EVE Online Fanfest, QEN, and research co-operation with CCP

The fourth EVE Fanfest, an event giving the EVE Online players an opportunity to meet each other and the game developers, was held in Reykjavik 1. – 3. November. There were two interesting revelations in the event, which also sparked discussion in panels and roundtables, a part of which I’ll try to summarize here. The first one had to with a soon-to-be-published white paper on the EVE player democracy, and why it actually might not be wise to call it democracy after all. The second was about the soon-to-be-published EVE Online Quarterly Economics Newsletter, Vol.1, No.1.

In addition to these two interesting matters, there’s also the reason why a representative of Helsinki Institute of Information Technology (HIIT) was present at EVE Fanfest this year, namely, a recently formed agreement of research co-operation between HIIT and CCP, the operator of EVE Online.

Council of Stellar Management a.k.a. player democracy

The idea of some sort of player representatives, it seems, is actually old in EVE. A few years ago, there was a group of player representatives, hand-picked by CCP, that were in some manner meant to voice the players’ opinion to the developers. For one reason or another, the system was abandoned later on. Similar system is now in use in Star Wars Galaxies.

Now CCP’s plans to introduce a democratic system, in which the players get to elect representatives into a “Council of Stellar Management”. In the very short, any player can be a candidate for the council, and every account gets one vote. In an EVE-ish manner, there are no restrictions on how the players can promote their candidacy, as long as it happens within the restrictions placed by the EULA – so bribing and blackmailing, among other things, are ok (inside the game at least). The council’s role is, simply, “to make EVE better”. I figure there are at least three reasons for CCP to want to introduce such a system. First, it sounds cool, fun, and unique. Second, it would be a medium, far better than e.g. any noise-ridden forums, through which the players could voice their opinions and needs to CCP. Third, it could help CCP in building up trust in the eyes of the players, especially considering the couple of incidents they’ve had this year.

After the presentation of the idea, it got shot down in a panel discussion. The playing of devil’s advocate was done by fifty years of combined MMO experience, a.k.a. Richard Bartle and Jessica Mulligan. From my point of view, their main argument had to do with how the council is promoted: it would be misleading to call the council a democracy, as the developers of MMOs always have the final say in everything. The council, though elected by a democratic process, does not actually get a vote, only a say. This is opposite from what people usually think of as a democracy, thus potentially leading to expectations that can’t be fulfilled. Of course, even infeasible ideas by the council still do carry an important message – that there’s something in the game that the players wish to change for some reason – but, to be sure, the council has no power over the developers. Therefore, names such as democracy or government are inappropriate.

A white paper on CCP’s plans, written by Petur J. Oskarssen, was due to be published last week, so I guess it can be expected any day now. I’ll update this post with a link as soon as I have it. It remains to be seen how the white paper actually presents the idea after the discussion and comments in Fanfest.

UPDATE #2: The white paper is available now.

EVE Online Quarterly Economics Newsletter, Vol.1, No. 1

After publishing a couple of analyses on the markets of EVE Online (found here and here), the recently nominated lead economist of CCP, Dr. Eyjólfur Guðmundsson, has now turned to the macroeconomics of EVE. The Fanfest-goers witnessed a preview of the main results of the first Eve Online QEN. The lecture is also available via EVE TV (behind the Friday tab). More in-depth analysis should be in the paper, which should also be downloadable this week, and I’ll post an update once I get the link.

The results that were shown dealt with demographics (in the in-world sense), monetary system and the price level. The actual paper is supposed to also include some wealth distribution analyses, but those were not shown in the presentation.

Perhaps the two most interesting results were the distribution of players between the more secure and the less secure areas of EVE, and the price level discussion. By the measures that were used, the vast majority of players seems to prefer the less risky areas of the virtual world. 78 % of the characters were located on a given time at high security areas, where they are mostly safe from attacks of other players. According to this measure, the players prefer less risk to smaller payoffs. Some critique was voiced regarding the snapshot nature of the analysis, but the number likely gives a general indication of the player distribution anyway.

The price level measures shown at the presentation gave evidence of a more or less constant deflation in EVE. As all the players are, or have the possibility to be, both producers and consumers, a general inflation measure (similar to CPI) is somewhat difficult to form. As a workaround of this problem, the presentation included four different price indices: minerals (i.e. raw materials), primary and secondary production (i.e. intermediate product and end product producer indices), and consumer prices. What should be termed consumption (as opposed to investment) in a virtual world is open to discussion, but I figure this is also true for many real-world consumer products. Also, somewhat different methods of computing the indices were used, that is, chained and fixed-bundle approaches. The changes in the player’s preferences seemed to be quite fast, considering the very different results the different weight update methods gave. The bundle weights were in all cases based on observed shares of total expenditure. Regardless of the measure that was used, the indices showed evidence of deflation during the course of the last 12 months.

UPDATE #1: The QEN was actually published while I was writing this blog entry :)

The fundamental reason underlying the observed deflation is still an open research question, which actually leads to the last thing I had in mind, namely the

Research co-operation between HIIT and CCP

The main reason for there being a researcher from HIIT at the Fanfest was the recent launch of research co-operation between HIIT and CCP. The co-operation agreement allows us at HIIT to conduct empirical studies of a virtual economy in a way that’s not been possible before. EVE Online is a particularly interesting case for us, as all players take part in the same instance of the virtual world, effectively making the number of agents far larger than in most other virtual economies. There’s been a lot of theoretizing on virtual economies during the last few years, and we now hope to get some flesh around those bones, and get the results published on academic forums.

This was a brief announcement only, and there’ll be more to post later on.

Currency intervention in Second Life – Analyses and doomcasts

The Ludvig von Mises institute, an advocate for the Austrian line of economic thought, recently published an article in which Matthew Beller analyzes the Second Life (SL) economy. I’m happy to see such work done on virtual economies, and published on a forum that I suppose also some mainstream economists read.

There are also some other analyses of the SL economy available. Randolph Harrison has previously written an article that’s somewhat related to the Beller’s article, both of them dealing with Linden Lab’s tendency to intervene in the “foreign exchange” markets of Linden Dollars, the internal currency of SL.

Beller analyzes the supply of L$, and notes that there’s a considerable amount of L$ created by Linden Lab without economic production or flow of real wealth into the SL economy. The author calls this the L$ budget deficit. Such budget deficit occurs when the weekly L$ stipends, handed out by Linden Lab to the premium members (or residents) of SL, exceed Linden Lab’s revenues. On the aggregate, 33 % of the current L$ supply has been created by running such budget deficits in the last year and a half.

Beller argues that this shows that Linden Lab has been inflating the L$ supply, and gives two possible future scenarios, both of which will lead into a slump of some sort:

  1. Linden Lab will stop running significant deficits, which results in less L$ available to spend, which in turn results in slowing economic activity.
  2. Linden Lab will continue running budget deficits until the residents will recognize the L$ frailty and there will be a “crack-up boom”, in this case a run away from L$ and for the US$. In this scenario, if Linden Lab wishes to maintain their peg on L$/US$, they would basically have to have considerable US$ reserves to be able to stop L$ from depreciating.

There’s a parallel here to Randolph Harrison’s blog posts about half a year ago. In the post, Harrison condemned Linden Lab’s exchange rate interventions. Though the main arguments in his post have to do with other issues, he also notes that Linden Lab have been constantly selling L$, maintaining their peg on the currency – where maintaining the peg has also meant collecting seigniorage.

According to his analysis, Linden Lab will have to start buying back L$ in the future. Using the time series of L$ purchased per user, which has had a declining trend, the author concludes that in order to maintain Linden Lab as a net seller of L$, the growth of the user base of has to reach rates that are, basically, impossible to reach.

There are two scenarios here, both dealing with the interventionist policies in SL, and both leading to similar consequences. As I see it, the first of the scenarios is laid on the idea that a L$ backed up by a US$ transaction will not in any way dilute the value of all L$ in circulation, whereas the L$ created without a US$ transaction will. To follow author’s logic, a reader should make the assumption that the continuing process of injecting non-US$-backed L$ in the economy will, at some point, inevitably lead into the users wanting to get rid of their L$ holdings.

In the second scenario, I figure the assumptions (or observations) behind the conclusions are the following:

  • The producers of virtual items, and providers of services, desire, on the aggregate, cash their revenues out of SL – that is, they produce to make real-world profits
  • The amount of spending in SL is decreasing compared to the desire to cash out
  • The number of new users, assumed to spend on the aggregate, will have to grow unrealistically fast in order to maintain the demand of L$ above the supply

I think there’s good material in both of the analyses: they’re among the few transparent and analytical approaches to virtual economies I’ve seen so far. Both are quite pessimistic regarding the outcomes of currency intervention carried out by Linden Lab. If L$ starts to depreciate, according to one of the scenarios presented here or otherwise, I believe that Linden Lab will not start buying back L$ on a large scale – so an escalation to a major depreciation may actually take place quite easily.

Plans to link the internal markets of Second Life and Entropia Universe announced

Anshe Chung, proclaimed “the virtual Rockefeller” by Business 2.0, has published plans to launch an inter-virtual world financial market, linking the internal markets of Second Life (SL), Entropia Universe (EU) and IMVU. The public launch of the service is set at early June. As I understant, the service makes it possible for e.g. users of SL to use their Linden Dollars for purchasing shares of portfolios consisting of EU assets.

The press release states that RMT is not involved. A possibly related piece of old news: Chung got one of the EU banking licenses a while back.

It’s interesting to see how this works out, where do the virtual world operators stand in this, and whether there’s demand for such markets or not. If the markets thrive, there are interesting related issues – for example, the currency unit of EU, the PED, is pegged to USD, whereas the Linden Dollar is floating. Anyway, the markets inside virtual ecomomies and their external connections are getting more complex, which I guess is not a surprise.

(via 3pointD)

How big is the RMT market anyway?

Monopoly money -- photo by goat_girl_photosQuite a few very different estimates of the volume of real-money trade of virtual items (RMT) are available. Part of the reason for the different estimates is the very rapid growth of the user base, and the number of potential real-money traders, in virtual worlds. New service types also spring up, and there are difficulties in defining what RMT should actually include. Regardless of what we decide to include in it, measuring it directly is practically impossible. In this post I review the existing estimates, speculate on what exactly might be included in them, and try to sum up some of the latest regional figures to produce an estimate of the grand total size of the RMT market.

In what follows, I conclude that what I consider as an important part of RMT is often excluded when RMT is discussed. In many cases it is also very hard to say exactly what the RMT market estimates include. Summing up the current RMT estimates from Asia and extrapolating from the recent Station Exchange statistics, I estimate the total worldwide RMT volume to reach USD 1,820M 2,090M (see comment below). The main contribution to the figure comes from the Asian market.

Perhaps a disclaimer is in place here. This article includes very rough estimates and heavy generalizations, all of which are not explicitly stated but should be apparent. I also rely exclusively on secondary sources of information.

What exactly is real-money trade of virtual property?

Even though many are used to thinking of virtual property as something that exists in full-fledged virtual worlds and MMOGs only, it is hard to find qualitative differences between buying a three-dimensional object in Second Life and a two-dimensional object in Tencent QQ or IRC-Galleria. According to Fairfield’s definition, both are virtual property. While virtual worlds get more attention in the press, other community services may have significant RMT as well.

In addition to including only the virtual world RMT, it seems that RMT is often restricted to the so-called secondary markets. These include the user-to-user trade in e.g. eBay or in dedicated marketplaces and, more recently, also the IGE-type trade through a middleman.

Apart from the secondary market of MMOG items, there is also a kind of primary market for virtual property, in which the operators of certain services sell virtual items to users for real money. Such services are usually free to use. Typically players buy the service’s own internal currency with real money, and use that currency for microtransactions inside the service. Examples of virtual property sold at such primary markets include the virtual furni, basically tradable personalization of the users’ accounts, in the Habbo Hotel; the cars, car paintings, weapons, and other items sold in Kart Rider (played mainly in Korea); the land sale in the hyped Second Life; and the virtual personalization items of accounts in Tencent QQ, the Chinese instant messenger service.

In addition to the typical MMO game avatars, there are other types of “avatars” as well, e.g. desirable Tencent QQ and ICQ numbers, which can be sold. The Q-coins, the internal money of Tencent QQ, are also an interesting issue. If virtual items sold by the operators are to be included, what part of the trade that happens with Q-coins should be counted in? The purchases of personalization items at the least, because they are clearly virtual property. However, the Q-coins are used also as e-payment for purchases of “real” goods and services, so all of the Q-coin trading cannot be included.

The list of controversial examples could be easily continued. Even if the trade from the operator to user is not counted in, there are quite a few open questions concerning the RMT concept. Drawing the line between the primary and secondary markets is artificial, since the motivations to purchase items in either market are likely overlapping, and in both cases real money changes hands for virtual property. Why should the real-money trade of virtual property as a term include only certain kind of real-money trade of virtual property?

The scope of some popular estimates

Some of the current estimates are collected in the following table.

RMT estimates

Edward Castronova published the first estimate of the volume of RMT in 2001. The figure was based on measuring the daily volume of Everquest-related RMT transactions, and estimating the yearly volume based on the daily volume. On one day at one web site, the total volume of RMT auctions was USD 12,900. The site in question was presumably eBay. On a yearly basis, assuming constant trade volume, this adds up to about USD 5M. As for an estimate of the total RMT of Everquest in 2001, doubling the figure to about USD 10M has been proposed.

This first well-known estimate of the RMT market is, to my knowledge, the only one for which the source of the data and the exact method of aggregating the data are known. Assessing the scope and the reliability of the following estimates is harder.

Castronova gave a slightly more recent and more complete estimate in a New Scientist interview in 2004. He estimated the total volume of eBay and Korean ItemBay RMT transactions of MMOG virtual items and currency to add up to USD 100M. The rest of Asia, all other types of virtual property, and all other marketplaces are not included in this figure.

Slightly later in 2004, Steven Salyer, the president of IGE, announced that the volume of the secondary RMT market reaches USD 880M. In a Guild Wars interview, (available only via web.archive.org) Salyer said that the figure is an estimate of the annual worldwide secondary market size. It would seem safe to assume that the figure includes only the kind of virtual property IGE is providing, i.e. MMOG currency, items and avatars/accounts.

The same types of virtual property were included in the USD 100M estimate by Castronova earlier. We could form an estimate of 2004 total MMOG secondary market RMT excluding eBay and Itembay: this amount would be USD 780M. The share of RMT done in eBay was small, maybe surprisingly so.

The huge Asian RMT market

The volume of the Asian RMT market is very high in some recent estimates. An estimate of the Korean RMT market only, given by Korean Game Development and Promotion Institute (KGDI) (see the English source by Jun Sok Huhh at Gamestudy), is as high as USD 830M. The Gamestudy article also mentions poker and traditional Korean card games, but it is not clear whether these are included in the KGDI estimate or only in the separate survey that was made on RMT composition. If online gambling is actually included, it’s not surprising the estimate is very high. Steven Davis assumes that the estimate contains only the secondary market, though the original English source does not include this information.

There is also a way to estimate the Korean primary virtual property market. In Korea, free-to-use services are very popular. The operator of the service collects revenues from selling virtual property inside the service. Summing the revenues of some of the largest players in the Korean primary virtual property market gives an estimate of the total primary market in the country.

The popular Kart Rider by Nexon is free to play, but better cars, decorations and such can be bought using real money. The revenue model in other Nexon’s games (e.g. the 50-million-user Maplestory) is basically the same. The total revenue of Nexon includes virtual item sales, advertisement sales and licensing etc. It has been reported that “most” of Nexon’s revenues (USD 230M in 2005) come from virtual property microtransactions. I assume three quarters of the revenues can be accounted for virtual property sales. Nexon’s share of primary virtual property market is USD 170M. Cyworld, a Korean social network -type service, provides free accounts, or users’ 3D rooms. However, personalizing one’s account with virtual items is not free. The estimate of Cyworld’s total virtual item sales in 2006 is about USD 100M. The estimate for the total Korean primary RMT market is thus USD 270M.

According to Chinese government statistics (via PlayNoEvil), the total value of “virtual item trading” in China was about USD 901M in 2006. The figure reportedly includes the domestic consumption in online gaming (i.e. excluding gold farming), which takes about USD 514M of the total. The rest probably come from e.g. the Q-coin purchases in Tencent QQ. However, it might be very hard to actually separate the different uses of Q coins, so the figure may include also non-RMT transactions.

Extrapolating the EU/US secondary market

The recent white paper on Station Exchange (SE) also sheds some light into the secondary market MMOG RMT volume. The sanctioned RMT of Everquest II virtual property in Station Exchange reached USD 1.87M during one year. A total of about 40,000 users were registered on the two servers in which RMT was sanctioned. The introduction of Station Exchange had very little or no effect on unsanctioned RMT on other servers. It might be fair to say that RMT is not very much more likely to occur on the sanctioned server. I assume that on the average one Everquest II user uses about as much money on RMT as any other MMOG user. If the reported USD 1.87M is as large a part of the total RMT as the 40,000 users are of the total MMOG user base, we can form an estimate for the grand total worldwide secondary market. According to Bruce Woodcock, MMOGs had about 12.5M users in June 2006, so the estimate of the secondary market total RMT is about USD 585M. The 12.5M users include only paying users of MMOGs, i.e. users with a monthly subscription or similar.

Though the monthly subscription is in dominant use in EU/US games, the 12.5M users include also subscribers of popular Asian games. To produce a EU/US estimate, only EU/US users should be included, but is impossible to find exact numbers of the regional distribution of MMOG users. As an estimate, I exclude the users of both Lineages I and II, and the users of Chinese servers of WoW from the 12.5M total users, and end up in 6.1M users (all user statistics come from the above-mentioned source). My extrapolated estimate of the size of MMOG secondary RMT market in the EU and US is therefore about USD 285M.

The virtual property in the sanctioned RMT servers of Everquest II is traded also at other marketplaces, which is not taken into account here. This figure is only a lower limit for the total RMT volume. There are obviously many other sources of errors in this number, including any differences between an average MMOG user and an average SE-server Everquest II user.

An Asia-centric aggregate estimate of the worldwide RMT market

Let’s make some assumptions and try to come up with an estimate of the total worldwide RMT volume.

  1. Take the total value of gaming-related virtual item trading given by the Chinese government statistics as-is. The figure is USD 514M.
  2. Take half of the remaining USD 387M reported by the Chinese government, because e.g. some of the Q-coins are used for purchases of “real” goods and services, and some are used to purchase virtual assets. This figure is USD 193M.
  3. Take the RMT volume reported by the Koreans as-is. Assume it does not overlap with the Chinese government estimate. Assume it does not include gambling, but includes both the primary and the secondary markets. The figure is USD 830M.
  4. Take the extrapolation from SE numbers as the EU/US MMOG secondary RMT market value. This figure is USD 285M.
  5. Update: Take the estimate of the Korean primary RMT market that was derived from the revenues of Nexon and Cyworld. This estimate is USD 270M.

Summing up the estimates 1-4 1-5, the worldwide grand total is about USD 1,820M 2,090M. (see comment on the updates below)

As commented in the beginning, this figure lies on heavy assumptions, but then again, is transparent in the sense that its sources are traceable. Especially, it should be noted that the EU/US estimate includes only the MMOG secondary market, whereas the Asian market includes also other types of RMT.

Vili Lehdonvirta also contributed to this article.

SOE releases white paper on Station Exchange

Yesterday, Sony Online entertainment (SOE) announced the findings of a white paper on their official EverQuest II virtual property auction site Station Exchange. What’s interesting, there are now verified numbers of RMT available. The findings of the paper are based on data collected during the first year of operation of Station Exchange: from its launch in June 2005 to June 2006. From the numbers in the press release, it seems that most of the transactions were quite small, and most of the virtual assets were sold with a set price instantly. Interestingly, the paper also reports that most of the players who earned money in Station Exchange did so by selling items they had quested or crafted in EQII, not by buying items and selling them at a higher price. The total transactions amounted $1.87M. Some traders seemed to make a hefty profit selling virtual items, two traders reaching profits over $37k. Apart from the top sellers, the paper reports that “many more” players earned between $200 and $500 per month during the year. The average spending in Station Exchange was little over $60. There is also some data on the user base of Station Exchange. The 34-year-olds were the biggest buyers, while 22-year-olds were the biggest sellers. The traders turned out to be mostly males, though the average consumption did not depend on gender very much. Perhaps, as a very rough generalization, a typical buyer would be a little older player who needs something for e.g. a quest asap, and a typical seller would be a little younger player who makes relatively small profit (covers the subscription fee plus some more) by selling assets accumulated during game play. More information can be found in the SOE press release. Edit: commentary & the white paper can be found at Gamasutra