1
‘The effects of unexpected unconventional monetary
policies of the ECB on the level of stock prices in the
Netherlands’
By:
Joris Brouwer (s2368382)
January ‘17
Abstract
This paper investigates the relationship between unexpected unconventional monetary policies executed by the European Central Bank and the level of stock prices in the Netherlands. Furthermore, also the main drivers behind this relationship will be analysed. It is found in this paper that unexpected unconventional loosening (tightening) monetary policies have a significant positive (negative) impact on the level of stock prices in the Netherlands. Moreover, by applying a variance decomposition along the lines inspired by Campbell and Ammer (1991) it is found that the largest part of the reaction of stock prices to unexpected unconventional monetary policies is the result of changes in future expected excess returns of stocks.
Keywords: Financial crisis, stock pricing, unconventional monetary policies. JEL code(s): E58, E52, G12, G13.
2 1. Introduction
This research paper will examine the relationship between unexpected unconventional monetary policies executed by the European Central Bank and the level of stock prices, with specific attention to the stock market in the Netherlands. This relationship has been examined by previous literature in the case of the United States by Bernanke and Kuttner (2005) and also for the whole Euro area by Haitsma et al. (2015). In addition to what has been examined before, this research paper will focus mainly on the effects of unexpected and unconventional monetary policies conducted by the European Central Bank. These unconventional monetary policies have been employed by the European Central Bank as a result of the fact that the standard monetary policies to fight the financial crisis had started to lose their effectiveness. Especially after the financial crisis, it has been questioned by many scholars what kind of effects these unexpected unconventional monetary policies might have on financial markets and what kind of consequences these effects should have on conducting unconventional monetary policies in the first place. However, as is also stated by Laeven and Tong (2010), a full understanding of the effects of monetary policy on financial markets is required to be able to make normative statements about the design of monetary policies that take into account the response of financial markets. Therefore, this research paper will focus on the effects of unexpected unconventional monetary policies on the level of stock prices. More specifically, this research will investigate whether the relationship between these unexpected unconventional monetary policies and the level of stock prices is positive as is to be expected based on the findings of other scholars.
3 increase the understanding about the intermediate effects that unexpected unconventional monetary policies have on stock markets.
Next to this, this paper also endeavours to investigate what the main drivers behind the relationship between unexpected unconventional monetary policies and the level of stock prices in the Netherlands are. Bernanke and Kuttner (2005) have examined these drivers for the United States, but for European countries and especially the Netherlands these factors have not been examined up to this point. The main drivers that have been found by Bernanke and Kuttner (2005), that will also be examined in this research paper, are: expected future dividends, future expected interest rates and expected excess returns. Hereby this paper will try to create more insights about these main drivers behind the relationship between unexpected unconventional monetary policies and the levels of stock prices in the Netherlands.
Finally the structure of this research paper will be as follows. First of all, the economic theory behind the relationship between unexpected unconventional monetary policies and the level of stock prices will be explained more extensively. Secondly, an overview of the research field and previous literature will be presented and also some hypotheses regarding the expected outcomes of the research will be presented. These hypotheses will be based on economic theory and previous literature that will be discussed in this paper. Thirdly, the methodology that will be used in this research will be reflected upon. Fourthly, the data that has been used for this research and the adjustments that have been made to this dataset will be discussed. Fifthly, the results of the research performed in this paper will be presented and commented upon. Finally, the last section of this research paper concludes and a discussion of the outcomes that have been obtained will be presented to provide directions for future research.
2. Economic theory and hypotheses
2.1 Economic theory
4 financial markets, including stock markets, play an important role within the transmission mechanism of monetary policies. This is also argued by Ioannidis and Kontonikas (2006) and Bernanke and Kuttner (2005). One of the reasons for this connection is that new information is incorporated very quickly by financial markets to keep prices up-to-date which make financial markets sensitive to unexpected changes in monetary policies.
As is also argued by Ioannidis and Kontonikas (2006), the easiest way to explain why monetary policies may have significant effects on stock prices is by making use of the discounted cash flow model, also called discounted dividend model. This model is frequently used to assess the value of a stock. According to the discounted dividend model, the price of a stock that is traded on a stock exchange can be determined by calculating the value of the expected future dividends. The formula for the dividend discount model is as follows:
𝑃𝑡= ∑ (𝑟−𝑔𝐷0 ) 𝑡 + 𝑃𝑁 (1+𝑟)𝑁 𝑁 𝑡=1 (1)
Where N is the number of periods the investor holds on to the stock, r is the discount rate used by market participants to discount future dividends, D stands for the expected future dividends and P is the stock price at a given moment in time. The growth rate of dividends on stocks is incorporated into the model with the symbol g.
From equation (1) it can be seen that monetary policies can affect stock prices through the real interest rate and expected future dividends. From economic theory it is known that in the long run monetary policies should have no effect on real variables. In the long run monetary policies should only affect nominal variables and should have no impact through the real interest rate on the levels of stock prices. However, in the short run this effect might be different as people might not respond rationally to a change in monetary policy or due to the fact that the prices are not fully flexible in the short run. As is stated by Patelis (1997): ‘stocks are claims on future economic output, so that if monetary policy has effects on the real economy then stock markets should also be influenced by monetary conditions’. This statement from Patelis (1997) sums up why it is of substantial importance to investigate this relationship. In this section this channel will be used to explain why this relationship is of substantial importance and deserves considerable attention. Furthermore, another channel through which monetary policies could affect the levels of stock prices will be discussed.
5 rates imply that there will be more investments and general activity in the economy. Consequently there will be more growth of the economy which will lead to a positive reaction of stock prices. This positive reaction of stock prices is based on the fact that due to the expansion of the economy, expected cash flows will increase which will in turn increase the implicit value of the stock according to the dividend discount model.
Furthermore, the future real interest rates are used as the discount rate for future dividends in the dividend discount model. As a result of this, future dividends increase as well. Therefore, a decrease in this rate will increase the future dividends which in turn increase the stock price as can be seen from equation (1). This channel is extensively discussed by Ioannidis and Kontonikas (2006).
6 on the relationship between unexpected unconventional monetary policies and the level of stock prices within the Netherlands.
To conclude, there are two main channels through which unexpected conventional and unconventional monetary policies can influence the level of stock prices. These include an impact through changes in the real interest rate as a result of unexpected unconventional monetary policies. Interrelated with the real interest rate is that loosening (tightening) monetary policies will have a positive (negative) impact on the levels of stock prices through increased (decreased) future dividends as a result of the lower (higher) real interest rate which is used as a discount rate for future dividends. Finally, there is an ambiguous impact on the levels of stock prices through changes of expected excess returns.
3. Literature review 3.1 General
Monetary policies used by the European Central Bank are aimed to control inflation in the Euro area. The other objectives set by the European Central Bank are subservient to this main goal. This has been laid down in the Treaty on the Functioning of the European Union. The quantitative definition of this objective which has been stated by the European Central Bank’s Governing Council reads as follows: ‘price stability is defined as a year-on-year increase in the Harmonised Index of Consumer Prices (HICP) for the euro area close to, but below 2%’. Therefore, the monetary policies executed by the European Central Bank are mainly focussed on reaching this objective and controlling inflation. To reach this objective the European Central Bank has several instruments at its disposal to influence short-term interest rates and steer inflation. These instruments are open market operations, standing facilities and minimum reserve requirements. Especially the first two of these instruments are used to influence the short-term interest rate so as to reach the inflation target.
7 capital. These will also be the main points of interest of this research paper. Although this topic is discussed regularly within the existing literature there does not seem to be an analysis as the one proposed in this paper. So far there has not been research that combines the effects of monetary policies used by the European Central Bank in the context of the financial crisis which also explains the main factors driving this reaction of the stock markets. In the paper of Haitsma et al. (2015) the impact of unconventional monetary policies of the European Central Bank during the crisis is examined. In their research Haitsma et al. (2015) use the EURO STOXX 50 index to explain surprise effects of unconventional monetary policies. To gain insights about the effects of these unexpected unconventional monetary policies used by the European Central Bank as they are discussed by Pattipeilohy et al. (2013), this paper endeavours to investigate the influence that these monetary policies have had on the level of stock prices in the Netherlands. To be able to create a clear picture of the existing literature, first a general overview of the monetary policies employed by the European Central Bank to fight the financial crisis will be presented. After this, the existing literature on the relationship between monetary policy changes and levels of stock prices will be discussed. Finally, the existing literature on the main drivers behind this relationship will be discussed.
3.2 European Central Bank monetary policies during the crisis
8 Union and especially the Netherlands. For a clear understanding of this relationship, existing literature on the relationship between monetary policies and levels of stock prices will be discussed in the next part of this literature survey.
3.3 Effects of monetary policies on stock markets
9 Furthermore, as is discussed by Cassola and Morana (2002), the effect of monetary policy on stock markets is strong yet temporary, whereas the effect of monetary policy on inflation is of a more structural nature. As a result of this, it seems important to study the relationship between monetary policies and financial markets as well as the main drivers behind this relationship. As argued by Cassola and Morana (2002), this could prevent monetary policy authorities from implementing monetary policies aimed at financial market responses that are incompatible with its price stability objective.
To conclude, the relationship between monetary policy and its effect on financial markets has been researched frequently, however not for the Netherlands. To contribute to the literature this paper will try to examine this relationship as thoroughly as possible for the stock market in the Netherlands. As discussed before, previous literature has come up with mixed results on the relationship between monetary policies and the level of stock prices. Therefore, this paper will also make a contribution to the literature by investigating this relationship in more depth to create more insights into these mixed results. Next to this, the main drivers behind this relationship for the Netherlands will be examined. Therefore in the next section previous literature on these drivers will be discussed.
3.4 Drivers of monetary policies’ effects on stock prices
10 as Bernanke and Kuttner (2005) for the response of financial markets to general economic variables contributes to the strength of the argument put up by Bernanke and Kuttner (2005).
A more recent contribution to the literature has been made by Gospodinov and Jamali (2015). In their research Gospodinov and Jamali (2015) reach similar conclusions as Bernanke and Kuttner (2005); however Gospodinov and Jamali (2015) find that long-term changes in the levels of stock markets as a result of monetary policy tend to be dominated by the effect of monetary policy on expected future dividends.
3.5 Hypotheses
Based on the previous literature and economic theory which have been discussed so far, a couple of hypotheses have been formulated. These are based on previous results found by other scholars and are extrapolated to the Netherlands. On the basis of these hypotheses the results of this research will be investigated and commented upon. Based on existing literature and economic theory the following hypotheses have been formulated:
- H1: Unexpected unconventional monetary policies of a tightening nature used by the European Central Bank have a negative effect on level of stock prices in the Netherlands.
- H2: Unexpected unconventional monetary policies of the European Central Bank have a different effect on the level of stock prices in the Netherlands than unexpected conventional monetary policies.
- H3: Unexpected unconventional monetary policies employed by the European Central Bank have had different effects on the level of stock prices during the financial crisis when compared to times of economic stability.
- H4: The effects of monetary policies in general, executed by the European Central Bank, on the level of stock prices in the Netherlands have been different during the financial crisis when compared to times of economic stability.
- H5: Expected future dividends are the main driver behind the relationship between unexpected unconventional monetary policies and the level of stock prices.
4. Methodology
11 method similar in nature to the one used by Bernanke and Kuttner (2005) and Haitsma et al. (2015). By using this method it is possible to investigate the differences between effects of unexpected conventional and unexpected unconventional monetary policies. Also, a comparison can be made between the effects of unexpected unconventional monetary policies during the financial crisis and during times of economic stability. This comparison will also be made for unexpected monetary policies executed by the European Central Bank in general.
4.1 Impact of monetary policies on the level of stock prices
First of all, to measure the surprises of conventional monetary policies a strategy close to the one proposed by Bernanke and Kuttner (2005) is used. The theory behind the method used by Bernanke and Kuttner (2005) is that futures prices include the market’s expectations of future monetary policy rates. So, if futures prices change after monetary policy announcements, market participants had not foreseen this. As the effects of monetary policies by the European Central Bank are investigated, this paper uses three-month Euribor futures rates to proxy unexpected conventional monetary policies like Haitsma et al. (2015). As a result the indicator of unexpected conventional monetary policies can be stated as:
∆𝐼𝑡𝑐,𝑢= (𝑓𝑠,𝑡− 𝑓𝑠,𝑡−1) (2)
Where ∆𝐼𝑡𝑐,𝑢 represents the conventional monetary policy surprise at day t. One remark has to be made about this formula. In their paper Bernanke and Kuttner (2005) propose a correction of this variable to correct for the number of days in each month. The method employed by Bernanke and Kuttner (2005) is to multiply the differences in Euribor futures rates by 𝐷−𝑑𝐷 . However, in the dataset that is used in this paper the Euribor futures rates are observed at the last day of each month which makes multiplying by this factor impossible as the denominator will always take a value of 0. Therefore, this addition will not be employed in this research paper.
Besides the dates of the announcements of the European Central Bank to describe conventional monetary policy decisions, the expected part of a conventional monetary policy change can also be explained by the following equation:
12 As this paper endeavours to investigate the effects of unexpected unconventional monetary policies, a proxy for the unexpected part of these unconventional monetary policies is also required. To be able to account for this, the method proposed by Rogers et al. (2014) will be used. In their paper Rogers et al. (2014) use the spread between German and Italian 10-year bond yields. If the spread changes as a result of unconventional monetary policy this means that monetary policy changed in a different way than what had been anticipated by the market. To account for this, the formula used by Rogers et al. (2014) is:
∆𝐼𝑡𝑢,𝑢 = (𝑦𝑠,𝑡𝐼 − 𝑦
𝑠,𝑡𝐷) − (𝑦𝑠,𝑡−1𝐼 − 𝑦𝑠,𝑡−1𝐷 ) (4)
Where ∆𝐼𝑡𝑢,𝑢 represents the unexpected changes in unconventional monetary policy conducted by the European Central Bank.
In this paper the dependent variable are the changes in the levels of stock prices in the Netherlands. These will be calculated using a log-transformation:
𝑅𝑡𝑁𝐿 = log (𝑝𝑡
𝑁𝐿
𝑝𝑡−1𝑁𝐿) (5)
Where 𝑝𝑡𝑁𝐿 represents the closing price of the Dutch stock index AEX.
Finally, as we have all the variables needed for the equation to measure the impact of monetary policies on level of the stock prices in the Netherlands we can construct the main equation of this research paper:
𝑅𝑡𝑁𝐿 = 𝛼 + 𝛽
1∆𝐼𝑡𝑢,𝑢+ 𝛽2 ∆𝐼𝑡𝑐,𝑢+ 𝛽3 𝐶𝑟𝑖𝑠𝑖𝑠𝑡 ∆𝐼𝑡𝑢,𝑢+ 𝛽4 𝐶𝑟𝑖𝑠𝑖𝑠𝑡∆𝐼𝑡𝑐,𝑢+ 𝛽5𝐶𝑟𝑖𝑠𝑖𝑠𝑡+ 𝜀𝑡(6)
13
4.2 Drivers behind the relationship
To be able to investigate what the main drivers behind the relationship between unexpected unconventional monetary policies and the level of stock prices in the Netherlands are, a method similar in nature to the method of Bernanke and Kuttner (2005) will be used. The method used by Bernanke and Kuttner (2005) is deduced from the one employed by Campbell (1991) and Campbell and Ammer (1993). In this approach log-linearization methods are used to split excess equity returns, in this case excess stock market returns in the Netherlands, into different components. These components are represented by news about real interest rates, future dividends and future excess returns. However, in this approach the different component’s parameters will be related to the unexpected unconventional monetary policy strategy used by the European Central Bank. By using this method the contributions of the real interest rate, future dividends and future excess returns can be investigated. The equation which makes use of the decomposition is:
𝑅𝑡+1𝑁𝐿 = 𝛼 + 𝛽1 𝑒𝑡+1𝑑 + 𝛽2 𝑒𝑡+1𝑟 + 𝛽3 Ṙ𝑡+1𝑁𝐿 + 𝜀𝑡+1 (7)
Where 𝑒𝑡+1𝑑 represents expected future discounted dividends, 𝑒
𝑡+1𝑟 represents the expected real
discount rate and Ṙ𝑡+1𝑁𝐿 represents expected excess future returns. By using this equation the main drivers behind the relationship between monetary policies and the levels of stock prices in the Netherlands can be identified.
Equation (7) can be deduced by making use of log-linearization methods which are extensively discussed in the paper of Bernanke and Kuttner (2005). These methods are based on the approach that has been taken by Campbell and Ammer (1993) and are discussed in detail in section 3 of the paper by Bernanke and Kuttner (2005).
14 future dividends in explaining the relationship between unexpected unconventional monetary policies and the level of stock prices. This trade-off is discussed by both Bernanke and Kuttner and Campbell and Ammer (1993) in their papers, however they give no further explanation upon the reasons behind this.
To prohibit this overweighting of future dividends in explaining the relationship and as a result of the fact that the data on expected future dividends on the Dutch AEX index were readily available and do not have to be forecasted, future dividends will be included directly as a variable in the forecasting VAR in this paper. These forecasts give the expected returns on investments in the AEX based on historical returns from the third quarter of 2016 onwards till 2045.
4.2.1 Forecasting VAR
In a similar fashion to Bernanke and Kuttner (2005) and Campbell and Ammer (1991) a VAR method is used to capture the dynamic relationships between the excess equity returns, the real interest rate, future dividends and future expected returns. The real interest rate has been calculated as follows:
𝐼𝑡𝑟 = 𝑅
𝑡𝑓− (𝐶𝑃𝐼𝑡𝑁𝐿− 𝐶𝑃𝐼𝑡−1𝑁𝐿) (8)
In which 𝑅𝑡𝑓is the one month bill yield in the Netherlands and CPI represents the Consumer Price Index for the Netherlands.
15
4.2.2 Decomposition of the VAR
To be able to interpret the results of the forecasting VAR that has been set up, the VAR model needs to be decomposed into the contributions of the variances of the different variables which have been included. Therefore a variance decomposition will be applied to the VAR model. According to Campbell and Ammer (1991), a problem that may occur is that some of the variables that are included in the model could be correlated with one another. This could lead to ambiguity in the notion of the variance decomposition according to Campbell and Ammer (1991). The equation which is proposed by Campbell and Ammer (1991), which is also used by Bernanke and Kuttner (2005), is given by equation (9):
𝑉𝑎𝑟(𝑅𝑡+1𝑁𝐿 ) =
𝑉𝑎𝑟(𝑒𝑡+1𝑑 ) + 𝑉𝑎𝑟(𝑒
𝑡+1𝑟 ) + 𝑉𝑎𝑟(𝑅̇𝑡+1𝑁𝐿) − 2𝐶𝑜𝑣(𝑒𝑡+1𝑑 , 𝑒𝑡+1𝑟 ) − 2𝐶𝑜𝑣(𝑒𝑡+1𝑑 , 𝑅̇𝑡+1𝑁𝐿 ) +
2𝐶𝑜𝑣(𝑅̇𝑡+1𝑁𝐿 , 𝑒𝑡+1𝑟 ) (9)
One way to interpret the variance decomposition is to report all the values of the coefficients in equation (9). This has been done by Bernanke and Kuttner (2005). As can be seen from the covariances included in equation (9), correlation could constitute a problem when using this method. Campbell and Ammer (1991) argue that this does not have to be a problem as the fact that some of the drivers behind excess equity returns are highly correlated could be an important stylized fact on its own. However, the purpose of this paper is to find the main drivers behind the relationship between unexpected unconventional monetary policies executed by the European Central Bank. Therefore, the main aim is to make sure that there are no problems with correlation between the components to prevent ambiguity in the interpretation of the variance decomposition. This will give the clearest insights into which components have the largest contributions.
16 paper. In this paper the Cholesky method will be used solely to perform the decomposition of the forecasting VAR. This is done to circumvent the problem of correlation between the variables that are included in the VAR and because it better fits the dataset. Therefore this paper differs from the papers by Bernanke and Kuttner (2005) and Campbell and Ammer (1991).
The Cholesky method will be applied using 10 periods of forecasts which indicate that the decomposition of the variances will be calculated for the next 10 periods. The order for the Cholesky method is also of importance and will be as follows: real interest rate, future dividends and excess equity returns. This order represents the way in which the drivers directly have an effect on the relationship examined and is based on economic theory. The channel with the highest expected direct effect, the real interest rate, comes first and is followed by future dividends and excess equity returns. The real interest rate is also the factor which can be influenced by monetary policies of the European Central Bank as is explained in the section on economic theory. Furthermore, the real interest rate influences future dividends and expected excess equity returns. Finally, the expected dividends influence the expected excess equity returns as these will increase when the expected future dividends are higher. Therefore, this is the appropriate order of variables for the Cholesky decomposition.
The results of this Cholesky variance decomposition will create more insights on which of these drivers are the most important in explaining the relationship between unexpected unconventional monetary policies and the level of stock prices in the Netherlands. The results of the Cholesky variance decomposition are presented in table 2 of section 6.
5. Data
5.1 General data descriptions
In this section the data that has been used to conduct the research in this paper will be discussed. The descriptive statistics of the data and the variables that have been used for equation (6) are summarized in table 3. Furthermore the descriptive statistics that have been used for the Cholesky decomposition are summarized in table 4. Both tables are included as an appendix to this paper. From the Jargue-Bera tests it is visible that the data is non-normally distributed, however in this paper only t-tests and variance analyses are performed which both can be labelled as robust against non-normality. Therefore this has no further implications on the regressions that are performed in this paper.
log-17 transformed changes in the daily levels of the AEX) on this index will constitute the dependent variable in the equations examined in this paper. The AEX index consists of the 25 most actively traded securities on the Dutch stock exchange. Furthermore, the AEX index consists only of Dutch companies.
The Euribor three-month futures rates will be used to proxy for unexpected conventional monetary policies executed by the European Central Bank. To proxy for the effects of unexpected unconventional monetary policies by the European Central Bank the yield spread between German and Italian government bond yields will be used as has been done by Rogers et al. (2014). It might seem more logical to use intraday changes on German bond yields as a proxy for policy shocks in the euro-area. However, as is explained by Rogers et al. (2015), the actions of the European Central Bank over the period examined in this paper have focussed on lowering intra-euro sovereign bond spreads. Normally these intra-euro area sovereign bond spreads represent default risks and risk premiums which should be considered as separate from monetary policy. But, due to the abnormal circumstances of the euro-area during recent periods, the intra-euro spreads can be thought of as more reliable indicators and appropriate measures of unconventional monetary policies executed by the European Central Bank.
Furthermore, the real interest rate is used as an explanatory variable in this research as the real interest rate is the variable that is influenced by the monetary policies of the European Central Bank. The real interest rate is used as an explanatory variable as this is the risk-free rate that an investor could obtain if he would invest in government bonds instead of investing in the stock market. The real interest rate is calculated as the difference between the one month bill rate in the Netherlands and the change in the Consumer Price Index.
Finally, the future dividends on the AEX index will be included to investigate what the main drivers are behind the relationship between monetary policies and stock market levels.
5.2 Error assumptions and other assumptions
18 the Breusch-Pagan-Godfrey test has been used. For equation (6) it turns out that heteroscedasticity is indeed a problem. To correct for this, HAC-standard errors have been used to make the coefficients and standard errors more reliable.
Besides heteroscedasticity, serial correlation can also constitute a problem while performing regression analyses. In the case of serial correlation the independent variable depends on lagged values of the dependent or independent variables. Given the fact that the data used in this paper, like for example interest rates, tend to be highly persistent over time it is highly probable that serial correlation constitutes a problem within the dataset. The Breusch-Godfrey Serial Correlation LM test has been used to test for serial correlation. After performing this test it turns out that serial correlation is present for the data used in equation (6). To correct for this and to make the coefficients and standard errors obtained more reliable, HAC-standard errors have been used. In the papers of Campbell and Ammer (1993) and Bernanke and Kuttner (2005) it is not explicitly stated that they have tested the variables on serial correlation. This test has been included in this paper to ensure that serial correlation does not constitute a problem while performing the regressions analysis. Therefore, the method employed in this paper differs slightly from the method used by the papers stated before.
Thirdly, non-stationarity of the variables used might cause problems when running regressions. Non-stationarity could cause spurious regressions and unreliable estimators, test statistics and predictors. To test for the stationarity of the variables used in this research, the Augmented Dickey-Fuller test has been used to find out if the variables have a unit root. As a result of the test, it can be concluded that non-stationarity does not constitute a problem.
Finally, a problem that occurs frequently when performing regression analysis is endogeneity. This implies that a variable that is used in the regression is correlated with the error terms. Some of the most common reasons for this limitation are reverse causation and omitted variables. However, as has been stated by Bernanke and Kuttner (2005), the focus on unexpected monetary policies allows circumvention of the problems of endogeneity and simultaneity. Therefore, it can be assumed that endogeneity does not constitute a problem for the dataset that is being used in this research paper.
6. Results
19 the equation that investigates the main drivers behind this relationship. Besides showing the main results of the empirical analyses that have been performed, this section will also comment on the validity of the hypotheses that have been formulated at the beginning of this research paper.
6.1.1 Results on the relationship between monetary policies and the level of stock prices in the Netherlands
In this part of the paper the results with respect to the relationship between unexpected conventional and unexpected unconventional monetary policies and the level of stock prices in the Netherlands will be discussed. The main focus will be the effects of unexpected unconventional monetary policies on level of stock prices. Furthermore, it has been investigated if there are differences between these effects during periods of the financial crisis and periods of economic stability. The results of equation (6), which has been used to investigate these results, are shown in table 1. The numbers in parentheses represent the t-statistics of the coefficients where *** indicates significance at the 1% level, ** indicates significance at the 5% level and * indicates significance at the 10% level.
Regressor Coefficient
Intercept 0.0055
Conventional monetary policy surprise Unconventional monetary policy surprise Crisis dummy x conventional monetary policy surprise (1.444) 0.0271 (0.8715) -0.0607*** (-4.3612) -0.0155 (-0.3179) Crisis dummy x unconventional monetary
policy surprise -0.4160*** (-2.8734) Crisis dummy -0.0121 (-1.0677) R-squared 0.1616
Table 1: Results of equation (6). (Rounded to 4 decimals).
20 and equal to -0.4160 during the financial crisis. These coefficients can be interpreted by taking into account the log-transformation that has been applied to the changes in stock price. This has been done by using the formula in equation (10).
𝛥% 𝑅𝑡𝑁𝐿 = 100·(𝑒𝛽1− 1) (10)
Using formula 10 one finds that a 1 percentage point increase in the spread between German and Italian government bonds as a result of unconventional monetary policy leads to a decrease of 5.889% in the level of stock prices in the Netherlands during normal times. The effect of a 1 percentage point increase in the spread between German and Italian government bonds during times of financial crisis is even bigger with a decrease of 34.032%.
So the effects of unexpected unconventional monetary policies from the European Central Bank on the level of stock prices in the Netherlands is not only statistically significant, but could also in terms of size have had a very significant impact. Moreover, it can be stated that there is a positive (negative) relationship between unexpected unconventional loosening (tightening) monetary policies and the level of stock prices in the Netherlands. This is in line with the results of previous research which have predicted a positive (negative) relationship between loosening (tightening) unexpected monetary policies and the level of stock prices. This also means that unexpected unconventional loosening monetary policies which have been implemented by the European Central Bank in response to the financial crisis have increased stock prices. With the results of table 1 we can find that a 1 percentage point decrease in the spread between German and Italian government bond yields leads to a 6.258% increase in the level of stock price in the Netherlands during times of economic stability. Furthermore, a 1 percentage point decrease during the financial crisis leads to an increase of 51.589% in the level of stock prices.
21 actual effects of unexpected unconventional monetary policies on the level of stock prices in the Netherlands have been more modest than the changes based on 1 percentage point which are obtained from the model that is used in this paper.
Finally, from the results in table 1 it becomes clear that only unexpected unconventional monetary policies have a significant negative influence on the level of stock prices in the Netherlands when compared to the effects of unexpected conventional monetary policies.
6.1.2 Hypotheses about the relationship between monetary policies and the level of stock prices
The hypotheses that have been formulated earlier on in this paper will now be verified based on the results of the empirical analyses that have been performed. In this section the first four hypotheses will be discussed.
Firstly, H1 states that unexpected and unconventional tightening monetary policies executed by the European Central Bank will have a negative effect on the level of stock prices in the Netherlands. From the results in table 1 it is visible that there is indeed a significant negative relationship between an unexpected and unconventional monetary policy measure of a tightening nature and the level of stock prices within the Netherlands. The coefficient is significantly negative, both during periods of economic stability and during the financial crisis. Next to this, this also proves that there is a significant positive relationship between unexpected unconventional loosening monetary policies and the level of stock prices in the Netherlands. Therefore it can be concluded that hypothesis 1 holds and that the findings of this research paper are in line with the main findings of previous literature that has been performed on this topic.
22 change in futures prices does not result in a significant changes in the level of stock prices. Therefore, it can be concluded that there is a substantial difference between the effects of unexpected unconventional and unexpected conventional monetary policies of the European Central Bank when it comes to its effects on the level of stock prices in the Netherlands. So, it can be concluded that H2 holds and that there are significant differences between the effects of unexpected unconventional and unexpected conventional monetary policies on the level of stock prices in the Netherlands.
Thirdly, H3 states that the effects of unexpected unconventional monetary policies executed by the European Central Bank have had different effects during the financial crisis when compared to periods of economic stability. The impact of the financial crisis has been included in a two-way manner in equation (6). It has been included as an interaction term in combination with unexpected conventional and unexpected unconventional monetary policies, but also as a stand-alone dummy variable. This last addition is included to capture the whole effect of the financial crisis on the relationship. When looking at the interaction terms it is clear that the interaction variable for the unexpected and unconventional monetary policies is statistically significant both during periods of financial crisis and during periods of economic stability. However, the magnitudes of the two coefficients differ substantially. The negative effect of unexpected unconventional monetary policies of a tightening nature is much stronger during the financial crisis. It is found to be 5.889% in normal times and 34.032% during the financial crisis. Furthermore, it is also possible to look what the differences are between the positive effects of loosening unexpected unconventional monetary policies when comparing the effect during the financial crisis and during times of economic stability. When interpreting the coefficients of table 1 it can be found that the positive effect is 6.258% during normal times compared to 51.589% during the global financial crisis. So, also in the case of unexpected unconventional monetary policies of a loosening nature the effect during the financial crisis has been significantly larger. Therefore, it is concluded that there are indeed substantial differences between the effects that unexpected unconventional monetary policies have had on the level of stock prices in the Netherlands and H3 holds.
23 employed by the European Central Bank. However, when looking at the total effect on the financial crisis which is captured by the coefficient that represents the complete effect of the dummy variable, it becomes clear that this coefficient is statistically insignificant. This means that for the general relationship between unexpected monetary policies and the level of stock prices in the Netherlands, the financial crisis has not had a significant impact. This has not been examined specifically in earlier papers so it is not possible to make a valid comparison to previous literature when it comes to the impact of the financial crisis. Therefore we can conclude that H4 does not hold for the Netherlands during the time period examined in this paper.
6.2.1 Results on the drivers behind the relationship between unexpected unconventional monetary policies and the level of stock prices in the Netherlands
Now the results on the main drivers behind the relationship between unexpected unconventional monetary policies and the level of stock prices in the Netherlands will be discussed. To be able to investigate these drivers, a forecasting VAR has been set up and a Cholesky decomposition has been applied to the variance of the forecasting VAR. The results are displayed in table 2 where the contributions of the variances of the three variables which are the real interest rate, future dividends and future excess returns are shown in percentages to the total variance of excess returns as a result of unexpected and unconventional monetary policies. All the contributions add up to 100% for every period so that it is possible to see what the main driver behind the relationship is.
Period Standard Error Future excess
returns (% of total variance)
Real interest rate (% of total variance) Future dividends (% of total variance) 1 20.7536 93.1145 2.4584 4.4271 2 21.4702 93.5491 2.3072 4.1438 3 21.5006 93.2885 2.5068 4.2047 4 21.5324 93.0730 2.6748 4.2522 5 21.5546 92.9078 2.8065 4.2857 6 21.5729 92.7615 2.9245 4.3140 7 21.5901 92.6233 3.0367 4.3399 8 21.6065 92.4912 3.1449 4.3639 9 21.6222 92.3649 3.2492 4.3859 10 21.6373 92.2442 3.3497 4.4062
24 From the results that are displayed in table 2 it is visible that future excess returns are the most important driver of changes in the level of stock prices in the Netherlands as a result of unexpected unconventional monetary policies of the European Central Bank. From the results that are shown in table 2 it can be seen that future excess returns account in almost all periods for a contribution of about 92% or 93%. So around 93% of the variance in the log-transformed changes in the levels of stock prices in the Netherlands as a result of unexpected unconventional monetary policies can be explained by changes in the future excess returns on these stocks.
When looking at the contributions of the other variables that have been included in the forecasting VAR it is visible that both future dividends and the real interest rate have a very small contribution to the Cholesky variance decomposition. The contribution of the real interest rate lays around 2% or 3% of explaining the total variance. This means that changes in the real interest rate as a result of unexpected unconventional monetary policies have a very small direct effect on the changes in the level of stock prices in the Netherlands. However, it seems to be the case that the contribution of the real interest rate is growing in later periods at the expense of the contribution of excess returns.
Finally, the contributions of expected future dividends that will be paid on the stocks account for around 4.2% of the total variance. Therefore the effects of unexpected unconventional monetary policies executed by the European Central Bank do not affect the level of stock prices in the Netherlands very substantially through the channel of expected future dividends.
When comparing the findings of this paper with respect to the contributions of the different components to the total variance of the forecasting VAR one has to be careful. One has to take into account that the findings of Bernanke and Kuttner (2005) and Campbell and Ammer (1991) have been retrieved by making use of equation (9). Furthermore, the time period which is being examined is different for the other paper. The data used in these papers is of a more historical nature whereas the investigation that has been performed in this paper is of a more recent nature. Nevertheless, it is still interesting to investigate how the results of this paper compare to those found by Bernanke and Kuttner (2005) and Campbell and Ammer (1991). It can be concluded that the findings in this paper correspond to a large extent with the results found by Bernanke and Kuttner (2005) and Campbell and Ammer (1991).
25 (contributions of above 100% are possible in their research due to the use of equation (9), which allows for correlations between the components, when decomposing the variance).
Next to this, the contribution of the real interest rate has also been found to be very small or even negligible in previous literature. This corresponds with the low contributions of the real interest rate which have also been found by this paper.
At last, there is one difference with respect to the contributions found by this research paper when compared to the previous literature. Future dividends are found to have a very small contribution to the total variance by this paper. However, Bernanke and Kuttner (2005) and Campbell and Ammer (1991) find more substantial contributions for future dividends of respectively 24.5% and 14%. However, this might be the result of the fact that the contribution of future dividends is overstated in the findings of Bernanke and Kuttner (2005) and Campbell and Ammer (1991) as has been discussed in the section on methodology.
6.2.2 Hypothesis on the drivers behind the relationship
The hypothesis that has been formulated based on previous literature and economic theory which was concerned with the main drivers behind the relationship between unexpected unconventional monetary policies and changes in the level of stock prices in the Netherlands will now be discussed.
H5 stated that expected future dividends were expected to be the main driver through which unexpected unconventional monetary policies would have effects on changes in the level of stock prices in the Netherlands. From the results in table 2 it is visible that not future dividends, but future excess returns on the stocks are the main driver behind the relationship just described. The Cholesky variance decomposition shows that around 93% of the variance of changes in the level of stock prices is attributable to changes in future excess returns of the stock and only around 4.2% is attributable to future dividends. Furthermore, an even smaller percentage of around 3% can be attributed to effects through the real interest rate. This is probably a result of the fact that changes in the real interest rate that are caused by unexpected and unconventional monetary policies are transitory and not very persistent, as is also argued by Bernanke and Kuttner (2005). Therefore, it can be concluded that H5 does not hold.
7. Conclusions and discussion
26 futures markets are used to account for the expectations of these monetary policies. Furthermore, the approach adopted by Rogers et al (2014) is used to account for the unconventionality of monetary policies. The spread between German and Italian government bond yields has been used in this paper to find out whether monetary policies employed by the European Central Bank were of an unconventional nature. This relationship between unexpected unconventional monetary policies and the level of stock prices in the Netherlands has been compared to the effects of unexpected conventional monetary policies. Moreover, the relationship between unexpected unconventional monetary policies and the level of stock prices is compared during the financial crisis and periods of economic stability.
Besides investigating the validity of the relationship between unexpected unconventional monetary policies of the European Central Bank, this paper also endeavoured to create more insights about the main drivers behind this relationship. More specifically, to examine through which channel the monetary policies investigated in this paper affect the level of stock prices in the Netherlands. The three channels which have been examined are the real interest rate, future dividends and future excess returns on stocks. The contributions of these three variables to the relationship between unexpected unconventional monetary policies have been examined by making use of a forecasting VAR and a Cholesky variance decomposition to determine the contributions.
27 unexpected unconventional monetary policies outweigh the negative impact of their tightening counterparts. More research is needed to verify whether this relationship holds for all the countries in the whole monetary union so as to make sure that there will be no conflicting effects after possible redesigning of monetary policies by the European Central Bank.
Secondly, it has been found that the relationship between unexpected conventional monetary policies and the level of stock price in the Netherlands is insignificant. This seems to imply that, although the adjustments to monetary policies that are made by the European Central Bank were unexpected, they do not surprise market participants in such a way that it has immediate effects on the level of stock prices. This indicates that it is the unconventional component of unexpected unconventional monetary policies which causes the significant reaction of stock prices.
Thirdly, this paper finds evidence of the fact that the financial crisis has not had a significant influence on the relationship between unexpected monetary policies and the level of stock prices in the Netherlands in general. However, in the case of unexpected unconventional loosening (tightening) monetary policies, a significant positive (negative) relationship is observed. An important remark that has to be made about this is that the effect is a lot stronger during the financial crisis. The coefficient changes from -0.0607 to -0.4160. This indicates that unexpected unconventional loosening monetary policies came with greater benefits during the financial crisis. The fact that unexpected conventional monetary policies have not had a significant influence during the period of financial crisis seems to contribute even more to the fact that it is the unconventionality of monetary policies that is responsible for the significant reaction of stock prices. Possible solutions and directions for future research on this topic have been provided in the previous section.
29 8. References
Bernanke, B., Kuttner, K., 2005. What explains the stock market’s reaction to Federal Reserve policy? Journal of Finance 60, 1221–1257.
Campbell, J., 1991. A variance decomposition for stock returns. Economic Journal 101, no. 405: 157-179
Campbell, J., Ammer, J., 1993. What moves the stock and bond markets? A variance decomposition for long-term asset returns. The Journal of Finance, 48, no. 1: 3-37.
Cassola, N., Morana C., 2002. Monetary policy and the stock market in the EURO area. Working paper no. 119. European Central Bank, Frankfurt.
Gali, J., Gambetti, L., 2013. The effect of monetary policy on stock market bubbles: Some evidence. Unpublished working paper. Barcelona Graduate School of Economics, Barcelona.
Gospodinov, N., Jamali, I., 2015. The response of stock market volatility to futures-based measures of monetary policy shocks. International Review of Economics and Finance 37, 42-54.
Haitsma, R., Unalmis, D., De Haan, J., 2015. The impact of the ECB’s conventional and unconventional monetary policies on stock markets. Working paper no. 483. De Nederlandsche Bank, Amsterdam.
Ioannidis, C., Kontonikas, A., 2006. Monetary policy and the stock market: some international evidence. Working paper. University of Glasgow, Glasgow.
Kontonikas, A., Maio, P., Zekaite, Z., 2016. Monetary policy and corporate bond returns. Working paper. University of Glasgow, Glasgow.
Laeven, L., Tong, H., 2010. U.S. monetary shocks and global stock prices. IMF working paper.
30 Pattipeilohy, C., Van Den End, J., Tabbae, M., De Haan., 2013. Unconventional monetary policy of the ECB during the financial crisis: An assessment and new evidence. Working paper no. 381. De Nederlandsche Bank, Amsterdam.
31 9. Appendix
Table 3: Descriptive statistics variables of equation (6)
This table reports the values of the descriptive statistics of the variables which have been included into equation (6). Rounded to 4 decimals.
Log_Aex_index Conv_mon_pol_surprise Unconv_mon_pol_surprise
Mean 0.0022 -0.0139 -0.0072 Median 0.0102 0.001 0.0000 Maximum 0.1483 0.475 1.14 Minimum -0.3162 -0.961 -1.08 Std. Dev. 0.0543 0.1727 0.2048 Skewness -1.3866 -1.9104 0.1478 Jarque-Bera 327.7906 872.9027 1068.24 Probability 0.0000 0.000 0.0000 Observations 218 218 218
Table 4: Descriptive statistics variable Cholesky decomposition
This table reports the values of the descriptive statistics of the variables that have been used in the Cholesky decomposition. Rounded to 4 decimals.
Excess Returns Real Interest Rate Future Dividends
