• As global economies recover, central banks are shifting towards ‘normalising’ their bloated balance sheets. Given the scale of the previous build-up, many investors fear that this could be hugely disruptive to markets.
• Yet, this is unlikely. The Fed’s balance sheet is unlikely to return to its pre-crisis size and the path towards its steady-state size is likely to be a gradual one. The move is well anticipated by market participants and the Fed can use both rates and its balance sheet size to calibrate overall monetary conditions.
• What is more, normalisation will (only) take place against a backdrop of economic recovery and the Fed will deliberately choose to remain “behind the curve” (i.e. tolerate an overshoot of the inflation target).
• On a positive note, the release of collateral into the market is set to expand private sector balance sheets and lubricate the financial system.

Ever since the US Fed and the ECB (and several other central banks) ventured beyond the ‘zero interest rate limit’ and engaged in Quantitative Easing (QE), investors and commentators alike feared that bloated balance sheets would lead to a surge in inflation. Persistent negative output gaps and declining commodity prices on the back of excess supply ensured that this did not come to pass and the world’s central banks found themselves battling a slide into deflation instead. Now, the world economy is finally on the mend and markets are looking forward to ‘balance sheet normalization’, again associating extra-large balance sheets with long-term inflation risks.

Leaving aside the issue of whether the current soft patch in the US economy represents a pothole on the road to recovery or a more worrisome sinkhole, the question of what the adequate balance sheet size should be in a post-crisis economy is an important one. What is more, if the answer is along the lines of “smaller than now”, a host of other related questions emerge.

While investors continue to enjoy, indeed clamour, for the ‘central bank put’, central banks themselves are happy to ‘remain behind the curve’ as a form of insurance against a relapse into recession. On the other hand, they also share a concern about 1) systemic distortions which persistent, extraordinarily low interest rates (across the curve) may introduce into the financial systems and 2) re-establishing their conventional policy toolkit in time for the next recession. The Fed is furthest along the path towards ‘normalization’, having raised interest rates three times from the low so far. In 2014, it stated that balance sheet reduction would only occur once interest rate normalization was “well under way”. This was understood to mean interest rates between 1.0% and 1.5%. With the Fed Funds rate currently at 1.0%, this leads market participants to anticipate the balance sheet unwind to begin in early-to-mid 2018.

What is the appropriate size of an economy’s central bank balance sheet? Strict economic theory would argue that it doesn’t matter, as money is simply a ’veil’. Relative price changes will determine the exchange rate but that’s about it. Yet, the increase of the Fed’s balance sheet from a pre-crisis $900 bn to a high of $4.5 trn at end-2015 arguably had tangible economic effects, confirmed by several studies.

Former Governor Ben Bernanke recently argued that the Fed balance sheet should be at about $4 trn by 2027. Fed staff estimated that it could be reduced to $2.7 trn by 2025. This is close to what many market practitioners assume. But should it be reduced at all? And if so, by how much and on what basis?

One way to approach the question is by looking at the liability side of the Fed’s balance sheet and proceeding by backwards induction. The Fed’s  liabilities consist chiefly of $1.5 trn of notes and coins in circulation and $2.3 in bank reserves (forming together ‘base money’), plus government deposits and some other items. The large bank reserve item is the counterpart to the Fed’s asset purchases. Its expansion from near-zero prior to the crisis has had several advantages. For one, it appears to have fostered a more stable interbank interest rate environment. Second, this implies that the banking system is less reliant on the Fed for intra-day credit, reducing the latter’s credit risk exposure. And finally, maintaining a large pool of excess reserves allows the Fed to alter liquidity in the system through the Interest On Excess Reserves (IOER) rate without affecting the Federal Funds benchmark rate. To safeguard these benefits, some $500 bn of bank reserves will likely be needed on a permanent basis (the Fed’s projections assume $100-300 bn, Bernanke has argued for $1 trn). Restoring government deposits to $300 bn (currently depleted because of the debt ceiling) and assuming currency growth at the rate of nominal GDP growth, implies a normative goal of $2.5 trn for the Fed’s liabilities, all else equal. Exact timing and balance sheet size will vary with the path of interest rates (as they affect the pace of MBS prepayments), the rate of nominal growth and last but not least, the asset mix the Fed chooses during the phase-out period and the pace at which it stops reinvesting proceeds from maturing assets. Variations in these parameters can be easily simulated, but for most reasonable assumptions, they shift the desired endpoint by just 1-2 years.

The Fed’s $4.5 trn assets consistsof $2.5 trn Treasuries, $1.8 trn MBS holdings plus some other small items. The majority of USTs have a maturity of 1-5 years ($425 bn mature in 2018 and $352bn in 2019), but since the Fed has stated its intention to return to an all-Treasury balance sheet, it will likely let its MBS holdings roll off first. However, at the point at which its liabilities approach the desired size (as discussed previously), the Fed will still hold some MBS. In order to maintain the size of its assets, it is thus likely to continue to let MBS roll off, while beginning to buy Treasuries to compensate.

Are there any constraints on the pace of asset disposals? A Fed study has shown that the cumulative effect of QE programs was to reduce 10-year bond yields by some 120 bps. It is thus feared that selling Treasuries or MBS too rapidly could cause market dislocation or undesirable change in rates.  For her part, Janet Yellen has suggested that the expectation of balance sheet normalisation has already increased bond yields by 15 bps (said to be equivalent to two 25 basis bps increases in the Fed Funds rate).

The market impact of the Fed’s balance sheet unwind is thus likely to be limited because 1) its terminal size is likely to be greater than the pre-crisis level and 2) the path the Fed takes towards this point is likely to be very gradual. Indeed, in its latest minutes, the FOMC said that it would announce a set of gradually increasing caps on the amounts of securities that would be allowed to run off every month and that “only the amounts of securities repayments that exceeded the caps would be reinvested each month”. These limits would be set at a low level initially and be raised every three months.

In addition, there are other reasons why the unwind will likely have a more muted market impact than the initial build-up. For one, it is well anticipated by markets and takes place against the backdrop of a sustained recovery, which in turn underpins market sentiment. Second, while the US recovery is more mature than elsewhere, other central banks such as the ECB, are also approaching an inflection point in their monetary stance, thus limiting the extent of policy divergence. Third, the Fed will have two policy tools at its disposal to fine-tune monetary policy: it can use its rate tool to counteract any adverse effect the balance sheet reduction may have and overall monetary conditions will be determined by the combination of rates and balance sheet size. The Fed can thus simultaneously aim to control both ends of the yield curve.

Finally, there is a different way to look at the issue: balance sheet reduction releases collateral back into the market system which is constantly reused in a process that resembles that of money creation. As new regulations require banks to hold more “high quality liquid assets”, the availability of good collateral has become a key constraint for financial intermediation. Like excess reserves, which also count as high quality liquid assets, good collateral is often reused two to three times, increasing bank balance sheet size and lubricating the financial system. Together with managing the IOER rate, this allows the Fed to enhance liquidity in the financial system at the same time as it reduces its balance sheet.

On balance, Fed balance sheet adjustment is unlikely to be disruptive, if only by design. What is more, it doesn’t take place in a vacuum but against of backdrop of other economic and political developments, which may require an adjustment of risk premia. Its effects may thus be difficult to disentangle from other yield-enhancing developments. The Fed will likely use a combination of rate adjustments (Fed Funds, IOER) and balance sheet management (across maturities, instruments etc) to counter any threat to market stability.

Note: What happens mechanically when Treasuries come due is that the Treasury pays the Fed the par amount from its deposit account. The value of assets in the form of Treasuries thus decrease sand the value of liabilities in the form of government deposits goes down by an equivalent amount. At this point, there is no change in bank reserves yet and no direct effect on market yields. In practice, the Treasury will want to maintain a certain balance in its deposit account in order to meet upcoming payments. As a result, it will thus issue debt to the private sector and the purchasing bank will see its deposit account debited, while that of the government will be credited with the same amount. All else equal, the increased supply of Treasury debt to the market would drive up yields. The end result is an unchanged position in the government deposit account, a reduction in the amount of Treasuries held on the asset side and an equivalent reduction in the amount of bank reserves on the liabilities side. Note that if the Fed instead of letting its Treasuries roll off actually sold them, a different mechanism would be in effect. In that case, the asset reduction would be directly matched by a reduction in bank reserves on the liability side.