Large-scale asset purchase programmes have become an integral part of the set of unconventional tools central banks use to achieve their domestic objectives.
When faced with negative output gaps and policy interest rates approaching their eﬀective lower bound, these programmes have given central banks an essential tool to continue providing additional monetary accommodation, and prevent their domestic economies from falling into a deﬂationary trap.
Yet, in a world of integrated ﬁnancial markets, purchase programmes do not only have domestic eﬀects. This has been recognised by policymakers in the past. For example, there has been a lively debate about the eﬀects of the Federal Reserve’s purchase programmes on ﬁnancing conditions in emerging market economies. However, less attention has been paid to this aspect in the case of the euro area – that is, the international spillovers of the ECB’s asset purchase programme (APP).
This is what I would like to discuss in my remarks today. Speciﬁcally, I plan to cover two topics. First, I will review what we know about the APP’s eﬀect on net capital ﬂows out of the euro area. And second, I will look at whether these capital ﬂows have depressed exchange rates and, as a result, diverted demand away from other economies.
My main message is that the ECB’s APP does appear to have triggered substantial capital ﬂows across borders. But it is far from clear whether this explains the depreciation in the eﬀective exchange rate of the euro. In fact, asset purchases aﬀect the exchange rate in broadly the same way as conventional monetary policy –through expectations of interest rate diﬀerentials.
Moreover, while unconventional monetary policies may drive down exchange rates, it does not make them a zero-sum game. The evidence suggests that the demand-boosting eﬀects dominate – that is, the spillovers from our unconventional monetary policies have boosted growth and inﬂation prospects, not only in the euro area but also globally.
Post-APP international capital ﬂows
Let’s start, then, by trying to get a sense of how strong international capital ﬂows have been since we began our large-scale asset purchases. In other words, to what extent has there been a rebalancing towards foreign assets? Here some stylised facts from the euro area balance of payments data are instructive.
On slide 1 you can see that capital ﬂows have indeed been considerable in the past few years. At their peak around the middle of 2016, net capital outﬂows – measured here in terms of 12-month moving sums – reached nearly 5% of euro area GDP. Never before in the history of the euro area have capital ﬂows been so high.
Of course, in the absence of a counterfactual, we do not know how much of this is actually the result of our monetary policy measures. After all, investment decisions reﬂect a number of factors, including risk perceptions and, importantly, the yield that investors can earn abroad. Often, capital ﬂows also simply reﬂect new investments and savings – just think about the euro area’s growing current account surplus – or they can be distorted by tax and regulatory arbitrage.
Yet, it is striking to see that the turnaround in capital ﬂows from net inﬂows to net outﬂows started in mid- 2014, just when we had announced our credit easing package and when expectations were gradually building among market participants that we would also begin purchasing government bonds.
Slide 1 also tells a second story – that among the various investment opportunities available to investors, virtually all net capital outﬂows have been concentrated in purchases of foreign long-term debt securities. Equity investment, by contrast, has hardly changed in net terms over time.
At face value, this suggests one tentative conclusion: large-scale purchases of sovereign bonds encouraged some investors to rebalance their portfolios towards the closest substitute – bonds issued by other sovereigns. This could be described as textbook quantitative easing (QE), not least because these ﬂows likely reﬂect the impact of our measures on bond prices in the euro area. According to ECB estimates, our monetary policy measures have contributed to reducing euro-area long-term risk-free rates by around 80 basis points since June 2014. As a result, spreads over bonds issued by non-euro area sovereigns have increased notably.
For example, towards the end of last year the spread between ten-year US Treasuries and equivalent German Bunds hit levels above 200 basis points – its highest level since the fall of the Berlin Wall. On average last year, the spread was around 60 basis points higher than during the six months leading up to the launch of our credit easing package. And the observed spread already reﬂects the likely impact that increased purchases of foreign bonds has had on their prices. In other words, while our measures can be expected to have contributed to a widening of spreads, foreign rebalancing is likely to have caused the spreads to narrow. The net eﬀect is what we see reﬂected in prices today.
ECB internal model analysis conﬁrms that yield spreads are an important driver of international portfolio ﬂows. On slide 2 you can see that yield diﬀerentials between euro area government bonds and those issued by non-euro area governments of advanced economies can explain a signiﬁcant part of net debt outﬂows over the course of last year. Risk aversion can explain much of the remainder.
Now, net outﬂows in debt securities can mean two things: either euro area investors increasingly moved domestic funds abroad, or foreign investors – those resident outside the euro area – have sold euro area bonds. Both mechanisms are part of the type of portfolio rebalancing that policymakers typically have in mind when thinking about asset purchases. In practice, they can reinforce each other and this is what we have been seeing in the euro area.
Consider slide 3. Here you can see that euro area investors have been a major driving force behind the net outﬂows. In fact, since the start of the APP in March 2015, net purchases by domestic investors have been almost entirely in the form of long-term foreign ﬁxed income securities. There have been anticipation eﬀects, but the chart strongly suggests that net debt outﬂows accelerated as the APP continued and decelerated only when we decided to reduce the pace of our monthly purchases in December last year.
Although it is unclear to what extent these ﬂows ultimately inﬂuence bond prices, the pattern of international capital ﬂows at least encourages discussion about the stock versus ﬂow eﬀects of central bank asset purchases – an interesting angle that should be explored further in future research.
On slide 4 you can see that euro area investors did not only choose to rebalance into the closest substitute, namely government bonds, but also predominately into the safest of the issuers. Around two-thirds of net outﬂows went into securities issued by issuers in the United States, United Kingdom, Japan, Sweden and Canada. By contrast, and contrary to what has often been discussed in the wake of the Federal Reserve’s asset purchase programmes, there was hardly any investment into emerging market economies, including the BRICs – Brazil, Russia, India and China.
So euro area investors deliberately chose to redirect some of their funds into the closest substitute. Slide 5 shows the extent of this rebalancing more clearly. Since the start of the APP, euro area investors alone accounted for more than half of foreign purchases of US debt securities. History suggests that these shares are unusual for the euro area.
The other side of the coin is the behaviour of non-euro area investors. You can see this on slide 6. In net terms, foreign investors have been neither large sellers nor large buyers of euro area assets in recent months – the solid blue line is close to balance.
But this masks a signiﬁcant diﬀerence between asset classes: foreign investors have been large buyers of euro area equity, while they were actively selling euro area bonds throughout most of last year, thereby reinforcing the actions of domestic investors.
Broadly speaking, this means two things: ﬁrst, our policy measures have boosted conﬁdence in the euro area’s growth prospects, which has brought foreign investors back to euro area stock markets. This conﬁdence eﬀect has also been conﬁrmed in empirical studies.
You can see clearly on slide 6 that this trend essentially started after the announcement of the outright monetary transaction (OMT) programme in mid-2012 and that inﬂows accelerated further in the aftermath of our credit easing package of June 2014. By the end of 2014, shortly before we announced purchases of government bonds, annual inﬂows into euro area stock markets by non-residents had reached 4% of euro area GDP – the highest on record. Policy stimulus and measures to repair the bank lending channel were clearly seen as having a positive eﬀect overall on the euro area economy.
The second thing to note is that some selling of euro area bonds by non-residents is a mechanical feature of the ECB’s asset purchase programme. What I mean by that is that foreign investors are holding a relatively large share of outstanding euro area government bonds. For example, when we started the APP in March 2015, non-euro area investors were holding nearly 75% of German Bunds with maturities of seven to ten years. So it should not come as a surprise that around 45% of our purchases of government bonds have been with investors resident outside the euro area. And if these foreign investors decide not to reinvest the proceeds from the sale in the euro area, a net outﬂow is recorded.
This feature makes the ECB’s asset purchase programme structurally diﬀerent from its counterpart in Japan, for example, where at the start of the Bank of Japan’s quantitative and qualitative monetary easing (QQE) in April 2013 less than 10% of outstanding Japanese government bonds ( JGBs) were held by foreign investors. As a result, it is mainly domestic investors that have signiﬁcantly reduced their JGB holdings in response to QQE.
And as you can see on slide 7, they did not rebalance their portfolios towards foreign bonds: in the ﬁrst two years after the start of QQE there were hardly any net outﬂows in debt securities despite a massive widening in international spreads following the US “taper tantrum” in mid-2013. Instead, QQE initially led to strong inﬂows into Japanese equities – a conﬁdence eﬀect similar to the one I have just described for the euro area. However, these inﬂows reversed relatively quickly.
In the United States, by contrast, the holding structure of the Treasury market is more similar to that of the euro area bond market, but actual purchase patterns are notably diﬀerent. Speciﬁcally, although non- residents owned around 60% of marketable US Treasuries in 2008, empirical evidence suggests that they were not the ones selling directly and indirectly to the Federal Reserve under its various rounds of QE. You can see this on slide 8. In fact, perhaps with the exception of the ﬁrst round (“QE-1”), foreign investors continued to pile into US Treasuries, in particular after the third round (“QE-3”).
This is not to say that quantitative easing by the Federal Reserve did not boost international capital ﬂows. In fact, numerous studies, albeit mostly reliant on event studies over short periods of time, tend to suggest that it triggered sizeable ﬂows into emerging markets, in particular after QE-2. Yet, net portfolio ﬂows, as you can see on slide 8, paint a broadly diﬀerent picture of post-QE portfolio rebalancing than that seen in the euro area.
I see three main factors that can explain the diﬀerence in post-QE debt ﬂows in the United States and the euro area.
First, the yield diﬀerential: ten-year US Treasuries and equivalent German Bunds were generating, on average, broadly the same return, in local currency, during the Fed’s QE programmes from late 2008 to early 2013. But, as I hinted at before, yield diﬀerentials were quite diﬀerent in recent times. Since the start of the APP in March 2015, ten-year US Treasuries have been yielding, on average, around 170 basis points more than ten-year Bunds – an enormous diﬀerence that, to a large extent, reﬂects the divergence in monetary policy cycles that emerged in the wake of the euro area’s sovereign debt crisis. As a result, portfolio rebalancing during the ECB’s asset purchase programme has been more attractive for investors than it was during the Fed’s QE programmes.
The second factor is negative rates. In addition to yield diﬀerentials, the absolute yield level may also matter for investors, particularly if rates are negative. It may be no coincidence that net bond outﬂows were among the largest when ten-year German Bund yields hit a low of nearly -20 basis points last summer. In a recent survey among foreign central banks, 70% of the respondents reported that negative interest rates in the euro area had encouraged them to adjust their allocations to the euro.
This emphasises the strong ampliﬁer eﬀect of negative rates – they are highly eﬀective in boosting portfolio rebalancing eﬀects, both domestically and across borders, exactly when monetary accommodation might be needed the most.
And third, we have political uncertainty and the role of the US dollar in the international ﬁnancial system. The US dollar is the main reserve currency and US Treasuries are the world’s most important safe haven asset, and as such beneﬁt from a liquidity advantage. Hence, in periods of elevated uncertainty, such as after Brexit, investors may overweight US Treasuries and underweight European assets. So, rebalancing out of US Treasuries is always likely to be more limited than out of any other safe security.
At this point I would like to draw three preliminary conclusions.
One is that unconventional policy measures can signiﬁcantly boost conﬁdence, leading foreign investors to review their international portfolio allocations, in particular with regard to equity.
A second conclusion is that the extent to which investors rebalance their portfolios towards foreign bonds is likely to depend on the structure of local bond markets, in particular the holding composition, and the risk- adjusted return investors can expect to receive from switching domestic and foreign bonds.
This means that there can also be noticeable diﬀerences between euro area countries, as can clearly be seen when looking at the recent behaviour of Japanese investors. Slide 9 shows that Japanese investors accumulated large positions in French securities in the aftermath of the launch of the APP, while at the same time selling German Bunds – a rebalancing within the currency union that likely reﬂected yield and safety considerations.
And a third conclusion, more speciﬁc to the euro area, is that any future decline in the pace of our asset purchases is unlikely to have much direct negative impact on emerging markets, given that ﬂows have been very limited.
Capital ﬂows and exchange rates
So, the evidence is fairly clear that asset purchase programmes are associated with net capital ﬂows, and probably also with changes in relative asset prices. This brings me to the second part of my remarks: the cross-border eﬀect of asset purchases on exchange rates.
Two questions stand out. Do large-scale asset purchase programmes lead to exchange rate depreciation in the initiating country? And, if so, does monetary easing produce a beggar-thy-neighbour eﬀect for other economies?
The ﬁrst point to note is that, during announcements of asset purchase programmes, the currencies of all major central banks have indeed tended to depreciate. You can see this on slide 10, which shows, for all countries, the cumulative short-term impact of selected major announcements of asset purchase programmes and negative rates, or both.
Of course, exchange rates often change before actual announcements in anticipation of policy actions. In the case of the euro area, for example, studies of a larger number of events that include such prior signals and comments by policymakers ﬁnd that the APP may have led to a depreciation of the euro of up to 12% against the US dollar.
What is more, ECB in-house research suggests that negative rates not only amplify portfolio ﬂows, they also amplify exchange rate movements. ECB staﬀ estimate that the reduction in the deposit facility rate from 0% to -0.2% in the second half of 2014 led, cumulatively, to a 1.9% depreciation of the euro against the US dollar, of which 0.5 percentage points were due to the non-linear eﬀects of interest rates moving into negative territory.
These non-linear eﬀects could be related to the ampliﬁcation of the carry trade channel, where currencies with negative funding costs may become the preferred short leg for carry trade strategies. Negative rates can therefore amplify the initial volatility-supressing eﬀect of asset purchases on carry trades. The lower the rates, the stronger these eﬀects can be expected to be. This applies not only to the euro area but, as you can see on slide 10, also to Japan.
It is tempting to think that currency movements related to asset purchase programmes are a direct consequence of their impact on the supply of and demand for currency when investors rebalance their portfolios across borders. The truth, however, is that the direction of causality is much less clear in practice. Indeed, a recent Deutsche Bundesbank study ﬁnds that actual purchases under the APP have not led to statistically signiﬁcant euro exchange rate movements.
I see two main reasons for this.
First, since exchange rates are asset prices that react in real time, they incorporate forward-looking expectations – most importantly expectations about short-term interest rate diﬀerentials. For instance, as you can see on slide 11, since the launch of the APP, the EUR/USD exchange rate has often closely mirrored two- year interest rate diﬀerentials between Germany and the United States. Capital ﬂows, by contrast, have tended to respond to interest rate diﬀerentials with a delay.
In other words, rather than driving exchange rate depreciation, capital ﬂows typically take place after this depreciation has already happened. One reason for this delayed response may be that, during asset purchase announcements, asset managers achieve an immediate change in portfolio weights without active sales or purchases via “passive” rebalancing – that is, the change in the exchange rate and asset prices.
This disconnect between the exchange rate and capital ﬂows is consistent with the empirical ﬁnding that the exchange rate is mostly aﬀected by the signalling channel of monetary policy – that is how policy decisions aﬀect the expected future path of short-term interest rates. ECB research ﬁnds that the depreciation of the euro exchange rate in response to the APP announcement in January 2015 occurred mainly through this signalling channel, including the added credibility asset purchases provide to that expected path. The portfolio rebalancing channel, by contrast, played much less of a role. In this sense, the eﬀect of asset purchases on exchange rates is not fundamentally diﬀerent from conventional interest rate policy.
The second reason to be sceptical about capital ﬂows aﬀecting the exchange rate is linked more to the structure of foreign exchange markets. For example, a large share of euro foreign exchange transactions – 84% – is initiated outside the euro area, notably in the City of London. As a result, they do not necessarily give rise to capital ﬂows that aﬀect the euro area. They are undertaken oﬀshore. The fact that foreign exchange trading in euro is so active in ﬁnancial centres outside the euro area helps explain why there is often little correlation between cross-border capital ﬂows and the euro’s exchange rate in the short to medium term.
In addition, portfolio outﬂows from the euro area – although large in size relative to economic activity – remain very small relative to overall transactions in euros in the foreign exchange market. They account at most for 0.2% of daily transactions in the euro-dollar pair, for instance. And to the extent that such investments are hedged, the link between capital ﬂows and exchange rate becomes even more tenuous.
Exchange rate spillovers and currency wars
Still, even if other eﬀects of unconventional monetary policy, rather than capital ﬂows, are the main inﬂuence on the euro exchange rate, the question of economic spillovers remains. This brings us to the “currency war” criticism of monetary policy.
Those taking this line argue that currency depreciation caused by unconventional policy easing can divert domestic expenditure away from foreign imports, thus creating a zero-sum game.
Clearly such arguments need qualifying.
One way to do so is to use NiGEM, a well-known global macroeconomic model with an articulated external sector and international trade and ﬁnancial linkages. Given the diﬃculties of capturing large-scale asset purchases in this model, policy easing at the lower bound is proxied by central banks imposing negative interest rates.
What you can see on slide 12 are two things. Scenario 1 portrays the impact of a multilateral, simultaneous cut in policy rates by the G-3 central banks, bringing rates into negative territory in all jurisdictions, including in the US. Scenario 2 involves a deepening of negative rates in the euro area only.
The scenarios illustrate two things.
First, the exchange rate channel is less important than most observers believe. Look at how similar the response of economic activity in the euro area is under the two scenarios, even though the exchange rate depreciation is less than half under the multilateral case. Clearly, the expansionary eﬀect of monetary policy easing is the dominant factor – in practice probably even more so as evidence is growing that the exchange rate pass-through to ﬁnal consumer prices has steadily fallen in recent years.
Second, unilateral policy actions are not a zero-sum game. Scenario 2 demonstrates that, although the currencies of our trading partners appreciate, a unilateral cut in the euro area interest rate is a net positive for all countries individually, and for the global economy as a whole. That is, if unconventional monetary policy helps prevent a country stuck at the lower bound from being a drag on global demand, all countries can beneﬁt. This has also been well documented in the academic literature.
In the case of the euro area, the demand-augmenting eﬀects are clearly visible. For example, since we introduced our credit easing package in mid-2014, net exports have hardly contributed to GDP growth. The recovery has been driven almost entirely by domestic demand, supported by a virtuous circle between rising employment, labour income and consumption.
Certainly, if we abstract for a moment from the fact that mark-up adjustments could have taken place, the fall in the euro exchange rate may have helped exporters hold onto their market shares as world trade stalled.
But by far the most dominant eﬀect of our monetary policy has been to encourage these domestic growth forces.
In other words, far from “beggaring its neighbours”, monetary policy easing by the ECB has added to global demand and thereby stabilised the global economy, in particular after the euro area had been a major drag on global growth for many years. In the last three years the euro area’s share of global growth has become much more signiﬁcant, contributing 0.5 percentage points, on average, to the 1.9% growth in advanced economies.
Moreover, we also responded to a number of signiﬁcant common, global shocks, such as the protracted period of low oil prices and the slowdown in growth in emerging market economies, which threatened not only the achievement of our price stability objective, but also that of many other central banks. So, the combined eﬀects of stabilising the euro area economy – a signiﬁcant source of demand for many countries – and cushioning global shocks contributed to putting global growth back on it current path.
Let me conclude.
Asset purchase programmes have been instrumental in easing monetary policy at the eﬀective lower bound in major advanced economies. A wealth of evidence suggests that these policies have successfully boosted economic growth and inﬂation prospects. At the same time, they have clearly had spillover eﬀects on other countries via capital ﬂows and relative asset price movements.
The evidence does not suggest, however, that these capital ﬂows have led to major exchange rate movements. Rather, exchange rates have responded to forward-looking interest rate diﬀerentials. Asset purchase programmes, together with negative interest rates, may have exacerbated those diﬀerentials through signalling and non-linear eﬀects, but their eﬀect on exchange rates is, by and large, not fundamentally diﬀerent from conventional policy.
What is more, currency depreciation is a side-eﬀect of policy and neither its main transmission channel, nor its objective. Monetary policy actions aimed at supporting domestic price stability objectives in advanced economies have been clearly positive for the global economy, mainly by stimulating employment, incomes and, ultimately, economic growth.
This is particularly the case in the euro area, which, thanks to a rich set of carefully calibrated monetary policy measures, is now adding to, rather than subtracting from, global growth. In this sense, the view that asset purchase programmes in large advanced economies have encouraged harmful currency wars is misleading. The world of monetary policy is not an arena where central banks engage and compete for advantage.
Thank you for your attention.