The effects of negative interest rates. A brief reply to Miles Kimball

Miles Kimball has developed a very intriguing idea. He wants there to be an exchange rate between physical cash and electronic money (deposits) and a steeply negative interest rate on deposits to stimulate spending in the Eurozone. If an exchange rate between cash and deposits can be set to engineer an annual depreciation of physical cash of say 6%, then interest rates could be set at -6% on overnight interest rates set by the central bank. He thinks this would have a greater impact on spending than a 5% of GDP cash transfer from the ECB to the general public, leaving interest rates unchanged.

Before I go through the thought experiment as to how people might respond, it is worth being clear about the relatively predictable effects of the alternative policy – cash transfers to households. If all households received several thousand euros from the central bank, we know the range of outcomes: the effect on GDP over the subsequent 12 months is somewhere between 3% of GDP and >5% of GDP – but probably closer to 3% of GDP. The longer-term effects of course could be much greater, because the economy gets out of recession much quicker – so people don’t lose jobs, businesses don’t fail etc. Also, in contrast to the frequent counter-argument that much will be saved or used to repay debt, repayment of private sector debt is in fact a positive consequence: part of post-crisis stagnation is high savings preferences and a debt over-hang. Repeated cash transfers would alleviate this.

Ok, so the economic consequences of cash payments from the ECB are predictable – because we have the precedent of ‘tax rebates’ etc.

What about the effects of -6% interest rates on o/n rates and a depreciating rate on physical cash? The most striking initial observation, is that we have no clue what the effect would be (as the research by Chris Carroll, cited here makes clear).

Here are some further, more specific, thoughts:

1) A 6% negative policy rate is likely to make the yield on most fixed income assets negative, and possibly many borrowing rates negative. But this is not a simple or straightforward matter. Credit risk premia may be very large and the yield curve very steep – particularly as cash is now a rapidly depreciating asset – and many liabilities will not be.

2) For the same reason, the impact on borrowing is, in fact, highly unpredictable. The impact on actual lending rates is unlikely to be equivalent to the decline in policy rates.

3) It is likely that significant resources (both time, labour and capital) will be diverted from productive activities to finding substitutes for cash, deposits and other fixed income assets.

4) Negative interest rates will cause major asset price volatility. For example, you would expect a huge increase in demand for positively-yielding equities. Enterprising financial firms might even offer checking accounts on equity funds. You could easily get an equity bubble and bust in short order. Whether this would boost demand or not is a complete gamble: net wealth is declining because of negative rates on fixed income assets and stock market gains are a wealth illusion for many holders. It’s possible there is some ‘wealth effect’, thanks to an equity bubble – but this is a terrible basis for a demand stimulus. Have we learnt nothing?

5) The impact on the supply of credit and the financial state of the banking sector is highly uncertain. Banks cannot simply substitute safe assets for risky assets to increase their profits. Their ability to lend depends on demand for credit, credit risk, and the future path of rates. The stock of reserves is fixed by the central bank – and the value of these reserves will now be falling significantly. If the non-financial private sector is trying to find substitutes to holding deposits, banks might be out of business. The incentive to disintermediate is huge.

6) It is true that the CB could allow banks to borrow more reserves at negative rates – but if this rate is substantially lower than the rate that banks earn on reserves, this is essentially a helicopter drop for banks, which could be done anyway – and would be sub-optimal.

7) The premise of this policy also denies the possibility that at a certain point, income effects dominate substitution effects and lower interest rates actually cause people to save more rather than consume more (the donut problem). We may already be at that point, in which case negative rates would further depress demand and possibly cause deflation.

Given 1-7, negative policy rates may fail to cause negative rates on longer maturities. Why? Because people will assume that because of the significant welfare costs associated with negative rates and their questionable efficacy, they will not last. After all, countries with negative policy rates are already facing growing opposition – and no evidence of positive effects.

The conclusion is clear to me. Floating cash and charging steeply negative interest rates on deposits is very clever. It stimulates demand in some very naive economic models, where everything is equilibrated by a single real interest rate. But this is not a necessary property of theory. As Keynes pointed out, the market for savings and investment is not brought into equilibrium by interest rates  – and not because interest rates cannot go negative.

When I think through plausible behavioural responses to this novel policy, the probability of it completely backfiring seems quite high. On the other hand, the alternative – a money-financed tax cut – is a predictable stimulus.

Postscript
Miles Kimball’s work is persistently original and brilliant, and though we disagree (for the time being!) on the merits of negative interest rates, it is well worth reading. I have also not done justice to the subtlety of his argument. For more detailed thoughts on negative interest rates from Miles, I highly recommend this.