Let us recall the four landmark dates of the ECB’s QE:
• 22 January 2015: an asset purchase programme is announced starting in March 2015 and ending in September 2016 at a monthly pace of €60bn per month. The essential component of the programme chiefly related to the purchase of Public Sector bonds (PSPP) but also included purchases of ABS (ABSPP) and Covered Bonds (CBPP3)
The ECB’s QE was characterised by size of the volume of purchases
Among other things, the ECB’s QE differed substantially from the Fed’s QE programmes in that the volume of sovereign bond purchases to date was much greater than net issuance of long-term securities (in the case of the Fed, its purchases never exceeded the net issuance of long-term securities). Over the twelve-month period ending in July 2017, the countries of the eurozone issued €124bn in long-term bonds while the Eurosystem purchased €806bn of such securities: net issuance of long-term securities after central bank purchases was therefore -€681bn – which is very negative. Each eurozone country found itself in a situation where the ECB’s purchases far exceeded the net issuance of long-term securities. The impact on the term premium was therefore very significant and widespread.
Although the central banks of the Eurosystem held very few government bonds prior to 2015, the percentage of long-term government bonds held by the Eurosystem today is over 25%. Since 2015, the speed of the rise in securities holdings by the Eurozone central banks is equivalent to that of the rise in the Bank of Japan’s holdings of JGBs. On the other hand, the situation was very different for the Fed’s QE since its three QE operations merely restored the Fed’s proportion of long-term Treasury securities holdings to about 25% of the market. The decision taken during the 20 September FOMC meeting to normalise the Fed’s balance sheet is expected to gradually reduce the Federal Reserve’s holdings of Treasury securities to unusually low levels.
Non-residents and European banks have reduced their share of holdings
The increase in the weight of sovereign bond holdings attributed to the Eurosystem has mainly occurred at the expense of non-residents and eurozone banks. In the light of negative interest rates and the depreciation of the euro which followed the announcement of QE in 2015, non-residents dramatically reduced their exposure to the European sovereign bond markets. The proportion of government bonds held by non-residents fell sharply in Germany, France and Italy following the start of the ECB’s QE in early 2015. However, the statistics currently available do not differentiate between non-residents belonging to the eurozone (for instance, French investors holding Italian debt) and non-residents not belonging to the eurozone (for instance, Asian investors holding Italian debt).
It is worth noting that the peak level of holdings of government bonds denominated in euros by European banks was reached in February 2015, that is, the month prior to the start of the ECB’s QE. In February 2015, European banks held €1.885 trillion in government bonds denominated in euros. This total had been no more than €1.566 trillion in August 2017, i.e. a drop of more than €318bn relative to February 2015. Note that European banks primarily reduced their holdings of bonds issued by their own countries while their holdings of sovereign bonds issued by other countries have remained relatively stable since early 2015. That being said, European banks still hold far more sovereign bonds than they did during the previous decade: the question of the link between the banks and governments will presumably be raised again in the future. As a side point, note that the decline in the holdings of govies by the banks remained relatively limited in Italy relative to other countries (Italian banks still hold €423bn in government bonds).
What was the impact on the market of the announcements relating to QE?
It is not easy to estimate the precise impact of QE on the yield curve and spreads for a number of reasons:
• The markets usually anticipate announcements related to QE and, unlike key interest rates, it is hard to measure what is already priced in by the markets in terms of QE-related announcements.
The few academic studies that have dealt with this subject have generally shown that announcements of unconventional programmes by the ECB have caused sovereign spreads in the eurozone to fall. Falagiarda and Reitz argue that communications about the SMP and OMT programmes substantially reduced government bond yield spreads (2015, “Announcements of ECB Unconventional Programs: Implications for the sovereign spreads of stressed euro area countries”). One of the ECB’s own working papers argues that the announcement of the PSPP resulted in 10-year yields declining by 30 to 50 basis points (depending on the approach) for the core countries of the eurozone and by roughly twice as much in countries such as Italy and Spain (2015, “Asset purchase programmes and financial markets: lessons from the euro area”).
The following charts position ECB announcements relative to changes in German bond yields and sovereign spreads. Several remarks are appropriate here:
• Sovereign yields and spreads had already fallen significantly in H2 2014 in anticipation of the announcement of QE. They continued to fall for a few months following the announcement of 22 January 2015.
• The announcement of the extension of QE on 8 December 2015 was followed by a sharp drop in bond yields but a widening of sovereign spreads.
• The announcement of the increase in the pace of QE (increase from €60bn to €80bn per month) of 10 March 2016 was followed by a further decline in long bond yields and contained the widening of spreads following the risk aversion peak in early 2016.
The announcement concerning the extension of QE at a reduced pace (return to €60bn per month) was followed by a pick-up in rates and a steepening of the German yield curve and interrupted the widening of sovereign spreads subsequent to the resignation of Matteo Renzi as Prime Minister of Italy.
Another view of the market impact: from the nominal quantity of securities held to the amount of duration held
In his speech of 2 October (“Maintaining price stability with unconventional monetary policy measures”), Peter Praet underscored that the impact of QE should not be reduced to the nominal amount of the securities held by the ECB: “It is important to emphasise that it is the amount of duration that we extract relative to the amount of duration that otherwise would be in the market that produces the impact [on interest rates]. In this respect, keeping the size of the APP portfolio constant at a certain nominal level does not necessarily safeguard a given amount of accommodation. By keeping the portfolio of assets acquired under the APP constant in nominal terms, it is difficult to prevent the ageing of the portfolio, i.e. its gradual loss of duration as the securities held in the portfolio mature. Although re-investments of the proceeds from principal payments of the maturing securities can offset this endogenous decay of the average portfolio duration, this substitution effect is generally not strong enough to maintain the average maturity of the portfolio[...] As time passes, the endogenous loss of duration in the central bank portfolio is bound to exert increasing upward pressure on the term premium.”
Each eurozone country found itself in a situation where the ECB's purchases far exceeded the net issuance of long-term securities
The shift from a thought process focused on the nominal level of securities held to one that is focused on amount of duration in the portfolio signals a real change and probably some form of learning on the part of the ECB. On 8 December 2016, in addition to the announcement of the extension of QE until December 2017 at a reduced pace, the ECB also announced:
1. the broadening of the maturity range of the Public Sector Purchase Programme (PSPP): from this point, the securities purchased must have a remaining maturity ranging from 1 to 31 years (the range was 2 to 31 years prior to this adjustment).
2. the removal of the yield constraint (prior to this adjustment, the yield to maturity had to be above the interest rate on the ECB’s deposit facility).
As a result, a very large number of short-term German bonds became eligible and the average maturity of German bond purchases fell from January 2017. This coincided with an upturn in bond yields in Germany. As a side point, we might mention that during the implementation of the Fed’s four main quantitative programmes (QE1, QE2, QE3 and Operation Twist), it was presumably Operation Twist (sale of bonds with short maturities to buy bonds with long maturities at a constant rate) that coincided the most with a long episode of weakness of US long-term yields.
In order to measure the impact on long-term interest rates of the Fed’s quantitative programmes and of the balance sheet shrinking, the Fed’s economists (“The Effect of the Federal Reserve’s Securities Holdings on Long-Term Interest Rates”, FEDS Notes, April 2017) combined the par amount of securities holdings and their maturities by calculating “total holdings of Treasury securities in terms of ten-year-equivalents”, i.e., by calculating the par amount of 10-year securities that would have the same weighted duration as the stock of Treasury securities held by the Fed.
The quantity of German securities held by the Eurosystem under the PSPP calculated in terms of “ten-year-equivalents” have topped out since the beginning of 2017even as the Eurosystem purchased €121bn in the first nine months of the year. It is probably no coincidence that this topping out of “ten-yearequivalent” Bund amounts coincided with a pick-up in German rates.
What should we expect from the evolution of Eurosystem holdings of Bunds calculated in 10-year equivalent?
By making the following assumptions (all of them being obviously debatable):
• the QE is extended for 9 months at €30 bn/month (as suggested by a Bloomberg piece on Oct. 13 “« ECB to Consider Cutting QE Purchases in Half Next Year »)
• the weight of the PSPP within the QE remains the same
• the reinvestments of maturing German securities purchased under the PSPP account for €5bn per month on average
• the net PSPP purchases and reinvestments of German securities occur at an average maturity of 6 years and half (the average maturity of the long-term securities issued by the German Treasury these last years has been 7 years and half but under the scenario considered here, the gross issuance will only represent one third of the gross purchases of German purchases from the ECB and the Bundesbank and the latter will have to buy shorter-dated bond that have not been purchased for the time being).
We come to the conclusion that the Eurosystem holdings of Bunds calculated in 10-year equivalent should virtually stagnate in 2018. In the case of lower reinvestments or lower net purchases (if the ECB opts for a 9 months QE extension at 20 bn per month) or a lower share of the PSPP in the QE or even lower maturities for gross purchases would even lead to a slight decline of the Eurosystem holdings of Bunds calculated in 10-year equivalent. As underlined by Peter Praet, we will have to take more and more the ageing of the portfolio into account.
“As time passes, the endogenous loss of duration in the central bank portfolio is bound to exert increasing upward pressure on the term premium”.
It is likely that the ECB’s QE will end in 2018. By its magnitude, it will have weighed much more than the euro sovereign bond market than the Fed’s QE operations, with net purchases significantly exceeding net issuance of long-term govies. The share of euro long-term sovereign bonds held by the Eurosystem increased from a few % in early 2015 to more than 25% today. Non-residents and European banks have reduced their positions in this market, although sovereign bond holdings by euro area banks remain large. The strength of the ECB’s QE has weighed heavily on the German yield curve, bringing the 10-year German yield to a historical low of -0.18% in July 2016, but there will be upward pressure on the term premium from the effects of the ageing of the ECB’s securities portfolio and from the signal sent that rate hikes in the coming years are conceivable again.