Update, June 5, 2019: On this topic, be sure to check out this more recent blog post: “Ruchir Agarwal and Miles Kimball—Enabling Deep Negative Rates to Fight Recessions: A Guide.” The paper of that title expands on the idea that, except for the concern about disintermediation or “debanking,” it is quite speculative to think we are anywhere close to an effective lower bound on interest rates. An effective lower bound other than a concern about debanking would require essentially the conversion of the entire national debt into paper currency. For the US, that would require the storage of something on the order of $20 trillion as paper currency, which is a Herculean task. Long before that, the Fed would be worried about the consequences to the banking system. It is those worries about the consequences of deep negative rates for the banking system to form the effective lower bound.
Once the convenience of electronic money for a large fraction of transactions (especially large transactions and business to business transactions) is recognized, the key to enforcing the zero lower bound in many stark formal models that assume a traditional paper currency policy is the emergence of some equivalent to money market mutual funds backed by paper currency. That is, in the models, it is the “electrification” of paper currency by funds that have paper currency as their key asset and an electronic means of exchange as their key liability that creates the zero lower bound.
But in the real world, the business model for money market mutual funds backed by paper currency is not that promising, given the danger that the government is likely to be annoyed at such funds thwarting a negative interest rate policy, and is likely to act against such funds. Some things are relatively easy to do in opposition to a government, but setting up a money market mutual fund with access to the electronic payments system is not one of them. Even Bitcoin is much more vulnerable to any government action against it than some of its enthusiasts think. And something trying to look as much as possible like a regular money market mutual fund, but backed by paper currency is much more vulnerable to the government than Bitcoin.
Thus, when a central bank lowers its target rate and interest rate on reserves, it is not large scale paper currency storage by the equivalent of money market mutual funds backed by paper currency that is the first symptom of the effective lower bound kicking in, but expanded small-scale storage of paper currency by households and businesses keeping more of their liquid balances as paper currency instead of in bank accounts, or the fear by banks of such a shift into paper currency. That is, the first symptom of the effective lower bound kicking in is disintermediation or fear of disintermediation.
One can debate just how important banks are to the functioning of the economy, but given how much economic activity currently runs through banks, it seems likely that trying to move all of that economic activity outside of banks within too short a time period is likely to be disruptive. So sudden disintermediation is probably something to be avoided. And banks’ fear of households and businesses shifting toward small-scale paper currency storage is likely to lead them to cut deposit rates less than the extent to which other interest rates fall when the paper currency interest rate is kept at zero in a traditional paper currency policy. This could lead to a reduction in the net interest margins that are a mainstay of bank profits.
If banks are extremely well capitalized going into a period of negative interest rates (that is, with a lot of equity on the liability side of their balance sheets), a strain on bank profits and consequent effect on equity levels will have very little effect on the probability of insolvency. But if banks go into a period of negative interest rates with relatively low levels of equity, the strain on profits could lead to more serious undercapitalization, with a consequent heightened probability of insolvency. And low bank profits can be a political problem even in situations where they are not a serious economic problem at all.
One solution to the strain on bank profits and consequent reduced capitalization is to subsidize banks. This is the approach I discuss in “How to Handle Worries about the Effect of Negative Interest Rates on Bank Profits with Two-Tiered Interest-on-Reserves Policies.” But the other approach is to lower the paper currency interest rate, to avoid most of the problem of a profit squeeze on banks in the first place. That is the approach I discuss in “If a Central Bank Cuts All of Its Interest Rates, Including the Paper Currency Interest Rate, Negative Interest Rates are a Much Fiercer Animal.” In other words, if a central bank begins to see the particular strains on bank profits that are symptoms of the effective lower bound in action, it should consider eliminating the effective lower bound itself by lowering the paper currency interest rate, as discussed in “How and Why to Eliminate the Zero Lower Bound: A Reader’s Guide.”