Predictability of Stock price movements : evidence from the Amsterdam Stock Exchange

(1)

Predictability of Stock price movements

Evidence from the Amsterdam Stock Exchange

Name: Yunis Lankarani

Student number: 10826947

Supervisor: Drs. P.V.Trietsch, M.Phil.

Study: BSc Economics and Business

Track: Economics and Finance

Number of credits thesis: 12 ECTS

(2)

Statement of Originality

This document is written by Yunis Lankarani who declares to take full responsibility of the

contents of this document.

I declare that the text and the work presented in this document are original and that no

sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of

(3)

Abstract

This paper aims to answer the following research question: Can the price movements of the

Amsterdam stock exchange price index be predicted using macroeconomic variables? The

Arbitrage Pricing model is used to check the relationship in a monthly time series data over the

period of 1983 - 2018. Five macroeconomic variables examined in this paper to predict the

AEX index price movements are the interest rate, the exchange rate, industrial production, the

CPI, and the money supply. Two statistical tests, namely, the Ordinary Least Squares (OLS)

regression analysis and the Granger Causality test are applied to examine the relationship. The

regression coefficients indicate that the exchange rate is positively and significantly related to

the AEX price index. The interest rate, industrial production, the CPI, and the money supply

coefficients match the expected hypothesis but insignificantly related to the AEX price index.

The Granger Causality test concludes that at two variable lag level the exchange rate and the

money supply are significant in forecasting the AEX price index.

(4)

Chapter 1: Introduction

1.1 Motivation

One of the fundamental issues in financial markets is understanding how firm characteristics and economic forces affect the stock price movements. Prediction of the stock price movements is an attempt to determine the firm value in the future on a stock exchange market. The successful

prediction will result in future profits, which every investor desires. However, it is not an easy task, and some consider that it is not possible to beat the market because unexpected shocks affect the financial markets on a daily basis. There has been plenty of research done regarding the predictability of the stock prices. For instance, Campbell and Yogo (2006), attempted to determine whether the traditional regression methods used to test stock predictability were valid. The research on the predictability of monthly stocks prices by Kandel and Stambaugh (1996), stated that the statistical measures such as regression analysis are not strong enough to predict the future stock movements due to the efficiency and random walk of the stock market.

The main focus of this research will be to answer the following question. Can the price movements of the Amsterdam stock exchange price index be predicted using macroeconomic variables? Predicting the stock price movements is testing whether the AEX price index is subject to the changes in the macroeconomic variables. This research is relevant since, Amsterdam Stock Exchange (AEX) with a market capitalization of 500 billion euros, and volume of 99,174,792 makes it the largest stock exchange market in the Netherlands (See Appendix figure 1 for the AEX Price Index graph). AEX is part of the Euronext stock exchange. Euronext is located in 6 countries Belgium, UK, Portugal, Ireland, France and the Netherlands. The headquarters are located in Amsterdam with a total market capitalization of 4649 billion USD.

1.2 Summary of Existing literature

Whether reliable predictor variables of the stock prices exist is still a debate. Using the publicly available information to predict the stock price movements violates semi-strong form of the efficient market hypothesis. Several researchers have focused their studies on whether the efficient market hypothesis hold and the publicly available firm-specific information can help investors to predict the future stock price movements. For example, Keim and Stambaugh (1986) concluded that few observable firm-specific financial variables such as default risk, size of a firm are significantly related to the stock market price movements. Another research done by Pontiff and Schall (1998) indicated that the book to market ratio is significantly related to the future stock prices. According to Lewellen (2004), the book to market ratio, the dividend yield and the earnings-price ratio have a significant predictive effect on the stock prices. Hong et al. (2007) pointed out evidence that the industry portfolio returns in the U.S. stock markets are a good predictor of the future stock price movements.

(6)

The previous paragraph summarized the recent researches based on the relationship between the stock prices and the firm-specific financial information. However, this paper aims to find whether the future stock price movements are subject to the macroeconomic variables. According to the researchers explained below, it is evident that the stock price movements are subject to changes in the macroeconomic variables. However, there are multiple views on this regarding the phenomenon.

Some studies observed that the macroeconomic variables are an inconsistent predictor of the stock prices over the different periods of time. Many scholars were unable to find any evidence for the stock predictability through macroeconomic variables (Chan et al., 1998). For instance, Durham (2001) is one of those studies. For one single variable, some studies found that macroeconomic variables can help to predict the stock price movements while the other studies contradicted this results. For instance, Balvers et al. (1990) and Flannery and Protopapadakis (2002) who analyzed the macroeconomic variables and concluded a strong link with the stock prices. Moreover, there has been extensive research regarding the phenomenon suggesting that there might be concerns regarding the data mining. Therefore, due to the mixed evidence, there is no unique conclusion regarding whether the macroeconomic variables can predict the stock price movements.

1.3 Sub-questions

There are additional questions that need to be discussed in order to answer the research question. Firstly, whether the stock price movements are predictable according to the Efficient Market Hypothesis (EMH)? It is crucial because the EMH states that the future stock price movements are unpredictable based on the publicly available and private information. Also, the past prices do not predict the future price movements. Additionally, whether the technical analysis (TA) can be used to predict the stock price movements will be discussed. The TA is a tool to help the investors to predict the stock price movements based on the past observations, but it is against the EMH. Another question is which macroeconomic factors affect the stock price movements?

1.4 Data & Methodology

The objective of the data analysis is to empirically examine the impact of changes in the macroeconomic variables on the AEX price index. The data sample consists of 422 monthly observations of the dependent and the independent variables using time period of 1983-2018. The dependent variable is the AEX price index, and the independent variables are the interest rate on Dutch government bonds, the Eurodollar parity exchange rate, GDP (Industrial Production) of the Netherlands, the inflation rate (CPI), and the money supply of the Netherlands.

The model that applies to this research is the Arbitrage Pricing Theory (APT) model. The APT is one of the well-known models to test the relationship between the macroeconomic variables and stock prices. It adds to the CAPM multiple independent variables. CAPM is too simple since it includes only one independent variable. Assumptions of the CAPM such as perfect capital markets and perfect competition also form the basis of this model.

(7)

Two main econometric analyses will be examined in this study are the Ordinary Least Squares (OLS) regression analysis and the Granger Causality test. Before conducting these tests, the unit root test will be implemented to check whether the data is stationary and multicollinearity test will be used to make sure that the independent variables are not highly correlated with each other. These two tests need to hold before implementing the OLS regression and the Granger Causality analysis. The OLS checks whether there is a relationship between the macroeconomic variables and the AEX price index. The Granger Causality test checks whether one-time series is significant in forecasting the other.

Three residual tests will be investigated to check the predictive power of the OLS.

Correlogram for residuals is used to test whether the residuals are white noise meaning that they are mean zero, no serial correlation, independent, and stationary. Heteroskedasticity test to check whether the standard errors for certain observations are biased. Also, normality test will be conducted to verify if the residuals follow a normal distribution.

1.5 Structure

This paper will look to answer the research question as follow: First, in chapter two the literature review will summarize findings of the topics such as the EMH, how the TA applies to the stock and lastly, which macroeconomic factors affecting the stock prices will be discussed. Next, methodology section will be introduced in chapter three. The methodology section will include data descriptions and the expectations as well as the model that will be used to answer the research question. Chapter four will present the summary of the empirical test results and discussions on the findings. Lastly, the conclusion will be reached whether the price movements of the AEX index can be predicted using the macroeconomic variables.

(8)

Chapter 2: Literature Review

2.1 Are stock price movements predictable according to the EMH?

This section aims to summarize the previous findings on the EMH. Three sub-sections will be discussed in order to conclude if the price movements are predictable according to the EMH. The first section, background information and definitions. The next section discusses the previous pieces of literatures on how to test the EMH. Lastly, previous findings on whether the stock markets are efficient will be summarized.

2.1.1 What is the EMH?

The idea behind the EMH is that the markets are efficient and the stock prices reflect the best available information (Fama, 1970). Bachelier (1900) concluded that the stock prices are based on the present and the past actions, and it has no relationship with the future prices. Addition to this, stocks are correctly priced, and higher returns come with higher risks. The cheaper the information, and higher the competition among the investors, the more efficient markets will be. More precisely, if the prices could be predicted based on the publicly available information, then every investor would use this information to make profits. As a result, in the efficient markets, there is no possibility to beat the market persistently, and no alpha is available. Alpha is a return on investment compared to the market index.

The EMH is critical because it is welfare improving (Bodie, Kane & Marcus, 2011). With the correct prices, investors or firms can correctly calculate the expected return for the projects or observe which stocks to buy or sell. Accurate stock prices lead to more efficient economic asset allocation. When the prices are uninformative, people do not know what to expect from the market and investors are less willing to invest or trade. As a result, financing becomes more expensive and the economic growth reduces as well as the income.

The EMH is highly related to the Random Walk Hypothesis (RWH) (Fama, 1965). The RWH states that the past stock price performance is uninformative about the future performance and the prices should change upon the announcement of any news related to the firm or the market. Also, no reliable price drifts should occur after the announcement just because the stock prices are independent and follow the random walk in efficient markets.

Three main points stressed in the EMH definition (Timmerman & Granger, 2014). The first point is the reliability of the selected information. It indicates whether the investors can believe the piece of information. The next point is that if the investors have skills to exploit the information to make a positive return on time. The last point is that making alpha after all the transaction costs to the brokers and the other parties.

There are three versions of the EMH, namely the weak-form, the semi strong-form and the strong-form (Bodie, Kane & Marcus, 2011). The weak-form hypothesis implies that the prices reflect all information that can be derived from the past prices. The semi strong-form hypothesis says that all

(9)

the publicly available information that relates to a firm’s future already reflected in the stock prices. This information includes prices, fundamentals, forecasts, economic and financial reports. The last form of the EMH is the strong-form. It is extreme and unrealistic, states that all the relevant information about the company reflected on the stock price including the publicly and the privately available information. Usually, information about the firm’s future is very private, and only the insiders possess this information, and it is not reflected in the stock prices.

2.1.2 How to test the EMH?

For many decades, academics are trying to come up with the models to test the EMH. According to Timmerman and Granger (2014), complex empirical methods must be used to test the EMH theory, but it is very problematic. Fama (1970), conducted three EMH tests based on the data before 1960 using probabilistic models and tested the three forms of the EMH. He found that the first two tests fail to reject the null hypothesis. Therefore, the weak form and the semi-strong form can be concluded to be efficient. However, he failed to statistically conclude that the strong-form of the EMH holds due to the monopolistic access of the information to the insiders.

Jensen and Benington (1970), indicated that the most empirical evidence support that there are two types of EMH tests. The first type is statistical in nature, that means statistical tools being used to determine the price fluctuations. In short-run the stock prices are roughly independent for the long price series thus the stock market is efficient. The second type of test includes mechanical trading rules. It is based on the long-run price series, and the evidence is very consistent with the random walk, and thus efficiency. These two types of tests have their drawbacks due to the statistical procedures. The hypothesis that for a short period of time, the price of the securities may not fluctuate randomly failed to be rejected. Similarly, the hypothesis that stated the price fluctuations follow a random pattern in more extended periods of the time can also not be rejected. It could be that either the variance or the mean have reacted to the new information.

2.1.3 How efficient are the stock markets?

Academics usually argue that the major stock exchange markets may be considered efficient markets (Fama, 1995). Fama (1970) and Malkiel (2005), indicated that the markets are efficient with few exceptions. Another research done by Maskin and Laffont (1990) stated that there is a strong evidence to believe that the markets are not efficient when the investors have irrational expectations.

Cooper (1982) analyzed the price behavior of the American and the British stock markets and concluded that there is evidence to support the random walk. However, the exact behavior of the stock prices is unpredictable in the stock markets mentioned above. Another research by Solnik (1973) argued that there are larger deviations from the EMH in European stock markets than in the U.S. stock markets. Also, the European financial markets have smaller correlation coefficients, and it is not essential from an investor point of view. Nevertheless, those deviations are more significant for the

(10)

daily price changes, but the transaction costs would eliminate those deviations for the daily trading strategy.

Lim et al. (2008) investigated the efficiency of the Asian stock markets prior, during, and post 1997 financial crisis by applying the rolling bicorrelation test statistics. They found that during the crisis period, the Asian markets were inefficient however post-crisis, markets started to improve in terms of efficiency. They explained the inefficiency during the crisis by overaction of the investors to the external news and higher correlation of the Asian stock markets during the crisis period.

2.2 Can Technical Analysis be used to predict stock price movements?

Technical analysis (TA) is one of the common tools used by the investors to predict the stock price behavior. Han et al. (2013) stated that the TA is a financial analysis tool used to evaluate the future security prices by analyzing the current and the past prices and volume movements. “Technical analysts are sometimes called chartists because they study the records or charts of the past stock prices, hoping to find patterns they can exploit to make a profit” (Bodie, Kane & Marcus, 2011). According to Lo et al. (2000), many academics have proven that the TA for being theoretically useless, but it is still one of the favorite tools for predicting the stock prices among the investors. The EMH theory does not support this idea because the TA violates the weak-form of the EMH.

Taking into consideration the fact that the TA violates the EMH. There is still on-going debate among investors whether this technique can be used to predict the stock price behavior. For instance, Lakonishok and LeBaron (1992), argued that with the use of simple trading rules significant changes in price behavior of Dow Jones Industrial Average can be predicted using a significantly large sample of data. However, Jensen and Benington (1970) concluded that the TA is pointless because markets are efficient. According to Qian and Rasheed (2007), there are three essential assumptions of the TA. Actions in the market reflect everything, prices follow trends, and history repeats itself.

According to Sehgal and Gupta (2007), TA is a subjective method to predict the stock price movements because whether an investor can apply these tools to predict the stock prices depends on the skill level. Therefore, it purely depends on the intelligence and the ability of the analysts. However, the traders argue that the TA works well in the stock markets due to the short-term and long-term trends.

2.3 Which macroeconomic factors affect stock price movements?

This section aims to summarize the previous findings on the relationship between the macroeconomic variables and the stock prices. Macroeconomic factors have an impact on a large population and not just a small number of the individuals. It is associated with the overall economy at the national level. Change in the macroeconomic variables are shocks to the economy which impacts the direction of the stock market as well. Some of the essential macroeconomic variables directly affect the stock prices are the interest rates, the exchange rate, GDP, the inflation rate and the money

(11)

supply. There are many other numerous macroeconomic variables which directly affects stock returns, but these five variables are widely recognized. Following sections will summarize the findings listed in tables 1 and 2.

Table 1: Previous Researches

Macroeconomic variable Positive Negative

Interest rate Nasseh and Strauss (2000) Nasseh and Strauss (2000) Ratanapakorn and Sharma (2007) Geetha et al. (2011)

GDP/industrial Production

Morck et al. (2000) Nasseh and Strauss (2000) Bilson et al. (2001) Chen et al. (1986)

Morck et al. (2000)

Exchange rate Ma & Kao (1990) Young (1972) Ma & Kao (1990) Doong et al. (2005) Geetha et al. (2011)

Inflation/CPI Bilson et al. (2001) Firth (1979)

Humpe and Macmillan (2009) Bilson et al. (2001)

Fama and Schwert (1977)

Money Supply Palmer (1970)

Homa and Jaffee (1971) Hamburger and Kochin (1972)

Flannery and Protopapadakis (2012)

Mahedi (2012) This table represents previous findings relating to the impact of macroeconomic variables on the stock prices.

Table 2: Previous Researches continued

Author Variable Year Region Method

Palmer (1970) Money supply 1959-1969 U.S. OLS regression

Homa and Jaffee (1971)

Money supply 1954-1969 U.S. OLS regression

Hamburger and Kochin (1972)

Money supply 1950-1969 U.S. OLS regression

Franc and Young (1972)

Exchange rate 1967-1971 U.S. OLS regression

Fama and Schwert (1977)

Inflation 1953–1971 U.S. OLS regression

(12)

Chen et al. (1986) GDP Inflation 1953-1978 U.S. OLS regression

Ma & Kao (1990) Exchange rate 1973-1983 Countries with floating exchange rates

OLS regression analysis

Morck et al. (2000) GDP 1926-1995 China, Malaysia, Poland, Denmark, Iceland, US

OLS regression

Nasseh and Strauss (2000)

Interest rate GDP

1962-1995 Europe Johansen

cointegration tests

Bilson et al. (2001) GDP 1985-1997 Emerging countries

OLS regression

Doong et al. (2005) Exchange rate Post 1997 Asia Granger Causality test

Ratanapakorn and Sharma (2007)

Interest rate 1975-1995 U.S. Granger Causality test Humpe and Macmillan (2009) Inflation 1965-2005 US Japan Contingent analysis

Geetha et al. (2011) Interest rate Exchange rate Post 1997 Malaysia, China U.S. Multivariate Vector Error Correction (VEC) models Flannery and Protopapadakis (2012) Inflation Money supply

1980-1996 U.S. GARCH model

Mahedi (2012) Money supply 1999-2011 Germany UK

Variance

Decomposition and Impulse Response Analysis

This table represents previous findings relating to the impact of macroeconomic variables on the stock prices based on authors, time period, region and method.

2.3.1 Interest rate

The interest rate is the discount rate used to determine the present value of the future cash flows which directly affects the stock prices. If the security is priced correctly at the time of the purchase, the interest rate changes will lead to the stock price fluctuations. An increase or decrease in discount rate changes the present value (PV) of the cash flows and thus the stock prices.

(13)

Ratanapakorn and Sharma (2007) examined the interest rate behavior using the Granger Causality test in the U.S. stock market during the period of 1975 and 1999 and found a negative relationship between the stock prices and the interest rates. Nasseh and Strauss (2000) examined the interest rate and the stock price relationship in six European stock markets using Johansen

cointegration tests between 1962 and 1995. They observed positive coefficients for the short-term interest rates and negative coefficients for the long-term interest rates on the stock prices. Another research by Geetha et al. (2011) investigated Malaysia, China and U.S. post Asian financial crisis of 1997 using multivariate Vector Error-Correction (VEC) models and concluded a negative link between the stock prices and the interest rates both in the short-run and the long-run.

2.3.2 Gross Domestic Product

The Gross Domestic Product (GDP) is another macroeconomic variable that influences the stock prices because GDP is an important indicator of the health of the economy. GDP is good estimator to determine the economic performance of a country. More precisely, it measures the value of all final goods and services provided in a given time period.

Morck et al. (2000) implemented the multivariate regression analysis using a data from 1926 to 1995 and concluded that the stock prices and GDP in developed countries move in opposite direction, but in less developed economies the stock prices and GDP seem to follow the same trends. They stated three reasons for these findings. First, less developed countries have more correlated fundamentals which make the stock prices to move together. Second, poor and uncertain protection of private property rights in small countries which lead to the more market-wide price fluctuations. Third, low GDP countries provide less protection for the investors from the corporate insiders which makes publicly available information less valuable.

Besides testing the interest rates, Nasseh and Strauss (2000) also found a positive relationship between GDP and the stock prices. Bilson et al. (2001) examined emerging stock markets using the OLS regression analysis between 1985 and 1997. They concluded a positive link between the stock prices and the real future activity which are the GDP and the industrial production. Another research by Chen et al. (1986) tested the EMH using macroeconomic information over the period of 1953-1978 in the U.S. stock markets. They found that the GDP and the industrial production are positively related to the stock price movements. They used APT model to conduct this test.

2.3.3 Exchange rate

Another crucial macroeconomic factor that influences the stock price movements is the exchange rates. There are two definitions of the exchange rate (Pilbeam, 2013). The price of foreign currency in terms of the domestic currency and the price of domestic currency in terms of the foreign currency. Depreciation (Appreciation) of the currency leads to a decrease (increase) in the purchasing power of the money. It is crucial because, the return on the stock investments are based on both

(14)

domestic rates of return and the expected change of the currency value in which stock sold (Ma & Kao, 1990).

Franc and Young (1972) examined 280 U.S. companies using the Rhomberg model on a data from 1967 to 1971 and concluded that the stock prices are sensitive to the exchange rate fluctuations. Ma and Kao (1990) stated that the stock returns are exposure to the two different kinds of the foreign exchange risks, namely, the exchange rate levels and the exchange rate changes in countries with the floating exchange rates. They found that the stock prices are positively related to the exchange rates in import-dominated economies and negatively related in export-dominated economies from 1973 to 1983. Doong et al. (2005), investigated the Granger Causality test on the relationship between the stock prices and the exchange rates in six Asian countries. They concluded that the exchange rates and the stock price movements are negatively related in all the six countries. Another research by Geetha et al. (2011) found a negative link between the stock prices and the exchange rates in Malaysia, U.S. and China.

2.3.4 Inflation rate

Next variable that affects the stock prices is the inflation rate. The inflation is defined as an increase in overall prices and a decrease in purchasing power of the money. To be more specific, money becomes less worthy and the cost of goods increase. There are two types of inflation, namely, the anticipated inflation and the unanticipated inflation (Parks, 1978). Depending on the kind of the inflation, the stock prices can react positively or negatively to the shock. Also, the stock price movements depend on firms ability to hedge the inflation risk (Bilson, Brailsford & Hooper, 2001).

Fama and Schwert (1977) investigated the link between the stock prices and both the

unanticipated and the anticipated inflation in the U.S. stock market using the OLS regression analysis on a data ranging from 1973 to 1983. They concluded a negative relationship in both cases. Firth (1979) examined the Fischer hypothesis in the British stock market from 1955 to 1976. He concluded a positive link between the stock prices and the inflation. Chen et al. (1986) found that the inflation has no significant effect on the stock prices in the U.S. economy. Humpe and Macmillan (2009) stated that inflation unpredictability affects the discount rate which in turn changes the PV of the future cash flows and thus affects the stock prices negatively. Their evidence is from the U.S. and Japanese stock markets using contingent analysis between 1965 and 2005. Also, Flannery and Protopapadakis (2012) concluded a negative relationship over the period of 1980 to 1996 using the GARCH model.

2.3.5 Money supply

Money supply will be the last variable to be examined in this paper. The money supply is the total amount of money in the economy. The money supply is the most liquid asset in the economy, and there are various types of it including M1, M2 and M3. However, each country might use different classifications. M1 is the narrowest type of the money supply and includes coins and notes that can quickly be converted to cash. M2 is a broader type and includes M1 plus short-term demand

(15)

deposits and the money market funds. M3 is the broadest one and includes M2 plus the long-term time deposits. M3 is no longer reported by many central banks including the Federal Reserves.

Palmer (1970) investigated the S&P500 index from 1959 to 1969 using the OLS regression analysis. He concluded that the money supply is considered to be one of the crucial determinants of the stock prices and follows the same trend with it. Homa and Jaffee (1971) examined quarterly data between 1954 to 1969 using the OLS regression analysis concluded a positive relationship between the stock prices and the money supply in the U.S. stock market. Another research in the U.S. between 1950 and 1969 by Hamburger and Kochin (1972) found that when the money supply increases, the money demand increases as well over the period. This expansion leads to an increase in economic activity and thus increases the stock prices which is considered as a direct portfolio effect.

Flannery and Protopapadakis (2012) indicated a negative relationship between the money supply and the stock prices in the U.S. stock market. They used the GARCH model to find a correlation between the stock prices and the money supply over the period of 1980-1996. Mahedi (2012) examined the money supply and the stock price movements in German and British stock markets and concluded a negative relationship over the period of 1999 to 2011 using the Variance Decomposition and Impulse Response analysis.

(16)

Chapter 3: Data & Methodology

3.1 Data & Descriptions

In this section, the expected impact of the dependent on the independent variables will be described as well as the data and the descriptive statistics. The dependent variable is the AEX price index which consists of the market value weighted index of 25 stocks. The independent variables are the interest rate, the exchange rate, industrial production, the CPI, and the money supply. Each of the independent variables has a unique relationship with the stock prices. Following paragraphs will summarize the coefficient expectations between the AEX price index and the stated macroeconomic variables.

The first independent variable is the interest rate. The interest rate on Dutch Government bonds with ten years maturity will be used as a proxy for the interest rate. The interest rate on government bonds is a risk-free rate, more precisely it is the rate of return on investing government bonds or depositing at the central banks. When the interest rate increases, the discount rate which companies use to discount future cash flows increases as well. Therefore, higher interest rate expected to decrease stock prices. Hence, the expected interest rate coefficient is negative.

The exchange rate is the Euro/Dollar exchange rate parity adjusted for the Dutch guilder before 1999. The Netherlands is in the European Union since 1999 and uses Euro as a currency officially since 2002. Before this period, the Dutch guilder was an official currency in the country. Depreciation (increase) of the exchange rate negatively affects the stock prices due to the anticipated inflation. Additionally, companies which import scarce material will be affected most by this

depreciation, because weak domestic currency increases the import costs. The higher costs lead to lower income and thus lower stock prices. On the other hand, domestic exporters benefit from the depreciation of the local currency because the demand for the domestic goods by the foreign consumers increases. The general effect is insignificant, but in this paper expected exchange rate coefficient is positive. This is because total exports in European Union exceed total imports and has a direct positive impact on the stock prices.

The GDP is one of the ways to measure economy of the Netherlands. The industrial

production will be used as a proxy for the GDP. Higher GDP or industrial production also leads to the higher stock prices. The reason is that there are more transactions and outputs in the economy and thus the stock prices should be high due to the increased productions and the cash flows. The coefficient expectations are a positive between industrial production and the stock price movements.

The CPI will be used as a proxy for the inflation. The inflation can positively or negatively affect the stock prices depending on the firms' ability to hedge the risk and whether the inflation is anticipated or unanticipated. The anticipated inflation is positively related to the stock prices because firms expect it and plan their activities accordingly. It results in increased earnings due to the higher prices in the economy which leads to the higher cash flows and the dividend payments thus increase

(17)

the stock prices. The opposite occurs when there is an unanticipated inflation shock in the economy, resources move from investments to consumption due to the higher prices in the marketplace. Also, inflation leads to a higher nominal interest rate. The PV of the cash flows discounted using nominal interest rate. Therefore, higher discount rate leads to the lower stock prices. Additionally, increase in unexpected inflation level decreases sales and revenues as well as income leading to the lower stock prices. The coefficient expectations in this paper are based on the unanticipated inflation and thus negative.

The money supply will be the last independent variable. In this model M2 monetary

aggregate will be used as a proxy for the money supply. The reason is that M1 is too narrow and M3 is too broad. Therefore, M2 is considered to be a better measure of the money supply among three. There are at least three mechanisms the stock prices are affected by the money supply. Firstly, changes in the money supply result in future inflation uncertainty. This uncertainty is negatively related to the expected future stock price movements. Secondly, changes in the money supply may impact the economic activity positively which in turn affects the stock prices positively. Lastly, positive relationship due to the portfolio theory. Portfolio theory suggests that increase in the money supply leads to portfolio shift from noninterest-bearing money to the stock markets. The expected money supply coefficient is positive in this paper.

Table 3 represents the summary of the coefficient expectations between the AEX price index and the macroeconomic variables. Table 4 represents the descriptive statistics of the dependent and the independent variables. In total 422 monthly observations in a time period between 1983 to 2018 collected from the well-known data source DATASTREAM.

Table 3: Hypotheses

Variables Expected coefficient

AEX (Dependent variable)

Interest rate (𝑩_𝟏) Negative

Exchange rate (𝑩_𝟐) Positive

Industrial production (𝑩_𝟑) Positive

CPI (𝑩_𝟒) Negative

Money supply (𝑩_𝟓) Positive

This table represents the hypotheses and the coefficient expectations. All variables are defined in section 3.1.

𝐻0: 𝐵_,= 𝐵_.= 𝐵_/ = 𝐵₀= 𝐵₁= 0

(18)

Table 4: Descriptive statistics

Variables Obs Mean Min Max STD

AEX 422 309.009 47.930 687.730 167.319 INT 422 4.927 0.030 9.220 2.418 EXCH 422 0.863 0.636 1.398 0.143 IND 422 83.707 57.720 107.800 13.158 CPI 422 76.690 53.190 102.310 15.590 M2 422 416769.600 103969.000 878008.000 251357.200 This table represents the descriptive statistics of all the variables between time period of 1983-2018. All variables are defined in section 3.1.

3.2 Methodology

In order to check the relationship between the AEX price index and the independent

macroeconomic variables, following procedures will be implemented. Firstly, data will be converted to the logarithmic form, and the first differences will be taken. The reason for the logarithmic form is to scale the variables and make the relationship more linear. The reason for taking the first differences is to make data stationary which will be discussed in chapter 4.1 in more detail. Therefore, the model will be as follow:

𝐷𝐿𝑜𝑔(𝐴𝐸𝑋) = 𝛼 + 𝛽_,𝐷𝐿𝑜𝑔(𝐼𝑁𝑇) + 𝛽_.𝐷𝐿𝑜𝑔(𝐸𝑋𝐶𝐻) + 𝛽_/𝐷𝐿𝑜𝑔(𝐼𝑁𝐷) + 𝛽₀𝐷𝐿𝑜𝑔(𝐶𝑃𝐼) + 𝛽₁𝐷𝐿𝑜𝑔(𝑀2) + 𝜀

Where:

AEX = AEX Price Index INT = Interest Rate EXCH = Exchange Rate IND = Industrial Production CPI = Consumer Price Index

M2 = Money supply or Monetary Aggregate 2 DLog=First logarithmic difference

Two econometrical analysis will be implemented in this paper to test the model, the OLS regression analysis and the Granger Causality test. Before conducting these two tests, Augmented Dickey-Fuller (ADF) unit root test on both the dependent and the independent variables will be applied. The purpose of the unit root test is to check whether the variables are stationary. It is important to use stationary variables when dealing with the OLS regression analysis and the time series data. Also, an important assumption of the Granger Causality test states that the variables should be stationary. The second test, multicollinearity is an important assumption of the OLS

(19)

regression analysis. Multicollinearity will be tested to examine whether the independent variables are correlated with each other.

After completing both the ADF unit root and the multicollinearity tests, the OLS regression analysis will be implemented to statistically test the impact of each independent macroeconomic variables on the AEX price index. Next, to check the predictive power of the OLS three residual tests will be conducted. The first test will be the correlogram for the residuals or the white noise test. This test aims to confirm whether the residuals are white noise, more precisely stationary, independent and mean zero. The second test is to check for heteroscedasticity, because it leads to bias estimation. Last residual test will be the normality check to see if the residuals are normally distributed. Lastly, the Granger causality analysis will be tested. The purpose of this analysis is to check whether one-time series is significant to predict another using two variable lags. More precisely whether Y (the dependent) variable is caused by X (the independent) variable or vice versa.

(20)

Chapter 4: Data Analysis

4.1 Unit Root test

This section will examine whether the variables are stationary or not stationary. When analyzing time series data, it is essential to use stationary variables. The reason is that when the variables are not stationary, there is a possibility to compute large R-square. R-square represents how many percentages of variations of the dependent variable can be explained by the independent variables in the whole model. Therefore, the ADF unit root test will be implemented to check if the variables are stationary. The idea behind testing the unit root is that if the absolute value of the test statistics is lower than the 5% critical value, then the test fails to reject the null hypothesis and thus the variables are not stationary. The first test examines the logarithmic variables by testing the following hypothesis.

The following hypothesis tested: 𝐻0: Unit root (not stationary) 𝐻1: Not unit root (stationary) Table 5: Unit root test

Variables Test Statistics 5% Critical Value

LAEX -2.382 -2.873 LINT -1.155 -2.873 LEXCH -2.042 -2.873 LIND -2.176 -2.873 LCPI -0.862 -2.873 LM2 -1.757 -2.873

This table represents unit root test results at the logarithmic variables level.

Table 6: Unit root test

Variables Test Statistics 5% Critical Value

DLINT -18.671* -2.873 DLEXCH -12.413* -2.873 DLIND -20.266* -2.873 DLCPI -31.885* -2.873 DLM2 -15.669* -2.873 DLINT -21.509* -2.873

This table represents the unit root test results at the first logarithmic differences variables level. *represents significance at the 5% critical value for one-sided test. All variables are defined in section 3.1.

(21)

Since every variable has an absolute value of the test statistics lower than the absolute valued of 5% critical value in table 5. There is statistical evidence to infer that this test fails to reject the null hypothesis. Then, it can be concluded that all the variables are unit root or not stationary because it is problematic when dealing with a time series. Therefore, to analyze the model better, the first

differences of each variable should be used.

The first differences of every logarithmic variable have an absolute test statistics value larger than the 5% critical values in table 6. Therefore, it can be concluded that there is statistical evidence to reject null hypothesis and thus variables are stationary.

4.2 Multicollinearity

Multicollinearity arises when the independent variables are highly correlated with each other. Multicollinearity is problematic because it increases the variance of the coefficient estimates and makes them very sensitive to small changes in the model. Multicollinearity leads to biased standard errors of the fitted coefficients upwards. Therefore, it results in a downward biased student t-statistics and creates a problematic situation when evaluating a regression model. The most extreme version of multicollinearity is when the significant t statistics is not reached but the OLS concludes a significant overall regression relationship.

The method to check for multicollinearity is analyzing the correlation coefficients or by computing the Variance Inflation Factors (VIF). The correlation coefficients represent how the independent variables are correlated with each other in table 6. Usually, coefficients higher than 90% is a sign of multicollinearity. The VIF test tells that to which extent the variance of a regression coefficient is inflated due to the multicollinearity in the model. The rule of thumb is that if the VIF equal to five or more, then there is multicollinearity. Given the values both in table 7 and 8, it can be concluded that there is no statistical evidence for multicollinearity between the independent variables in the model.

Table 7: Correlation Coefficients

Variables DLINT DLEXCH DLIND DLCPI DLM2

DLINT 1.0000

DLEXCH -0.0405* 1

DLIND 0.0268* 0.0077* 1

DLCPI -0.0180* 0.0591* -0.0434* 1

DLM2 -0.0068* -0.0461* 0.0298* 0.0087* 1

This table represents correlation coefficients between the independent variables for multicollinearity check. *represents correlation less than 90% and thus no multicollinearity. All variables are defined in section 3.1.

(22)

Table 8: Variance Inflation Factors (VIF) Variables VIF DLEXCH 1.01* DLCPI 1.01* DLIND 1.00* DLM2 1.00* DLINT 1.00*

This table represents the Variance Inflation Factors. *represents. VIF < 5 and thus no multicollinearity. All variables are defined in section 3.1.

4.3 Regression Analysis

Since both the unit root and the multicollinearity tests confirmed that the data is stationary and there is no high correlation between the independent variables, then the OLS regression analysis using the model stated in the methodology section can be examined. The purpose of the regression analysis is to test which independent variables are related to the AEX price and in which direction. Table 9: OLS regression analysis

Variables (AEX) Coefficients (Std. Error) T-statistics P-value Constant 0.0018 (0.0013) 1.33 0.185 DLINT -0.0058 (0.0205) -0.28 0.776 DLEXCH 0.3586 (0.0914) 3.92 0.000* DLIND 0.0614 (0.1050) 0.58 0.559 DLCPI -0.1237 (0.6528) -0.19 0.850 DLM2 0.3464 (0.2186) 1.58 0.114 R-square 0.0416 Number of obs 421

This table presents the OLS regression analysis results of the model. *represents significance at the 5% level for one-sided test. All variables are defined in section 3.1.

(23)

Multiple regression analysis outputs presented in table 9. It is important to note that all the variables are transformed into logarithms, and the first differences were taken. The reason for that is to make the trend linear, the discrepancies between variables less apparent and make the data more stationary as discussed previously. Model in this paper is a log to log model therefore 1% change in the independent variables will change the dependent variable by β%. The p-value is the probability of the null hypothesis being rejected or fail to reject using 5% significance level in this paper.

The exchange rate coefficient 0.3586 indicates that 1% increase (depreciation) of the

exchange rate will shift the AEX price index up by 0.3586%. The p-value is less than 5% significance level. Therefore, it can be concluded that there is statistical evidence to reject the null hypothesis. The exchange rate positively and significantly related to the AEX price index. The rest of the variable coefficients, the interest rate (-0.0058), industrial production (0.0614), the CPI (-0.1237) and the money supply (0.3464) did match the hypothesis expectations, but they are not statistically significant to reject the null hypothesis. Therefore, we cannot conclude that the changes in those variables are subject to the AEX index price movements. Table 10 summarizes the OLS regression analysis. Table 10: Summary of the OLS regression analysis

Variables Match the Hypothesis Significant

DLINT YES NO

DLEXCH YES YES

DLIND YES NO

DLCPI YES NO

DLM2 YES NO

All variables are defined in section 3.1.

4.4 Residual tests

There are three residual tests need to reject the null hypothesis which are the OLS

assumptions in order to perform the OLS regression analysis correctly. Firstly, the correlogram for the residuals or the white noise test to check whether the residuals are white noise. Next,

heteroscedasticity test to confirm the absence of heteroscedasticity. Third, normality test to check if the residuals are normally distributed.

4.4.1 Correlogram for Residuals

The white noise test of the residuals needs to be derived from the regression model. This test checks whether the variables are white noise meaning that there is no serial correlation, and the mean of the residuals are zero. Also, the residual variables are random, independent and stationary. Given the white noise test results in table 14 (see appendix) it can be concluded that there is no statistical evidence to reject the null hypothesis. Therefore, we can conclude that there is white noise in the

(24)

residuals meaning that no serial correlation, the mean of residuals are zero and the residual variables are random, independent, and stationary.

The following hypothesis tested: 𝐻0: white noise

𝐻1: no white noise

4.4.2 Heteroscedasticity test

Another test on the residuals will be heteroskedasticity test. Heteroskedasticity is a serious problem linked to the OLS regression because the OLS regression assigns equal weights to all observations. This leads to larger standard errors for certain observations to be biased. The following hypothesis tested:

𝐻0: No Heteroscedasticity 𝐻1: Heteroscedasticity

Table 11: Heteroscedasticity test

Chi-square 0.46

Prob> Chi-square 0.4983 This table presents the Heteroscedasticity test result of the model.

4.4.3 Normality test

Jarque-Bera Normality test is used to check whether the data set is well modeled, and the residuals are normally distributed. It is a more critical test when the data size is less than 30 observations, but it can be applied to any sample size.

The following hypothesis tested: 𝐻0: Normally distributed 𝐻1: Not normally distributed

Table 12: Jarque-Bera Normality test

Chi-square 1457

Probability 0.000*

This table presents the Jarque-Bera normality test result of the model. *represents significance at the 5% level for one-sided test.

From figure 2 (see appendix) and table 12, it can be concluded that the p-value is less than 5% significance level. Therefore, there is statistical evidence to reject the null hypothesis and thus the

(25)

residuals do fulfill the normality assumption. However, we can rely on the regression analysis since the sample size is larger than 30.

4.5 Granger Causality Test

Granger (1988) developed Granger Causality analysis to test whether one-time series is significant in forecasting another. In order to find a causal relationship between the macroeconomic variables and the AEX price index, Granger Causality test using two variable lag level will be tested. A critical assumption of the Granger Causality test is stationary data assumption. This assumption must hold in order to continue with the Granger Causality test. It was concluded in section 4.1 that the model is stationary through unit root test.

The following hypothesis tested:

𝐻0: X(independent) does not granger cause Y(dependent)/ Y(dependent) does not granger cause X(independent)

𝐻1: X(independent) does granger cause Y(dependent)/ Y(dependent) does granger cause X(independent)

Table13. Granger Causality test results

Null Hypothesis P-value

DLINT does not Granger Cause DLAEX DLAEX does not Granger Cause DLINT

0.8648 0.0596

DLEXCH does not Granger Cause DLAEX DLAEX does not Granger Cause DLEXCH

0.0288* 0.1908

DLIND does not Granger Cause DLAEX DLAEX does not Granger Cause DLINT

0.0665 0.7814

DLCPI does not Granger Cause DLAEX DLAEX does not Granger Cause DLCPI

0.2469 0.6925

DLM2 does not Granger Cause DLAEX DLAEX does not Granger Cause DLM2

0.0044* 0.5961

This table presents the Granger Causality analysis of the model. *represents significance at the 5% level for one-sided test. All variables are defined in section 3.1.

From table 13 it can be concluded that taking two variable lag level, there is statistical evidence to infer that the exchange rate and the money supply are significant in forecasting the AEX

(26)

price index. The p-value of 0.0288 and 0.0045 are both less than 5% significance level. Therefore the exchange rate and the money supply are statistically significant at two variable lag level to forecast the AEX price index. Rest of the variables fails to reject the null hypothesis and thus they are not statistically significant in forecasting the AEX price index. Additionally, the stock prices have no statistically significant effect to predict any macroeconomic variable used in this paper.

(27)

Chapter 5: Conclusions

5.1 Main findings

The main purpose of this paper was to answer the following research question. Can the price movements of the Amsterdam stock exchange price index be predicted using macroeconomic variables? Chapter 2 introduced theories to predict the stock price movements. According to the EMH, stock prices are not predictable based on information available neither publicly nor privately. The TA was the next tool which was discussed in this paper. The TA analysis stated that the stock price movements can be predicted if the investors have excellent skills to examine graphs and charts of the past stock prices thus it is a subjective method. The next sub-question discussed which macroeconomic factors affect the stock prices based on the previous researches. Five mostly used macroeconomic variables were conducted, and many opinions introduced on how these variables influence the stock price movements.

Chapter 3 and 4 examined how the interest rate, the exchange rate, industrial production, the CPI, and the money supply can predict the stock price movements. The APT used to test the

relationship between the independent and the dependent variables. Two main econometric analysis, the OLS regression analysis and the Granger Causality tests are implemented. Firstly all the variables converted to the logarithmic form for scaling purposes. Next, unit root analysis was conducted on each variable to check whether the data is stationary. This test is essential when dealing with the time series data. Also, when applying the Granger Causality test to the model, the data must be stationary. However, this test failed to reject the null hypothesis with the logarithmic variables. Therefore, the first differences of the logarithmic variables were computed to make the data more stationary, and the unit root test statistically approved that the first logarithmic differences were stationary. Next,

multicollinearity test using correlation coefficients and VIF test conducted. Both tests confirmed the absence of multicollinearity.

The OLS regression analysis results found that there is a positive relationship between the euro-dollar exchange parity and the AEX price index. This relationship was positive and statistically significant at 5% level. This can be due to the domestic exporters benefit from depreciation of the local currency because demand for domestic goods by the foreign consumers increase and thus increases the cash flows and the stock prices. Rest of the variable coefficients met the expectations but failed to reject the null hypothesis and thus were statistically insignificant.

Following the OLS regression analysis, three residual tests on the model implemented to check the fit of the model. First white noise test rejected null hypothesis which confirms that the residuals are mean zero, no serial correlation, independent, and stationary. The second test was heteroscedasticity test which also rejected the null hypothesis meaning that the residuals were homoscedastic thus fit of the model is statistically good. Lastly, normality test failed to reject the null hypothesis, but this is not a critical issue as the sample size were large enough.

(28)

The final test conducted in this paper was Granger Causality test in order to check whether one-time series is significant in forecasting another. The result of the test concluded that at two variable lag level the exchange rate and the money supply are significant in forecasting the AEX price index.

Overall, based on the OLS analysis conducted throughout the research it can be concluded that one out of five selected macroeconomic factors, namely, the exchange rate have a significant predictable effect on the stock prices movements of the AEX price index. Additionally, Granger causality affirms that the exchange rate and the money supply can be used to forecast the stock price movements at two variable lag level.

5.2 Limitations

Limitations of this paper were as follow: Availability of the data, the difficulty of measuring psychological factors, and the non-stationary relationship between the variables. These limitations results in less conclusive outcomes of this paper.

The first limitation is that stock prices released daily, but in this paper monthly stock prices were examined. The reason for the use of the monthly data is that there is no available daily

macroeconomic data. Macroeconomic data is usually released quarterly or yearly. Therefore, GDP could not be used which could probably give more significant results. Additionally, when discounting the PV of the cash flows, the best estimate is the money market rate and not the risk-free interest rate but this data was not available for the chosen time period. Also, the data size of 422 observations for the AEX price index is not large compared to the other big stock exchanges such as NYSE.

Another limitation is measuring the psychological factors. The psychological factors can have a significant impact on the stock prices, but they are difficult to measure because every investor observes information and makes decisions differently when it comes to the investment opportunities. Lastly, because the data was non-stationary and non-linear, logarithms and their first differences had to be taken to perform better OLS and Granger Causality tests. If the data was stationary, then the different outcomes could be concluded. Additionally taking the first logarithmic differences eliminates the multicollinearity issue which could possibly be the case between the CPI and the money supply.

5.3 Further research

To sum up, the results obtained in this paper should not be decisive for future investment decisions. Further research can be done with a more accurate model to examine the effect of

macroeconomic variables on the stock prices using daily data. In contrast, using more macroeconomic variables that might affect the stock price movements such as the unemployment rate, the foreign exchange reserves, the world oil prices, the international trade and other similar factors. Also, the specific factors could be examined to find the stock price movements. In general, the firm-specific microeconomic variables have more impact on the stock prices than the macroeconomic

(29)

variables because changes in the firm-specific variables can be diversified by the investors when considering investment opportunities.

(30)

REFERENCES

Bachelier, L. (1900). Théorie de la spéculation. Gauthier-Villars.

Balvers, R. J., Cosimano, T. F., & McDonald, B. (1990). Predicting stock returns in an efficient market. The Journal of Finance, 45(4), 1109-1128.

Basu, S. (1977). Investment performance of common stocks in relation to their price-earnings ratios: A test of the efficient market hypothesis. The journal of Finance, 32(3), 663-682.

Berk, J. B., & DeMarzo, P. M. (2007). Corporate finance. Pearson Education.

Bilson, C. M., Brailsford, T. J., & Hooper, V. J. (2001). Selecting macroeconomic variables as explanatory factors of emerging stock market returns. Pacific-Basin Finance Journal, 9(4), 401-426.

Bodie, Z., Kane, A., & Marcus, A. J. (2011). Investment and portfolio management. McGraw-Hill Irwin.

Brock, W., Lakonishok, J., &LeBaron, B. (1992). Simple technical trading rules and the stochastic properties of stock returns. The Journal of finance, 47(5), 1731-1764.

Campbell, J. Y., &Yogo, M. (2006). Efficient tests of stock return predictability. Journal of financial

economics, 81(1), 27-60.

Chan, L. K., Karceski, J., &Lakonishok, J. (1998). The risk and return from factors. Journal of

financial and quantitative analysis, 33(2), 159-188.

Chen, N. F., Roll, R., & Ross, S. A. (1986). Economic forces and the stock market. Journal of business, 383-403.

Cooper, J. C. (1982). World stock markets: some random walk tests. Applied Economics, 14(5), 515-531.

Doong, S. C., Yang, S. Y., & Wang, A. T. (2005). The dynamic relationship and pricing of stocks and exchange rates: Empirical evidence from Asian emerging markets. Journal of American

Academy of Business, 7(1), 118-123.

Durham, J. B. (2001). The effect of monetary policy on monthly and quarterly stock market returns. Fama, E. F. (1965). The behavior of stock-market prices. The journal of Business, 38(1), 34-105. Fama, E. F. (1970). Efficient capital markets: A review of theory and empirical work. The journal of

Finance, 25(2), 383-417.

Fama, E. F. (1995). Random walks in stock market prices. Financial analysts journal, 51(1), 75-80. Fama, E. F., & Schwert, G. W. (1977). Asset returns and inflation. Journal of financial economics,

5(2), 115-146.

Firth, M. (1979). The relationship between stock market returns and rates of inflation. The Journal of Finance, 34(3), 743-749.

Flannery, M. J., & Protopapadakis, A. A. (2002). Macroeconomic factors do influence aggregate stock returns. The review of financial studies, 15(3), 751-782.

(31)

Franck, P., & Young, A. (1972). Stock price reaction of multinational firms to exchange realignments. Financial Management, 66-73.

Geetha, C., Mohidin, R., Chandran, V. V., & Chong, V. (2011). The relationship between inflation and stock market: Evidence from Malaysia, United States and China. International journal of economics and management sciences, 1(2), 1-16.

Hamburger Michael, J., & Kochin, L. A. (1971). Money and Stock Prices: The Channels of Infuences. The journal of Finance, 27, 231-249.

Han, Y., Yang, K., & Zhou, G. (2013). A new anomaly: The cross-sectional profitability of technical analysis. Journal of Financial and Quantitative Analysis, 48(5), 1433-1461.

Homa, K. E., & Jaffee, D. M. (1971). The supply of money and common stock prices. The Journal of

Finance, 26(5), 1045-1066.

Hong, H., Torous, W., &Valkanov, R. (2007). Do industries lead stock markets?. Journal of Financial

Economics, 83(2), 367-396.

Humpe, A., & Macmillan, P. (2009). Can macroeconomic variables explain long-term stock market movements? A comparison of the US and Japan. Applied Financial Economics, 19(2), 111-119.

Jensen, M. C., &Benington, G. A. (1970). Random walks and technical theories: Some additional evidence. The Journal of Finance, 25(2), 469-482.

Jones, C. M., & Kaul, G. (1996). Oil and the stock markets. The Journal of Finance, 51(2), 463-491. Kandel, S., & Stambaugh, R. F. (1996). On the predictability of stock returns: an asset-allocation

perspective. The Journal of Finance, 51(2), 385-424.

Keim, D. B., & Stambaugh, R. F. (1986). Predicting returns in the stock and bond market Journal of

financial Economics, 17(2), 357-390.

Laffont, J. J., &Maskin, E. S. (1990). The efficient market hypothesis and insider trading on the stock market. Journal of Political Economy, 98(1), 70-93.

Lewellen, J. (2004). Predicting returns with financial ratios. Journal of Financial Economics, 74(2), 209-235.

Lim, K. P., Brooks, R. D., & Kim, J. H. (2008). Financial crisis and stock market efficiency:

Empirical evidence from Asian countries. International Review of Financial Analysis, 17(3), 571-591.

Lo, A. W., Mamaysky, H., & Wang, J. (2000). Foundations of technical analysis: Computational algorithms, statistical inference, and empirical implementation. The journal of finance, 55(4), 1705-1765.

Ma, C. K., & Kao, G. W. (1990). On exchange rate changes and stock price reactions. Journal of

(32)

Mahedi, M. (2012). Impact of the macroeconomic variables on the stock market returns: The case of Germany and the United Kingdom. Global Journal of Management and Business

Research, 12(16).

Malkiel, B. G. (2005). Reflections on the efficient market hypothesis: 30 years later. Financial Review, 40(1), 1-9.

Morck, R., Yeung, B., & Yu, W. (2000). The information content of stock markets: why do emerging markets have synchronous stock price movements?. Journal of financial economics, 58(1-2), 215-260.

Murphy, J. J. (1999). Technical analysis of the financial markets: A comprehensive guide to trading

methods and applications. Penguin.

Nasseh, A., & Strauss, J. (2000). Stock prices and domestic and international macroeconomic activity: a cointegration approach. The Quarterly Review of Economics and Finance, 40(2), 229-245. Palmer, M. (1970). Money supply, portfolio adjustments and stock prices. Financial Analysts

Journal, 26(4), 19-22.

Parks, R. W. (1978). Inflation and relative price variability. Journal of Political Economy, 86(1), 79-95.

Pilbeam, K. (2013). International finance. 4th ed. Basingstoke: Palgrave MacMillan.

Pontiff, J., &Schall, L. D. (1998). Book-to-market ratios as predictors of market returns. Journal of

Financial Economics, 49(2), 141-160.

Qian, B., & Rasheed, K. (2007). Stock market prediction with multiple classifiers. Applied Intelligence, 26(1), 25-33.

Ratanapakorn, O., & Sharma, S. C. (2007). Dynamic analysis between the US stock returns and the macroeconomic variables. Applied Financial Economics, 17(5), 369-377.

Sadorsky, P. (1999). Oil price shocks and stock market activity. Energy economics, 21(5), 449-469. Sehgal, S., & Gupta, M. (2007). Tests of technical analysis in India. Vision, 11(3), 11-23.

Solnik, B. H. (1973). Note on the validity of the random walk for European stock prices. The journal

of Finance, 28(5), 1151-1159.

Timmermann, A., & Granger, C. W. (2004). Efficient market hypothesis and forecasting. International Journal of forecasting, 20(1), 15-27.

(33)

APPENDIX

Figure 1: AEX Price Index (1983-2018)

Figure 2: Histogram Normality 0 100 200 300 400 500 600 700 800 1/ 1/ 83 4/ 1/ 84 7/ 1/ 85 10/ 1/ 86 1/ 1/ 88 4/ 1/ 89 7/ 1/ 90 10/ 1/ 91 1/ 1/ 93 4/ 1/ 94 7/ 1/ 95 10/ 1/ 96 1/ 1/ 98 4/ 1/ 99 7/ 1/ 00 10/ 1/ 01 1/ 1/ 03 4/ 1/ 04 7/ 1/ 05 10/ 1/ 06 1/ 1/ 08 4/ 1/ 09 7/ 1/ 10 10/ 1/ 11 1/ 1/ 13 4/ 1/ 14 7/ 1/ 15 10/ 1/ 16 1/ 1/ 18

AEX Price Index

0 5 10 15 20 D e n si ty -.2 -.15 -.1 -.05 0 .05 Residuals

(34)

Table 14: Correlogram for residuals – White noise test . 40 -0.0003 0.0155 45.295 0.2606 39 -0.0773 -0.0577 45.295 0.2259 38 -0.0412 -0.0994 42.511 0.2829 37 -0.0153 0.0013 41.72 0.2730 36 -0.0050 -0.0140 41.612 0.2396 35 0.1085 0.0858 41.6 0.2053 34 0.0490 0.0655 36.172 0.3675 33 0.0361 0.0568 35.067 0.3703 32 0.0264 0.0030 34.469 0.3505 31 -0.0799 -0.0700 34.15 0.3187 30 0.0792 0.0598 31.234 0.4040 29 0.0036 0.0442 28.378 0.4978 28 -0.0529 -0.0646 28.373 0.4448 27 0.0326 0.0204 27.104 0.4582 26 -0.0336 -0.0155 26.623 0.4293 25 -0.0408 -0.0466 26.113 0.4016 24 0.0014 0.0198 25.364 0.3862 23 0.0020 -0.0179 25.363 0.3319 22 0.0093 0.0338 25.361 0.2801 21 -0.0219 -0.0625 25.322 0.2335 20 -0.0240 0.0149 25.108 0.1973 19 0.0095 -0.0253 24.853 0.1654 18 0.0522 0.0939 24.813 0.1302 17 -0.0332 -0.0785 23.609 0.1304 16 0.0204 0.0481 23.122 0.1105 15 -0.0799 -0.0702 22.939 0.0854 14 0.0329 0.0209 20.138 0.1259 13 0.0016 -0.0106 19.664 0.1039 12 0.0262 0.0279 19.663 0.0737 11 -0.0028 0.0013 19.364 0.0549 10 -0.0265 -0.0372 19.36 0.0359 9 0.0771 0.0841 19.057 0.0247 8 0.0688 0.0566 16.486 0.0359 7 -0.0898 -0.0620 14.444 0.0438 6 -0.0457 -0.0771 10.974 0.0892 5 0.0878 0.1052 10.077 0.0731 4 -0.0409 -0.0569 6.7777 0.1481 3 0.0903 0.0964 6.065 0.1085 2 -0.0343 -0.0395 2.5872 0.2743 1 0.0702 0.0703 2.0867 0.1486 LAG AC PAC Q Prob>Q [Autocorrelation] [Partial Autocor] -1 0 1 -1 0 1

Predictability of Stock price movements : evidence from the Amsterdam Stock Exchange