Central Banks: to shrink or not to shrink

Should central banks shrink their swollen balance sheets? And, if they do, how should it be done and how far should it go? As pointed out in our previous essay (“Central banking: edging closer to normality”, 11.03.17) these are questions much debated among central banks.

Or, rather, the second of the two questions is debated. As for the first one, the decision generally seems now to have been made. And the collective wisdom is that yes, central banks should eventually shrink their balance sheets. The Fed is actively discussing the modalities, the Bank of England has said it will begin to do so once Bank Rate has reached 2%, ECB spokesmen (and Euro Area politicians) are openly venting the topic and the Riksbank has made it clear that it will retain some, but not all of its purchased assets. Only in Japan is the central bank actively discouraging talk of slimming, presumably because it has not yet reached its 2% inflation target, nor is it likely to do so for some years at least.

We agree: central banks should shrink their balance sheets. There are two reasons for this. First, we have to assume that quantitative easing (QE) will henceforth be a routine part of central bank policy toolbox. But, unless balance sheets now shrink, central banks will enter the next downturn (and one day there will be a downturn) with already swollen balance sheets. Moreover, and perhaps more importantly, if they already own a large part of the government bond market, what should any further purchases comprise?  If they buy more government bonds, they risk ending up with a dominating position in the sector and a poorly functioning market, with perhaps unknown consequences.  But if not, they have to enter other asset classes, eg REITs, ETFs or equities, which have hitherto largely been seen as inappropriate for the authorities to hold.

Second, there is the practical consideration that if central banks do not slim their balance sheets, banks will retain large excess reserves. This makes them to an extent independent of the central bank’s policy interest rate, since these reserves can be lent in the inter-bank market.

The central bank is not without counter-measures here:  it can control the level of reserves by paying interest on excess reserves, or by temporarily removing them through reverse repos, or through higher interest rate frozen accounts (called the Term Deposit Facility in the USA). Indeed the Fed is currently using all three of those options. The problem with all three is that it keeps the reserves there; and that it becomes costly. The higher you raise the policy interest rate, the more you have to pay banks to keep their excess reserves undeployed. This can eventually become politically sensitive – banks are not necessarily the most popular institutions in the world, and few people will understand why it is worth paying them billions of dollars/pounds/euros of taxpayers’ money in order not to lend out their reserves.

That being said, it should be noted that shrinking their balance sheets is a double step into the unknown for central banks. In the first place, large central bank balance sheets are not new.  But, previously they have tended to occur in wartime, to finance the war effort. Afterwards, the balance sheet is left unchanged, and while the economy grows (through one or both of economic recovery and, as after World War Two, via inflation), it shrinks in relative terms. Now, we are talking about actively shrinking the balance sheets of several large central banks in a time of very low inflation and limited growth.

Second, as noted in our previous essay, QE was a step into the unknown for central banks. Something had to be done, this was something, it was done; the urgency of the situation overcame any concerns about the risks. But in a likewise fashion, there is also a lot of uncertainty about what its opposite, quantitative tightening (QT), will do. Logically, if QE lead to lower bond yields, an injection of liquidity in the economy and a reallocation of assets from bonds to equities, then QT should have the opposite effect. But that assumes a symmetry between the two processes, something which is by no means certain.  And this time there is no similar sense of urgency or crisis to stiffen central bank sinews into action – action which as we have observed may have unexpected consequences and entail unknown risks.

All this being said, it is clear that at least the Fed intends to begin slimming its balance sheet, most likely in early 2018, although the process could begin in late 2017 as well. At the moment, much of the discussion concerns whether the Fed should gradually cease reinvesting maturing assets; or cease investing all maturing assets; or change the composition of its assets; or sell some assets; or even  sell all its QE assets (though this last isn’t really being discussed).

In order to decide what to do, the authorities should first consider how the process impacts the economy. The problem here is that nowadays, apart from the ECB, no major central bank ascribes any importance to money – the monetarism of the 1980s and 1990s is no longer in fashion, and it only survives in Frankfurt courtesy of the Bundesbank’s dogged adherence to it – yet it is exactly through money that QT initially works.

Each time the Fed presents a maturing bond to an issuer, that issuer must pay the nominal value of the bond. That means drawing down the issuer’s bank account and transferring the money to the Fed. The issuer’s bank account is money; but funds held by the Fed are by definition not part of money supply. So, for each bond maturing and presented by the Fed for redemption, the stock of broad money will contract. Moreover, this also means that the banking system balance sheet will contract, with potentially negative impacts on the economy.

This argues for caution in moving ahead. The best option for the Fed would therefore be to gradually cease reinvesting maturing assets, carefully gauging the impact of this policy on activity.

And there is another reason to move carefully. The Fed has made it clear that it wishes to restore primacy of monetary policy to the Fed funds rate. Based on studies by the Federal Reserve Bank of San Francisco, ceasing all reinvestment of maturing assets over the next two years would be equivalent to raising the Fed funds rate by 155 basis points. On the Fed’s own projections (the ‘dot plot’), the Fed expects to raise the Fed funds rate by 162 basis points over the next two years. Having a similar-sized impact from QT hardly amounts to restoring primacy to the policy interest rate.

Of course, much also depends on by how much the Fed – and other central banks – intend their balance sheets to shrink. Or, put differently, what is the new target size? The Fed’s balance sheet currently stands at $4.46tn, roughly speaking equivalent to 23% of GDP. In August 2008, before the Great Recession, it stood at just short of $933bn or 6.3% of GDP.

Once again, there are a wide range of different options. The Fed could

(a) return to the pre-crisis size in dollar terms – shrinking the balance sheet by $3.49tn

(b) return to the pre-crisis size as a percentage of GDP – shrinking the balance sheet by $3.28tn

(c) aim for the size where the balance sheet would have been if it had continued to grow at its pre-crisis pace – shrinking the balance sheet by $3.13tn (assuming a growth rate equivalent to the average of the last five pre-crisis years)

(d) aim for a size that takes into account the growth of cash in circulation (cash being a liability of the Fed) – shrinking the balance sheet by $2.7tn

(e) get rid of all excess bank reserves – shrinking the balance sheet by $2tn

(f) aim for another size, where the Fed retains some assets purchased under QE

These sizes vary greatly. But they are all, even option (e), substantial – by way of comparison, the U.S. commercial banking system balance sheet currently totals $16.2tn, of which loans and leases in bank credit total $9.1tn. Even the smallest reduction discussed above would thus have a significant negative impact on the banking system balance sheet; and the easiest way for banks to accommodate that is to shrink the amount of outstanding credit.

A much more likely development is therefore option (f), i.e., retaining some of the assets purchased under QE. This also does seem to be what at least some Fed spokesmen are talking about. The presidents of the New York and San Francisco Feds have talked of a four- or five-year period to complete the balance sheet shrinkage. Assuming only ceasing reinvestment over this period, this would (given the maturity profile of the Fed’s assets) amount to shrinking the balance sheet by $1.2-1.4tn.

This leaves us with the mostly likely development being that the Fed will at some stage this year announce the phasing-in of an end to reinvestment of maturing assets, and that ultimately its balance sheet over the next 2-3 years will shrink by at least $1.2tn. However, there are so many unknowns (both known and unknown) in this process that a cautious process, carefully gauging the impact as they go, is devoutly to be hoped for.

The great experiment of QE is about to be replaced by the potentially even riskier one of QT.  The Fed, as befits the world’s leading central bank and controller of the global reserve currency, will be in the lead.  The scope for mistakes is substantial.  The main consolation for the Fed as it undertakes this delicate manoeuvre is that as it does so, it will not be short of advice from other central bankers, the markets and no doubt the shy, retiring and taciturn current incumbent in the White House.


Gabriel Stein is Managing Director, Developed Markets Research at 4Cast RGE