The effect of Quantitative Easing on the stock market in the UK

(1)

1

The effect of Quantitative Easing on the stock market in the UK.

Bachelor Thesis: Economics and Finance University of Amsterdam

Timon Kager – 10508791 Supervisor: Ryan van Lamoen

26 June 2018

Abstract

This paper examines whether unconventional monetary policy of the BoE, also known as Quantitative Easing (QE), has had an effect on the stock market in the UK. To analyze the influence of QE on stocks, a time series analysis was used. The variables in this analysis have been checked for stationarity using the Augmented Dickey-Fuller test. The results show that QE has had no effect on the stock market in the UK. Since this does not correspond with findings from previous literature, this result offers new insights into the effect of

unconventional monetary policy. This also provides scope for follow-up research into the impact of QE.

(2)

2

Statement of originality

This document is written by Timon Kager 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.

(3)

3

Table of content

1. Introduction………4

2. Literature review………6

2.1 Effects of QE on financial markets………6

2.2 The transmission channels of QE………..8

2.3 Hypotheses ………...10

3. Dataset………..10

4. Methodology….………11

4.1 Expectation of the regression coefficients………13

5. Results ………..15

6. Conclusion ………18

7. References ……….20

(4)

4

1. Introduction

The financial crisis of 2007-2008 caused an economic recession worldwide. After the bankruptcy of the American Lehman Brothers, in September 2008, instability in financial markets reached a peak. Banks no longer dared to lend each other money and the conditions for lending were tightened. As a result, the confidence of investors and consumers in the economy collapsed completely (Joyce, Tong, & Woods, 2011, p.200).

These consequences of the crisis were also present in the United Kingdom (UK). Therefore, the central bank of the UK, the Bank of England (BoE), was forced to ease monetary policy. The conventional measure of the Monetary Policy Committee (MPC), to lower the Bank Rate from 5 percent to the zero lower bound of 0.5 percent, was not effective enough to stabilize financial markets (Joyce, Lasaosa, Stevens, & Tong, 2011, p.113). That is why the BoE applied an unconventional monetary policy in 2009 to prevent a greater

economic depression (Joyce, Miles, Scott, & Vayanos, 2012, p.272).

The unconventional policy is also referred to as Quantitative Easing (QE). QE means that the central bank expands its balance sheet by buying assets on a large scale. In the UK, the BoE bought in particular government bonds, also known as gilts. By buying up these gilts the demand and the amount of money in circulation went up. As a result, the prices of the bonds increased and it became more attractive for private sector institutions to sell these securities to the BoE. With the increased liquidity of the sold gilts, these financial institutions were then able to issue new loans. This stimulated the number of investments, the nominal spending and thus the entire economy (Joyce & Spaltro, 2014, pp.7-8).

Although QE can improve financial market instability, it can also increase the

probability of bubble formation. When the MPC buys gilts, the prices of these securities rise. Due to the negative relationship between bond prices and yields, the latter will fall. Private sector institutions such as commercial banks, pension funds and insurance companies see the yield of gilts declining in their portfolio. In order to keep the average return on the portfolio constant, the private parties are looking for riskier assets with a higher yield (Buch,

Eickmeier, & Prieto, 2014, p.12). This is also called search for yield and can result in bubble formation. The increased demand for riskier securities, such as corporate bonds and stocks, causes a price increase of these assets. When these riskier assets are overvalued with respect to their fundamental value, a bubble is created. This undesirable side effect can lead to instability in several financial markets. That is why QE is not used as a standard measure to stimulate economic growth (Martinez-Miera & Repullo, 2017, p.351).

(5)

5

In the UK a lot of research has been done into the influence of unconventional measures on inflation, output and bank lending (Kapetanios, Mumtaz, Stevens, & Theodoridis, 2012). However less attention has been paid to the effect that QE has on

equities. This has been researched for Japan. Kurihara (2006) examined in his paper whether the balance sheet expansion of the Bank of Japan (BoJ) has a direct influence on the stock market in Japan. Joyce et al. (2011) also looked at the influence of a QE announcement on the stock market in the UK. Possibly, in addition to this direct effect, there is also an indirect effect via variables that are closely linked to the stock market. However, there is not much existing literature on these indirect effects of QE. This paper aims to fill this gap by

investigating both, the direct and indirect effects of QE on the stock market in the UK. It is examined at the extent to which QE contributes to this indirect effect on the stock market. In addition, it is being investigated whether a possible impact of QE has been positive or

negative. In this way it can be evaluated whether the QE policy of the BoE has also stimulated the stock market in the UK.

For several market participants it is of value to know what the influence of QE on the share price is. Investors will have an interest in this information in determining their strategy to trade shares. Furthermore, it gives extra insight to policymakers, such as central bankers, about what the consequences of QE can be. It is true that stimulating the stock market does not fall under the QE policy. Nevertheless, it is important for these central bankers to know whether QE causes unwanted effects in this market.

This thesis will use a time series analysis to examine the effects of QE on the stock market. As a dependent variable in the regression equation, the FTSE100 was taken, as a proxy for the British stock market. Four explanatory variables have been chosen, which have an effect on the FTSE100. These drivers are the sterling/US dollar exchange rate, the Federal funds rate, the yield on 10-year government bonds and the US stock market. QE is added as a dummy variable to determine the direct effect. The possible indirect effects are examined by the interaction between the QE dummy variable and the explanatory variables. On the basis of a Chow test, it is then checked whether these interaction variables have a joint influence on the stock market. Use is made of daily data from DataStream. The period that is analyzed is from 31 December 2004 to 24 May 2018. Within this timeframe, QE was applied by the BoE from 2009 to 2012. Based on the regression results, a statement is finally made as to whether QE has an effect on the stock market in the UK.

This thesis is structured as follows. Section 2 consists of the review of literature on the effect of QE on various financial markets. Also the transmission channels are discussed in this

(6)

6

section and a hypothesis is formulated. In section 3 the data is described and in section 4 the methodology is explained. The results are then discussed in section 5 and section 6 provides a conclusion.

2. Literature review

This section describes the effects of QE on various financial markets. Furthermore, the transmission channels of the BoE are discussed which the central bank uses to influence the economy. Thereafter, a hypothesis is formulated about the impact of QE on share prices in the UK.

2.1.Effects of QE on financial markets

Quantitative Easing was applied in several countries in response to the financial crisis of 2007-2008. The Federal Reserve (Fed) announced on 25 November 2008 that it would implement this unconventional monetary policy in the United States (US) (Fawley & Neely, 2013, p.60). The Bank of England followed this example and reported on 5 March 2009 that it would undertake a Large Scale Asset Purchase (LSAP) program (Joyce et al., 2011, p.123) Later, on 22 January 2015, Mario Draghi announced, on behalf of the European Central Bank (ECB), that an Expanded Asset Purchase Program (EAPP) would also be applied in the Eurozone. This was an extension of the earlier Public Sector Purchase Program (PSPP), started in September 2014 (Claeys, Leandro, & Mandra, 2015) This had an effect on several financial markets.

The stock market was one of these markets where QE had an effect. Although this market was not part of QE, there are strong signs of a positive relationship between share prices and this unconventional monetary policy. For example, the Standard & Poor’s 500 (S&P500), the stock index in the US, rose by more than 50 percent after the application of QE by the Fed in 2008. This, while during the financial crisis this index fell by 30 percent. In addition to the S&P500, the stock index in the UK also rose after the BoE had implemented a QE policy. Between 2008 and 2009 this index, the FTSE100, fell by 29 percent, but in

October 2011 an increase of 30 percent was observed. Lima, Vasconcelos, Simão, and de Mendonça (2016) investigated this growth on equity markets in the US and the UK. Using an Autoregressive-Distributed Lag (ARDL) model, they concluded that QE had a positive effect on the stock markets in these countries.

This effect of QE on equity markets may be reinforced by the international

(7)

7

lead to an increase in the index in the other country. Eun and Shim (1989) have found empirical evidence for this interdependent relationship. Their research also showed that a change in share prices in the US has the greatest influence on stock markets in other countries.

In addition to the impact on the stock market, QE initially had an effect on the bond market in the UK. This was because the BoE increased its balance sheet by buying mainly government bonds from private sector institutions. During the crisis these gilts were for the most part held by insurance companies, pension funds and commercial banks (Benford, Berry, Nikolov, & Young, 2009, p.96). Because these assets were bought on a large scale, the

aggregate demand increased. As a result, the prices of the gilts rose and the yield declined. Medium to long-term yields in particular declined by 100 basis points (Joyce et al., 2011, pp.155-156). The result was that it became more attractive for private parties to sell the assets to the BoE. With the increased liquidity of the sold bonds, these financial institutions were then able to issue new loans. This stimulated consumption, the number of investments and thus the entire economy (Joyce & Spaltro, 2014, pp.7-8). The increased liquidity also ensured that, in addition to bonds, the expenditure on other products also increased. This caused inflation in several markets. This made a first step to reach the 2 percent inflation target of the BoE (Joyce & Tong, 2012, F348).

The BoE announced on 5 March 2009 that it would buy gilts for an amount of £75 billion. This amount was gradually expanded to a total of £200 billion in 2010. Eventually in 2012, the BoE had bought up assets for £375 billion in total over two QE rounds, QE1 and QE2. (Fawley & Neely, 2013 p.63). According to Joyce et al. (2011), the UK’s asset purchases program was similar in size to that of the US. However, in addition to this

similarity, there were also differences. Especially in the impact on various financial markets. Gagnon, Raskin, Remache, and Sack (2011) for example, found that the effect on Mortgage-Backed Securities (MBS) yields was greater in the US than in the UK. This was because the Fed initially expanded its balance sheet by purchasing MBS from the private sector. With this, it wanted to lower the mortgage rates and increase the amount of credit for buying houses. This aimed to stabilize the housing market. In addition to the MBS, the Fed also increased its balance sheet by purchasing US Treasury debt. The purchased Treasury debt would reduce long-term interest rates and make investment more attractive. In this way the Fed wanted to stimulate the economy (Fratzscher, Lo Duca, & Straub, 2018, p.333)

The MBS and US Treasuries were bought in large quantities in three different QE rounds. In QE1 the Fed spent more than $1.000 billion on MBS and $300 billion on US Treasury debt. This was done over a period of one year, between 2009 and 2010. From 2010

(8)

8

onwards, US Treasury spending was increased by $600 billion in QE2 and in 2012, during QE3, by $45 billion a month until the year 2013. Thereafter this amount was gradually reduced (Patrabansh, Doerner, & Asin, 2014, pp.5-9).

According to Hancock and Passmore (2011), mortgage rates have fallen in the US with 100 basis points as the result of the Fed’s MBS purchases. This had caused house prices to rise after QE1. This price increase continued after more MBS were purchased during QE2 and QE3. Fear of a bubble in the housing market was the result, as this could be the beginning of a new crisis. Huston and Spencer (2018) have investigated using the Generalized Sup Augmented Dickey Fuller (GSDAF) technique whether there actually was a housing price bubble from 2009. However, they did not find any significant evidence for this. In addition to the prices on the housing market, equity and bond prices also rose explosive during the same period. Although clear signals were observed of overvalued stock prices, it was also

concluded for this market that there was no bubble. On the other hand, some evidence was found that between 2011 and 2013 the bond market formed a bubble. Here, however, the nuance has to be made that the last 30 years more often there was explosive growth in bond prices and this has not caused many problems in this market (Huston & spencer, 2018, p.374).

For the Eurozone, it has also been examined whether the ECB’s QE policy has led to bubbles in financial markets. The ECB announced in September 2014 to start an EAPP. This, with the same objective as the Fed and the BoE, to bring inflation back to the target level. Also in the Eurozone, unconventional monetary policy has led to explosive growth in

government bond prices. Lamoen, Mattheussens, and Dröes (2017) investigated whether these prices actually deviated from their fundamental values. With GSADF and Generalized Sup Phillips-Perron (GSPP) tests they have found empirical evidence for this deviation.

2.2.The transmission channels of QE

The Bank of England can exert a direct effect on the economy by adapting the policy rate. It can also do this by using a QE policy. With the QE policy the BoE can affect several factors simultaneously through different transmission channels. For example, the signaling channel can be used to adjust the long-term interest rate. The expectation of economic agents about short-term interest rates also plays a role here (Gern, Jannsen, Kooths, & Wolters, 2015, p.207).

The BoE can provide information to market participants about the current or future economic situation by announcing bond purchases. A large scale asset purchase (LSAP) program announcement is therefore a signal that the central bank will ease its monetary

(9)

9

policy. For the effectiveness of this signal it is important that the BoE credibly commits itself to this program. As a result, economic agents will adjust their expectations about the future short-term interest rate downwards (Christensen & Rudebusch, 2012, F.386). According to the expectations hypothesis of the term structure, described by Mishkin, Matthews, and Giuliodori (2013, pp.538-539), the long-term interest rate is the average of expected short-term interest rates. So when market participants expect low short-short-term interest rates, the long-term interest rate will also be low. The intention to keep this long-long-term interest rate low also indicates that the economic situation is bad and that the easing of monetary policy will continue in the near future (Bauer & Rudebusch, 2014, pp.242-243).

Besides the expectation about the short-term interest rate, there is another factor that can lower the long-term interest rate. This is the term premium. This premium is the

compensation for interest rate risk. Investors who hold assets with a longer duration run more interest rate risk and require a higher compensation (Rudebusch, Sack, & Swanson, 2007, p.241). By reducing the term premium, the long-term interest rate will also decrease. The central bank can influence this through the portfolio rebalancing channel (Gern et al., 2015, p.208). When the BoE buys assets from private sector institutions on a large scale, these are mainly government bonds with a longer duration. As a result, a considerable portion of these types of assets disappear from the market and the amount of money in circulation increases. This reduces the risk of a longer duration and as a result the offered term premium on these assets falls. The LSAP program of the central bank also increases the price of these assets, causing the yield to fall. This is also called the portfolio balance effect (Gagnon, Raskin, Remache, & Sack, (2011), p.42). This effect can affect both, the purchased gilts and other types of assets. In exchange for the sold bonds, the private sector receives short-term assets, namely bank deposits. Because these assets have a shorter duration than the bonds, the portfolio of these institutions is no longer in balance. To restore this, they use the increased liquidity to buy other long-term assets, such as corporate bonds and shares. As a result, there is also a downward pressure on the yield of these assets, which then increases the price (Joyce et al., 2012, pp.278-279).

As described, QE can cause interest rate cuts through the signaling and portfolio rebalancing channel. However, two other factors that can influence an LSAP program are the net export and aggregate output. This is done through the exchange rate channel. This channel has also become a component by which QE can influence the economy through major

(10)

10

The fall in interest as a result of an LSAP program ensures that domestic products become less attractive compared to foreign products. As a result, domestic products will decrease in value and depreciate the domestic currency (Mishkin et al. 2013, p.540). When the exchange rate is defined as domestic over foreign currency, it will rise as a result of the depreciated domestic currency (Pilbeam, 2013, p.5). Subsequently, it becomes more attractive for foreign investors to buy in the domestic country, which increases net exports and

aggregate output (Mishkin et al. 2013, p.540).

2.3.Hypothesis

Existing literature indicates that the QE policy of the BoE also influences financial markets that are not part of this. One of these markets is the stock market. When an announcement of QE is made, this may be a signal for market participants that the central bank keeps interest rates low in the future. This makes it plausible that these economic agents are adjusting their trade strategy. Due to these adjustments, QE is expected to have a direct effect on the stock market in the first instance.

In addition, QE also influences multiple economic drivers via the transmission channels of the BoE. These variables also have an effect on share prices. Therefore, the expectation is that the interaction between QE and the economic drivers creates a joint indirect effect on the stock market in the UK.

One of the objectives of QE was to increase inflation on various financial markets. Despite the fact that the stock market is not part of the QE policy, it is possible that stock prices are also positively influenced by these unconventional measures. Namely, as a result of QE, the yields on assets with a low risk decrease. This is an incentive for investors to replace these assets for securities with a higher return such as shares. Due to this portfolio balance effect, it is expected that the effect of QE on share prices will be positive.

3. Dataset

In order to investigate whether the QE policy of the Bank of England has had an effect on the stock market in the UK, a time series analysis is used. This regression method examines the effect of QE on UK share prices over time. The share prices in the UK are therefore the dependent variable in the regression equation.

The influence of QE is investigated on the basis of a number of economic drivers. These drivers are the independent variables and consist of the exchange rate, the yield on bonds in the UK, the interest rate in the US and also the stock market in the US. QE was also

(11)

11

added as a dummy variable to estimate the direct effect. The indirect effect is analyzed using four interaction terms. These interaction terms consist of the QE dummy combined with the chosen explanatory variables. The data for these variables consists of 2795 daily observations from Datastream. The timeframe that is being investigated runs from 31 December 2004 to 24 May 2018. Within this period QE was applied by the BoE from 11 March 2009 to 31 May 2012.

The Financial Times Stock Exchange Index (FTSE100) has been chosen as a

benchmark for share prices in the UK. This index contains the one hundred most capitalized companies in the UK, which are listed on the British stock exchange, the London Stock Exchange. Within the period of interest from 2004 to 2018, 2795 daily observations were made of this index. This is the size of the sample for the FTSE100 and for the independent variables.

The exchange rate between the UK and the US is one of the explanatory variables. Use is made of the spot exchange rate, which indicates how many British pounds have to be paid for 1 US dollar. A second explanatory variable is the yield on bonds with a 10 year duration. These long-term bonds were issued by the government of the UK and were mainly owned by the private sector during the crisis. Thirdly, the Federal funds rate was taken for the interest rate in the US. This is the interest at which American banks can lend cash reserves to each other and is determined by the Federal Reserve. The US stock market has also been added as an explanatory variable. For this the S&P500 index is used as an indicator. This index consists of the five hundred most capitalized companies in the US, listed on the New York Stock Exchange (NYSE) and the NASDAQ.

4. Methodology

To estimate the effects between stock prices in the UK and the explanatory variables, the following empirical model is used:

RstockUKt = b0 + b1ΔREXt + b2 ΔFFRt + b3 ΔGBYt + b4RstockUSt + b5QEt +

b6 ΔREX*QEt + b7 ΔFFR*QEt + b8 ΔGBY*QE + b9RstockUS*QEt + εt regression model (1)

b0 = The constant in the regression model

RstockUKt = The return on the FTSE100 in period t

(12)

12

ΔFFRt = The change in the Federal funds rate in period t

ΔGBYt = The change in the yield on bonds with a 10-year maturity in period t RstockUSt = The return on the S&P500 in period t

QEt = Dummy variable in period t, this has value 1 when QE is applied and zero otherwise ΔREX*QEt = Interaction variable between ΔREX and QE in period t

ΔFFR*QEt = Interaction variable between ΔFFR and QE in period t ΔGBY*QEt = Interaction variable between ΔGBY and QE in period t

RstockUS*QEt = Interaction variable between the S&P500 and QE in period t εt = The errorterm in period t

A number of things stand out in the regression model (1). Firstly, the returns on shares in the UK and the US are used instead of the prices. This is done to take stationarity into account. Stationary variables are variables whose probability distribution is constant over time. This is a requirement before use can be made of a time series analysis. When variables are not stationary but contain a unit root, this results in unreliable outcomes and predictions. Stock returns typically do not contain a unit root. Therefore stock prices are converted to returns using the following formula:

Stock return = 𝑃𝑡−𝑃𝑡−1

𝑃𝑡−1

∗

100% (2)

Pt is the price of the share in period t and Pt-1 the price of the share one period before t. What can be seen further in the regression model (1) is that the first difference is taken from the Sterling/US dollar exchange rate (ΔREXt), the Federal funds rate (ΔFFRt) and the bond yield (GBYt). This is indicated by the Δ sign. By taking the first difference one looks at the growth or change of a variable between the periods t and t-1. Usually the growth of a variable is more constant over time than the absolute values. Therefore the deltas of the REXt, FFRt and the GBYt are taken to ensure that these variables are also stationary. To statistically exclude the presence of a unit root in one of the variables, the Augmented Dickey-Fuller (ADF) test is used.

When the stationarity of the variables is checked with the ADF test, regression model (1) is applied. With this time series analysis the coefficients of the variables are estimated. This is done with both, OLS regression standard errors and heteroskedasticity-robust standard errors. The use of robust-standard errors is done with the aid of the Newey West estimator and

(13)

13

is to solve autocorrelation. Subsequently the Chow test is performed with the interaction variables from regression model (1). This determines the joint effect of these interaction terms on RstockUK. Additionally, the regression models (2), (3), (4) and (5) are used. Each of the explanatory variables is regressed on QE in these models. In this way it can be examined what the contributions of QE are to the effect that the explanatory variables have on RstockUK.

ΔREXt = a0 + a1QEt + ut regression model (2)

ΔFFRt = c0 + c1QEt + vt regression model (3) ΔGBYt = d0 + d1QEt + wt regression model (4) RstockUSt = f0 + f1QEt + rt regression model (5)

4.1.Expectation of the regression coefficients

Beforehand, it can be argued on the basis of existing literature whether certain independent variables will have a positive or negative effect on RstockUK.

The exchange rate is likely to have a positive effect on equity returns in the UK. In this study, this rate is defined as the amount of pounds needed to buy one US dollar. The UK is regarded as the domestic country and the US as the foreign country. With this definition, a depreciation of the pound leads to an increase in the exchange rate. Because the pound becomes cheaper, it becomes more attractive for foreign investors to invest in the UK. This increases the demand for British stocks and the returns increase. Therefore, it is expected that the coefficient for the exchange rate will be positive (Pilbeam, 2013, pp.148-176).

For the Federal funds rate and the yield on bonds with a 10-year maturity, the

expectation is that both variables will have a negative coefficient. An increase in the Federal funds rate makes it more expensive for investors to borrow money from banks in the US. As a result, these foreign investors will invest less in the US as well as in the UK. Due to this decrease in the number of investments, the demand for shares also decreases. This reduces the price and the return on these financial products. The negative coefficient of yield on bonds is plausible because a drop in the price causes the yield on bonds to increase. This will allow investors to include more gilts in their portfolio and fewer assets with a higher risk, such as shares.

The returns on shares in the US is an independent variable that is expected to have a positive impact on equity returns in the UK. This is due to the aforementioned interdependent

(14)

14

relationship between the US and other international equity markets (Eun & Shim, 1989, pp.251-254).

QE is added as a dummy variable in regression model (1). This has value 1 when the BoE applies the unconventional monetary policy. This was between 11 March 2009 and 31 December 2012. In other periods, this variable takes the value 0. The direct relationship between the dummy variable and the stock returns is expected to be positive. The reason for this is that a QE announcement by the BoE is a signal that interest rates will be kept low in the future. The resulting portfolio balance effect ensures that investor include more riskier assets, such as share, in their portfolio. The increased demand for these shares increases the values and therefore the return grows. The coefficient for QE is therefore estimated to be positive.

The interaction terms between the explanatory variables and the QE dummy are included in the regression equation to estimate the indirect effect of QE on equity returns in the UK. The values of these interaction terms are positive or negative depending on the impact that QE has. The interaction between the exchange rate and QE can have both, a positive and a negative effect on stock returns. Although the exchange rate is expected to have a positive effect on equity returns, QE is putting downward pressure on the exchange rate (Joyce et al., 2011, p.206). When the effect of QE dominates, the exchange rate will drop and the pound appreciate. This makes it more expensive for a foreign investor to invest in the UK. As the demand for shares in the UK declined, the price and return also fall. Therefore, the coefficient will be negative, when the QE effect dominates and positive when the exchange rate effect on stock returns predominates.

The Federal funds rate has a negative effect on stock returns. The expectation is that QE will reinforce this effect, so that the coefficient of the interaction variable ΔFFRQE will also be negative. By applying QE, this is a sign that the central bank wants to keep interest rates low to make investing more attractive. Therefore, the expectation is that QE has a negative effect on the ΔFFR.

The expected negative effect of ΔGBY on RstockUK will probably be strengthened by QE. Because more gilts are purchased during QE, the yield will drop further, increasing the demand for shares and therefore RstockUK.

The last interaction variable in regression model (1) is the RstockUSQE. The variable RstockUS is likely to have a positive effect on RstockUK due to the interdependent

relationship between international equity markets (Eun & Shim, 1989, pp.251-254).

According to Lima et al. (2016) has QE a positive effect on equity markets. Therefore, QE is expected to reinforce this effect of RstockUS on RstockUK.

(15)

15

5. Results

The results from the ADF test are shown in table 1. This table shows that the variables RstockUK and RstockUS are significant for two lags and with a significance level of 1%. This shows that the returns on shares in the UK and in the US are stationary variables. However, the interest rate in the US (FFR), the yield on 10-year bonds in the UK (GBY) and the exchange rate (REX), contain probably a unit root. Yet, these variables are stationary at the first and second lag when the delta is taken. Both, for lag(1) and lag(2), ΔREX, ΔFFR and ΔGBY are stationary at a significance level of 1%.

Table 1. Unit root test: Augmented Dickey-Fuller (ADF) test results Test statistic Z(t) Lag(1) Lag(2) RstockUK -31.776* -23,155* FFR 0.573 ΔFFR -33.067* -13.373* REX -1.589 ΔREX -25.093* -13.437* GBY -0.860 ΔGBY -26.276* -14.698* RstockUS -33.360* -22.165* *indicate significance at 1%

(16)

16

The time series analysis in regression model (1) was then applied over the period 31 December 2004 to 24 May 2018. The results of this regression are shown in table 2.

Table 2. Stock return in the UK and its determinants (2004-2018) (Dependent variable: Return on stocks in the UK)

Model 1 Model 1 with Newey West estimator Constant -.0040654 -.0040654 (.0198605) (.0198641) ΔFFR -1.43116*** -1.43116* (.2368106) (.7501765) ΔREX -2.607201 -2.607201 (5.07262) (7.532792) ΔGBY -.773169* -.773169 (.4139472) (.4951133) RstockUS .5532407*** .5532407*** (.017983) (.0401113) QE -.0093813 -.0093813 (.0406073) (.037941) ΔFFRQE -1.817872 -1.817872 (3.17027) (3.266899) ΔREXQE 6.218887 6.218887 (10.25473) (13.10992) ΔGBYQE .7613269 .7613269 (.7556887) (.9419179) RstockUSQE .0986394*** .0986394* (.0329016) (.0510681) R2 0.3568 Adj. R2 0.3547 F 171.65 70.94 Prob.>F 0.0000 0.0000

Notes: nine determinants of return on stocks in the UK, see regression model (1), are used in column 1. Column 2 is based on regression model (1) but heteroskedasticity-robust standard errors are used. Standard errors are reported in parentheses. *,** and *** indicate significance at 10%, 5% and 1%, respectively.

(17)

17

The applied regression model (1) has a R2 _{of 0.3568. This means that 35.68 percent of the}

variation in the dependent variable RstockUK is explained by the variation in the model. Table 2 also indicates that a number of effects on stock returns in the UK are significant. For example, the Federal funds rate has a significant effect on equity returns in the UK. When the Federal funds rate rises by 1 percent, the returns on shares in the UK fall by 1.43 percent. One of the reasons for this decline is that it becomes more expensive for US investors to borrow. As a result, the demand for shares declines both, in the domestic country and in the foreign country. However, with robust standard errors, the effect of ΔFFR on RstockUK is only significant at a level of 10%.

What table 2 further shows is that the variable RstockUS is also significant at a 1% level. This means that a 1 percent increase in equity returns in the US results in a 0.55 percent increase in the return on shares in the UK. As expected, this effect is positive. Because the S&P500 has an international influence on other equity markets, including the FTSE100, these follow the trend of the stock market in the US. Also when compensating for autocorrelation with robust standard errors, this effect appears to be significant at a significance level of 1%.

Furthermore, at a significance level of 10%, the variable ΔGBY is also significant. When the yield on bonds with a 10-year duration rises by 1 percent, the return on shares decreases by 0.77 percent. This negative effect can be explained with the preference of an investor for the highest possible return. When the yield on gilts rises, investors will include more bonds in their portfolios and less shares. However, this effect is not significant when robust standard errors are used.

The exchange rate between the UK and US has a negative impact on equity returns in the UK. An increase in the exchange rate of 1 percent means that RstockUK drops by 2.61 percent. This is not in line with the expectation that the exchange rate and the return on shares are positively correlated. However, this effect is not significant at the significance levels used. The QE dummy variable and the interaction variables ΔFFRQE, ΔREXQE and ΔGBYQE are also insignificant. This could mean that in the period from 2004 to 2018 QE had no direct or indirect influence on equity returns in the UK. However, another possibility is that QE does have an effect on stock returns, but it takes some time before this reveals itself. Market participants do not respond directly to the announcement of QE. Due to this lagged response, an announcement of QE at time t may only affect share returns at time t+1. Unlike other interaction variables, RstockUSQE is significant. The positive coefficient is consistent with the expectation that RstockUS has a positive effect on RstockUK and QE reinforces this effect.

(18)

18

The Chow test on the interaction variables shows that the interaction terms together have a significant effect on the stock returns in the UK. The null hypothesis that the

coefficients ΔFFRQE, ΔREXQE, ΔGBYQE and RstockUSQE are zero, could be rejected with a p-value of 0.0365<0.05. As an extension of the Chow test, it was investigated how QE contributes to the effect of the explanatory variables on RstockUK. This is done by regressing each of the explanatory variables on the dummy variable QE. The results of these regressions are shown in Table 3. QE was applied in the period from 11 March 2009 to 31 May 2012. During this period, the dummy variable was equal to 1 and QE would therefore cause a drop in the exchange rate of 0.00017 percent. Although the decrease is consistent with the existing literature that QE exerts a downward pressure on the exchange rate, the estimated coefficient is insignificant. This applies to each of the significance levels used. The other results in table 3 are also not significant, which means that the effect of QE on ΔFFR, ΔGBY and RstockUS is not unequal to zero.

Table 3. Effect of QE on exchange rate, Federal funds rate, yield 10 yr government bond and return on shares in US (2004-2018)

(Dependent variables: column (A): ΔREX, column (B): ΔFFR, column (C): ΔGBY, column (D): RstockUS)

(A) (B) (C) (D)

Constant .0000822 -.0025931 -.0005097 .0153249

(.0000871) (.0015909) (.0011114) (.0226231)

QE -.0001732 .0016584 .0014696 .0606964

(.0001774) (.0032398) (.0022632) (.0460914)

Notes: Standard errors are reported in parentheses. *,** and *** indicate significance at 10%, 5% and 1%, respectively.

6. Conclusion

Following the financial crisis of 2007-2008, the Bank of England has applied an

unconventional monetary policy called Quantitative Easing. This policy aimed to restore stability on financial markets and stimulate economic growth in the UK. QE mainly

influences these markets through transmission channels. However, this indirectly also exerts an effect on markets that are not part of QE, such as the stock market. This paper examined the impact of QE on the stock market in the UK. This was done with the aid of a time series analysis. Considered is the direct impact of a QE announcement on this market and the indirect effect of QE through a number of economic drivers. These drivers are the Federal funds rate, the yield on bonds with a 10-year duration, the exchange rate and the returns on

(19)

19

shares in the US. For this purpose the period from 31 December 2004 to 24 May 2018 was analyzed, in which the BoE applied a QE policy from 11 March 2009 to 31 May 2012.

The results show that QE did not have a direct significant effect on shares during this period. This is not in line with the expectation that QE has a positive direct effect on the stock returns. An explanation for this is that the effect of the QE announcement on the stock market at the same time was examined in the time series analysis. However, it may take some time for investors to respond to this announcement, so that a direct effect of QE is not immediately reflected in the price of a share at the same time. To be able to solve this, we have to look for lagged effects. This is an interesting possibility for follow-up research.

Furthermore, the results show that the interaction variable between QE and the returns on shares in the US has a significant effect on equity returns in the UK. The effects of the interaction variables between QE and the other economic drivers are not significant. Finally, the Chow test shows that the used interaction terms jointly have had a significant effect on the stock market in the UK. This is in line with the expectation that the interaction between QE and the economic drivers will have a joint effect on equity returns in the UK. However, the contribution of QE to the indirect effects on equity returns is not significant. From this it can be concluded that QE did not have an indirect effect through the chosen economic drivers. This conclusion does not correspond with the hypothesis, based on existing literature, that QE has a positive effect on the stock market. This contradiction makes it interesting to continue analyzing the effects of QE. One suggestion for this is by examining in the future whether there are variables that are more sensitive to the effect of QE. Due to these sensitive economic drivers QE may have a significant effect on shares.

(20)

20 7. References

Bauer, M. D., & Rudebusch, G.D. (2014). The Signaling Channel for Federal Reserve Bond Purchases. International Journal of Central Banking, 10(3), 233-289.

Benford, J., Berry, S., Nikolov, K., Robson, M., & Young, C. (2009). Quantitative easing. Quaretrly Bulletin // Bank of England, 49(2), 90-99.

Buch, C.M., Eickmeier, S., & Prieto, E. (2014). In search for yield? Survey-based evidence on bank risk taking. Journal of Economic Dynamics and Control 43(C), 12-30.

Claeys, G., Leandro, Á., & Mandra, A. (2015). European Central Bank Quantitative Easing: The detailed manual. 02.

Christensen, J.H.E., & Rudebusch, G.D. (2012). The Response of Interest Rates to US and UK Quantitative Easing*. Economic Journal, 122(564), F385-F414.

Eun, C., & Shim, S. (1989). International Transmission of Stock Market Movements. Journal of Financial and Quantitative Analysis, 24(2), 241-256.

Fawley, B., & Neely, Christopher J. (2013). Four stories of quantitative easing. Review, 95(1), 51-88.

Fratzscher, M., Lo Duca, M., & Straub, R. (2018). On the International Spillovers of US Quantitative Easing. Economic Journal, 128(608), 330-337.

Gagnon, J., Raskin, M., Remache, J., & Sack, B. (2011). The Financial Market Effects of the Federal Reserve’s Large-Scale Asset Puchases*. Economic policy review, 17(1), 41-59 Gern, K.-J., Jannsen, N., Kooths, S., & Wolters, M. (2015). Quantitative Easing in the Euro

Area: transmission Channels and Risks. Intereconomics, 50(4), 206-212.

Hancock, D., & Passmore, W. (2011). Did the Federal Reserve’s MBS purchase program lower mortgage rates? Journal of Monetary Economics, 58(5), 498-514.

Huston, J. & Spencer, R. (2017), “Quantitative Easing and asset bubbles,” Applied Economics Letters, http://dx.doi.org/10.1080/13504851.2017.1324604.

Joyce , M. A., Lasaosa, A., Stevens, I., & Tong, M. (2011, September). The financial Impact of Quantitative Easing in the United Kingdom. Bank of England Quarterly Bulletin, 50(3),113.

Joyce, M., Miles, D., Scott, A., & Vayanos, D. (2012, November). Quantitative Easing and Unconventional Monetary Policy – an Introduction*. Economic Journal, 122(564) F272-F279

(21)

21

Joyce, M., Spaltro, M. (2014) Quantitative easing and bank lending: a panel data approach, (Bank of England Working Paper No. 504). Retrieved from Bank of England website:

https://www.bankofengland.co.uk/working-paper/2014

Joyce, M., & Tong, M. (2012). QE and the Gilt Market: A Disaggregated Analysis*. Economic Journal, 122(564), F348-F384.

Joyce, M., Tong, M., & Woods, R. (2011). The United Kingdom’s quantitative easing policy: Design, operation and impact.(Research and Analysis)(Report). Bank of England Quarterly Bulletin, 51(3), 200.

Kapetanios, G., Mumtaz, H., Stevens, I., & Theodoridis, K. (2012). Assessing the Economy-wide Effects of Quantitative Easing*.(Report). Economic Journal, 122(564), F316-F347.

Kurihara, Y. (2006). The relationship between exchange rate and stock prices during the quantitative easing policy in Japan. International Journal of Business,11(4), 375-386. Lima, L., Vasconcelos, C.F., Simão, J., & de Mendonça, H.F. The quantitative easing effect

on the stock market of the USA, the UK and Japan. (2016). Jouran of Economic Studies, 43(6), 1006-1021.

Martinez-Miera, D., & Repullo, R. (2017). Search for Yield. Econometrica, 85(2), 351-378. Mishkin, F.S. (2001). The Transmission Mechanism and The Role of Asset Prices in

Monetary Policy (NBER Working Paper No. 8617). Retrieved from National Bureau of Economic Research website: http://www.nber.org/papers/w8617

Mishkin, F.S., Matthews, K., & Giuliodori, M. (2013). The Economics of Money, Banking & Financial Markets (pp. 335-547). Edinburgh Gate: Pearson Education Limited. Patrabansh, S., Doerner, M., & Asin, S. (2014). “The Effect of Monetary Policy on Mortgage

Rates,” FHFA Working paper 14-2, 2014.

Pilbeam, K. (2013). International Finance (pp. 5-176). Basingstoke: Palgrave MacMillan. Rudebusch, G.D., Sack, B.P., & Swanson, E.T. (2007). Macroeconomic Implications of

Changes in the Term Premium, 89(4), 241-269.

Van Lamoen, R., Mattheussens, S., Dröes, M.I., Faculteit Economie en Bedrijfskunde, & Finance. (2017). Quantitative Easing and Exuberance in Government Bond Markets: Evidence from the ECB’s Expanded Asset Purchase Program. 2017.

(22)

22 8. Appendix

Output ADF test

MacKinnon approximate p-value for Z(t) = 0.0000

Z(t) -31.776 -3.430 -2.860 -2.570 Statistic Value Value Value Test 1% Critical 5% Critical 10% Critical Interpolated Dickey-Fuller Augmented Dickey-Fuller test for unit root Number of obs = 2096

(23)

23

(24)

24 Output regression model (1)

Z(t) 0.573 -3.430 -2.860 -2.570 Statistic Value Value Value Test 1% Critical 5% Critical 10% Critical Interpolated Dickey-Fuller Augmented Dickey-Fuller test for unit root Number of obs = 2096

_cons -.0040654 .0198605 -0.20 0.838 -.0430083 .0348774 RstockUSQE .0986394 .0329016 3.00 0.003 .0341255 .1631534 dGBYQE .7613269 .7556887 1.01 0.314 -.7204397 2.243094 dFFRQE -1.817872 3.17027 -0.57 0.566 -8.034187 4.398444 dREXQE 6.218887 10.25473 0.61 0.544 -13.88875 26.32653 QE -.0093813 .0406073 -0.23 0.817 -.0890047 .0702422 RstockUS .5532407 .017983 30.76 0.000 .5179793 .5885022 dGBY -.773169 .4139472 -1.87 0.062 -1.584843 .0385053 dFFR -1.43116 .2368106 -6.04 0.000 -1.895502 -.966818 dREX -2.607201 5.07262 -0.51 0.607 -12.55368 7.339274 RstockUK Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 3617.1476 2,794 1.2946126 Root MSE = .914 Adj R-squared = 0.3547 Residual 2326.59519 2,785 .835402221 R-squared = 0.3568 Model 1290.55241 9 143.394712 Prob > F = 0.0000 F(9, 2785) = 171.65 Source SS df MS Number of obs = 2,795

(25)

25

Output regression model (1) with heteroscedasticity-robust standard errors

Output regression model (2)-(5)

_cons -.0040654 .0198641 -0.20 0.838 -.0430153 .0348844 RstockUSQE .0986394 .0510681 1.93 0.054 -.0014957 .1987746 dGBYQE .7613269 .9419179 0.81 0.419 -1.085601 2.608255 dFFRQE -1.817872 3.266899 -0.56 0.578 -8.223659 4.587916 dREXQE 6.218887 13.10992 0.47 0.635 -19.48725 31.92503 QE -.0093813 .037941 -0.25 0.805 -.0837767 .0650141 RstockUS .5532407 .0401113 13.79 0.000 .4745899 .6318915 dGBY -.773169 .4951133 -1.56 0.118 -1.743995 .1976572 dFFR -1.43116 .7501765 -1.91 0.057 -2.902118 .039798 dREX -2.607201 7.532792 -0.35 0.729 -17.37762 12.16322 RstockUK Coef. Std. Err. t P>|t| [95% Conf. Interval] Newey-West

Prob > F = 0.0000 maximum lag: 0 F( 9, 2785) = 70.94 Regression with Newey-West standard errors Number of obs = 2,795

_cons .0000822 .0000871 0.94 0.346 -.0000886 .000253 QE -.0001732 .0001774 -0.98 0.329 -.000521 .0001746 dREX Coef. Std. Err. t P>|t| [95% Conf. Interval] Total .04496466 2,794 .000016093 Root MSE = .00401 Adj R-squared = -0.0000 Residual .044949321 2,793 .000016094 R-squared = 0.0003 Model .000015339 1 .000015339 Prob > F = 0.3290 F(1, 2793) = 0.95 Source SS df MS Number of obs = 2,795

_cons -.0025931 .0015909 -1.63 0.103 -.0057127 .0005264 QE .0016584 .0032398 0.51 0.609 -.0046942 .008011 dFFR Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 14.9954555 2,794 .005367021 Root MSE = .07327 Adj R-squared = -0.0003 Residual 14.9940488 2,793 .005368439 R-squared = 0.0001 Model .001406684 1 .001406684 Prob > F = 0.6088 F(1, 2793) = 0.26 Source SS df MS Number of obs = 2,795

(26)

26 Output Chow test

_cons -.0005097 .0011114 -0.46 0.647 -.0026889 .0016695 QE .0014696 .0022632 0.65 0.516 -.0029681 .0059073 dGBY Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 7.31810866 2,794 .002619223 Root MSE = .05118 Adj R-squared = -0.0002 Residual 7.31700402 2,793 .002619765 R-squared = 0.0002 Model .001104639 1 .001104639 Prob > F = 0.5162 F(1, 2793) = 0.42 Source SS df MS Number of obs = 2,795

_cons .0153249 .0226231 0.68 0.498 -.029031 .0596808 QE .0606964 .0460914 1.32 0.188 -.0296725 .1510653 RstockUS Coef. Std. Err. t P>|t| [95% Conf. Interval] Total 4745.22351 3,494 1.35810633 Root MSE = 1.1653 Adj R-squared = 0.0002 Residual 4742.86886 3,493 1.35782103 R-squared = 0.0005 Model 2.35465942 1 2.35465942 Prob > F = 0.1880 F(1, 3493) = 1.73 Source SS df MS Number of obs = 3,495

Prob > F = 0.0365 F( 4, 2785) = 2.56 ( 4) RstockUSQE = 0 ( 3) dGBYQE = 0 ( 2) dFFRQE = 0 ( 1) dREXQE = 0