Believe those who are seeking the truth. Doubt those who find it. Andre Gide

Friday, September 13, 2013

Confessions of a rehypothecating fractional reserve banker

I've been trying to wrap my mind around the new 4-letter-word in finance: rehypothecation. I found out that it seems to be related to an old 4-letter-word: fractional reserve banking. I want to argue that these phrases do not deserve to be viewed as cuss words. At least, that's what I think so far. Let me explain why.

An acquaintance approaches you asking for a money loan of $100. He sheepishly offers his vehicle as collateral for the loan. The market value of the vehicle just happens to be $100. (Your acquaintance would prefer not to sell his vehicle because he only needs the cash on a short-term basis, say, one month). You both agree on a one-month loan at an (annualized) interest rate of 5%.

The technical term for this is hypothecation--i.e., when a borrower pledges an asset as collateral to secure a debt. The borrower retains ownership of the asset, but the asset is "hypothetically" under the control of the creditor, who is granted permission to take possession of the asset if the borrower defaults.

In the example above, your loan is 100% secured by your acquaintance's vehicle. But let's imagine instead that the vehicle is only worth $10. After talking with some friends who know your acquaintance a bit better than you do, you decide to go ahead with the $100 loan, secured by the $10 vehicle.

What a nice guy you are. But not everyone thinks so. There are people who rail against your recklessness. Some even call you a fractional reserve banker (I talk a bit about fractional reserve banking here).

A fractional reserve banker? Yes. Let me relabel you a bank and your acquaintance a depositor. The depositor is in possession of $10 in cash (not a vehicle) and he goes to the bank to borrow money (not necessarily cash). The depositor opens an account with the bank and deposits his $10 of cash. The loans officer credits the depositor's account with $90 of electronic digits. (In the old days, the $90 would have taken the form of banknotes and the $10 deposit would have been in the form of specie.) The depositor now has $100 in money to play with (he can buy stuff using his debit card).

Some observations. First, banks do not lend cash. Banks create money. More precisely, they transform illiquid promises (the depositor's IOU) into liquid payment instruments (bank liabilities). Second, fractional reserve banking is absolutely critical to this process. If the bank was restricted to lending only up to the value of its cash deposits, there would be no point to banking (apart from serving as secure repositories). Insisting on a 100% reserve requirement is like insisting that you are not permitted to lend your acquaintance more than the value of his collateral. A restriction like this would certainly make the loan safe. But are such restrictions efficient? (And if you've ever made an unsecured loan to anyone, you have practiced the absolute worst form of fractional reserve banking.)

Well, alright, but what does any of this have to do with rehypothecation? Rehypothecation occurs when a creditor uses the borrower's pledged asset for his own use (e.g., selling it, or using it as collateral for his own borrowing). Rehypothecation plays a big role in the so-called shadow banking sector. The practice is often likened to fractional-reserve banking and is widely blamed for the failure of Lehman Brothers and MF Global; see here.

But just like fractional reserve banking, rehypothecation has its upside. To see this, let  me return to my original example of you and your acquaintance.

Returning to that story, recall that the agreement is to lend your $100 cash to your acquaintance for one month at 5% interest, collateralized by his $100 car. But you know what? It's not entirely clear that you won't be needing some of that cash yourself over the month. You don't think you will, but you might. Hmm, what to do if you do need the cash?

Just before signing the loan agreement with your acquaintance, you come up with this idea. You explain the circumstances to your acquaintance and ask him whether he would be willing to let you use his car as collateral for your own loan, should you find yourself strapped for cash. You acquaintance says sure, but what's in it for me? You offer to lower the interest rate on his loan to 2%. Agreed. (For an example, consider section IV-F in this brokerage account agreement issued by the discount retail broker Scottrade.)

Notice something interesting here. Suppose that you trust your acquaintance fully to repay the loan. Then, you might say, no collateral is needed to support repayment. I want to suggest, however, that the creditor may nevertheless ask for collateral and an associated rehypothecation right. The purpose of the collateral in this case is not to support repayment of debt between broker and client (you and your acquaintance), but to support the broker's (your) promise-making ability in some future transaction with some less trusting third party. The rehypothecation right essentially allows the broker to use deposited collateral as "money on demand."

To see how rehypothecation relates to fractional reserve banking, imagine that you find yourself borrowing $100 mid-month from some third party using your acquaintance's vehicle as collateral. There is at that point $200 in outstanding debt obligations that are supported by only $100 in assets. If rehypothecation rights are granted to the third party (in exchange for lower financing costs) and if the third party in turn uses the same collateral to secure a $100 loan from some fourth party, then we have $400 in debt supported by $100 in assets. And so on.

At each stage in this process, rights over the collateral are passed on to the last creditor in the chain. All previous debts are rendered unsecured; which is to say, the debts are supported by the debtors' desire to maintain their reputational capital. Creditors become more trusting. Is this a bad thing?

What can go wrong, of course, should be obvious: some event happens in which a debtor is either unwilling or unable to fulfil a promise. Those creditors that are secured will emerge relatively unscathed. But unsecured creditors will pay the price. None of this has anything to do with fractional reserve banking or rehypothecation, per se. It is the nature of unsecured credit; that is, credit supported by trust (in the willingness and ability of debtors to make good on their promises).

A credit crisis is, as the Italians used to say, un mancamento della credenza; literally, a suspension in the general belief that any promises will be kept. Credit (derived from credere, or to believe) plays an important role in financial markets and in the payment system. Legislation that restricts or prohibits unsecured lending would surely make financial markets safer. But at what price? There are no financial crises in a society ruled by financial autarky, in particular.


  1. "If the bank was restricted to lending only up to the value of its cash deposits, there would be no point to banking (apart from serving as secure repositories)"

    It could still intermediate by sourcing funds and then lending onwards.

    1. Sure, but so could a regular financial market, where savers meet borrowers. To the extent a third party might facilitate this process, I would label such third party a "broker," and not a banker.

    2. An agency that transforms illiquid assets into liquid payment instruments.

  2. You are on the right path, keep going. Include real world processes, like commission to handling agents (bankers), bankrupcy procedings inside fractional reserve banking (FRB), processing clerings of colateralization (rehypothecation).

    Your 'unsecured' credit is not credit anymore since previous creditor got funds to cover as if he is not creditor anymore, the problem is that such process takes away funds from original $100 by handlers of the process in the form of comission/ bonus even tough face value stays the same-$100.
    Bankrupcy is supposed to remove such credit issuing fees from the face value which is possible to do only trough FRB without having any real world economic effect. It is a proverbial "free lunch" that neoliberals and goldbugs repetedly claim it doesnt exist.
    FRB provides for "free lunch".

    Credit creation is money creation (not value creation, that comes later), paying off credit is money destruction. In bankrupcy procedengs, a bankrupcy judge orders wiping off money created by FRB without ever needed to be returned and cleared from the balance sheet. The wiping out the balance clears money given to the debtor and as comission to handlers.
    In a bankrupcy all creditors and debtors, secured and unsecured, clear their balance sheet without anyone loosing within FRB.
    Bankrupcy allows for keeping money created by issuing credit into economy instead of destroying it.

    The problem is with MERS (Mortgage Electronic Recording System), a corporation designated by banks to record all such transactions between banks. MERS was supposed to keep records of all unsecured transactions which would be secured at the time of the transaction. MERS is not compatible with US court system. MERS was avoiding local administration fees that requiers recording all asset mortgaging (hypotecation) with the government.
    Since securing hypothecation waas not recorded with government, a judge can not clear 4 times unsecured debt, only the first recording. Now there is no knowing who is holding secured and who is holding unsecured debt and in what order, in order to be cleared in a bankrupcy proceedings.

    FRB allows for avoiding destruction of money which happens with repayment of such credit. Inflation makes repayment of all credits easier, reduce inflation, there will be more bankrupcy need.
    This crisis was initiated by lack of inflation to makes debt burden smaller.

  3. The unsecured lending question is totally separate from the full versus fractional reserve argument. That is, advocates of both fractional and full reserve believe in leaving it lenders to decide for themselves whether they want to demand collateral in exchange for a loan.

    1. Ralph:

      Maybe. Let me see if I understand you.

      Let's consider the example of 100% reserve banking. A depositor deposits $100 cash in a repository. The repository is not permitted to issue liabilities made demandable for that cash in excess of $100. However, the repository is permitted to issue separate liabilities that are secured in various degrees (possibly not at all) by other assets owned by the repository. (Of course, because this latter set of liabilities is not convertible into cash on demand, they are likely to be less liquid than demandable liabilities).

      On the other hand, I am fond of my perspective too; namely, that less than fully-secured lending (fractionally collateralized) resembles fractional-reserve banking/lending.

      I will continue to think about it. Thanks.

    2. Secured or not it is not about where funds for loan come from, while fractional vs. 100% reserve is only about where funds comew from. Secured or not is about risk.
      As Ralph says, it is a totaly separate issue.

      100% reserve banking used to exist in all communist countries but then inflation had to be created by state money printing, which was btw, much more effective in developing economies. Dirigist creation of money was very effective once developement path was known from capitalist examples.

      Capitalism/ free market without money creation is a failing perspective.
      Taking a surplus and not returning it to the economy is self destructive.Fractional reserve banking allows for capital accumulation and making sure all production is sold. So employees get their full value for work while capitalist get capital accumulation at the same time. Without credit money creation that would not be possible in the long run. Somebody has to print money at everhigher rate in order for capitalism to give any inch of prosperity.

      100% is communist country operation and fractional reserve banking is capitalist system. Communist country would have state print money and distribute it as needed, while banks print money and look for viable returns.

  4. I often deposit money at my bank. My employer does it roughly twice a month. I also borrow money from my bank. I do it when I use my credit card rather than my debit card for making a purchase. I have never borrowed 10 times the amount of a deposit. My credit limit on my credit card is not 10 times the amount of my deposit balance.

    I think the standard perspective is better. Depositors make deposits and the bank lends an amount equal to the deposit less any reserve it chooses or is compelled to hold. For example, a depositor deposits $100, which is a loan to the bank. The bank, holds 10% as a reserve and then makes a $90 dollar loan.

    In a banking system many deposits are not of currency, but rather checks. And loans are not made by handing out currency to the borrower, but rather giving them a check or else a balance in a checking account. When the deposited check (or electronic payment) clears, the bank ends up with a credit in its account at the Fed. Though banks keep most reserves in the form of currency, that is vault cash.

    An individual bank that operated on the basis you describe, accepting a 10% deposit of currency and lending $100 would very rapidly be reserve deficient because the borrower would spend the money. Those receiving the money would deposit in their own banks. Once the checks cleared, the bank would have to cover the payment with $100 of reserves. And they only had $10.

    A bank that operated in the traditional manner would receive a $100 deposit, lend $90, the borrows spend the money and when the checks clear, the bank has $10 of reserves left, a $90 dollar loan, and owes the depositor $100. Now, the bank might require collateral for the loan, but that isn't the deposit. And banks sometimes do require borrowers, especially firms, to keep some fraction of a loan as a deposit. But it isn't really collateral. In effect, it is a reduction in the net amount of the loan.

    Anyway, what each bank needs to do is find customers willing to hold deposits in the bank. These can be payment media, like checkable deposits or banknotes, or they could be savings accounts or CDs. And that, along with the owners' capital investment, is now much lending, securities and loans, the bank can undertake. That the bank also needs some cash to operate, is really no more significant than the need to have a building. But regulation has created a fixed ratio between reserves and transactions accounts.

    As for rehypthecation, I thought the issue was not so much making a loan with less than 100% collateral. It is rather than someone owns a T-bill, and borrows against it. There is a haircut, and so the loan is less than the value of the T-bill. But then, the lender sells the commercial paper (or keeps it.) Then the holder of the commercial paper gets a loan and uses the commercial paper as collateral. It is good collateral, because it is secured by the T-bill. Then the lender now has a loan, commercial paper, secured by other commercial paper, secured by a T-bill. And then, the holder of that commercial paper gets a loan against it. It is good because it is secured by commercial paper that is secured by a T-bill. The same T-bill is securing commercial paper worth more than the T-bill.

    But the firm that owned the T-bill didn't get a loan for more than the value of the T-bill.

    1. Bill, you may think that the standard perspective is "better," but I think it's wrong!

      An individual bank that operated on the basis you describe, accepting a 10% deposit of currency and lending $100 would very rapidly be reserve deficient because the borrower would spend the money. Those receiving the money would deposit in their own banks. Once the checks cleared, the bank would have to cover the payment with $100 of reserves. And they only had $10.

      Think of the bank I describe as the consolidated banking sector. When the borrower makes a purchase, he debits his bank account and credits the bank account of the merchant. Reserves play no role in this story.

      As for rehypothecation, what you describe seems to correspond to what I described, no?

  5. One important difference is that in fractional reserve banking, the depositors are compensated for the use of their money. Where fractional reserve banking gets into trouble is through excessive leverage.

    In rehypothecation the asset is often being held in custody and is contractually pledged for one purpose, while being pledged again and again, even downstream. And no compensation and often no disclosure is given to the asset holder so that title becomes obscure.

    1. Jesse:

      Did you read the Scottrade Broker Agreement form that I linked to above? The idea that there is "no disclosure" or that the parties involved in these transactions do not fully understand what is going on seems highly unlikely to me.

      Of course, when something goes wrong, people like to cry "foul." But, you know how it is.

  6. David:

    A $100 loan secured by only $10 of collateral is not a loan. It is a gift of $90 to the "borrower". If a sane banker lends $100 to a borrower who offers only $10 of collateral, it is only because the banker believes that the borrower's credit is good enough to be equivalent to a $90 lien on real collateral. In effect, every $100 loan is secured by $100 worth of collateral, once we recognize that a person's credit and character can serve as substitutes for actual property.

    I'm glad you recognize that a bank is not limited to lending up to the value of its cash deposits. For example, American colonists used to take the deed to their $100 farm to the local land bank and "deposit" the deed in the bank. The bank would print $50 worth of notes and lend them to the colonist. Those banks hardly ever had any cash actually deposited with them, but they lent huge amounts.

    1. Mike: I am in 100% agreement with what you said in your first paragraph. Indeed, it is what I teach: all debt is "secured" in one manner or another. The terminology is such, however, that "secured" only means "secured by some tangible property" to most people. And if they take this view, they are misled into thinking that the whole credit system is built on a house of cards.

      I like the example you report in your second paragraph. This is precisely how I understand banking. Bill Woolsey, would you care to comment?

  7. David,
    This was interesting. Perhaps you could expand it to discuss what, exactly, is a "collateral shortage"?

  8. Perhaps I should be more specific. There is a process of hypothecation of illiquid assets. Financial intermediaries carry it out. This process matches liquidity demand and supply. "Excess demand for safe assets", or a "collateral shortage" implies there is something wrong with this financial intermediation. What is that something? Does it explain all asset price movements, or does it rely on some assumption of market segmentation? What accounts for the persistence of this market failure?

    1. Hi Diego. Thanks.

      A "collateral shortage." I did write a bit about this here:

      But I can't really remember what I wrote, so let me try to answer separately here.

      A fundamental friction in financial markets is a lack of commitment (the inability or unwillingness of individuals to honor their promises).

      Exchange media (including collateral) are objects that enhance commitment and, therefore, are valued not only for any intrinsic properties (like their ability to generate income), but also for the ability to enhance intertemporal trade.

      Good exchange media, collateral objects in particular, are generally scarce. The private sector goes to great lengths to create good collateral (tranching of assets, for example). But as we saw in the last crisis, what we thought was good collateral turned out not to be in the repo market.

      I do not like the term "excess demand for safe assets." A variety of factors can influence the desired portfolio composition of agents. In uncertain times, there may be an enhanced demand for good collateral assets. And the supply of good collateral may be low. This will be reflected in the price of good collateral (yields on such objects will be very low).

      What is "wrong" with financial intermediation is the lack of trust (credit). But that's not something one can fix by waving a wand. There may, in this case, be some role for a good government (oxymoron?) to supply assets (U.S. treasuries, for example).

      Anyway, didn't mean to blab for so long, but hopeful this gives you some idea of where my thoughts are current at on the subject.

  9. David,
    Implicit in your comments is a failure of safe collateral "technology" (i.e. tranching). However, prior to the advent of that technology, safe asset collateral mostly came in the form of illiquid bank loan portfolios. So we need three explanations:

    1) Is the failure of shadow bank safe collateral supply permanent; and if so, what can the government do about it without creating distortions?

    2) Even if it is permanent, why can't the system function with traditional bank collateral?

    3) Are risk asset price movements since 2009 consistent with explanations of a persistent safe collateral shortage?

    1. Diego, excellent questions.

      Before I answer, be clear that I am only speculating that an apparent "shortage" of good collateral is playing some role in recent events. There are likely many other forces at work. Also, I am speaking from the perspective of what I know can happen in theory.

      1) I doubt that it is permanent. Trust, once destroyed, can be rebuilt, albeit, slowly. A relatively trustworthy government can issue substitutes, like US treasuries, without creating major distortions.

      2) Bank capital is not sufficient, and has not been in history (there were bank panics prior, no?). In any case, the missing collateral has probably hit the shadow banking sector hardest. Money managers with large deposits would rather invest in repo if they can get good collateral -- FDIC does not insure large deposits.

      3) I don't know. One thing to keep in mind though, I can generate in my models a "flight to safety" even in the absence of risk. E.g., suppose you know with 100% certainty that MBS is crap -- there is no risk there. But you will still want to substitute out of MBS into treasuries, driving down the latter's yields.

  10. David,

    Paul Krugman wrote a blog post a couple of days ago entitled,

    "Slackers at the Fed"

    Your thoughts

    1. I tend to agree with him.

      But I can see why the very large and growing balance sheet poses a concern. Central bankers need to be concerned about the inflation risks. I just think these risks are overstated by some at the moment.

    2. Well, I'm not as sure as he seems to be about the potential benefits of "easier" monetary policy. But I do think various fiscal and regulatory reforms might get things moving more briskly.

  11. David,

    I am going to bring my criticism from the other side of the spectrum. I actually think that you don't take the argument far enough.

    With rehypothecation it is important to have the institutional details. Consider this in the context of a repurchase agreement. Suppose that the two of us enter into a repurchase agreement. I give you a $100 bond and I agree to repurchase it from you later. Before it is time for repayment, you realize that you could use cash now. As a result, you rehypothecate that $100 bond. Now suppose that I default. When I default you keep the $100 bond. The repurchase agreement that you entered into with a third-party is not harmed by this transaction. Even in the event that you have to default, the third party will receive the $100 bond. Thus, even in the event that we BOTH default on our obligation, the third party is unharmed by the transaction.

    Okay, but let's change this around. Suppose that you default with the third party. The third party now gets the bond. I show up to repurchase the bond from you and you say that you don't have it. Ah ha! This must be the disaster everyone is speaking of. Except, of course, that it is not. A failure to provide the bond when I show up to repurchase it is just that, a "failure." It is not a default. In fact, you are not legally allowed to "default" on your end of the repurchase agreement. You must provide me with a sufficiently similar asset to the bond that I sold to you. If you fail to do so, you have to pay me a penalty until you do.

    So where is the harm in this? Critics make two arguments that I think are actually missing the point:

    1. Some argue that in the midst of a financial crisis, it is hard to buy "safe" assets of the kind that are needed to resolve a repurchase agreement. In other words, in the second scenario above you would have a problem finding a similar bond to what I sold you. As a result, we would expect a large number of repo failures. I suppose if the third party above was the one to fail, then it would result in two failures: the third party failing to provide the bond to you and you failing to provide the bond to me. But each of us would be compensated for these failures. And it is not clear how the failures would have broader macroeconomic consequences. In fact, the failures are the result of macroeconomic factors, not the cause.

    2. The second criticism involves the overvaluation of the underlying assets. For example, suppose in the scenario above that the bond I gave you is really worth $90 rather than $100. We don't have to assume that this is fraudulent. Perhaps we both think its worth $100, but it's really only worth $90. Now I sell you the asset with the promise to repurchase it later. You do the same with a third party. The third party, thinks that they will be able to potentially rehypothecate too, if necessary. However, before they can do so, everybody learns that the asset is really only worth $90. You and I rationally default. The third party is now stuck with the $90 asset approximately 10% less wealth. Now some argue that this creates a reduction in aggregate demand because wealth declined. Others argue that this reduces aggregate demand because the value of transaction assets have declined. But it doesn't matter what you call it. In this scenario, the problem was that everybody believed that the assets were worth more than what was actually the assets true value. Thus, even if these critics are correct about the mechanism and the business cycle dynamics, it is NOT rehypothecation that was the cause of the problem, but rather the mistaken asset valuation.

    It is hard to argue that rehypothecation is bad.

  12. David,
    I was struck by this comment: "Money managers with large deposits would rather invest in repo if they can get good collateral"

    "Rather invest" is a function of price. Prices adjust. So the persistence of "excess demand" for private safe collateral implies some sort of discontinuity in preferences. That is, agents favor repo over deposits to such an extent as to render deposits unprofitable at prices they require to switch.

    On your last point, if you know with certainty MBS are crap, the conditional probability of high yield being crap is probably 100%. It is difficult to imagine "flight to safety" scenarios in which the "certainty" involved also creates room for a massive risk asset boom, as we have seen since '09.

    1. Diego:

      The statement I made came from something Gary Gorton once explained to me.

      Consider a firm that has $500M cash on hand, that it may have to spend in a short while, but in the mean time wants to keep safe with ready access. I asked Gary why the firm would not just deposit the cash in a bank. Answer: it is not safe (deposit insurance does not cover). In the event that the bank gets into trouble, the money could be tied up for a long time, which defeats the purpose of having it readily available. I asked whether a large deposit might not be put in a segregated account with senior claim. Evidently, this might violate other debt covenants that the bank has. And so, depositing the money with a dealer overnight collateralized by Treasury or MBS turned out to be preferable. It's not just an issue of price, as you say (although price certainly factors into it, via the haircut, etc.). It is an issue of getting "bank like" services outside the cumbersome realm of chartered banking.

    2. OK this answers a question that's been nagging me all week. Why aren't reserves a perfect substitute for treasuries that yield zero? If the story above is correct, then the "shortage" of treasures seems to be a bank/finance regulation problem, not a monetary policy problem.

  13. David,
    "In the event the bank gets into trouble" is the operative phrase.

    So, 5 yrs after the crises, corp. Treasurers don't trust banks. But they do trust equities and high yield. This has something to do with the asymmetry of safe, s.t. asset returns. That is, they can get you fired, but they'll never get you promoted.

    I'm with you so far.

    Here's where I diverge. What is the implication of "excess demand for safe collateral"? Presumably, it increases the risk premium of non-safe collateral. This, in turn, raises the cost of capital (real interest rates) to the risk-bearing part of the economy, slowing investment and spending. I argue this is a huge leap. First, risk appetite is apparently quite strong out there, at least for large-institution debt and equity, as well as for duration risk. Second, there are a number of more important reasons (agency, market power) for why some credit carries higher risk premia than it should; safe asset demand seems small in comparison. I'm thinking, for instance, credit card rates.

    In the end, the "problem" we're trying to solve is a dearth of investment, as Stephen Williamson points out. Is "safe asset demand" holding back investment in a minor or major way? I don't think anyone has the answer, but my guess is "quite minor".

    1. Diego,

      I have written down models where a recession is caused by a sudden downward revision in the forecasted return to capital spending. The effect is to cause a portfolio substitution, away from capital and into government securities. The price level drops.

      If the expectation shock is rational, then an intervention can at best mitigate some of the worst consequences of the shock (e.g., in the event that debt is nominal). But as you suggest, the big question has to do with what has caused the pessimistic outlook? An increase in the supply of government debt is not going to solve that problem.

      Alternatively, if the expectation shock is "irrational," then, not surprisingly, an intervention (guided by a wise and benevolent authority, of course) can solve a lot of problems.

      I presume you are closer to the former view. And in that case, the effect of scarce collateral problem is of second-order importance. But as you say, we cannot be sure as of yet. Thanks.

  14. David,
    Thanks for clearing that up. Your two explanation of low return expectations make sense, and choosing between them in a rigorous way seems difficult.

    One of the things I appreciate about your blog is that you are comfortable identifying and accepting uncertainty.