The end of QE leaves a very big hole.
Central banks routinely deny that there is a connection between quantitative easing (‘QE’) and:
- All time high bond and credit valuations;
- Close to all time high equity valuations and ownership rates;
- All time high corporate, consumer and government leverage;
- All time low realised and implied volatility; and
- Record short futures positioning in the VIX.
If there truly is no connection then it’s been a mighty big coincidence, to say the least. In our view, it is much more likely that since 2008, as central banks globally have added USD15 trillion of assets to their balance sheets they have forced the private sector off the “risk-free” bond curve and into exotic sectors such as EM debt and DM High Yield, and even into short volatility and long momentum positions.
The Size of the Hole
None of this can be proven by the data. What can instead be demonstrated, is the scale of the hole that the private sector will have to fill. We have looked at the ten largest central banks’ balance sheets and developed estimations for how they might develop out to the end of next year. Figure 1 shows what the chart looks like.
The deceleration in asset growth is projected to start now with the Federal Reserve’s (the Fed) balance sheet contraction beginning this very month. The Fed, the European Central Bank (ECB) and Bank Of Japan’s (BOJ) balance sheets are forecast to decline very gradually next year, on the assumption that the Fed sticks to its stated plan and the Eurozone starts to taper purchases from January 2018 and stops all new asset purchases by the end of the year. But since the aggregate balance sheet is hardly shrinking rapidly, there’s no problem, right?
Well, we don’t think it’s that simple. If the central banks aren’t growing their balance sheets but governments are still running deficits and growing the total bond issuance outstanding, then someone else has to step in to buy the net new issuance from now on. We have to evaluate three things – the scale of the private sector step up, the room that the various private sector actors have to step up, and indeed their willingness to do so.
Figure 1. Top 3 Central Bank Balance Sheets Declining Very Gradually in our Model
Source: Central Bank financial statements, IMF, Man GLG estimates; as of October 2017. We factor in the Fed’s stated intention regarding its portfolio run-off, and make estimations based on recent trends and central bank statements for the others. Black vertical line indicates the start of the estimation period. Further detail available on request.
Figure 2 shows our estimate of the scale of assets that the private sector will have to step up to buy. The dark blue bars on the left are the monthly net asset purchases by the top ten global central banks so far this year. The light blue bars are the forecasts on the basis already described – so the balance sheets grow but barely. The dotted black line is the year to date run rate of monthly asset purchases. The grey bars are the monthly amount the private sector will have to buy in order to fill the hole left by the retreat of the CBs. Finally, the red line is the cumulative total of the grey bars – and this is the point of the chart – showing that between now, October 2017, and the end of 2018 it is estimated that the private sector will have to buy USD2.5 trillion of bonds that it otherwise could have spent buying more exotic products.
We should be clear that our concern centres on the asset side of the central bank balance sheet rather than the liability side. To appreciate the distinction we have to quickly re-cap how QE works. When the central bank buys government bonds it adds an asset (the bond) and to pay for it credits a liability account – at the Fed it’s called “Other deposits held by depository institutions” – which are bank reserves. Bank reserves are not money – they are excluded from most definitions of money, in particular they are excluded from M0, M1, M2, M3 and MZM. In fact they are only really central bank money, in that they form the monetary base. This may appear an arcane distinction but as Figure 3 shows it’s very relevant when we consider what causes expansion and contraction of the money supply.
Figure 2. Monthly Changes in Global Aggregate Central Bank Balance Sheets
Source: Central Bank financial statements, IMF, Man GLG estimates; as per Figure 1 and as of October 2017. Light blue bars indicate the start of the estimation period, which begins in September due to delays in data releases.
As Figure 3 shows, Federal Reserve assets grew by a factor of 5 during the QE process, but M2 (money) was completely unaffected, growing just in line with nominal GDP. We can therefore state clearly that QE is not money printing, it is reserve printing. And this may explain why central bankers generally appear so relaxed about reducing excess reserves in the system, because they know that unless they get the signalling wrong it will not reduce the money supply, just as adding excess reserves did not expand the money supply. So they can shrink the liability side of the balance sheet with impunity up to the point that there are no or very few excess reserves. In the Fed’s case, fully USD2.1 trillion of the USD2.2 trillion of bank reserve liabilities are excess reserves – so theoretically they could shrink their balance sheet by this much without affecting banks’ ability to lend and therefore the money stock.
However, what is not clear to us is how central banks can get comfortable with the shrinkage of the asset side of the balance sheet. The whole reason QE was supposed to work, and in our view did work, was that it forced changes to the private sector’s asset mix – via the so-called portfolio balance channel. What is clear is that assuming a) the stock of bonds in issue continues to grow and b) the Fed continues its policy of not fully reinvesting proceeds of maturing bonds, then c) the asset composition of the private sector is going to have to change to incorporate more “risk-free” assets.
Now USD2.5 trillion is a lot of money to find, even for the banking system in aggregate, but it is absolutely possible that the private sector does step up where the CBs step back. To take one likely candidate, the US banks could easily increase their treasury holdings. Figure 4 shows a snapshot of their balance sheets before and after QE.
Figure 3. QE is Not Printing Money
Source: Federal Reserve, Man GLG calculations; as of October 2017.
Figure 4. Table of the Assets and Liabilities of US Commercial Banks
|May 07||Sep 17||∆|
|Treasury and agency securities||2,272||2,495||223|
|of which MBS||1,086||1,798||712|
|of which non-MBS (i.e. TREASURIES)||1,186||697||-489|
|Real estate loans||3,373||4,241||868|
|Note: treasuries as % of Total Assets||14%||4%|
Source: Federal Reserve H8; as at end September 2017.
The only part of the balance sheet that shrank was the banks’ holdings of US Treasuries. Their share of the overall balance sheet (assets) accordingly fell from 14% to under 4%. On the assumption that they decided or could be prevailed upon to take their treasuries weight back up to say 15%, they would have to buy USD1.6trillion of bonds at today’s balance sheet size, and more if we factor in some balance sheet growth over the next couple of years. We would argue that this process is not just possible, but is likely in the event that the government sector needs to repress the private sector into owning the debt. And there are of course many other potential buyers of newly minted treasuries in other markets – the Chinese and Japanese institutions between them were majority owners of Treasuries for the two years after the GFC, for example (see Figure 5).
Figure 5. Falling Chinese and Japanese Share of US Treasury Holdings
Source: Bloomberg, Man GLG calculations; as of October 2017.
Our key point remains: QE worked to reduce risk premia by introducing a massive new buyer of risk-free assets. As that buyer steps away we have little doubt that risk-free assets the Fed is not buying will be bought by the private sector, but we wonder what effect this new allocation will have on credit. And to be clear, we doubt it’s benign. But again, that’s for later.