Should we worry about the power of monetary policy in the future?

I’m going to exploit this question as an excuse to talk about an interesting Job Market Paper written by my student and co-author Mengbo Zhang of UCLA.

“Loan Market Power and Monetary Policy Passt-through under Low Interest Rates”

Background. There has been a secular decline in real rates in advanced economies over the past decades — see figure below from NBER discussion of work by Rachel and Summers. This decline has been a source of concern for central bankers, in light of lessons obtained from the new-Keynesian model. The nature of the concern is two-fold: one has to do with magnitude of possible responses and the second about the pass-through of policy rates.

Concerns about magnitude. First, there’s is a well understood concern that if real interest rates approach zero, nominal rates must follow to guarantee an acceptable amount of price stability. The problem with this is that nominal rates will likely be closer to the zero-lower bound on policy rates. Therefore, monetary policy is likely to hit its limits more frequently.

This concern is formally understood from the work of Gauti Eggertson, Taisuke Nakata, Ivan Werning, and many others.

This concern has to do with how much rates can fall in a point in time, but not about the power of monetary policy. Put it differently: take two paths of policy rates that only differ in levels. Standard models will predict similar effects. The concern is about the admissible magnitudes of the drop, but not about the power.

One important related question is what is the actual “zero” of monetary policy — see the point below. However, let me postpone this comment to the very end.

Decline in bank profitability. The second concern fits the following narrative: as nominal rates continue to fall, bank profitability will decline. Thus, it is believed, that effectiveness of monetary policy will decline also. One way to quantitatively interpret this narrative is to say that the pass-through from monetary policy rates to deposit rates will fall, as policy rates are lower, all as a result of lower bank profitability. The policy pass-through is a measure of the change in deposit rate to changes in monetary policy rates.

Why deposit rates? Well, because deposit rates affect consumer aggregate demand, according to the insights of the new-Keynesian model. If the policy concern is valid, this means that in a low-interest rate environment, a reduction in policy rates is less effective in stimulating output. Thus, not only should we worry about interest rates hitting zero more frequently, but also about the pass-through. This is the concern in Zhang’s job market paper is interested in validating.

Zhang’s job market paper. Zhang (2020) is an empirical and analytic investigation of the second concern, that is, that low bank profitability induced by low nominal rates will reduce the monetary policy pass-through to deposit rates. The context is the US economy: the message of the paper is that the concern about lower bank profitability is valid only in locations where the bank loans market is highly competitive. Hence, the overall decline in the pass-through from policy rates to deposit rates depends on the distribution of market concentration in the loans market. The message is interesting for two reasons: First, because it implies that to qualify the concern central bankers should look at concentration across counties and, furthermore, to consider an aggregate pass-through that takes into account market concentration. Second, and this is the most striking feature, to gauge the effects on the deposit pass-through, we should look at concentration in the loans market, not in the deposit market!

The paper first presents evidence regarding the change in the pass-through rate and then builds a model that rationalizes why the pass-through rate can increase with lower rates as markets are concentrated. The figure below is a snapshot. Focusing on different banking units — US counties — the figure presents a scatter plot of the measured pass-through rates. The x-axis and y-axis correspond to the pre- and post-crisis measurements. As we can see from the figure, Zhang finds evidence that the policy pass-through rate has changed across counties in a detectable pattern: In areas where the pass-through was high prior to the crisis, we have observed a decline in the pass-through rate. In areas where the pass-through was low, the pass-through has increased. The pattern is super clear.

Zhang then successfully demonstrates a systematic connection between the initial pass-through rate and the degree of bank concentration. In particular, concentration in the loans market — measured via the HH index — is correlated with high initial pass-through rates. It is important to state that the evidence is about concentration in the loans market, not in the deposit market. This aspect is instructive for the model. Because Zhang finds that areas where the pass-through was originally low, were areas with high concentration in the loans market, this reflects a spill-over from concentration in loans markets, which is a different market than the deposit market.

In Zhang, households can substitute deposits for government securities. Importantly, borrowers cannot access the rate of policy tools. The mechanism in Zhang works because monetary-policy rates carry two effects on banks. On the one hand, they reduce bank profits on their securities portfolio, but on the other hand, they reduce the marginal funding costs on deposits.

In areas where loan market competition is fierce, changes in nominal policy rates do not translate into profits by much, do to the high degree of competition which prevents charging large spreads above marginal funding costs. Thus, bank profits in the loan market are not very sensitive to changes in rates, but the decline in rate erodes profits on the security portfolio. This force translates into lower deposits rates for borrowers, to offset the decline in the securities portfolio. By contrast, in area with substantial bank concentration in loans, reductions in marginal funding costs do translate into greater markups, that offset the decline in the securities portfolio. As a result, deposit rates don’t need to fall that much.

Aggregating across areas, weighting by market power, Zhang finds that the current pass-through has not fallen by much.

Oliver Wang’s job market paper. A recent paper on the subject is Oliver Wang’s job market paper, a study intimately related to Zhang. In Wang’s model, deposits and currency are imperfect substitutes as means of payment. The mechanism in Wang roughly works as follows: to stay in the bank, bank equity must earn an appropriate return on equity, given by some real interest rate. The return on equity (ROE) is given by the following simple identity:

ROE=return on loans+interest differential * leverage.

When nominal rates are above that real rate, the deposit rate and loan rates can be equalized, and the return on equity will equal the rate on loans.

However, when policy rates are below the required return on equity, the banking system must earn some for of quasi-rents, as an interest rate differential between loans and deposit rates. Critical to this are some form of leverage constraints or reserve requirements. In any case, the effects of low interests will be passed on as a tax on intermediation. Hence, the effect of the monetary policy pass-through to deposit rates should be interpreted as a logic of tax incidence. What is interesting about Wang’s model is that the elasticity of substitution between currency and deposits increase as rates fall, due to the nature of substitution. Thus, the deposit side becomes more rate elastic, as with any tax, the effect on prices shows up on the more inelastic side.

Thus, the pass-through to deposit rates will be lower as rates fall.

Now, back to the first concern.

Bigio-Sannikov. There are some interesting papers on the zero-lower bound that share a related logic. For example, in my own work with Yuliy Sannikov, “A model of intermediation, money, interest and prices” households can transform currency into deposits freely. Bank ROE is zero because there is free entry and there are no leverage constraints. Thus, the elasticity of substitution is infinity at zero-nominal deposit rates. As a result, when policy rates are negative, the entire decline in rates is passed on to borrowers. In that paper, the effect of policy rates becomes contractionary at negative rates. A similar force feature is found (independently) in Brunnermeier and Koby, also. However, this is a secondary point of my paper.

Now, in this paper of mine, zero is zero. That seems not to be the case in the data and something addressed in some recent papers.

The real zero. Two beautiful papers come to mind when asking what is the real zero of monetary policy rates, one is Ulate (2020) and the other is Balloch and Koby (2020). Both papers share in common with Zhang, that the deposit market is not perfectly competitive. All three papers build on the “The deposit channel of monetary policy” by Itamar Drechsler, Alexi Savov and Phillips Schnabl, a paper that I have learned to appreciate with time. By introducing imperfect competition, the elasticity of substitution for deposits and competing assets is not infinity at zero rates. This is important because it tells us what is the actual “real zero” where monetary policy begins to carry adverse effects. That real zero, can be much below zero. What I like about Ulate and Balloch&Koby is that by looking at data, they carefully identify the value of the negative rate on reserves that reverses the effects of monetary policy. The claims seems to be that the real zero is about -0.5%. Some more room.

Peter Paz. Another interesting and related paper is the Job Market Paper by Peter Paz from NYU. In this paper, Paz studies the pass-through of monetary policy combining recent methods to identify monetary policy with bank-level data on bank capitalization and a careful attempt to match estimations with a structural modeling. I see this paper as a modern incarnation of a famous paper by Kashyap and Stein (2000) “What Do a Million Observations on Banks Say about the Transmission of Monetary Policy?”. Paz finds evidence of a balance-sheet channel at the bank level which is actually different from the liquidity channel envisioned by Kashyap and Stein.