By David Llewellyn-Smith, founding publisher and former editor-in-chief of The Diplomat magazine, now the Asia Pacific’s leading geo-politics website. Originally published at MacroBusiness
After many years of sizable increases in central bank balance sheets (Figure 2), major changes in central bank purchases are approaching, as i) the Fed is due to announce balance sheet reduction in September and begin implementing it in October, ii) the ECB will probably announce a reduction in its net purchase volume in October (to begin in Jan-18) and will probably end its net asset purchases in H2- 2018. In addition, the BoJ net asset purchases have fallen and we expect them to continue to fall, even though the BoJ now pursues a yield-targeting policy. Overall, advanced economy (AE) net asset purchases are likely to fall from $100bn/month currently to roughly zero by end-2018 (Figure 1) and aggregate AE central bank balance sheets are likely to peak soon (in dollar terms around end- 2018 and relative to GDP in mid-2018).
This note discusses the trajectories of central bank asset purchases, their drivers, exit strategies, potential implications and risks. The change in central bank purchases is mostly due to a better economic backdrop in AEs – much lower unemployment rates and somewhat higher inflation – even though political reasons and technical factors also play a role. Central banks will tread carefully and the direct impact of global tapering on the real economy will likely be modest. But there is a material risk in our view, that major asset price corrections could be triggered by this global tapering (Figure 3).
If global tapering were to puncture asset bubbles, this could magnify the effects of global tapering on the real economy, by discouraging investment, lifting corporate defaults and leading households to retrench and could therefore be associated with a significant global economic slowdown, too.
Why should global tapering matter?
In our view, the direct effects of global tapering on global growth and inflation are likely to be modest. The overall impact of global tapering would, however, crucially depend on the financial market reaction. Major corrections in asset markets would likely cause a more severe slowdown in global growth, ie the causation would run from asset markets to the real economy.
Such corrections may be caused by global tapering or simply triggered by it. In the latter case, global tapering probably only determines the timing of the fallout, which would in any case have taken place.
Why would asset prices fall?
The footprint of central banks in financial markets is very large. The combined asset holdings of the five central banks with recent QE programmes (BoE, BoJ, ECB, Fed, Riksbank) amount to $11.5trn, or 26% of the outstanding values of those assets classes.
Our colleagues in Credit Strategy eloquently laid out the argument for why asset valuations could be threatened by global tapering, based on portfolio balance effects of these purchases and the view that many assets are overvalued currently.
The portfolio balance channel
The decline in CB net asset purchases will lead to a rise in net supply of the assets that central banks were previously buying, which would need to be absorbed by the private sector (Figure 10). The net supply of these assets to the private sector is due to rise from -$200bn in 2016 and 2017 to $1.2trn in 2018 and $1.9trn in 2019 (Figure 11). Within that, the net supply to the private sector of US Treasury securities is set to rise by more than $230bn between 2016 and 2018 (2017 issuance will probably be distorted by debt ceiling issues), and by about $170bn for US MBS. Eurozone net sovereign bond issuance to the private sector could rise most sharply (from admittedly very negative amounts) as the ECB phases out its APP.
Standard portfolio balance theory suggests that for a rise in net supply to be absorbed, prices of assets would have to fall (strictly, be lower than they would otherwise have been) and the more so, the less substitutable they are with other assets and the less elastic demand is to decreases in their price, both in the nearterm (which mostly depends on the market’s liquidity) and in the long-term.
Substitutabilities between different assets and demand elasticities of investors to changes in asset prices are key. The bulk of central bank holdings is in sovereign bonds – e.g. the Fed currently owns $2.5trn of US Treasury securities (18% of the outstanding), the ECB owns €1.8trn (23%) of Eurozone sovereign bonds and the BoJ ¥431trn of Japanese government securities (46%).
But we would expect the largest asset price moves to often be in other assets, notably those which are more risky and for which demand tends to be more volatile. US high-yield corporate debt, Eurozone periphery sovereign bonds, Eurozone corporate bonds, global equities, EM assets are among the asset classes in focus.
In some of these markets, central banks are also very present: for instance, the Fed owns 29% ($1.8trn) of agency MBS, the ECB 27% of Eurozone covered bonds and 15% of Italian sovereign bonds. For all of these net supplies to the private sector are also set to rise quite sharply, even though from deeply negative levels in the case of Eurozone covered bonds and Italian sovereign bonds most recently. The ECB also buys a significant share of the gross issuance of Eurozone IG nonfinancial corporate bonds. In others (e.g. equities or EM assets), the effect of central bank purchases are indirect.
One concern is that global tapering would lead to higher implied and realized volatilities of many asset prices (which have tended to be very low of late, at least in the equity market). Increases in these volatilities could turn out to be self-reinforcing, as rising realized volatility could induce investors to try to exit their positions, which in turn could lead to further price declines and reinforce higher volatilities.
Overall, we would therefore expect, to the extent that investor risk aversion rises and/or the economy weakens during the global tapering episodes, the prices of sovereign bonds may fall less or indeed rise (ie yields fall) as central banks retreat from those markets, due to safe-haven demand for these assets, as other asset prices fall (more). Like our colleagues in credit strategy, we stress that the drivers and the effects are potentially global.
While the supply to the rest of the market (i.e. excluding the central banks) of the securities central banks now hold would be rising under global tapering, the availability of central bank reserves would be declining, other things equal. However, we are not unduly worried about this aspect, including as there are significant ‘excess’ reserves currently, central banks and regulators are aware of these issues.
It is also worth noting that we do not expect global central banks to become significant net sellers of private sector assets for the time being, ie the stock of financial assets held by AE central banks will continue to rise for some time and only start to decline very slowly in 2019.
In general, we think that both stocks and flows matter for asset prices. Indeed, unless markets are completely myopic, portfolio balance theory implies that all current and anticipated future asset stocks (relative to the demand for these assets at given prices and yields, that is, relative to income or wealth), should matter for asset prices and the real economy. Because current flows and anticipated future flows drive anticipated future asset stocks, flows matter even in the ‘stock-oriented’ portfolio balance model. In very liquid, orderly markets, only unanticipated flows or changes in flows affect asset prices and yields today – through their impact on future anticipated asset stocks relative to the future demand for these assets at given yields and prices.
In practice, there are no perfectly liquid markets, and it is therefore likely that even fully anticipated reductions in central bank purchases (‘the flow effect’) could push down asset prices at least temporarily, as financial markets try to digest the rising net supplies. In liquid markets (US Treasuries, Bunds, JGBs etc.) we expect that much of that effect would be temporary and modest. This is not true for illiquid markets like those for Eurozone periphery sovereign debt and most EM securities.
Are asset markets overvalued?
As long as financial markets are orderly and relatively fairly valued, changes in relative asset supplies should only have a moderate effect on asset prices and through that on the real economy. It is, of course, very difficult to ascertain whether assets are fairly valued. But we are concerned that a number of key asset valuations appear rather stretched, according to a range of metrics. For instance, the value of the US stock market to GDP is at historically high levels and the cyclically-adjusted P/E ratios are similarly elevated (Figure 15). Moreover, we have gone for an unusually long period without a stock market correction (Figure 16).
If past, present and anticipated future central bank policies have been, at least in part, responsible for pushing asset prices away from their fundamentals, then the retreat of central banks from these policies would strengthen the case for expecting a pullback in asset prices – in that case, the retreat of central banks would cause the fall in asset prices.
But even if central banks were not a cause for their overvaluation, central bank pullback could be a drag for valuations. In that case, the bubble would burst eventually anyway, and global tapering would only affect the timing of it – it would be the trigger.
The timing of any asset price reaction is indeed another source of major uncertainty. With perfectly efficient markets, we would expect all financial markets to adjust as soon as new information becomes available. However, in fully efficient financial markets, relative asset demands and supplies – both stocks and flows – are not among the drivers of fundamental asset valuations and major asset bubbles are unlikely. If one believes that asset valuations can stray from their fundamental valuations (as they currently appear to), one may therefore also be sympathetic to the view that financial markets react only (or mostly) as purchase changes are realized and not just as new information gets released.
Good stuff, that. My own view is that the Great Taper will end the global business cycle, but not before we get one more march higher in DM asset values as bond prices rally around tapering into lowflation.
I wouldn’t be hanging around in EM assets to find out what effect it will have there henceforth.