"Paper losses don't exist until you sell."

The resignation of Citigroup chief executive Charles Prince and news of additional writedowns at the firm have raised the profile of hazy Wall Street phrases such as Collateralized Debt Obligations and Mark-to-Model.

Why aren't losses being seen? Even now, they aren't being seen.

The answer, we now know, is mark-to-model. What does "mark-to-model" mean? Let's use something even we can understand - a sports card example. Suppose you and I are collectors (investors) in sports cards, but not baseball cards, the "prime" sports card market. No, we collect professional golfer cards - the "subprime" sports card market instead.

Why would we collect professional golfer cards? Simple.., we're looking for an a fancy way of saying "excess return" - and by using leverage we believe we can buy these illiquid sports cards and sell them later to someone else for more money. Ok, ok, there are a couple of problems with this scheme you can see already.

*First, the market for professional golfer cards is far, far riskier and smaller (illiquid) than the market for professional baseball player cards.

*Second, since they so rarely traded, it's difficult to value the cards on a day-to-day (even week-to-week) basis.

Ok, so back to the sports card market. Let's say that baseball cards are "highly liquid," meaning that, like stocks or U.S. Treasuries, they trade every day. These trades provide a way to instantaneously value our portfolio of baseball cards at any time. To value our baseball card portfolio, we simply look up the most recent trade and record the value. This is called "Mark-to-Market."

This "liquidity" is particularly helpful if we are using leverage (meaning, if we are using borrowed money to buy baseball cards with the hope that the borrowed money will increase our return when we decide to sell) since it allows us to closely monitor and track exposure and adjust the amount of leverage we are using accordingly.

So how do we know, at any given time, what our leveraged portfolio of professional golfer cards are worth? They rarely trade, so we can't look up similar cards that have recently traded in the market! Well, unfortunately, in our professional golfer card portfolio we can't mark-to-market because these cards trade so infrequently!

So how do we value our portfolio then?

Simple, we have some mathematicians build us a model that values the cards based on how each golfer performed last year, the tournaments in which they made the cut, their overall earnings and rankings among their peers, and rating.

Wait, did I say, "a model that values the cards based on how each golfer performed last year"? Yes.

But what if a professional golfer's card in our portfolio is a guy who last year ranked fourth overall in earnings and won two tournaments, but suddenly gets injured this year, fails to finish a few tournaments, and slips down to 40th in overall earnings?

Hmmm.., good question. For that we would rely on that separate "professional golf card agency" we mentioned to "re-rate" this card. Then we would simply input that revised rating into our models and adjust the value accordingly.

But what if the rating agency, for a variety of reasons, chooses not to re-rate the card?

Then we have a situation where the value of the card that is being spit out by our model is in no way even close to the true market value of the card.

Wouldn't that be a problem if we suddenly feared that all the ratings of our cards were too high? Wouldn't our model be insufficient? Might we not be over-leveraged in cards that have very little real market value? Yes, yes and yes.... That's when the shit start hitting the fan !

And that is precisely where we are right now with respect to CDOs.

The credit ratings agencies' ratings are key in the mark-to-model values, and so far very few CDOs have been re-rated in a way that reflects the surging subprime default rates.

There's an old investment saying that says: "Paper losses don't exist until you sell." That old saying is being tested in real time, right now!