The virtual tax question

Are virtual assets tax-free? Should Blizzard pay taxes for the gold it earns by selling items to World of Warcraft players? Professor Theodore P. Seto’s analysis of the taxation of virtual assets is the first to distinguish between businesses and ordinary users. In this post, I attempt to explain and comment on his paper.

The discussion on how virtual assets are and should be treated by the taxman is becoming increasingly relevant as business based on virtual assets grows. The taxman wants to make sure that businesses involving virtual asset transactions are treated on an equal basis with other businesses. At the same time, publishers and gamers want to make sure that ordinary gameplay is not hampered by interventions from tax authorities.

There is no question that transactions where virtual assets are sold for real money are taxable in terms of income tax as well as sales tax, where applicable. But in some cases it may also be necessary to tax transactions where virtual assets are exchanged for other virtual assets, in order to ensure equal treatment and prevent loopholes. For example, under a system where only real-money trades (RMT) are taxable, a Second Life entrepreneur would have the advantage of being able to choose the timing of their taxable income (to coincide with deductions), while other businesses have to report sales during the accounting period in which they actually took place. The solution is that the Second Life entrepreneur’s virtual-to-virtual transactions should be made taxable. On the other hand, we certainly don’t want the virtual-to-virtual transactions of an EVE Online corporation or a World of Warcraft player to be subject to this rule. Where to draw the line between taxable and non-taxable transactions is the big “virtual tax question”.

I’ve previously blogged about Swedish Tax Agency’s attempts to draw the line here and here. There are also a few related papers in the VERN bibliography (Bryan Camp, Leandra Lederman). I am not a tax lawyer, but I follow the topic since it is crucially important to both virtual asset businesses as well as those who wish to have nothing to do with real-money trading.

Seto’s answer

Greg Lastowka informs us of a new working paper on the topic by Professor Seto of Loyola Law School, titled When is a Game Only a Game? The Taxation of Virtual Worlds. Seto summarises the previous discussion by Camp and Lederman and provides an increasingly nuanced answer to the virtual tax question. I think it’s a great paper and well worth a read.

Seto’s answer involves two types of distinctions. The first distinction is between users who are subject to cash method accounting rules and users who are subject to accrual method accounting rules. Very broadly speaking, this is a distinction between individuals and real corporations that participate in virtual economies.

The second distinction is Seto’s typology of “virtual worlds”, although I would extend it to say that it is a typology of online services containing virtual assets (i.e., virtual asset platforms), regardless of whether they involve avatars or any other feature associated with virtual worlds. It consists of the following types (pp. 13-14):

  1. Platforms with non-redeemable, non-convertible currencies: e.g., Monopoly, World of Warcraft
  2. Platforms with redeemable or convertible currencies: e.g., Second Life

Using these distinctions, Seto outlines a set of rules, which I have attempted to summarise and explain below. According to Seto, these rules represent not only where he thinks the line should be drawn, but also where U.S. tax law actually draws the line, if interpreted correctly. After presenting the rules, I follow with some of my own comments and analysis.

Tax rules for corporations

Accrual method taxpayers must […] report game credits or other items received in virtual worlds when they are earned, regardless of the type of world or game involved. This should be true even in the case of items earned in worlds with non-redeemable, non-convertible currencies[.] (p. 22)

In other words, for corporations, virtual-to-virtual transactions are taxable, regardless of the type of platform. If a Second Life entrepreneur (whose business is subject to accrual method accounting rules) has an profit in Linden Dollars at the end of the accounting year, they must pay tax on it. Any expenses, whether in Linden Dollars or U.S. dollars, can of course be deducted first. This implies that when the accounting year is over, the company balance sheet will contain assets in Linden dollars. If and when those assets are eventually converted into U.S. dollars, there is no need to pay income tax on them again.

Interestingly, this rule seems to imply that professional U.S. gold farmers must report their loot as taxable income, regardless of whether they intend to convert the loot to real money or not. I wonder if they are equally entitled to report armor repair costs as expenses.

One point that was left ambiguous to me is whether looted, crafted and self-designed items are considered taxable income immediately upon their creation, or only later if and when they are converted into virtual currency.

Tax rules for individuals

Cash method taxpayers engaging exclusively in in-world transactions in [worlds without redeemable or convertible currencies] should not be subject to federal income tax. For such taxpayers, the game is only a game. Transactions in worlds with redeemable or convertible currencies, by contrast, should be treated as generating [taxable income]. Real money trades, even by cash method taxpayers, similarly trigger changes in net worth; they therefore are and should continue to be taxable under standard tax timing rules. (p. 25)

In other words, for Type 1 platforms (non-redeemable, non-convertible currencies), Seto’s rule is that individuals need not report earnings unless and until they use real-money trading to convert them into cash. If a MMORPG or Monopoly player sells something on a RMT market, it is considered taxable income.

On Type 2 platforms (e.g., Second Life) virtual-to-virtual transactions are taxable even for ordinary individual users. If an individual has earned Linden Dollars, they must report the earnings in their tax statement. If and when the profit is realised into U.S. dollars, there is no need to pay tax on it a second time.

How bright is the line?

Do the rules outlined above settle the virtual tax question once and for all? According to Seto, “[f]or cash method tax payers, whether a world’s currency is redeemable or convertible may therefore conveniently serve as a bright-line test of how the world should be treated for tax purposes” (p. 25). This is very similar to the position taken by the Swedish Tax Agency. But the problem is this: how do you determine “whether a world’s currency is redeemable or convertible”?

Edward Castronova has argued for a regime where the legal status of virtual-to-virtual transactions is determined by the operator of the platform. If the operator wants to make a platform for economic exchange, they state this in their terms of service and accept the fact that virtual-to-virtual transactions within the platform will be taxed. If, on the other hand, the operator wants to make a game where in-game transactions are strictly not taxable, they must state in their terms of service that all real-money trading is forbidden.

But according to Seto, existing U.S. tax law takes a more practical approach. The question is “not one to be resolved by hypertechnical analysis of a world’s terms and conditions” (p. 23). Instead, Seto considers, among other things, whether a currency is “readily and routinely convertible into cash” (p. 24). This approach is understandable, since we know that in practice RMT takes place even in MMOs where rules forbid it, and the objective is to avoid tax loopholes. But it again forces us to return to the question of where to draw the line. If 20% of users regularly conduct RMT transactions, does that constitute “readily and routinely”? Does it matter what actions the operator takes to enforce the RMT ban? What if the operator permits RMT and provides a marketplace for it, but only a fraction of the users use it? What if other types of RMT are forbidden but it is permissible to purchase game time cards (GTCs) using in-game currency? These are real examples. To me, it looks like the line proposed by Seto and Swedish Tax Agency to distinguish between different types of virtual asset platforms is not quite so bright.

What about operators?

Seto’s analysis of the taxation of virtual assets is the first to distinguish between businesses and ordinary users. One more type of participant that might be interesting to consider is the operator of the platform (Seto uses the term “sponsor”). Linden Research is active in the real-money market for Linden Dollars, selling L$s for U.S. dollars. According to my reading of Seto’s rules, they would have to report as income not only their realised sales, but also the L$s they create out of thin air that can potentially be sold in the future. That’s a rather arbitrary number.

More frequently, operators are active participants in their in-game virtual-to-virtual markets. According to my reading of Seto’s rules, Blizzard (no doubt subject to accrual method accounting rules) would have to report as income any gold it “earns” by selling items to WoW players. Clearly this conclusion does not make any sense, so what is the exception that allows us to avoid it?

The development aspects of gold farming

The Hindu Gold Dubloon -- photo by Swamibu

Professor Richard Heeks from University of Manchester recently made available a working paper titled Current Analysis and Future Research Agenda on “Gold Farming”: Real-World Production in Developing Countries for the Virtual Economies of Online Games. Heeks summarizes pretty much everything that is currently known on gold farming. The approach is systematic and includes meticulously going through e.g. the estimates on on RMT market volume and aggregate spending on gold farming products, trends of RMT market prices and their effect on gold farming, the stakeholders in the gold farming industry, and the virtual world operators’ incentives in reacting to gold farmers.

The paper can be found via the VERN bibliography. Instead of listing more of its contents here, I’d like to point out the viewpoint of development studies, which is novel in these circles at least to myself, and which is interestingly visible in many parts of the paper.

According to the sources referred to by Heeks, there are probably something like 400,000 gold farmers in 2008, serving something like 5-10 million customers and earning around USD 145 per month. By far most of the farmers are located in China. The gold farming enterprises share many features with typical developing country enterprises: the gold farms are mainly micro-enterprise-sized and based on the informal sector. They are, however, run by more entrepreneurial individuals than the norm, pulled to the market by available input factors and profits rather than pushed by the unavailability of options.

From the development studies point of view, the interesting question is, what is the socioeconomic effect of all this? Heeks seems to believe it has been positive: compared to the alternatives the gold farmers face, the pay, working conditions and prospects seem as good as or better. On the macro level, the impact is likely positive: instead of substituting one form of employment for another, the gold farms seem to create new employment, increasing national income. Since much of the gold farming products and services are bought by foreigners, the gold farms also affect the balance of trade positively.

As one of the possible future research area, Heeks mentions the further investigation of socioeconomic consequences of gold farming and, if there turn out to be developmental benefits, the strategies that countries interested in developing a gold farming sub-sector should employ. I believe this to be a fresh point of view, and one that may be shunned by many of the anti-RMT folk out there.

The paper offers a good summary to anyone interested in to RMT and related phenomena, regardless of attitude towards gold farming.