Goldsmiths were the forerunners to modern bankers. Originally, they would issue receipts to certify gold was deposited in their vaults. These eventually gave rise to fractional reserve banking, as goldsmiths used a portion of the gold to make loans.
Well, we might be on our way back to the original version, but instead of keeping our gold safe, banks will be keeping our dollar, Swiss franc, yen, and euro notes safe!
Believe it or not, this story is tied to a current puzzle facing investors and monetary policy makers alike: how low can nominal interest rates go? Answer: if they go too low, we will revert to the early days of the goldsmiths.
Turning to the current circumstance, several central banks have set their deposit rates below zero, including the ECB (-0.20%), the Danish National Bank (-0.75%) and – after abandoning the exchange rate floor – the Swiss National Bank (-0.75%). Markets have followed these central banks through the looking glass, with Swiss bond yields below zero up to 10-year maturities, and German government bonds offering negative yields up to 5 years.
This seems upside down. Lenders paying borrowers to borrow? Isn’t there a zero lower bound (ZLB) for nominal interest rates?
While a debate is now under way, our answer is that we have probably hit the sustainable interest rate floor and that attempts to go meaningfully lower (aside from temporary moves) are likely to prove counterproductive: instead, depositors will shift into cash and banks may be happy to shrink their balance sheets.
Why? Recall that bankers run what is called a “spread business.” They rely on receiving higher returns on their assets than they pay on their liabilities. In the simplest case, that means charging borrowers a higher rate for loans than they pay customers for their deposits. That is their spread.
This spread business usually doesn’t depend on the level of interest rates. That is, without going any further, one might naively conclude that if the bank’s interest rate spread is 3 percentage points (a fairly representative level), then with assets yielding ‒1%, they would offer depositors ‒4%. The bank’s interest margin has several parts, including the cost of screening and monitoring borrowers, the return to the bank’s owners, and the cost of providing services to depositors. This last bit, which includes providing payments services, account statements, personalized customer service and the like, is important for our story, so we will come back to it.
Now ask yourself the following question: how would you feel if your bank were to charge you 4% of your balance each year for the privilege of holding a checking account? You will gladly pay for some banking services: it would be inconvenient to live without things like check writing, automated payments, and web access to your account. Let’s say that those services cost 0.5%, roughly the upper end of the expense ratio for money market mutual fund (MMMF) accounts. As for the remaining 3.5%, our guess is that many businesses and households would try to figure out how to avoid it. And, we strongly suspect banks (or their competitors) will accommodate them.
The financial system is capable of creativity and innovation in satisfying client demands. When double-digit inflation drove market interest rates well above the regulatory ceiling on deposit rates in the 1970s, MMMF accounts flourished. By holding commercial paper liabilities, these were able to offer market-based interest rates.
In the case of negative deposit rates, the alternative asset is currency, which bears a zero nominal interest rate. People have made quite a bit out of the costs of storing large quantities of currency, and we agree that it is expensive for most individuals. But currency storage is like gold storage. Like the goldsmiths of yore, banks can rent out vault space.
Suppose, for example, that banks were to offer customers the ability to convert their deposit accounts into “cash reserve accounts” (CRAs) that only hold currency. Like the sweep accounts banks created in the 1990s to minimize reserve requirements, CRAs would minimize the holding of central bank deposits. The funds would be available just like a checking account during the business day, but would be swept into cash currency rather than being held as reserves at the central bank. Overnight, the piles of currency belongs to customers, during the day they are the asset balancing the CRA liability on the bank’s own account. Critically, the cash in the bank’s vault assigned to the CRAs is not available to be lent out. That is, the textbook version of the deposit expansion multiplier is not operating when depositors are holding currency.
There may be some legal or regulatory impediments to a chartered bank creating CRAs. If that is the case, as it was with the MMMFs, we expect that some clever lawyers will find a way to create a nonbank subsidiary to do it. Or, as has been suggested, maybe an exchange-traded fund.
All of this leads us to conclude that the sustainable lower bound on nominal interest rates is the marginal cost of supplying CRAs. That is, the cost of currency storage and moving the currency around. This business probably has some significant scale economies (if there were only one CRA, it wouldn’t even have to move currency around!). We would be surprised if the CRA cost significantly exceeds 50 basis points. If we’re right, then there isn’t a zero nominal interest rate bound, there’s a minus-fifty-basis-point nominal interest rate floor. Below that level, depositors eventually will switch into CRAs.
Banks’ behavior may reinforce that of depositors for at least two reasons. First, in some cases, formal or informal agreements could preclude charging depositors significant fees. As a result, banks may have difficulty recovering the costs of offering conventional deposit accounts – both those associated with the negative interest rates they will have to pay on their assets and the ones arising from servicing customers. Second, deposit balances themselves are likely to be more volatile, forcing banks to hold a higher fraction of their assets in highly liquid instruments that have the most negative yields. As a consequence, the deposit business will be less and less profitable the more negative interest rates go.
The economic implications of this logic are not very pleasant. Instead of being stimulative, setting nominal interest rates persistently too low could result in tighter monetary conditions.
A simple way to see this is to recall that rising currency holdings reduce the textbook “money multiplier” – the ratio of a monetary aggregate such as M2 to the monetary base composed of central bank liabilities. In the chart below, we have plotted the M2 money multiplier starting in 1928 along with the same quantity starting in 2007 – allowing us to compare behavior in the Great Depression with that in the recent financial crisis. You can see how the M2 multiplier actually plunged deeper and faster in late 2008 and early 2009 than it did during the banking panics of 1930 to 1932. At the end of 2014, the multiplier stands at just over one-third its 2006 level. That is a much larger drop than occurred in the early 1930s. We argued elsewhere that the Federal Reserve’s aggressive stimulus prevented this recent collapse from leading to another depression.
M2 Money Multiplier: The Great Depression vs. the Recent Financial Crisis
(monthly, 1928 to 1936 and 2007 to 2014)
Source: Cecchetti and Schoenholtz, Money, Banking and Financial Markets 4e, and FRED.
The U.S. Federal Open Market Committee is not contemplating reducing its deposit rate below zero, so this is unlikely to be a problem in the United States. But continental Europe is a different story. There, we see some real risks. If the European Central Bank tries to cut rates much further (and keep them there), we would expect the broad money multiplier to fall. Furthermore, if – as is likely – negative rates hurt profitability, banks facing capital shortfalls will be squeezed even further. Such weak banks generally do not lend. In other words, the attempt to lower deposit rates to stimulate the euro-area economy could lead to precisely the opposite outcome.
Our bottom line is fairly simple. We doubt that the lower bound on nominal interest rates is much below zero over any extended period of time. Trying to keep nominal rates below the cost of currency storage and movement would convert bankers back into goldsmiths, tightening rather than loosening monetary conditions.