The role of emerging foreign markets in the price discovery of cross-listed stocks

(1)

THE ROLE OF EMERGING FOREIGN MARKETS IN THE PRICE

DISCOVERY OF CROSS-LISTED STOCKS

University of Amsterdam – Amsterdam Business School

Master in International Finance 2017

Master Thesis

Author: Haiping Liu

Student No.: 10679391

Supervisor: Rafael Matta

(2)

i Abstract

The study of price discovery in the stock market has gained significant research interests over the past decades. Though many research efforts have been made on this topic, the role of emerging markets has not been greatly investigated yet as most literature focused on the developed markets. This research study was peculiar because it investigated the role of emerging markets in the price discovery of stocks cross-listed in developed markets. The literature review shows that both domestic and foreign markets play significant roles in the price discovery process; however, in the case of cross-listed stock, the domestic markets tend to dominate the process. The study used the Vector Autoregression (VAR) methodology to investigate the role of emerging market. Data were collected on three Indian stocks cross-listed in the New York Stock Exchange. The result of the study shows that emerging markets determine price discovery of stocks cross-listed in developed markets.

Keywords: Stock Market, Price Discovery, Emerging Market, Developed Market, Cross Listing, Cross Listed Stocks, Share Price

(3)

ii

List of Tables

Table 1: Descriptive statistics summary of prices in NSE ... - 15 -

Table 2: Descriptive statistics summary of prices in NYSE ... - 16 -

Table 3: Correlation coefficients ... - 17 -

Table 4: Correlation coefficients ... - 20 -

Table 5: ADF unit root test on NSE prices ... - 21 -

Table 6: ADF unit root test on NYSE prices ... - 21 -

Table 7: Result of Johansen cointegration tests ... - 22 -

Table 8: Lag Order Selection Criteria test ... - 23 -

Table 9: Result of VAR model analysis ... - 24 -

Table 10: Root of Characteristic Polynomial ... - 27 -

(6)

v

List of Figures

Figure 1: Graph of USD/INR Exchange rates ... - 17 -

Figure 2: Graph of Wipro share price changes ... - 19 -

Figure 3: Response to Cholesky One S.D. Innovation ... - 25 -

(7)

vi

List of Equations

(8)

vii

List of Abbreviations and Acronyms

AIC: Akaike Information Criterion

BRICS: Brazil, Russia, India, China, and South Africa BSE: Bombay Stock Exchange

FPE: Final Prediction Error

HQ: Hannan-Quinn Information Criterion IEX: The Investors Exchange

INR: India Rupee

ISE: International Securities Exchange Holdings Incorporation

NASDAQ: National Association of Securities Dealers Automated Quotations NSE: National Stock Exchange

NYSE: New York Stock Exchange SC: Schwarz Information Criterion TSX: Toronto Stock Exchange U.K. The United Kingdom

U.S.A. The United States of America USD: United States Dollar

(9)

- 1 -

Section One: Introduction

1.1 Background of the Study

Under the globalization of financial markets, many firms choose to cross list their shares on foreign stock exchange markets. Cross listing is believed to be beneficial for firms, and some empirical evidence shows that cross-listed firms are rewarded with higher valuation, enhanced transparency, reduced agency costs, and pure cash flow effects among others (Lang, Lins, and Miller 2003, p. 319).

The prices of cross-listed stocks should be highly consistent under the fundamental pricing theory law of one price since they reflect the same asset; any price deviation should be eliminated by arbitrage activities in the efficient markets. A explained by Ansotegui, Bassiouny, and Tooma (2013, p. 4), price discovery is the process whereby the market attains equilibrium prices. It is through this process that leads to the determination of spot prices of stocks in the equity market. Ansotegui, Bassiouny, and Tooma (2013, p. 4) further explained that price discovery is an essential process in the stock exchange.

On the other hand, the cross listing is defined as the process whereby companies list their equities in one or more foreign stocks exchanges in addition to the domestic one. Cross listing is believed to have numerous advantages including the opportunity to trade the company’s shares in multiple time zones and multiple currencies. With this, the listing company benefits from greater ability to raise capital as well as higher level of liquidity. Many researchers including Ansotegui, Bassiouny, and Tooma (2013, p. 4), Chouinard and D'Souza (2004, p. 1) and Frijns, Gilbert, and Tourani-Rad (2007, p. 4) have documented the dramatic increase of stock cross listing over the past recent years. Besides, Chouinard, and D'Souza (2004, p. 4) asserted that cross listing enables local companies to access broader investor base as well as increase the marketability of the security.

(10)

- 2 - In most cases, stocks from emerging markets are cross-listed in well-established foreign markets. Even though previous researchers have extensively studied the dynamic of price discovery for cross-listed companies, little efforts have been put on emerging markets, and as a continuation, this thesis examined the price determination process over domestic emerging market and foreign market. This paper was designed to provide insights into the role of emerging markets in the price discovery of stocks cross-listed in well-established foreign markets.

1.2 Statement of the Problem

The phenomenon of stock cross listing has been on the increase, and that trend is anticipated to continue in the coming years due to the perceived benefits. As explained by Ansotegui, Bassiouny, and Tooma (2013, p. 4) and Claessens, Klingebiel, and Schmukler (2002, p. 1), most companies in the emerging markets choose to cross list their stocks in large and well-established international markets. However, not many researchers focus how the domestic market influences the price discovery in the foreign markets. Unfortunately, most research studies have focused only on developed markets, and little emphasis on emerging ones was done. Furthermore, the possibility of emerging markets dominating the price discovery in well-established foreign markets has not been investigated, which calls for the need to understand the contribution of emerging markets in well-established foreign markets.

1.3 Research Objectives

This paper was guided by the following three major research objectives:

1) To determine whether domestic emerging markets dominate price strategy of cross-listed stocks in foreign markets.

2) To determine whether emerging markets influence price discovery of stocks cross-listed in well-established foreign markets.

(11)

- 3 - 3) To assess the contributions of the domestic emerging markets in the price discovery at the

foreign markets.

1.4 Research Questions

This paper was guided by the following three major research questions:

1) To what extent does domestic emerging market dominate price strategy of cross-listed stocks in a foreign market?

2) Does emerging markets influence on the price discovery of stocks cross-listed in well-established foreign markets?

3) What are the contributions of the domestic emerging markets in the price discovery at the foreign markets?

1.5 Research Hypotheses

The research hypotheses were developed based on the theories and findings of the literature review. Two hypotheses were developed concerning the two research questions outlined in section 1.4 above.

Hypothesis 1: The local emerging market is dominant in price discovery of cross-listed firms in foreign established markets.

In this case, the research was designed to study the influence of domestic emerging markets in the price discovery of stocks cross-listed in established foreign markets. In essences, it was designed to determine whether prices in the local market influence prices in the foreign market in the price discovery process.

Hypothesis 2: Emerging markets have a significant influence on the price discovery of domestic stocks cross-listed in the foreign markets.

(12)

- 4 - With a highly developed internet connection, the sharing of information is height effective. The developments in the emerging markets would influence the price discovery of domestic stocks cross-listed in foreign markets.

1.6 Scope of the Study

The study was conducted in one emerging market and one well-established market namely India and the United States of America respectively. India was chosen because it is one of the rapidly growing and most aggressive emerging markets in the world today. India has two major stock exchanges, the National Stock Exchange (NSE) and the Bombay Stock Exchange (BSE), where trading of all Indian stocks takes place. They follow the same trading mechanism, settlement process, and trading hours and these two exchanges also cross-list some foreign stocks. National Stock Exchange is chosen in this study as a representative of the emerging markets.

The United States of America represents one of the most advanced stock markets in the market. It has some leading exchanges such as Chicago Stock Exchange (CSE), Miami International Securities Exchange (MISE), New York Stock Exchange (NYSE), and NASDAQ among others. In this study, the New York Stock Exchange is chosen to represent the developed markets.

Under this scope, three Indian companies which are listed in both New York Stock Exchange and National Stock Exchange are selected as research objects: HDFC Bank, Tata Motors Limited, and Wipro Limited.

1.7 Significance of the Study

This research would contribute to a proper knowledge and understanding of where price discovery occurs in the stock market. This information is essential to academicians, researchers as well as investors and policy makers. This paper attempts to contribute to the proper understanding of the role of emerging markets in the price discovery process, which is lacking from existing studies.

(13)

- 5 - Another significance of this research study is that it leads to a proper understanding on how information is priced in the stock market. In particular, it focuses on the significance of information from the domestic emerging markets and their usefulness in determining stock prices. In addition, the paper is essential because it provides a broader framework for propagating the mechanism of price discovery in foreign markets. This paper would also bring attention on of the roles of emerging financial markets in the global market. It highlights on the mechanism of price transmission from emerging markets to foreign markets.

To sum it up, the main contribution of this study is that it provides clear empirical evidence of the influence and role of emerging markets in the price discovery in the well-established foreign markets.

1.8 Overview of the Study

The study is organized into seven sections. Section One presents an introduction to the study by discussing the research background, the statement of the problem, research questions, hypothesis, as well as significance and scope of the study. Section Two presents the literature review, where related literature was reviewed to highlight the theoretical and philosophical underpinning of the study. Section Three presents and discusses the methodological approaches used to conduct the study. Section Four discussed the data and descriptive statistics while Section Five presents the data analysis results. Lastly, Section Six provides the robustness check while the last section presents the discussion and conclusions that are derived from the results of the research study.

(14)

- 6 -

Section Two: Literature Review

2.1 Introduction

This chapter presents a literature review that was conducted on the topic of price discovery in the stock market. The literature review focuses on identifying the philosophical and theoretical underpinning of the phenomenon being studied. In addition, the focus is on identifying what other researchers and writers have established about the role of emerging markets in the price discovery cross-listed stocks in the foreign markets. Literature from trusted sources such as journals, peer-reviewed articles, and books was reviewed in regards of the topics, background on the cross listing, mechanisms of price discovery, and the influence of emerging markets on price discovery.

2.2 Background on Cross Listing

There are numerous pieces of literature about cross listing and price discovery in the stock market; however, the majority focuses on the United States and United Kingdom markets (Feng and Stewart 2016, p. 22). Such pieces of literature have discussed various motivations for cross listing, especially in the developed markets. Feng and Stewart (2016, p. 22) asserted that cross listing is a common phenomenon because it is hypothesized to add value to the firm. Another hypothesis, the information environment hypothesis, states that the requirement for information disclosure is more stringent when stocks are cross-listed in foreign developed countries than in their domestic markets.

It is a common practice globally for companies from developing markets to cross list their stocks in developed markets where there is good governance, high level of accountability as well as good information standards (Feng and Stewart 2016, p. 22). This helps companies from countries with poor governance to improve their access to capital, to increase firm value and to lower the cost of capital. They further explained that another main reason for cross listing in developed market is the need to eliminate the existing market segmentation. Cross border barriers such as information

(15)

- 7 - problems and regulatory restrictions often segment the global market. According to Feng and Stewart (2016, p. 22), firms that are based in emerging markets often derive higher returns when their securities are cross-listed in developed markets such as U.S.A. and U.K.

2.3 Mechanisms of Price Discovery

A number of existing literature illustrates the mechanism of price discovery, especially for cross-listed stocks. As explained by Chen, Choi, and Hong (2013, p. 2), price discovery is the primary method by which the market prices information. It is important to understand where and how price discovery occurs especially when securities are a trade in multiple markets.

As explained by Ansotegui, Bassiouny, and Tooma (2013, p. 6), both domestic and foreign market directly contribute to price discovery of cross-listed stocks; however, the domestic market tends to dominate the process. They asserted that the ‘informative share’ of each market is the most important factor that contributes to price discovery of stocks. They defined informative share as the proportion of long-term variations in stock returns explained by each market. They further explained that the most important aspect of price discovery is the adjustment of prices caused by the flow of information across the markets.

2.4 The Influence of Emerging Markets on Price Discovery

Previous studies revealed the influence of developed markets on the price discovery of cross-listed stocks; however, only a few pieces have discussed the influence of emerging markets. According to Chen, Choi, and Hong (2013, p. 2), a research study by Hasbrouck (1995) showed that developed markets such as NYSE play a leading role in the price discovery and information share of cross-listed stocks. They further explained the result of a research study by Bacidore and Sofianos (2002) showing that for international cross-listing price discovery primarily occurs in the domestic

(16)

- 8 - markets. This provides hints to a possible influence of emerging markets especially of their stocks are cross-listed in international markets.

However, in another empirical study based on Canadian stocks cross-listed in United States market, Eun and Sabherwal (2003, p. 552) found that the Toronto and United States prices are adjusted simultaneously to restore equality in the market; they explained that these adjustments in maintaining prices occur on both exchanges. Their findings indicate that even though the foreign market (U.S.A) dominated the price discovery process, the domestic market (Toronto) also had a significant contribution (Eun and Sabherwal 2003, p. 552), which further affirms that foreign markets have a significant influence on the price discovery process. Furthermore, an empirical study by Solnik, Boucrelle, and Le Fur (1996, p.17) showed that most price discovery process occurs in the domestic market.

Chen, Choi, and Hong (2013, p. 4) explained that price discovery in cross-listed stocks might be affected by the differential adverse selection risks of the cross-border as well as various agency walls. Using the Canadian listings on Toronto Stock Exchange (TSX), they demonstrated that domestic stocks suffer more from information asymmetry as compared to cross-listed stocks, which eventually affect their price discovery.

In a study of the dynamic price transmissions in BRICS stock markets, Visalakshmi and Lakshmi (2016, p. 77) found an existence of high level of dependence structure between the BRICS stock markets and the global market. In essence, they argued that the structural dependence is information asymmetric which directly affects the price discovery process in the global market. Further, they illustrated that there is a remarkable co-movement between BRICS and well-developed stock markets such as the United States of America. This is a clear indication of the possible influence of emerging markets on price discovery. Their study further suggested that the level of stock market co-movement is dependent on regional factors (Visalakshmi and Lakshmi 2016, p. 78).

(17)

- 9 - In the short run, various destabilizing factors in the emerging market cause varying levels of vulnerability in the foreign market, which is eventually passed to their price discovery process (Visalakshmi and Lakshmi 2016, p. 87). This could lead to short-term speculation gains and arbitrage opportunities in the cross-listed stocks.

Similarly, Agarwal, Liu, and Rhee (2007, p. 2) explained that for cross-listed stocks, price discovery occurs in the domestic market but with significant information sharing between the two markets. This follows the result of a study of three German blue-chip stock cross-listed in New York Stock Exchange. They further explained that most domestic markets have the largest proportion of price discovery occurring in the foreign markets. In another similar empirical study, Kim, Szakmary, and Mathur (2000, p. 1380) showed that home market price is the most important factor of price discovery in the foreign market (Agarwal, Liu and Rhee 2007, p. 2). The research by Agarwal, Liu, and Rhee (2007, p. 5) provided empirical evidence showing that, for cross-listed stocks, closing prices in the domestic markets are fully incorporated in the opening prices in the foreign markets. They explained that trading in foreign markets has little explanatory power in explaining overnight price changes in cross-listed stocks. In essence, this study indicated that prices of cross-listed stocks in foreign markets could be largely explained by trading in the domestic markets.

Wang and Wu (2014, p.43) explained that emerging domestic market is a dominant force in the price discovery of cross-listed stocks. Their study followed the results of a research study by Yang et al (2005, p. 137) that used data from emerging markets like Taiwan, Korea, Singapore, as well as Hong Kong. In particular, the study showed that foreign markets (U.S.A.) simply adjust their prices to those of domestic markets. The results of this study provided a strong evidence of the influence of emerging markets in the price discovery of cross-listed stocks.

(18)

- 10 -

Section Three: Methodology

This chapter presents a discussion on the methods, tools, and techniques that were used to study the role of emerging markets in the price discovery of cross-listed stocks. The techniques, tools, and methods were carefully chosen to provide the best answers to the research questions illustrated in section 1.4. Data types, data source, methods of data analysis and model specification are discussed in the following parts.

3.2 Data Type

This part describes the sources and type of data that were used in the analysis. The sample consists of the intraday transaction of three Indian stocks listed on both NSE and NYSE. Three Indian companies were chosen to represent the emerging markets, and their performance and price discovery in India were compared with the results in United States market. The chosen three Indian companies are HDFC Bank, Tata Motors Limited, and Wipro Limited, from which data was collected about their intraday transactions; these data consisted of dates and corresponding open prices for each of the company in the two stock markets. The author chose to work with daily open prices to eliminate the intraday variations in prices that could have affected the result of data analysis.

3.3 Preliminary Analysis

Preliminary analysis of the collected data was conducted to identify some existing trends and relationships. The preliminary analysis was done to provide a pictorial comparison of price movements in the two stock exchanges to provide a glimpse of any possible influence of emerging markets in the price discovery of stocks cross-listed in developed markets.

(19)

- 11 - In addition to the graphical presentation, simple correlation analysis was carried out in the preliminary study, on the shares as traded on both stock exchanges. This is a useful statistical analysis tool in evaluating the strength of the relationship between two variables. Furthermore, it helped determine any possible connection between the two variables.

3.4 Data Analysis Methods

This part presents a brief description of the statistical methods that were used to analyze the data set. The study of the role of emerging foreign markets in the price discovery of cross-listed stocks was based on the use of time series analysis. Despite the differences in trading locations and the base currency, the cross-listed stocks represent the same firm equity value. According to the law of one piece, the market value of stocks of the same firm should be identical; hence trend of price movement on both stock exchanges.

Furthermore, because these firms are home-based in emerging market, the fundamental of the emerging market should significantly influence prices in the foreign stock markets. Due to the systematic relationship with the emerging market, the author expected remarkable co-movements of stock prices in the two markets.

The theoretical relationship in the price discovery process illustrated above can be tested empirically using co-integration analysis and Vector Autoregression (VAR) method.

In order to use the above technique, a prerequisite is that prices in the underlying emerging markets and developed markets are cointegrated. Besides, the short-term deviation of prices in developed markets should be corrected by adjustments of prices in the emerging markets. These two tests enable us to determine the role of emerging markets in the price discovery in developed markets.

(20)

- 12 - Vector Autoregression (VAR) model was used to test for the role of emerging foreign markets in the price discovery of cross-listed stocks. In this case, the lagged values can be interpreted as the price discovery in the developed market. A VAR of multiple time series can explain own lagged values of multiple variables and all other variables included in the model.

In summary, the methodology consisted of analyzing the error-correction mechanism between the two markets using co-cointegration and VAR model.

3.4.1 Tests for Stationarity and Cointegration

A necessary condition for using the error correction model is that the time series data is stationary and that long-run cointegration exists. Therefore, Unit Root Test was conducted to test for the stationarity. The test results indicated that the time series data was non-stationary as shown in Tables 5 and 6. Cointegration of the time series data was also tested, and the results can be found in Table 7.

3.4.2 Model specification

The study of the role of emerging foreign markets in the price discovery of cross-listed stocks used Vector Autoregression (VAR) model. This model is used to describe a set of endogenous variables as a linear combination of their past lagged values; here, the lag values were interoperated as the price discovery. This model is chosen due to its good forecasting capabilities. In addition, it is easy to use and does not require the researcher to specify the endogenous and exogenous variables. The following equation specified the model:

Equation 1: VAR model

𝑌𝑌𝑡𝑡 = 𝐶𝐶 + 𝛽𝛽1𝑌𝑌𝑡𝑡−1+ 𝛽𝛽2𝑌𝑌𝑡𝑡−2+ ⋯ + 𝛽𝛽𝑝𝑝𝑌𝑌𝑡𝑡−𝑝𝑝+ 𝑒𝑒𝑡𝑡

Where:

(21)

- 13 - 𝛽𝛽𝑖𝑖is a fixed vector coefficient;

C is a fixed vector intercept terms, and et is a fixed vector error term.

A key determinant in the role of emerging markets in the price discovery is the values of 𝛽𝛽𝑖𝑖. The

values of 𝛽𝛽𝑖𝑖 will be considerably small if the emerging market dominate price discovery, which is an

indication of the inability of the market to correct any differences with the developed market. However, high relative values between the two markets indicates that the former strong adjust errors in prices.

(22)

- 14 -

Section Four: Data and Descriptive Statistics

In this section, the sources and types of data used in the analysis are presented, as well as how the data was collected. Data type and the source were a very crucial consideration in this research study to ensure that the results are accurate, relevant, and justifiable. In addition, a summary of some key descriptive statistics to the data is also presented, such as mean, median, minimum value, maximum value, standard deviation as well as the number of observations.

4.2 Data Type and Source

In this research study, the author relied on the daily price data of selected Indian stocks cross-listed in New York Stock Exchange. The author chose to work with daily price data since it represents a significant element of price discovery; price discovery is expressed by changes in prices of the stock. Therefore, in general, the data consisted of daily open price from each stock from January 2, 2013, to August 4, 2017.

Data was collected from both the National Stock Exchange (NSE) of India and New York Stock Exchange (NYSE) of United States. The data was collected from Indian companies whose stocks are trading in both bourses. The three companies selected are HDFC Bank, Tata Motors Limited, and Wipro Limited. They were chosen to cut across major economic sectors in emerging markets. So, for each of these three companies, the daily open price data were collected from both NSE and NYSE stock exchanges.

The intraday foreign exchange rate was obtained for Indian Rupee (INR) to USA Dollar (USD) for the period between January 2, 2013, and August 4, 2017. The exchange rate data were then used to convert the daily open price of stocks trading in NSE to U.S. dollars to match units for those in NYSE. Investing.com provided the exchange rate data.

(23)

- 15 - 4.3 Summary of Statistics

This subsection is going to present key summary statistics of data that is for use in the analysis, as presented in tables, frequency distributions, and graphs. In addition to the descriptive statistics, this section also provides the cross-correlation table between the variables. Tables 1 and 2 below provide the summaries of descriptive statistics about the data.

Table 1: Descriptive statistics summary of prices in NSE

HDB TTM WIP

Mean 15.60697 6.692167 8.082214

Standard Error 0.116815 0.037105 0.036705

Median 15.58538 6.72855 8.255262

Mode #N/A #N/A #N/A

Standard Deviation 3.937191 1.250616 1.23712 Sample Variance 15.50147 1.56404 1.530465 Kurtosis 0.654292 -0.82288 1.683174 Skewness 0.804419 0.17621 -0.94188 Range 19.87511 5.745926 6.942316 Minimum 8.306428 4.104527 3.933932 Maximum 28.18153 9.850453 10.87625 Sum 17729.52 7602.302 9181.395 Count 1136 1136 1136

Source: Author’s Calculation

The above table provides a summary of key descriptive statistics for the daily open stock price of the three selected stocks as traded on the National Stock Exchange of India between January 2, 2013, and August 4, 2017.

(24)

- 16 - Table 2: Descriptive statistics summary of prices in NYSE

HDB TTM WIT Mean 54.80995 34.16563 5.540343 Standard Error 0.446471 0.209217 0.023397 Median 57.06 34.02 5.8 Mode 59.6 28.3 6 Standard Deviation 15.18658 7.116443 0.795847 Sample Variance 230.6321 50.64376 0.633373 Kurtosis -0.39962 -0.83363 -0.50696 Skewness 0.310752 0.321324 -0.56723 Range 71.94 30.64 3.565 Minimum 26.88 20.62 3.54 Maximum 98.82 51.26 7.105 Sum 63415.11 39529.63 6410.177 Count 1157 1157 1157

The above table provides a summary of key descriptive statistics for the daily open stock price of the three selected stocks as traded on the New York Stock Exchange of United States of America between January 2, 2013, and August 4, 2017.

(25)

- 17 - Figure 1: Graph of USD/INR Exchange rates

Source: Author’s Calculations

The graph above shows changes in the USD/INR exchange rates between January 2, 2013, and August 4, 2017. This graph shows that the U.S. dollar significantly gained against the Indian rupee between 2013 and 2017. This confirmed the need to convert the prices of NSE into U.S. dollar base.

4.4 The Cross-Correlation Analysis

The below table was constructed to provide the results of cross correlation analysis between stocks listed in New York Stock Exchange (NYSE) and the National Stock Exchange (NSE) of India. It is a useful statistical analysis tool in evaluating the strength of the relationship between two variables. The below table shows the cross correlation of prices of the selected three stocks listed in both NSE and NYSE stock markets. The graph shows a positive relationship between the two exchanges.

Table 3: Correlation coefficients

HDB TTM WIP

Correlation coefficient 0.973 0.927 0.682

56 58 60 62 64 66 68 70

II III IV I II III IV I II III IV I II III IV I II III

(26)

- 18 - This table shows the correlation coefficient between NSE and NYSE for each of the three stocks. We can see a high level of positive correlation between the three stocks. The high positive correlation coefficient is a strong indication of the co-movement of prices between the two stock exchanges.

(27)

- 19 -

Section Five: Data Analysis Results

This section presents the results of the data analysis. The data collected from the three stocks listed in both NYSE and NSE bourses were analyzed with statistical software EViews following the criteria established in Section Three. This chapter only presents a summarized version of the data analysis as well as interpretations and of the results of data analysis.

5.2 Preliminary Analysis

The results of the preliminary data analysis are shared and discussed in this part. In the preliminary analysis, the prices of each pair of cross listed stock were compared. The data is presented graphically below to provide a pictorial impression of the movements of prices in both stock exchanges and the result for listed company show a remarkable co-movement of stock prices between the two bourses. The figure 2 below shows the graph of prices of Wipro Limited at both NSE and NYSE.

Figure 2: Graph of Wipro share price changes

3 4 5 6 7 8 9 10 11

I II III IV I II III IV I II III IV I II III IV I II III

2013 2014 2015 2016 2017

NYSE NSE

(28)

- 20 - The figure above shows the co-movements of Wipro Limited share prices as listed in both NSE and NYSE stock markets; with eyeballing we can see a positive relationship between the two bourses.

Conducted correlation analysis was also conducted in the preliminary to provide a clear picture of the relationship between the two stock markets regarding share price movements of the selected stocks. The below table provides a summary of coefficients of correlation for each of the three stocks.

Table 4: Correlation coefficients

HDB TTM WIP

Correlation coefficient 0.973 0.927 0.682

Table 4 shows the correlation coefficient between NSE and NYSE for each of the three stocks. The table shows a high level of positive correlation for all the three stocks. The high positive correlation coefficient is a strong indication of the co-movement of prices between the two stock exchanges.

5.3 Unit Root Test

Unit Root Test is carried out using the augmented Dickey and Fuller (ADF) method, to test for the existence of a unit root in the time series data. This test was used to confirm whether the daily open prices for each stock have a unit root; the existence of a unit root is a confirmation of nonstationarity in the dataset. The result of the ADF unit root test is documented in Tables 5 and Table 6 below.

(29)

- 21 - Table 5: ADF unit root test on NSE prices

Table 5 shows the result of ADF Unit Root Test on prices of the three stocks listed NSE. In this test, the null hypothesis is that the time series data has a unit root, which is upheld if the probability value (p-value) is less than 5%. The p-value for Wipro Limited, Tata Motors Limited, and HDFC Bank are 99.98%, 24.45%, and 39.318% respectively, which are all greater than 5%; therefore, we fail to reject the null hypothesis and conclude that data is not stationary. Therefore the study is further processed with the first difference of the time series which is stationary.

Table 6: ADF unit root test on NYSE prices

Method Statistic Prob.**

Im, Pesaran and Shin W-stat 1.71190 0.9565

** Probabilities are computed assuming asympotic normality Intermediate ADF test results

Max

Series t-Stat Prob. E(t) E(Var) Lag Lag Obs

HDB 1.8300 0.9998 -1.530 0.745 1 22 1134 TTM -2.1009 0.2445 -1.532 0.735 0 22 1135 WIP -1.7753 0.3931 -1.532 0.735 0 22 1135 Average -0.6821 -1.531 0.738

Method Statistic Prob.**

Im, Pesaran and Shin W-stat 1.22757 0.8902

** Probabilities are computed assuming asympotic normality Intermediate ADF test results

Max

Series t-Stat Prob. E(t) E(Var) Lag Lag Obs

WIP -2.3027 0.1713 -1.532 0.735 0 22 1156 TTM -1.8422 0.3601 -1.530 0.745 1 22 1155 HDB 1.3778 0.9990 -1.532 0.735 0 22 1156

(30)

- 22 - Table 6 shows the result of ADF unit root test on prices of the three stocks listed NSE. In this test, the null hypothesis is that the time series data has a unit root, which is upheld if the probability value (p-value) is less than 5%. The p-value for Wipro Limited, Tata Motors Limited, and HDFC Bank are 17.1%, 36.01%, and 99.9% respectively, which are all greater than 5%; therefore, we fail to reject the null hypothesis and conclude that data is not stationary. To proceed with the model, I made the data stationary by getting the first difference.

5.4 Cointegration Analysis

Test results for cointegration of the time series is presented in this part. Testing for cointegration is essential as it forms part of my model as explained in section 3.4.1. The Johansen co-integration approach is adopted to test the null hypothesis that there is zero number of cointegration vectors between the daily stock prices. In addition, this test was used to confirm whether the prices in emerging market and developed markets have long run the association. The results of Johansen cointegration test for both NSE and NYSE are documented in Table 7 below.

Table 7: Result of Johansen cointegration tests Stock Eigenvalue Trace Statistic Probability Value

WIP 0.0624 76.53 0.000

HDB 0.0193 25.69 0.0011

TTM 0.01433 21.997 0.0045

Table 7 shows the results of Johansen cointegration test for all the three selected stocks. According to the criteria, the p-value for Wipro Limited, HDB bank, and Tata Motors Limited are all less than 5%. Thus the null hypothesis cannot be rejected; we can rather conclude that the stock prices in NYSE and NSE are cointegrated; this implies that they have a long-term association.

(31)

- 23 - In addition, using the results of the Trace Statistics, which is very high in the first three stocks, and adequate in the last two, we can conclude that all the three price series are cointegrated. Also, the trace statistics are strongly significant and show that a stationary combination of prices exists in the data series. Using the Johansen test results, we reject the possibility of non-cointegration vectors in the data series.

5.5 The Vector Autoregression (VAR) Model

VAR Model is employed to investigate the role of emerging foreign markets in the price discovery of cross-listed stocks. The model was specified in section 3.4.2. In this model, the lagged values are used to interpret the price discovery for each of the selected stocks. In essence, the lagged values represent a relative good proportion of price discovery in the developed markets.

Before using the VAR model, the number of appropriate lag periods to use has to be decided, where the author used the Lag Order Selection Criteria test. In this test, the LR, FPE, AIC, and HQ criterion confirmed that two-period lag is the best for the model as documented in table 8 below.

Table 8: Lag Order Selection Criteria test

Following the results above, used one period lag to analysis the data set using VAR model and a summary of the results are indicated in table 9 below. Complete results are documented in

Lag LogL LR FPE AIC SC HQ

0 1423.868 NA 0.000275 -2.523279 -2.514357* -2.519908 1 1437.494 27.18050* 0.000270* -2.540362* -2.513597 -2.530249* 2 1441.004 6.988061 0.000270 -2.539492 -2.494884 -2.522636 3 1442.541 3.055679 0.000272 -2.535122 -2.472671 -2.511524 4 1446.508 7.870862 0.000272 -2.535063 -2.454769 -2.504723 5 1448.653 4.248807 0.000273 -2.531772 -2.433634 -2.494690 6 1450.092 2.843695 0.000274 -2.527226 -2.411246 -2.483402 7 1450.859 1.513587 0.000275 -2.521489 -2.387665 -2.470922 8 1453.807 5.807105 0.000276 -2.519622 -2.367955 -2.462313

(32)

- 24 - Appendices A to E. Because the original data set is non-stationary, is calculated the one period lag VARs on the first difference to make it stationary.

Table 9: Result of VAR model analysis

Wipro Ltd HDFC Bank Tata Motors Ltd

D(NSE(-1)) D(NSE) D(NYSE) D(NSE) D(NYSE) D(NSE) D(NYSE)

-0.02595 0.00459 -0.088 0.023 0.024 0.234 std error 0.0295 0.01422 0.029 0.103 0.029 0.1335 t-statistics -0.8797 0.3228 -2.97 0.228 0.797 1.754 D(NYSE(-1) 0.2997 -0.0527 0.0099 0.0202 -0.001 0.0815 std error 0.0618 0.0298 0.0086 0.029 0.006 0.0202 t-statistics 4.852 -1.7691 2.254 0.049 2.108 0.264 C -0.00301 0.00193 0.014 0.0498 0.0008 0.0053 std error 0.00547 0.00264 0.0065 0.0224 0.0004 0.0202 t-statistics -0.5502 0.7328 217133 2.219 0.181 0.264 Source: Author’s Calculations

Table 9 provides a summary of key VAR model statistics for each of the three selected stocks as calculated by EViews. In this model, the price of stocks trading in the NSE was used as independent variables while those of stocks trading in NYSE as the dependent variable. Analysis and interpretations of these figures are provided in the paragraphs that follow in the next page.

The t-statistic is significant if it is greater than two are. From the table above, for the Wipro Limited, the t-statistics in one period NSE lags is not significant; however, it is significant for the one period NYSE lag. This is a clear indication that the lag can be used to explain price discovery of Wipro stocks in the NYSE market.

(33)

- 25 - For the Tata Motors, the t-statistics are not significant (less than 3) for both NSE and NYSE. This indicates that the one period lag cannot be used to explain the price discovery in the NYSE market for the two stocks. Lastly, for the HDFC Bank and Infosys limited, the t-statistics are greater than three meaning that they are significant and the one period lag can be used to account for the price discovery.

5.5.1 Interpretation of the VAR model results

To interpret correctly the results of the VAR model documented in table 9 above, it is essential to estimate the Impulse Response Function. A graph of impulse response function for the VAR model was constructed. We were interested in checking how fast changes in the NSE market cause changes in NYSE market for each cross-listed stock. Therefore, the ordering of the variables was from NSE to NYSE. The result of the Impulse Response Function test is given by Figure 3 below.

Figure 3: Response to Cholesky One S.D. Innovation

Response of D(NYSE) to D(NSE) Response of D(NSE) to D(NYSE)

Figure 3 shows the result of the Impulse Response Function test. The left graph indicates responses of changes in NYSE stocks to changes in similar stocks in NSE. The right one shows how stocks in NYSE respond to stock price changes in NSE.

(34)

- 26 - From the first graph, it can be clearly seen that there is an immediate response to a change in NSE prices. However, the second graph shows little response to changes in NYSE prices. In essence, the prices of stocks cross-listed in NYSE are contemporaneously affected by shocks or changes in prices of similar stocks in NSE. Furthermore, as we can see the changes occur almost immediately. Therefore, shocks in the NSE markets are almost immediately transmitted to the NYSE market. This can be further interpreted to imply that changes in the emerging markets almost immediately cause price discovery in the developed markets.

On the other hand, the right figure shows that there is little impact of changes in NYSE to changes in NSE. This can be concluded to imply that shocks in the developed markets do not cause immediate contemporaneous changes to prices in the emerging markets. With these results confirming the possible influence of emerging market on price discovery in developed country, we proceed to the next chapter where I performed the robustness check for the VAR model.

(35)

- 27 -

Section Six: Robustness Checks

This section presents robustness checks on the model and additional results that further confirms how emerging markets influence price discovery of stocks cross-listed in developed markets. The robustness checks were included in the empirical study to determine whether the coefficients of the model are plausible and robust, that is then taken as a clear indication of structural validity. The set of robustness checks used here examined the sensitivity of inferences made on different lags of emerging market stocks cross-listed in New York Stock Exchange.

6.2 Model Stability

The first robustness check that was performed is to determine the stability of the model. The author checked for the stability of the model using the Root of the Characteristic Polynomial method and a summary of the results is documented in Table 10 below. Complete result for each stock is documented in Appendices C to E.

Table 10: Root of Characteristic Polynomial

HDFC Bank Tata Motors Wipro Ltd

Root Modulus Root Modulus Root Modulus

-0.0902 0.0902 0.0785 0.0785 -0.0787 0.0787

0.0223 0.0223 0.0268 0.0268 0.000108 0.000108

Table 10 shows a summary of the results of Root of a Characteristic Polynomial check for model stability for each of the three selected stocks namely HDFC Bank, Tata Motors, and Wipro Limited.

(36)

- 28 - The model is considered stable f the absolute value of the modulus is less than one. The tables show that, for HDFC Bank, Tata Motors, and Wipro Ltd, the modulus values are all less than one, meaning that they are all inside the unit circle. This confirms that the VAR system is stationary and hence satisfies the stability conditions.

6.3 Residual Diagnostics

To confirm the further robustness of the model, the VAR was performed residual diagnostic checks using the constructed model. For the stationary Vector Autoregression (VAR) model to be correctly specified, the residuals should be white noises; otherwise, the model would not be considered robust. Therefore, once we obtained the VAR estimates in EViews, we can easily obtain the residual correlations and covariance matrices for diagnostics. However, for robustness check, we focused only on one residual test namely autocorrelation test. The result of the two-period lag autocorrelations test is shown in Table 11 below.

Table 11: Result of VAR Residual serial correlation LM Test

Lags

HDFC Bank Tata Motors Wipro Ltd

LM-Stat Prob LM-Stat Prob LM-Stat Prob

1 10.79 0.029 9.25 0.055 5.54 0.23

2 10.66 0.031 11.68 0.0199 6.26 0.18

Table 11 shows the result of VAR Residual serial correlation LM Test for each of the three stocks namely HDFC Bank, Tata Motors, and Wipro Limited. In this test, the null hypothesis is that the residuals do not have serial correlations and the selection criterion is a probability value greater than 5%. Looking at the graph, the probability values for HDFC Bank and Tata Motors are all less than 5% so we fail to reject the null hypothesis. The presence of serial correlation in the residual is an

(37)

- 29 - indication of the robustness of the VAR model. However, for the Wipro Limited, the probability value is greater than 5%, so we reject the null hypothesis. This indicates that the VAR model with one period might not be robust for the Wipro stock. A change of lag period might help improve the robustness of the model.

(38)

- 30 -

Section Seven: Discussion and Conclusion

This section provides discussion and conclusion on the study of the role of emerging foreign markets in the price discovery of cross-listed stocks. The results of this study is described the context of the existing literature and discussed its various limitations.

7.2 The Discussion

This study used the daily open stock prices for three stocks traded on both the National Stock Exchange (NSE) of India and the New York Stock Exchange (NYSE) of U.S.A. The NSE and NYSE stock markets were used to represent emerging market and developed market respectively. The aim of the study was to determine whether emerging market has a significant influence on the price discovery of stocks cross-listed in developed markets. The study used the daily open prices of three stocks namely HDFC Bank, Tata Motors, and Wipro Limited from 2 January 2013 to August 4, 2017.

The methodology of the study was based on the Vector Autoregression (VAR) model. Tests for stationarity and cointegration were done and the results showed that the time series data was nonstationary and cointegrated. Because the VAR model uses stationary data, the series was then differenced before applying the model. Furthermore, Lag Order Selection Criteria test was performed and it showed that the one period lag was most suitable for the VAR model.

The result of the VAR model analysis confirmed that emerging markets have a significant influence on the price discovery of stocks cross-listed on developed markets. This was further confirmed by the results of the Impulse Response Function test; the test showed that stocks in the developed markets quickly responded to price changes in the emerging markets. Therefore, the lag values can be used to interpret price discovery of stocks cross-listed in the developed markets. On

(39)

- 31 - the other hand, prices in eh emerging markets did not quickly respond to price changes in the developed markets.

The available pieces of literature have not extensively discussed the role of emerging markets in stock price discovery. Some of the available literature indicates that both developed and emerging markets influence stock price discovery. The result of this study provides a further confirmation to this. However, the majority of the literature has pointed out to the dominant role of developed markets in the stock price discovery. Even though the result of this study shows little influence, the role of the developed markets cannot be refuted. This is because, in this study, the developed market was not used as the independent variable hence the result might diverge.

7.3 Implications and Recommendations

This result of this study has shown that the fundamental of emerging markets is an important contributor to price discovery process in eh developed market. Therefore, traders in the developed markets should closely watch the developments and fundamentals of the emerging market. This information is also beneficial to trade analysts, planners, and advisors who are interested in developing good investment portfolios.

7.4 Limitations of the Study

It is also essential to understand some limitations of this research. First, the scope of the research was limited to only three selected stocks from emerging market; this is considered a smaller number to represent the entire emerging market. Secondly, the scope of emerging and developed markets was limited to India and United States only; it would have been better if other markets were included in the analysis. Another limitation of this research is that it only relied on the measure of daily open prices, which might not include all fundamentals of price discovery.

(40)

- 32 - 7.5 Conclusion

This research used the VAR model to study the role of emerging markets in determining stock price discovery in the developed market. Data were collected about three stocks from emerging market also cross-listed in the developed market. In conclusion, the study confirmed that emerging market has a significant influence on the stock price discovery in developed markets. This research study is essential because it contributes to the current literature in many ways. In particular, the result clearly shows that emerging markets significantly contribute to stock price discovery in developed markets. This affirms that the fundamental of emerging markets should be closely considered when trading stocks cross-listed in developed markets.

(41)

- 33 - References

Agarwal, S., Liu, C. and Rhee, S.G. (2007). Where does Price Discovery Occur for Stocks Traded In Multiple Markets? Evidence from Hong Kong and London. Journal of International Money and Finance, vol. 26, no. 1, pp. 46-63.

Ansotegui, C., Bassiouny, A. and Tooma, E., (2013). An investigation of Intraday Price Discovery in Cross-Listed Emerging Market Equities. Investment Analysts Journal, vol. 2013, no. 77, pp. 55-67.

Chen, H., Choi, P.M.S. and Hong, Y., (2013). How Smooth is Price Discovery? Evidence from Cross-Listed Stock Trading. Journal of International Money and Finance, vol. 32, no. 4, pp. 668-699.

Chouinard, E. and D'Souza, C., (2004). The Rationale for Cross-Border Listings. Bank of Canada Review, 2003, vol. 7, no. 2, pp. 23-30.

Claessens, S., Klingebiel, D. and Schmukler, S.L., (2002). The future of stock exchanges in Emerging Economies: Evolution and Prospects. Brookings-Wharton Papers on Financial Services, vol. 2002, no. 1, pp. 167-212.

Eun, C.S. and Sabherwal, S., (2003). Cross-Border Listings and Price Discovery: Evidence from US-Listed Canadian Stocks. The Journal of Finance, vol. 58, no. 2, pp. 549-575.

Feng, S. and Stewart, J., (2016). A Review of Market Segmentation and Inefficiencies of the Chinese Stock Market. International Journal of Finance & Banking Studies (2147-4486), vol. 4, no. 4, pp. 18-28.

(42)

- 34 - Frijns, B.P.M., Gilbert, A. and Tourani-Rad, A. (2007). Price Discovery, Cross-Listings and Exchange

Rates: Evidence from Australia and New Zealand. Department of Finance Auckland University of Technology.

Kim, M., Szakmary, A.C. and Mathur, I. (2000). Price Transmission Dynamics between ADRs and Their Underlying Foreign Securities. Journal of Banking & Finance, vol. 24, no. 8, pp. 1359-1382.

Lang, M.H., Lins, K.V. and Miller, D.P. (2003). ADRs, Analysts, and Accuracy: Does Cross Listing in the United States Improve a Firm's Information Environment and Increase Market

Value? Journal of Accounting Research, vol. 41, no. 2, pp. 317-345.

Solnik, B., Boucrelle, C. and Le Fur, Y. (1996). International Market Correlation and Volatility. Financial Analysts Journal, vol. 52, no. 5, pp. 17-34.

Visalakshmi, S. and Lakshmi, P. (2016). BRICS Market Nexus for Cross Listed Stocks: A VECX Framework. The Journal of Finance and Data Science, vol. 2, no. 1, pp. 76-88.

Wang, M.C. and Wu, Y.C. (2014). Where Does Price Discovery Occur? An Empirical Study of Taiwan’s ADRs and Their Underlying Foreign Stocks. International Journal of Financial Research, vol. 5, no. 3, p. 43.

Yang, S.Y., Doong, S.C., Wang, A.T. and Chang, T.L. (2005). Return and Volatility Intra-Day Transmission of Dually-Traded Stocks: The Cases of Taiwan, Korea, Hong Kong, and Singapore. Journal of Economics and Management, vol. 1, no. 2, pp. 119-141.

(43)

- 35 -

Appendices

Appendix A: THE VAR Estimates for Wipro Limited

This table shows the result of one period lag VAR Estimates for Wipro Limited stocks using Eviews; the t-statistics is greater than 2 hence significant.

Vector Autoregression Estimates Date: 08/09/17 Time: 17:01

Sample (adjusted): 2/05/2013 8/04/2017 Included observations: 1134 after adjustments Standard errors in ( ) & t-statistics in [ ]

D(NSE) D(NYSE) D(NSE(-1)) -0.025947 0.004588 (0.02950) (0.01422) [-0.87969] [ 0.32277] D(NYSE(-1)) 0.299711 -0.052673 (0.06177) (0.02977) [ 4.85170] [-1.76912] C -0.003011 0.001933 (0.00547) (0.00264) [-0.55018] [ 0.73268] R-squared 0.021702 0.002943 Adj. R-squared 0.019972 0.001179 Sum sq. resids 38.39038 8.918029 S.E. equation 0.184238 0.088798 F-statistic 12.54479 1.668896 Log likelihood 310.6153 1138.283 Akaike AIC -0.542531 -2.002263 Schwarz SC -0.529215 -1.988947 Mean dependent -0.002404 0.001826 S.D. dependent 0.186106 0.088850

Determinant resid covariance (dof adj.) 0.000266

Determinant resid covariance 0.000265

Log likelihood 1451.475

Akaike information criterion -2.549339

(44)

- 36 - Appendix B: THE VAR Estimates for HDFC Bank

This table shows the result of one period lag VAR Estimates for HDFC Bank stocks using Eviews; the t-statistics is greater than 2 hence significant.

(45)

- 37 - Appendix C: THE VAR Estimates for Tata Motors Limited

This table shows the result of one period lag VAR Estimates for Tata Motors Limited stocks using Eviews; the t-statistics is greater than 2 hence significant.

(46)

- 38 - Appendix C: Root of Characteristic Polynomial for HDFC Bank

This table shows the result of Root of Characteristic Polynomial test on HDFC Bank stocks.

Appendix D: Root of Characteristic Polynomial for Tata Motors

This table shows the result of Root of Characteristic Polynomial test on Tata Motors stocks. Roots of Characteristic Polynomial

Endogenous variables: D(NSE) D(NYSE) Exogenous variables: C Lag specification: 1 1 Date: 08/10/17 Time: 06:18 Root Modulus -0.090242 0.090242 0.022317 0.022317

No root lies outside the unit circle. VAR satisfies the stability condition.

Roots of Characteristic Polynomial Endogenous variables: D(NSE) D(NYSE) Exogenous variables: C Lag specification: 1 1 Date: 08/10/17 Time: 07:21 Root Modulus 0.078451 0.078451 0.026761 0.026761

(47)

- 39 - Appendix E: Root of Characteristic Polynomial for Wipro Ltd

This table shows the result of Root of Characteristic Polynomial test on Tata Motors stocks. Roots of Characteristic Polynomial

Endogenous variables: D(NSE) D(NYSE) Exogenous variables: C Lag specification: 1 1 Date: 08/10/17 Time: 07:25 Root Modulus -0.078728 0.078728 0.000108 0.000108

The role of emerging foreign markets in the price discovery of cross-listed stocks