## A straightforward explanation of the present financial crisis (part 1)

Perhaps you’ve noticed that some funky stuff is going down in the financial markets.

Most normal people seem to understand finance the way I understand physics. I understand that, if you’re holding a brick of lead and you drop it, it will hit your foot. I seem to recall that the brick will accelerate at 9.8 meters per second squared. Apparently this has something to do with general relativity. Perhaps if I boned up on it for a year or two, I could gain some vague understanding of GR. But I am happy to leave it to the physicists.

If you think of finance this way, I’m afraid you have been misinformed. First, finance is not a science. Secondly, finance is not intrinsically complicated. And third, unless you are a physicist, your life will be pretty much the same whether you understand relativity or not. “You may not be interested in war,” Trotsky once said, “but war is interested in you.” The same is surely true of finance.

A better way to think of our present financial system is to compare it to Microsoft Windows. Windows is indeed complicated. It is astoundingly, brilliantly, profoundly complicated. If there is anyone in the world, even at Microsoft, who understands all of Windows, he or she must have a brain the size of a basketball. (One would think this individual would have at least been photographed at some point.) It took them six years to ship Windows Vista, and I really have no idea how the hell they did it.

But Windows is not intrinsically complicated. It is a time-sharing kernel with a graphical user interface. A smart undergraduate can write an time-sharing kernel, with basically the same functionality as Windows, as homework in a semester. The same can be said of the GUI, which is perhaps a bit more code but is certainly less complex. (I wouldn’t have the same student write both. I do feel this would be a little much.)

Windows is complicated because it is both ancient and mature. It has been around for a long time (20 years), and it has been used to solve just about any problem an OS could conceivably solve. Including quite a few that it has no business at all in solving.

Exactly the same can be said of the modern financial system. The difference is that Western finance is ten times as old as Windows. If not twenty or thirty. The result is that, as with Windows, people can spend their entire lives trying to understand a single small corner of it—say, real estate, or commodities, or bonds, or whatever. To actually understand all of modern finance, you would need a brain the size of a beachball. This would certainly attract notice, and you might have some trouble getting laid.

Learning Windows (or Linux, or OS X, or whatever—they are all humongous) is an awful way to understand operating systems. You can put as much work into it as you want, and all you will learn is a tiny scrap of the basketball. So even the people at Microsoft who work on Windows all day do not have degrees in Windowsology. They learned OS the same way I did: by studying the principles, which are not at all complex, and then writing a toy OS.

Unfortunately, if there is any equivalent of the toy-OS approach in finance or economics, I have not run into it. But this is UR, and we never let a small problem like this stop us.

Certainly the closest thing to a toy economics is the Austrian School. But when you read the Austrians, you are still getting a good dose of Windowsology. Even a very thorough Austrian treatment, like Rothbard’s, exists to describe the financial system as it is, not as an undergrad would build it. Worse, the founding treatise of the modern (Misesian) Austrian school dates to 1912, and the Austrians have spent most of the century since just fending off their many enemies. So you are actually getting a sort of Hapsburg Windowsology. It is an entertaining task to try to map Mises’ terms to the contents of today’s WSJ, but I can’t say it’s especially easy or productive.

So I thought I’d give the toy-OS method, which (thanks to Conrad Roth) I call nitroeconomics, a spin in explaining what’s up with all these SIVs, CDOs, and of course those famous subprime loans. Ideally, the less you know about finance, economics or accounting, the better. If you do have some understanding of this area, please try to suppress it.

Of course our toy financial system will exist only in our shared imaginations. However, we do need to imagine building it, because we are assigning the task to our imaginary undergraduate. One easy way to do so is to set our actors not in the real world, but in a virtual world.

This is kind of cool, because as everyone who hasn’t been living in a cave for the last ten years knows, virtual worlds really do have real economies. (Here is one blog, unfortunately quite Windowsological, devoted to virtual economics.)

Unfortunately, the virtual worlds that exist today tend to have very primitive financial systems (think Windows 1.0). They are more focused on killing orcs or chatting with giant inflatable penises or whatever, than options and equities and mortgage-backed securities. And what they do have tends to be copied from the “big room,” which gives us more Windowsology.

Nitroeconomics (if you want to sound more scientific you can call it synthetic economics) is different. It is set in the virtual world of Nitropia, which doesn’t exist but easily could. Orcs and inflatable penises are no strangers to Nitropia—there is no economics without fun—but the main point of the place is that it has a financial system which is simple and makes sense.

We can use nitroeconomics to understand real situations in the real world, such as the subprime crisis, with a simple three-step process. First, we set up a scenario in Nitropia which (we claim) is analogous to the real-world scenario we’re trying to understand. Second, we figure out what would happen in Nitropia. Third, we connect this back to the real-world result. While the process is certainly nontrivial and offers great latitude for error, it is also a lot easier than Windowsology. Especially if you know nothing about Windows.

The first cool thing about Nitropia is that it has no financial system at all. Unlike other, inferior virtual economies, it does not distinguish between “money” and other virtual objects. A monetary token in Nitropia is an object like any other—a magic sword, an inflatable penis, or whatever. A player in Nitropia who has a lot of money just owns a lot of these tokens. There is no special, separate “bank balance.”

Today I am going to carefully skip over the issue of what “money” is, and why some goods are “money” and others are not. This is a fascinating question and we will return to it. However, the answer does not help us understand the subprime crisis.

Rather, I will just declare by personal ukase that the monetary token in Nitropia is the cowrie. Cowries have the following properties:

• Every cowrie is identical to every other cowrie.
• Any number of cowries can be trivially stored or carried.
• To remain alive, a Nitropian must eat one cowrie a day.
• Cowries are not (directly) useful for any other purpose.
• Nitropia contains a fixed number of cowries—$$2^{32}$$ or $$4,294,967,296$$.
• Eaten cowries reappear at a random point in Nitropia.

In other words, the cowrie system is a closed-loop financial system. You probably understand intuitively why this is ideal. But we’ll certainly return to the subject.

Nitropia is divided into two parts: Nitro City, where Nitropians hang out and chat with their inflatable penises, and the Dungeon of Yendor, which is infested with orcs and other monsters, but also contains magic swords and other useful trinkets. At the beginning of time, all cowries are in Yendor, so Nitropians need to kill if they want to eat. A small, monsterproof door in Nitro City leads down into Yendor.

The second cool feature of Nitropia is a frictionless market. The market only works in Nitro City—there is no coverage in Yendor, so to speak. In Nitro City, any Nitropian can teleport any object (cowries, magic swords, inflatable penises, etc.) instantly to any other Nitropian. There are no delivery or other transaction costs.

Moreover, Nitropians can solve the ancient who-gives-first problem by conducting specified exchanges, in which all goods are as described or no trade takes place. And Nitropians of an especially geeky bent can upload trading bots which make automatic trades, allowing them to design any market mechanism whatsoever.

So we’ll assume that for every object O, there are market bid and ask prices in cowries. The ask price is the minimum number of cowries that anyone who has an O will accept in exchange. The bid price is the maximum number of cowries that anyone who wants an O will pay. For common objects, ask and bid will probably be close. Of course, prices change all the time with the rich, complex and constantly-changing lives of Nitropians.

The third cool feature of Nitropia is a system of formal promises, in which Nitropians can promise to perform some action in future. Promises in Nitropia are transferrable objects, just like cowries or magic swords. The beneficiary of the promise is the holder of the object. Thus for any promise there is a promisewriter and a promiseholder.

A common standard promise is a promise to deliver some quantity of cowries Q, at some maturity time M. We call one of these promises a ticket. The difference between the present time N and the maturity time M is the term of the ticket.

Promises in Nitropia are automatically enforced. If a Nitropian writes some promise and fails to deliver, she is said to choke. When a Nitropian chokes, her account is liquidated. All goods in it are sold on the open market. All promises written by the liquidated Nitropian are cancelled, prorated by the current ask price of the promised good, and refunded to the promiseholder from the proceeds of the liquidation sale.

Nitropians who write promises can also accept restrictions. A restriction, also automatically enforced, is a promise with no beneficiary. For example, a promisewriter may promise to stay out of Yendor (for obvious reasons). Restrictions may also include disclosure of any and all information about the promisewriter, including objects possessed and other promises written. In general, restrictions are useful because a promise from a Nitropian who accepts certain kinds of restrictions may well sell for more cowries. More on this subject later.

Okay. We now know enough about Nitropia to understand the subprime mortgage crisis, or at least its simplified Nitropian equivalent.

The cause of the crisis is a pernicious bit of financial engineering called term transformation. If Nitropians are behaving sensibly, they will avoid term transformation. Nitropians are just people, however—fat, lonely, pasty-faced people with no life outside their computers—and we can’t be too surprised when they do non-sensible things. Why high-powered jock traders on Wall Street were playing the term-transformation game is another question entirely, and we’ll save it for another day.

To understand term transformation, we first need to understand the market for tickets.

To understand a market for any good is to understand why it costs what it does. Since all prices are set by supply and demand, we need to know the motivations of the buyers and sellers. Of course, the seller of a ticket can be its original writer, or someone else who bought the ticket and now wants to sell it again. But for simplicity, we’ll start by assuming the only participants are the ticketwriter and the initial ticketholder.

We’ve defined a ticket so that there are only three variables: the term to maturity, the quantity of cowries to deliver, and the identity of the ticketwriter. The identity of the ticketwriter determines the probability that the ticketwriter will choke before delivering the cowries.

Assume for the moment (we’ll un-assume it in a minute) that no ticketwriters choke. For these magical risk-free tickets, which cannot exist in nature (unless Nitropia’s sysadmin changes the rules), we have only two variables: quantity and maturity.

We can eliminate the quantity by dividing the current market price P of the ticket (in cowries) by the delivery quantity Q of the ticket (in cowries). This gives us a nice unitless number, P/Q. What can we say about this number?

I assert that for any ticket of any nonzero term, P/Q will be less than 1.

Suppose P/Q is greater than 1. Then the ticket buyer will be exchanging P cowries at time N for Q cowries at time M, where P > Q and M > N. But, since cowrie storage is free, she would do better by just holding onto her P cowries until M. If P/Q is 1, she won’t lose, but she won’t gain either. Why bother? Since all exchanges require both buyer and seller, none will happen.

We can define P/Q as the discount of the ticket. Then we can use the continuous interest formula to normalize this to an interest rate or yield, which by Nitropian convention is specified not per year, but per day. (Life is fast in Nitropia.)

For any ticket, if we know any two of discount, yield and term, we can compute the third. Thinking in terms of yields rather than discounts does not eliminate the term variable. But it lets us ask a very interesting question: in the Nitropian ticket market, how will the yields of tickets with different terms compare? (Remember, we are still assuming that no ticketwriters choke.)

The answer is that yields at longer terms will be higher. This is not as easy to show as the fact that there will be no negative yields. But it’s worth working through.

To understand why longer terms mean higher yields, we have to consider the reason that Nitropians buy and sell tickets in the first place. We are assuming that these instruments will even exist. But why should they? Why is there a market for tickets?

A ticketwriter is promising to deliver Q = (P + X) cowries later, in exchange for P cowries now. She cannot fulfill this promise without engaging in some activity which is profitable. Please note that this is not a value judgment. It does not make the activity good or bad. It just means that it takes in P cowries now, and produces P+X cowries at M = (now + T).

(In fact, the ticketwriter will only write the ticket if she expects her activity to produce (Q + Y) cowries, because otherwise she makes nothing at all on the deal. Everyone’s gotta eat.)

The ticketbuyer is exchanging P cowries now for Q = (P + X) cowries at M. His motivations are simpler. He always has the alternative of just storing the P cowries until M. But, if he uses the ticket instead, he will get X more cowries.

So we have both buyer and seller, and the transaction happens. But what yield does it happen at? The yield is defined by the discount, which is defined by the price of the ticket, which is set by supply and demand among buyers and sellers. Contra most medieval thinkers, there is no just or righteous interest rate, derived from the Bible and the eternal laws of God. If we want to know what the cowrie yield is, we have to actually instantiate Nitropia. And, like any price, this yield may change over time.

And even at any one time in the Nitro City market, there is not just one yield. There may be different yields for different terms. In other words, there is a yield curve. We are now ready to see why this curve slopes upward—i.e., yields at longer terms are higher.

How do you actually turn a profit in Nitropia, anyway? Not that the ticketbuyer cares at all how this feat is accomplished—but there is one obvious productive process. Nitropians can descend into Yendor, kill monsters, and relieve them of their cowries and other good stuff. Of course, the monsters fight back, so there is no sure thing. And in order to fight monsters, you need weapons—magic swords, inflatable penises, or whatever.

So the ticketseller might use P, the cowries she gets for selling a ticket, to rent a magic sword, with which she descends to level 15 of Yendor, clears out a nest of foul orcs, confiscates their cowrie stash, and returns to Nitro City, where she returns the sword and pays off her ticket.

The trick is that all this takes time. Unless you find a way to hack the Nitropia servers, there is no such thing as an instantaneous profitable process.

So we can ask: in Nitropia, what will the yield be for a one-second ticket? Unless someone can figure out a way to turn P cowries into P+X cowries in one second, the yield will be zero. Thus (as we saw) no one will write any such tickets. And if there is some way to profit in one second, we can always imagine an epsilon term which is shorter. Thus, the Nitropian yield curve intercepts zero term at zero yield—regardless of the details of Nitropia.

We can generalize this observation by noting that there are always more kinds of longer productive processes than shorter ones. If you want to descend to level 417 of Yendor and take on the evil dragon Angstrom, it will take you three weeks round-trip. Whereas if all you are doing is heading down to 113 to knock off some random ogre, you can do it in a week.

But a ticketseller can construct the equivalent of one three-week dragonslaying expedition by stringing three one-week ogre-whacking trips back to back. Because it takes three weeks to slay a dragon, there is no way for the profitable process of dragonslaying to affect the market for one-week tickets. The converse, however, is not the case.

There is no law that says that dragonslaying has to be more profitable than ogre-whacking. However, if it is not at least as profitable as ogre-whacking, no one will slay dragons. If ticketbuyers still want three-week tickets, ogre-slayers will sell them three-ogre tickets. The short-term yield is a floor for the long-term yield. The Nitropian yield curve can be flat, but it cannot invert.

In theory, there might be no profitable processes at all in Nitropia. For example, suppose there are no cowries or goodies at all in Yendor. The yield curve will start at the origin and remain there. In practice, any world complex enough to be any fun at all, virtual or real, will have long-term processes that are more profitable than their short-term counterparts. Because any short-term process can produce long-term tickets but the converse is not the case, there will be more sellers at later maturities, and the extra competition will push up yields.

Now it’s time to relax the unrealistic assumption that killing monsters is risk-free. Obviously, it is anything but. Monsters have teeth and claws, they bite and scratch, they shoot fireballs and crush your bones in their huge armored pincers. If you never return from Yendor, not only do you not fulfill your promises, you have no assets to liquidate. The monsters keep it all.

This is tricky because, suddenly, comparing tickets written by different monster-hunters is comparing apples to oranges. Wall Street seems to have real trouble with this concept, but assessing risk is always a matter of subjective opinion. How do you know whether a hunter will return? You don’t. How do you know her probability of returning? You, um, guess.

So you guess. Let’s say you think your ticketseller has a 97% chance of coming back. Then, the expected value of her ticket is not Q, but (Q * 0.97). This simply lowers the price P that you are willing to pay for the ticket. So that we can calculate objectively, we still define the discount as P/Q. The risk of choking simply appears as a higher yield.

But wait. For most of us, a 97% chance of Q is not at all the same thing as a 100% chance of (Q * 0.97). Risk tolerance is subjective and nonlinear. This creates a market for risk dispersal.

If you, the ticketbuyer, want a 100% chance of (Q * 0.97), there is no way to achieve it. Uncertainty cannot be converted into certainty. However, you can produce a much more desirable probability distribution by buying not one ticket from one hunter with a 97% chance of returning, but 100 slices of 1/100th of a ticket from each of 100 hunters, each of whom has a 97% chance.

A tricky transaction! At this point you, the ticketbuyer, are tempted to just let your cowries sit around “under the mattress.” But there is an alternative. You can buy one ticket from a ticket balancer, who in turn deals with 100 ticketbuyers and 100 ticketselling hunters.

The balancer does not hunt herself. She does not go anywhere near Yendor. She buys 100 tickets of Q cowries with one-week terms from hunters who she estimates have a 97% chance of returning. She then sells 100 tickets of (0.97 * Q) cowries to risk-averse buyers. Since the buyers are risk-averse, by definition they accept lower yields than she pays the hunters, and she profits. As described the balancer’s risk of choking is quite high, but maintaining a small buffer of, say, (10 * Q) cowries will make it very low, due to the law of large numbers.

What does a ticketbuyer need to know in order to buy a ticket written by a balancer? He needs to know that (a) the balancer will not wander into Yendor and get knocked off by an orc herself, and (b) the balancer holds enough tickets with sufficient expected value, plus some buffer, to have a high probability of delivering on the tickets she has written. There are no risk-free tickets in Nitropia, but a balancer can reduce risk to an arbitrarily low epsilon, of course at the price of lower yield.

Thus, the balancer must submit to restrictions and disclosures in order to write tickets which are perceived as low-risk. Rather than forcing every buyer of her tickets to do his own “due diligence,” she will probably hire an auditor who will inspect her restrictions and disclosures, and testify to her credibility. The auditor is then responsible for maintaining its own credibility in the eye of the public ticketbuyer.

Okay. This concludes our discussion of normal, healthy finance in Nitropia. (Note that we have not mentioned equity, i.e., stock. Our “tickets” are bonds, zero-coupon bonds to be exact. Briefly, a more sophisticated financial system would replace our one-size-fits-all liquidation procedure with different priorities of tickets, where high-priority tickets are paid off early if the ticketwriter chokes, and thus earn lower yield. Stock is a sort of zero-priority ticket for especially funky hunting expeditions whose return is wildly unpredictable. Its holders get no fixed payment at all, just whatever’s left over after everyone else is paid. Which could be huge, or it could be nothing at all.)

Now it’s time to transform some terms. Term transformation (this vice goes by other names which are easier to Google, but let’s keep the suspense for a little while) happens when a ticket balancer buys tickets which mature later than the tickets she writes.

At first, term transformation seems to make no sense at all. Then it makes perfect sense. Then you see what a truly awful idea it is.

Recall that in our explanation of ticket balancing, our balancer bought 1-week tickets from hunters who were going on 1-week expeditions, i.e., whacking ogres. She then sold 1-week tickets to her buyers, and the whole thing worked out perfectly. Her hunters, or at least most of them, came back with the cowries, which she promptly delivered. No sweat.

Suppose that instead of ogre-whackers, our balancer dealt with 3-week dragonslayers. Remember, since the Nitropian yield curve slopes upward, the yield on dragonslaying is higher than the yield on ogre-whacking. In other words, dragonslaying is more profitable per week than ogre-whacking. Not because dragons are wealthier than ogres, though they are. But just because if it wasn’t, no one would bother slaying dragons.

Thus, term transformation is an obvious way for the balancer to pocket a little extra scratch. She is still paying 1-week yields, but she is earning 3-week yields. But, um, there’s a problem—when her ticketholders come back in a week, what does she say? “Um, we’ve had, some, um, bad weather. Also the ogres are unexpectedly recalcitrant. Also, um, actually, I was just heading out. Can I call you tomorrow? I’m sure we’ll get this whole thing cleared up.”

This might fly in the real world. But not in Nitropia, whose ice-cold, iron-hard law is enforced not by mere flesh, but merciless and inexorable software. As soon as the balancer breaks her first promise, she is locked out of her account and her assets go straight to the block. Ouch.

So how can term transformation happen? It seems that we are trying to teleport cowries not through space, but backward through time. Even in Nitropia, there is just no way to convert tomorrow cowries into today cowries. Besides, as we’ve seen, balancers need to be audited, and even if our balancer is so foolish as to think she can pull cowries magically out of the future, no auditor could conceivably agree.

We cannot understand term transformation without looking at the motives of the buyer. After all, the auditor exists only because the buyer demands this service. If the buyer accepts that term transformation is good and sweet and true, surely the auditor will agree.

First, we notice that, since the balancer is after all earning a higher yield from her dragonslayers, she can pay a higher yield to her customers. What will probably happen in practice is that she will split the profit with them. So the balancer wins and the ticketbuyers win. In fact, since term transformation creates a higher demand for long-term tickets, thus driving down the yield (if not quite to ogre-whacking levels), the dragonslayers win, too. Everyone wins. It’s a win-win-win situation. At least if we can solve this pesky liquidation problem.

Second, we notice that liquidation is not as bad as it seems. Especially in Nitropia, where the process is fully automated and demands no ridiculous, periwigged judicial official. What matters to the ticketbuyers—and hence to the auditors—is that they get their cowries. And, since the liquidation process has plenty of assets (the dragonslayer tickets) to sell, there is no reason at all to think that the ticketbuyers will take a haircut.

Quite the contrary, in fact. Remember that, to deal with the vagaries of monster-hunting, any balancer must maintain a small cowrie buffer to get on the good side of the law of large numbers. With this buffer, plus the proceeds from selling the dragonslayer tickets on the open market, the ticketbuyers may even profit when the balancer chokes.

Of course this makes no sense. Liquidation cannot be profitable. If liquidation is profitable, there is no need to liquidate. Ergo, the balancer herself can just sell the dragonslayer tickets, using the proceeds to redeem the tickets she has written. She has thus earned a week of dragonslayer yield, not ogre-whacker yield, and has still come out ahead. Everyone still wins.

Thus, the auditor doesn’t need to worry at all about the term of the balancer’s assets. The auditor can just verify that the total present market price (in cowries, of course), of these assets meets or exceeds the number of cowries that the balancer is obligated to deliver. As long as this criterion, which we can call scalar solvency, is met, there is no problem.

Furthermore, this idea that the balancer’s customers all show up after a week and start pounding on the doors, demanding their cowries, is not at all realistic. We have spent a considerable amount of effort in understanding the motives of ticketwriters. We need to think more about the end buyers.

Why do buyers buy one-week tickets, rather than three-week tickets, even though the latter produce a higher yield? It can only be because they feel they might need their cowries in one week, rather than three.

Perhaps they will. But perhaps they won’t. The future is full of uncertainty. If buyers knew what they would exchange their cowries for next week, it’s very likely that they would just exchange them now. They lose a week’s yield, they gain a week’s use of an inflatable penis. But Nitropians hold cowries and/or cowrie tickets, rather than magic swords or inflatable penises, because they are not sure what they want to buy in the future, or when they want to buy it. And because cowries cost nothing to store, they are a good choice for a rainy-day fund.

So it’s very likely that the balancer’s ticketholders, when they get their cowries back in a week, will simply want to buy another week’s ticket. The rainy day has not arrived. The cowries go back in the kitty. We call this rolling over the ticket.

Rolling over is really quite convenient for everyone concerned. If all her buyers roll over, as they may well, the balancer does not have to sell any dragonslayer tickets before their time. She can make this even easier by stipulating that, by default, her tickets roll over. If one of her customers really wants his cowries after only a week, the request will of course be satisfied promptly and courteously. If all the customers demand redemption, there will perhaps be a little more of a Chinese firedrill as dragonslayer tickets are sold, but the balancer maintains a comfortable margin of scalar solvency, so again, everyone’s ticket is redeemed.

Furthermore, now that we’ve made term transformation work so well, what is this whole week thing about? The number no longer bears any resemblance to anything. We have long since given up on the ogre-whacking. The ogre-axes have been beaten into dragonswords. Surely, rather than insisting that her customers wait a whole week whenever they want their cowries, she can shorten the term—say, to a day. Or an hour. Or…

In fact, the magic of term transformation has solved a problem that we didn’t think was solvable at all. It has produced a yield on a zero-term ticket. As we adopt automatic rollover and we adjust the term of the balancer’s tickets down to epsilon, term transformation creates a fabulous new construct, the demand ticket.

A demand ticket is an instrument that pays the bearer a quantity Q of cowries on demand. It is just like holding cowries, only better. Because until you actually demand your cowries, your demand ticket produces a yield. And no mere in-and-out orc-mugging yield, either. But the maximum, bona fide, long-term, dragonslaying yield.

Sometimes the dragon wins, of course, and we do have to subtract that. And the balancer, like everyone, needs to eat. But overall, term transformation produces a win for everyone. The end buyer gets maximum yield and total flexibility, the balancer gets more of a taste, the auditor still has to audit, and the dragonslayers get more investment at better rates. Perhaps the ogre-whackers aren’t too happy, but they just have to bring their game up to the point where it’s as profitable as dragon-slaying. If that can’t be done, perhaps the ogre ecosystem could use a bit of a rest, anyway. And aren’t we, the brave hunters of Nitropia, stronger and better men and women if we pit all our sinews against that noblest of beasts, the dragon?

Alas. If only it was true. If term transformation really did work, not just Nitropia but indeed the real world would be a cleaner, shinier, happier place. (I’m afraid it’s quite inarguable that if the Americans and Europeans of 1908 could see their countries now, they would be both amazed at the achievements of science and engineering, and appalled at how shabby, filthy, ugly and dangerous their great cities had become. If you can’t imagine what this could possibly have to do with term transformation, I understand. But I will get there.)

So what is the problem? What could possibly be the problem?

In the above defense of term transformation, which I do hope is not a strawman—I really did throw in every argument I have heard or can think of, and if readers know of others they should feel free to append—there are a few subtle, but extremely fatal, holes.

First, let’s introduce the concept of financial hygiene. Financial hygiene is a lot like regular hygiene. If you are the only one who goes around licking doorknobs, you are unlikely to catch diseases from doorknobs. But if you find yourself licking doorknobs, it’s probably not an idea you (being a sensible person) came up with yourself. Which means that other people are probably doing it too. Sooo…

Financial hygiene is a criterion for prudent herd behavior. A strategy is financially hygienic if the strategy works just as well whether a whole herd adopts it, or just a maverick individual. As we’ll see, term transformation is to hygienic finance as a pile of oil-soaked rags is to a sanitary napkin. Term transformation is a great idea—so long as no one else is doing it.

We’ll start by going back to our magical risk-free tickets. What’s truly amazing is that term transformation is so dangerous that even when all tickets are risk-free, it can detonate instantly and without warning. Mind you, risk does not make the problem any better. In fact, it makes it much worse. But first things first.

With risk-free tickets, we no longer need the services of our balancer. Instead, ordinary ticketbuyers can transform their own terms. This does not make the problem worse, but it makes it easier to see.

In Nitro City, where all transactions are frictionless, why in the world would anyone hold cowries? If long-term tickets are risk-free, keeping your cowries “under the mattress”—even for a millisecond—is just throwing away cowries. Just hold the longest-term tickets, with the highest yield, until you are ready to spend your cowries. Then, instead of exchanging cowries for an inflatable penis, exchange your long-term tickets for cowries, and immediately exchange the cowries for the penis. With this strategy, you are always earning maximum yield.

Or will it? It’s not clear that the yield will be all that great. After all, when everyone is buying long-term tickets, how much can they yield? And let’s not even start on the consequences for the dragon ecology. Presumably the beasts are apex predators, after all.

But there is a worse problem. The problem is that when you buy a ticket of term T, you are signalling the market that you intend to exchange cowries now for cowries at now + T. This demand is naturally satisfied by a dragonslayer who commits to the opposite exchange.

If this is not really the transaction you intended to perform, you are sending a false signal. When you send a false signal, you should not be surprised to get an ugly result. When you send a false signal as part of an entire herd which is sending the same false signal, you should be surprised not to get an ugly result.

Our fallacy in designing this wonderful edifice of term transformation was to assume that it is possible for an unlimited number of parties to signal a market in one direction, without affecting the market price. In other words, we were committing the most common error in economics, by assuming an objective or absolute price. Of course there is no such thing. All prices are set by supply and demand, and if you add sellers to a market without adding buyers, the price will go down.

Let me slip briefly into the language of normal, Windowsological finance. To buy a ticket of term T is to lend for a period of term T. To sell a ticket of term T is to borrow for that period. And these same terms apply whether or not you are the original buyer or seller of the ticket.

So, when you buy a ticket of term T, but you really intend to be able to use your money at T/2, or T/6, or whenever the heck you feel like it or need to, you are assuming that you will be able to find another lender at T/2, or T/6, or whenever, who will acquire your loan. For example, an equivalent transaction is to write your own ticket at T/6, with a term of 5T/6, and sell it. You keep the original ticket, and use its proceeds at T to pay off the ticket you wrote. (Don’t try this at home, kids.)

If everyone who lends at term T really intends to lend at term T, then a few individuals whose plans turn out differently will have no problem in correcting their mistake. Peoples’ actions may differ from their plans, but they are likely to at least cluster symmetrically around them.

However, if every single Nitropian who lent at N for term T really intended to lend (for an extreme and unrealistic example) at T/6—when N + T/6 rolls around, the market for 5T/6 tickets will become um, well, kind of nasty. There will be quite a few sellers, and no buyers at all. In the Nitro City ticket market on N + T/6, for tickets maturing at T, the yield will be immense. And the price will be derisory.

This is the opposite of financial hygiene: financial contagion. Our herd of Nitropians all thought they could earn a dragonslayer yield on an orc-hunter schedule. Well, they did earn a dragonslayer yield—until the market opened at N + T/6. Then they saw a picture that looked rather like this. And they suddenly realized that they had no yield at all, but massive losses. And all this with 100% risk-free tickets.

The example of everyone selling at N + T/6 is slightly exaggerated. But it’s less exaggerated than you might think, for a couple of reasons. First, the whole point of a rainy-day fund is to have a stash of cowries for a rainy day. And a rainy day, by definition, rains on everyone. Second, when the curve starts to head for Antarctica, everyone who has those N + T tickets sees the light and wants out ASAP.

This is called a panic. Historically speaking, the association between financial panics and term transformation is about as solid as the association between smallpox and the smallpox virus. Although I suppose you do have to have the smallpox virus to get smallpox, whereas panics can in theory be caused by any widespread error in planning that mimics term transformation. For example, the sudden discovery of an asteroid that was about to hit Nitropia at N + T/2 would certainly have a rather dramatic effect on the yield curve. On the other hand, if there are any cases of financial panics in history that were not associated with term transformation, I am not aware of them. Perhaps readers will be so kind as to enlighten me.

Now let’s put the risk and the balancer back in, and see how it changes the game. Not for the better, as you can imagine.

If you scroll back up to our rosy explanation of how well term transformation works, you can see how the same error infects the case with risk and balancer. At every step of the process, we were assuming that another buyer can be found for the ticket, at the same pleasant prices.

Of course this is not so at all. The balancer just adds a layer of indirection to the same basic process of breakdown and panic. The result is an even more spectacular trainwreck. In a way it’s almost beautiful, unless of course you’re in it.

Using the familiar logic of financial hygiene, we can assume that in a market where term transformation is active, our one poor balancer will not be the only agent performing risk dispersal for the dragonslayer market. Rather, we need to consider the entire category of balancers in this space as the herd.

The panic can start in many ways, but the simplest is a wave of redemption in demand tickets. There is no way to predict this wave, because by replacing true tickets with demand tickets, the term-transforming balancers have essentially jammed a price signal. They have no idea what the future pattern of demand ticket redemption will look like, either within their own customer base or across Nitropia as a whole.

To redeem the demand tickets they have written, the balancers must in turn sell dragonslayer tickets. Of course this drives the dragonslayer yield up and the price down.

The result is that the balancers enter scalar insolvency. The total market price (in present cowries) of the tickets they own, plus their cowrie buffers, no longer exceeds the sum (in present cowries) of their current obligations. So the owners of their demand tickets will take a haircut. They strive to avoid this in the usual way, by withdrawing ASAP. The early bird gets the worm. Flap, bird, flap!

But to even calculate this haircut, to liquidate the balancers and pay their creditors, more dragonslayer tickets need to be sold. So the price drops even further. Eventually, someone realizes that ordinary, vanilla, untransformed, financially hygienic demand for cowries at N + T continues to exist and puts a floor under the market, and there are buyers again. If there is still a Nitropia to buy them in.

This nasty, nasty process is a feedback loop between the dragonslayer tickets, the institutions that own them, and the customers to whom they are obligated. It can be triggered by any unexpected pattern of selling, anywhere in the circle. For example, if dragonslayer attrition rates start going up from 3% to 5% to 7% to 15% to 20%—perhaps because demand for dragonslayer tickets has been so high that all kinds of completely unqualified warriors decide to rent a sword and descend into Yendor in search of dragons—the same thing will happen.

Next week, we’ll look at how unhygienic finance in the real world, and just how and when it got to be so popular. Clearly, what we have here is a case of regulatory failure. Is this another piece of Reagan-era napkin economics, coming back up to haunt us like a bad burrito? Did Hillary do it, with her cattle futures? Or is it one of those New Deal things? When will it not be all FDR, all the time, here at UR? If you don’t know, please take a ticket and come back next Thursday. (And do avoid the comments if you truly cherish your suspense, because I assume someone will spill the beans.)