The response of tail risk perceptions to unconventional monetary policy : an analysis for the Euro area

(1)

0

THE RESPONSE OF TAIL RISK PERCEPTIONS TO

UNCONVENTIONAL MONETARY POLICY

AN ANALYSIS FOR THE EURO AREA

ANNELIE PETERSEN

11120185 August 2016

MSc in Economics, Amsterdam School of Economics Monetary Policy, Banking & Regulation

Master’s thesis under the supervision of prof. dr. Sweder van Wijnbergen

KEYWORDS

Unconventional Monetary Policy, Tail Risk Perceptions, Euro Area, Extreme Downside Risk, Option, Skewness, Risk Reversal

(2)

1

1_{I would like to express my gratitude to Andrei Lalu and Folkert Mulder for the useful comments, remarks and}

engagement through the data process of this master thesis. I am also grateful to my supervisor, prof. dr. van Wijnbergen, for useful suggestions throughout the writing of this thesis.

ABSTRACT

1

The reversal of the housing boom and the collapse of the sub-prime mortgage market in the US resulted in a crisis of global dimension in 2008. Currently, major central banks have implemented unconventional policy measures to deal with the economic and financial crisis that materialized in the aftermath of the bursting of the global credit bubble. This study evaluates the response of tail risk perceptions in financial markets to the implementation of unconventional monetary policy by the European Central Bank. This paper employs an option-implied empirical proxy for downside risk perceptions, referred to as risk reversals, and uses information gleaned from out-of-the-money equity index (Euro Stoxx 50) options. By means of an event-study framework, the results of this study suggest that UMP announcements did not ameliorate investor expectations associated with extreme downside risks. As a matter of fact, the ECB’s monetary policy even seems to have aggravated tail risk perceptions

.

STATEMENT OF ORIGINALITY

This document is written by Student Annelie Petersen 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)

2

“THE PROBLEM WITH QUANTITATIVE EASING IS THAT IT

WORKS IN PRACTICE, BUT NOT IN THEORY ”

(4)

3

1. INTRODUCTION

The reversal of the housing boom and the collapse of the sub-prime mortgage market in the US resulted in a crisis of global dimension in 2008. Within the euro area, the economic and financial collapse escalated into a sovereign debt crisis in 2010 (Fratzscher, Duca & Straub, 2014). Consequently, major central banks dropped their policy rates towards an effective zero

lower bound (ZLB) until they could meaningfully ease no further. In other words, the

conventional measures had largely lost their potency. Currently, major central banks have implemented unconventional policy measures to deal with the economic and financial crisis that materialized in the aftermath of the bursting of the global credit bubble (Pattipeilohy et al., 2013; Arteta et al., 2015). Consequently, there is a prevailing belief that unconventional policy actions undertaken by various central banks over the past years have helped in alleviating some of the most immediate downside risks to both the financial markets and the global economy (Hattori et al., 2015).

Previous studies mainly introduce and predate the ongoing unconventional policy measures by the European Central Bank (ECB) and these authors generally concentrate on the transmission of conventional policies, i.e. changes in short-term policy rates, to the broader economy. The beginning of the literature on the impact of recent unconventional measures, however, mainly focuses on other variables like the slope of the yield curve and asset prices (Gagnon et al., 2015). Thus far, the literature remains relatively silent on the impact of Unconventional

Monetary Policy (UMP) on risk appetite and risk perceptions and the evidence provided is

mainly anecdotal. This study, by contrast, is specifically interested in evaluating the impact of recent unconventional measures on perceived tail risks2 and therefore contributes to the understanding of market assessment of downside risks and whether it changes around nonconventional policy innovations.

For central banks to mitigate both the perception of tail risk and realized tail risk is essential in restoring investor confidence and stabilizing stock markets, since tail events quickly spread panic across financial markets causing a downward spiral of declines in a broad spectrum of investments (Kim & Zhang, 2014). Importantly, these changes in beliefs prevail long after the event itself has passed and since tail risk perceptions affect both prices and choices, this persistent change in beliefs produces long-lasting effects on investments,

2_{In this master thesis, tail risk represents the loss at the most negative part of a distribution, or, in other words,}

(6)

5

employment and output (i.e. the real economy). The goal of this thesis is to examine the broader impact of the unconventional policies, in this case on the perception of tail risk by market participants in the euro area. Therefore, the research question of this study is

‘‘What are the effects of the ECB’s Unconventional Monetary Policy Measures on Tail Risk

Perceptions in the Euro Area?’’

This paper relates to a number of strands of the empirical literature studying the impact of central banks’ unconventional policies on financial markets, using daily data. Daily data allows for a more precise identification of the effects of unconventional monetary policy on financial variables (Wright, Scotti & Rogers, 2014). This empirical study analyses and quantifies the impact of the most important ECB’s non-standard policy measures on tail risk perceptions in the euro area. We gauge downside risk perceptions using information extracted from

out-of-the-money (OTM) equity index (EURO STOXX 50) options. Specifically, we rely on the

difference between the option-implied volatilities of (deep) OTM puts and calls of the same maturity and ‘moneyness’, often referred to as a Risk Reversal (RR). RRs therefore serve as a proxy of how market participants perceive the risk of a stock market crash. In essence, this measure captures the option-implied skew in the equity return distribution and the associated skewness risk premium (Fratzscher, Duca & Straub, 2014).

Although managing tail risk is not part of the ECB’s (inflation) mandate and the ECB does not focus on developments in stock markets, it can be an effect of the unconventional monetary policy measures as conducted by the ECB in the last couple of years (i.e. UMP can have spillover effects). The effect of non-conventional policy measures on tail risk has only been researched upon in the United States, whereas other existing literature in the field mainly focuses on the effect of UMP on the yield curve and on asset prices. The main contribution of this master thesis is to elucidate why tail risk fluctuates and whether monetary policy, both conventional and unconventional, can mitigate tail risk. This paper therefore attempts to contribute to the gap in the empirical literature on tail risk perceptions in the euro area.

The modelling strategy in this study consists of an event study methodology (i.e. using impulse dummies) to capture the announcement effects of policies. The results of this study suggest that UMP announcements did not ameliorate investor expectations associated with extreme downside risks, suggesting that the transmission of the QE announcements to volatility measures originated from investors’ repricing of the overall level of volatility rather than the

(7)

6

extreme crash risks. As a matter of fact, monetary policy as conducted by the ECB even seems to have aggravated tail risk perceptions.

The remainder of this paper is organized in the following way. To set the stage, Section II provides a theoretical framework explaining the volatility smile and skew. Section III provides a brief account of the ECB’s unconventional policy measures along with the main transmission channels and the link to the real economy. Section IV specifies the empirical proxy extracted from derivatives prices as employed in this paper along with the empirical method. Section V describes the data process. Section VI presents the results of the event-study framework. Section VII provides a discussion of the results and suggestions of further research. Section VIII concludes.

2. THEORETICAL FRAMEWORK

2.1 Why options smile

As the recent crisis unfolded, international investors became increasingly concerned with the performance of portfolios in distress events, or so-called tail events in the market. Essentially, tail risk may play an important role in asset pricing. A centerpiece of the financial pricing paradigm is option-pricing theory, which has witnessed tremendous growth of activities over the past two decades using options and other derivatives. Traditional models for stock returns, including the Black-Scholes option pricing model, assume that returns follow a geometric Brownian motion. In essence, this implies that over any discrete time interval, the return on the underlying assets is lognormally distributed and the volatility of the underlying is constant. However, empirical studies indicate in particular that the Black-Scholes approach fails to capture more extreme price movements and stochastic variability in the volatility parameter. Consequently, stock return distributions are skewed and kurtotic relative to a normal distribution. In other words, the empirical asset return distribution in this paper is assumed to exhibit fat-tails (kurtosis) and skew (van Oordt and Zhou, 2014; Hardy, 2001).

If the underlying risk-neutral distribution of stock returns were purely log-normal, options that expire on the same date should have the same implied volatility regardless of the underlying strike prices. However, implied volatilities strongly differ among various strike prices and this discrepancy is known as the volatility smile. In particular for a given expiration, the strike price bias implicates that options with strike prices further away from the price of the underlying

(8)

7

asset command higher prices and consequently yield higher implied volatilities. The Black-Scholes model therefore systematically misprices (deep) in-the-money and (deep) out-of-the-money options. In general, at-the out-of-the-money options tend to have lower volatilities than (deep) in- or out-of-the-money options.

FIGURE 1: THE VOLATILITY SMILE

Notes: in this graph K stands for the Strike Price of an option, whereas S refers to the stock

price of the same option. Volatility increases as the option becomes increasingly in-the-money or out-of-the –money.

Source: Author’s graph based on Nowak & Sibetz (2012)

Graphing implied volatilities against strike prices, for a given expiration date, therefore yields a volatility smile (see figure 1). American markets did not show a volatility smile until the stock market crash of 1987. Kozlowski et al. (2015) explain this phenomenon by stating that transitory shocks have persistent effects on beliefs since the shocks remain in the agents’ dataset. Agents thereby re-estimate the distribution from which the data is drawn, since the experience permanently changes their assessment of risk. Extreme events (e.g. tail events), which are rare, therefore lead to permanent changes in beliefs and long-run outcomes. Hence, it is believed that investors reassess the probability on a left-tail event, which ultimately leads to higher prices of out-of-the-money options. Additionally, investors may have a preference to write either call or put options generating a disparity in put and call volatilities. In this case, the implied volatility pattern that arises is called a volatility skew. Essentially, this study focusses on the volatility skew and whether it changes around unconventional monetary policy moments (Corrado & Su, 1996; Christoffersen, 2012).

(9)

8

3. LITERATURE REVIEW

This section provides a brief account of the monetary accommodation by the European Central Bank and related empirical literature along with the transmission mechanism of monetary policy to tail risk perceptions. Lastly, it explains the effects of tail risk perception on the real economy.

3.1 Overview of Policy Innovations by the European Central Bank

Since the eruption of the global financial crisis in the end of 2007, central banks of hard-hit advanced economies dropped policy rates towards an effective zero lower bound until they could meaningfully ease no further. However, market stress still remained high whilst the main policy transmission tools had been exhausted (de Pooter et al., 2015). Hence, central banks around the world have adopted accommodative unconventional monetary policies to help support real economic activity and bolster economic growth (Arteta et al., 2015).

Nowadays, nearly all major central banks have several monetary policy tools at their disposal, as materialized in the following categories: short-term interest rate policies, ‘’helicopter drops’’ of money, asset purchase programs and collateral policies for lending programs. Whereas the short-term interest rate policies are considered to be the main policy transmission tool. The potency and effectiveness of these tools depend on the credibility of the central bank about its future behaviour conditional on the state of the economy (Brunnermeier & Sannikov, 2012; Micossi, 2015). Once the financial crisis struck, peripheral countries (i.e. the GIIPS countries: Greece, Ireland, Italy, Portugal and Spain) experienced remarkable surges in their sovereign bond yields. When the crisis progressed and intensified, special instruments and facilities were introduced for refinancing both banks and sovereigns under severe stress. These non-standard measures focused on restoring a proper liquidity allocation (Falagiarda & Reitz, 2015; Micossi, 2015). In essence, the ECB introduced their programs to lower the government bond yields. The ECB programs included direct purchases of government debt, conditional commitment to purchase government debt (SMP, OMT) and lending to banks in expectation that they would buy government bonds (LTRO).

Early on in the crisis the ECB undertook some measures in order to support the interbank money market in the euro area: (i) an unrestricted liquidity provision through fixed rate tenders

(10)

9

with full allotment (FRTPFA), which allowed banks to get unlimited access to central bank liquidity at the main refinancing rate, contingent on proper collateral; (ii) extension of the list of acceptable collateral assets for refinancing operations (COLL); (iii) extension of the maturity of long-term financing operations (LTRO) in order to remove uncertainty and enhance liquidity conditions for banks; (iv) liquidity provision in foreign currencies through currency swap agreements with other central banks to improve bank’s foreign currency funding (FOR) (Falagiarda, McQuade & Tirpák, 2015).

Enhanced Credit Support (ECS)

In May 2009 the ECB adopted a new programme. The Enhanced Credit Support (ECS) programme included the four types of liquidity operations as highlighted above and the central bank extended its operations to the covered bond market. The so-called Covered Bond Purchase Programme (CBPP1) involved the outright purchase of covered bonds and was complemented in November 2011 by CBPP2. The program aimed to strengthen the functioning of the covered bond market in the euro area, which played an important role in the refinancing of banks (Falagiarda, McQuade & Tirpák, 2015).

Securities Market Programme (SMP)

On May 10, 2010, the ECB established a program particularly constructed to address sovereign-debt tensions by directly purchasing government debt of distressed countries. The so-called Securities Market Programme (SMP) involved the purchase of unspecified amounts of euro area sovereign debt to ensure depth and liquidity in ‘’dysfunctional’’ secondary markets. While the program announcement stated that purchases might include both private and sovereign debt, only the latter has been purchased. As previously mentioned, this was in response to the European Sovereign Debt Crisis which started from Greece and contaminated the whole EMU. Hence, speculation was growing that one or all of these GIIPS-countries would be forced out of the monetary union, thereby increasing the capital outflows from southern to northern euro area members, fragmenting the credit markets across national borders and lowering consumer and investment demand. These developments passed through to the real economy and lowered output and employment along with the negative impact on public finances and balance sheets of banks. During the first phase of the program, which comprised the time span from May 2010 up till April 2011, the program consisted of purchases of Greek, Irish and Portuguese debt. A period of relative market calm, entailing the summer of 2011, resulted in the fact that the program was mostly inactive in that period. On August 7, 2011, the

(11)

10

ECB expanded the program due to the rising peripheral borrowing costs and the deterioration of market perceptions concerning the creditworthiness of Italy and Spain. Consequently, the SMP operations have been carried out in two big waves: when referred to the first SMP the period from May 2010 to April 2011 is considered whereas the second SMP comprises the period from August 2011 till February 2012 (de Pooter et al., 2015; Falagiarda & Reitz, 2015; Micossi, 2015).

Outright Monetary Transactions (OMT)

On September 6, 2012, the ECB discontinued the SMP and substituted it by the Outright

Monetary Transactions (OMT) program whereby the ECB would be willing to intervene for

unspecified and unrestrained amounts in secondary sovereign-bond markets on request of a government demanding for financial assistance. With this announcement, the ECB succeeded in credibly positioning itself as a lender of last resort standing behind euro area sovereigns in case of large idiosyncratic financial shocks, using funds of the European Financial Stability

Facility (EFSF) or the European Stability Mechanism (ESM). These interventions, however,

would only be initiated after the bond-issuing country agreed to an economic programme with the EFSF or ESM, entailing specific domestic measures (e.g. fiscal adjustments). So the OMT programme, de facto, involves a strict and effective conditionality principle, which could materialize in full macroeconomic adjustment programmes or ‘precautionary’ programmes. In terms of the declared objective of the program the ECB’s motivation was to safeguard ‘’an appropriate monetary policy transmission and the singleness of the monetary policy’’3_through

lowering bond yields. In principle, the ECB tried to flatten the yield curve by influencing the long-term yields which ultimately has a downward effect on sovereign borrowing costs. Additionally, the OMT programme fosters confidence to investors in sovereign-bond markets. In fact, the program implementation has never been tested and financial markets incorporated the ECB announcement at face value. Yet, even though the official announcement of the program was on September 6, there were earlier indications regarding the likelihood of the introduction of a new monetary program of such kind. Firstly, on July 26, 2012, Draghi stated the following well-known sentences at a conference:

‘’Within our mandate, the ECB is ready to do whatever it takes to preserve the euro. And believe me, it will be enough.’’ Draghi, 2012a

(12)

11

Moreover, the introductory statement of the ECB to the August 2, 2012 press conference indicated that the Governing Council considered undertaking additional non-standard policy measures in order to repair the policy transmission (Draghi, 2012b). Consequently, the market may have already expected a program of such kind to be implemented. In short, it can be noted that the OMT was communication without intervention, contrary to the SMP, which mainly entailed intervention with little transparent communication (de Pooter et al., 2015; Falagiarda & Reitz, 2015; Micossi, 2015).

3-year Long-Term Refinancing Operations

On December 8, 2011, the ECB announced Long-Term Refinancing Operations (LTROs), whereby banks obtained full allotment loans against a variety of collateral, with a collateral-specific haircut schedule. The loans are usually backed up by collateral through the national central banks of the particular banks. Essentially, full allotment loans indicate that banks do not face quantity limits on their borrowings. The objective of the LROs is to avoid a severe credit crunch or collapse of the banking system as Eurozone banks are strapped for cash. This program complemented the Eurosystem’s regular open market operations, also known as the

Main Refinancing Operations (MRO) , which mainly served to both steer short-term interest

rates and signal the monetary policy stance, along with the management of liquidity. The LTROs therefore provide longer-term refinancing to the financial sector (ECB, 2016a)

Forward guidance

Since July 2013 the ECB embraced forward guidance (FWG) as a new monetary policy communication strategy intended to bolster confidence. Clear communication by the central bank about the monetary policy strategy has been used as a tool in order to avoid surprises that might disrupt financial markets and cause significant asset price fluctuations. Hence, FWG influences the expectations formed by the public. Put differently, investors can have greater confidence in their spending and investing decisions as they know where the economy might be headed (Falagiarda, McQuade & Tirpák, 201).

Credit Easing Package

In June 2014 the ECB announced additional measures, usually referred to as the Credit Easing

Package (CEP), in order to strengthen the monetary policy transmission mechanism by

supporting lending to the real economy. The CEP was based on two pillars: (i) a series of

(13)

12

the non-financial private sector of the euro area; (ii) outright purchases of asset-backed

securities (ABSPP) in parallel with a new covered bond purchase programme (CBPP3)

(de Pooter et al., 2015; Falagiarda, McQuade & Tirpák, 2015).

Expanded Asset Purchase Programme

Since January 2015 the ECB adopted the expanded Asset Purchase Programme (APP), henceforth QE, which complements the existing private sector asset purchase programmes (CBPP3 and ABSPP) by a purchase programme for public sector securities (PSPP). These extensions served to address the risks of a too prolonged period of low inflation. In particular, under the PSPP the ECB purchased sovereign bonds of euro area countries in distress to restore the transmission mechanism of monetary policy and curb soaring risk premia (Fratzscher, Duca & Straub, 2014; Falagiarda, McQuade & Tirpák, 2015)

3.2 Unconventional Monetary Policy Transmission.

Since the ECB embraced unconventional policy, there has been a growing body of empirical literature to analyze its transmission mechanisms. However, the efficacy of the unconventional monetary policy as conducted by the ECB are only marginally understood at best (de Pooter et al., 2015). In principle, the literature discusses several transmission channels that underpin the global financial market impact of the ECB’s non-traditional policies. Certain main transmission channels are the signaling, portfolio-rebalancing and market functioning channel.

According to the ‘‘signaling channel’’, central bank announcements can signal commitment to monetary stimulus, which subsequently will lower the expected path of future short-term rates. This channel emphasizes the role of private agents’ expectations of future economic conditions and policy actions (Hattori et al., 2013; Falagiarda & Reitz, 2015).

In a related vein, in the portfolio-rebalancing channel central banks influence yields by influencing the relative supplies of assets with different maturities. This channel emphasizes the imperfect substitutability among different assets (i.e. assets are not perceived as perfect substitutes by investors) and reflects a degree of market segmentation. Moreover, agents are heterogeneous and hold different portfolios. Consequently, purchases carried out by the central bank will entail a rebalancing of investors’ portfolios. This is, investors with preferences for certain bonds with specific maturities or safety characteristics (i.e. ‘‘preferred habitat investors’’) switch to close substitutes of the purchased securities. In essence, the purchase of a particular security by the monetary authority reduces the amount of that security held by

(14)

13

private agents. As a result asset prices rise and interest rates decline. This creates a more favourable economic environment (Hattori et al, 2015; de Pooter et al., 2015; Falagiarda & Reitz, 2015).

Another important channel of unconventional monetary policies is the so called liquidity premia channel, also referred to as the market functioning channel. In times of financial panic, markets usually are characterized by poor market liquidity and therefore assets contain a higher risk premium. The presence of a central bank acting as a protagonist in the financial markets could substantially enhance market functioning and liquidity which subsequently leads to lower risk premia. This new role of the central bank might make investors more willing to behave actively in markets, knowing that they could sell the financial products to the monetary authority if necessary. This could in turn reduce the liquidity risk premiums embedded in asset prices and thereby lowering their yields (Falagiarda & Reitz, 2015; de Pooter et al., 2015). Nonetheless, transmission channels are not mutually exclusive and can work in parallel (Fratzscher, Duca & Straub, 2014).

In order to shed light on the effects of monetary policy on investors’ perceptions in the stock markets in the euro area, it is needed to understand the operative transmission mechanism to the real economy of monetary policy decisions. While existing literature of the transmission channels of asset purchases tends to focus on core variables of monetary policy like money market rates (Angelini et al., 2011), interbank rates (Abbassi & Linzert, 2011) covered bond markets (Beirne et al., 2011), and the slope of the yield curve (Gagnon et al., 2011), more recent work has started looking at the broader transmission of monetary policy to other assets and markets (Hattori et al., 2015). Hence, these other studies look at broader transmission via their impact on financial sector risk-taking. Several empirical studies of monetary policy transmission studying the area of macro finance include Bernanke and Kuttner (2005). In their research they find an empirical link between the federal funds target rate and returns of stock markets, namely a hypothetical unanticipated 25 basis-point rate cut would lead to an increase in stock market prices of 1%. Put differently, higher interest rate reduce equity prices meaning that tight money may reduce the willingness of stock investors to bear risk. Birru and Figlewski (2010) found that announcements concerning the federal funds rate targets are associated with declining uncertainty in comparison with ordinary days, thereby measuring changes in the S&P500 risk-neutral distribution. In turn, Bekaert, Hoerova & Duca (2013) decompose the VIX to test the hypothesis that expansionary monetary policy can be successful in curbing risk aversion.

(15)

14

Similar to this study, these papers try to estimate the effects of monetary policy beyond the immediate impact on the yield curve and therefore try to identify the broader policy impact of unconventional monetary policy. In particular, the effect on market participants’ attitudes towards risk will be highlighted. Before considering the results of unconventional monetary policy on tail risk perceptions, it is important to think in terms of the specific transmission mechanisms at work. The response of equity market tail risks accentuate the risk-taking channel of monetary policy as the commitment to monetary easing (i.e. low funding rates) may have relaxed the risk-bearing constraints of financial intermediaries (Hattori et al., 2015). Another specific example is the suggested transmission mechanism of Brunnermeier and Sannikov (2012), who suggest that central bank purchases can serve as insurance against tail events in combination with clear communication and commitment. In their model the central bank can signal its commitment to redistribute tail risk, which can have an immediate effect on market expectations and pricing of downside risks. Moreover, Birru and Figlewski (2010) argue that non-conventional policy announcements possibly reduce uncertainty, which curtails the fear of a collapse of the stock markets and consequently leads to a decline in the option-implied perception of tail risk. Finally, the research of Bekaert, Hoerova & Duca (2013) argue that monetary easing may dampen risk aversion, which is closely related to - but not the same as - perceived tail risk. However, an increase in risk appetite and financial sector risk-taking suggests a decrease in the insurance against tail-events and therefore a decrease in perceived tail risk.

3.3 Tail risk as a macro risk

‘’the main aim of the OMT is to remove tail risk to overcome monetary and financial fragmentation of the euro area that would stem from a redenomination risk. And we would do

it in a size that would be adequate to achieve its objective.’’

Draghi, 2012c

The recent economic meltdown has sparked a renewed interest in (left) tail risk. Traditional economics and portfolio investment strategies often rely on normal bell-shaped curves where the most probable market states are centered in a bulge. The ‘tails’ on the edges, both left and right, represent the low probability, high impact events (i.e. busts on the left, booms on the right). Due to the unpredictable nature of financial markets and the fact that unforeseen events

(16)

15

are everywhere (financial crises, terrorist attacks and pandemics), tail risk has been a frequent topic of conversation amongst market participants. These unpredictable market shocks resulted in ‘‘fatter tails’’ and caused uncertainty, which is not desirable in an economy still recovering from the financial crisis. Prominent economists, such as Olivier Blanchard, therefore argue that mitigating both the perception of tail risk and realized tail risk is essential in restoring investor confidence and stabilizing stock markets (Kim & Zhang, 2014). In essence, tail events quickly spread panic across financial markets, which can cause a downward spiral of declines in a broad spectrum of investments. In a related vein, the 2008 financial crisis can be marked as one of the deepest downturns with long-lasting effects on labor market, credit markets and output. For example, in 2005, no one ever raised the possibility of financial panic on this scale. Now, the experience of this extreme crisis changed the agents’ assessment of risk. Some investors, like underfunded pension funds, have limited resources to withstand another market shock. Consequently, option prices still reflect a heightened probability of tail events due to the fact that, given this backdrop and these market fears, ‘tail risk’ hedging against extreme downside risk is in demand. Importantly, these changes in beliefs prevail long after the event itself has passed and since tail risk perceptions affect both prices and choices, this persistent change in beliefs produces long-lasting effects on investments, employment and output (i.e. the real economy). Therefore we could argue that tail risk perceptions are indeed important for the real economy (Kozlowski, Veldkamp & Venkateswaran, 2015; Gerstein, 2012).

Moreover, Bhansali (2008) argue that in a finance-driven economy such as the U.S. (i.e. also applicable to Europe), tail events can be hedged with macro instruments which respond to central bank policies. Subsequently, both easing and tightening of monetary policy largely impacts the real economy through recessions and expansions. Therefore, monetary policy may influence the tails which develop during the presence of regimes of alternately conventional and nonconventional monetary policies. More specifically, the overarching conclusion from Roache & Rousset (2013) is that both tail risk and the perception of tail risk diminishes in the immediate aftermath of an event that serves to ease monetary policy through unconventional means. Since monetary policy affects tail risk perceptions, tail events have become a macro event (Bhansali, 2008). Hence, monetary policy could definitely be seen as an insurance against tail risks.

(17)

16

4. METHODOLOGY

4.1 Empirical Proxy for Tail Risk Perceptions

This section clarifies all the main ingredients in order to fully comprehend the empirical proxy for tail risk perceptions. It commences with an explanation on the different volatility measures along with the main indices. Then, it explains the proxy for tail risk perceptions along with the implications of the precondition of equal moneyness.

4.1.1 Conceptual difference between implied volatility and historical volatility

.

We rely on proxies for tail risk perceptions computed from option prices on the EURO STOXX 50 index. Besides these, we also reckon with some other well-known measures of volatility. It is, however, crucial to understand the concept of volatility, which can be seen as a measure of the level of uncertainty prevailing in various markets. In essence, there are two distinct approaches to measure volatility. Volatility can either be historical or implied. On the one hand, historical volatility comprises estimating the standard deviation of historical closing prices for any specific security over a given timeframe. Implied volatility, on the other hand, is derived from option prices and this type of volatility displays the estimates and assumptions of market participants engaged in a trade, based on a given option price (Gómez-Puig & Sosvilla Rivero, 2014). In principle, a stock index option has a valuation model with a number of parameters. All but one of them are known or are able to be estimated with a high degree of accuracy. The unknown parameter is the index’s expected future volatility. By equating the market price of this index option to its model value and solving for volatility, we may single out the implied (by the option price) volatility (Whaley, 2000). Therefore market participants consider implied volatility indices as a meaningful tool for measuring investors’ sentiment. Accordingly, Sirioupoulos & Fassas (2009) reason that implied volatility indices encompass information about possible future volatility beyond that incorporated in historical volatility.

4.1.2 Implied volatility indices

In the same line of thought and for comparison purposes we may therefore look at volatility indices based on implied volatility. These volatility indices are referred to as fear index or investor fear gauge and are a direct measure of expected stock market risk. The higher their value, the greater the fear and consequently the larger is market uncertainty (Whaley, 2000; Grimaldi, 2010).

(18)

17

This can also be seen in figure 2. Investors set the level of these implied volatility indices, albeit indirectly. Market participants demands for call and put options set prices and these prices, in turn, are used to imply these indices (Whaley, 2000).Whereas the VIX is the best measure of uncertainty in the U.S. stock market, Standard and Poor’s 500 (S&P500), the VSTOXX is the best measure of uncertainty for euro area stocks and is based on the EURO STOXX 50. However, we cannot use these implied volatility indices as a proxy for tail risk, since they do not focus solely on the downside risk, as it is a symmetric measure of risk (Hattori et al., 2015).

4.1.3 Risk Reversals

Risk Reversals In contrast to the VSTOXX index, we therefore need a non-symmetric

measure which puts its focus more on the crash risk of euro area stock markets. Risk Reversals (RR) can be a suitable proxy for capturing the extent to which the equity return distribution is asymmetric (left-skew). Hence, it is a relevant proxy for picking up tail risk in equity markets. Risk reversals are a measure of the cost of hedging against downside risk and are generally

FIGURE 2: SHARPLY RISING EQUITY MARKETS HAVE OFTEN BEEN ASSOCIATED WITH LOW IMPLIED VOLATILITY

Notes: This figure shows the development of the EURO STOXX 50 price and volatility index over time. The left axis indicates the price index, whereas the right axis illustrates the volatility index

(19)

18

expressed as the difference between the implied volatility (IV) of an out-of-the-money (OTM) call and the IV of an OTM put of the same ‘’moneyness’’ and maturity. These risk reversals are used as an indicator of how market participants perceive the risk of a stock market crash. In essence, this measure captures the option-implied skew in the equity return distribution (i.e. the concept of tail risk suggests that the distribution is not normal, but skewed, and has fatter tails). As highlighted before, the difference with other option implied volatility indexes, for example the VIX index, is that this VIX index does not specifically capture downside risks, as it is considered to be a symmetric measure of expected volatility. Following Hattori et al. (2015) the risk reversal will be estimated as follows:

𝑅𝑅_𝑥𝛿 = 𝐼𝑉_{𝐶𝐴𝐿𝐿}𝑥𝛿 − 𝐼𝑉_𝑃𝑈𝑇𝑥𝛿 (1)

Where 𝛿 represents the option’s delta, which indicates the sensitivity of the value of the option to the price of the underlying, and 𝑥 represents the option’s maturity. The implied volatility of an OTM call and an OTM put will be the same in case of a symmetric distribution (i.e. the risk reversal will be zero). Risk reversals will only be non-zero if the (risk-neutral) distribution of equity is skewed. In the case of negative tail-risk there will be a negative skew, which indicates that the IV of the OTM put is higher than the IV of the call. In this case out-of-the-money puts have a higher likelihood of being exercised than out-of-the-money calls, meaning that their price or IV is higher. In essence, a stock index put option can be used to offset the losses on an investor’s portfolio in a market downturn, which makes it rather logical to purchase a put option when markets are expected to experience a downfall (Brunnermeier, Nagel & Pedersen, 2008; Hattori et al, 2015; Campa, Chang & Reider, 1998). Figure 3 shows the time series of 10δRR and 25δRR, expressed as absolute values

Importantly, according to Hattori et al. (2015) these proxies exhibit high and significant correlation with other (perhaps more direct) measures of tail risk. Although risk reversals do not immediately measure the probability mass in the negative tail of the implied distribution itself, it perfectly captures the perceptions of tail risk in stock and option markets. On top of this, other more precise or direct measures require more crude information in order to construct the entire option-implied densities. This information tends to be available only at lower frequencies, which renders them unsuitable for analysing the effects of unconventional monetary policy. Additionally, the development of academic literature established an

(20)

19

increasingly important role for derivative markets as a source of supplementary information, instead of the fixation on the underlying time series of, among others, stock prices (Dunis & Lequeux, 2001; Campa, Chang & Reider, 1998).

FIGURE 3: MEASURES OF TAIL RISK

J

Notes: This figure shows the tail risk gauges used in this study. Risk reversals are typically expressed as the difference in

implied volatility of an OTM call and the implied volatility of an OTM put of the same moneyness and maturity. In this study, risk reversals for two degrees of moneyness are considered, 25δ options (OTM) and 10δ options (deep OTM). Because of the negative skew in equity prices, OTM puts tend to be more expensive than OTM calls, hence the risk reversals are negative. For ease of exposition, this graph shows the absolute values of risk reversals and therefore the series in the figure are positive. Vertical lines indicate big economic events.

Source: Bloomberg | Author’s own graphs

4.1.4 Moneyness

The moneyness of an option refers to the relative position of the current price of an underlying asset (in this case the EURO STOXX 50 index) with respect to the strike price of an option. Therefore, moneyness indicates whether exercising leads to profits. In this case:

𝑀𝑜𝑛𝑒𝑦𝑛𝑒𝑠𝑠 = 𝑆 𝐾 =

𝐸𝑈𝑅𝑂 𝑆𝑇𝑂𝑋𝑋 50 𝑃𝑅𝐼𝐶𝐸 𝐼𝑁𝐷𝐸𝑋 𝑆𝑇𝑅𝐼𝐾𝐸 𝑃𝑅𝐼𝐶𝐸 𝑂𝑃𝑇𝐼𝑂𝑁

(2)

In essence, moneyness is a threefold classification. The option is said to be in-the-money (ITM) if the derivative would make money if it were to expire today. If the option would not make money it is said to be out-of-the-money (OTM), while if the current price and strike price are equal, it is said to be at-the-money (ATM). Thus, an option has positive moneyness if it is ITM, zero moneyness when it is ATM and negative moneyness if it is OTM. When quantifying moneyness, there are two conventions depending on the type of option (i.e. call or put).

(21)

20

Essentially, the moneyness of ITM and OTM strikes is inversely related for calls and puts. Call

moneyness increases if the spot price of the underlying increases relative to the strike price of

the option, whereas put moneyness increases when spot decreases relative to the strike (Christoffersen, 2012; Siriopoulos & Fassas, 2009).

When quantifying an empirical proxy for tail risk perceptions, it is wise to look at deep OTM options. For example, one could define tail risk as those events in which the asset price moves more than three standard deviations from its current price. In order to shed light on perceptions of downside risk, i.e. moving more than 3 standard deviations below its current price, it is necessary to focus the sample directly in the tail. Therefore we need far OTM options, since the other options carry no or little information of tail events (Skoglund & Chen, 2015). Thus, considering risk reversals based on options which are OTM or deep OTM provides that it would take an extreme swing in stock prices for such options to end up in-the-money and be exercised, considering their (very) low sensitivity to incremental movements in the EURO STOXX 50 (Hattori et al., 2015)

Another closely related measure of moneyness is the delta of an option. Whereas the moneyness of an option is defined as the difference between the underlying asset price and the exercise price, the option’s delta is encountered as the option price sensitivity to the underlying asset. Additionally, the delta can be regarded as an approximation to financial markets’ assessment of the probability that the option ends up in-the-money. In more elementary terms, the delta increases along with the chance that an option will be exercised. In essence, deltas under 0.5 are OTM strikes, ATM strikes have deltas at about 0.5 and ITM strikes have deltas above 0.5. Moreover, deltas closer to 0 represent strike prices that are deeper-out-of-the-money. Similarly, deltas closer to 1 comprise deeper in-the-money strike prices.

Since the delta is a different metric for moneyness, this study builds it data on risk reversals with the same maturity and delta. We consider risk reversals for two degrees of moneyness in our analysis, 25δ options (OTM) and 10δ options (deep OTM). In today’s markets, smiles are often asymmetric (i.e. volatility skew) as one can observe significant risk reversals. As illustrated in figure 4, risk reversals are a good measure of the volatility skew. In this figure Nowak & Sibetz (2012) show the conceptual difference between the 10δRR and the 25δRR whereas the two additional points of 10 delta options yield a better calibration as far out of the money options possibly can have even higher than extrapolated implied volatility, yielding higher risk reversal values.

(22)

21

FIGURE 4: VOLATILITY SKEW

Notes: This figure shows that the risk reversal can serve as a measure of the volatility

skew, which can be build using the points in the volatility surface for any time to expiry. In essence, the 10 delta option points yield a better calibration as far out of the money options can have higher than extrapolated implied volatility. The figure graphs the risk reversal as the difference between volatilities, in delta terms, for call and put options that are equally out of the money.

Source: Author’s graph based on Nowak & Sibetz (2012)

4.2 Empirical framework

This section explains the methodology of this study.

4.2.1 Regression Analysis

As previously mentioned, equity options on the S&P 500 did not show a volatility smile before the Crash of 1987 (Jackwerth & Rubinstein, 1996). Conceptually, the volatility smile implicates that large downward movements are more common than normal distributions imply. Hence, Han (2008) argue that investor sentiment alters the way options are priced. During times of crisis, the demand for out of the money put options will outweigh the supply, which generates an upward pressure on put options (Gârleanu et al., 2009). In order to ascertain whether monetary policy has been effective in curbing perceived tail risks, the empirical analysis resorts to a regression setup in order to gauge the response of the European financial markets to announcements of unconventional measures of the ECB. Effective monetary policy might lead to changes in economic agent’s behavior (i.e. effective monetary policy implicates

(23)

22

that the implied volatility of put options declines because of the lower urgency of hedging against downside risk). Put differently, a more formal analysis is conducted to quantitatively assess how tail risk responded to UMP policy announcements. More specifically, this study is interested in the average levels of the crash risk measure before and after the announcement dates. Consequently, standard event-study regressions measure the response of the average risk gauges to the announcement of policies.

In order to estimate the impact of the different UMP policy announcements and the likelihood of a significant shift in the mean before and after the event, a standard event-study framework is considered for different event windows. Specifically, the following regression is considered:

𝑦𝑡 = 𝛼 + 𝛽𝐷𝑡+ 𝜖𝑡 (3)

where the dependent variable 𝑦𝑡 refers to the 10- and 25-delta Risk Reversal series for 1, 3 and

6 months and 𝛼 is the mean Risk Reversal value for the given timeframe. In other words, the time series is regressed on the mean value of the dependent variable and on the explanatory variable of interest, respectively the constant term 𝛼 and the dummy variable 𝐷_𝑡. The impulse dummy can take the following forms, if b is the announcement date:

𝐷_𝑡 = { 1 𝑓𝑜𝑟 𝑡 ≥ 𝑏

0 𝑓𝑜𝑟 𝑡 < 𝑏 (4)

This model tests the conjecture of constant coefficients across the two subsets of data (pre-and post-event data) and analyzes the null hypothesis that the parameters (i.e. mean) do not vary over the subsamples defined by the specified announcement dates. Given the prespecified model, it is necessary to formulate both the null and alternative hypothesis, respectively:

𝐻₀: 𝛽 = 0 (6)

𝐻1: 𝛽 ≠ 0 (7)

The null hypothesis is tested by assigning statistical significance based on the F-test for the null that the coefficient on the UMP dummies are equal to zero. With the announcements categorized, the empirical design requires to specify a length of the sample period (or observation interval), event window, estimation window and post-event window (see figure 5). Following Hattori et al. (2013), a 45-day observation interval is employed, comprised of 22 pre-event days, the event day, and 22-post event days. Importantly, it is essential for the estimation window and event window not to overlap (McKinlay, 1997). Additionally, it is

(24)

23

typical to set the event window length to be larger than one, even if the event being considered is an announcement on a given date. Strictly speaking, this design renders estimators for the parameters to be not influenced by the information around the event. Put differently, UMP measures are partly anticipated by financial markets well before they are indeed implemented. This renders it difficult to pinpoint the exact effects of the announcements of UMP using event-study frameworks. Right before an announcement, information spreads gradually to markets and investors revise likelihood and size of programs over the preceding time before an announcement. This can be marked as the anticipation effect of monetary policy. Motivated by these considerations, this study eliminates the event windows, since including this window in the estimation of the pre- and post-event parameters could lead to a bias in the results when testing the difference in means (i.e. the pre-event mean RR will show a bias upwards when the RR values of the immediate days before the announcement dates are included).

5. DATA

This section will describe the process of acquiring the data for the proxy of tail risk perceptions in the euro area. In order to conduct this research we have to compute risk reversals from option prices on an equivalent of the S&P 500 for the euro area. Therefore, we rely on option prices on the EURO STOXX 50, previously known as the Dow Jones EURO STOXX 50, which is Europe’s leading blue-chip index containing the top stocks in the Eurozone. A blue-chip index is a stock index that involves the shares of the top-performing publicly traded companies with reputations for quality and reliability. The index covers 50 stocks from 12 Eurozone countries4. The EURO STOXX 50 index is licensed to financial institutions to serve as underlying for a

4_{The Eurozone countries covered in the index are Austria, Belgium, Finland, France, Germany, Greece, Ireland,}

Italy, Luxembourg, the Netherlands, Portugal and Spain.

FIGURE 5: TIMELINE OF AN EVENT STUDY

Notes: Sample period = estimation window + event window + post-event window. Source: McKinlay (1997)

(25)

24

wide range of investment products (e.g. options) and these derivatives are the world’s leading euro-denominated equity index derivatives (Weizmann, 2007).

The timespan considered would be from the start of non-standard monetary policy measures as conducted by the ECB. Therefore, we rely on announcements of the different policies during various episodes, commonly referred to as –among others- SMP, OMT and LTRO. We augment announcement dates analysed by previous research (Falagiarda, McQuade & Tirpák, 2015). Hence, our sample period is from March 2007 through January 2016.

Indicators of unconventional monetary policy.

The most important policy events are hereby given:

1. Securities Market Programme SMP 10/05/2010

2. Draghi Speech DRAGHI 26/07/2012

3. Outright Monetary Transactions Programme OMT 06/09/2012 4. Asset Purchase Programme APP 22/01/2015

Equity Options. We obtain options data from Thompson Reuters’ DataStream, which provides the option transaction price history on exchange-listed equity options in the Euro Area. When quantifying our tail risk proxies, bilinear interpolation (e.g. extension of linear interpolation for interpolating functions of two variables on a regular grid) is adopted to standardize both the delta and the maturity of the options data. For instance, the 6 months delta call-implied volatility is interpolated from four call options, with deltas straddling the 25-delta and maturities straddling the 180 days. Essentially, a similar procedure is used for put options. As indicated in figure 6 the key idea is to interpolate first in the direction of the delta, and then again in the other direction (of the maturity). In essence, the four red dots show the initial four call options and the green dot is the implied volatility of the call, on a specific day, interpolated over the two variables on the x and y axis. Ultimately, both put and call options with the same moneyness and maturity were paired in order to obtain the daily value for the risk reversal.

For comparison purposes, we would like to extract a daily risk reversal value from (deep) out-of-the-money puts and calls for different deltas and maturities. We rely on the data from Thompson Reuters DataStream to construct the 10δRR and 25δRR for different maturities (i.e. the 1 month RR, 3 months RR, 6 months RR).

(26)

25

FIGURE 6: BILINEAR INTERPOLATION OF OPTION DATA

Notes: Figure 4 shows the technique of bilinear interpolation used in order to

standardize both the delta and the maturity of options data. In essence, the four red dots (𝑄11, 𝑄12, 𝑄21, 𝑄22) show the initial four call options and the green dot (𝑃) is

the implied volatility of the call, on a specific day, interpolated over the two variables on the x and y axis.

Source: Author’s own graph

Other option-implied volatilities to construct our tail-risk proxies are obtained from Bloomberg. These tail risk proxies are constructed from IV series for both call and put options in basis points with specific deltas and maturities. Ultimately, the Risk Reversal series comprise the same deltas and maturities (i.e. the 1 month RR, 3 months RR, 6 months RR), but considers a smaller timespan, since longer-dated data was not available in Bloomberg. Moreover, the implied volatility index measure relies on the VSTOXX, obtained from STOXX innovative global indices.

Limitations. This section describes the process of parameterising the acquired data to find implied volatilities with a constant moneyness and time to maturity. As being limited by the availability of data, we therefore made use of an alternative approach utilizing the concept of bilinear interpolation on data from Thompson Reuters’ Datastream. However, due to stale IV quotes in corresponding neighbourhood of the interpolated values for both the delta and the maturity, there are some instances of consecutive dates with ‘stale’ risk reversal values (i.e. flat

(27)

26

exhausting different types of interpolation routines (i.e. locally adaptive quadratic splines, which makes the volatility surfaces differentiable at the ‘connection’ points) does not significantly change the data nor the results, considering that lack of differentiability is not the problem.

Furthermore, the Bloomberg data covers a timespan that does not include all major monetary policy moments within the euro area. However, this data might be useful in order to compare and complete the interpolated data, since the Bloomberg data is relatively more precise.

Consequently, table 1 and 2 in appendix 10.2 list the flat segments within a sample period of 45 days of the four major economic policy decisions of the ECB for both the 10δRR and the 25δRR series. Since the problem for the flat segments is missing information, it should be checked whether these problems arise around the events of interest in order to preclude the risk of biased results. Subsequently, the four main UMP moments of this study (i.e. SMP, OMT, Draghi Speech, QE) are analyzed from different data sources. This is, the QE event will be analyzed from the Bloomberg data, whereas the other events will be analyzed from the data of Thompson’ Reuters Datastream (i.e. the QE event cannot be evaluated from this data source, since the flat segments are too close to the announcement date, see appendix 10.2).

6. RESULTS

6.1 Interrupted time-series analysis

A good starting point for any data analysis involves exploring the underlying initial data without imposing any preconceived structure on the outcome. So, before turning to the empirical model, we have a closer look at the data to gain preliminary information. In various disciplines it is also called the ‘eyeball technique’ or ‘eyeball method’. Importantly, inferences from data subject to limitations, as described before, should be carefully drawn.

In order to obtain a first initiation whether tail risk perceptions change due to unconventional monetary policy, figures 7 till 10 give a first-cut answer by comparing risk reversal values on the Euro Stoxx 50 before and after a policy moment. These figures show the average readings of the 1-month and 6- month 10-delta risk reversals5 for the four major unconventional monetary policy moments for different event windows before and after the announcement. If

(28)

27

monetary policy is effective in curbing tail risk perceptions, the implied volatility of put options drops as a result of the lower urgency of hedging against downside risk. Theoretically, effective monetary policy will implicate a rise in the mean of the risk reversal after the policy announcement.

Firstly, figure 7 shows the impact of the SMP on risk perceptions in the euro area. Both the 1-and 6-month figures (weakly) implicate a cut-off point at the event date. In other words, the central bank stepped in when the situation was most wanting (i.e. risk reversal was at the lowest point). In essence, the risk reversal deteriorated in the run-up to the SMP announcement and the SMP announcement seem to break the downward trend of the RR series as the decline in the value of RR does not significantly weaken after the policy. For both the 1-month and 6-month series the shortest windows still display a rise in hedging costs of downside risk as the mean of the series is lower after the event. The 10-day and 20-day event window illustrate mixed patterns. Whereas the 1 month series present a slight improvement, the 6 month series give the opposite. In fact, the Securities Market Programme entailed intervention with little transparent communication, which makes it very hard for investors to correctly anticipate on the policy as conducted by the ECB.

Secondly, figure 8 shows the impact of the Draghi speech on European financial markets. Whereas previous literature mainly found that Draghi largely influenced sovereign bond spreads, the influence on downside risk appears to be much smaller (at first sight). Whereas the 1-month RR series for the short windows display a modest drop in hedging costs after the announcement, the effect dies out in the longer window. Surprisingly, the 6-month series show a deterioration in the hedging costs, but this becomes almost negligible in the longer event-windows. Ostensibly, it appears that the Draghi Speech did not boost the confidence of risk averse investors.

Thirdly, figure 9 basically entails the same patterns as observed for the Draghi Speech. These graphs also indicate that the central bank stepped in at the worst situation. It is however impossible to compare the graphs with the counterfactual of no monetary policy announcement.

Fourthly, the QE event indicate a temporary effect of the policy announcement. The short windows present a modest improvement of investor sentiment, which dies out in the longer event windows. Contrary to the previous policy moments, the investors seemed to have anticipated the announcement, since the risk reversal values were increasing before the announcement.

(29)

28

FIGURE 7: IMPACT OF ECB SMP POLICY ANNOUNCEMENT ON RISK PERCEPTIONS

FIGURE 7A FIGURE 7D

FIGURE 7B FIGURE 7E

FIGURE 7C FIGURE 7F

Notes: Figure 7A-F show the average readings of 10-delta risk reversals (10δRR) for different event windows before and

after the Securities Market Programme announcement date associated with the ECB unconventional policy measures. Risk reversals are typically expressend as the difference of IV of an OTM call and the IV of an OTM put of the same moneyness and maturity. Hence, an increase in the RR value indicate a drop in hedging costs. Figure 7A-C show the 1-month 10δRR, respectively for 5, 10 and 20 days before and after the SMP announcement date. Figure 7D-F show the 6-month 10δRR, respectively for 5, 10 and 20 days before and after the SMP announcement date.

(30)

29

FIGURE 8: IMPACT OF DRAGHI SPEECH ON RISK PERCEPTIONS

after the famous Draghi Speech associated with the ECB unconventional policy measures. Risk reversals are typically expressed as the difference of IV of an OTM call and the IV of an OTM put of the same moneyness and maturity. Hence, an increase in the RR value indicate a drop in hedging costs. Figure 8A-C show the 1-month 10δRR, respectively for 5, 10 and 20 days before and after the SMP announcement date. Figure 8D-F show the 6-month 10δRR, respectively for 5, 10 and 20 days before and after the SMP announcement date.

(31)

30

FIGURE 9: IMPACT OF ECB OMT POLICY ANNOUNCEMENT ON RISK PERCEPTIONS

after the Outright Monetary Transactions Programme announcement date associated with the ECB unconventional policy measures. Risk reversals are typically expressed as the difference of IV of an OTM call and the IV of an OTM put of the same moneyness and maturity. Hence, an increase in the RR value indicate a drop in hedging costs. Figure 9A-C show the 1-month 10δRR, respectively for 5, 10 and 20 days before and after the SMP announcement date. Figure 9D-F show the 6-month 10δRR, respectively for 5, 10 and 20 days before and after the SMP announcement date.

(32)

31

FIGURE 10: IMPACT ECB QE6_{POLICY ANNOUNCEMENT ON RISK}

PERCEPTIONS

after the Outright Monetary Transactions Programme announcement date associated with the ECB unconventional policy measures. Risk reversals are typically expressend as the difference of IV of an OTM call and the IV of an OTM put of the same moneyness and maturity. Hence, an increase in the RR value indicate a drop in hedging costs. Figure 10A-C show the 1-month 10δRR, respectively for 5, 10 and 20 days before and after the SMP announcement date. Figure 10D-F show the 6-month 10δRR, respectively for 5, 10 and 20 days before and after the SMP announcement date.

Source: Bloomberg | Author’s own graphs

6_{It should be noted that the data on the last event is obtained from Bloomberg as opposed to the previous policy}

(33)

32

As previously mentioned, the VSTOXX index is a symmetric measure of option-implied volatility and therefore it is not an adequate proxy for extreme downside risk. However, for comparison purposes, this study also looks at implied volatility indices in order to get a first impression on the impact of the policy measures on financial markets. Knowledge about the impact of monetary policy on markets is critical for the calibration of future policy. In order to get an idea whether the investors’ risk appetite changed due to unconventional policies, figures 11 and 12 show the value of the investor fear gauge (i.e. VSTOXX) before and after a policy moment. Consistent with figure 7 till 10, figure 11 and 12 show the average readings of the VSTOXX for the four major unconventional policy moments for different event windows before and after the announcement.

In essence, euro area equities benefit from positive confidence effects which, in turn, precipitate a fall in the VSTOXX, if monetary policy is effective. Whereas figure 11 illustrates the influence of the SMP event and the Draghi Speech, figure 12 shows the effect of both the OMT and the QE event. In particular, figure 11 implicates a negative impact of the SMP announcement on risk appetite. The graphs may partly reflect abating concerns of market participants over the inflation prospects and long-run growth of the euro area. The upper left panel of figure 11 shows that market uncertainty hardly responded to the SMP announcement, although the trend of the VSTOXX reflects a decrease in market uncertainty in a short window around the announcement. In contrast, the longer event windows (panel 11B and 11C) even indicate an increase in market uncertainty. Hence, the SMP announcement may not have guided markets in the right direction, which may be partly ascribed to the little transparent communication in tandem with the considerable market expectations of unconventional monetary policy. Additionally, the impact of the Draghi Speech is quite similar to figure 8. At first sight, the effect of the ‘Whatever it takes’ speech on implied volatility proxies were rather muted for both the risk reversal and the VSTOXX, which is in contrast with the impact of the Draghi Speech on other economic variables like the sovereign bond spreads.

The other ECB announcements, respectively the OMT and QE announcement, show relatively substantial improvements in investor sentiment (i.e. a decline in the VSTOXX) for both the shorter and longer windows. All panels suggest that markets partly anticipated the policies before the actual announcements considering that information spread gradually in the week prior to the ECB’s Governing Council meetings. Hence, the purchase of sovereign debt largely influenced European financial markets.

(34)

33

FIGURE 11: IMPACT OF THE ECB SMP POLICY ANNOUNCEMENTS AND THE DRAGHI SPEECH ON RISK PERCEPTIONS MEASURED BY THE

VSTOXX

Notes: Figure 11A-F show the average readings of the VSTOXX for different event windows before and after the Securities

Market Programme and the Outright Monetary Transactions Programme announcement dates associated with the ECB unconventional policy measures. An increase in the VSTOXX value indicates an increase in investor uncertainty. Figure 11A-C show SMP event, respectively for 5, 10 and 20 days before and after the SMP announcement date. Figure 11D-F show the OMT event, respectively for 5, 10 and 20 days before and after the OMT announcement date.

Source: STOXX global indices | Author’s own graphs

(35)

34

FIGURE 12: IMPACT OF THE ECB OMT AND QE ANNOUNCEMENT ON RISK PERCEPTIONS MEASURED BY THE VSTOXX

Notes: Figure 12A-F show the average readings of the VSTOXX for different event windows before and after the Draghi

Speech and the Quantitative Easing announcement dates associated with the ECB unconventional policy measures. An increase in the VSTOXX value indicates an increase in investor uncertainty. Figure 12A-C show Draghi Speech event, respectively for 5, 10 and 20 days before and after the Speech date. Figure 12D-F show the QE event, respectively for 5, 10 and 20 days before and after the QE announcement date.

The response of tail risk perceptions to unconventional monetary policy : an analysis for the Euro area