Tuesday, July 12, 2011

Sweden Again

So, the otherwise good Matt Yglesias is at it again, hyping Sweden's famous "negative interest rate" experiment.

As I've noted before, this entire story is the result of a misunderstanding and nothing that Matt has said on the subject is true.

To recap:

1) Sweden never paid a negative interest rate on overnight reserves, and in fact they never stopped paying positive interest rates on overnight reserves.

2) Paying interest on reserves does not cause banks to "hoard reserves" and refuse to loan them out because banks do not loan their reserves to the public. Ever.


  1. "Paying interest on reserves does not cause banks to 'hoard reserves' and refuse to loan them out because banks do not loan their reserves to the public. Ever."

    This is either the fallacy of composition or just an inaccurate statement, depending on in what sense you mean it. The banking system as a whole does not lend out reserves to the public, but individual banks do, in the sense that is relevant for their incentives.

    When a bank makes a loan, it acquires an asset for which it must either create an offsetting liability or reduce an existing asset. In the latter case (e.g., when someone gets a cash advance from an ATM), the asset reduced may well be a reserve balance. If so, the Fed will typically offset the reduction in total reserves via an OMO, but since the OMO involves a different bank (as it presumably will, otherwise the first bank would have liquidated a different asset in the first place), it doesn't affect the incentives faced by the first bank. The bank has to choose between two different assets, a reserve balance and a loan, that have different risk/return/liquidity characteristics, and the return on the reserve asset will certainly affect the bank's decision about which asset to hold. More generally, the bank might reduce a different asset, such as a T-bill, but since the assets are part of the same portfolio, the return on the reserve asset affects the characteristics of the overall portfolio and thus still affects bank's decision whether to make a loan.

    The case where the bank creates an offsetting liability might seem different but it's really not, because in most cases, the offsetting liability will soon be withdrawn, and the bank will then have to reduce an asset. Presumably the borrower took out the loan with the intent of spending it. Unless the borrower's vendor happens to have an an account at the same bank, the bank will have to settle the transaction by transferring reserves to the vendor's bank. From the point of view of the lending bank, it is loaning its reserves, even though the reserves remain in the banking system.

  2. It's not clear to me why you think what you're saying means that banks loan out their reserves, but you've gone to some trouble so I'm going to create a post exploring this tonight.

  3. They key part you seem to be missing is that the bank created the reserves it's loaning out.

    If Bank A starts with no reserves, and makes a $10k loan, it creates a $10k deposit account and a $10k loan account. The $10k deposit is then spent on a new car, and the car dealership banks at Bank B.

    So now Bank A is back to having no reserves, which is what it had to begin with. It didn't loan out its reserves - it created reserves which got moved to another bank.

  4. I don't understand your example. A bank with no reserves can't make a loan. If a bank has no reserves, then either it has no deposits, or else it is failing to meet the reserve requirement. Either way it can't legally make a loan.

    And banks don't create reserves (except when they take currency as a deposit, if the currency was previously in circulation).

    I don't see how there is any creation of reserves in your example. When the car purchase settles, Bank A has to transfer reserves to Bank B. Either Bank A already had those reserves before it made the loan, or else it obtained them (via either new deposits or borrowing in the federal funds market) at the time it made the loan (or immediately after). The only times reserves are created are (1) when the Fed does an OMO and (2) when someone deposits currency in a bank account.