new type of technology or alter the availability of certain raw materials to keep things interesting — not unlike real-life technology shocks. But, Guðmundsson says, the players will often speculate on what the game designers are planning, affecting the markets.

“We’ve even seen large alliances trying to manipulate aspects of the market to control the supply and affect prices,” Guðmundsson says. “It’s a lot like OPEC.”

In some ways, the economy of Eve Online is a libertarian experiment on a grand scale. There are few overarching rules. Labor markets quickly bounce back from recession because there’s no minimum wage. Players can voluntarily band together to create all sorts of innovative arrangements, including corporations, trade alliances and financial institutions.

Eve Online’s banks aren’t supported by a central bank or lender of last resort. Much like Ron Paul has proposed for the United States, there’s no fractional reserve banking, in which banks need to keep only a portion of their deposits on hand at any time and can lend the rest out freely.

“That increases the burden on banks to be diligent and efficient,” Guðmundsson says. On the downside, the financial system is sometimes ripe for abuse — one large bank, EBank, collapsed in 2009 when its founder seized its virtual funds and traded them for real-life cash on the black market.

A lack of oversight has caused havoc in some games. In summer 2007, the game Second Life experienced a banking crisis not unlike the one that would grip the United States the following year. Thousands of players had entrusted money in virtual banks that had turned around and invested in lucrative land deals and casinos. When the deals soured all at once, panic ensued. Ordinary depositors found they couldn’t withdraw cash from ATMs, and one bank, Gingko Financial, lost its customers $750,000 in real-life money. The game designers had to step in and ban certain types of lending.

Real-world applications?

As fascinating as these virtual economies can be to observe, however, experts are still divided on whether they have much to teach us about the real world.

“To this day, we haven’t seen anything in these virtual environments that violates fundamental economic theories,” says Ted Castronova of Indiana University Bloomington, one of the first scholars to study virtual worlds. After all, the fact that prices rapidly adjust to supply and demand in Eve Online isn’t shocking. Economists have studied artificial currencies in other contexts, such as cigarette trading in prisons.

Some economists, however, are starting to observe intriguing phenomena. A 2010 study by Cornell’s Bloomfield looked at an unregulated stock exchange in Second Life, called SLCapex, which raised $145,000 from investors. He found that the market tended to favor large investors over small investors. And, he realized, existing economic theories could not explain why issuers continued to raise so much money in such a setting.

Another scholar who studies virtual worlds, Magnus Thor Torfason of Harvard Business School, says it’s surprising that there’s not more cheating in virtual worlds — as when the head of EBank simply absconded with the bank’s funds. That, in itself, may reveal something interesting about how norms and trust shape markets. “If you break these contracts, there’s not much that can be done to you,” Torfason says. “So one thing I’m looking at is, why isn’t [cheating] more prevalent?”

Others say that the fullness of the data available in online worlds allows for the study of concepts that can’t be studied nearly as accurately in real life.

“In real life, if you want to know what’s happening with car sales, you might call up a handful of car dealerships and ask what their sales are like this month,” says Dmitri Williams, a researcher at the University of Southern California. “In a virtual world, you just know everything. There’s no sampling, there’s no error. It’s perfect information.”

Few economists, however, are as ambitious as Varoufakis. He says that, to date, most economic studies of virtual worlds have merely mapped existing concepts onto the games. “They’re just imposing their own prejudices rather than genuinely trying to extract new knowledge from these worlds,” he says.

Varoufakis argues that the games in Valve’s Steam platform may have a great deal to reveal about theoretical economics. Take general equilibrium theory, which economists rely on to model economies. This well-known theory is based on a barter system, which rarely occurs in the real world and is hence hard to test empirically. But a complex barter economy does exist in many of Valve’s games.

So Varoufakis has been probing abstract economic theories by studying how Valve’s games work. “You find a huge variety of startling and interesting phenomena,” he says.

Consider, for example, the concept of arbitrage. Economic theory suggests that markets will usually tend toward equilibrium — if there’s an opportunity for someone to buy low and sell high, enough people will do it that these imbalances will disappear. This is difficult to track in real life. Yet this summer, while working for Valve, Varoufakis used data from the game Team Fortress 2 to measure how often this actually happened. As it turned out, the economy can veer far from equilibrium surprisingly often.

Another thing that virtual worlds could, in theory, allow economists to do is run massive experiments. At the moment, when economists want to run financial experiments, they are typically limited to rounding up college students and putting them in a lab. While that has led to some discoveries — market participants, it turns out, don’t always act rationally — it also has real limits.

“A couple dozen people in a computer lab isn’t enough to test things like the development of banking,” Bloomfield says. “We can’t test how things like property rights and the ability to write contracts and taxes and regulation affect growth.”

Virtual worlds offer one possible way forward. In his essay explaining why he took a job at Valve, Varoufakis noted that the field of econometrics is a “travesty,” in part because of “our inability to run experiments on a macroeconomy such as rewinding time to, say, 1932, in order to see whether the U.S. would have rebounded without the New Deal.” But, he added, in a digital economy, it was possible to alter values, rules and settings, and go “back to observe how the community responds, how relative prices change, the new behavioural patterns that evolve. An economist’s paradise indeed.”

It seems economists are only just warming up to the possibilities. “My academic colleagues initially get startled when I tell them I’m working for Valve,” Varoufakis told me. “They know I don’t have any actual interest in video games. But I only need to talk to them for a few minutes, and they quickly get excited, asking, ‘Well, what if you tried this . . . ?’ ”

Getting along isn’t always easy

One problem, however, is that it’s less clear whether video game companies are willing to cooperate with economists and academics salivating over the possibilities.

“The last thing these companies want to do is partner with academics,” Castronova says. “The academics will come in and say this is interesting, these people are killing monsters, there’s this market, but you know what? This business of killing monsters is interfering with my intellectual pursuit. Can we change that? The game people look at you like you’re crazy.”

Trust between the two groups is low, adds USC’s Dmitri Williams, because for so long scholars were only concerned with whether video games were making teenagers violent.

Indeed, Guðmundsson, the economist behind Eve Online, says he is reluctant to conduct the sort of theoretically interesting experiments many academics would like to see. “If I started changing the parameters, the players would be upset,” he says.

Varoufakis, for his part, has run into a slightly different dilemma. Whenever he tries to run randomized control trials on, say, different pricing strategies for Valve, the players will rapidly catch on and spread the word on online message boards. “They very quickly suss out what we’re doing,” he says. But, he adds, Valve’s customers often enjoy reading about his experiments after the fact.

Other social scientists who study virtual worlds say that getting cooperation from game companies is, as Williams politely puts it, “nontrivial.” (When data is shared, it is always made anonymous.) And when companies do go to academics for help, it’s usually with a narrow set of questions: Why do players quit? What causes them to spend money? Which players in the game are most valuable?

That rift has spurred a few economists to try to create their own games. Bloomfield is working on a platform, called the Synthetic Economy Research Environment, that could enable economists to produce games that simulate large-scale economic phenomenon like a central bank.

One problem, however, is that it’s difficult for academics to create a game as popular as those such as Eve Online. “Academics don’t always have the best understanding of what makes things fun,” Castronova says. “You look at games, see the thing you’re interested in, and think I’ll just make another one of these games. It’s like seeing something scientific in Michelangelo’s ‘David’ and saying, ‘I just need to sculpt another “David.” ’ ”

How games and life intersect

Apart from studying the economic systems of video games, some academics are interested in how online worlds function within the broader real-life economy. Americans are set to spend $2.9 billion on “virtual goods” in 2012, according to trade figures, and the market in Asia is even bigger. That’s still a fairly small slice of the global economy, but academics think virtual communities could raise interesting questions if they get bigger.

“Just for example,” Castronova says, “Facebook has an entire currency system that isn’t taxed or regulated. At what point does that threaten what the Federal Reserve does?”

There’s also a question of whether actions in online worlds count as real-life economic activity. “Say someone is playing Eve Online for a whole week and not providing services in real life,” Guðmundsson says. “That would hurt GDP [the measure of real-life economic growth], but it would increase the Gross User Product in the virtual world. So did overall value creation really decline?”

Similarly, if these worlds grow big enough, could they allow players to “work” online, doing small tasks in exchange for money? Would that help alleviate unemployment? Can video games enable people to find skills that they never knew they had? Eve Online players frequently discuss ways in which gaming skills translate into real life. Chief executives say they have honed management abilities after learning to corral fellow players inside virtual corporations that span star systems.

Those questions all tantalize Guðmundsson. “I once said if Eve Online closed tomorrow,” he says, “I would have enough material to do research for the next 10 years.”

But not right now. There’s no time. After all, he’s got a galactic economy to oversee.