Confused Sumner on banking reform
(Very busy period for me, so blogging frequency is low and I have literally a ton of things to catch up with)
Scott Sumner just wrote a couple of posts about banking on Econlog. And there is quite a bit to say about them.
And I don’t understand the point about “creating such accounts out of thin air”. The whole subject of “money creation” and the “money multiplier” is surrounded by confusion, and quite frankly, stupidity. All well-informed observers understand what banks do. Whether you prefer to call that money creation, a money multiplier, or something else depends on how you prefer to define terms. I prefer to define ‘money’ as the monetary base, but most economists prefer a broader definition of money—cash plus bank deposits. No one claims that banks create base money out of thin air. I happen to think the term ‘money multiplier’ should be dropped as it causes more confusion than it is worth.
I totally agree with this. Well, apart from the fact that there are indeed people who really believe that banks create money out of thin air. And some others, such as most elaborate endogenous money theorists, who believe (also wrongly) that banks create money out of thin air, but only indirectly, through extending credit and only then borrowing the required reserves at the central bank, which has to oblige to maintain money market rates within its target range.
But I am a little bewildered by the fact that a libertarian such as Sumner would be willing to reform and constrain banking in a top-down, Keynesian and Martin Wolf-type, fashion. This sounds really contradictory to me. He wishes to exclude banks as much as possible from the monetary system, thereby giving the central bank (a centralised institution outside of the market realm and which has no real independence from the state unlike many people believe) a very large power over money and credit. This also goes against Milton Friedman’s late views that we should end central banking.
In his post, Sumner claims that “multiplier instability is not a problem today, under our current fiat money regime. The Fed can offset any fall in the multiplier, keeping the money supply unchanged.” It can of course, but doesn’t mean that this new money circulates in the real economy. And IOR is not the only factor that drives whether or not banks hold a large share of the monetary base or not.
Martin Wolf’s banking reform proposal isn’t recent and I have already commented on it. But Sumner doesn’t seem to get the accounting behind Wolf’s (or indeed Positive Money’s) investment accounts. He claims that:
Under Wolf’s proposal, banks would be creating deposits “out of thin air” just as much as today. You could deposit $1000 in a bank investment account. Someone might borrow $900 of that money. The borrower might then deposit that $900 in another bank, which then loans out $810, etc., etc. As long as you have bank accounts not backed by reserves (i.e. investment accounts), banks will be creating deposits in pretty much the same way they do today.
No. Investment accounts are effectively like mutual funds or P2P lending or direct investment in bonds, etc. It is a reserve transfer. You know that you don’t have access to your cash anymore. You just end up with a claim on this cash that has a maturity date (i.e. a financial asset), which could possibly be marketed later to try to recover reserves before maturity, at risk of losses. Investment accounts involve an asset swap on the asset side of a balance sheet. Purchasing a house is the same thing as putting your money in an investment account. The borrower, on the other hand, credits both the asset side of his balance sheet (i.e. he acquires some new cash) and his liability side (i.e. he now also owes an equivalent amount plus interests). If the borrower makes the choice to place this newly acquired cash into his investment account, then the same process starts over again: an asset swap for him, and a double entry on the new borrower’s balance sheet. Most importantly, there is normally no maturity mismatch in those accounts. Only one person has an on-demand claim on those reserves at any one time.
What Sumner is describing is a normal fractional-reserve account. Money placed in those are still supposed to be available on demand (or at short-notice) to the depositor. But in reality, they are ‘loaned out’ to borrowers, creating a maturity mismatch, and hence the so often confusing and misunderstood money multiplier (as several people effectively hold claims to the exact same reserves).
When he says that investment accounts should be backed by reserves, he is confused. Any account, or indeed investment, backed by reserves implies no cash transfer. And hence no investment in the first place. One cannot buy a house and keep the cash used to buy that house. One cannot have his cake and eat it. Of course some particular assets are near-money. But they are nevertheless not reserves.
As for his proposal that banks should be required to back bank accounts with T-bonds, I agree with Bill Woolsey’s comment:
Requiring transaction accounts to be “backed” by T-bills or reserves is not at all the same thing as requiring FDIC insured accounts to be “backed” by T-bills or reserves.
It would be possible to have transaction accounts that are not FDIC insured. They could be used for transactions but could be “backed” by a variety of assets. In my view, trying to prevent runs by regulating bank assets is unlikely to be effective. It just drives financial innovation aimed at regulatory arbitrage.
And not all FDIC insured accounts are transactions accounts. Savings accounts and Certificates of Deposit that are FDIC insured would be matched on bank balance sheets by reserves or T-bills under such a proposal.
Sumner’s second piece, published a few days ago, is also rather curious. It tries to explain that the 2008 crisis could have been avoided if stricter capital requirements had been applied to banks. First, this is another strange libertarian position, and second, I thought Sumner used to argue instead that banking was irrelevant and that the crisis could have been avoided if the Fed had offset the NGDP fall.
Scott is (and many commenters are) unfortunately quite confused here, mixing up things like subordinated debt, contingent convertible (CoCo) bonds, TLAC requirements… Some sub debt had been accounted for as regulatory capital since Basel was first introduced several decades ago, as Tier 1, 2 or 3 capital depending on their features. They miserably failed to absorb losses during the crisis. CoCos are not an idea of Calomiris and Herring. Some banks have been issuing them since 2009 and regulators have progressively devised rules to include them as part of regulatory capital. High conversion trigger CoCos are now usually considered ‘additional Tier 1 capital’ (AT1). There remains a lot of doubts as to how those instruments will perform during a crisis, and as to whether or not there is no legal recourse to prevent conversion (also see some criticisms against CoCos here).
Finally, TLAC (‘Total Loss Absorbing Capacity’) is regulators’ latest invention: a layer of debt that is senior to CoCos but junior to senior debt and which is ‘bail-inable’ in case of bankruptcy (in order to reduce the amount of losses that the bank’s equity base, and hence depositors, has to bear). Banks’ capital requirements have become ridiculously complex (see my earlier post on this topic) and no one has any idea how things will unravel during a crisis (see this interesting PwC publication on TLAC limitations). Whether CoCo or TLAC bonds are the miraculous tools that could have prevented the crisis is highly unclear (see this older post on the limitations of artificial regulatory capital triggers).
Sumner wishes (rightly) to reduce moral hazard in the banking system. But those recently taken banking capital reforms do not really solve that problem, and in fact adds a lot more complexity and opacity to the system.