My favorite talk of the first four days of GDC was Sulka Haro’s talk on The Evolution of Habbo Hotel’s Virtual Economy. Habbo Hotel is a virtual world that’s been around for almost 10 years now; its economy has gone through a lot of phases, each of which came with its own set of surprises. I doubt I’ll be able to do it justice here (Sulka said he’d post the slides on his blog, I’ll add an update when that happens), but I’ll give it a shot.
The phases that the economy has gone through so far:
- No in-game currency at all.
- Emergent unofficial currencies.
- A paid currency.
- A tradable paid currency.
- Dual official currencies.
- An official secondary market.
When the game started, there was no in-game currency. If you wanted to buy an item, you paid for it right there with real money, by sending a certain text message.
This had some problems: for one thing, there were social engineering hacks (people convinced others to send plausible-looking text messages that ended up buying items for somebody else), but also there weren’t enough price points. There were only two prices; for the cheap one, the carriers took off a large chunk, while the expensive one was too expensive for most of the items in game.
Perhaps in reaction to this second problem, players developed their own currency. From the beginning, players could trade items to each other after they purchased them; rather than trading desired items directly, they eventually settled on using a particular cheap chair as the standard currency. (So a player would sell you this thing for 7 chairs, that thing for 10 chairs, etc.) And, in fact, this unofficial currency system persists to this day, though the reference items have changed over time.
In 2001, they launched the game in the United Kingdom. Cell phone penetration wasn’t high enough there (especially among kids) to use it as the exclusive payment source, so they needed to do something else; but, again, they couldn’t afford to be bled try by transaction fees for the smaller amounts. So they need a way to split a 10-pound payment into smaller chunks, which led to:
At this stage, they introduced credits, and switched item purchase to work in terms of credits. One credit cost about 15 cents; cheap items cost 1 credit, standard items cost 2–3 credits, premium items cost 4–15 credits.
Here is where the speaker started putting up increasingly complex diagrams explaining the economy; I’m not up for trying to recreate them myself, which is one of the main reasons why I’m hoping the slides will appear soon! But one of his main points was that their economy went from only having one pool (the pool of objects) to having two pools (adding the pool of credits, which drained into the pool of objects). This made it harder to predict revenue: people could either purchase items on the secondary market or purchase items from saved up credit without having to spend real money to purchase credit. (Also, as he pointed out here: all the goods in their economy at this stage were persistent, which makes them a great value proposition, especially in the presence of the ability to trade, but also led to item value inflation.)
So, to help with predictable revenue, they added the Habbo Club. This is a monthly subscription; it’s paid for in credits, so it added another sink from the credit pool. It lets people customize their avatars; that way, subscriptions and item purchases don’t cannibalize each other.
In 2005, they noted that the number of players who were trading but not paying increased faster than the number of players who were purchasing new items: basically, there was too big an item pool in circulation. They thought about banning trading, but it’s great content for the game, so trading drives sales. Also, people who trade the most monetize the most. (At which point he made a side note: expect a power law in distribution of, e.g. paying players: metrics based on averages can give you very misleading guidance.)
So they wanted to continue to allow trading, but rethought it a bit:
At this step, they started allowing people to trade credits, not just trade items. The point here is that the most liquid currency is the most desirable currency: so you want people to be trading your currency, instead of trading chairs or rubber ducks.
And this, indeed, led to more buying of credits. I have some notes here that I don’t quite understand: they say that before, “buyers = spenders” (because you couldn’t spend credits if you didn’t buy them yourself), and that, before, “traders > buyers” (most of the action happened in the item pool). Afterwards, it remained the case that “traders > buyers”, but now, “spenders > buyers” (because there were more ways to get credits to spend). With the effect (this is the part I don’t entirely get) that there were more buyers. (I guess because the buyers were feeding not only the trading pool but the new pool of non-buying spenders?)
There were, of course, unanticipated side effects. Before this change, the smallest unit of currency was a rubber duck, which people had decided was worth about .1 credit; after this change, a duck ended up worth 1 credit. So the cheapest items went up in value, and the total value of the item pool went up significantly.
He also showed a picture of a room full of gold bars that was a problem; I don’t quite understand either why this happened (why wouldn’t rich people store their assets in credits?) or why it was a problem.
At any rate, they had inflation. Which raised the question: is there any way to establish a sink for persistent goods after the fact?
The answer that they ended up with was the “ecotron”: you could put five items in, and it would give you a random higher-value object back. This turned out to be very successful: players liked it a lot, low value items started disappearing, which drove more demand for high-value items.
But, even with that, inflation was a problem. The credit sales were still unstable, and new players had a hard time entering the market: the very presence of items that people were selling for $200 discouraged them from participating. Also, inflation served to discourage people from buying credits: because of inflation, their value dropped as soon as you bought them.
A side note on inflation: you have to be careful as to how you measure it: as the user base grows, the size of your pools naturally grows, so there’s a moving target here. The metric they ended up using was the ratio of liquid cash to the number of people trading in the secondary market. (Another metric he mentioned was a “consumer price index”: the value of a standard basket of items.)
At here, he also mentioned something about having inflation arise from giving away the paid currency for free. If I’m remembering correctly, Habbo Hotel never did this, it was something that they considered and rejected. But they still wanted to have something they could give to users as a reward. So:
They introduced a secondary currency, called pixels. It had a different role than credits, the paid currency. Specifically:
- Get by buying or trading.
- Use to buy persistent goods, expendables and services. (I think, I can’t read that last word in my notes.)
- Its purpose is monetization.
- Get by performing in-game actions.
- Use to buy expendable items, and for discounts on the credit market.
- Its purposes are user retention and conversion (? I can’t read that last word in my notes), and to support the primary market.
I don’t entirely understand this—in particular, if my notes are correct and you can use pixels for discounts on the credit market, then doesn’t that lead to inflation concerns?
One question that people sometimes ask: we only have one currency in the real world, why does Habbo Hotel have two? His answer is to reject the premise: aside from different countries’ currencies, we also have frequent flyer points, stocks, stock options, mortgages, and in general no end of financial instruments that can play the role of currencies. (Also, by his count, World of Warcraft has 39 currencies!)
The two currencies and the two sorts of items ended up working well together: e.g. players could build a race track out of persistent items purchased with credits, but to drive on the race track, people would need to purchase expendable items with pixels.
Anyways, continuing along the line of reasoning that led them to make credits tradable, they asked the question: if removing credit frictions leads to increased sales, and if trading correlates to sales, then wouldn’t we expect that removing trading friction would also increase sales? It should: removing friction increases participants, and participants correlate to sales. So:
The introduced an official market for purchasing items. Before this, to buy an item, you had to run around looking for somebody who could and would sell it to you; this was a pain for the buyer, and it also negatively affected the experience of players who didn’t want to participate in the transaction. In contrast, the official marketplace was a lot easier to use, and players felt a lot safer; the result was that a lot more people participated.
There was also quite a lot there for traders to love. The official marketplace gave you lots of information about the average sales price of each item, both right now and historically; players would spend hours just browsing the sales catalog, gathering information and looking for (and finding!) arbitrage opportunities.
They also used this this as an opportunity to introduce a further sink into the economy. Listing items wasn’t free: you had to buy a ticket to list an item (they came in packs of 5 selling for 1 credit), which let you list an item for sale for 48 hours. And, if the item sold, the company took 1% of the sale as a commission; but the commission was a minimum of 1 credit, so in practice commissions were actually quite a bit higher than that.
And, the final takeaways:
- Economics helps you create a sustainable virtual economy, which is necessary for a sustainable virtual world.
- Think of economics in a broad sense: don’t just focus on prices, also focus on making the world a safer place for players to invest time and money.
This post has not been revised since publication.