Let’s take the very simplest form of a derivative – a forward contract. A forward contract gives you the obligation to buy or sell at a point in the future for a price determined today – it’s a form of price insurance. I’m sure we can agree that the generic ‘impression’ isn’t a fungible good. The value of a top-of-page placement on a well-regarded site is very different from the value of a below-the-fold spot on a forum archive full of spam. So you can’t trade an impression for an impression without knowing more about it.

Let’s assume that I’m buying a forward contract on 10MM of these generic “impressions,” to be delivered on a predetermined date in the future. So I’ve got the obligation to buy the impressions at the price agreed to in the contract. Since impressions aren’t fungible, how should I price that forward contract to reflect the value of what I’m getting? Since any 10MM impressions can fulfill the contract, and since I know that only a fool would fulfill a contract with higher-value inventory than he has to, I have to assume that my contract will return 10MM extremely-low-value impressions, and therefore it’s worth almost nothing (and there’s no point in buying any.)

Of course, no one would buy or sell a generic ‘impressions’ contract. So let’s get more specific – assume you’re buying a forward contract on 10MM impressions from a specified ad unit on the front page of Yahoo! That *sounds* more fungible – you could trade one impression on the front page of Yahoo for another impression at a standard price, right? But in fact you can’t. Because the seller (not necessarily Yahoo – he could be buying and reselling impressions) has no incentive to send you any more valuable inventory than he has to, he’s going to send you the least monetizable impressions he’s got – for instance, countries that monetize poorly. The guy who sold you that Yahoo contract will just buy impressions in real-time on the spot market, separate the low-value ones from the high-value ones using geotargeting, send you the low-value ones, and keep the high-value ones for his own ads. In order to not get taken advantage of, you’ve got to assume the 10MM impressions you’ll receive are the 10MM lowest-value impressions and price the contract accordingly – near-worthless.

Even if you think you’ve absolutely nailed down the contract – “10MM impressions from an ad unit from Yahoo’s front page, all from U.S. males between 25-34, spaced evenly over the course of the day, with the primary browser language set to English” – someone out there’s got a little proprietary information about the users behind those impressions and will use it to separate the good from bad ones and just sell you the bad ones.

The only way you can avoid this is to insist that the forward contract be sold only by the original publisher and that the original publisher not do any prior analysis to separate the high-value impressions from the low-value impressions – but we’ve already got this, it’s called a guaranteed insertion order. It turns out the only type of forward that could exist already exists. (Even this contract is based on levels of trust inappropriate for the data age – I wouldn’t be at all surprised if sharp and unethical publishers were doing a little pre-analysis and sending their worst impressions to advertisers who buy CPM and don’t measure.)

There’s only one exception I can think of in derivative-land that might work, and that’s an advertising price index future, based on an extremely wide collection of inventory. I’m skeptical about that for a different reason – not enough volatility. But that’s another comment, and I’ve written enough.

By the way, it’s ironic that you used the example of diamonds, which, like ad inventory, are liquid but not fungible. Every attempt (and there’s been a few) to create diamond futures has gone very badly.

That’s where data comes in – and I’m super excited about that,

The real issue, in the medium term, seems to me not one of liquidity but more about managed expectations. Today, markets and exchanges are the inventory liquidation of last resort, yet there is so much inventory they are awash in impressions. Most large web sites optimistically desire to sell their inventory at a substantial mark-up over the value they receive from the exchange. As long as direct sales forces are able to command such large premiums selling outside the marketplace, it is unlikely that the optimization of a network or exchange can generate enough lift to overcome it. The result is that no publisher with unique and differentiated inventory would provide inventory to fund the futures markets that people are building. The result is silos that prevent true marketplaces from emerging and efficient pricing from dominating the market.

We are a long way from market efficiency. Of course, that creates opportunity!

Fungibility is different from liquidity. A good is liquid and tradable if it can be easily exchanged for money or another different good. A good is fungible if one unit of the good is substantially equivalent to another unit of the same good of the same quality at the same time and place.

Fungibility does not imply liquidity, and liquidity does not imply fungibility. Jewels can be bought and sold (the trade is liquid), but individual diamonds are not interchangeable (diamonds are not fungible). Zimbabwean dollar bank notes are interchangeable in London (they are fungible there), but they are not easily traded there (they are not liquid in London).

Unfortunately I can’t read the article because of NY Times walled garden. Sucks for them.

We should get lunch and catch up – despite my belief that this stuff will fail, I’m still a securitization geek and would love to talk about it.