Moldbug,

Interesting and worthwhile discussion. Thx.

Yes. The problem is that MT is such a large parameter that the difference between MT-on and MT-off states, the liquidity signal if you like, completely drowns out the solvency signal, and replaces it with meaningless noise.

That’s what MBS prices are now: meaningless noise. Except in the sense that they feed back into real estate prices, they are not an indication of solvency risk. The Hayekian price signal cannot be received, even by the most sensitive of antennas.

That’s why the Mellonist attempt (which seems defeated, for the moment – at least Napoleon got 100 days, “Hank the Hammer” had only two) to solve the problem via liquidation can’t work. The solvency signal will not reappear until you either pull the whole mortgage market down to its flat MT-off level, liquidating to M0, or pull it back up to MT-on with fresh FRNs.

Orthodox Austrians recommend the former, which is crazed. The latter is the only sensible alternative. And it’s what will happen, one way or another. So I’m just proposing that it be done wholeheartedly and for good, rather than in the usual bandaid fashion: pull the switch back to MT-on and then close out MT.

Moreover, when you use CB insurance to eliminate liquidity risk, you have a choice between formal liquidity insurance and wink-and-nod liquidity insurance. The former creates moral hazard. The latter, well…

You are absolutely right about the deposit pool. My view is that extremism in the pursuit of stability is no vice. Like Hazlitt, I would rather not split the difference between right and wrong.

Most bankers are not constitutionally prone to extremism, but fortunately there is another view of the issue. As I said earlier, you can model any MT banking system as a non-MT system which is balanced by government loans.

So our present deposit system is equivalent to one in which Y (the bank) stores its checking deposits as cash in the vault, and then borrows money from X (the Fed) to make its 30-year mortgage loans.

This makes the problem transparent: how, in a financial system heavily dependent on government loans, do you close out said loans and return to a free market, in which every loan has a private lender and a private borrower? The answer strikes me as simple: pull the whole structure back onto X’s balance sheet, mortgages and all.

This leaves a situation in which Y’s customers have tons of cash, Y is just a cash warehouse with no debt, and X has tons of future payments of uncertain default risk. X can auction those payments and see what the actual risk is. Then we are back to a free market. And a free market is a stable market. (This ignores the feedback effect on home prices, which might require further intervention.)

In real life, of course, banking is anything but simple. But it is sometimes good to have a simple model in one’s head, when dealing with big, complicated realities.

“I.e., solvency, which is of course the heart of banking.”

So I think we’ve in agreement on that fundamental point. MT is then a risk parameter or switch as you referred to it, overlaid on capital underpinning as part of a system of risks.

Actually, I was thinking of none of the above with respect to the pragmatism of converting to a non-MT system. I believe that the demand for immediately en-cashable deposit balances (money of 0 maturity if you will) is one of the strongest structural demand functions in banking, particularly as demanded by retail banking customers. If solvency is the heart of banking, these deposits are key to the central nervous system and, of course, to MT. The MT parameter positions such deposits as self-insuring, in the sense of actuarial probabilities of regular in and out flows, subject to capital and solvency adequacy, decent franchise value, and ongoing relationships with retail banking customers. Note that these funds do tend to be a stable source of funding for commercial banks with sound enterprise value, as opposed to wholesale sources of funding. By pragmatic I mean that deploying these funds in corresponding assets of 0 maturity seems unduly extreme to me, although I can see it would be purist in the sense of a non-MT system. And I mean pragmatic in the sense that non-MT is indeed an extreme point on the continuum of asset-liability liquidity risk in the current world where limits on various risk types tend to be non-zero as a matter of business course. But I think your point is simply that MT risk should be excluded from the risks that bankers assume with their capital.

I am sorry but this is nonsense. Maybe the financiers can believe this if it makes them feel better about themselves but most of the money did not go to the subprime borrower.

All the subprime borrower (most of whom bought at the peak of the market in 2006-07 btw) got was an unaffordable house they could not afford. The Wall St traders and execs got billions in cold hard cash bonuses.

Today the subprime borrowers are getting kicked out of their houses. The traders get to keep their bonuses. Now tell me who benefited from the boom times?

“Even in a non-MT system, equity as a source of funding must be invested in assets with some maturity and liquidity profile. Funding supporting a bad asset at a non-MT bank must be repaid by liquidating equity investments. This runs down equity and solvency. Depositors realize that they must stand in line for repayment of their deposits according to maturity, in case of bad assets. Obviously they can’t get their money out before maturity. There is no bank run in this sense, because they are locked in. But perceptions of solvency over the long term still matter, and even if they get their money back, such depositors may not fund new assets on a non-MT basis because of this risk.”

Indeed. Non-MT accounting is not a recipe for eliminating banks, or bankers! The banker’s answer to solvency risk is (a) prudent diversification, (b) excess capital, and (c) independent accounting. None of these requires a central bank or other liquidity insurer.

Basically, in a non-MT system with a fixed or near-fixed (eg, a physical gold standard) interest rates should be stable and market-determined, asset prices should not fluctuate systematically, and the price of future payments should reflect default risk and nothing else.

The key to thinking about non-MT banking is to realize that in the absence of a liquidity insurer, lenders have an incentive to actually match the maturity of their investments to *their own* liquidity needs. If you need an investment that matures in one year, you won’t buy a five-year bond. And since there is minimal demand for one-month, one-week, or one-day loans (there are few productive uses of capital at this duration) the ultra-short lending market basically disappears. If people need to hold cash, they just hold cash (giro banking).

At first this seems counterintuitive, because yield curves in a non-MT system will still slope upward. The rate on the five-year-bond is higher. But a rational investor will match correctly to avoid a liquidity crunch. The game theory is that if you are the *only one* mismatching your maturities, you can profit. But you won’t be the only one, so you shouldn’t.

So the liquidity risk migrates out to the edge of the system and disappears, instead of collecting in a central single point of failure as it does now. And bankers can focus on the problem that they understand: default risk.

Ie, solvency, which is of course the heart of banking. Basically, our present system tries to factor out liquidity risk, which is in Taleb’s fourth quadrant, ie intrinsically unamenable to modeling, by putting it all in one basket and having the government hold that basket. I can see how this seemed like a good idea in 1913. Or even in 1694. But it’s just wrong.

“A full non-MT system of course is not pragmatic…”

There are several ways to interpret this statement, which I’m afraid I’ll have to ask you to defend!

What I think you mean is that a *transition* to a full non-MT system is not pragmatic, and I think you mean “pragmatic” in the sense of “politically plausible.”

This, of course, is true, so you have an easy out. I am a blogger, not a banker, so I am interested in hypotheticals which are not politically plausible. Not everyone is, and I respect that.

If you mean that a non-MT banking system is unstable or otherwise unworkable, or even socially undesirable, some clarification of the point is needed.

If you mean that a transition from MT banking to non-MT banking is impractical, you are correct – but only with the strawman assumption that the transition is done in the worst possible way, ie reducing the money supply to M0.

The straightforward way to close out MT is for the liquidity insurer to buy the securities whose liquidity, and indirectly whose solvency, it has been insuring. This means the whole banking system, as described above. The Fed becomes the RTC. Nor does this imply eternal Bolshevism: what is bought can and should be sold.

Other band-aids, of course, can be found. And no doubt will be.

I just wanted to stress the fact that regulators need to also be extremely vigilant.

We all know that the final buyer of mortgages was getting tainted goods from predator mortgage agents in the field.

This system was all wrong. If the regulator is vigilant, and detects and corrects this problem in due time, we would not be having this discussion.

Improved mouse traps are well and necessary, but, keeping the guard up and being vigilant, can go a long way –remember that mice will always be trying to defeat the next mouse trap, and history tells us that they always do.

However, you shouldn’t fall into the trap of thinking that only rich people are capable of corruption. The problem is that if you look at where most of the money ended up, it didn’t end up in the pocket of some rich person. The problem with the system that got set up is that you were better off if you went to a bank and lied about how much money you made, how much your house was worth, and how likely you were to pay your money back, and most of the money ended up in the hands of people with bad credit.

“The key to the problem is that it’s mathematically impossible to distinguish a low price of S-MBS as caused by astronomical default risk, from a low price of S-MBS as caused by the astronomical interest rates of the MT-off state.”

Whether risk premiums for credit and liquidity are differentiable is not critical to the issue of what the logical risk hierarchy is as between solvency and liquidity risk. In fact, one can distinguish between pricing for liquidity and credit risks to some degree by comparing floating rate note spreads (no or low interest rate risk) for different maturities of the same credit.

“A non-MT bank is not dependent on rollovers at all. Its cash-flow pattern will be perfectly intact even if its clients redeem all liabilities at maturity and never, ever roll over. Thus, there is no feedback and no wave.”

Even in a non-MT system, equity as a source of funding must be invested in assets with some maturity and liquidity profile. Funding supporting a bad asset at a non-MT bank must be repaid by liquidating equity investments. This runs down equity and solvency. Depositors realize that they must stand in line for repayment of their deposits according to maturity, in case of bad assets. Obviously they can’t get their money out before maturity. There is no bank run in this sense, because they are locked in. But perceptions of solvency over the long term still matter, and even if they get their money back, such depositors may not fund new assets on a non-MT basis because of this risk. Non-MT hedging precludes conventional bank runs, but may still cause depositors to run away from the franchise once they get their money back. This is an issue for banks that are actually solvent who still want to grow. So liquidity is still an issue and solvency still drives liquidity.

I’m not disputing the fact that liquidity and MT are fundamental issues, or that MT affects pricing of assets comprehensively. They just aren’t the core issue, which is solvency. And MT doesn’t account for all liquidity risk, as noted above. Even in a non-MT system, liquidity of equity investment and funding liquidity in terms of ready access to funding for new assets (again, a function of ongoing solvency) remain important. The solvency issues that cause an interruption to funding midstream in an MT system would still be present when borrowers want to rollover their financing in a non-MT system (the requirement for borrowing doesn’t suddenly stop system wide when current contracts mature). And both maturity structure and pricing would adjust systematically for both borrowing and depositing in a non-MT system, with a severe but non-apocalyptic pricing and financing term adjustment (the idea of 30 year fixed pricing on a mortgage is a joke to begin with, but something reasonable like 5 years isn’t, and can be funded in normal risk environments). A full non-MT system of course is not pragmatic, but given the obvious role of short funding in the current unfolding disaster, look for Basel and future US regulatory guidelines on liquidity to be much more severe than in the past, although mostly in relation to wholesale funding maturity constraints. This associates with the now predicted demise of the stand-alone investment bank model, although I doubt you’d be comforted much to see their current wholesale short-funding reliance replaced by the retail deposit base of a parent commercial bank.

“But my point is that the case of actual solvency risk is binding with or without maturity transformation. The same perception will materialize by maturity at the latest. Maturity transformation tends to increase liquidity risk because the proof of solvency is delayed, and the risk of perception increases.”

I think you’re still missing the feedback loop at the heart of the problem. You’re not by any chance an electrical engineer? If so, think of it as a flipflop. Feedback creates memory. The circuit has multiple equilibrium states.

The price of our questionable security – the S-MBS – is set, like all things, by supply and demand. The supply of S-MBS doesn’t change much. But the demand, because of the Diamond-Dybvig dual equilibrium, has two stable states: MT-on and MT-off.

When MT flips off, it reduces the S-MBS as an asset class to a price level which, compared to its relatives the R-MBS and T-MBS, indicates a preposterous default risk. In fact, it is just a preposterous interest rate – corresponding roughly to whatever the price of any MBS would be in a world without MT.

MT supplies a very, very large percentage of the world’s demand for high-maturity assets. This is why, if you eliminated MT across the entire financial system, but preserved the existing formal money supply (ie, M0), you would see preposterous interest rates for long-term money. Or to it another way, the price of *even risk-free* 2038 dollars in 2008 would be preposterously low.

When MT shuts down even in one asset class, basically what you see is that the demand for 30-year loans has to be fulfilled by 30-year lenders. How many people want to buy a 30-year CD? Does this product even exist?

The key to the problem is that it’s mathematically impossible to distinguish a low price of S-MBS as caused by astronomical default risk, from a low price of S-MBS as caused by the astronomical interest rates of the MT-off state (which, in our example, applies only in the case of the market for S-mortgages).

Thus, for our real-world bad mortgages, the present market price does not indicate the default risk of the security. It simply indicates the extent to which holders of these mortgages have accepted the MT-off price. Which, obviously, sucks. (The situation is exacerbated, of course, by the feedback loop between new mortgage availability and real-estate prices – but let’s not go there.)

If MT is off to begin with, and 30-year interest rates are actually set by the interaction of supply and demand for 30-year money – a loan requires a 30-year lender and a 30-year borrower – none of this can happen. There is no phase change that can descend suddenly and unpredictably without warning.

Of course, momentum can drive any asset price up and/or down. You can have a default-risk panic, or any other class of panic, in any asset. But the market tends to converge on a single equilibrium price. There are no multiple equilibria. Diamond and Dybvig are not in the building.

Another way to see the essential instability of MT is to see the way a maturity-transforming bank depends on the rollover behavior of depositors, which of course it does not control. If the bank is solvent, it can meet a wave of withdrawals (failures to roll over) by selling assets – but this wave of selling feeds back into the market for those assets, causing the effects described above.

A non-MT bank is not dependent on rollovers at all. Its cash-flow pattern will be perfectly intact even if its clients redeem all liabilities at maturity and never, ever roll over. Thus, there is no feedback and no wave.

And – perhaps most important – if you knew that there was no lender of last resort, you would not lend your own money to MT banks. Because who needs bank runs? Thus, it is the old story of the solution creating the problem.

Without MT, the concept of “liquidity” is not particularly meaningful. “Liquidity,” in the sense commonly found today, represents the presence of MT-created demand in a loan market.

When the MT engine shuts off, markets shut down. Those who hold MT-disabled securities on their books at the MT-on price do not, in general, care to sell in an MT-off market. So there are no transactions, or almost none. And when there are, the prices seem to have no relationship to the default risk of the loan.

Basically, the hellacious interest rates we’d see, if the whole financial system switched to MT-off, the quantity of money was fixed at M0, and the Fed did not fill in the gap with lending of its own, are why any attempt to turn MT off and keep it off, without producing dreadful debt deflation, requires an equally unprecedented level of monetization. The payoff, however, is a stable financial system with a fixed, closed-loop money supply and a complete absence of central planning.