1
The impact of QE announcements on stock market volatility
Do central bankers accomplish their goal of stabilising prices?
Abstract
This study investigates whether QE announcements have an impact on stock market volatility by performing an event study analysis. In addition, two transmission channels are studied through which the QE announcements impact the stock market volatility. By analyzing the t-statistics of the abnormal returns in the event study, contradictory results are revealed. The announcement day as event shows no statistically significant impact on stock market volatility. By combining the announcement and post-announcement day as event, the three main QE announcements appear statistically significant at the 1% level. Despite the difficulties in isolating the true event, being significant at the 1% confidence shows a clear relationship between QE announcements and stock market volatility. This would imply that the central bank fails to accomplish their goal of stabilising prices. In order to measure the possible transmission channels, an OLS regression and Chow-test will be performed. After performing both tests, it can be concluded that the impact of QE announcement on volatility goes via the discount rate channel. The coefficient shows a negative relationship at the 1% confidence level. The rebalancing channel appears to be insignificant in both the regression and the Chow-test.
Keywords: stock market volatility, quantitative easing, transmission channels, announcements JEL classification: E52, E58,
2 1. Introduction
Since the collapse of Lehman Brothers in 2008 the dynamics in the economy and financial markets have changed dramatically. This event triggered the worldwide financial crisis which led to a crisis in the real economy. For governments and central banks these are challenging times because of the deteriorating unemployment rates and GDP growth numbers. In order to get the economy back on track, conventional policies weren't enough. On 13 September 2012 the Federal Reserve (FED) announced it would initiate a new $40 billion dollar asset purchasing program in order to stabilise prices on financial markets and to stimulate the economy. Did the FED really stabilise prices?
The Federal Reserve's main goal is to maximize employement, stable prices and moderate long-term interest rates. In order to achieve this goal conventional measures are normally used like changing the interest rate. However, in urgent times the central bank can choose for unconventional policies such as asset purchases. This phenomenon, also known as quantitative easing (QE), expands the central bank's balance sheet in order to increase the amount of money in the economy. (Bernanke, Reinhard and Sack (2004)). According to Blinder (2010) the conventional policy instrument, the federal funds rate, is more reliable and powerful than quantitative easing. However, when the nominal interest rate is cut all the way to zero the central bank can stimulate the economy by asset purchase programs.
These asset purchasing programs have dominated the financial markets and the newspapers on a regular basis in the past few years. Every word said by the president of the FED at the Federal Open Market Committee (FOMC) got attention from market participants and reporters. The big question was whether there would be any sign or hint that the central bank would expand its monetary policy. The implications of these words could be seen everywhere. All of the most substantial markets were impacted by these expectations and announcements. Not only equity markets were impacted; also forex, commodity, bond and derivatives markets were influenced by these meetings and announcements. Therefore, market participants from all over the world were impacted by the actions of the central bank.
3
said about the action of the central bank and their actions can be evaluated. Although this study only focuses on the announcement effect, it might be useful to see whether the intended action of the central bank really accomplishes.
This paper sheds light on the announcement effects of QE. Given its overwhelming importance, the main focus of this study is the impact of the QE announcement on stock market volatility. Nowadays volatility is for professional the core of their business. Also private investors and risk managers have to understand the concept of volatility otherwise they won´t beo n top of their game. Market makers earn their money on making spreads which get stimulated by higher volatility. Therefore, to beat the market everyone should know the concept of volatility and the determinants of volatility. The aim of this paper is to see whether there exists a significant relationship between the announcement event of QE and stock market volatility. In order to measure the relationship an event study analysis will be performed with 3-day event windows. In addition, event studies are based on the foundation of the efficiënt market hypothesis so the results of this study can be linked to this hypothesis. Two robustness checks will be performed. Firstly, the announcement day will be combined with the post-announcement day to form a new event day. It sound splausible that investors have to digest new informaiton and that they will anticipate on this news later. Secondly, the estimation window of the market model will be altered to see whether the results obtained are robust. Furthermore, the transmission channels through which this impact is established will be investigated. A regression analysis will be performed which includes dummies to account for the event dates and includes control variables for macroeconomic effects. The significance of channel proxies will indicate whether their is a true significant relationship between the channel and volatility. To see whether the finding is robust, a chow test will be performed to see whether there exists a structural change. The results of this study gives a clear picture on the effectiveness of the instrument and whether the central bank achieved its goal of price stability. To my understanding, this relationship hasn’t been studied extensively before despite its significance and importance in today’s practice. Therefore, the implications of this study will be beneficial for central bankers and practitioners and will add to their knowledge about the effectiveness of QE announcements.
4
stock market volatility. This result should be interpreted with caution since more noise can potentially come in as the event window gets wider. The two transmission channels investigated, rebalancing and discount rate channels, have different results. These results show that the existence of the rebalancing channel is not statistically significant. However, the coefficient of this channel indicates that there exists a positive relationship which is in line with previous research. On the other hand, there exists a statistically significant negative relationship for the discount rate channel. This finding is contradictory to existing theory which expected a positive relationship. After performing a Chow-test, the null hypothesis can be rejected at the 10% significance level. This implies that there is a structural break in the data caused by the QE event. For volume, this null hypothesis cannot be rejected.
The rest of the paper is structured as follows. In section 2, the theoretical framework will be discussed. This gives an overview of the research done on this topic and provides us with a theoretical background from which predicted results will be derived. Section 3 describes the data and methodology used throughout this study. In section 4 the results will be provided and discussed. Here, the predicted outcome will be related to the actual outcome and explanations will be given. Then, in section 5 the conclusions of this study will be drawn. Finally, in section 6 the limitations ofthis study are discussed as well as some future research possibilities.
2. Theoretical framework
In the past decades many academic papers have been written about quantitative easing and its implications and effects on the economy and financial markets. Throughout the literature, different definitions of quantitative easing have been used. According to Blinder (2010), ''quantitative easing refers to the composition and/or size of the central bank's balance sheet that are designed to ease liquidity and/or credit conditions.'' Joyce et al. (2011) refer to quantitative easing as ''a programme with large scale asset purchases, with the aim of injecting additional money and so increasing nominal spending growth to a rate consistent with meeting the CPI inflation target with the medium term.'' The baseline of QE is injecting money in the financial system in order to meet the central bank's goals.
5
stimulating the economy. In September 2008 the FED started the first round of buying securities which is called QE1 ($600 bln). When the economy started to recover slightly, the FED stopped the purchasing program which would harm the economy again. Because the economy started to deteriorate again, the central bank announced a second round of asset purchases called QE2 ($600 bln). Finally, in september 2012 the FED announced another purchasing programme called QE3 ($40 bln per month). The securities bought by the FED are mainly U.S. Treasuries and mortgage backed securities (MBS). Figure 1 shows the development of the FED’s balance sheet in time.
Also in Europe the phenomenon of quantitative easing can be found although this is less clear and more debated. Since 2003 the ECB conducted its Long Term Refinancing Operations (LTRO). With this operation, banks can refinance their obligations against very low rates. Although the strict definition of QE doesn't apply here (asset purchasing programme), this programme eases liquidity and injects additional money in the economy which can be seen as QE. Since the announcement of QE in the US is the main focus of this research, we will leave the LTRO topic and debate untouched.
Figure 1: The development of the FED’s balance sheet.
6
Leduk (2011) reported a decline of around 100 basispoints. In order to evaluate the effects in the UK, Joyce et al (2011) peformed an event study and found that the 10-year gilt yields were lowered by around 100 basispoints. From these articles it is clear that the easing policies were effective in depressing bond yields. Other asset classes received less attention from academics. Neely (2011) and Szczerbowicz (2011) evaluated the effect of QE on money market rates. Both studies didn't find any significant effects in the US during QE1 and in several other countries. Furthermore, several studies found that quantitative easing depreciated the home currency of the country initiating it (Neely, 2011; Glick and Leduc, 2011; Joyce et al., 2011). Quite interestingly, almost no research has been done about the effect on equity markets and on the volatility effect of the QE purchases on the equity market which will be the main topic throughout this study.
One study investigated the relationship between the FED's QE purchases and the stock volatility. The authors of this paper, Tan and Kohli (2011), researched how big the effect of QE purchases on stock volatility is and whether this effect is statistically significant. They argue that quantitative easing policies initiated by the FED keeps the stock market volatility artificially low. This point is also being made in the paper of Cole (2011). The explanation for this effect can be found in several factors. Firstly, the process of QE is a stable and long period which reduces the uncertainty about future policy actions. Secondly, by stimulating the economy and financial markets asset prices go up which ultimately leads to lower volatility. This reasoning is backed by a research conducted by Whaley (2008). In this paper, the author studies the relation between the stock market (S&P500) and the VIX (volatility index) and found that the volatility decreases when the stock market goes up and vice versa. Tan and Kohli (2011) conducted an event-study with different data sets (QE1 & QE2) and found that the QE purchases had a significant negative impact on stock market volatility.
7 inflation is less clear and not significant.
2.1 Stock market volatility
From the previous literature review it can be seen that most research focused on the impact of QE purchases on interest rates. It is obvious that such a policy action affects more asset prices since the financial markets are united in many senses and are interrelated. However, the effects on the other asset prices did not receive much attention because the effects are too ambigious. This study wants to shed some light on the impact of the QE announcement on the stock market volatility in the US. Abken and Nandi (1996) define volatility as a measure of the dispersion of an asset price about its mean level over a fixed amount of time. In this case, the asset price will be the broad stock market S&P500. Stock market volatility and volatility in general has become more important in past decades. First, more and more financial products who come to the market have a volatility component in their pricing. The most straightforward example is options. In option pricing, volatility is an important ingredient which impacts the option price significantly. The option price and its volatility have a positive relationship; a higher volatility leads to a higher option price and vice versa. For many trading strategies involving options, volatility is very important. Secondly, policy makers pay a lot of attention on volatility because of their goals. As discussed before, one of the main goals of central banks is price stability. A low level of volatility ensures more stable prices and therefore central banks closely monitor volatility. The impact of the QE announcement on volatility can therefore show whether the action of the central bank is succesful to achieve their goal of price stability. However, price stability should be defined more narrow because the concept is too broad. Central banks aim to stabilize prices over the whole economy with a main focus on inflation. The stock market can also be seen as an 'asset price' which is related to the broad concept of 'prices'. Furthermore, all information on the economy is reflected in the stock market according to the efficient market hypothesis. Therefore, stock markets and its volatility are of great importance for central bank policy makers.
8
Figure 2: Development of the VIX index 2007-2013
2.2 Transmission channels
In order to assess the impact of QE, previous literature investigated channels through which monetary policy effects the financial markets. There are several ways in which the actions of the central banks effect the financial markets which in the end effects the real economy. These channels are generally known as transmission channels.
A channel in which QE purchases affect asset prices is the signalling channel or sometimes called news channel. This channel has received a lot of attention in the the academic world. Joyce et al. (2011) define this channel as ''anything economic agents might learn from the central bank's QE announcements about the underlying state of the economy and the future path of interest rates.'' According to Eggortson and Woodford (2003) unconventional policies initiated by central banks can reduce long term bond yields only if the central bank is credible. Krisnamurthy and Jorgensen (2011) argue that the signalling channel has an impact on all interest rates because interest rates are related to the federal futures fund rate in the US. If this futures fund rate is low and the central banks are committed to this low level, interest rates will drop. However, Joyce et al. (2011) argue that the effect of this channel on bond markets are ambigious since it can signal lower future policy rates but also higher inflation in the future. This channel is also said to have an impact on other asset classes since their respective discount rates are affected by this channel.
9
want. Normally, the spreads in liquid markets is also smaller than in illiquid markets. Previous studies, for example Joyce et al. (2011), examined the liquidity premia channel to see what effect QE purchases had on the yields in the bond market. Because the central bank is a credible and large player, the market gets more liquid because of the purchases and therefore the illiquidity premia should become smaller. This ultimately results in a declining bond yield. This channel could affect yields significantly because in normal times markets are deep and liquid, but in periods of crisis the markets get less deep and illiquid.
After identifying a possible effect of the QE purchases announcement it is interesting to see from which channel this effect comes from. There are several channels through which QE purchases can impact stock market volatility. The first channel is the rebalancing channel. As was discussed earlier, Tobin introduced the concept of imperfect asset substitutability in his article (Tobin, 1969). Since investors have certain preferences and asset classes have different characteristics, investors have to rebalance their portfolios when an announcement is made regarding the future state of the economy and financial markets. This will have an impact on the volatility since investors have to do rebalancing transactions. Furthermore, Schwert (1990) argued in an early paper that trading volume and stock market volatility moved together. As more investors participate in the market, stock markets get more volatile. In order to react to the news and reflect these changes in investors’ personal portfolios, they have to rebalance them. This rebalancing effect induces more investors to do transactions on the stock market which increases the trading volume. Since trading volume has a positive relationship with stock market volatility, it can be expected that volatility increases because of this rebalancing channel. As schwert argues in his paper, more reasons for increased volume can be given. The main reasons are 1) reacting to news, 2) using price information as input for trading strategies and 3) portfolio insurance frictions. Furtermore, Burton (2005) assumes in his article that markets have a high degree of efficiency. He concludes this from his study in which professional fund managers were not able to consistently outperform the passive strategies. This shows that the market incorporates a high degree of available information. All in all, the effect of this channel should increase stock market volatility.
10
for this cost of debt is the 10 year government bond of the home country. Knowing that QE purchases influence the bond yield (bond yields decrease), this has also an effect on the valuation of the stock. In general, it can be expected that stock prices will increase and therefore, according to Whaley (2008), stock market volatility will decrease. This means a positive relationship is predicted.
The next sextion will discuss the methodology used throughout this study to measure the effects of interest..
3. Data and methodology
In order to test the effects of QE on stock market volatility several difficulties arise. Unconventional monetary policies get a lot of attention from market participants and policy makers. To see whether the action of the central bank achieved its goals, the real effect from the policy should be measured. This can be quite a challenging task because financial markets focus to a large extent on expectations. During earlier Federal Open Market Committee meetings (FOMC) central bankers can hint on future stimulating actions of the central bank which will be reflected in the stock market. By talking and hinting, expectations about another round of stimulus are formed which will be reflected in the financial markets since practitioners anticipate on possible future actions. Because of this, the ''real'' effect of the quantitative easing is hard to measure because it is hard to observe when market expectations are formed. In order to deal with this problem, previous studies in this field mainly use the event study methodology to capture the announcement effect of the stimulus action (Joyce et al., 2011; Krishnamurthy and Jorgensen, 2011). Bernanke and Kuttner (2005) argue that unexpected policy actions help mitigate potentional difficulties with endogeneity and simultaneity, and therefore the ’real’ effect can be disentangled better.
Previous articles dealt with the problem of expectations in several ways. Joyce et al. (2011) used data from the Reuters Poll of Economists about expected asset purchases by the Bank of England. They measured the expectations about the purchases right before and right after the QE announcement. The difference between these two values was interpreted as the unexpected QE announcement effect. Bernanke and Kuttner (2005) used the federal fund rate futures contract to measure the surprise effect of the policy action. The difference in price between the day before the announcement and the day after the announcement can be seen as the ‘surprise’ effect. Other authors state that it is a heroic task to measure the ‘’expectations’’ and they just focus on the daily announcement effect. This will be the course of action in this paper as well.
11
which all concern QE or hints on QE. Two events focus on ’hints’ about possible quantitative easing policies in order to capture the rising expectations within the market. These events will be treated as a QE policy although the FED only used words, not actions. The focus will be spread between three rounds of QE: two events about QE1, two events about QE2 and three events about QE3. In the last two rounds the concept of ‘hinting’ is used since markets are familiar with the FED pulling the trigger and therefore expectations are formed which weren’t in place at the time of the first round of QE. The following event dates are used:
25 November 2008: Announcement of QE1
At this day the Federal Reserve chairman Ben Bernanke announced the plan to purchase $100 billion in agency debt and $500 billion in mortgage backed securities.
18 March 2009: Expansion of QE1
On this day the FOMC decided to expand the current operation. They announced the decision to purchase $300 million long term Treasury bonds and increase the QE1 purchase amount of agency debt and MBS with respectively $200 billion and $1.25 trillion.
27 august 2010: Hint QE2
During the speech Bernanke hints on QE2: “Macroeconomic projections are inherently uncertain, and the economy remains vulnerable to unexpected developments. The Federal Reserve is already supporting the economic recovery by maintaining an extraordinarily accommodative monetary policy, using multiple tools. Should further action prove necessary, policy options are available to provide additional stimulus. Any deployment of these options requires a careful comparison of benefit and cost. However, the Committee will certainly use its tools as needed to maintain price stability--avoiding excessive inflation or further disinflation--and to promote the continuation of the economic recovery.’’
3 November 2010: Annoucnement of QE2
12 26 March 2012: Hint QE3
Bernanke hints on a possible third round of QE. He says: “To the extent that this reversal has been completed, further significant improvements in the unemployment rate will likely require a more-rapid expansion of production and demand from consumers and businesses, a process that can be supported by continued accommodative policies.’’
13 September 2012: Announcement QE3
The Federal Reserve pulls the trigger and announces QE3. Bernanke says: “To support a stronger economic recovery and to help ensure that inflation, over time, is at the rate most consistent with its dual mandate, the Committee agreed today to increase policy accommodation by purchasing additional agency mortgage-backed securities at a pace of $40 billion per month. The Committee also will continue through the end of the year its program to extend the average maturity of its holdings of securities as announced in June, and it is maintaining its existing policy of reinvesting principal payments from its holdings of agency debt and agency mortgage-backed securities in agency mortgage-backed securities. These actions, which together will increase the Committee’s holdings of longer-term securities by about $85 billion each month through the end of the year…”
12 December 2012: Expansion of QE3
During the FOMC the decision to expand QE3 is announced. The committee argues: “To support a stronger economic recovery and to help ensure that inflation, over time, is at the rate most consistent with its dual mandate, the Committee will continue purchasing additional agency mortgage-backed securities at a pace of $40 billion per month. The Committee also will purchase longer-term Treasury securities after its program to extend the average maturity of its holdings of Treasury securities is completed at the end of the year, initially at a pace of $45 billion per month.”
13
directly in the stock market. However, since not everybody can access the information at the same time and at the same cost, the assumption can be violated. In order to deal with isolating a QE event, many researchers use a small event window (Gagnon, 2010; Joyce et al., 2011; Krishnamurthy and Jorgensen, 2011). Most commonly authors use 1-, 2- and 3-day windows. In this study a 3-day window is used to isolate the QE event. The true event is considered the announcement day. No other news which impacts the stock market are found within the established windows. This means we can say safely that the QE announcement or hint is the only observed event going on in the market. Different event windows will be used to check whether the results are robust.
In order to assess the impact of QE announcement on stock market volatility, the following model will be used:
(1)
Where AR is the abnormal return of the volatility at time t, ri,t is the actual return and is the expected return. According to Pynnönen (2005), several methods can be employed in order to estimate the expected return. The most standard model is the constant mean model, where the expected return is estimated by the sample mean. Furthermore, multi-index models estimate the
expected return with the following model:
= α + β1I1,t + ··· + βpIp,t where the Ip,t represents index returns. In this study, the market model will be used to come up with a good estimation. The formula for this model is as follows:
(2)
14
standard error in order to come up with the t-statistic. The t-test will be measured according to this formula:
(3)
where SE represents the standard error. The null hypothesis states that the event has no impact which can be tested by the given t-statistic.
After identifying the impact of the QE announcement on stock market volatility, the transmission channels through which this impact originates will be investigated. The two channels (portfolio rebalancing and discount channel) both need a proxy in order to measure the effect on the impact of QE on stock market volatility. In order to obtain results about the impact of the transmission channels, the research methodology of Krishnamurthy and Jorgensen (2011) will be taken into consideration. In their article, they investigate the transmission channels of the impact of QE purchases on interest rates. In order to assess the channels, they use rate changes or changes in derivatives contracts to assess the significance of the corresponding channel. The main idea is to check whether the changes on QE announcements days are different from changes of other days. This is done by regressing the change of the variable to the event and see whether the change is different during the event. The event will be represented as a dummy variable.
The model used to assess the impact is as follows:
(4)
15
channels were the focus is on there should be a control variable included in the OLS regression. Furthermore, the Crude Oil spot price is used as a proxy for the oil price and will also serve as a control variable.
In order to increase the statistical significance, a robustness check will be performed. In this case, the data is split in two subsamples which are both regressed upon volatility. Then, the significance can be calculated by comparing the residual sum of squares.
In this study data from financial markets will be used. For all variables included in the research model, daily data from november 2007 till may 2013 is available. This data is obtained by using Thomson Reuters Datastream and by data files on the website of the Federal Reserve attached to the annual report. Because markets are closed at different days, only data is used when all relevant markets were opened. The rest of the data has been excluded from the sample. The descriptive statistics can be found in table 1.
Table 1 shows the descriptive statistics of the dependent, independent and control variables. For all variables there are 1377 observations from the period november 2007 – may 2013. Interesting to see is the distribution of the dependent variable. With a mean of 24,88 its average is closer to the minimum value than to the maximum value. This indicates that the VIX index is a ‘’fear’’ indicator. If the market starts to panic the VIX value will go through the rooftop. During the fall of Lehman Brothers, the VIX started to skyrocket. Another interesting observation is the minimum value of expected inflation. Throughout this study it will be used as a control variable. However, it is surprising to see that investors expected deflation. Deflation is for a central bank the worst situation they could be in. However, the accomodative policies of the central banks are the driving factor for investors to start thinking about inflation. Therefore, the minimum value of expected inflation can be linked directly to the monetary policy of the central bank. The variable volume refers to the daily volume of the S&P 500.
Event studies in general have to be performed carefully. As was discussed previously, this methodology is based on the market efficiëncy hypothesis. However, markets are not always perfect which can lead to biases and drawing wrong inferences. Furthermore, the choice of event window is crucial in drawing conclusions from the event. Within this window the event should be isolated so no other event influences the actual observed event. Finally, the results are as good as the input is. Many academics refer to event studies as “garbage in garbage out.” Proper data is fundamental for a fruitful event study.
16 4. Results and discussion
In this section the results will be discussed and the outcomes will be linked with the theory section and then discussed. First of all, the event study has been performed from which the results can be found in table 2. In order to get the results, the event window had to be chosen. According to newspapers no significant news effect appeared on the three day window event which had a big impact on the variables researched in this paper. Therefore, the true effect could be observed by the 3-day window. Then, the expected return of the event had to be estimated. In order to come up with a reliable and good estimator, the market model which has been discussed in the theory part, has been used. Previous research indicated that a regression based on a time period of a year with daily data would be sufficient to come up with an estimator which doesn’t have a bias. The raw data of the VIX and the S&P 500 have been translated into returns to make the data more comfortable to work with. In addition, the daily VIX data has been regressed to the S&P 500 index returns. In all seven events we can see that the relationship between the VIX index and the S&P 500 index is negative. This implies that when the S&P 500 goes up, the VIX index en therefore volatility will go down. This result is in line with the result of Whaley (2008) who found that an increase of 1% in the S&P 500 would lead to a decrease in the VIX of -2,99%. However, on the other hand the author argues that a 1% decline in the S&P 500 would lead to an increase in the VIX by 4,493%. This shows that the VIX index is indeed a ‘’fear’’ indicator with a skewed distribution and fat tails. Furthermore, in six of the seven events a negative expected return is estimated. This derives logically from the slope estimate which indicate a fall in the index and therefore negative expected returns. Finally, the R-square shows that around 70% of the variation of the S&P 500 can be explained by the VIX index. This statistic confirms that it is safe to state that the VIX is a good proxy for stock market volatility.
To assess the impact of the QE announcement on stock market volatility, the actual and estimated return should be compared. From this, the abnormal effect of the QE can be seen. The
Table 1
Standard
Variable n Mean Std. Dev. Minimum Maximum
Vix 1377 24,88 10,99 11,30 80,86 Treasury yield 1377 2,89 0,81 1,43 4,30 Expected inflation 1377 0,84 0,96 -0,87 3,15 Oil price 1377 86,71 19,75 30,81 145,66 Volume 1377 4,49 1,31 1,03 11,46 Summary statistics
17 Table 2
Results of the event study analysis for the 3-day event window
Panel A: Results of the market model estimation
Event Date Intercept Slope Std. Error
QE1 25-11-2008 0,002 -2,774 0,043 QE1 expansion 18-3-2009 0,001 -2,476 0,044 QE2 hint 27-8-2010 0,004 -5,453 0,040 QE2 3-11-2010 0,005 -5,555 0,039 QE3 hint 26-3-2012 0,005 -4,918 0,044 QE3 13-9-2012 0,004 -4,688 0,042 QE3 expansion 12-12-2012 0,004 -5,674 0,038
Panel B: Results of the event study
Event Date E(r) AR AR t-test
QE1 24-11-2008 -17,769 11,896 2,797** 25-11-2008 -1,672 -8,147 -1,916* 26-11-2008 -9,720 9,720 2,285** QE1 expansion 17-3-2009 -7,881 1,160 0,266 18-3-2009 -5,093 3,279 1,018 19-3-2009 3,291 5,745 2,337** QE2 hint 26-8-2010 4,622 -2,112 -0,522 27-8-2010 -8,739 -1,929 -0,477 30-8-2010 8,366 2,922 0,723 QE2 2-11-2010 -3,827 2,636 0,669 3-11-2010 -1,662 -7,656 -1,944* 4-11-2010 -10,325 5,008 1,271 QE3 hint 23-3-2012 -1,069 -3,748 -0,858 26-3-2012 -6,375 2,597 0,594 27-3-2012 1,790 7,537 1,726 QE3 12-9-2012 -0,752 -2,965 -0,711 13-9-2012 -7,356 -3,720 -0,892 14-9-2012 -1,507 4,781 1,147 QE3 expansion 11-12-2012 -3,314 0,323 0,084 12-12-2012 0,023 2,417 0,634 13-12-2012 3,845 -0,021 0,005 * = significant at 10% level ** = significant at 5% level *** = significant at 1% level
18
results are mixed since four of the seven events show a positive abnormal return where the other three events show a negative sign. In order to check whether the observed effect is really the observed effect and not just pure luck, the t-test should be performed. This statistic divides the abnormal return by the standard error of the sample. Because of the relatively large sample, the t-statistic has more than 200 degrees of freedom which suggest that the relevant t-t-statistic is 1.96 at the 5% confidence level and 1,65 at the 10% confidence level. The corresponding null hypothesis is that there is no significant impact of the event.
19
seeks for the impact of the announcement so the announcement day will nonetheless be regarded as the true event date.
Considering the other insignificant events, an extension of the beforementioned argument can be made. It could be the case that market participants already expected and anticipated to a new hint or new round of quantitative easing. This problem is hard to solve as was discussed in the methodology part since it is hard to disentangle market expectations from the market.
In order to check whether these results are robust, the announcement day and the day after are combined in order to create a new event period. It is plausible to assume that investors have to digest the information given by the FED chairman and thus won’t react directly to the news. The day after the announcement investors can react to new insights and this could influence the volatility. Combining the pre-announcement day and the announcement day could also be used as a robustness check. It is plausible to assume that investors anticipate to the announcement. However, you have to be careful with expectations since it is hard to observe when these expectations are formed. In addition, important news was released in most of the pre announcement days which blurs the true effect of the event. Therefore, the only robustness check will be the announcement day and the post announcement day. By using 2 days as the event, it is quite tricky to keep the announcement as the only observed event. Although no significant news releases or macroeconomic numbers were published, it is possible that other influences might kick in. However, the impact of the QE announcement can be seen as substantial so for now it is assumed that it is the only observed event in the event window.
20
Furthermore, another robustness check will be performed to see whether the results differ substantially when the estimation period is changed from 252 days to 120 days. As was discussed earlier, the input has a significant impact on the output obtained. Table 4 shows the results of the robustness check. With a shorter estimation period, it can be seen that the results do not differ much. In this case, the QE1 announcement is significant at the 10% level. With the longer estimation period, QE2 was also significant at the 10% level. With a shorter estimation period this event becomes insignificant. These results support the conclusion that QE announcements do not have a significant impact on stock market volatility.
After this striking result, the transmission channels through which volatility gets impacted are investigated. In order to assess the two possible mechanisms, treasury yield and volume will serve as proxies for the rebalancing channel and the discount rate channel. As discussed in the theory section, several articles have argued about influencers of stock market volatility (Schwert (1989); Officer (1973)). Because of this, it is hard to disentangle the effect of the proxy on the explained variable. In order to get an unbiased estimation of the transmission channel, several control variables have to be included. As Schwert (1989) mentioned, inflation is an important determinant of stock market volatility. He argued that if inflation diverges from the expectation embedded in market, stock markets get more volatile. Furthermore, oil prices are also a determinant of stock market volatility. This seems more logical since the majority of companies in the world are dependent on oil. Difficulties in this industry hurts companies in other industries and in the end also
Table 3
Results of the event study with announcement and post-announcement day combined as event
Event E(r) AR AR t-test
QE1 -11,580 15,823 3,721*** QE1 expansion -1,881 2,673 0,614 QE2 hint -0,823 1,056 0,261 QE2 -12,471 14,803 3,759*** QE3 hint -5,041 6,159 1,411 QE3 -9,216 11,257 2,700*** QE3 expansion 3,465 -4,004 -1,052 * = significant at 10% level ** = significant at 5% level ***= significant at 1% level
21
customers. These effects are likely to have an impact on the stock market volatility. Because of this, both variables will be included in the OLS regression.
First, all data was turned into daily returns and then dummy variables were made for six of the seven events. Since there is a constant included in the regression, the seventh dummy should be avoided otherwise the dummy variable trap will kick in. After this, a regression model was build up with the expected inflation and the oil price as the control variables.
From Panel A of table 5 the regression results can be obtained. Since the transmission channels will be investigated, the focus will be on the proxies volume and treasury yield. The corresponding coefficients reveal that the relationship between stock market volatility and volume is slightly positive although this is not significant. This means that the portfolio rebalancing channel cannot be observed with statistical significance. Even though this finding is not significant, it is interesting to see that an increase in volume increases volatility. This is in line with previous research of Schwert (1990) who found that trading volume and volatility are positively related to eachother.
Furthermore, according to the table, treasury yield has a statistically significant negative relationship with a coefficient of -0,369 and a t-statistic of -14,374. This means that stock market volatility increases when the discount rate decreases. This is in contrast with the predicted positive relationship. Logical thinking would suggest that higher stock prices would lead to lower stock market volatility. As Whaley (2008) argues in his article, the VIX index is also known as the ‘’fear’’ index. Higher stock prices should be negatively correlated with this index since investors have less fear when prices rise. Only investors who have short positions are feared when markets and stock
Table 4
Results of the event study with a 120-day estimation period
Event E(r) AR AR t-test
QE1 -1,459 -8,360 -1,735* QE1 expansion -4,741 2,927 0,720 QE2 hint -9,281 -1,388 -0,303 QE2 -1,985 -7,334 -1,651 QE3 hint -5,777 1,998 0,463 QE3 -9,298 -1,778 -0,456 QE3 expansion 0,005 2,435 0,626 * = significant at 10% level ** = significant at 5% level *** = significant at 1% level
22
prices go up but for now we exclude them from this study. Therefore, this result is quite surprising and in contrast with theory. A possible explanation could be that investors tilt their portfolios more towards government bonds which triggers more selling pressure on the stock market. Investors who do not like the QE announcement might be afraid and switch to more riskles sassets. As Whaley (2008) described, selling pressures have a bigger impact on volatility than purchasing pressures. If this is the case, than the central bank fails to communicate their intentions properly to the investors.
Table 5
Results of the OLS regression for the transmission channels and Chow test
Panel A: Results of the OLS regression De pe nde nt variable = Coeffient (t -stat) Intercept 0,002 1,208 Treasury yield -0,369 -14,374*** Volume 0,037 1,527 Expected inflation 0,019 0,790 Oil price -0,151 -5,917*** QE1 dummy -0,057 -2,35**
QE1 expansion dummy -0,037 -1,507
QE2 hint dummy -0,006 -0,228
QE2 dummy -0,010 -0,416
QE3 hint dummy -0,018 -0,756
QE3 dummy -0,019 -0,781
R-squared N
Panel B: Results of the Chow test Dependent variable = Treasury yield Volume * = significance at 10% level ** = significance at 5% level *** = significane at 1% level
23
These actions of the central bank are taken to enlarge the trust in the market and the economy. As Blinder et al., (2010) mention in their article, this communication fail could possibly come from a lack of credibility and commitment. However, more investigation has to be done in order to come up with the right explanation. All in all, according to these results it can be said that the QE announcements impact the stock market volatility negatively with statistical significance. This regression also shows that the QE1 dummy is statisically significant at the 5% level. This indicates that this event has a significant impact on stock market volatility. This is in line with the results found in the announcement day event study and the combined day event study.
In order to check whether these results are robust, a Chow-test has been performed. This test sheds light on the parameters within the regression. This test wants to test the assumption that the parameters are constant for the entire sample. This tests is also known as a parameter stability test or the analysis of variance test (Brooks, 2008). Essentially, the data is split in two sub-periods from which three regressions will be performed. One regressions is performed on the entire sample and two regressions are run on both sub-periods. By comparing the residual sum of squares the test statistic can be calculated from which inferences can be drawn.
This test statistic can be calculated with the following formula:
(5)
24 5. Conclusion
In this paper some light has been shed on the implications and effects of the unconventional central bank policy called QE. Previous research focused mainly on conventional central bank policies or investigated the impact of QE on interest rates. The main goal of this paper is to see whether QE announcements have a significant impact on stock market volatility. Given its increasing
importance, this relationship should get more attention from the academic field. The methodology used to observe whether QE announcements impact volatility is the event study analysis. Seven events have been selected which all relate to announcing or hinting on a possible QE stimulus. Using the announcement day as event, none of the events had a statistical significant impact on stock market volatility. This result is robust to changing the estimation period from 252 daily returns to 120 daily returns. However, when changing the event window from the announcement day to a combination of the announcement day and the post-announcement day, striking results can be seen. The abnormal returns of the three main QE announcements appear to be statistically significant at the 1% level. The conclusion resulting from this is quite surprising since it would indicate that the central bank fails to accomplish their goal of price stabilization. This result has to be interpreted with some caution since other factors might kick in as the event window gets wider. However, being significant at the 1% level shows at least some relationship between QE
announcements and stock market volatility.
In order to see via which transmission channel the QE announcements influence stock market volatility, an OLS regression has been performed. The two hypothesized channels,
rebalancing and discount rate, are proxied with respectively volume and treasury yield. In order to deal with macroeconomic influences, expected inflation and oil price are added as control variables. Dummy variables have been used to account for the QE events.. The regression results show that the rebalancing channel is not statistically significant even though its’ positive coefficient which is in line with previous theory. The discount rate channel appears to have a negative statistically significant relationship with stock market volatility. As a robustness check, a Chow-test has been performed to assess whether the parameters are constant for the entire sample. Three regression has been performed, one for the whole sample and two for the sub-samples. The QE announcement events are included as a dummy which is 1 at the event day and 0 otherwise. The results from this test are supportive towards the results earlier found. The test statistic for the rebalancing channel is not statistically significant and therefore the null hypothesis cannot be rejected. There is no
25
via the discount rate channel. All in all, the results are too striking to come up with a final
conclusion. At the very short time period the central bank accomplishes its goal of stabilising prices but when using longer event windows the opposite is true.
6. Limitations and future research
A lot of research has been devoted on monetary policy and its implications on the financial markets and the real economy. However, unconventional measures are relatively new to the academic world and therefore a lot of research can still be done in this field. As is the case with monetary policy in general, it is hard to resolve the endogeneity problem. It is hard to find out whether the central bankers are triggered to act because of market circumstances or that market circumstances react to monetary policy actions. In order to evaluate whether central banks accomplish their goals, this is of great importance to know. This paper followed the course of others in the field and ignored the problem of endogeneity. As it could influence results or the interpretation of the results, this can be seen as a limitation. Furtermore, this paper tried to deal with market expectations as good as
possible. By using the VIX index as dependent variable, including hinting in the event selection and using the expected inflation as control variable, the possible effect of market expectations have been lowered. However, for event studies it is important to know when expectations are formed and why they are formed. Therefore, the lack of market expectation treatment is a limitation to this study. Since many more research papers have problems to identify market expectations, it would be an area to devote more attention and research on. Finally, as is a problem for all event studies, selecting the right event window is a limitation to this study. Although robustness checks have been
performed, the true event is hard to isolate. Combined with market expectations it is a serious issue which should get priority on the research agenda in order to develop the monetary policy literature.
26 7. References
Abken, P, Nandi, S, 1996, Options and volatility, Economic Review, 21-35
Benford, J, Berry, S, Nikolov, K and Young, C, 2009, Quantitative easing, Bank of England Quarterly Bulletin, 49, 90-100
Bernanke, B, Kuttner, K, 2005, What explains the stock market’s reaction to Federal Reserve policy?, Journal of Finance, 60, 1221-1257
Bernanke, B, Reinhart, V and Sack, B, 2004, Monetary policy alternatives at the zero bound: an empirical assessment, Brookings papers on Economic Activity, 2, 1-100
Blinder, A, 2010, Quantitativeeasing: entrance and exitstrategies, Federal Reserve Bank of St.
Louis Review, Sept./Oct. (2010 Homer Jones Memorial Lecture)
Blinder, A, Ehrmann, M, Fratzscher, M, Haan, J de, Jansen, D, 2008, Central bank communication and monetary policy: a survey of theory and evidence, Journal of Economic Literature, 46, 910-945
Brooks, C, 2008, Introductory econometrics for finance, 2nd edition, Cambridge University Press.
Burton, M, 2005, Reflections on the efficiënt market hypothesis: 30 years later. The Financial Review, 40, 1-9
Calomiris, C, Tallman, E, 2010, In monetary targeting, two tails are better than one, Bloomberg Businessweek, November 18.
Cole, C, 2011, Is volatility broken? Normalcy bias and abnormal variance, Artemis Capital Management
Eggortson, G, Woodford, M, 2003, Zero bound on interest rates and optimal monetary policy, Brookings Papers on Economic Activity, 1, 139-233
Eichengreen, B, Tong, H, 2003, Stock market volatility and monetary policy: what the historical record shows, Asset Prices and Monetary Policy, 108-142
27
Gagnon, J, Raskin, M, Remache, J and Sack, B, 2010, Large-scale asset purchases by the Federal Reserve: did they work?, Federal Reserve Bank of New York, Staff Report No, 441.
Glick, R, Leduc, S, 2011, Are large-scale asset purchases fueling the rise in commodity prices?, FRBSF Economic Letter, 4
Joyce, M, Lasaosa, A, Stevens, I and Tong, M, 2011, The financial market impact of quantitative easing in the United Kingdom, International Journal of Central Banking, 7, 113-161
Kearney, A, Lombra, R, 2004, Fed funds futures and the news, Atlantic Economic Journal, 31
Koller, T, Goedhart, M, Wessels, D, 2010, Measuring and managing the value of companies, 5th edition, McKinsey and Company
Krishnamurty, A, Vissing-Jorgensen, A, 2011, The effect of quantitative easing on long term interest rates, Brookings Papers on Economic Activity, 2, 215-265
Neely, C, 2011, the large-scale asset purchases had large international effects, Federal Reserve Bank of St .Louis Working Paper 2010-1018C. http://research.stlouisfed.org/wp/2010/2010-018.pdf
Officer, R, 1973, The variability of the market factor of the New York Stock Exchange, Journal of Business, 46, 434-453
Pynnönen, S, 2005, On regression based event study, Contributions to Accounting, Finance and Management Science, 143, 327-354
Schwert, G, 1989, Why does stock volatility change over time?, Journal of Finance, 44, 1115-1153
Schwert, G, 1990, Stock market volatility, Financial Analysts Journal, 46, 23-24
Szczerbowicz, U, 2011, Are unconventional monetary policies effective?, Working paper.
http://celeg.luiss.edu/files/2009/07/CeLEG_Szczerbowicz.pdf
Tan, J, Kojli, V, 2011, The effect of Fed’s quantitative easing on stock volatility, Working Paper.
28
Tobin, J, 1969, A general equilibrium approach to monetary theory, Journal of Money, Credit and Banking, 1, 15-29
Whaley, R, 2008, Understanding VIX, Working Paper.
