Should the European Central Bank do more and go negative?

The European Central Bank is torn between those who say that it is already doing too much, mostly in Germany, and those who say it should do more. One of the actions the ECB should pursue, in the opinion of the latter group, is to bring the deposit rate into negative territory. 
The attitude of the ECB towards this step has been mixed. President Draghi said: “(…) we are technically ready [to set a negative deposit rate] and it is part of our artillery (…) if we see a low rate of inflation. (…) [but] we do not think it is going to materialise because we see that inflation expectations are firmly anchored at 2%, or less than 2%.” [1] His statement was followed by those of many others ECB representatives to the effect that, yes, the rate on the deposit facility of the ECB could be brought in negative territory, but there is reluctance from the ECB to go in that direction. In this post I give my understanding of this ambivalent position of the ECB and conclude that indeed, if forced by persistent, downward deviations from its inflation target, the ECB could go negative. However, as I wrote in a previous post, this move figures, in my view, quite low in the rank of measures that the ECB could take.
The first thing to realize is that negative rates are an exceptional, I would indeed say odd, phenomenon, both conceptuallyand empirically. Conceptually, the real rate of interest has a long term rendez-vous with the return on capital, or the marginal efficiency of capital if you prefer the Keynesian name, which should be ex-ante positive if investment decisions follow an economic logic. Barring a persistent and relatively high deflation, a positive return on capital should translate into a positive nominal interest rate. Empirically, the cases of negative nominal rates are very rare and nearly unprecedented when it comes to monetary policy rates. Anders Jørgensen and Lars Risbjerg [2]indeed write:”“For the first time in its nearly 200-year history, one of Danmarks Nationalbank’s interest rates is negative. Negative monetary-policy interest rates are also unique in an international perspective.” [3] The other known case is that of the Sveriges Riksbank, which lowered the rate on its deposit facility to -25 basis points between July 2009 and September 2010. However, the use of the deposit facility with the central bank by Swedish banks is minimal, so the measure hardly had any practical impact.
The fact that negative rates are exceptional is consistent with another fact: they cannot go too deep into the negative domain. The reason is straightforward: there is a central bank liability, i.e. banknotes, that intrinsically carries a zero rate of interest and this would establish an arbitrage opportunity if interest rates would be brought too far below zero. The lower bound for interest rates is not literally zero, as there are (warehousing, transport,…) costs in “investing in banknotes” to profit from its zero return, but these costs do not allow bringing the return on other assets much below zero. Minus 25 basis points is a reasonable estimate of the absolute lower level to which interest rates can be brought, before the arbitrage with banknotes starts to be effective. It is not by chance that both Danmarks Nationalbank and the Sveriges Riksbank did not exceed this lower bound.
The two previous comments imply that we are necessarily discussing about a small step, which in principle should have little effect. Question marks are often raised about the real effect of the standard monetary policy move: a quarter per cent increase or cut of rates. The effect of monetary policy is, however, regained noting that the central bank can increase the potency of monetary policy by cumulating a number of quarter point cuts or increases. This possibility is not available, however when moving to negative rates: only one shot is possible. Of course, there may be disproportionate effects just because the move is so exceptional, but when you move from economics to psychology, forecasts are no longer difficult, they become quasi impossible. Still, it is useful to venture into some possible economic effects of this necessarily small move, also looking at the case of Denmark.
A clear impact of the decision of Danmarks Nationalbank to bring the remuneration of its certificates of deposit to -20 basis points in July of 2012 was on the Krone/€ exchange rate: as desired, the krone depreciated and lessened the pressure on the Danish central bank to defend with foreign exchange interventions the parity against the € in the Exchange Rate Mechanism. The size and the meaning of this impact has to be carefully considered, however, before concluding that the effect on the exchange rate is something that could induce the ECB to follow the Danish example. In terms of size, the impact depreciation of the Danish currency was less than 0.5%, which is not surprising recalling that the rate could only be cut by 20 basis points. In terms of meaning, the impact on the exchange rate was relatively important for the Danmarks Nationalbank, which manages very tightly its exchange rate against the €, but a 0.5% exchange rate change for a freely floating currency like the € is a daily occurrence, not something to be excited about.
Similar considerations can be made for the short-term rates: the CITA, i.e. the Danish short term interest rate swap, and the Danish Treasury bill rates came down as a result of a “portfolio balance effect”, when the rate on certificates of deposit was made negative, but the effect was small (with something of a unitary elasticity) and gradually dissipated over time. Consistently with the limited duration of the effect, the yields on the maturities beyond 5 years increased on impact instead of coming down, as can be seen in chart 1, taken from an interesting paper on the Danish experience.
                       Chart 1: Danish 3-months interest rates and zero coupon curves
Source: Jørgensen A.,  Risbjerg L. Negative Interest Rates
In trying to extract information from the Danish case that would be useful to shed light on the possible use of negative rates by the ECB, one has also to take into account the different central bank liquidity position in the two jurisdictions. In the €-area the still large excess liquidity derives from very large provision of liquidity by the ECB in excess of reserve requirements and autonomous factors, which goes into current account and the deposit facility (for a total of about 180 billion on Nov 27th). This means that, given the remaining tensions in the € area, the ECB still carries out a large intermediation function between banks in the periphery, which do not find all the liquidity they need in the market, at least at reasonable cost, and banks in the core, which are loath to lend liquidity to banks in the periphery. In Denmark, instead, the central bank does not need to carry out such intermediation function and therefore there is no contemporaneous large “lending” and “borrowing” by the central bank. This difference is important: if the ECB would make the deposit rate negative while keeping the Main Refinancing Operation rate (MRO), at which it lends to banks, unchanged, the corridor between lending to and borrowing from banks would become larger and thus make central bank intermediation more costly, which would mean a tightening rather than an easing of monetary policy. This effect would, of course, not be there if both the deposit and the MRO rate were lowered.
The area of biggest uncertainty in forecasting the effect of a negative ECB deposit rate has to do with the behaviour of banks, especially on their lending. Indeed, on the liability side, it seems unlikely that banks would be able to translate negative rates onto retail depositors: this did not happen in Denmark and, thinking about the €-area, one can identify three kinds of reasons why it would not happen there as well:
  •  IT systems of banks may not be able to accommodate negative rates,
  • The millions of current account contracts with retail customers, which were clearly written with positive or zero remuneration in mind, may not accommodate negative rates,
  • There is likely to be great resistance on the side of customers to the idea that someone deposits 100 € in a bank and gets at the end of the year less than 100 €.
As I mentioned above, the forecast about the effect of a negative ECB deposit rate on the cost of bank lending is very difficult. In the Danish experience, lending rates may have, overall come  slightly down, thus affecting bank margins, but in many market commentaries the fear is expressed that, in the €-area, banks may increase the lending rates to recover the profitability lost because of the negative return they would have on its deposit with the central bank. For a 20 basis point negative rate this would amount to a yearly loss of about half a billion on the outstanding amount of circa 250 billion of deposits and current accounts they hold with the central bank. Of course, this estimate requires an assumption of how much banks would be allowed to keep on their current accounts with the ECB, if the return on these was not also made negative. Indeed banks would try to maintain as much as possible on current accounts, with a zero return, if deposits were to have a negative return. It is useful, in this respect, to recall that current accounts in Denmark were rationed.
Chart 2: Danish retail rates
Source:National Bank of Denmark
Two final sources of uncertainty would be the impact on the turnover on the money market and on the functioning of money market funds. Both are likely to be negative, but the size of the negative effect is hard to estimate.
In conclusion, there are significant uncertainties for at most a modest benefit. If the ECB choice was an investment choice, one would say that the Sharpe ratio is unfavourable, as the expected return is low with respect to its risk. Of course, as I mentioned already, the cost benefit analysis could be very different if one had a clear view that a negative deposit rate would have strong positive confidence effects. While I declare here my lack of competence on psychological effects, my long experience makes me very wary of justifying a measure of uncertain economic effects with difficult-to-forecast psychological benefits.

[***] Research assistance was provided by Mădălina Norocea.
[1] Mario Draghi, Press conference 7th November 2013
[2] Jørgensen A.,  and Risbjerg L.; Negative Interest Rates, National Bank of Denmark, Monetary Review, 3rd Quarter 2012
[3] The Danish Central Bank lowered the rate on its certificates of deposits to -20 basis points in July 2012 and then raised it to -10 basis points in January 2013. No changes have been made since.