What the SNB should have done

I have got a lot of questions about what I think about the Swiss central bank’s (SNB) decision last week to give up its ‘floor’ on EUR/CHF – effectively revaluing the franc by 20% – and I must admit it has been harder to answer than people would think. Not because I in anyway think it was a good decision – I as basically everybody else thinks it was a terrible decision – but because I so far has been unable to understand how what I used to think of as one of the most competent central banks in the world is able to make such an obviously terrible decision.

One thing is that the SNB might have been dissatisfies with how it’s policy was working – and I would agree that the policy in place until last week had some major problems and I will get back to that – but what worries me is that the SNB instead of replacing its 120-rule with something better seems simply to have given up having any monetary policy rule at all.

It is clear that the SNB’s official inflation target (0-2%) really isn’t too important to the SNB. Or at least it is a highly asymmetrical target where the SNB apparently have no problems if inflation (deflation!) undershoots the target on the downside. At least it is hard to think otherwise when the SNB last week effectively decided to revalue the Swiss franc by 20% in a situation where we have deflation in Switzerland.

Try to imagine how this decision was made. One day somebody shows up in the office and says “we are facing continued deflation. That is what the markets, professional forecasters and our own internal forecasts are telling us very clearly. So why not test economic theory – lets implement a massive tightening of monetary conditions and see what will happens”. And what happened? Everybody in the SNB management screamed “Great idea! Lets try it. What can go wrong?”

Yes, I am still deeply puzzled how this happened. Switzerland is not exactly facing hyperinflation – in fact it is not even facing inflation. Rather deflation will now likely to deepen significantly and Switzerland might even fall into recession.

What was wrong with the ‘old’ policy?

When the SNB implemented its policy to put a ‘floor’ under EUR/CHF back in 2011 I was extremely supportive about it because I thought it was a clever and straightforward way to curb deflationary pressures in the Swiss economy coming from the escalating demand for Swiss franc. That said over the past year or so I have become increasingly sceptical about the policy because I think it was only a partial solution and it has become clear to me that the SNB had failed to articulate what it really wanted to achieve with the policy. Unfortunately I didn’t put these concerns into writing – at least not publicly.

Therefore let me now try to explain what I think was wrong with the ‘old’ policy – the 120-floor on EUR/CHF.

At the core of the problem is that the SNB really never made it clear to itself or to the markets what ultimate nominal target it has. Was the SNB targeting the exchange rate, was it targeting a money market interest rate (the key policy rate) or was it targeting inflation? In fact it was trying to do it all.

And we all know that you cannot do that – it is the Tinbergen rule. You cannot have more targets than you have instruments. The SNB only has one instrument – the money base – so it will have to focusing on only one nominal target. The SNB never articulated clearly to the markets, which of the three targets – the exchange, the interest rate or inflation – had priority over the others.

This might work in short periods and it did. As long as the markets thought that the SNB would be willing to lift the EUR/CHF-floor even further (devalue) to hit its 2% inflation target there was no downward (appreciation) pressure on EUR/CHF and here the credibility of the policy clearly helped.

Hence, there is no doubt that the markets used to think that the floor could be moved up – the Swissy could be devalued further – to ensure that Switzerland would not fall into deflation. However, by its actions it has become increasingly clear to the markets that the SNB was not about to lift the floor to fight deflationary pressures. As a consequence the credibility of the floor-policy has increasingly been tested and the SNB has had to intervene heavily in the FX market to “defend” the 120-floor.

A proposal for a credible, rule-based policy that would work

My proposal for a policy that would work for the SNB would be the following:

First, the SNB should make it completely clear what its money policy instrument is and what intermediate and ultimate monetary policy target it has. It is obvious that the core monetary policy instrument is the money base – the SNB’s ability to print money. Second, in a small-open economy particularly when interest rates are at the Zero Lower Bound (ZLB) it can be useful to use the exchange rate as an intermediate target – a target the central bank uses to hit its ultimate target. This ultimate target could be a NGDP level target, a price level target or an inflation target.

Second, when choosing its intermediate target it better rely on the support of the markets – so the SNB should announce that it will adjust its intermediate target to always hit its ultimate target (for example the inflation target.)

In this regard I think it would make a lot of sense using the exchange rate – for example EUR/CHF or a basket of currencies – as an intermediate and adjustable target. By quasi-pegging EUR/CHF to 120 the SNB left the impression that the FX ‘target’ was the ultimate rather than an intermediate target of monetary policy.

By stating clearly that the exchange rate ‘target’ is only a target implemented to hit the ultimate target – for example 2% inflation – then there would never be any doubt about what the SNB would trying to do with monetary policy.

I think the best way to introduce such an intermediate target would have been to announced that for example the EUR/CHF floor had been increased to for example 130 – to signal monetary policy was too tight at 120 – but also that the SNB would allow EUR/CHF to fluctuate around a +/-10% fluctuation band.

At the same time the SNB should announce that it in the future would use the ‘mid-point’ of the fluctuation band as the de facto ‘instrument’ for implementing monetary policy so to signal that the mid-point could be changed always to hit the ultimate monetary policy target – for example 2% (expected) inflation.

That would mean that if inflation expectations were below 2% then the Swiss franc would tend to depreciate within the fluctuation band as the market (rightly) would expect the SNB to move the mid-point of the band to ensure that it would hit the inflation target.

This would also mean that there would be a perfect ‘ordering’ of targets and instruments. The expectations for inflation relative to the inflation target would both determine the expectations for the development in the exchange and what intermediate target SNB would set for EUR/CHF. This would mean that under normal circumstances where SNB’s regime is credible the market would effectively implement SNB policy through movements in the exchange rate within the fluctuation band.

As a consequence the SNB would rarely have to do anything with the money base. Of course one can of course think of periods where the SNB’s credibility is tested – for example if a spike global risk aversion causes massive inflows into CHF and push the CHF stronger even if inflation expectations are below the inflation target. That said the SNB would never have to give up “defending” CHF against strengthening as the SNB after all has the ability to print all the money it needs to defend the peg.

Of course this is the ability that has been tested recently, but I believe that the appreciation pressure on CHF has been greatly increased by the SNB failure to move up the target in response to the clear undershooting of he inflation target. Hence, the reluctance to respond to deflationary pressures really has undermined the peg.

Had the SNB moved up the EUR/CHF peg to 130 or 140 six months ago then there would not have been the appreciation pressures on the CHF we have seen and the SNB would not have had to expand its balance sheet as much as have been the case.

The ‘regime’ I have outlined above is any many ways similar to Singapore’s monetary regime where the monetary authorities use the exchange rate rather than interest rates to implement monetary policy. In such a regime the central bank allows interest rates to be completely market determined and the central bank would have no policy interest rate.

This would have that clear advantage that there would never be any doubt what target the SNB would be trying to hit and how to hit it. This of course is contrary to the ‘old’ regime where the SNB effectively tried to have both an exchange rate target, an interest rate target and the inflation target. This inherent internal contradiction in the system I believe is the fundamental reason why SNB’s management felt it had to give it up.

Unfortunately the SNB so far has failed to put something else instead of the old regime and we now seem to be in a state of complete monetary policy discretion.

I hope that the SNB soon will realise that monetary policy should be rule-based and transparent. My suggestion above would be such a regime.

Update: I realise that I really should have dedicated this blog post to Irving Fischer, Lars E. O. Svensson, Bennett McCallum, Robert Hetzel and Michael Belongia. Their work on monetary and exchange policy greatly influenced the thinking in the post.

Did Bennett McCallum run the SNB for the last 20 years?

Which central bank has conducted monetary policy in the best way in the last five years? Among “major” central banks the answer in my view clearly would have to be the SNB – the Swiss central bank.

Any Market Monetarist would of course tell you that you should judge a central bank’s performance on it’s ability to deliver nominal stability – for example hitting an nominal GDP level target. However, for an small very open economy like the Swiss it might make sense to look at Nominal Gross Domestic Demand (NGDD).

This is Swiss NGDD over the past 20 years.

NGDD Switzerland

Notice here how fast the NGDD gap (the difference between the actual NGDD level and the trend) closed after the 2008 shock. Already in 2010 NGDD was brought back to the 1993-trend and has since then NGDD has been kept more or less on the 1993-trend path.

Officially the SNB is not targeting NGDD, but rather “price stability” defined as keeping inflation between 0 and 2%. This has been the official policy since 2000, but at least judging from the actually development the policy might as well have been a policy to keep NGDD on a 2-3% growth path. 

Bennett McCallum style monetary policy is the key to success

So why have the SNB been so successful?

My answer is that the SNB – knowingly or unknowingly – has followed Bennett McCallum’s advice on how central banks in small open economies should conduct monetary policy. Bennett has particularly done research that is relevant to understand how the SNB has been conducting monetary policy over the past 20 years.

First, of all Bennett is a pioneer of NGDP targeting and he was recommending NGDP targeting well-before anybody ever heard of Scott Summer or Market Monetarism.  A difference between Market Monetarists and Bennett’s position is that Market Monetarists generally recommend level targeting, while Bennett (generally) has been recommending growth targeting.

Second, Bennett has always forcefully argued that monetary policy is effective in terms of determining NGDP (or NGDD) also when interest rates are at zero and he has done a lot of work on optimal monetary policy rules at the Zero Lower Bound (See for example here). One obvious policy is quantitative easing. This is what Bennett stressed in his early work on NGDP growth targeting.  Hence, the so-called Mccallum rule is defined in terms the central bank controlling the money base to hit a given NGDP growth target. However, for small open economies Bennett has also done very interesting work on the use of the exchange rate as a monetary policy tool when interest rates are close to zero.

I earlier discussed what Bennett has called a MC rule. According to the MC rule the central bank will basically use interest rates as the key monetary policy rule. However, as the policy interest rate gets close to zero the central bank will start giving guidance on the exchange rate to change monetary conditions. In his models Bennett express the policy instrument (“Monetary Conditions”) as a combination of a weighted average of the nominal exchange rate and a monetary policy interest rate.

SNB’s McCallum rule

My position is that basically we can discribe SNB’s monetary policy over the past 20 years based on these two key McCallum insights – NGDP targeting and the use of a combination of interest rates and the exchange rate as the policy instrument.

To illustrate that I have estimated a simple OLS regression model for Swiss interest rates.

It turns out that it is very to easy to model SNB’s reaction function for the last 20 years. Hence, I can explain 85% of the variation in the Swiss 3-month LIBOR rate since 1996 with only two variables – the nominal effective exchange rate (NEER) and the NGDD gap (the difference between the actual level of Nominal Gross Domestic Demand and the trend level of NGDD). Both variables are expressed in natural logarithms (ln).

The graph below shows the actually 3-month LIBOR rate and the estimated rate.

SNB policy rule

As the graph shows the fit is quite good and account well for the ups and down in Swiss interest rates since 1996 (the model also works fairly well for an even longer period). It should be noted that I have done the model for purely illustrative purposes and I have not tested for causality or the stability of the coefficients in the model. However, overall I think the fit is so good that this is a pretty good account of actual Swiss monetary policy in the last 15-20 years.

I think it is especially notable that once interest rates basically hit zero in early 2010 the SNB initially started to intervene in the currency markets to keep the Swiss franc from strengthening and later – in September 2011 – the SNB moved to put a floor under EUR/CHF at 120 so to completely curb the strengthening of Swiss franc beyond that level. As a result the nominal exchange rate effectively has been flat since September 2011 (after an initial 10% devaluation) despite massive inflows to Switzerland in connection with the euro crisis and rate of expansion in the Swiss money supply has accelerated significantly.

Concluding, Swiss monetary policy has very much been conducted in the spirit of Bennett McCallum – the SNB has effectively targeted (the level of) Nominal Gross Domestic Demand and SNB has effectively used the exchange rate instrument to ease monetary conditions with interest rates at the Zero Lower Bound.

The result is that the Swiss economy only had a very short period of crisis in 2008-9 and the economy has recovered nicely since then. Unfortunately none of the other major central banks of the world have followed the advice from Bennett McCallum and as a result we are still stuck in crisis in both Europe and the US.

PS I am well-aware that the discuss above is a as-if discussion that this is what the SNB has actually said it was doing, but rather that it might as well been officially have had a McCallum set-up.  

PPS If one really wants to do proper econometric research on Swiss monetary policy I think one should run a VAR model on the 3-month LIBOR rate, the NGDD gap and NEER and all of the variables de-trended with a HP-filter. I will leave that to somebody with econometric skills and time than myself. But I doubt it would change much with the conclusions.

%d bloggers like this: