Arnold Kling  

Is Repo Lending Inherently Unstable?

Microcredit versus Immigration... Schools of Macro...

The blogger at Synthetic Assets (Carolyn Sissoko, I assume) writes,

Liquid assets are supposedly "safe" - but for the problem that liquidity itself is inherently ephemeral. How precisely do the regulators imagine that collateral posted by a systemically important financial institution (SIFI) is going to protect the lenders? If the SIFI goes down, there is, in the absence of central bank intervention, a fire sale. And if they're counting on central bank intervention to make it possible for collateral to function to protect the borrowers from losses (e.g. via a PDCF or TSLF), why not just rely on traditional central bank lending to banks in a crisis?

If this criticism is valid, as I believe it is, then the whole structure of our financial system is wrong.

Comments and Sharing

COMMENTS (7 to date)
JeffM writes:

Many years ago, I wrote the liquidity policy for the bank where I then worked. As I was writing it, I realized and incorporated into the policy that "liquidity" refers to two quite different although related concepts.

One refers to the attribute that, with probability close to 1, an asset can be turned into money at fair market value virtually instantaneously with very low transaction costs. In this sense, a stock listed on a major exchange is "liquid" although its fair market value may be quite unpredictable. Notice that liquidity in this sense depends in part on the existence of a very efficient market. It also implies an institutionally and exogenously defined "market period."

The other concept refers to the attribute that, with probability close to 1, an asset will turn into a known amount of money within a known time at virtually no cost. In this sense, the contractual payments on a portfolio of 30-year, self-amortizing, non-delinquent mortgages that are all 6 months or less from maturity are "liquid" within at most 31 days. Notice that liquidity in this sense does not depend on the continued efficient operation of a specific market; indeed there may not be a market on which any ostensible "fair market value" could be realized before receiving that value through the ordinary course of business. Such liquidity varies with respect to arbitrarily defined periods. In this sense, the aggregate of what is liquid within a month cannot be less than what is liquid within a week and will normally be greater.

I do not deny that the two concepts are related. Many assets are liquid in the first sense only so long as they are liquid in the second sense. I only insist that the concepts are not identical. But I have found, at least among regulators and lawyers, almost universal lack of understanding of the distinction. They think liquidity is one thing.

Moreover, the need for liquidity cannot be assessed intelligently without understanding what is the effective (as opposed to the legal) maturity structure of liabilities. Failure to understand this leads to excessive stress on leverage as a measure of risk. Of course, leverage is easy to explain to the press and politicians and judges.

As you can see, on these topics, I have strong opinions, which is probably why no one pays attention to them.

JeffM writes:

This is a follow-up to the above comment now that I have read with more care the blog entry from which the quote that triggered this thread was extracted.

I have several doubts as to the validity of that entire blog entry. For example, it may be true that a loan from Bank A to Bank B secured by a loan previously made by Bank B is contractionary for the economy as a whole whereas an unsecured loan from Bank A to Bank B is expansionary, but it is not obviously true. Keynes may have thought it true, but that does not make it true. I too can quote authority: "bye-gones are always bye-gones." The obvious effects of securing an interbank loan with a loan outside the banking system are: (1) to move reserves to a bank that perceives opportunities to lend from a bank that does not, and (2) to potentially change which bank may be funding the loan outside the banking system. I admit, however, that I have not pondered what might be downstream effects so perhaps the blog entry is correct, or perhaps the blog entry does not quite say what was intended.

In any case, the proposition that the collateral securing Bank A's loan to Bank B necessarily loses value if Bank B fails is absurd on its face. Let's say the collateral is a T-bill. As we saw in 2008, the T-bill may very well appreciate in value if a Lehman fails.

MG writes:

If you make the collateral a highly liquid security trading within an arbitrageable class of similar securities, you subject the collateral to a valuation haircut that depends on its historical volatility (which includes a few fire sale crises), and you insist on daily mark to market, you are likely to be OK. I think I just described a well-functioning repo market. And in the case of an apocalyptic fire sale, what is so bad about having the central bank intervene by simply buying liquid collateral at a fire sale discount? That seems like a reasonable form of intervention, and it is hardly a bailout of the banking system.

JeffM writes:

I do not disagree with MG in any major sense.

However, my point is that if the lending bank takes short-term paper of good quality as collateral for its interbank lending, whether there is a firesale or not for such collateral is likely to be irrelevant to the lending bank. It just waits for maturity and takes no loss. If the lending bank is sound and nevertheless suffers a run, the purpose of a central bank is to act as a lender of last resort against good collateral. If we are going to class all loans of last resort from the central bank as inappropriate "bailouts," we may as well dispense with a central bank.

The problem in 2008 was not just that much of the layering of credit within the financial sector was normally unsecured, it was that huge debtors had too little good paper to use as collateral when collateral was required.

Collateralization does not do everything to eliminate systemic risk arising from the failure of very large financial institutions. For one thing, the normal operations of a banking system generate "involuntary creditors," and modern US banking laws somewhat exacerbate the extent of involuntary interbank credit. But sensible collateralization of intentional interbank lending would minimize the risk of chain reactions of failure and would minimize the costs of what failures do occur.

Where MG and I may disagree is that, in my opinion, relying on the marketability sense of liquidity does not result in effective "circuit breakers" when panic seizes a market. Self-liquidating paper, the real bills sense of liquidity, is more effective in a systemic crisis.

Shayne Cook writes:

Thank you both (JeffM and MG) for informative comments.

To MG specifically, you state:
"And in the case of an apocalyptic fire sale, what is so bad about having the central bank intervene by simply buying liquid collateral at a fire sale discount?"

I concur, but that isn't what the Fed did during 2008 and 2009. Instead, it purchased "toxic assets" at full face value, not the "fire-sale" prices offered by the market. I considered that an error on the part of the Fed then, and still do.

In the context of the Synthetic Assets blog post, such Fed actions, or any other similar central bank actions, reverse some roles. Instead of the central bank controlling the financial system, the financial system controls the central bank. I'm convinced that is a sub-optimal state of affairs.

MG writes:


I agree with you, but notice that Arnold's provocative question seems to postulate some existential inability to run almost any kind of a repo market without tainted intervention. I don't think a well functionng repo market is inherently unworkable or inevitavly likely to lead to moral hazard. Food for thought...In JPM's letter to shareholders the year after 2008, Jamie Dimond stated that there was a misconception regarding what the Fed did in the context of JPM's BSC purchase. He statted that many thought the Fed took in the toxic stuff and left JPM with the better assets. He said it was the opposite, that what the Fed took in was creditworthy assets that would have had to be needlessly liquidated at a fire sale price. I wonder if most of that was collateral? What JPM bought (cheap, for sure) was the more toxic stuff, which may been non-repoable, and JPM was to be running big risks while sitting on this risk for a while.

Shayne Cook writes:


Thank you for the explanation.

"... Arnold's provocative question seems to postulate some existential inability to run almost any kind of a repo market without tainted intervention."

If that is what Arnold (or the Synthetic Assets blog author) is postulating, I (like you) wouldn't agree. I guess I interpreted their point as being just that it matters significantly how and when the central bank intervenes in such circumstances. That also seems to be consistent with and follow JeffM's explanation of the liquidity characteristics of different assets.

On JPM/Jamie Dimon ...
If I were forced to invest in one of the major U.S. banks (and I would have to be under duress to do so), it would be in JPM due to James Dimon. He's a very sharp fellow. But I have to doubt whether either Jamie Dimon or the Fed had any concept of which Bear-Sterns assets were "creditworthy" and which were "toxic", either in March 2008 or even into mid/late-2009. At best, either the Fed or Jamie or both were "over-paying" for most of those assets, relative to what the market pricing was/is.

Comments for this entry have been closed
Return to top