A virtual economy is the system of production, trade, and asset ownership that emerges inside a persistent online world, such as a massively multiplayer game. Although its goods exist only as entries in an operator's database, a virtual economy exhibits prices, wages, inflation, wealth inequality, and cross-border trade with real-world economies. Beginning in the early 2000s, economists demonstrated that these economies could be measured and analysed with the standard tools of the discipline — and, in doing so, established that digitally scarce assets command genuine, dollar-denominated value.
This research programme, which came to be called virtual economics, is a direct intellectual ancestor of token economies. The practice of pricing a world's internal currency against the dollar and estimating its aggregate output prefigures the way blockchain communities are valued today through network-level measures such as gross network value.
The founding studies were written between 2001 and 2005, when the field's open question was whether virtual economies could be measured at all. That question was settled. What the two decades since have supplied is a record of the other half of the problem the same researchers identified: who sets a virtual economy's policy, and what recourse its participants have when that policy changes.
Measuring Norrath
The founding empirical study of the field is Edward Castronova's 2001 analysis of Norrath, the world of the game EverQuest. Drawing on 616 currency auctions, 651 avatar auctions, and a survey of 3,619 players, it was the first systematic economic measurement of a virtual world — one with roughly 12,000 self-described "permanent residents" and about 60,000 people present at any given moment.
The results were striking. Castronova estimated Norrath's gross national product at about $135 million, or $2,266 per capita — enough to rank it the 77th richest "country" of 171 in World Bank tables, roughly on par with Russia and richer than China or India. The in-game currency, the platinum piece, traded on eBay-style markets at about $0.01072, nominally above the Japanese yen and the Italian lira; the average avatar produced 319 platinum pieces per hour, an hourly wage of roughly $3.42. Avatar accounts sold for $500 to $1,000, and hedonic pricing put a shadow price of about $13 on each character level, allowing real value creation per hour of play to be expressed in dollars.
The study also documented recognisably macroeconomic dynamics. Castronova framed real-money trade as foreign trade akin to tourism exports — earthlings spending dollars on goods that never leave Norrath — with a single trading site implying gross exports above $5 million a year, an economy that operated entirely underground because the game's publisher claimed all in-world items as its intellectual property. A 29-item price index fell 29 percent over three quarters, deflation that players experienced as the world's challenge, and entertainment value, declining. Wealth was highly unequal, yet residents judged the system fair by a norm of equality of opportunity rather than of outcomes.
Virtual currencies as real currencies
Building on this work, Hiroshi Yamaguchi's 2004 analysis argued that game currencies with observable real-money exchange rates — "meaningful" currencies such as EverQuest's platinum piece, as opposed to closed play-money like the Monopoly dollar — fulfil all three textbook functions of money: medium of exchange, measure of value, and store of value. Limited geographic validity and lack of intrinsic value do not disqualify them, since the same is true of national fiat currencies; in Yamaguchi's formulation, virtual currencies are currencies.
Yamaguchi showed that the incentive for real-money trading is built into game design itself: because in-game skill accumulates with time rather than talent, time-poor but cash-rich players rationally buy items from time-rich but cash-poor players, each side exploiting a comparative advantage. The virtual/real exchange rate is set by players' opportunity costs, and because players themselves continually issue new currency, virtual money is "destined to depreciate" as total avatar wealth grows. With no central bank controlling the money supply and no interest rates to reward saving, virtual currency values are far more volatile than those of major earth currencies. Yamaguchi classified them as Local Exchange Trading Systems (LETS) freed from geographic boundaries — an early sign, he argued, of a "global LETS" that could become a numeraire for value on the internet — and noted that online worlds offer economists a quasi-controlled laboratory.
The economics of designed scarcity
In a companion theoretical paper, Castronova identified what he called the "puzzle of puzzles": players pay to be constrained, because achievement requires scarcity and challenge — inverting the usual economic logic that constraints are bads. Under the subjective theory of value, virtual assets are as economically real as tangible goods: farmed items command genuine prices, so their value is legitimate subject matter for economics. The paper modelled how people allocate time between worlds, documenting "immigrants" who spent more hours per week in the virtual world than in paid employment, and warned that large-scale migration into virtual economies challenges GDP measurement and erodes real-world tax bases. It also observed that game operators are profit-constrained "dictators" who can unilaterally change economic policy and destroy asset values, creating legitimacy tensions as players self-identify as citizens rather than customers — a governance problem that structures such as decentralized autonomous organizations would later attempt to address.
Vili Lehdonvirta's 2005 synthesis named the emerging field "virtual economics" and argued that virtual worlds should be studied as economies in their own right, not merely as simulations of the real one. Because the marginal cost of producing most virtual assets is zero, scarcity is a design choice, and demand curves take unusual shapes: virtual goods largely behave as Veblen goods, valued for exclusivity and status. This explains "mudflation" — in Norrath, items were produced faster than they were destroyed, so what would count as growth in a real economy destroyed value instead. Lehdonvirta found that microeconomic concepts such as markets, comparative advantage, and subjective value transfer well to game worlds, while macroeconomic models such as the national income identity break down and need world-native replacements. By 2005, digital scarcity had already become a business model: the Finnish operator Sulake derived most of its roughly EUR 15 million in revenues from selling virtual furniture in Habbo Hotel.
Two decades of operator policy
Castronova treated the operator's position as the field's unresolved governance problem rather than an incidental detail. The published record since supplies four documented instances of that power being exercised, in four different directions, by four different operators. What they share is not the direction but the procedure: each was an announcement.
Publish: CCP Games and EVE Online
On 27 June 2007 CCP Games appointed Dr Eyjólfur Guðmundsson as EVE Online's in-world lead economist, recruiting him from the deanship of the Faculty of Business and Science at the University of Akureyri. CCP described the role as the first time an MMOG "has commissioned a dedicated real world economist to operate at this level of economic monitoring and research for a virtual world", and the stated rationale was disclosure rather than intervention: "by ensuring everyone has access to the same data, we will enhance the player experience and facilitate economic stability in EVE". The promise was quarterly reporting on inflation, growth and price trends. Nineteen years on, the series is still running and has moved to a monthly cadence: the report covering July 2026 was published on 7 August 2026 and covers mining, production, destruction, regional activity, price indices and money supply, with the underlying figures released as a downloadable data archive rather than only as charts. It is the longest continuous public statistical series any virtual economy has produced, and it exists because one operator decided that it should.
Regulate: Linden Lab and Second Life
Second Life developed a lending sector paying interest on Linden dollar deposits, and it failed. Linden Lab's standing policy on in-world "banks" records the reasoning in its own words: "Before this policy was enacted, several of these 'banks' defaulted on their obligations", and "unsustainably high interest rates offered by many of these banks make it likely that others will default as well, which will have a negative effect on Second Life's virtual economy". The remedy was to prohibit "offering interest or any direct return on investment (whether in L$ or other currency) from any object, such as an ATM, located in Second Life, without proof of an applicable government registration statement or financial institution charter". The mechanism is the interesting part. Rather than construct a virtual banking regulator, the operator made access to the activity conditional on a real-world one, importing an external authority it did not have to build or defend.
Abolish: Blizzard and Diablo III
Blizzard shipped Diablo III in 2012 with both a gold auction house and a real-money auction house built into the game, then removed both on 18 March 2014. The announcement is unusually explicit about the reason, and the reason is not that the market malfunctioned. The houses had been designed to "provide a convenient and secure system for trades", but the conclusion reached was that the system "ultimately undermines Diablo's core game play: kill monsters to get cool loot". This is the clearest instance of the profit constraint binding against the economy rather than through it: a functioning market with real willing buyers and sellers was closed because it competed with the product it was embedded in.
Re-denominate: Sky Mavis and Axie Infinity
The fourth case tests whether a public ledger changes the answer. Smooth Love Potion (SLP), the reward token of the play-to-earn game Axie Infinity, is an ERC-20: holders custody it themselves and can trade it without the operator's participation. Its supply was nonetheless uncapped, which is Lehdonvirta's mudflation with an on-chain balance sheet. On 11 January 2024 Sky Mavis capped it at 44 billion and paired the cap with a buyback fund targeting 2 per cent annual deflation, noting that "in 2023 SLP overshot the 2% target and was actually 2.8% deflationary". Two details matter more than the number. The cap is not a contract constraint: "this cap will be enforced via social contract and via our in-game emission mechanisms". And the change was "authored by multiple team members at Sky Mavis", who also "oversee decisions around the buyback fund", with no token-holder vote at any stage. Self-custody moved the asset out of the operator's database. It did not move the monetary policy.
Functioning as money, without owning it
Yamaguchi's conclusion that meaningful virtual currencies satisfy all three textbook functions of money is a claim about what they do. The issuer's contract makes the opposite claim about what they are, and both hold at once. Linden Lab's Terms and Conditions, effective 2 May 2024, state that "Linden Dollars are not currency or any type of currency substitute or financial instrument, and are not redeemable for any sum of money from Linden Lab at any time". What a holder acquires is instead "a limited license granted to you by Linden Lab to access and use Content, applications, services, and various user-created features in Second Life", one that "may not be sublicensed, encumbered, conveyed or made subject to any right of survivorship or other disposition by operation of law or otherwise". Section 3.1 reserves the policy power in terms that need no interpretation: "Linden Lab has the right to manage, regulate, control, and/or modify the license rights underlying such Linden Dollars as it sees fit", and the licence may be revoked "at any time without notice, refund or compensation".
The gap between the economic description and the legal one has two halves, and they close separately. A bearer asset on a public ledger closes the custody half: a holder's balance stops being an entry the issuer can revise, and the exchange rate Yamaguchi measured stops depending on the issuer's forbearance. The SLP case shows that it closes nothing else. Setting the issuance schedule, the fee, the sink and the cap is a distinct power, and moving the asset on-chain leaves that power exactly where it was unless something is built to move it. That something is governance, which is why the problem the virtual-economy literature raised in 2002 resolves into a DAO question rather than a cryptographic one.
Relevance to Caper
Every caper is, in effect, a small open economy: a community that raises capital, produces work, and trades with the wider Radix economy, with its token serving as both internal currency and claim on the venture's future. Just as Castronova priced Norrath's platinum piece and computed its per-capita GNP, a caper's economy is measured through its token — the market price on its perpetual bonding curve reflects the community's output and expectations, and aggregate metrics such as gross network value descend directly from the virtual-economy tradition of measuring digitally native economies (see Token valuation).
The two halves are worth separating in Caper's own terms. On custody, a caper's token is a Radix-native resource held in the holder's own account, and the redemption path is not a facility the issuer extends: sell_token is declared PUBLIC in the protocol's method-authorisation table, so selling back into the curve is a method any holder may call. The one exception is stated in the source rather than in a policy that could later be revised, the root $XRD caper being refused by name because its governance tokens are non-salable by construction.
On policy, the honest comparison is narrower than "no operator". A fee-setting method exists and is restricted to an admin role. What differs is where the limit sits: the collateralization peak that role may set is bounded inside the method itself to a maximum of 0.15, and the source states the reason, which is that the founder's ceiling of at most five per cent is derived from that peak, so leaving it unbounded "would be a setting an admin badge could raise, not a property of the curve". Set beside a supply cap "enforced via social contract", that is the substance of the difference. It is not the absence of an administrator but a bound the administrator's own transaction has to satisfy, with everything discretionary left to proposals and their on-chain execution.
References
- Edward Castronova (2001). Virtual Worlds: A First-Hand Account of Market and Society on the Cyberian Frontier. CESifo Working Paper No. 618; Gruter Institute Working Papers on Law, Economics, and Evolutionary Biology 2(1).
- Edward Castronova (2002). On Virtual Economies. CESifo Working Paper No. 752; published in Game Studies 3(2), 2003.
- Hiroshi Yamaguchi (2004). An Analysis of Virtual Currencies in Online Games. The Journal of Social Science (International Christian University), No. 53, pp. 57–76.
- Vili Lehdonvirta (2005). Virtual Economics: Applying Economics to the Study of Game Worlds. Proceedings of the 2005 Conference on Future Play, Lansing, MI.
- CCP Games (27 June 2007). EVE Online Appoints In-World Economist. Press release.
- CCP Games (7 August 2026). Monthly Economic Report — July 2026. EVE Online news.
- Linden Lab. Policy regarding inworld banks. Second Life official policy.
- Linden Lab (effective 2 May 2024). Second Life Terms and Conditions, §3.1.
- Blizzard Entertainment (2013). Diablo III Auction House Update. Blizzard News.
- Sky Mavis (11 January 2024). Updates To SLP's Monetary Policy!. The Lunacian.