The Impact of FED Quantitative Easing Announcements on The Different Sectors Represented in The Equity Markets of The United States of America
Afonso Mendes Ashwood Madeira
Universiteit van Amsterdam – Faculty of Economics and Business Economics and Finance Bachelor Thesis
Statement of Originality
This document is written by Student Afonso Mendes Ashwood Madeira who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
Table of Contents
Abstract ... 4
Introduction ... 4
How exactly does Quantitative Easing work? ... 6
Literature Review ... 8 Is QE an Effective tool? ... 8 Effects on Bonds ... 9 Effects on Stocks ... 11 Effects on Volatility ... 13 Luca-Moench Effect ... 13 Intermezzo ... 14 Hypothesis ... 15 Methodology... 15
Stock Market Data ... 15
Event Study Method... 16
Announcement Dates ... 16 Operations... 17 Testing ... 18 Empirical Results ... 19 Cross-Sectional Results ... 19 25th of November 2008. ... 20
Time Series Results ... 22
CAARs. ... 22
Pre, Post & Event AARs and CARs. ... 23
Financial Sector. ... 25
Positive and Negative Annoucements ... 27
Positive Announcements. ... 27 Negative Announcements. ... 29 Conclusion ... 30 Recommendations ... 31 References ... 33 Tables ... 35
Abstract
This thesis aimed to understand the impact of FED quantitative easing announcements on the different sectors within American equity markets. To do so event studies were conducted on each of the 11 GICS sector indexes, over 19 different quantitative easing announcement dates. Results were mixed. No single announcement managed significantly impact each sector during the same event window. Energy, Information Technology, Consumer Discretionary, and Consumer Staples show statistically significant cumulative average abnormal returns. Sectors reacted differently from one another. Real Estate and Energy sectors on average displayed negative abnormal returns over the entire event window, whilst Information Technology, Consumer Discretionary sectors on average showed positive abnormal returns over the entire event window. Financial and Consumer Staples sectors showed contrasting results over the event window. The Financial sector on average showed negative pre-announcement abnormal returns, but positive announcement and post-announcement abnormal returns, whilst the Consumer Staples sector displayed the opposite (positive pre-announcement whilst negative during and post announcement). Results stayed the same when analyzing solely positive or negative
announcements, with the exception of the Financial sector. Financial sector abnormal returns were positive with positive announcements, and vice versa.
Introduction
One wouldn’t need a room full of economists to understand the impact the years 2007/2008 had on the economic systems encompassing the developed, and undeveloped world for that matter. The “Great Recession” as it was later dubbed, grew to become the largest economic and financial crisis in recent memory, and the second largest in history, after the “Great Depression” of the 1930s. As asset bubbles burst and previously high financial ratings plummeted, central banks sought to stabilise markets. Traditional economics sets forth a series of conventional monetary
tools that central banks, such as the Federal Reserve System (FED), can use to curtail the impacts of developing financial crises, and eventually use to promote economic growth. One of these more classical tools, and the FED’s main operating instrument, is put simply; lowering interest rates. The FED did just that in hopes of regaining control amid the “Great Recession". By December 16th 2008, the had Fed significantly tightened its policy rate, reaching an unsurpassed low of a 0-25 basis point range on its federal funds target rate (Rosa, 2012, pp.1). Although the Fed had reached the "zero lower bound on its main operating instrument” (Rosa, 2012, pp.1), economic prospects were progressively worsening. The FED would need more unconventional tools in other to help ease sinking monetary conditions (Rosa, 2012, pp.1).
One new tool the Fed used (and the focus of this paper) was “Quantitative Easing”. Quantitative Easing policies aim at growing the monetary base. This can be achieved by asset purchase or lending programs (Fawley and Neely, 2013). The FED’s own quantitative easing policy involved conducting large scale asset purchases from domestic American markets (Yang and Zhou, 2017). The FED tasked the FOMC (Federal Open Market Committee) with implementing this policy. Appropriately, the FOMC established the LSAP (Large-Scale Asset Purchases) program (Rosa, 2012, pp.1), and began its equivalent of quantitative easing on the 25th of November 2008. This LSAP program purchased a substantial amount of assets varying from longer-term Treasury securities to housing agency debt and mortgage-backed securities, consequently expanding central bank reserves (Rosa, 2012).
The FED stated that it’s LSAP program was merely to improve credit conditions; a “credit easing” program (as dubbed by the Feds chairman Ben Bernanke). Unlike in pure quantitative easing the FED purportedly sought to merely to “improve the functioning of long-term bond
(Fawley and Neely, 2013, pp.55). Such programs are considered a “special case of quantitative easing if they increase the monetary base” (Fawley and Neely, 2013, pp.52), which results showed it did.
The FED was certainly not the only central bank to have used quantitative easing polices as a tool in response to the effects of the Great Recession. The BoE (Bank of England), ECB
(European Central Bank) and, BoJ (Bank of Japan), all used varying quantitative easing policies to alleviate recessions and bolster their economies. It’s important to note, that the FED’s and BoE’s quantitative easing programs/policies differed fundamentally to those of the ECB’s and BoJ’s. The Fed and BoE focused mainly on bond purchases instead of (the ECB’s and BoJ’s focus on) direct bank lending (Fawley and Neely, 2013, pp.56). Bond markets are traditionally more important than banks in the United States and England, and central banks had to choose methods that provided liquidity and support to their respective economic structures and financial systems (Fawley and Neely, 2013, pp.56). Studies conducted by Gagnon, Raskin, Remache, and Sack (2011) on the LSAP program revealed that long term U.S bond yields were indeed
decreased.
How exactly does Quantitative Easing work?
Quantitative easing essentially acts as a new tool to promote liquidity when policy rates reach their zero-lower bound. It is also tasked with increasing “the growth of nominal spending to a rate consistent with meeting the inflation target” (Dale et al, 2010, pp.11). As a central bank begins buying assets/securities and replacing them with its previously stored reserves, it increases the amount of money supply in the economy and reduces the supply of specific
assets/securities. As a result, it reduces liquidity premiums in the market. This in turn encourages investors and lenders to take on/make riskier loans, as less riskier alternatives aren’t available
(Mamaysky, 2018), promoting “greater levels of lending and borrowing” (Dale et al, 2010, pp.11). In the short-term, targeted asset/security prices would increase as their supply decreased, whereas in the medium to long-run non-targeted asset/security prices would react to the
increasing risk undertaken by lenders and investors as they diversify into alternatives
(Mamaysky, 2018). This process is known as the “wealth channel”. What’s more the increase in prices consequently reduces yields and lowers the costs of raising funds for companies on capital markets (Dale et al, 2010, pp.11).
Quantitative Easing also helps a central bank demonstrate that it is willing to do whatever is needed to maintain inflation targets, thus anchoring inflation expectations (Dale et al, 2010, pp.11), this is known as the “confidence channel”. The wealth channel however does open up the economy to price bubble risks. With the new increase in prices; equity, housing and bond prices are all susceptible to over pricing and potential creation of pricing bubbles as a result of quick monetary market shifts (Huston, & Spencer, 2018). Understanding the way in which high frequency asset markets (in particular domestic stock markets) react as a result of central bank’s initial announcements and eventual implementation of quantitative easing can therefore prove important. Due to the higher speed of price changes in equity markets, unlike in other asset markets (like housing markets for instance), the effects of in this case quantitative easing announcements should more visible.
This thesis aims to investigate and understand the reactions of American public equity markets to quantitative easing policy announcements issued by the FED and its operating entities. Grasping the way in which stock markets swing with announcements could prove important if a central bank is contemplating implementing a new quantitative easing policy. Especially when markets are exceptionally sensitive/volatile and any other potential deviations could trigger larger
problems, for investors, companies, and as such the economy overall. This research will also give some insight into the way the majority of investors at the time (during the thick of a
recession) viewed the use and implementation of quantitative easing policies, as stock markets in many respects represent the majority view of investors. What’s more, this research could also be beneficial for investors or traders who are attempting to interpret future movements in markets by looking back at market reactions.
Literature Review Is QE an Effective tool?
Although quantitative easing has a solid theoretical framework, the question still stands; Does quantitative easing actually work, and is it effective? Current research is mixed, and at times contradicting.
Rosa’s empirical results suggest that the effects of large scale asset repurchases by central banks are “not statistically different from an unanticipated cut in FED fund rates target” (Rosa, 2012, pp.2). This implies that quantitative easing can act as another effective tool, especially in times when policy rates are “stuck at the zero-lower bound” and central banks need another way to stimulate the economy (Rosa, 2012, pp.2). This is in line with the theoretical understanding presented further above and demonstrates quantitative easing’s capability.
De Haan and van den End (2018) however presented results that implied Quantitative Easing “can have perverse effects, which may work against the objectives of the central bank” (Haan & End, 2018, pp.59). These “perverse effects” came in the form of fast growing asset prices.
Although De Haan and van den End (2018) found that quantitative easing heightened asset prices and as such raised inflation, reducing potential threats of deflation (- a main goal of quantitative easing). The very growth of asset prices, could however be or turn into a pricing bubble, and thus
“central banks should closely monitor the signals that asset prices and interest rates give for future price stability” (Haan & End, 2018, pp.59). Failing to do so, the potential pop of an asset price bubble, developed by quantitative easing, could further deteriorate the economy. Meaning quantitative easing actually made matters worse overall.
De Haan and van den End do go on to note, that central bank policymakers such as Draghi (of the ECB) and Yellen (of the FED) recognise and admit that quantitative easing can promote financial imbalances. Even so, aforementioned policy makers believe that such risks are imperative in order to reach inflation objectives, and that such potential side-effects should be counteracted with macroprudential policy (Haan & End, 2018, 45).
Furthermore, empirical research also found that even though quantitative easing is effective at boosting asset prices, increasing asset prices don’t in turn necessarily always promote higher inflation (Haan & End, 2018, 60). According to Haan and van den End this suggests that
quantitative easing doesn’t have a 100% correlation with increasing inflation, and as such is not always effective as a deflationary tool/policy (Haan & End, 2018, 60). What’s more, Haan and van den End demonstrated that (via multiple economic channels) the purchase of government bonds to reduce bond yields (which in turn should raise inflation), can in the long-run, as a result of zero-bound interest rates, threaten economic growth and consequently inflation goals (Haan & End, 2018).
Effects on Bonds
As expected and stated theoretically above, Neely (2015) indicated that along with forward guidance, large scale term asset repurchase programs managed to “reduce expected long-term U.S. bond real and nominal yields and long-long-term foreign bond yields in dollars” (Neely, 2015, pp.110). Furthermore, these corresponding bond price increases had a correlation with
announcements, “too large to have been generated by chance” (Neely, 2015, pp.110). This shows that the unconventional policies implemented by the FED managed to “reduce long-term interest rates and the value of the dollar” (Neely, 2015, pp.110). Neely added that "central banks are not toothless when short rates hit the zero bound” (Neely, 2015, pp.110), meaning that quantitative easing policies, are again regarded to be an effective tool for central banks. Moreover, data showed that quantitative easing announcements associated with the amount of assets purchased had bigger impacts on the asset prices than any other announcements relating to either minor delays or reductions in purchases (Neely, 2015). Interestingly, FOMC announcements not related to the unconventional policies had inconsistent and only minor average impacts on asset prices, particularly at a high frequency (Neely, 2015).
Huston and Spencer (2018) further complement Neely’s results, finding that bond markets in response to the FED’s expansionary policies had also risen. However, upon closer inspection they discovered “evidence of a large and persistent bubble in Treasury bonds in 2011–2013 and again approaching…June of 2016” (Huston & Spencer, 2018, pp.374). This links back to Haan’s and van den End’s (2018) research that as previously stated, suggested quantitative easing came with “perverse effects” (bubbles) that could threaten the objectives of a central bank’s
expansionary policy. Mamaysky’s (2018) research revealed high variability in domestic short, medium, and long-term bond prices following the days of a quantitative easing announcements by the FOMC, BOE or ECB. Mamaysky goes on to state that “the majority of the QE effect on bonds happened quickly” and as described by previous research papers cited (Mamaysky, 2018, pp.39). Overall research seems to reach a consensus; quantitative easing works on bonds in the way it was theorised to, and announcements seem to have an almost immediate impact on bond prices.
Effects on Stocks
If bond prices were affected by quantitative easing, especially surrounding the announcements, could the same be observed for equity markets?
Bernanke and Kuttner found that broad stock market indexes such as the CRSP value-weighted index (a solely American stock index), on average witnessed a 1% price increase, for an
“unexpected cut of 25-basis-point in the federal fund rate” (Bernanke and Kuttner, 2005). These results were robust regardless of the choice of time window chosen to measure the market’s reaction, or the removal of outliers. What’s more Bernanke and Kuttner suggest that there is evidence of “larger market response to policy changes that are perceived to be relatively more permanent, and a smaller response to unexpected inaction on the part of the FOMC” (Bernanke and Kuttner, 2005, pp.1253). These results imply that the stock market would indeed react to quantitative easing, as it (as previously stated above) is an alternative way to cause a change in policy rates when traditional methods are no longer possible.
Huston and Spencer (2018) found contradicting results with regard to stock prices since the lows of the 2008 financial crisis. They suggested that there has been a dramatic increase in equity prices since the 2008-2009 trough, but different models contradict on whether there was an overvaluation/bubble of those prices (Huston and Spencer, 2018). Huston and Spencer (2018) do however suggest that FED quantitative easing played a large part in these price hikes. This research ties in with that of Bernanke and Kuttner. Quantitative easing does indeed affect equity market prices, but does this happen gradually or do announcements (like with bonds) play a central role?
Rosa assessed how different styles of announcements affected stock prices (represented by the S&P500 index). A more “dovish LSAP” announcement (one that is considerably more subdued)
demonstrated on average “a stock price increase of 0.9%”, which was seen as “substantial” due to this type of “dovish” statement representing “a lower bound of the impact of asset purchases on U.S. asset prices” (Rosa, 2012, pp.8). When assessing all types of LSAP announcements (including those deemed “dovish”, neutral, or aggressive) Rosa found stock prices to increase on average by 3% (Rosa, 2012, pp.8). Rosa then conducted the same style of research on the BOE’s quantitative easing announcements. Rosa indicated that “the response of U.K. asset prices is broadly in line with the reaction of U.S. asset prices to the Fed’s asset purchases” (Rosa, 2012, pp.14). However, there was one notable exception. U.K. domestic stock market prices (as represented by the FTSE100) did “not react to react to QE shocks” (Rosa, 2012, pp.14). These results were found to be robust and statistically significant. Joyce, Lasaosa, Stevens, and Tong (2011), also found similar results to Rosa with regard to the effects of the BOE’s quantitative easing policies on respective domestic equity prices. Using the “FTSE All-Share Index” Joyce et al found that “equity prices did not react in a uniform way in response to QE news” (Joyce et al, 2011, pp.143). Quantitative Easing announcements overall then tend to have a somewhat
immediate impact on equity prices, with American equities reacting stronger than that of England’s.
Mamaysky (2018), conducts research in similar ways to Stevens and Tong (2011), however focuses on a more inter-day long-term post announcement period. His results show that stock prices (at both country and industry levels), and equity implied volatilities, react over the weeks following quantitative easing announcements. Stock indexes (S&P500, Euro Stoxx 50, and FTSE100) representing the American, European, and British equity markets, had statistically significant and large aggregate returns of 49%, 28%, and 20% respectively (Mamaysky, 2018). Furthermore, Mamaysky saw another significant drop in the implied volatilities of the same
stock market indexes. These also amounted to a rather large economically significant drop of “53% in the US, 65% in Europe and 47% in the UK” (Mamaysky, 2018, pp.43). Mamaysky indicated that the effects of quantitative easing announcements found on stocks and their implied volatilities, were substantially larger than those found in previous studies, suggesting further research is required to fully understand the scale of impact (Mamaysky, 2018).
Effects on Volatility
Mamaysky mentioned the implied volatilities of stock market indexes, but what are those
exactly? One of these ‘implied volatilities’ is the CBOE Volatility Index or VIX (ticker symbol). It is an interpretation of the expected volatility in United States stock markets, calculated with the use of S&P 500 index options. The higher the price on the VIX, the higher the expected volatility, and as such stock markets are more sensitive and vulnerable.
Yang and Zhou (2017) found that FED quantitative easing policies in the treasury bond market seemed to have a significant spill over impact upon the VIX and thus the American stock market volatility (Yang and Zhou, 2017). Neely (2015), Huston & Spencer (2018), and Mamaysky (2018) (as stated above), found a strong relation between increasing bond prices and quantitative easing announcements in the short run. Combining this with the spillover channels Yang and Zhou (2017) discovered between treasury bond volatility and American stock market volatility, one could extrapolate and suggest that price changes in bond markets could also have spillover effects on equity prices.
Luca-Moench Effect
One interesting phenomenon pointed out by Mamaysky (2018) was the Lucca-Moench Effect, named by Lucca and Moench (2015) as the “pre-FOMC announcement drift”. Lucca and Moench (2015) revealed that since September 1994 till March 2011, during a 24 hour window
prior to a scheduled FOMC announcement, the S&P500 index had increased on average by 49 basis points (Lucca and Moench, 2015). Furthermore, over the ensuing trading days, pre-FOMC returns don’t drift back and, are significantly larger than other returns outside the 24-hour window preceding FOMC announcements (Lucca and Moench, 2015). Respectively, the significance of pre-FOMC returns is so large that since 1994 it encompasses 80% of annual realised excess stock returns, and what’s more, similar significant effects/returns are witnessed on other major foreign stock markets, such as the FTSE100 (Lucca and Moench, 2015).
In line with these findings Mamaysky (2018) suggest that any future research into the effects of quantitative easing upon stock markets should also consider assessing the 24-hour period prior to a quantitative easing announcement.
Intermezzo
Current research seems to lack on the effects quantitative easing announcements have upon the different sectors within equity markets, ergo, this thesis aims to understand the effects of quantitative easing announcements have upon different sectors within equity markets. Furthermore, due to the suggestion of Mamaysky (2018) and the work of Lucca and Moench (2015) on pre-announcement drifts (as seen above), this thesis will attempt to understand both the pre and post-announcement effects of quantitative easing announcements on different sectors within equity markets.
The FED’s quantitative easing policies have been touted as effective, and thus it’s policy’s and effects on the American stock market will be used for this research, so as to avoid a bias
generated from a poorly executed regime. Although the quantitative easing strategies used by the BOE and the FED (as explained further above) are similar, the scale of the BOE’s quantitative easing program was shorter than that of the Fed’s. Therefore, and due in part to the research of
Rosa (2012), this thesis will not investigate the BOE’s quantitative easing announcement impacts upon their domestic stock exchange.
Hypothesis
Given previous research and theoretical understanding presented above, it’s expected that that quantitative easing announcements will trigger an average abnormal price increase the American equity market as a whole.
The financial sector should have a stronger response to FED announcements in comparison to other sectors due to the nature of large scale asset repurchases and their direct impact on the balance sheets of banks and other financial institutions.
What is more, the impacts of announcements on all sectors are presumed to happen on the day of the announcement.
Pre-announcement drifts as researched by Lucca and Moench (2015), are also expected to affect abnormal returns in the days leading up to the announcements.
Methodology Stock Market Data
For statistical methods explained further below, an American stock market index was needed in order to benchmark the performance of the American stock market as a whole. For this the S&P500 composite price index was used.
In order to understand how different sectors within equity markets responded, it’s important to define each separate sector. The ‘Global Industry Classification Standard’ (GICS) classifies the general economy into 11 separate sectors and is widely used for classification within the
these separate sectors. The GICS sector indexes used, as well as additional information can be found on table 1.
Event Study Method
In order to understand how exactly Quantitative Easing announcements impacted these
individual sectors in the United States, multiple event studies were conducted. An event study is a statistical method that calculates the impact of an event (announcement) on the value of a firm, set of firms, asset ext. It does so by calculating the abnormal returns during an event window. An event window constitutes of a set time before, during and after the specific event. This paper used an event window of 7 days; with 3 of those days being pre-event, 1 day for the event, and 3 days post-event. To identify the day and whether its pre or post event, results are labelled with a time “t”; where negative “t” numbers indicate pre-event days, positive numbers post-event days, and zero for the event (announcement) day.
Abnormal returns are calculated by deducting “Normal Returns” from the Realised Returns of the sector equity indexes.
“"#$%&'() +,-.&$/” = “+,()2/,3 +,-.&$/” − “5%&'() +,-.&$/”
“Normal Returns" are the returns expected of the sectors, given no announcement. These were calculated using a market model and an estimation window (explained further below).
Announcement Dates
Quantitative Easing announcement dates around which to base the event studies, were acquired from Fawley and Neely’s (2013) research. Mamaysky (2018) also used an altered version of these same announcement dates. Fawley and Neely also included a short description of each announcement, which was useful for deciphering whether an announcement was positive or negative. In total 19 announcements were used (table 2). Announcements were limited to the
initial inception and heavy use periods of quantitative easing, so as to avoid periods where the economy had substantially recovered, and to be able to compare results with previous literature. In order to calculate the starting and ending dates for the event window, 3 trading days were deducted and added to the announcement dates. The estimation window was calculated by deducting 70 trading days from the starting date of the event window, for each announcement. Dates and additional information can be found in table 2.
Operations
“Datastream” was used to obtain the closing pricing data of the market index and each independent market sector index, for each estimation and event window.
Returns were then calculated for the benchmark market index and the separate market sector indexes. Returns were calculated by dividing ‘today’s’ closing price, by ‘yesterday’s’ closing price and then deducting 1.
+,-.&$ = (‘-%3(8’/’ :)%/2$; <&2:, / ‘8,/-,&3(8’/’ :)%/2$; <&2:,) − 1 With the list of appropriate returns for each event and estimation window for each sector, the market model was employed to calculate the “Normal Returns”. Using only the returns in the estimation window, a single variable regression was performed for each market sector index, using the benchmark stock market index (S&P500) returns, and market sector index returns.
+,-.&$/Sector Index= J + LSector Index∗ (+,-.&$/S&P500)
This regression was run for each separate sector index, over each of the estimation periods (11
sectors x 19 estimation windows = 209 regressions). For each regression standard error and
R-square was also calculated. This produced a regression that could then be used to calculate the Normal (expected) Returns for each of the sector indexes during the event windows. This was
done by inserting the appropriate S&P500 return for each day in the event windows, into each separate regression equation.
5%&'() +,-.&$/Sector Index, t = J + LSector Index∗ (+,-.&$/S&P500, t)
Sector index Normal returns were then deducted from Realised Returns, over each event window day to calculate all the abnormal returns (7days x 19 event windows x 11 sectors =1,463
Abnormal Returns). All abnormal returns are displayed in the ‘Raw Excel Results Data’ section.
Testing
In order to understand whether the generated abnormal returns where sizeable, significance tests were conducted on each abnormal return. This was done by dividing each abnormal return by its regression standard errors, in order to obtain a t-statistic. Each AR t-statistic was then tested at a 5% significance level. In the ‘Raw Excel Results Data’ section, each AR labelled with a ‘Yes’ in blue fill, was significant.
In each event window, for each firm, the total Cumulative Abnormal Returns (CAR) were calculated. This was done by summing independent stock index abnormal returns (AR) in in each event window for each sector. Additionally, buy-and-hold aggregated Cumulative
Abnormal Returns were also calculated. This was done by adding abnormal returns through the event window, as days passed by, replicating a ‘buy and hold’ return over the event window (i.e., ‘buy-&-hold’ CAR at t=-3 is equal to ARt=-3, ‘buy-&-hold’ CAR at t=-2 is equal to ARt=-3 + AR
t=-2,’ buy-&-hold’ CAR at t=-1 is equal to ARt=-3 + ARt=-2 + ARt=-1). Total Cumulative Abnormal
Returns for each firm per announcement are equal to the ‘buy-&-hold’ CAR on final event window days (t=3).
Average Abnormal Returns (AAR) were determined for each stock index, on every event window, by calculating the arithmetic mean of the event window abnormal returns.
Two similar types of Cumulative Average Abnormal Returns (CAAR) were calculated, one using the total (end of event window, t=3) Cumulative Abnormal Returns (CAR) and the other using the separate Buy-and-Hold CARs. The total end of period CAAR was calculated by finding the arithmetic mean of the total end of period CARs for each sector, whereas the Buy-and-Hold CAAR was the arithmetic mean of each of the separate Buy-Buy-and-Hold CARs. Each CAAR was then divided by it’s appropriate standard deviation adjusted for degrees of freedom, in order to find it’s t-statistic and then appropriate p-value. This was done for both time series data (1 sector index over all event windows) and cross-sectional data (1 event window over all sectors). All CAARs were then tested for statistical significance. T-statistics were adjusted for degrees of freedom and tested again at a 5% significance level.
Average Abnormal Returns, and CARs were also calculated for pre (t=-3 to t=-1), post (t=3 to
t=1), and announcement (t=0) abnormal returns for each sector index over all event windows.
These results are displayed in tables 6 and 7. Positive and Negative announcements
In order to assess whether the type of announcement impacted the sectors differently, the 19 announcements were split into separate groups (a positive and negative announcement), and the same event study and testing methodology was carried on each group of announcements.
Empirical Results Cross-Sectional Results
Each sector index preformed in a varied way across announcements. While one announcement could generate positive abnormal returns for e.g. the Health care sector, another sector could have negative abnormal returns. Some correlation between certain sectors and the
25th of November 2008. This date marked the first quantitative easing announcement in the FED’s history. The FED announced that it would begin large scale asset purchases amounting to 600 billion dollars. In theory this
announcement should be considered positive news as the FED is responding to
the financial crisis mentioned in the introduction. Prior to this announcement American equity markets as benchmarked by the S&P500 were particularly volatile. The “S&P500 Returns” graph above demonstrates the erratic returns of the S&P500, as calculated by the daily closing prices (MRKT Return (PI)) and the daily average of the intraday high and low price (MRKT Return
(Average)). Additionally, a graph of the daily high and low prices of the VIX is included to show how volatile the market was in the lead up to the announcement. The 25/11/2008 event window impacted the most sectors compared to any other announcement event windows. With 7 out of the 11 sectors exhibiting significant abnormal returns (a=5%) within the event window. Plotting the ‘Buy-&-Hold’ CARs of this event period for each of the sectors that exhibits significant abnormal returns reveals that there is no clear direction for CARs across sectors. Some sectors
-10% -5% 0% 5% 10% 8/13/08 9/13/08 10/13/08 11/13/08 S&P500 Returns (12/08/2008 to 19/11/2008) MRKT Return (PI) MRKTReturn (Average) Linear Trendline 0 30 60 90 8/12/08 9/12/08 10/12/08 11/12/08 VIX Price Index
(12/08/2008 to 19/11/2008)
CBOEVIX(PH) CBOEVIX(PL) Linear (CBOEVIX(PH))
exhibit stable positive CARs and others negative CARs.
Interestingly, both the Real Estate and Financial sectors exhibit significant positive pre-announcement drifts on this date, however this is contrasted by a significant negative pre-announcement drift for the Health Care sector. Upon closer inspection of the Financial Sector positive abnormal returns continue into
the day of the event (t=0), potentially indicating that the financial sector received the news in a positive manner. This is not seen in the real estate sector where abnormal returns stayed near but below zero (-0.28%), stabling CARs.
Although certain sectors present interesting significant abnormal returns around the event day, a cross-sectional analysis of all the sectors on the first day of quantitative easing announcements presents no clear pattern across firms. This is the case for all announcements. No particular announcement was able to show abnormal returns across all sectors in a one positive or negative direction. Leading to question the stated hypothesis that quantitative easing announcements will trigger an average abnormal price increase in the American equity market as a whole. In order to attempt to test this statistically a cross-sectional significance test of each of the event based -7.00% -2.00% 3.00% 8.00% 13.00% -3 -2 -1 0 1 2 3
'Buy-&-Hold' CARs (25/11/2008 event window)
REAL ESTATE INFO TECHNOLOGY
FINANCIALS CONSUMER DISCRETIONARY
UTILITIES HEALTH CARE
INDUSTRIALS -7.00% -2.00% 3.00% 8.00% -3 -2 -1 0 1 2 3 Financial Sector (25/11/2008 event window)
CAARs (average total CARs across one event) was conducted. All 19-event based CAARs were not statistically significantly different from zero. Results are shown in table 3. This indicates that no one single announcement managed to have an abnormal impact on every sector in the market at the same time, whether positive or negative.
Time Series Results
CAARs. In order to understand how individual sectors reacted over all the different event windows, it was necessary to calculate CAAR values. CAAR values reflect how a sector on average reacted to an announcement. A positive CAAR indicates that on average a sector saw positive abnormal returns over the event window. A negative CAAR representing the opposite. CAAR values were calculated in two different ways. Total CAAR values only average the end of period CARs (t=3) over the 19 event windows, whereas ‘Buy-and-Hold’ CAAR values average the ‘Buy-and-Hold’ CARs over the event windows. The graph below displays both types of CAAR for each of the sector indexes. CAARs vary in both magnitude and direction. The Real Estate Sector, Energy, Consumer Staples, Utilities, and Materials all show a negative Total CAAR value, indicating that on average the cumulative abnormal returns over the event window, are negative, implying that the announcements potentially have an overall negative effect on
-2.0% -1.5% -1.0% -0.5% 0.0% 0.5% 1.0% Real Estat e Ener gy Info rmati on T echn olog y Fina ncial Cons umer Disc retio nary Cons umer Stap les Utili ties Healt h Car e Indu strials Mater ials Telec om S ervice s
Time Series CAARs
these sectors. For Information Technology, Financial, Consumer Discretionary, Health Care, Industrials, and Telecommunication Services CAAR values are positive implying
announcements potentially have a positive effect on the sectors. In order to test whether the impact of announcements on the sectors over the all periods was significant, both sets of CAARs were tested for each firm. Tables 4 and 5 display the results of the tests. When testing the total end of period CAARs, only the Energy sector proved to have a total CAAR significantly different from zero. Testing the ‘Buy-and-Hold’ CAARs determined that the Energy,
Information Technology, Consumer Discretionary and Consumer Staples sectors all had a ‘buy-and-hold’ CAAR significantly different from zero. These tests make sense when analyzing the ‘time series CAAR’ graph above. As other sectors had comparably low CAARs. Real Estate and Financial Sectors however still seemed to have a strong CAARs, but nevertheless were proved statistically insignificant. This was due to the high variability of CARs in these sectors. Real Estate and Finance sectors had the highest times series total CAAR standard deviations of any sectors at 0.046954 and 0.034988 respectively. ‘Buy-and-Hold’ CAAR time series standard deviations were also the highest of any sectors at 0.043851 and 0.026556 respectively. This could signify that both these sectors are particularly susceptible to whether the announcement was positive or not. As seen above for the financial sector.
Pre, Post & Event AARs and CARs. In order to understand the average abnormal returns for each sector just before, during, and after all of the announcements, the average abnormal returns pre (t=-3 to t=-1), during (t=0),, and post (t=3 to t=1), announcement for each sector were pooled, and the arithmetic mean was calculated. These values are presented in table 6. These same values are graphed below. The financial sector stands out, with an Event AAR (0.60%) substantially higher than that of any other sector. This is inline with results from the time series
CAARs. Real Estate, Energy sectors, as expected from CAAR results, exhibit negative AARs across all periods. Informationation Technology and Consumer Staples show postive AARs
across all periods of the event window, however the maginitue of these AARs is smaller than that of the Real Estate, Energy, and Financial sectors.Utilities, Health Care, Industiral Material sectors, show smaller AARs with no clear direction. This is inline with the tests conducted above, as these sector CAAR’s were rejected on both sets of significance tests. Like the Financial sector ,Telecomunication Services shows a positive AAR on the event day, however both sectors have negative pre annoucement AARs. These results in particular seem to conradict the pre-annoucement drift theory at the sector level. Although the market as a whole might show abnormal returns in the day before annoucements, these results suggest it isn’t the case in every equity sector. In order to expand on this, the ‘buy-and-hold’ CARs were also plotted for each firm across the different
periods in the event window. As expected the same sectors behaved in the same manner. Real Estate, Energy, Information Technology, -0.4% -0.2% 0.0% 0.2% 0.4% 0.6% Real Estat e Ener gy Info rmati on T echn olog y Fina ncial Cons umer Disc retio nary Cons umer Stap les Utili ties Healt h Car e Indu strials Mater ials Telec om S ervice s
Pre, Post & Event Sector AARs
PRE AAR Event AAR POST AAR
-1.40% -0.90% -0.40% 0.10% 0.60% Real Estat e Ener gy Info rmati on T echn olog y Fina ncial Cons umer Disc retio nary Cons umer Stap les Utili ties Healt h Car e Indu strials Mater ials Telec om S ervice s
Pre, Post & Event Sector 'Buy-and-Hold' CARs
Consumer Discretionationary, Materials and Telecomunitcation Services all display what could be considered pre-annoucement drifts. However, Financial and Consumer Staples seem on average to swing in price direction with the annoucement. This could suggest that both sectors were on average not expecting the annoucement to be what it was. If this is true, the Financial sector was on average expecting a negative annoucement, and the consumer staples was on average expecting a positive or neautral annoucement.
Financial Sector. Over every event window Financial sector abnormal returns averaged out to near zero percent (0.09%), with ‘buy-and-hold’ CAAR also averaging out to near zero (-0.04%). Interestingly however, end of period total CAAR averaged out to 0.63%. This indicates that the Financial sector did to some degree experience a positive return as a result of announcements, however this CAAR was not statistically significant (a=5%). Upon closer inspection, certain announcements did deliver interesting significant results. On the 12/10/2010 event window the Financial sector suffered significant negative post event (t=1 to t=3), however on the event day itself (t=0), the sector experienced positive significant abnormal returns. For a clearer picture the ‘buy-and-hold’ CARs were graphed, over the days (t). Although the announcement resulted in a positive return initially, the
post announcement returns were significantly (a=5%) negative, dropping buy-and-hold CARs from near 0% at t=0, to negative
3.72% at t=-3. Three days after the 12/10/2010 announcement came the 15/10/2010
announcement, which generated contrasting results. By plotting the 15/10/2010 the ‘buy-and--4.00% -3.50% -3.00% -2.50% -2.00% -1.50% -1.00% -0.50% 0.00% 0.50% -3 -2 -1 0 1 2 3
hold’ CARs over the days (t), one can see that the 15/10/2010 announcement had a significant (a=5%) positive impact on abnormal returns post event, but significant negative abnormal returns on the event day (t=0). But how could two announcements so close to each other have such significantly different impacts? By looking on Table 2, the announcement descriptions provided by Fawley and Neely (2013), makes things clearer. On 12/10/2010 FOMC minutes were released indicating that FOMC members believed more accommodation was needed in the near future (Fawley and Neely, 2013). This
announcement does not build confidence and suggests that the implemented policies are not delivering the results that were to be expected or required (overall a negative announcement). On 15/10/2010, Bernanke (the chairman of the Federal Reserve at the time) expressed that the FED is ready to further easy policy (Fawley and Neely, 2013) (a positive announcement). In line with this qualitative data, the negative post 12/10/2010 announcement drift abnormal returns, and the positive 15/10/2010 post announcement drift make sense, and suggest that the type of
announcement (whether positive or negative) play a part in the direction of ‘buy-and-hold’ CARs post announcement. Why abnormal returns are positive on the day of a seemingly negative announcement and vice versa with a positive announcement, remains unanswered.
-3.50% -3.00% -2.50% -2.00% -1.50% -1.00% -0.50% 0.00% 0.50% 1.00% 1.50% -3 -2 -1 0 1 2 3
Positive and Negative Annoucements
As suggested above with the Financial sector, not all QE announcements are the same. Some signify positive news, others negative. By splitting announcements into either positive or negative and running the same time series statistics, sector responses to positive QE or negative QE can be assessed. Table 2 displays which announcements were classified as positive or negative news.
Positive Announcements. Table 8 displays the time series total CAARs and their significance tests for positive announcements. Results indicate that only the energy sector shows a significant negative reaction to the positive announcements. This was the same when all announcements were combined (positive and negative). When calculating the ‘buy and hold’ CAAR’s for positive announcements results differed slightly (Table 9). Energy, Information Technology, Consumer Discretionary and Telecom Service sectors all displayed statistically significant ‘buy and hold’ CAAR’s. When combining all announcements, the Consumer Staple sector showed a significant reaction however Telcom Services did not. The graph below displays both types of
sector CAARs over positive
announcement dates. Similar to when all announcement dates were combined, the Real Estate, Energy, and Consumer Staples sectors showed a negative reaction to positive announcements, whilst Information -3.000% -2.000% -1.000% 0.000% 1.000% 2.000% REAL EST ATE ENER GY INFO TEC HNOL OGY FINA NCIA LS CONS UMER DISC RETI ONARY CONS UMER STAP LES UTIL ITIE S HEAL TH CAR E INDU STRIA LS MATE RIAL S TEL ECOM SER
TimeSeries CAARs (Posative annoucements)
Technology, Financial, Consumer Discretionary, and Telecom Services showed a positive reaction to positive announcements. On table 10, the average pre-event, during event and post-event abnormal returns across sectors during positive announcements are presented. These results are graphed below. Results for each sector swing in the same general direction as when all announcements where combined. The Financial sector again shows a large positive average
abnormal return on the day of positive announcements. Average negative pre-announcement drifts are witnessed for the Financial sector (previously seen) and also for the Real-Estate sector. Energy again shows a negative average abnormal return across all slots. The same overall results are again seen with the CARs, with the Financial sector again reacting on the day of
the announcement counter to its pre-announcement drifts.
-0.70% -0.20% 0.30% 0.80% 1.30% REAL EST ATE ENER GY INFO TEC HNOL OGY FINA NCIA LS CONS UMER… CONS UMER STAP LES UTIL ITIE S HEAL TH CAR E INDU STRIA LS MATE RIAL S TEL ECOM SER
Pre/Event/Post AAR (Posative annoucements)
PRE AAR Event AAR POST AAR
-2.30% -1.80% -1.30% -0.80% -0.30% 0.20% 0.70% 1.20% REAL EST ATE ENER GY INFO TEC HNOL OGY FINA NCIA LS CONS UMER… CONS UMER STAP LES UTIL ITIE S HEAL TH CAR E INDU STRIA LS MATE RIAL S TEL ECOM SER
Pre/Event/Post CAR (Posative annoucements)
Negative Announcements. Tables 12 and 13 display the CAAR results and significance tests for each sector during negative announcements. Real Estate, Energy, Information Technology, Financials, Consumer Discretionary and Consumer Staples sectors all show a statistically significant CAARS. However, most sectors seem to react in the same manner to positive or
negative announcements. For example, the Energy sector shows negative CAAR regardless of whether the announcement is positive or negative. There are however some exceptions. The Financial sector, displays negative CAARs to negative announcements and positive CAARS to positive announcements. What’s more, the Material sectors displays larger negative CAARS to negative announcements, than positive CAARs to positive announcements. When looking at the average pre-event, during event and post-event
abnormal returns across sectors for negative
announcements (table 14), the Financial sector again shows interesting results. Average Abnormal returns show a negative pre-annoucement -1.40% -0.90% -0.40% 0.10% 0.60% REAL EST ATE ENER GY INFO TEC HNOL OGY FINA NCIA LS CONS UMER DISC RETI ONARY CONS UMER STAP LES UTIL ITIE S HEAL TH CAR E INDU STRIA LS MATE RIAL S TEL ECOM SER
Time Series CAARs (Negative annoucements)
Total CAAR Buy&Hold - CAAR
-1.10% -0.90% -0.70% -0.50% -0.30% -0.10% 0.10% 0.30% 0.50% REAL EST ATE ENER GY INFO TEC HNOL OGY FINA NCIA LS CONS UMER DISC RETI ONARY CONS UMER STAP LES UTIL ITIE S HEAL TH CAR E INDU STRIA LS MATE RIAL S TEL ECOM SER
Pre/Event/Post AAR (Negative annoucements)
drift, however these abnormal returns remain negative through and after the negative annoucements unlike with positive annoucements. This is more clearly demonstarted when ploting the CARs. With Negative Annoucements the Real Estate sector also presents a positive
pre annoucment drift, however apon annoucement abnormal returns show a sharp decrease. This suggests that not all sectors desphire an annoucement in the same way, and as such react in a consitent way to annoucements regardless of what they bring. The Financial sector seems more responsive and senetitive to annoucments and style of annoucements overall.
Conclusion
Results suggest that across all sectors there are days of statistically significant abnormal returns in the event windows surrounding quantitative easing announcements. Energy, Information Technology, Consumer Discretionary, and Consumer Staples show statistically significant cumulative average abnormal returns over all 19 quantitative easing announcements. Real Estate and Financial Sectors also revealed substantial abnormal returns, however these were not
statistically significant at a=5%, due to the high variability of results across announcements. Sectors did not react in a uniform way to any of announcements, with some exhibiting positive abnormal returns, whilst others negative abnormal returns. What is more, no one single
-1.20% -0.70% -0.20% 0.30% 0.80% REAL EST ATE ENER GY INFO TEC HNOL OGY FINA NCIA LS CONS UMER DISC RETI ONARY CONS UMER STAP LES UTIL ITIE S HEAL TH CAR E INDU STRIA LS MATE RIAL S TEL ECOM SER
Pre/Event/Post CAR (Negative annoucements)
announcement caused all sectors to react abnormally in a statistically significant manner, regardless of whether each sector’s abnormal returns were positive or negative. This further suggested that different announcements impact different sectors in different ways. One interesting observation in the Financial and Consumer Staples sectors was how
pre-announcement abnormal returns were on average opposite to abnormal returns during and after the announcement. For example, the Financial sector on average exhibited negative abnormal returns in the pre-announcement window, but positive abnormal returns during the
announcement and post-announcement periods. This suggested that the pre-announcement drift theory could potentially not affect every sector in the same manner or at all. What is more, when splitting announcements into solely positive and negative announcements, no sector reacted differently to the style of announcement with the exception of the Financial sector, (the Financial showed negative average abnormal returns for negative announcements and positive average abnormal returns for positive announcements). All in all, although the results discovered where interesting no single trend over all sectors could be identified. Each sector reacted in its own way to announcements, with the Financial sector being the most sensitive and volatile to
annoucements.
Recommendations
An interesting route for future research could be to further break down sectors into industries. The GICS Sectors can be further broken down into 24 separate industries, which could give further insight into the reasons why certain sectors seem in particular to be more responsive to quantitative easing announcements.
Future research should also focus on the Financial and Consumer Staples sectors in order to further understand why exactly on average pre-announcement abnormal returns are opposite to
that of the event and post announcement periods. Such research could also help understand whether pre-announcement drifts as suggested by Lucca and Moench (2015) differ among sectors.
References
Bernanke, B., Kuttner, K. (2005). What Explains the Stock Market’s Reaction to Federal Reserve Policy? Journal of Finance 60 (3), pp. 1221–1257.
Dale, S., Proudman, J., & Westaway, P. (2010). The inflation-targeting regime in the United Kingdom: A view from Threadneedle Street. Oxford Review of Economic Policy, 26(1), pp. 3-14.
Fawley, B,W., Neely, C,J. (2013). Four Stories of Quantitative Easing. Federal Reserve Bank of St. Louis Review, 95(1), pp. 51-88.
Gagnon, Joseph; Raskin, Matthew; Remache, Julie and Sack, Brian. "The Financial Market Effects of the Federal Reserve's Large-Scale Asset Purchases'.' International Journal of Central Banking, March 2011b, 7(1), pp. 3-4.
Haan, L., End, J. (2018). The Signalling Content of Asset Prices for Inflation: Implications for Quantitative Easing. Economic Systems 42, pp. 45-63.
Huston, J., & Spencer, R. (2018). Quantitative easing and asset bubbles. Applied Economics Letters, 25(6), pp. 369-374.
Joyce, Michael A.S.; Lasaosa, Ana; Stevens, Ibrahim and Tong, Matthew. "The Financial Market Impact of Quantitative Easing in the United Kingdom:' International Journal of Central Banking, September 2011, 7(3), pp. 113-61.
Lucca, D., Moench, E. (2015) The Pre-FOMC Announcement Drift. The Journal of Finance 70, pp. 329-371.
Mamaysky, H. (2018). The Time Horizon of Price Responses to Quantitative Easing. Journal of Banking and Finance 90, pp. 32-49.
Neely, C. (2015). Unconventional Monetary Policy Had Large International Effects. Journal of Banking and Finance 52, pp. 101-111.
Rosa, C. (2012). How “Unconventional” Are Large-Scale Asset Purchases? The Impact of Monetary Policy on Asset Prices. Federal Reserve Bank of New York Staff Reports no.560.
Vuolteenaho, T., (2002) What Drives Firm-Level Stock Returns?. The Journal of Finance. Vol. 57, pp. 233-264
Yang, Z., Zhou, Y. (2017). Quantitative Easing and Volatility Spillovers Across Countries and Asset Classes. Management Science 63(2), pp. 333-354.
Tables Table 1
[Sector Indexes]
(GICS) Sector Ticker Number of Constituents
Real Estate IXRE 33
Energy SPN 31
Information Technology S5INFT 69
Financial SPF 69
Consumer Discretionary S5CONS 81
Consumer Staples S5COND 33
Utilities S5UTIL 28
Health Care S5HLTH 63
Industrials S5INDU 70
Materials S5MATR 25
Table 2
[Announcement Dates]
Announcement Dates Event Type Short Description Positive or Negative? 25/11/2008 FOMC Statement Release LSAPs first announced. Positive
01/12/2008 Bernanke Speech Expanding QE to
Treasuries proposed Positive
16/12/2008 FOMC Statement Release Expanding QE to Treasuries proposed by FOMC
Positive
28/01/2009 FOMC Statement Release FED confirms it will
expand to treasuries Positive
18/03/2009 FOMC Statement Release LSAPs expanded Positive
12/08/2009 FOMC Statement Release LSAPs dialed down Negative
23/09/2009 FOMC Statement Release LSAPs dialed down Negative
04/11/2009 FOMC Statement Release LSAPs reduced Negative
10/08/2010 FOMC Statement Release Balance sheet held stable
Positive
21/09/2010 FOMC Statement Release Suggestion of more QE Negative
12/10/2010 FOMC “minutes” Release FOMC members believed more accommodation was needed in the near future
Negative
15/10/2010 Bernanke Speech FED is ready to further easy policy
Positive
03/11/2010 FOMC Statement Release Second round of QE announced
Negative
22/06/2011 FOMC Statement Release Second round of QE finished
Positive
21/09/2011 FOMC Statement Release Maturity Extension program announced
Negative
20/06/2012 FOMC Statement Release Maturity Extension program expanded
Positive
22/08/2012 FOMC “minutes” Release FOMC members believed more accommodation was needed in the near future
Negative
13/09/2012 FOMC Statement Release Third round of QE
announced Negative
12/12/2012 FOMC Statement Release Third round of QE
expanded Positive
Note: These announcement dates were acquired from Fawley and Neely’s (2013) research. This includes all ‘announcement dates’, ‘event types’ and ‘ short description’.
Table 3
[Cross-Sectional CAAR significance tests]
Event Day Cross-sectional CAAR Std Dev t0 P-value Reject? (a=5%) 25/11/2008 1.42% 4.87% 0.96874562 0.355518191 NO 01/12/2008 0.20% 4.71% 0.140538733 0.891024914 NO 16/12/2008 -0.01% 3.00% -0.010577649 0.991768465 NO 28/01/2009 -0.40% 4.69% -0.286015596 0.780703575 NO 18/03/2009 -0.84% 6.46% -0.429145378 0.676915854 NO 12/08/2009 -0.34% 1.27% -0.88600413 0.396421481 NO 23/09/2009 -0.01% 1.57% -0.020917084 0.98372328 NO 04/11/2009 -0.07% 2.10% -0.110008692 0.914578887 NO 10/08/2010 0.41% 1.13% 1.214270436 0.252540807 NO 21/09/2010 -0.40% 1.80% -0.744717699 0.4735806 NO 12/10/2010 -0.34% 1.96% -0.571681583 0.580159625 NO 15/10/2010 -0.04% 1.16% -0.125927287 0.902285302 NO 03/11/2010 -0.03% 1.47% -0.063704128 0.950461235 NO 22/06/2011 0.03% 1.39% 0.064728558 0.949665812 NO 21/09/2011 -0.40% 2.42% -0.546435286 0.596747025 NO 20/06/2012 -0.28% 1.14% -0.802107473 0.441135476 NO 22/08/2012 -0.41% 1.07% -1.274859138 0.23118297 NO 13/09/2012 -0.15% 0.69% -0.720348849 0.487806442 NO 12/12/2012 0.18% 0.76% 0.80755815 0.438131949 NO
Table 4
[Time Series Total CAAR significance tests]
Sector Index CAAR Std Dev t0 P-value Reject?
(a=5%) Real Estate -0.868% 0.046954 -0.8059 43.08% NO Energy -1.877% 0.029215 -2.8007 1.18% YES Information Technology 0.623% 0.014959 1.8159 8.61% NO Financial 0.630% 0.034988 0.7852 44.25% NO Consumer Discretionary 0.889% 0.027205 1.4250 17.13% NO Consumer Staples -0.374% 0.013851 -1.1763 25.48% NO Utilities -0.075% 0.027256 -0.1198 90.60% NO Health Care 0.240% 0.021117 0.4960 62.59% NO Industrials 0.013% 0.020799 0.0271 97.87% NO Materials -0.526% 0.020121 -1.1385 26.98% NO Telecom Services 0.470% 0.020851 0.9818 33.92% NO Table 5
[Time Series ‘Buy-and-Hold’ CAAR significance tests]
Sector Index CAAR Std Dev t0 P-value Reject?
(a=5%)
Real Estate -0.683% 0.043851 -1.7950 7.49% NO
Energy -0.792% 0.017682 -5.1668 0.00% YES
Information Technology 0.458% 0.011616 4.5460 0.00% YES
Financial -0.037% 0.026556 -0.1609 87.24% NO
Consumer Discretionary 0.516% 0.017607 3.3804 0.10% YES
Consumer Staples -0.189% 0.010442 -2.0926 3.83% YES
Utilities 0.009% 0.019711 0.0519 95.87% NO
Health Care 0.055% 0.015259 0.4192 67.58% NO
Industrials 0.015% 0.013037 0.1363 89.18% NO
Materials -0.150% 0.014346 -1.2033 23.10% NO
Table 6
[Time Series AARs]
Sector PRE-AAR (t=-3 to t=-1) Event-AAR (t=0) POST-AAR (t=1 to t=3) Real Estate -0.12% -0.36% -0.05% Energy -0.12% -0.40% -0.37% Information Technology 0.10% 0.17% 0.05% Financial -0.12% 0.60% 0.13% Consumer Discretionary 0.11% 0.14% 0.14% Consumer Staples -0.01% -0.24% -0.04% Utilities 0.02% -0.05% -0.03% Health Care 0.03% -0.07% 0.08% Industrials 0.05% -0.12% 0.00% Materials -0.05% 0.02% -0.13% Telecom Services -0.03% 0.24% 0.11% Table 7
[Time Series CARs]
Sector PRE-CAR (t=-3 to t=-1) Event-CAR (t=0) POST-CAR (t=1 to t=3) Real Estate -0.47% -0.71% -0.89% Energy -0.22% -0.77% -1.38% Information Technology 0.25% 0.48% 0.66% Financial -0.48% 0.24% 0.31% Consumer Discretionary 0.28% 0.47% 0.76% Consumer Staples 0.06% -0.26% -0.42% Utilities 0.14% 0.00% -0.12% Health Care 0.05% 0.01% 0.08% Industrials -0.05% 0.03% 0.07% Materials -0.09% -0.13% -0.21% Telecom Services 0.15% 0.15% 0.34%
Table 8
[Time Series Total CAAR significance tests (positive announcements)]
Sector Index CAAR Std Dev t0 P-value Reject?
(a=5%) Real Estate -0.487% 0.064294 -0.2396 81.60% NO Energy -2.955% 0.034541 -2.7055 2.42% YES Information Technology 0.373% 0.016797 0.7020 50.04% NO Financial 1.781% 0.040765 1.3817 20.04% NO Consumer Discretionary 1.410% 0.036873 1.2092 25.74% NO Consumer Staples -0.616% 0.018725 -1.0410 32.50% NO Utilities 0.134% 0.036636 0.1152 91.08% NO Health Care 0.163% 0.026593 0.1934 85.09% NO Industrials 0.153% 0.027402 0.1771 86.33% NO Materials -0.099% 0.019283 -0.1622 87.48% NO Telecom Services 0.887% 0.020306 1.3814 20.05% NO Table 9
[Time Series ‘Buy-and-Hold’ CAAR significance tests (positive announcements)]
Sector Index CAAR Std Dev t0 P-value Reject?
(a=5%)
Real Estate -1.001% 0.058383 -1.4341 15.61% NO
Energy -1.122% 0.021500 -4.3652 0.00% YES
Information Technology 0.415% 0.013558 2.5602 1.27% YES
Financial 0.348% 0.033759 0.8629 39.12% NO
Consumer Discretionary 0.632% 0.023573 2.2434 2.81% YES Consumer Staples -0.218% 0.013462 -1.3577 17.90% NO
Utilities 0.162% 0.025648 0.5273 59.96% NO
Health Care 0.036% 0.019221 0.1547 87.75% NO
Industrials -0.032% 0.016645 -0.1594 87.38% NO
Materials 0.128% 0.013024 0.8252 41.21% NO
Table 10
[Time Series AARs (positive announcements)]
Sector PRE-AAR (t=-3 to t=-1) Event-AAR (t=0) POST-AAR (t=1 to t=3) Real Estate -0.40% 0.29% 0.14% Energy -0.18% -0.57% -0.61% Information Technology 0.15% -0.05% -0.01% Financial -0.12% 1.34% 0.27% Consumer Discretionary 0.07% 0.36% 0.28% Consumer Staples 0.06% -0.50% -0.10% Utilities 0.06% -0.06% 0.00% Health Care 0.03% -0.13% 0.07% Industrials 0.02% -0.04% 0.04% Materials -0.04% 0.26% -0.08% Telecom Services 0.01% 0.13% 0.25% Table 11
[Time Series CARs(positive announcements)]
Sector PRE-CAR (t=-3 to t=-1) Event-CAR (t=0) POST-CAR (t=1 to t=3) Real Estate -1.36% -0.92% -0.67% Energy -0.14% -1.12% -2.11% Information Technology 0.31% 0.39% 0.53% Financial -0.64% 0.99% 1.12% Consumer Discretionary 0.12% 0.57% 1.16% Consumer Staples 0.20% -0.32% -0.61% Utilities 0.34% 0.13% -0.01% Health Care 0.11% -0.05% -0.01% Industrials -0.25% 0.02% 0.17% Materials -0.02% 0.13% 0.27% Telecom Services 0.35% 0.15% 0.66%
Table 12
[Time Series Total CAAR significance tests (negative announcements)]
Sector Index CAAR Std Dev t0 P-value Reject?
(a=5%)
Real Estate -1.292% 0.016489 -2.3498 4.67% YES
Energy -0.679% 0.016478 -1.2367 25.13% NO Information Technology 0.901% 0.013022 2.0765 7.15% NO Financial -0.648% 0.023137 -0.8408 42.49% NO Consumer Discretionary 0.311% 0.008013 1.1641 27.79% NO Consumer Staples -0.104% 0.004653 -0.6714 52.09% NO Utilities -0.306% 0.012248 -0.7507 47.43% NO Health Care 0.327% 0.014359 0.6823 51.43% NO Industrials -0.143% 0.011106 -0.3868 70.90% NO Materials -1.000% 0.021085 -1.4222 19.28% NO Telecom Services 0.006% 0.021642 0.0082 99.37% NO Table 13
[Time Series ‘Buy-and-Hold’ CAAR significance tests (negative announcements)]
Sector Index CAAR Std Dev t0 P-value Reject?
(a=5%)
Real Estate -0.324% 0.016626 -1.5460 12.72% NO
Energy -0.419% 0.011195 -2.9733 0.42% YES
Information Technology 0.498% 0.009069 4.3563 0.01% YES
Financial -0.458% 0.014064 -2.5839 1.21% YES
Consumer Discretionary 0.381% 0.006196 4.8830 0.00% YES
Consumer Staples -0.155% 0.005502 -2.2337 2.91% YES
Utilities -0.158% 0.009461 -1.3287 18.88% NO
Health Care 0.076% 0.009190 0.6597 51.19% NO
Industrials 0.067% 0.007278 0.7274 46.97% NO
Materials -0.452% 0.015197 -2.3585 2.15% YES
Table 14
[Time Series AARs (negative announcements)]
Sector PRE-AAR (t=-3 to t=-1) Event-AAR (t=0) POST-AAR (t=1 to t=3) Real Estate 0.20% -1.08% -0.27% Energy -0.06% -0.21% -0.10% Information Technology 0.06% 0.41% 0.11% Financial -0.12% -0.22% -0.02% Consumer Discretionary 0.16% -0.11% -0.02% Consumer Staples -0.08% 0.06% 0.03% Utilities -0.04% -0.03% -0.06% Health Care 0.03% -0.01% 0.09% Industrials 0.08% -0.22% -0.06% Materials -0.06% -0.25% -0.19% Telecom Services -0.07% 0.36% -0.05% Table 15
[Time Series CARs (negative announcements)]
Sector PRE-CAR (t=-3 to t=-1) Event-CAR (t=0) POST-CAR (t=1 to t=3) Real Estate 0.52% -0.47% -1.13% Energy -0.30% -0.38% -0.56% Information Technology 0.19% 0.58% 0.80% Financial -0.30% -0.58% -0.59% Consumer Discretionary 0.46% 0.36% 0.32% Consumer Staples -0.10% -0.19% -0.21% Utilities -0.09% -0.14% -0.24% Health Care -0.02% 0.07% 0.18% Industrials 0.18% 0.03% -0.03% Materials -0.17% -0.43% -0.75% Telecom Services -0.08% 0.14% -0.01%
