A Tale of Two Tails: The Impact of ECB Unconventional Monetary Policy on Perceived Crash Risk

(1)

A Tale of Two Tails:

The Impact of ECB Unconventional

Monetary Policy on Perceived Crash Risk

Sytske de Jong

10088180

Abstract

This paper examines the effect of European Central Bank (ECB) unconventional monetary policy (UMP) announcements on (tail) risk perceptions in Euro Area financial markets, by using an event study approach. This research employs an option-implied proxy to measure perceived downside risk, the so-called risk reversal (RR), using information embedded in out-of-the-money (OTM) equity indices - (Euro Stoxx 50, Euro Stoxx Banks and Euro Stoxx Insurance) and currency – (EURUSD) option prices. The results support the hitherto unproven hypothesis that the ECB managed to curb tail risk perceptions in both the broad stock market and the exchange rate. The risk-mitigating impact of ECB UMP is larger for measures of extreme crash risk (10ΔRR) than for measures of moderate downside risk (25ΔRR). The results present evidence for the insurance channel of monetary policy; the ECB is successful in ‘tackling’ black swan - scenarios. The risk-mitigating effect of ECB announcements is more muted for the banking sector. Moreover, most key UMP - announcements lead to an increase in perceived tail risk in the insurance industry. Furthermore, the results suggest that asset purchase programs are more powerful in reducing tail risk perceptions than liquidity- providing refinancing operations.

JEL Classification: E44

-

E52

-

E58

-

E58

-

G12

-

G14

-

G21

-

G22

-

G23

Keywords: Tail Risk Perceptions

-

Unconventional Monetary Policy

-

Event Study

-

Risk Reversal

-

Euro Area Index Options

-

Banks and Insurance Companies – Monetary Surprise

Master of Science in Economics, Amsterdam School of Economics

Specialization: Monetary Policy and Banking

(2)

Statement of Originality

This document is written by student S.A.T. de Jong, who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

1. Introduction

“So what are policymakers to do? First and foremost, reduce uncertainty. Do so by removing tail risks, and the perception of tail risks.”

– Olivier Blanchard1

The fall of Lehman Brothers in October 2008 was the watershed moment for the financial sector; it marked the end of the era of Great Moderation. The pre-crisis years corresponded to a period of stable growth, low inflation and reduced volatility. Mainstream economists started to believe that central bankers had resolved the monetary policy battles of yesteryear (Caballero, 2013; Stark, 2011). However, the conventional toolbox proved inadequate during times of turbulence. The successive financial crises impaired the traditional monetary transmission mechanism (Szczerbowicz, 2012). The ECB, in line with several other central banks, resorted to unconventional monetary policies (UMPs) in an attempt to reduce uncertainty, stabilize the economy and ensure liquidity (Frommherz, 2016). A decade after the Great Financial Crisis (GFC), one could conclude that monetary policy has changed dramatically and permanently; central banks worldwide are now using their balance sheets as backstops to the financial system (Cour-Thimann and Winkler, 2013).

Previous studies have emphasized that the widespread presence of tail risks2_have

contributed to the depth, proliferation and capricious nature of the 2008 crisis (see, for example, Mishkin (2009) and Orlowski (2008)). Recent history has changed investors’ risk perceptions forever and these changes in beliefs produce “long-lasting effects on

investment, employment and output”, even long after the tail event itself has passed (Kozlowski, Veldkamp and Venkateswaran, 2015, p. 1). Crash risks threaten the stability of the financial system as a whole, thus mitigating both perceived and realized tail risk is essential in reducing systemic risk (i.e. a high probability of a systemic financial collapse) (Kim and Zhang, 2014; Chava, Ganduri and Yerramilli, 2014; Orlowski, 2010).

1

Former Chief Economist of the International Monetary Fund (IMF). See: Blanchard, O. (2009, January 31). (Nearly) nothing to 2

In this study, ‘tail risk’ represents the loss at the most left (negative) part of the probability distribution. The terms ‘tail risk’, ‘crash risk’ and ‘downside risk’ are used interchangeably in this paper.

(6)

ECB president Mario Draghi has been very vocal about “removing”3_{these tail risks; first}

during the sovereign debt crisis and later when deflation risk was rising in the Euro Area. Hence, it is crucial to evaluate whether risks perceptions, and especially the perceived risk of a stock market or a currency crash, were actually reduced. After all, UMP instruments are controversial, because they bring along various macroeconomic risks (Van Dijk and

Dubovik, 2018). The broadening of the ECB’s mandate, to engage in quantitative easing (QE), caused a violent public debate. Therefore, the following research question is highly relevant for policy makers and academics.

What is the impact of unconventional monetary policy announcements on risk perceptions in the Euro Area? And did the ECB actually managed to reduce perceived tail risk in the equity market as well as in the exchange rate?

This study goes beyond earlier work in three ways. The first contribution is that a more recent period is considered than the periods analyzed by previous studies on Euro Area risk perceptions, prior research has mainly focused on the pre-QE phase of UMP (see, for example, Fratzscher, Lo Duca and Straub (2014)). The second contribution is that this paper examines the impact of UMP on several financial markets (the broad stock market, the foreign exchange market and the financial sector), whereas the previous literature on risk appetite so far concentrated solely on measuring the effects on the broad stock market. Finally, and most importantly, this is the first paper to evaluate the impact of ECB UMP on perceived downside risk in the euro exchange rate and the Euro Area broad stock market. Prior studies almost exclusively focused on more symmetric measures of risk (implied- and realized- volatility) that that do not specifically capture perceived crash risk.

Risk perceptions can be gauged by using information embedded in option prices. Option markets form a gold mine of information about investors’ expectations of future asset returns, so market prices of equity index- and currency- options are at the heart of this study. This research employs some of the most well known market indices that serve as a barometer of Euro Area market sentiment; the Euro Stoxx 50 index4_{, the Euro Stoxx Banks}

3

See the quote of Mario Draghi, president of the ECB, on page 9 of this paper. 4

The Euro STOXX 50 is a regional index and comprises the 50 largest EA stocks from various subsectors (including some banks and insurance companies). The twelve Euro Area countries covered in the ES50 index are Austria, Belgium, Finland, France, Germany, Greece, Ireland, Italy, Luxembourg, the Netherlands, Portugal and Spain.

(7)

index5_{and the Euro Stoxx Insurance index}6_{. In this paper the euro – dollar nominal}

exchange rate (EURUSD) is employed as a measure of the euro exchange rate rather than a portfolio of currency pairs. The EURUSD is the most actively traded7_{currency pair and}

therefore the most liquid (this minimizes issues around data quality, see for example

Kearney, Cummins and Murphu, 2017). The empirical tail risk proxy employed in this paper is the option-implied volatility skew; the difference between the implied volatility (IV) of an out-of-money (OTM) call and the IV of an OTM put of the same moneyness and maturity. This measure is widely known as a ‘risk reversal’8_{and it reflects the cost of hedging against}

a left-tail event. In essence, the price of crash insurance serves as an indicator of how investors perceive the risk of a market and or currency crash.

The empirical strategy of this paper is as follows. As a first step, the effects of UMP announcements and key public statements (hereafter referred to as ‘announcements’) on risk aversion (measured by implied volatility) and perceived tail risks (risk reversals) are captured by a non-parametric event study that consists of a visual and descriptive

component. Analyzing the behavior of the (tail) risk proxies around the announcement dates is a natural starting point for assessing the effect of UMP on perceived (crash) risk. In order to gauge the statistical significance of the observed changes, a bootstrap technique is used to correct for small- sample bias. As a second step, a regression-based event study

framework is set up that controls for drivers of risk perceptions other than ECB

announcements. Both a multivariate regression model using announcement dummies and a cumulative abnormal changes regression analysis in the spirit of Fama, Fisher, Jensen, and Roll (1969) are used. The regressions are complementary to each other. The dummy

variables regression examines the actual response, while the abnormal changes regression evaluates to what extent the (tail) risk proxies deviate from the counterfactual scenario (e.g. the estimated path of the risk measures if the ECB had not intervened). Subsequently, two robustness checks are performed to ensure validity of the obtained results. First, the event windows are expanded to check for the possibility of an anticipation effect the day before the announcement. Then, the baseline analysis (Euro Stoxx 50 index and EURUSD

5

The Euro STOXX Banks is a sectorial index composed of the 26 largest EA banks. The nine Euro Area countries covered in the ESB index are Austria, Belgium, France, Germany, Ireland, Italy, the Netherlands, Portugal and Spain.

6

The Euro STOXX Insurance is a sectorial index composed of the 14 largest EA insurers. The seven Euro Area countries covered in the ESI index are Belgium, France, Germany, Italy, the Netherlands, Portugal and Spain.

7

The EUR/USD is the most actively traded currency pair, with a daily turnover of 1.17t trillion dollar it accounted for 23% of total foreign exchange trades in 2016 (BIS, 2016).

8

The terms ‘risk reversal, ‘price of crash insurance’ and ‘the cost of hedging against a tail event’ are used interchangeably in this paper.

(8)

exchange rate) is extended to a broader sample of announcements to examine the

robustness of the obtained findings and the impact of the most recent phase of UMP (2016-2018). Finally, the parametric event study is repeated for the Euro Area financial sector (both the banking and insurance industry).

The results of this paper contribute to the existing literature in several ways. First, the empirical findings demonstrate that ECB announcements significantly increased risk

appetite, while contemporaneously reducing market participants’ immediate perceptions of downside risks. To the best of author’s knowledge, this is the first study that provides evidence for the tail risk-mitigating impact of UMP on the euro exchange rate and the broad Euro Area stock market. The findings are consistent with the evidence provided by Anene and D’Amico (2017), Hattori, Schrimpf and Sushko (2013, 2016) and Roache and Rousset (2013) for the risk-taking channel of monetary policy in the United States. Second, the results suggest that UMPs had a more muted effect on perceived tail risk in the banking sector. Moreover, most key announcements even seem to have aggravated tail risk

perceptions in the insurance industry. Third, the risk-mitigating impact of ECB UMP is larger for measures of extreme crash risk (10ΔRR) than for measures of moderate downside risk (25ΔRR). The ECB is successful in ‘tackling’ black swan - scenarios. Yet, this effect is less prevalent in the financial sector. The ECB seems better equipped to curb extreme crash risk in the broad stock market and exchange rate than in the banking and insurance industry. Finally, the results suggest that asset purchase programs are more powerful in reducing extreme tail risk perceptions than liquidity- providing refinancing operations.

This paper aims to make a useful contribution to the current academic debate on the risk-taking channel of monetary policy (see the pioneering work of Borio and Zhu (2008) and Brunnermeier and Sannikov (2012)). The results suggest that the ECB is effective in cutting off extreme downside risks and that monetary policy announcements can provide a

backstop to destructive scenarios. Hence, this paper presents evidence for the insurance channel of monetary policy, which can be classified as being a sub-channel within the risk-taking transmission mechanism. The insurance channel has been largely overlooked in the literature but the results warrant further theoretical and empirical examination. The findings have potentially important implications for the conduct and design of monetary policy, as the insurance channel is likely the most powerful and unique tool of non-standard monetary policy (Ubide, 2014).

(9)

The remainder of this paper is structured as follows: Chapter 2 presents some theoretical background on important concepts mentioned trough out this paper. Chapter 3 presents a review of published event studies that analyze the impact of UMP announcements on risk perceptions in financial markets. Chapter 4 provides a description of the data and presents the summary statistics of the data set along with some preliminary observations. Chapter 5 discusses the event study methodology and its limitations and presents the event study framework used in this study. Chapter 6 presents the results of the non-parametric and parametric analyses performed on the baseline-9_{, robust-}10_{and extended-}11_{samples along}

with a discussion of the key findings. Finally, chapter 7 concludes by providing implications for monetary policy and suggestions for future research.

2. Theoretical Background

This chapter provides a brief overview of unconventional monetary policy in the Euro Area (section 2.1) and its main transmission channels (section 2.2). The aim of section 2.3 is to provide some theoretical background on the pricing of options; it also introduces the option-implied (tail) risk measures used in this study. Finally, section 2.4 explains why the ECB should be aware of tail risks despite the fact that it is not a part of its (inflation) mandate.

“The removal of tail risks related to unfounded fears regarding the euro was essential to fight fragmentation of the euro area markets. And that has been important because with fragmentation – I’ve stressed this many times – we are not in a position to deliver price

stability for the euro area as a whole.”

– Mario Draghi12

2.1. Unconventional Monetary Policy in the Euro Area

This section describes the different policy programs initialized by the ECB and Table 1 summarizes the key UMP announcements that are employed in this study.

9

Euro Stoxx 50 index and EURUSD exchange rate, from January 2, 2007 until March 6, 2017. 10

Euro Stoxx 50 index and EURUSD exchange rate, from January 2, 2007 until June 29, 2018. 11

Euro Stoxx Banks index and Euro Stoxx Insurance index, from January 2, 2007 until March 6, 2017. 12

President of the European Central Bank (ECB). See: Barber, L. and Steen, M. (2012, December 13). Interview with Mario Draghi. The Financial Times.

(10)

According to Houben (2017), ECB UMPs can be loosely grouped into three time periods and four types of instruments. Phase I instruments were created to combat the financial crisis (2007-2010), Phase II instruments were created to combat the sovereign debt crisis (2010-2013) and Phase III announcements were created to combat deflation risks in times of low growth and even lower inflation (2013-2016/2017). Building on the work of Houben (2017), this study proposes an additional, fourth, phase of UMP (2016/2017-present). Phase IV, or taper-, announcements are associated with the gradual winding down of QE activities (e.g. the slowing of the pace of purchases). The ECB UMP programs vary in nature and objective but they can generally be divided into four categories: credit and collateral easing, the negative policy rate, forward guidance and quantitative easing.

Phase I – Global Credit Crunch, Credit and Collateral Easing. Widespread financial

tensions severely impaired the functioning of the money market in August 2007; the period thereafter corresponded to a systemic shortage of liquidity (González-Páramo, 2009). The interbank market froze and the interest rate and bank lending channel broke down. The announcement of the first Longer Term Refinancing Operation (LTRO) on August 22, 2007, represented the start of ECB’s support to the banking sector (Rivolta, 2014). The ECB started to intervene more aggressively after the 2008 outburst of the global financial crisis by extending the list of eligible collateral for the refinancing operations. Moreover, it extended the LTRO from a 3- to a 6-month maturity to further enhance liquidity conditions to banks and remove uncertainty in their liquidity planning (Falagiarda, Mcquade and Tirpák, 2015). The Governing Council13_{announced the Enhanced Credit Support (ECS) programme}

on May 7, 2009 (Fausch and Sigonius, 2018). ECS consisted of the continuation of the liquidity- oriented measures and the introduction of the Covered Bonds Purchase

Programmes (CBPPs). The covered bond market is a primary source of funding for banks,

and it had virtually dried up in the wake of the fall of Lehman Brothers (Pattipeilohy, Willem Van Den End, Tabbae, Frost and De Haan, 2013). The CBPP was the ECB’s first attempt to directly stimulate the monetary system, whereas the refinancing operations only impacted the market indirectly trough supporting the supply of liquidity (Frommherz, 2016). In Phase I, the ECB effectively took over a large share of the interbank market, acting as lender of last resort (Reichlin, 2015).

13

The main decision-making body of the ECB is the Governing Council, which consists of the six members of the Executive Board plus the governors of the central banks of the 19 euro area countries. See ECB organization (www.ecb.europa.eu).

(11)

Phase II – Euro Area Sovereign Debt Crisis. The second phase of the crisis started in

2010, when the Greek drama unfolded. The euro sovereign debt crisis led to the financial fragmentation of the Euro Area and monetary stimulus packages were consequently not transmitted to distressed countries (Houben, 2017). The Governing Council introduced the

Securities Markets Program (SMP), the Very Long Term Refinancing Operations (VLTROs or

3-year LTROs) and the Outright Monetary Transaction program in response to the sovereign debt tensions in the Euro Area. With the Phase II instruments, the ECB attempted to repair the monetary transmission mechanism and reduce the tail-scenario of an euro break-up (Pericoli, 2016).

The SMP was introduced on May 9, 2010. The ECB tried to bring down yields in the

aforementioned countries by ensuring liquidity in secondary markets by directly purchasing Greece, Portuguese and Irish government bonds. The effects of the SMP purchases were fully sterilized, so that the monetary policy stance would not be influenced and it technically did not constitute monetary easing (Dewachter, Iania, and Wijnandts, 2016; Ambler and Rumler, 2016; Rogers, Wright and Scotti, 2014). On 7 August 2011, the scope of the SMP was extended to Italy and Spain, when markets were questioning their creditworthiness. The deterioration of sovereign credit ratings generated a dangerous correlation between governments and banks (Reichlin, 2015). The large balance sheet exposure of Euro Area banks to Euro Area sovereign debt resulted in a diabolic loop whereby sovereign risk and bank weakness reinforced each other (Beck, 2011). The Euro Area banking system was on the edge of a collapse. Mario Draghi declared in December 2011; “the most important thing for the ECB is to repair the credit channel.”14_{His remarks were considered as a strong hint}

that the ECB was preparing a liquidity-providing plan for Euro Area banks (Dewachter et al., 2016). One week later, the ECB announced LTROs with a maturity of three year (VLTROs) to provide liquidity to the Euro Area financial sector.

With the OMT program, the Governing Council signaled its readiness to intervene in secondary sovereign bond markets for unspecified amounts. Mario Draghi pledged to do “whatever it takes”15 to save the euro and to reduce the risk of a euro crash. The OTM program aimed to reduce the excessive risk premia and thereby the borrowing costs of distressed countries. Purchases under this program were conditional on the participation of

14

In a speech at the European Parlement (December 1, 2011). 15

(12)

the bond-issuing country in an economic programme with the European Financial Stability Facility (EFSF) and the European Stability Mechanism (ESM). Hitherto no country

participated in the OMT program, this possibly reflects sovereigns’ fear for imposed fiscal adjustments.

So, the OMT program formed “an effective backstop to remove tail risks from the euro area”, without ever being implemented.16

Phase III – Low Growth and Inflation Momentum. Overall, Phase II instruments were

successful in restoring the transmission of policy-rates to short-term sovereign bonds yields, but credit growth and inflation declined sharply in spite of that (Houben, 2017). The Governing Council introduced Forward Guidance (FG) and a series of Targeted Longer-Term

Refinancing Operations (TLTROs) along with a negative interest rate policy. Subsequently,

the ECB engaged in quantitative easing aimed at both the public and private sector under the general heading of the Asset Purchase Program (APP)17_{in order to fight the prolonged}

stagnation of the Euro Area economy and to fulfill its mandate of price stability by keeping inflation close to 2% (Pelizzon and Sottocornola, 2018).

From July 2013 the ECB started with Forward Guidance and interest rates on deposit facilities turned negative from June 2014 onwards. Once policy rates have reached the zero lower bound (ZLB), additional monetary stimulus can potentially still be achieved by

managing expectations about the future course of policy (Coenen et al., 2017). So, anchoring expectations about the future monetary strategy has become a policy tool in itself. With FG, the ECB provides transparent information to the markets on the evolution of the key policy rates and other macro economic variables, since clear central bank

communication on future policy is the key to reducing market uncertainty today.

The ECB introduced a series of TLTROs to provide funding cost relief for banks in order to enhance bank lending to the real economy, i.e. the non-financial private sector, in the zero lower bound environment (Beyer et al., 2017). The APP includes four sub-programs under which different types of securities - government, corporate, covered bonds and asset-backed – are purchased by the ECB (Dewachter et al., 2016; Falagiarda, Mcquade, et al.,

16

Mario Draghi in his ‘Introductory statement to the press conference’ (September 6, 2012). 17

APP consisted of the Asset-Backed Securities Purchase Program (ABSPP), a new Covered Bond Purchase Program (CBPP3), the Public Sector Purchase Program (PSPP) and the Corporate Sector Purchase Program (CSPP).

(13)

2015). On March 10, 2016 the Governing Council announced a further expansion of the APP18_{(Demertzis and Wolff, 2016).}

Phase IV – Cautious Steps towards Tapering. Later that year, on December 8, the

markets were taken by surprise by ECB’s announcement to cut monthly asset purchases from 80 billion to 60 billion euros. The reduced pace of purchases would start from April 2017 onwards. During the subsequent press conference, Mario Draghi made it very clear that “there is no question about tapering, (…) tapering was not discussed today”. But even without calling the policy move tapering, the ECB de facto started with the gradual winding down of central bank purchase activities. On June 14, 2018, the Governing Council decided that it would end QE by the end of that year, while pledging to keep the policy rates at their present levels until at least the summer of 2019. After three years of asset purchases, rates at the effective lower bound and cheap funding the ECB seemed finally confident that Euro Area inflation is on track to reach its goal.19

2.2 Unconventional Monetary Policy Transmission Mechanism

The process through which monetary policy interventions impact the economy has changed since UMPs were first introduced. Conventional monetary policy mainly operates by

impacting interest rates across different maturities, whereas unconventional measures include direct lending to governments and the private sector, forward guidance and outright purchases of a myriad of assets. From a theoretical perspective there are several

transmission mechanisms by which these measures are transmitted to financial markets. Previous research has identified the main channels trough which UMP could mitigate downside risk perceptions (see, for instance, Adrian et al. (2016), Silva (2015), Hattori, Schrimpf and Sushko (2013, 2016) and Adrian and Shin, 2009). These channels are considered in more detail below, with figure A. providing a stylized representation.

Although in practice, the monetary transmission mechanism is a complex web of economic interactions and the identification of causal relationships is difficult (Fiedler, Jannsen, Wolters, Hanisch and Hughes Hallett, 2016). Moreover, channels are not mutually exclusive and can work in parallel (Fratzscher, Lo Duca and Straub, 2014). Consequently, it is difficult to quantify the overarching impact of UMP on risk perceptions.

18

The combined monthly amount purchased was increased from 60 bn. to 80bn. . 19

See: Randow, J., Diamond, J.S. and Warren, H. (2018, July 13). When Will the ECB Pull Its Trillions From the Markets?

(14)

Roache and Rousset (2013) identify four channels through which UMP can affect asset prices and thus perceived risk: (I) the signaling channel, (II) the portfolio rebalancing channel, (III) the market liquidity channel and (IV) the confidence channel.

The ECB influences expectations of short-term rates through the signaling channel. It relies on agents having imperfect information about the prospective state of the economy (Zhou, 2018; Elbourne, Duijndam and Ji, 2018). Announcements can contain both explicit and implicit signals about how central bank policy will behave in the future (Woodford, 2012). Loose20_{UMP can signal the commitment of the ECB to maintain an accommodative policy}

for a longer period of time, which subsequently will lower the expected path of future short-term interest rates. As a result, bond yields and capital market rates will decrease whilst boosting asset prices, resulting in a decrease in risk premia (Lamers, Mergaerts, Meuleman and Vennet, 2016). Therefore forward guidance, which operates via the signaling channel, can have a substantial and mitigating impact on volatility expectations and risk premia of long-term rates (Hattori, Schrimpf and Sushko, 2016).

Central banks can influence long-term rates “by changing the quantity and mix of financial assets held by the public”21_{. This second channel is known as the portfolio-rebalancing}

channel, it works via relative demand and supply of government bonds and requires

frictions that preclude perfect arbitrage between long-term and expected short-term rates (Neely, 2015). Those frictions are typically preferred habitat investment behavior22_and

financial market segmentation; the reality that assets and bonds of different maturities are not regarded as being perfect substitutes in the eyes of market participants (Altavilla, Carboni and Motto, 2015). Central bank purchases create a shortage in some market segments. Due to imperfect substitutability among different assets, preferred habitat

investors have no tendency to sell bonds to the monetary authorities. Consequently, relative supply of the targeted securities decreases. This scarcity effect induces (I) a reduction of the term premium and (II) portfolio rebalancing among other investors towards more risky

assets, thereby increasing the price of a broad range of financial securities and reducing their yields (Gambetti and Musso, 2017; D ’Amico, English, Lopez-Salido and Nelson, 2011).

20

Reducing the policy rate and/or increasing the central bank’s balance sheet. 21

Ben Bernanke, former Chairman of the Federal Reserve (FED). See: Bernanke, B. 2010. “The Economic Outlook and Monetary Policy.” Speech presented at the Federal Reserve Bank of Kansas City Economic Symposium, Jackson Hole, Wyoming, August 27.

22

Market participants with preferences for certain bonds with specific maturities or safety characteristics (Hattori, Schrimpf and Sushko, 2013).

(15)

This so-called ‘search for yield’23_{mechanism is pivotal for the pass-through onto sectors}

that do not hold nor issue QE targeted securities and therefore not directly benefit from the UMPs (Albertazzi, Becker and Boucinha, 2018). The fall in long-term sovereign yields eases credit availability for firms and households, which in turn creates a more benign economic environment (Varghese and Zhang, 2018).

Figure A: Diagram summarizing the Transmission Channels of UMP

Notes: Figure A is a stylized representation of the unconventional monetary transmission mechanism. There are many ways to characterize the transmission channels (see the gray shaded cells) through which non-standard monetary policy affects various macroeconomic and financial market variables (see the white shaded cells). Grey thick arrows denote the effect of/on the mix of the exchange rate, asset prices and bank- and market rates.

Source: Author’s flow chart.

23

A term coined by (Rajan, 2006).

Unconventional Monetary Policy

Portfolio Rebalancing

Channel

Signaling

Channel Confidence Channel

Market Liquidity Channel Risk-Taking Channel Exchange Rate Channel

Bank & Market Rates

Risk Premia

Exchange Rate Asset Prices

Risk Aversion & Uncertainty

Tail Risk Perceptions

Insurance Channel

(16)

The exchange rate channel, according to which loose ECB announcements might lead to a depreciation of the euro, can be seen as a special sub-channel of the portfolio-rebalancing channel (Gambetti and Musso, 2017). Central bank announcements are a primary driver of exchange rate volatility as monetary policy surprises affects investors’ expectations of the future exchange rate path (Conrad and Lamla, 2010). Loose, or expansionary, UMP increases the supply of the currency and uncovered interest rate parity (UIP) causes the exchange rate to depreciate on the short term (when prices are sticky). However, the impact of an interest rate cut could be dampened by other factors that drive the currency up such as risk premia and market expectations. A decrease in the currency’s risk premium

suggests an increase in its attractiveness as an investment currency; capital inflows will lead to an appreciation of the exchange rate. Conversely, heightened uncertainty prompts

investors to seek refuge in typical ‘safe haven’ currencies such as the Japanese yen, the Swiss franc, and the U.S. dollar (De Bock and de Carvalho Filho, 2013). Recent literature suggests that the impact of the exchange rate channel of monetary policy has been growing significantly since the onset of the financial crisis (see, for instance, Pericoli and Veronese (2017)24_{and Glick et al. (2015)}25_{). Ferrari, Kearns, and Schrimpf (2017, p. 2) conclude that}

the negative interest rate environment has contributed to that development; “as the effective lower bound becomes increasingly binding, the exchange rate bears more and more the burden of adjustment”.

Central banks can improve the market functioning of bond markets via the market liquidity

channel, also referred to as the market liquidity and credit risk channel (Urbschat and

Watzka, 2017). The great financial crisis evolved into a liquidity crisis when the interbank market froze; liquidity premia increased exponentially.

Central banks can avoid that in crisis financial institutions fire sale non-liquid assets by acting as a backstop purchaser/ seller (Bindseil, 2013; Joyce, Tong and Woods, 2011). During the sovereign debt crisis, the ECB managed to improve market liquidity with its purchase programs by reducing liquidity risk premia and subsequently yields (Elbourne et al., 2018).

The ECB can influence the risk appetite of investors via the confidence channel. UMPs have a large and positive impact on investor sentiment provided that the announcement is

24

Pericoli and Veronese (2017) show that the impact of ECB monetary policy surprises on the euro exchange rate is much stronger than in the pre-crisis period.

25

Glick, Leduc and Francisco (2015) show that the impact of unconventional monetary policy on the dollar has been roughly three times that of conventional monetary policy prior to the Great Financial Crisis.

(17)

credible and decisive (Fratzscher et al., 2014; Lutz, 2015). Generally, loose policy reduces market volatility by convincing the market that the future economic environment will be better. The increase in confidence results in a decline in uncertainty and risk premia, which in turn boosts asset prices. But the confidence channel may work to offset the effects that are induced by the signaling channel (Roache and Rousset, 2013). Market participants may see some loose UMP announcements as an indicator that suggests that the economic outlook is less favorable than hitherto supposed. Investors realize that expectations are worse than currently priced in and risk premia will rise.

A series of recent studies (Hattori, Schrimpf and Sushko 2013, 2016; Borio and Zhu, 2012; Gambacorta, 2009; Adrian and Shin, 2009) has started emphasizing the impact of UMP on the pricing of downside risks and financial sector risk-taking as an additional channel of the monetary transmission mechanism. Central banks can influence the willingness of market participants to take on risk via the risk-taking channel, a term coined by Borio and Zhu (2012, p. iii). The authors argue that “recent changes in the financial system and prudential regulation may have increased the importance of the risk-taking channel and that prevailing macroeconomic models are not well suited to capturing it”. The channel underlines the role of the monetary authority in reducing either risk perceptions or risk-tolerance even when financial institutions are unconstrained, whereas the market liquidity channel and its risk mitigating effect only functions when the financial sector is liquidity constrained (Silva, 2015). Loose monetary policy induces market participant to take on risk, thereby influencing the degree of risk in the portfolio, the pricing of assets and ultimately influencing real

economic decisions (Bruno and Shin, 2015). Borio and Zhu (2012) distinguish three ways in which the risk-taking channel may be operative. Via (I) the impact of interest rates on valuations, incomes and cash flows, (II) the impact of sticky rate-of-return targets on portfolio choices and (III) the impact of central bank communication on investor behavior. Loose monetary policy can boost asset prices and profits, which in turn can increase risk appetite. A reduction in the interest rate can interfere with investment portfolios’ target rates of returns thereby stimulating search-for-yield behavior. Moreover, central banks can

compress risk premia with transparent and committed communication via a “transparency effect” and an “insurance effect” (Borio and Zhu, 2012, p. 244). Transparent

announcements reduce market uncertainty, while the latter concept indicates that the central bank is effective in cutting off large downside risks. Ubide (2014) have coined the

term insurance channel, which can be classified as being a sub-channel within the risk-taking channel category.

(18)

The insurance channel has been largely overlooked in the literature and yet it is likely the most powerful and unique tool of non-standard monetary policy (Ubide, 2014). In times of crisis, the public sector can take and redistribute risk when the private sector is very risk averse. Brunnermeier and Sannikov (2012) are the first to relate UMPs to tail risk

redistribution. The authors show in a theoretical framework that central bank lending and purchases can signal a commitment to redistribute future crash risk (provided that there is unconditional commitment and clear communication). Thus, UMPs can be regarded as an insurance against tail events in the eyes of market participants. The ECB can, by restoring risk aversion to normal levels, immediately impact perceived tail risks in the financial market.

2.3 Measuring (Tail) Risks with Options

This section describes how perceived tail risk can be measured using option pricing models.

Options give the option buyer the right, but not the obligation, to sell or buy the underlying asset at a prearranged price (the strike price) during an agreed-upon period of time. A call option is the right to buy an asset, whereas a put option is the right to sell at a given price. Hence, an investor who purchases a put option is limiting exposure to an abrupt downward movement in the underlying currency or market index.

2.3.1 Pricing Options

One of the most well known models for computing equity option prices is the Black – Scholes (B&S) model (see the work of Black and Scholes, 1973). The Black – Scholes formulas for call and put options are given by (2.1).

(2.1)

C(S_t,t)= StN(d1)− K −r(T−t )_N(d 2) P(S_t,t)= K−r(T−t )_N(_−d 2)− StN(−d1)

(19)

Where: d₁=ln S_t K +(r + 1 2 2_)(T_−t) T−t d₂= d₁− T−t And where:

C(S

_t

,t)

− theoretical call option price at time t ∈[0,T]

P(S

_t

,t)− theoretical put option price at time t ∈[0,T]

S

_t

− the price of the underlying asset at time t

K

− strike price

r

− risk-free interest rate

T

−t − time to maturity

N(x)

− cumulative distribution function of the standard normal distribution

σ − expected volatility of underlying instrument

Garman and Kohlhagen (1983) modified the B&S model for the foreign exchange market and extended the formula to deal with the prevalence of two interest rates; one for the domestic and one for the foreign currency (Huah and Wong, 2009). The only unknown variable within both models is the expected volatility

σ

, since the price of the underlying asset

_S

_t and the risk-free interest rate

_r

are observable in the financial market and strike price

_K

and time to maturity (

_{T −t}

) are specified in the option contract. The volatility of the underlying asset can be either estimated by computing the standard deviation of the time series of all historical closing prices of the stock in question or calculated from the market prices of traded options by reversing the B&S formula (Czech, 2018). The former is known as the historical or realized volatility of the option, whereas the latter is called the implied volatility.

One of the most important concepts for option traders is the moneyness of an option contract or its so-called intrinsic value. Its value can be indicated by (2.2) and it depends on the relationship between the price of the underlying asset

_S

_t and the strike price

_K

of the contract.

(20)

(2.2)

Call Option Intrinsic Value = S

_t

− K

Put Option Intrinsic Value = K

− S

_t

An option can be in-the-money (ITM), at-the-money (ATM) or out-of-the-money (OTM). OTM options have no intrinsic value (i.e. exercising the contract today is not profitable) so they are cheaper than derivatives that can be exercised at a profit (ITM options have a positive payoff and thus intrinsic value). The current price of the underlying asset equals the strike price when an option is ATM. This contract has no intrinsic value, but ATM options will be more expensive than OTM options because a smaller move in the underlying asset is needed to make the contract profitable. An option’s moneyness can be measured using the delta of an option (Kim and Zhang, 2014). The delta of an option is the first derivative of the option price

_K

with respect to the price of the underlying asset

_S

_tand it measures the sensitivity of the option premium to the volatility of the underlying (i.e. a 1-unit change) (Csávás, 2008; Galati, Melick and Micu, 2005). The market convention is to quote the delta of both a call and a put option as a percentage term between 0 and 100 (Malz, 1997). If an option has a delta of 25% and the current price of the underlying asset moves 1% (100 basis points), the market would expect the premium to move about 25 basis points (0.25 * 100). Options with a delta lower than 50∆ are considered OTM, equal to 50∆ are ATM and above 50∆ are ITM. The delta increases with the chance that an option will expire in the money and deltas closer to 0 or 100 represent contracts that are deeper in- or out-the-money.

Both the Black and Scholes – and the Garman and Kohlhagen model assume that asset returns follow a geometric Brownian motion, which means that the prices of the underlying security can be characterized by a lognormal distribution with a constant variance (Bańbuła, 2008). In practice, however, observed price changes seem to conflict with the normality assumption: the data distribution of financial market variables tends to be leptokurtic and negatively skewed (see Figure A). When a distribution is negatively skewed, it indicates that market participants see the risk of a large drop in asset prices as being more likely (Lewis, 2012). A leptokurtic distribution has more mass in the tails than a normal distribution; hence the market attaches greater probability to extreme outcomes and large swings in the asset’s price (Gereben and Pintér, 2005).

(21)

Figure A: The Moments of an Asset Return Distribution

Source: Author’s figure.

A non-normal distribution exhibits either kurtosis or skewness, or both. Kurtosis is known as the fourth moment of an asset return distribution or implied probability density function (PDF), and skewness or the option-implied skew as the third (Hattori, Schrimpf and Sushko, 2013). Most studies on the effects of ECB policy on risk perceptions thus far concentrated on the implied first (i.e. the mean or expected return) and the implied second (i.e. the standard deviation or volatility) moment.

Gereben and Pintér (2005 p. 9) argue that “it is plausible to presume” that the distribution describing investors’ expectations regarding future uncertainty, which proxies perceived risk, is also likely to suffer from non-normal data since the empirical asset return distribution in itself is leptokurtic. Option market data confirm this presumption and show that the implied volatilities of options at different strike prices but identical maturity differ, while they should be equal under the B&S assumption of constant volatility. This phenomenon is known as the volatility smile26_{; out-of-the-money and in-the-money options display higher}

volatilities than at-the-money options (Czech, 2018). Moreover, the volatility “smile” resembles more a volatility “smirk” due to its asymmetric distribution with its thick left tail (Hutchison and Sushko, 2013). Before Black Monday, the US stock market crash of 1987, equity options smiles were non-existent (Foresi and Wu, 2005). Before the 2008 crisis, market participants did not observe a volatility smile in the currency markets of developed

26

The “volatility” smile is a plot of the relationship between the implied volatility and the delta value of an option at a given maturity (Eitrheim, Frøyland, & Røisland, 1999).

Skewed Curve Positively skewed, “tail” pulled in positive direction. Normal Curve Kurtosis (K) = 3 Leptokurtic Curve K > 3 High kurtosis indicates “fat tails” in the end of the

(22)

countries27_{(Farhi et al., 2015). Both Black Monday and the GFC appeared decisive}

moments for the equity - and currency- option market, respectively. Volatility smiles reflect investors’ fear of another tail-event; the experience of a crash has permanently changed market participants’ assessment of risk (Kozlowski, Veldkamp, and Venkateswaran, 2015). Further, option market data reveal a volatility skew, the ‘skew’ refers to the observation that implied volatilities of out-of-the-money calls and puts differ due to investor preferences and thus demand in either of them (De Bock and de Carvalho Filho,2013).

2.3.2 Risk Reversals as a Proxy for Perceived Tail Risk

Put options provide protection against a crash (Husted, Rogers and Sun, 2016). During turbulent times, an increased demand for put options alone is insufficient evidence of a security being a safe or risky one; the divergence between the price of a call and a put option matters (Wong and Fong, 2013). Risk reversals measure the relative demand for upside (calls) and downside protection (puts) on the underlying asset (IMF, 2012). They are often described as an indicator of the cost of hedging against a market/currency crash or a left-tail event as it captures the option-implied skew in the asset return distribution (i.e. the third moment) and the corresponding skewness risk premium. Yet, it does not measure the physical mass in the negative tail of the implied distribution. However, risk reversals exhibit high and statistically significant correlation with more direct measures of tail risk that are in itself less suitable and convenient for the present analysis since the required information tends to be available only at lower frequencies (Hattori et al., 2013). Hence, risk reversals are a relevant proxy for perceived tail risk (e.g. an indicator of how investors perceive the risk of a market/currency crash). Risk reversals (RR) are expressed as the difference between the implied volatility (IV) of an OMT call and the IV of an OTM put of the same moneyness (as measured by delta Δ) and maturity (X) (see (2.3))28_.

(2.3)

RR

xΔ

= IV

Call xΔ

_{− IV}

Put xΔ

27

Since the 2008 crisis, currency option smiles are no longer symmetric. OTM puts on high interest rate currencies have become more expensive than OTM calls, indicating a high risk of large depreciations in those currencies (Farhi et al., 2015, p.2).

28

The risk reversal can also be considered as an options strategy of a simultaneous sale of an out-of-the-money put option and purchase of an out-of-the-money call option (Fratzscher, 2005).

(23)

Jurek (2009) argues that only far out-of-the-money options are relevant, as they provide the most direct measure of the cost of insuring against crashes.

This study considers risk reversals for two degree of moneyness: 10Δ options (deep OTM, measure for extreme downside risk) and 25Δ options (OTM, measure for moderate

downside risk).

Since equity return distributions are typically negatively skewed, equity risk reversals tend to be negative on average (Hattori et al., 2013). This is a unique feature of the equity options market, currency risk reversals change from positive to negative values over time (Carr and Wu, 2007; Foresi and Wu, 2005). For the equity market, a negative value of this tail risk proxy indicates that investors are more fearful of a stock market crash than of a market boom as they are willing to pay more (higher IV) for downside protection than upside protection. If perceived tail risk decreases, demand for crash insurance will fall along with the price of OMT equity put options and the equity index skewness becomes less negative. For currency risk reversals, the interpretation is less straightforward. Previous studies have emphasized that moments of the implied PDFs capture both investors’ risk appetite and market views as to the likelihood of particular exchange rate outcomes (Galati et al., 2005; Melick and Thomas, 1997). Galati et al. (2005, p. 991) argue that, in the absence of strong assumptions, it is impossible to distinguish between the two. The authors provide a simple “fire insurance” analogy29_{to support their claim. Based on currency risk reversals alone, we}

cannot distinguish whether a change in the price of crash insurance largely reflects a change in views on future exchange rates, altered preferences towards risks or both. A simultaneous analysis of the changes in risk reversals and a measure of risk aversion (implied volatility) is necessary to fully understand the impact of UMP on perceived crash risk. A negative EURUSD risk reversal implies that volatility for OMT EURUSD put options (e.g. insurance against euro depreciation and dollar appreciation) is greater than volatility for OMT EURUSD call options (e.g. insurance against euro appreciation and dollar

depreciation) (Brunnermeier, Nagel and Pedersen, 2008).30_{It indicates that investors charge}

a greater euro crash risk premium; market participants attach a greater probability to a depreciation of the base currency (euro) against the quote currency (dollar) (Czech, 2018;

29

“Suppose we observe an increase in the price of fire insurance. This increase might reflect the market view that fires are now

more likely; hence there is a greater need for insurance. Alternatively, the price increase might reflect a change in market sentiment regarding potential exposures in the event of a fire. The probability of a fire may not have increased; however, the market view of any loss associated with a fire may have increased. The observation that the price of insurance has increased does not allow us to determine whether fires are more likely, whether exposure is perceived as larger, or some combination of both.”

30

Risk reversals express “the cost of buying insurance against foreign currency appreciation, financed by providing insurance against foreign currency depreciation” (Brunnermeier et al., 2008, p.321).

(24)

Jurek, 2009). A currency risk reversal that becomes more negative reflects an increase in perceived risk in the exchange rate in most cases. This is, however, not the case for the QE-phase of UMP. A bias towards euro depreciation reflects that market participants see the ECB QE programme as credible (e.g. an increase in the monetary stance devaluates the currency). Moreover, risk reversals are useful for analyzing carry traders’ risk perceptions (Gagnon and Chaboud, 2007). Wong and Fong (2013) show that a negative EURUSD risk reversal in times of turmoil reflects that the dollar is looked upon as a safe haven currency by investors (and the euro is considered riskier).

2.3.3 Other Risk Measures: Implied Volatility and Realized Volatility

For comparison purposes, this study includes other well-known measures of volatility as additional measures of risk (not for the extended financial sector analysis). Both realized and implied volatility can be considered as symmetric measures of risk as they do not

specifically focus on downside risk (Hattori, Schrimpf and Sushko, 2016). Realized volatility is a proxy for the actual volatility of historical asset returns or the exchange rate and there are several ways of measuring it (Lyócsa, Molnár, Plíhal and Veka, 2017).31_Whereas

option-implied volatilities convey forward-looking information. Hence, Evaristo et al. (2016) argue that implied volatility indices are very useful risk indicators for the financial market in general and central banks in specific. The VIX and the VSTOXX are widely followed measures of market uncertainty in the U.S. and euro stock market; they are commonly used as investors’ gauges of ‘fear’.32_{Uncertainty or volatility is a market-based indicator that is closely related}

to investors’ risk appetite or risk aversion. There is academic disagreement on how to measure risk aversion (Kilponen, Laakkonen and Vilmunen, 2012; Luo, Biefang-Frisancho Mariscal and Howells, 2011). Many papers do so by employing implied volatility indices as a proxy for (global) risk appetite, despite the criticism that those indices may contain an element of financial market uncertainty as well (see, for example, Costa Filho (2016), Kilponen et al. (2012), Luo et al. (2011) and Brunnermeier and Nagel (2008)). It is hard to disentangle whether the dynamics of implied volatility indices are largely driven by a change in risk preferences or by a change in expected volatility. But Popescu and Smets (2010) found that, in the case of the German stock market, investors’ perceived risk explains most of the fluctuations in implied volatility indices rather than market uncertainty.

31

All measures of realized volatility estimate the quadratic variation or the integrated variance of a price process over time. The measures differ with respect to the size of the interval (intraday (5/10 minutes), daily or weekly), estimation methods and tuning parameters such as a kernel bandwidth or “block size” (Liu, Patton and Sheppard, 2012).

32

The VIX is a measure of the market’s expectation of volatility implied by S&P500 index options and the VSTOXX is based upon the Euro STOXX 50 index.

(25)

2.4 The Importance of Extremely ‘Unlikely’ Events

“We should always be humble by recognizing that our knowledge with regard to tail risks is limited.”

– Masaaki Shirakawa33

“There are known knowns. These are things we know that we know. There are known unknowns. That is to say, there are things that we know we don't know. But there are also

unknown unknowns. There are things we don't know we don't know.” – Donald Rumsfeld34

Tail risk, or rare- disaster risk, is a term that has been used broadly for extreme adverse events35_{in financial markets (Wong and Fong, 2013). In statistical terms, it is defined as the}

risk of a move of an asset or portfolio of assets greater than three standard deviations from its mean (The Economist, 2012). These three-standard deviations fluctuations have become more frequent over the last decennia and financial crises have occurred more commonly than suggested by a normal distribution of asset returns (Shirakawa, 2011). Leptokurtic or fat- tails arise from “animal spirits”36_{as investor behavior is ruled by “either excessive}

optimism or pessimism”, leading to extreme market swings (Orlowski, 2010, p. 2). This represents a challenge for asset pricing models and thereby investors’ forecasting of these variables (Gao et al., 2017; Andersen, Fusari and Todorov, 2016; Orlowski, 2010). The unpredictability of these tail events can have a devastating impact on the financial markets in general and portfolio returns in specific. Zhang and Schwaab (2016) argue that upraised tail risks alone can force institutional investors out of certain assets, thereby triggering a ‘flight to safety’ and causing a downward spiral of panic, chaos and fire sales. Long after the tail event itself has passed “it produces long-lasting effects on investment, employment and output” (Kozlowski, Veldkamp and Venkateswaran, 2015, p. 1). Hence, managing tail risks is an important part of investors’ overall risk-management strategy despite the fact that those risks that are not forecastable have the potential to cause the most damage (i.e. the unknown unknowns) (The Economist, 2012).

33

Former Governor of the Bank of Japan (BoJ). See: Shirakawa (2011). 34

Former Unites States Secretary of Defense. Response to a question at a US department of Defense (Pentagon) News Briefing on February 12, 2002.

35

A low-probability but high-impact event, in the left tail of the probability distribution. 36

A term coined by economist John Maynard Keynes. See: Keynes, J.M. (1936). The General Theory of Employment, Interest

(26)

The Great Financial Crisis was a prototype of a tail event, or a sequence of rare events, and it heralded a global economic recession. Yet, the prevalence of crash risks was clearly overlooked prior to the collapse of Lehman Brothers (Orlowski, 2010). When comparing the pre-crisis with the post-crisis years, Straetmans and Chaudhry (2015) show that both realized tail- and systemic risk surged over time and De Bock and de Carvalho Filho (2013) report that investors became increasingly more worried about tail events (i.e. perceived tail risk). Empirical evidence suggests that market participants nowadays are willing to pay extraordinarily high prices for tail risk insurance, as demonstrated by a significant and sharp surge in the price of hedging instruments like risk reversal strategies in the decade post-GFC (Pederzoli, 2017; Anene and D ’Amico, 2017; IMF, 2012).

Crash risks threaten the stability of the financial system as a whole, thus mitigating both perceived and realized tail risk is essential in reducing systemic risk (i.e. a high probability of systemic financial collapse) (Kim and Zhang, 2014; Chava, Ganduri and Yerramilli, 2014; Orlowski, 2010). Since the outburst of the GFC, financial stability monitoring has become a key priority for many central banks and monitoring tail risks has emerged as a point of departure for devising appropriate monetary and macro-prudential policies (Rompolis, 2017; Lucas, Schwaab and Zhang, 2017; Chen, Cummins, Sun and Weiss, 2015; Orlowski, 2010). At times, central bank communication and loose policy have been used to provide a

backstop to destructive tail-scenarios (Anene and D ’Amico, 2017). Yet, accommodative monetary policy in itself might heighten tail risks endogenously when the market believes that a low interest rate environment will be maintained and a situation of abundant liquidity will continue (Shirakawa, 2011).

3. Related Empirical Literature

This section provides an overview of the related empirical literature which can be loosely grouped into three streams: research on the impact of UMP on tail risk perceptions,

research on the impact of UMP on other risk measures and research on the impact of UMP on closely related financial variables.

The empirical literature on the financial market impact of unconventional monetary policy has been growing continuously in recent years. Most emphasis for the euro area has been, thus far, on measuring the effectiveness of ECB announcements on bond yields

(27)

(Pattipeilohy, Willem Van Den End, Tabbae, Frost and De Haan, 2013; Rivolta, 2014), government bond spreads (Dewachter, Iania and Wijnandts, 2016; Falagiarda and Reitz, 2015), asset prices (Haitsma, Unalmis and De Haan, 2015; Hosono and Isobe, 2014), the exchange rate (Fratzscher, Lo Duca and Straub, 2014; Rogers et al., 2014) and international spillovers (Falagiarda, Mcquade and Tirpák, 2015; Georgiadis and Gräb, 2016; Varghese and Zhang, 2018). Yet, there is not much evidence on the transmission of unconventional monetary policy to perceived tail risks in financial markets, with the exception of some previous studies that almost exclusively focused on non-standard policy measures of the FED.37_{This paper attempts to fill the research gap by studying the impact of ECB UMP}

announcements on market perceptions of crash risks, in both the equity and foreign exchange market.

3.1 UMP and Tail Risk Perceptions

A first stream of literature focuses on the impact of non-standard monetary policy on the hedging costs of downside risk. There are only a few empirical studies devoted to this topic.

The Equity Market. The existing literature analyses the impact of FED UMP

announcements on the perceived risk of a stock market crash using S&P500 index options and an event-study approach (Hattori, Schrimpf and Sushko, 2016, 2013; Roache and Rousset, 2013). These studies show that loose UMP significantly reduced tail risk

perceptions in the equity market and that the relative impact is greater for expectations of near term risks (1 month) than over a longer horizon (2 and 3 months). Moreover, the papers report that the mitigating impact of the announcements is larger for tail risk proxies than for implied volatility indices. Suggesting that “the transmission of policy announcements to volatility measures originated from investors’ repricing extreme downside risks rather than from the overall level of volatility” (Hattori et al., 2013, p. 18). Both Hattori et al. (2016) and Roache and Rousset (2013) study changes in the S&P500 risk-neutral distribution, whereas Hattori et al. (2013) employ S&P500 risk reversals as a proxy for crash risk. In a related study, Anene and D’Amico (2017) use a slightly different approach. They examine the time period spanning from 2013 until 2016 (in lieu of 2008-2012) and rely on the difference between the implied volatilities for the S&P500 index options with different moneyness rather than the same (as is the case in the work of Hattori et al. (2013)). The results of Anene and D’Amico (2017) can be viewed as complementary to the empirical findings of Hattori et

37

See, for instance, Anene and D’Amico (2017) and Hattori, Schrimpf and Sushko (2013, 2016), who study the impact of FED UMP on the stock-market tail risk, and Scotti et al. (2015), IMF (2013) and Roache and Rousset (2013), who demonstrate the impact of UMP announcements on FX market participants' tail risk perceptions.

(28)

al. (2016, 2013) and Roache and Rousset (2013), as the 2013-2016 sample is dominated by the zero lower bound and taper/ exit announcements. The authors find that tightening monetary policy, a larger- than-expected increase in the U.S. short-rate expectations38_,

significantly increases perceived tail risk at a longer horizon (2 years) and significantly decreases perceptions of crash risk at a short horizon (2 months). The rate hike is

interpreted as positive news by investors, “as it might reveal that the Fed is more upbeat about the economic outlook than previously thought” (Anene and D’Amico, 2017).

Hattori et al. broaden the scope of their FED research to UMP announcements of the Bank of England (BoE) and the ECB (VLTRO and OMT) in the BIS Quarterly Review (BIS, 2013) and conclude that all three central banks were effective in decreasing tail risk perceptions in the market.39_{However, their analysis appears to be visual only. Hence, the results on the}

effectiveness of the BoE and ECB should be interpreted with some caution. Further evidence for Europe is also sparse and prior research exclusively focused on the pre-QE phase of UMP. To the author’s knowledge, Lucas, Schwaab and Zhang (2017) and Frommherz (2016) are the only two papers that examine the impact of ECB UMP

announcements on downside risk perceptions in the Euro Area equity market. Frommherz (2016) concludes that direct purchase programs such as CBPP and SMP have a calming effect on the risk assessment of the market, whereas instruments that indirectly intervene in the financial market trough improved lending operations (both LTRO and VLTRO) further increase tail risk in the broad stock market. Lucas, Schwaab and Zhang (2017), on the other hand, show that VLTRO announcements had a temporary decreasing effect on perceived crash risk in the financial sector and that OMT announcements ended the acute phase of heightened tail risk. Frommherz (2016) builds on the work of Hattori et al. (2016) and studies the effect of ECB UMP announcements on tail risks in the broad German stock market during the 2006 – 2013 period using an event-study methodology. Crash risk is measured using both risk neutral densities and risk reversals derived from DAX index options. When mixing up all UMP events, the risk-increasing effect of the (V)LTRO announcements seems to dominate. However, Frommherz’s conclusion is not entirely convincing; the empirical evidence is rather thin. The estimated SMP announcements effects are statistically not different from zero (or even positive in the extended analysis). Also, most of her conclusions are based upon a two-day event window but she fails to test for the joint significance of the

38

Daily changes in the tail risk proxy are regressed on the surprise effect of a policy rate change within an event-study framework. Short-rate expectations are measured by changes in OIS rates of different maturities in response to FOMC announcements.

39

(29)

days within the window and does not report their cumulative effect (often, the

announcement effects of day 1 and 2 move in opposite direction). Thus, this issue needs to be further explored. Lucas, Schwaab, & Zhang (2017) examine the equity market more closely by assessing the impact of the VLTRO and OMT announcements on financial sector tail risk. The authors developed a novel high-dimensional non-Gaussian modeling

framework to estimate joint default risk for Eurozone banks, insurers and investment companies.

The Currency Market. The impact of UMP announcements on foreign exchange market

participants’ tail risk perceptions has been previously assessed only to a very limited extent. IMF (2013), Roache and Rousset (2013) and Rogers, Scotti, and Wright (2015) asses the

impact of FED UMP announcements on perceived tail risk in the dollar exchange rate within an event study framework. With tail risk they are referring to a large exchange rate swing, a price change expected with 5 percent probability and downside risk in the dollar exchange rate, respectively. Both IMF (2013) and Roache and Rousset (2013) found evidence that

non-standard monetary policy easing reduces perceived tail risk in the exchange rate, although the risk mitigating effect diminishes over time.40_{Rogers, Scotti, and Wright (2015)}

study the impact of FED UMP over a longer time horizon (2008-2015) and conclude that loose policy changes the option-implied skewness in the direction of dollar depreciation. This seems contradicting with the reduction in perceived tail risk, but it reflects relative demand for puts and calls (e.g. perceived crash risk decreases, while there is still an investor bias for dollar devaluation due to the expected effectiveness of QE). Both IMF (2013) and Rogers, Scotti, and Wright (2015) employ euro - dollar risk reversals as a proxy for crash risk, whereas Roache and Rousset (2013) study changes in the euro – dollar exchange rate risk-neutral distribution (RND).

An episode of abrupt exchange rate depreciation is equivalent to a period of heightened crash risk. The cost of insuring against extreme moves in the euro – dollar exchange rate, as measured by risk reversals and RNDs, provides valuable information regarding the

probability of such a crash or other tail scenarios in the foreign exchange market like a euro break-up (Gamboa-Estrada, 2017; Jurek, 2009). The possibility of a Euro Area collapse during the sovereign debt crisis led to a period of ultra high redenomination risk, i.e. the tail risk of an abandon of the euro for new domestic currencies and countries redenominating

40

IMF (2013) reports that earlier announcements in the UMP cycle appear to have had the largest effects for both the FED and the BoE.

A Tale of Two Tails: The Impact of ECB Unconventional Monetary Policy on Perceived Crash Risk