• 29 January 2010

Tracing Basis Through Virtual Spaces

Adam Chodorow - Sandra Day O’Connor College of Law, AZ State

Posted in

With the rise of virtual worlds and the generation of income with significant real-world value within them, debate has erupted regarding whether or how best to tax such income.  A consensus exists for the proposition that those who “cash out,” i.e., convert virtual wealth to real-world wealth, should be taxed on their gains.  The question of whether, and to what extent, activity that occurs entirely within virtual worlds should be taxed is more difficult.  Some argue that all in-world activity should be exempt from tax, while others, including myself, have argued for taxation under certain circumstances.  To date, the IRS has remained silent regarding this issue, creating a de facto cash-out rule.

Nina Olson, the National Taxpayer Advocate, recently noted the lack of guidance in this area and has suggested that “economic activities associated with virtual worlds may present an emerging area of [tax] noncompliance.”1 Concerned with the lack of guidance afforded IRS agents and taxpayers alike and concerned that non-compliance in this context could lead to non-compliance elsewhere, she has urged the IRS to “help taxpayers comply with their tax obligations by quickly issuing guidance addressing how taxpayers should report economic activities in virtual worlds.”2

This Editorial examines one of the issues that must be addressed, and soon.  Regardless of what decision is ultimately made regarding the tax consequences of in-world transactions, tax authorities must deal with the question of basis and how to trace it through virtual spaces.  The walls between virtual worlds and the real world are crumbling, and people can now cash out either licitly or illicitly with little difficulty.  Without a basis tracking system, someone who sells a virtual item for cash, an act that is undisputedly subject to tax, will not be able to determine gain or loss and therefore will be unable to comply with his tax obligations.  If the authorities move to tax in-world transactions, the need to determine basis becomes even more important as a substantially larger number of transactions involving virtual items will have real-world tax consequences.

It may be tempting to think that this issue is just a tempest in a teapot and not something on which the IRS should spend its limited resources.  However, the real-world market for virtual goods is large and growing as existing worlds expand and new virtual worlds come on-line.  By one recent estimate, the Gross Domestic Product of all virtual worlds is somewhere between $7 and $12 billion dollars, with annual real-world sales of virtual goods (e.g., taxable sales of virtual goods for real-world currency) estimated to be around $1.8 billion.  While once the domain of gamers, virtual worlds now attract wide variety of participants.  Numerous real-world businesses have established presences in virtual worlds, including Nike, Nissan, and IBM.  Politicians have also set up shop on-line.  If the past is any guide, more and more real-world activity will take place in these fora.  Given the increasing popularity and variety of virtual worlds, the need to clarify how the tax laws apply to virtual worlds will only increase.  It seems better to address these issues now, and in an administrative setting, than to wait until conflicts arise and the law is left to develop in the courts in the context of specific fact patterns that may not reflect or highlight larger issues at play.

In addition to solving a problem faced by those who cash out their virtual wealth, addressing basis in this new context affords us the opportunity to think broadly about basis, the role it plays in our tax system, and the myriad and often inconsistent ways we account for it in our tax laws.  In particular, the opportunity to develop rules for this new context may shed light on the propriety of our existing rules and inspire us to undertake the effort to rationalize those rules.

Any basis recovery system must do two things.  First, it must track previously taxed dollars such that any tax gain or loss matches economic gain or loss.  Second, it must take into account administrative difficulties, such as determining basis ab initio, keeping track of fungible assets with different bases, and accounting for the possibility of numerous tax-free exchanges.  These two concerns provide significant leeway to those who design the rules as can be seen by considering the variety of different basis tracking methods found in the Internal Revenue Code (Code or I.R.C.).  Indeed, the Code and Treasury regulations contain a number of different basis recovery rules that could be models for the recovery of basis in virtual goods.

Generally, the basis recovery models found in the Code follow one of two approaches.  The first, which I will call the tracing approach, grants each item its own basis and determines gain or loss property by property.  In the event of tax-free exchanges, basis is simply transferred from one asset to the next to preserve gain or loss until a taxable transaction occurs.  The key issue under this approach is whether taxpayers may designate which assets are sold or exchanged for other assets or whether some forced ordering rule applies.  Examples of this approach include the rules governing stock sales, tax-free reorganizations under I.R.C. § 368, and inventory accounting.  A simple example may help illustrate.  Assume a taxpayer owns 1 share of stock in Intel and 1 share of stock in Sun Microsystems, each worth $500.  He has a $300 built-in gain in the Intel stock and a $200 built-in loss in the Sun stock.  The taxpayer’s net economic position is a $100 gain.  However, under the tracing approach, each share must be considered separately.  The taxpayer may select which share to sell or hold, causing his tax gains or losses to deviate from his economic gains or losses.  If he exchanges the Intel stock for a share of equal value in some other company in a tax-free exchange, his basis in the Intel share will transfer to the new share, preserving his gain for taxation at a later date.

The second approach, which I will call the pooling approach, pools basis in a number of assets and allocates it back across those assets based on their relative fair market value.  Examples of this approach include the rules that apply to partnership interests, mutual fund shares, if the taxpayer so elects, and corporate formations under I.R.C. § 351.  Thus, a taxpayer who purchases interests in a partnership at different times and for different prices must average his basis.  For instance, if a taxpayer purchases a 1% interest for $100 and second 1% interest for $200, his total basis is $300.  If he later sells one of his interests, regardless of which one he sells, he must allocate to it half of his basis, or $150.  As a result, he will recognize gain or loss for tax purposes in proportion to his total economic gain or loss on the investment.

From a theoretical perspective, the tracing approach makes sense for non-fungible property.  Each item of property is separate and should therefore retain its own basis.  Transferring basis from one asset to another seems wrong as do rules that would force taxpayers to hold or sell certain assets such that their economic and tax gains and losses more closely align.  Taxpayers should have the liberty to decide what to sell and what to retain.  In contrast, the pooling approach makes sense for fungible assets.  A taxpayer who sells half his interest in a business has reduced his position by half.  It should not matter which specific shares or partnership interests he has sold.  Affording each identical interest a different basis amount seems artificial, and it provides taxpayers with significant opportunities to manipulate their tax liabilities so that they deviate from their economic position.

Nonetheless, a review of the different models found in the Code and the circumstances in which they apply reveals that the models employed are not driven by theory.  For instance, the tracing approach is applied to fungible assets, such as a single class of stock in a company, while the pooling approach is applied to non-fungible assets, such as general and limited partnership interests.  Taxpayers are given an election for mutual funds.  In the corporate tax context, pooling is required for transactions covered under I.R.C. § 351 regardless of fungibility of the assets contributed to a corporation, while tracing is the norm for corporate reorganizations under I.R.C. § 368.  The different rules are organized around the different types of assets and transactions, as opposed to the nature of the assets.

Digging a little deeper, one sees that the rules are often a function of historical circumstance or accident.  For instance, the different treatment of stock and partnership interests, both of which represent ownership of a business, likely reflects the fact that stock has been historically represented by separately identifiable certificates, thus highlighting the separate nature of individual shares, while partnership interests were typically spelled out in contracts, obscuring the fact that such interests could readily be divided and separately identified.  In the case of fungible mutual funds shares, where taxpayers are permitted to choose whether to trace or average basis, the impetus for allowing averaging came not from any conclusion about the fungible nature of such shares, but rather from the perceived difficulties mutual fund companies would have in reporting basis under a tracing approach.

This review not only reveals the incoherent nature of the rules from a theoretical perspective and set of rules, but it also highlights the instrumental nature of those rules.  That is, they are not foreordained by some internal tax logic, but rather they are merely a tool that can be designed in any number of ways to carry out their purpose, i.e., avoiding double taxation of income or the receipt of a double benefit.  This affords a significant degree of freedom to policymakers designing a basis recovery system for the virtual context.

This freedom is a good thing because none of the models found in the Code will work very well when applied to virtual worlds.  First, consider a tracing approach.  If in-world transactions are tax-free, as is de facto currently the case, virtual world participants may engage in countless tax-free exchanges before ultimately cashing out.  Thus, under a tracing approach, it is necessary to track basis through these exchanges.  This could be quite difficult, depending on the number of exchanges at issue.  While tracing basis might also be difficult for a taxpayer who engaged in a number of real-world I.R.C. § 1031 exchanges before ultimately selling his property for cash, important differences exist between such a person and a virtual world participant.  The real-world taxpayer must be engaged in a trade or business or profit seeking activity to take advantage of I.R.C. § 1031.  He is likely dealing with items of significant value (necessary to justify the cost of § 1031 transaction), and he is also likely to be advised by professionals who can help him pay close attention to basis as he goes.  In contrast, a virtual-world participant will usually be dealing with items of small value and will likely not be advised by tax experts as he goes along.  In addition, he may solely be participating for entertainment.  Accordingly, his attention to tax attributes will likely not rise to the level of his real-world counterpart.

A second problem with the tracing approach arises with the use of virtual currency.  While virtual currency functions as a currency within a virtual world, for real-world tax purposes, it is property.  This means that different batches of currency may have different bases and different holding periods.  Such currency is generally pooled together in a participant’s account.  If a virtual world participant purchases a virtual item with such currency in a tax-free exchange, a tracing approach would require him to determine which currency was used so that the basis and holding period of the currency could be transferred to the item purchased.  It is not clear whether virtual worlds provide mechanisms for doing this.  Thus, participants would need to keep a separate set of records reflecting their contemporaneous decisions.  As many participants are there for entertainment and the dollar amounts are small, it is not at all clear that this will happen.  Put simply, the administrative difficulties associated with the tracing approach make it an unattractive model for virtual worlds.

Even if tracing were possible, it is unclear that the tracing approach is appropriate.  Currency is fungible, and permitting participants to designate which currency they use in-world to purchase items gives them significant opportunities to manipulate the rules, causing their tax results to deviate from their overall economic results.  For instance, a clever participant will purchase items he intends to use in-world with low-basis currency.  He will purchase items he intends to sell for money with high-basis currency, thus minimizing the amount of gain he will have to realize.  Requiring pooling here would avoid this possibility and likely be less complicated from an administrative perspective than allowing or requiring tracing.

Unfortunately, a pure pooling approach would not work particularly well in the virtual context either.  While virtual currency is fungible, virtual items are not.3 For a pooling regime to work, it must be possible to allocate back to each item some portion of the pooled basis.  This is easy where goods are fungible because each good has the same value and therefore gets the same amount of basis.  However, when goods have different values, they should receive different amounts of basis based on their relative fair market value to prevent creating artificial gains and losses.  For instance, if one were to pool the $90,000 basis of a house worth $90,000 with the $10,000 basis of a car worth $10,000, absent a rule requiring allocation proportional to fair market value, the house would get half the basis ($50,000), creating a huge gain, while the car would get the same amount of basis, creating a huge loss, even though the taxpayer has suffered neither gain nor loss.  Instead, the house should get 9/10 of the basis ($90,000), as it reflects 9/10 of the value.  The car should get 1/10, or $10,000.

Setting aside our predisposition to treat non-fungible assets separately, the pooling approach could work with non-fungible assets where liquid markets exist and it is possible to easily value a taxpayer’s property.  However, in virtual worlds, where value is sometimes difficult to determine, pooling presents as much of an administrative nightmare as does the tracing approach.

Given the wide variety of basis accounting rules found in the tax code, and the instrumental nature of such rules, I argue that policymakers should not feel constrained to follow any of the models used to date nor limit themselves to one of the two basic approaches.  As has been done in the case of tax-free corporate transactions, the tax writers should feel free to create a hybrid system, where some types of transactions require tracing and others require pooling.  However, unlike the corporate tax context, where the decision regarding when to use what approach is based on the type of transaction (contribution under § 351 or reorganization under § 368), I would propose that the choice of regime be made based on the nature of the items involved, i.e., whether the items are fungible.

To reflect the non-fungible nature of most virtual goods, I propose that their basis should be separately tracked and treated, as occurs in the real world.  If a taxpayer has a sword and a shield with different bases because they were purchased for different amounts, he should be allowed to decide which to hold and which to sell.  Gain or loss for each item should be determined separately.  If in-world exchanges are tax free, basis should transfer from one item to the next to preserve gain or loss.  Thus, someone who buys a sword for $50, trades it for a shield in-world in a tax-free exchange, and then sells the shield for $60 must report a $10 gain, as the sword’s basis is traced to the shield.  If in-world exchanges are taxed, the basis in any item received in an exchange should be its fair market value, consistent with the rule that applies to real-world barter exchanges.

In contrast, a taxpayer’s basis in virtual currency should be pooled and averaged, reflecting the fungible nature of such goods and the administrative difficulties associated with tracing basis from a commingled pool of currency to any item acquired with such currency.  An example might help.  Assume a taxpayer has 200 gold coins.  He purchased the first 100 for $100, giving him a $100 basis in the lot, or a $1 basis in each coin.  He earned the second 100 in-world and has paid no tax on them.  Accordingly, he has no basis in those coins.  His total basis remains $100.  If he purchases a sword for 50 gold coins and then sells it for $50, he must pay tax on his gains.  Under a tracing approach, he would be allowed to decide which coins he used to purchase the sword, and their basis would be allocated to the sword.  If he chose to uses coins with a $1 basis, he would report no gain or loss on the sale.  Under a pooling approach, the taxpayer has used ¼ of his coins.  Accordingly, ¼ of his basis in the coins ($25) transfers to the sword.  Thus, when he sells the sword, he must report a $25 gain.  Pooling the basis in currency will prevent taxpayers from unduly manipulating their tax liabilities and having tax results that differ significantly from their economic position.


The rise of virtual worlds and the basis issues they engender present tax policymakers with both challenges and opportunities.  The challenge is to create a basis accounting system that ensures the appropriate amount of tax is collected at the appropriate time while being administrable.  In this regard, tax writers must consider both the nature of the transactions involved and the attitudes of those engaged in them.  Virtual worlds are relatively new and not well established.  Onerous tax rules are likely to curtail their use just when they should probably be nurtured.  While no set of rules will be perfect, given the growth of these worlds, remaining silent is no longer an option.

The opportunities are many.  First, tax writers are not burdened by a legacy approach to basis recovery and are therefore free to write on a clean slate.  Second, virtual reality differs from reality in some very important respects that may alleviate some real-world administrative constraints.  For instance, all transactions in-world are electronic, meaning they are already recorded.  It may be possible to add a basis-tracking feature to automate the process.  Similarly, virtual items could be encoded with a basis counter, allowing basis to be tracked within the items themselves and possibly transferred automatically in tax-free exchanges.  Authorities should consider carefully the nature of virtual worlds when designing the rules, as they may have greater liberty in this regard as well.

Finally, this exercise affords us the opportunity to re-examine the basis rules found elsewhere in the Code and regulations.  Those rules currently reflect historical accident and are therefore somewhat incoherent.  Focusing on basis in this new context may give us the impetus to re-examine the different treatment of stock, mutual funds, and partnership interests, among others, and perhaps attempt to rationalize and harmonize those rules around some central theory.



Copyright © 2010 Cornell Law Review.

Adam Chodorow is a Professor of Law at the Sandra Day O’Connor College of Law at Arizona State University, Tempe, Arizona.

This Legal Workshop Editorial is based on the following Law Review Article: Adam Chodorow, Tracing Basis Through Virtual Spaces, 95 CORNELL L. REV. 283 (2010).

  1. See Nina Olson, National Taxpayer Advocate 2008 Annual Report to Congress, Volume 1, The IRS Should Proactively Address Emerging Issues Such as Those Arising from “Virtual Worlds”, at 214, available at http://www.irs.gov/pub/irs-utl/08_tas_arc_intro_toc_msp.pdf (discussing the lack of guidance by the IRS in this area).
  2. Id. at 224.
  3. In some cases, virtual goods may be fungible. Multiple copies of the same item often exist in virtual worlds and may even be owned by one person. While an argument based on fungibility might extend to these assets, I do not suggest pooling basis in anything other than currency.

Post a Comment (all fields are required)

You must be logged in to post a comment.