VoxEU Column EU policies Monetary Policy

Restarting asset purchases in the euro area: Lessons from €2 trillion of ECB purchases

Recent economic performance in the euro area has once again raised the possibility of the ECB conducting asset purchases. This column sorts security-level portfolio holdings data by investor type and across countries in the euro area to study portfolio rebalancing during the ECB purchase programme from 2015-17. There was a material difference in the impact on investors by geography – with foreign investors selling more than half of purchases.

With inflation in the euro area slowing to just 1% in July 2019 and growing concerns about an economic slowdown, the ECB is reportedly considering cutting the deposit facility rate and launching a new bond-buying programme. In this column, we summarise several lessons based on the €2 trillion of purchases in the euro area from March 2015 to December 2017 that were part of the previous asset purchase programme. 

A large and growing literature studies the implications of the unconventional monetary policies that have been implemented since the Global Crisis as interest rates hit zero in many developed economies. By purchasing long-dated assets, such as government bonds, corporate bonds, and mortgage-backed securities, central banks may have been able to lower the borrowing costs of firms and households in an attempt to spur economic growth and inflation.

To understand the impact of unconventional monetary policies on financial markets, it is common practice to measure the impact on different asset prices in a short timeframe around key policy announcement dates (Krishnamurthy and Vissing-Jorgensen 2011). However, typically due to a lack of detailed portfolio holdings data, little is known about which investors rebalance their portfolios, whether risks got concentrated in certain sectors as a result, and how much different investors benefit from the programme.

In Koijen et al. (2019), we use new data on security-level portfolio holdings for all major investor sectors, including banks, insurance companies and pension funds, and mutual funds, and for all countries in the euro area from 2013Q4 to 2017Q4. For each sector in a particular country, we observe the quarterly holdings of government bonds, corporate bonds, asset-backed securities (including covered bonds), and equities, both in and outside of the euro area. We link these portfolio holdings to detailed data on prices and security characteristics. 

We also observe the ECB’s security-level monetary policy holdings and purchases. We use these data to measure which investors sell to the ECB, how investors’ portfolio rebalancing impacts the distribution of financial risk exposures across investors and geographies, how it impacts government yields, and which investors (again differentiated by investor type and geography) experience the largest appreciation of their asset portfolio.

Table 1 summarises the rebalancing activity across the different sectors and geographies. We group countries into two regions that we classify as vulnerable countries (Italy, Spain, Portugal, Greece, Cyprus, and Ireland) and non-vulnerable countries (all others), following Altavilla et al. (2017), but our analysis uses more granular country-level data. We report the changes in portfolio holdings across different asset classes – namely government bonds, corporate bonds, asset-backed securities and covered bonds, equities, and securities issued outside of the euro area or in foreign currency.

Four key important facts emerge from the table. 

  • First, the sector selling most to the ECB is the foreign sector – that is, investors domiciled outside of the euro area.
  • Second, the foreign sector does not reinvest the proceeds in the euro area. 
  • Third, of all sectors in the euro area, the banking sector sells the largest amount.
  • Fourth, long-term investors, such as insurance companies and pension funds, in fact purchase government bonds and thereby amplify the impact of the purchase programme.

The first fact implies that the demand elasticity of the foreign sector is particularly important for understanding any price effects. After all, what matters for the price response of purchase programmes is the size-weighted average demand elasticity across investors. Hence, if a non-trivial fraction of all bonds outstanding is held by the foreign sector, the price effects are more muted compared to an economy in which all bonds are held by domestic investors. Economically, it seems reasonable that the demand of domestic investors is more inelastic as they are typically funded using euro-denominated liabilities. In the presence of long-duration liabilities, it may be optimal to purchase bonds as yields fall to close the duration gap (Domanski at al. 2017).

However, the foreign holdings are sufficiently large that the rate at which they accommodate the ECB purchases is stable during our sample. As there is nothing that the ECB can do to target domestic investors (as they trade with foreign investors in centralised markets), our insights are important to understanding the price effects. In particular, if the foreign sector runs out of bonds to sell, the slope of the aggregate demand curve steepens, and price effects will be larger.

Table 1 Cumulative rebalancing from 2015Q2 to 2017Q4 by asset category (billions of euros)

Note: The holdings are the euro-area government debt holdings as measured in terms of market value as of 2015Q2. The share sold is the rebalancing in government debt relative to the holdings (in percentage points). The top panel reports the rebalancing for investor sectors in non-vulnerable countries and the second panel for investors in vulnerable countries. The third panel reports the rebalancing of the foreign sector and the ECB. The bottom panel reports net issuances. The flows are reported in billions of euros. The classification of vulnerable and non-vulnerable countries follows Altavilla et al. (2017).

The second fact, i.e. that the foreign sector, and the other sectors, do not reinvest large amounts in other assets in the euro area, suggests that the spillover to other asset classes in the euro area may be limited. 

We further explore whether the foreign sector sells differentially across vulnerable and non-vulnerable countries in Table 2. We find that the foreign sector sells approximately at the same rate across vulnerable and non-vulnerable countries. More broadly, in Koijen et al. (2019) we explore in detail whether different financial market risks (duration, sovereign credit, corporate credit, and equity risk) get concentrated in some sectors during the programme, but we do not find evidence of increased risk concentration.

Table 2 Cumulative rebalancing from 2015Q2 to 2017Q4 by asset category (billions of euros) for the foreign sector 

Note: The holdings pre-QE are measured in terms of the average market value from 2013Q4 to 2014Q4. The share sold is the rebalancing in government debt relative to the holdings (in percentage points). The classification of vulnerable and non-vulnerable countries follows Altavilla et al. (2017).

Next, we estimate the impact of the programme on government bonds yields in the euro area. As mentioned before, the typical approach uses event studies around key policy dates. A growing literature, however, suggests that prices may be slow to adjust fully to monetary policy surprises (e.g. Brooks et al. 2019, and Hanson et al. 2018). We instead use a low-frequency instrumental-variables estimator. In particular, the ECB purchases bonds across countries according to the capital key, which assigns equal weight to a country’s population and GDP shares. Under the assumption that population and the level of GDP are unrelated to the economic conditions that prompted the purchase programme, we can isolate an exogenous component to purchases that we use to identify the price impact. The yield impact averages to -47 basis points, ranging from -28 basis points to -57 basis points across countries.

We use this yield impact to estimate how much different investors gained based on their portfolio holdings in Table 3. The total valuation effect equals €377 billion, of which €179 billion is earned by investors in non-vulnerable countries, €67 billion by investors in vulnerable countries, and the remaining €131 billion by investors outside of the euro area.

Table 3 Bond portfolios by sector and country

Note: The table reports the size of the euro area fixed income portfolio (categories 1 to 5, 2015Q1) the quantity of duration risk and the duration of the bond portfolio held by each sector by country group. We compute the impact of a shock of 47 basis points on the portfolio (in billions of euros).

For banks, mutual funds, and insurance companies and pension funds in non-vulnerable countries, the duration of their portfolios exceeds the duration of the same institutions in vulnerable countries. These institutions in non-vulnerable countries, therefore, experience larger valuation effects as a result of the PSPP-induced decline in yields. Among all institutions in each region, insurance companies and pension funds experience the largest benefit due to the long-duration assets that they hold.

We conclude with two comments to put these results into perspective. First, as we do not have data on each sectors’ liabilities, we cannot estimate the overall impact on funding positions of intermediaries. It may well be the case, in particular for long-term investors, that the value of their liabilities increased by a similar, or even larger, amount. Second, there may be long-term costs to the programme. Foreign investors sell their securities at higher prices and therefore realise the capital gain. Investors in the euro area, by contrast, largely hold on to their assets until maturity and their long-term return (assuming that no governments default) is unaffected. Moreover, they subsequently reinvest their principal against lower yields going forward. Our sample does not allow us to assess the long-term costs and benefits, but the worsened investment opportunities for domestic investors is a potentially important cost. 

Authors’ note: The views expressed here are those of the authors and do not necessarily represent the views of the Banque de France or the Eurosystem.


Altavilla, C, M Pagano, and S Simonelli (2017), “Bank Exposures and Sovereign Stress Transmission”, Review of Finance, 21 (6), 2103-2139.

Brooks, J, M Katz, and H Lustig (2019), “Post-FOMC announcement drift in US bond markets”, working paper. 

Domanski, D, H S Shin, and V Sushko (2017), “The Hunt for Duration: Not Waving but Drowning?”, IMF Economic Review, 65 (1), 113-153.

Hanson, S G, D O Lucca, and J H Wright (2018), “The excess sensitivity of long-term rates: A tale of two frequencies”, FRB of NY Staff Report no. 810. 

Krishnamurthy, A, and A Vissing-Jørgensen (2011), “The Effects of Quantitative Easing on Interest Rates: Channels and Implications for Policy”, Brookings Papers on Economic Activity, 2, 215-265.

Koijen, R S J, F Koulischer, B Nguyen, and M Yogo (2019), “Inspecting the mechanism of quantitative easing in the euro area”, Banque de France Working Paper no. 601.

1,680 Reads