CHAPTER 1

• Introduction

FTSE Bursa Malaysia Kuala Lumpur Composite Index (KLCI) is the internationally recognized standards benchmark for measuring the Malaysian economy subsequent from the collaboration between the stock exchange of Malaysia and its index partner FTSE. FTSE Bursa Malaysia KLCI Futures (FKLI) is a Kuala Lumpur Composite Index (FBMKLCI) Futures contract allowing the market participants to buy or sell the futures contract today and settle it on a later date. The Futures contract value is derived from the underlying financial index, which is Kuala Lumpur Composite Index.

On 27^th May 2021, the FKLI trading volume recorded was the highest, with 66,687 contracts traded in a day. Generally, Index Futures contract is actively used by investors, traders, hedgers, arbitrageurs, and speculators as financial tools for price discovery, hedging, arbitrage, and asset allocation, with a total volume recorded at 3,497,416 in 2020 and 2,700,398 in 2021. The main objectives of derivatives instruments are to offer financial tools for risk mitigation by taking a different position, similar underlying instruments in a different market, and bringing in more financial trading liquidity (Shah and Malik, 2017). (Shah and Khan, 2019) confirmed that the derivatives market plays a crucial part in bringing in more market efficiency.

Researchers are widely measuring the lead-lag relationship to test whether the futures lead the spot, the spot leads the futures, or even if there is a two-way causality relationship between both. When index futures respond to the information currently available on the market instantly, it means there is a presence of lead-lag correlation.

Besides, the extension from EMH theory is always associated with “random walk theory”. Malkiel (1973) proposed the Random Walk hypothesis, where the theory assumes any changes in financial prices will have their own independent distribution. Although the EMH theory presumes the market price is adjusted to the available market information, Malkiel (2003) believes the latest information released is usually not predictable. As a result, the market prices will be adjusted randomly. The idea of a random walk is when the information flow is unhampered and revealed instantly on stock prices.

The efficient market hypothesis (EMH) theory describes that the current market prices fully reflect the overall market information. (Paul A. Samuelson and Eugene F.Fama 1970). Therefore, in the futures market, the current futures price that will be expired in the current month will account for the EMH theory. In contrast, the later far month will not receive much of an impact because the contract expiry is further from the present.

Rastogi and Agarwal. (2020) empirical evidence indicates a bidirectional volatility spillover effect is found between the spot and futures market, while stronger bidirectional volatility is stronger on the spot month contract. The lead-lag relationship also might exist when the participants in the market use the previous information to make an assumption and predict the futures price when both markets have a feedback relationship reflected by not only the new information but also including the past information. Torun et al. (2020) indicate that the causal information flow between the spot and corresponding futures may change depending on the time scale and market data though there are similarities in causal information flow for both from the result evidence. Suppose the institution investor analyses the market using fundamental and technical analysis, which is always based on the company’s past performance information. In that case, there will also be a lead-lag relationship caused by the asset management company’s prediction of the spot and futures. Therefore, fully understanding the dynamic relationship between FKLI and KLCI, including the lead-lag relationship, the long-run relationship from the original sequence, and the spillover effect between the volatility, will give the market participant to reap the full advantage of the instrument.

The lead-lag relationship has been studied extensively across other countries, while most research focuses on developed countries and only a few on developing countries. In developed countries, a study conducted by Gong et al. (2016) found that futures lead spot, while in developing countries, spot leads futures. Scholars by Ren et al. (2022), Jiang et al. (2019), Kharbanda and Singh. (2017), Judge and Reancharoen. (2014) and the result indicates that the futures lead spot. On the contrary, Shao et al. (2019) show that neither of the spot and futures markets plays a predominant role in the lead-lag relationship between the spot and futures market, consistent with the findings from Zhang and Lui (2018), Demir et al. (2018), Chen et al. (2017). However, Wang et al. (2017) empirical evidence proved that the lead-lag relationship only exists within one day period while neither of the markets is dominant. Bhat and Suresh V.N. (2017) indicate the lead-lag relationship result was mixed between the futures and spot prices in selected nifty companies in the Indian stock market. Kavussanos et al. (2008) and Floros and Vougas. (2007) indicates the futures lead spot in the Greece market. However, Alemany et al. (2020) research on the German market shows that the futures market leads the spot market in the short run, while both markets are not cointegrated in the long run. Sifat et al. (2021) have proved that using the high-frequency data, the spot market leads the futures market, consistent with Tauson et al. (2018), using the intraday data showed the spot leads futures in Malaysia. In conclusion, the exploration of lead-lag relationships in developing countries is still weak compared to developed countries.

 

• Problem Statement and Research Questions

In December 2019, the first virus outbreak was reported in Wuhan within the province of Hubei in China. However, later the disease spread to other countries outside of China. On 30^th January 2020, the World Health Organization announced that the Covid-19 outbreak was a global health emergency and declared COVID-19 as a worldwide pandemic on 11^th March 2020. By 31^st July 2020, 17,106,007 confirmed cases with 668,910 in total number of death had been established due to the virus. The global outbreak not only impacted the health system to its alarming stage but also included the global market, and economies internationally were affected as well. The unprecedented pandemic outbreak has caused economic and social disruption, and the impact is shocking; that not only led to the loss of human life worldwide but also brought more financial and health challenges to public health, food systems, and all other industries in the world.

Based on the following two years historical FKLI daily trading volume, the year 2021 showed a significant FKLI trading volume and recorded a new all-time high in daily trading volume, rising to 66,887 contracts during the pandemic period. Because the latest FKLI record-breaking performance that occurred during the pandemic outbreak period where there were lots of uncertainties in the market, it motivates this research study to investigate further the lead-lag relationship between both KLCI and FKLI during this study period, specifically before and during covid19 pandemic year to see the difference in the results. The increase in the daily trading volume could be because of the market participants who might have thought of using the futures market as the price discovery and better hedging tools during the uncertain market unless this research evidence states otherwise.

 

Furthermore, there is a market belief that the futures contract, current month contract, and far month contracts prices can be used to act as price discovery for the spot market due to the assumption that futures prices are unbiased estimators and futures always play as the major price discovery function Ren et al. (2022), Alemany et al. (2020), Jiang et al. (2019), Demir et al. (2018), Kharbanda et al. (2017), Judge and Reancharoen. (2014). However, based on the previous literature reviews that will be summarized in the next chapter point toward mixed results that either futures lead the spot or spot leads the futures. Hence, till this moment, the lead-lag relationship and price discovery between both markets are still ambiguous and might have changed since the previous discovery when the daily volume increased in the year 2021. The latest verification to quantify this belief is essential for the market participants to firm up their trading strategy. By knowing whether the research results of previous scholars remain applicable in the current market, the participants will be able to give a good idea of the price discovery function. Then, the participants will be assured of using the precise benchmark price for risk mitigation.

Finally, several pieces of literature are conducted related to the price discovery and lead-lag relationship between the spot and futures market. However, most of these scholars have been focusing more on the well-developed market scholars like Alemany et al. (2020), Shao et al. (2019), Zhang and Lui. (2018), Chen et al. (2017) instead of the emerging market. For the case of the Malaysia study, most of the research studies by Sifat et al. (2021) and Tauson et al. (2018) focused on high-frequency intraday day data. Therefore, this research aims to fill in the research gap by focusing on the daily closing price in the Malaysian market, specifically before the Covid-19 pandemic occurred in Malaysia from April 2018 to Jan 2020 and during the Covid-19 pandemic in Malaysia from March 2020 to January 2022 period. The following research questions have been classified to address the issue for this paper. Is there any presence of a lead-lag relationship between the Kuala Lumpur Composite Index (KLCI) and the FTSE Bursa Malaysia KLCI Futures (FKLI) pre and during the pandemic period in Malaysia? Is there any presence of long-run and short-run between the Kuala Lumpur Composite Index (KLCI) and the FTSE Bursa Malaysia KLCI Futures (FKLI)

• Research Aim and Objectives

For this research study, this paper objectives are as follows:

• To determine the presence of a lead-lag relationship between the Kuala Lumpur Composite Index (KLCI) and the FTSE Bursa Malaysia KLCI Futures (FKLI)
• To investigate the presence of long-run and short-run between the Kuala Lumpur Composite Index (KLCI) and the FTSE Bursa Malaysia KLCI Futures (FKLI)
• Significance of the Study

The FKLI futures contract specifications offer not only the spot-month contract but also the next month and the next two quarterly calendar months. The far-month contracts offering will be used as the indicators of the future spot rates. The study of the lead-lag relationship between KLCI and FKLI could assist fund managers, hedge fund managers and speculators in understanding the lead-lag relationship and taking advantage of the price difference while using the futures contract for prudent portfolio management to protect from any market uncertainties and price fluctuations from the underlying index.

Furthermore, the significance of this research study is the recent year observation period from Jan 2016 to December 2021, with a total of 1,566 observations using daily data intervals for the past five years. This study could be considered a continuation of the previous scholar Sifat et al. (2021), who only observed the 2017-2018 intraday period, while Tauson et al. (2018) observed one-month intraday data. Meanwhile, Jusoh et al. (2014) observation period is considered outdated due to the paper studying the previous ten years spot and futures market performance, which might not be relevant in the current market.

This research’s final significance is that most of the previous lead-lag relationships between the spot and futures market focused on the developed market. At the same time, only a few scholars were found to cover the developing market; this paper could contribute to the emerging market lead-lag relationship empirical research for the emerging market studies.

• Organization of Chapters
• Chapter 1: This chapter provides an overview and introduction of the research study. Further details on the problem statement, the motivation for conducting this research, research questions, research objectives, research hypotheses, and significance of the study.
• Chapter 2: This chapter summarizes the literature review, whichever is relevant to this research discussion. Based on this research study, the literature review area of analysis is within the investigation of the lead-lag relationship between the spot and the futures market. Further discussion on the approaches and the final section summarizes the overall literature review summary.

• Chapter 3: This chapter provides the fundamental framework for the lead-lag relationship that explains the overall research design, techniques, and data analysis. The research method’s structural background, including the measurement benchmark, will be detailed further in this chapter.
• Chapter 4: This chapter will discuss in detail the data analysis, which covers the overall data structure, frequency, interval, source of the data, factor analysis, and reliability analysis, including the components of descriptive analysis. The research findings also will be presented in this chapter.
• Chapter 5: This section will summarize the study’s conclusion, including the gathered study key findings, limitations from the proposed model structure, and further recommendations on study improvement that other researchers can explore.

 

CHAPTER TWO

LITERATURE REVIEW

2.1    Introduction

There are several number of research studies have been conducted previously to investigate the lead-lag relationship between the futures market and its underlying spot market based on the foreign and Malaysian markets. Given the research evidence, some results presented that the underlying spot market is leading the futures market while others are vice versa, that futures are leading the spot market. The evidence of whether the futures or spot market provides the price discovery plays a significant role in giving the respective market participant a better price direction for a viable underlying asset risk mitigation strategy.

Based on the renowned Efficient Market Hypothesis (EMH) theory by Fama (1960) indicates that all the available information on the market is already reflected in the asset prices. Hence, there won’t be any time delay between the spot and futures because the asset prices would adjust to the new price accordingly following the newly released information. It is hard to predict because different people might have different interpretations and assumptions. If the market is inefficient, there will be a presence of a lead-lag relationship between both markets – Stoll and Whaley (1990).

Due to the fact that EMH theory ignores certain conditions, such as assuming there are no transaction costs involved. Also, all investors and the markets are rational and will react rationally; this increases the doubt about the theory’s robustness. The extension to study whether human beings actually behave how they should behave rationally in the market is being reviewed by Shiller. (2003) and identified as behavioral finance. The theory reveals the real market movement might fail to reflect the economic fundamentals due to irrational investor behavior. When a group of investors interprets differently, an irrational systematic behavior will be formed into price deviation patterns. In addition, the information resources are accessible to both futures and spot market investors in parallel; however, due to the cost and carry model, both market reactions will have their own different pace of market influence. Flemming et al. (1996).

 

There are a few key advantages that weigh on the futures market as the price discovery as compared with the spot market, which indirectly will contribute to the futures market to lead the cash market.

There are four types of short selling offerings in Malaysia which are Regulated Short Selling (RSS), Intraday Short Selling (IDSS), Intraday Short Selling by Proprietary Day Traders (PDT), and Permitted Short Selling (PSS). Each short-selling category offers a unique mechanism that would prevent the participant from fully taking advantage of the short-selling benefit. However, the difficulty only relates to the equities market and not the Futures market that offers the ability to short-sell. Furthermore, because of the lower entry of costs in trading, the futures contract will ease the participation to trade in the futures market. Thirdly, the leverage contract size will benefit the investor to gain leveraged exposure to the underlying without having to enter into the underlying instrument. Finally, the futures market not only offers the near month, but it also offers the far month contract that gives the flexibility for the investor to enter into the distant month contract position with their future term market expectation, which could act as the predictor for the future spot price. The general conclusion arrives with the acknowledgment of the futures market as the price discovery role, along with the dismissal of the EMH theory that there shouldn’t be any lead-lag relationship between both markets. Further discussion on other scholars will be analyzed and discussed afterward.

 

2.2    Empirical Literature

Based on the literature reviews I have completed, mixed results were gathered from the earlier literature papers I reviewed. Selected researchers’ results debate the futures lead spot (Ren et al., 2022; Alemany et al., 2020; Jiang et al., 2019; Demir et al., 2018; Kharbanda and Singh, 2017; Judge and Reancharoen., 2014), while some highlight that spot leads future (Sifat et al., 2021; Tauson et al., 2018), besides the empirical evidence also found that neither of the spot and futures markets plays a predominant role. (Shao et al., 2019; Wang et al., 2017) and the final argument was that the results mixed that futures lead spot and spot leads futures. (Pradhan et al., 2021; Zhang and Lui. 2018; Chen et al., 2017; Bhat and V.N., 2017). Some of the areas of the argument behind the empirical results are when the spot market return is higher than the futures market, and the investors tend to re-adjust their portfolio according to the spot market, and it turns out the spot will lead the futures market. When the market is too efficient, the investors are unable to earn excess profits because both futures and spot are too efficient; therefore, there will be no lead-lag relationship between the two. Furthermore, some of the additional advantages of trading in the futures market include short selling flexibility, lower transaction costs, and lower initial capital outlay. While the spot is lagging, the futures is because the spot market transactions cannot be executed as quickly as the futures market (Silvapulle and Moosa, 1999).

 

2.3    Futures lead spot Lead-lag relationship

In 2014, Judge and Reancharoen. (2014) employed a cost of carry model to study the lead-lag relationship between the Thailand SET Index (SET) stock exchange with the SET50 Index Futures from 2006 to 2012 with a daily period. They used the Augmented Dickey-Fuller (ADF) unit root test to check for the data’s stationary properties, and the results provided evidence that both markets were stationary. Thus, the authors apply the Granger causality test, and the findings indicate that the SET50 Index leads the SET50 Index futures. Judge and Reancharoen. (2014) propose that the incorporation of new information in Thailand’s spot market is much faster as compared with the futures market.

Meanwhile, a study was conducted to investigate the relationship between the spot futures and the spot market by Kharbanda and Singh. (2017) in India. The authors focus on the FX market, specifically USD, GBP, EURO, and JPY, between February 2010 to Jun 2016 on a daily basis. Although this study area is within the FX market scope, but what is interesting for us to see is basically the empirical results out of this research due to the methodology employed by Kharbanda and Singh. (2017) is also applicable to the stock composite index. The lead-lag relationship is estimated following with few measures starting with stationary tests using the common model (Augmented Dickey Filler, Phillips-Perron, and Kwiatkowski-Phillips-Schmidt-Shin) because most of the economic time series are non-stationary (Gujarati et al., 2011) thus it is crucial to check for stationarity. The result confirms that all currencies’ futures and spot series become stationary at the first difference and integrated of order one I(1). Further exploration of the cointegration test using Engle and Granger’s residual-based approach shows that the residuals of the OLS equation formed are stationary, meaning both variables are cointegrated. Meanwhile, Johansen’s cointegration test confirms the long-run equilibrium relationship between the futures and the spot market. Finally, the author used the Vector Error Correction Model (VECM) to find out which market is leading. The result showed that the futures market makes fewer adjustments to reach equilibrium compared with the spot. The result is interesting for us to look at because it indicates that there is a long-run relationship between the futures and the spot market, which is consistent with Judge and Reancharoen. (2014) even though both research variables use a different market. The supportive argument behind the empirical results is because of the ease of market transaction access to the futures due to lower transaction cost, and it is much easier to enter into a futures position with low entry cash considering the leverage exposure per contract as compared to the actual spot that requires higher capital and longer time for trading execution. Another interesting fact pointed out was that the futures offer the participants’ price transparency and anonymity.

Demir et al. (2018) research study investigate the interrelationship among the spot, futures, and forward cotton markets in China during the major policy change period from 2011 to 2014. The research uses daily China Cotton Index 328 (CC328), Cotton futures contract, and Cotton forward contract by applying the stationarity test, Engle and Granger’s (1987), Vector Autoregressive (VAR) models in which the changes in causality are treated as random events governed by an exogenous process and Linear error correction model (ECM). First, the results indicate a linear and stable long-term relationship between the variables, which is highly influenced by the two policy shifts within the observation period. Furthermore, Demir et al. (2018) found that most futures lead forward prices, and the futures leading role faded with the introduction of the State reserve program for cotton in 2011. The overall conclusion is the government intervention distorted the way market participants interacted and was not sustainable. The results signify that the futures market plays a dominant role in the price discovery process. Hence, one important lesson learned from this paper is that the lead-lag relationship might not be sustained during certain market conditions due to the major policy changes that might impact investor behavior and affect the lead-lag relationship before or after the new policy implementation.

Jiang et al. (2019) conducted research in the year 2016 to study the non-linear causality between spot and options markets in China because the linear Granger causality, which is the classic methodology for measuring the lead-lag relationship, has been used extensively. This research explores the linear and non-linear lead-lag relationship among the Shanghai Stock Exchange Index (SSE50), SSE50 Index futures, and 50ETF Spot and Options Markets from 3^rd Feb 2017 to 28^th February 2018. Jiang et al. (2019) utilized the Nonparametric thermal optimal path (TOP) on 5 minutes and one-hour data due to the model’s ability to process high-frequency data. The major claim from this paper was no research had been conducted on the causal relationship of the selected market. Based on the research result gathered, the author highlights that 1-hour data is a better data frequency for analysis using the Granger causality for all three markets, and from the one-day test results, they found that the causalities vary over time; thus, the author uses the TOP method to study the short term causalities. The results signify that the 50ETF call options market leads the 50ETF spot market. In contrast, the 50ETF put-options market leads the index futures market, and the other series pairs are bi-causally related. This paper proved that the VAR and VECM models are able to disregard the linear relationships between the log-price series. Finally, the paper concluded that one of the reasons for the non-linear Granger causality is because of the volatility spillover, and there is a high likelihood that these relationships might change temporarily during the day. Finally, this study suggested that the lead-lag relationship is bi-directional over the sample period, and at certain high frequencies, the nonparametric Granger causality test is not able to give causalities, while the alternative way is to use the TOP method to study the relationship.

The price discovery role of the lead-lag relationship has also been examined in the context of a developed country, in this case, is Germany. This paper examines the relationship of the German market between the DAX30 index spot and DAX30 index futures by Alemany et al. (2020), which analyzed the price discovery process within a 5-min data scale. This study utilizes the regime-switching models (MS-VECM) and regime-dependent impulse response function to overcome the weakness of linear assumptions in the dynamic relationship between spot and futures prices with the advantage of endogenously determining the changes in the dynamic relationship without guessing exogenous structural changes. The results indicate that the dynamic causal effect differs among linear and non-linear models, including among the regimes. Unilateral interactions between these markets, the futures market leading the spot market, were found using the VECM. However, in the long run, it shows that the spot and futures markets are not cointegrated, while in the short run, the MS-VECM spotted two-way causality in the price discovery, with the futures market taking the leading part. Overall, this paper implies that the traditional models are not successful in capturing the lead-lag relationships that appear and validate that the nature of the price discovery process changes differently based on the regimes, which is kind of similar to Demir et al. (2018). Further advice is that the MS-VECM will provide a better view of the dynamic relationships and reduce the bias when disregarding the nonlinearities. The empirical evidence from this paper will be helpful for portfolio managers to understand the dynamic German relationship between both the spot and futures market for portfolio construction and portfolio risk management.

Ren et al. (2022) investigated the dynamic variation of intraday lead-lag relations between stock indices and their futures on the stock indices in multi-market. The authors utilized 1 min data from 16^th April 2015 to 31^st Dec 2020 for China market and 1^st September 2017 to 31^st December 2020 for Hong Kong and US Market. The Augmented Dickey-Fuller (ADF) test was conducted and found that not all of the time series is stationary, but the return series are shown to be stationary and integrated. Two methods are being applied to identify the dynamic intraday lead-lag relations: TOP to capture the non-linear lead-lag structure between the economic variables, and VECM IS, a method to examine the dependence structure between the two non-stationary time series. The study result provides robust conclusions from the TOP and VECM-IS model, where three markets reveal a piece of solid evidence that both futures and options demonstrate price discovery function over the spot market. In contrast, options proved to hold a firmer price leadership than the futures. In addition, the futures price leadership function reverses when the index return has a significantly larger mean value similar to the futures and options for all three China, Hong Kong, and US Market. Ren et al. (2022) emphasize that when the spot market generates a higher return than the futures market, the investor tends to re-adjust their portfolio strategies based on the spot market performance, which is an interesting pivotal point to highlight for market participants to have some ideas how they are going to re-adjust their intraday investment strategies while managing the portfolio risk.

 

2.4    Spot lead futures Lead-lag relationship

From the context of spot leads futures, Taunson et al. (2018) investigate the intraday lead-lag relationship between KLCI Index and FKLI market evidence from the emerging market using 15-second interval data from 2^nd Jan 2018 to 21^st Feb 2018. The author is interested in using the intraday data because none of the previous studies ever investigated using the high-frequency interval data for the case of Malaysia. Based on the empirical results, the spot and futures returns are identical with the spot returns more, and both series exhibit a significant first-order autocorrelation and are negatively skewed. The result from the ADF test indicates both series are stationary. The overall empirical result implies both markets contribute towards the price discovery, but the results leaning toward KLCI leads the FKLI. Taunson et al. (2018) suggest that the KLCI spot market is much more responsive to the market information as opposed to the FKLI Index futures market; thus, the market participants need to be concerned about the spot market information when trading the FKLI to arrive at the equilibrium price between both markets.

Further research extension on the Malaysia market was conducted by Sifat et al. (2021) using the same 15-s high-frequency data with a different period, September 2017 to August 2018, employed by a different methodology as compared to the previous study conducted by Taunson et al. (2018) to investigate the price discovery between the FKLI and KLCI index. This paper employed the Maximal Overall Discrete Wavelet Transform to evaluate the interdependence between the futures and spot returns. Cross-correlation and cross-coherence findings from this study explicitly lead towards the spot market as the price leader, with the futures market lagging by up to 1-min before the high convergence occurs between the two markets with just about 8-minute. At the same time, no discernible regime-specific or clustered patterns of cross-correlations were found that are changing over time. Besides, the results in testing the robustness checks using the DCC-GARCH, BEKK-GARCH, and VECM results survived. This paper manages to observe that the price discovery between futures and spot markets at a granular level is a scale-dependent phenomenon and records a shred of empirical evidence that the spot market is leading the futures market in price formation, which converges in between 1 to 8 minutes. In general, the research findings are against the efficient market hypothesis. The Wavelet coherence tests result signifies that there is no specific change in the evolution of the spot-future dynamics during the turbulent after the election period trading sessions. Therefore, there is a potential for statistical arbitrage in high-frequency trading.

 

2.5    Mixed results Lead-lag relationship

The lead-lag relationship in selected NIFTY companies listed on the Indian stock market between the spot and futures prices were also investigated by Bhat and Suresh. (2017) in 2017 for thirteen years from 9^th November 2001 to 31^st March 2014 for a daily price. The selection of the companies is based on certain conditions: those companies that are single stock futures are permitted for trading in the F&O segment since 9^th Nov 2001, included in Nifty, and permitted for trading in the F&O segment since Jan 2003. Those are included in the Nifty for a minimum period of 11 years. First, the Jarque Bera test was conducted to test the goodness of fit, and the result rejected the null hypotheses, and the data is confirmed to be non-stationarity. The Granger causality and Johansen’s cointegration results proved that there exists a lead-lag relationship between both futures and spot markets for all the selected companies. In the long run, the cointegration between futures and spot shows that futures prelude during the price discovery process. Finally, the VECM method verified that the futures have a significant role in price discovery and thus contribute to market efficiency.  The author suggests that the futures prices provide information about the fluctuations in the spot market in the long run relationship. Bhat and Suresh. (2017) conclude that a bidirectional and contemporary relationship exists between the futures and spot prices for the nineteen selected companies, while there is a presence of unidirectional relations for the remaining three companies.

Further investigation on the lead-lag relationship was carried out by Wang et al. (2017) to examine the relationship between CSI 300 Index spot and futures markets better to understand the connection between both markets in China. The author retrieved the data within daily and one-minute intervals from 16th April 2010 to 31st December 2014 with a total of 1144 observations. The TOP nonparametric method was implemented to identify the real-time lead-lag structures between pairs of time series. The authors found that the daily data indicated a volatile lead-lag relationship between both markets, while neither of the markets plays a dominant role in the price discovery. The one-minute data frequency result signifies that the futures market leads the spot market, but it is less than 5 min and does not depend on the market’s price trend. The author’s main argument is that index futures tend to move first in response to market news while dragging the stock index along with the reasoning of changes in the spot market are affected by the changes in the futures market. With the futures market leading the spot market, which lasts less than 5 min author’s assumptions are due to the limitation of the day trading in the spot market in China, which allowed the futures price to move faster than the spot price; therefore, the futures market accelerate resulting from the market information adjustment. Furthermore, the futures market may react to the latest market information efficiently because of the institutional investors that dominate the futures market; meanwhile, the individual investors mainly only participate in the spot market. The following argument was the degree of leverage that is attainable in the futures market could be the reason behind index futures leading the stock index and because traders are allowed to short-sell in the futures market, allowing the futures to lead the spot as compared with the limitation within the spot.

The relationship between spot and futures markets within the four-energy sector was also investigated by Chen et al. (2017) during the oil crisis versus the non-crisis period in the US market to address the research gap on price discovery in the energy sector under different scenarios such as oil crises and over the business cycle. The author designed the study by first testing the unit root test using the Augmented Dickey-Fuller (ADF) and employing the Johnsen test to test for the cointegration and estimate parameters of the common VECM model was applied to test for the causality between the energy variables. The empirical findings reveal both futures and spot prices showing the presence of unit root and stationary after the first difference. At the same time, the cointegration test indicates that both variables have at least one cointegration, thus correlated in the long run. The overall result shows the independent of oil shocks, business cycle, and transaction costs with a conclusion that the futures market is a biased estimator and a major contributor to price discovery between both markets. Still, the gasoline sector indicates that the spot market is leading toward major price discovery because it reacts more quickly to new information than the futures market. Some other arguments that are worth highlighting are even though futures played the primary price discovery function during the oil crisis periods, a deterioration in its price leadership was found during the business cycle. Furthermore, the spot market tends to reflect the new information more during recession periods than in expansion periods. Finally, the spot market always costs higher and tends to generate lower returns compared to futures with minimal impact if the author considers the transaction costs.

Pradhan et al. (2021) recent paper delve into investigating the relationship between spot and futures markets with the aim of determining if and to what extent the Indian commodity market is efficient and evaluating the stage of development of an emerging country by observing the current level of market efficiency. Pradhan et al. (2021) use the Autoregressive distributed lag (ARDL) technique following the previous study method by Pesaran et al. (2001) to examine the long and short-run dynamic bidirectional causality between spot and futures prices. The study employs daily futures and spot prices of aluminum, copper, crude oil, gold, nickel, silver, agricultural, livestock, and precious metals from 2009 to 2020 based on the consideration of the global financial crisis after effect from 2008 to 2009 and up-to-date data. From the empirical evidence extracted from the paper, it signifies that the Augmented Dickey Fuller (ADF) and Phillips-Perron (PP) results indicate the futures and spot prices of all variables are non-stationary at level data but stationary at 1% at first difference. From the Bounds testing for cointegration test, results show that there may be a long-run relationship between the six commodities and the agricultural index. The result from Johansen and Juselius cointegration test confirms that there is a long-run relationship between spot and futures prices in the commodity market. In the long run, the causality exposed unidirectional causality from spot to futures price for aluminum and copper only, while in the short run, the causality analysis exposed bidirectional causality from spot to futures price for copper and silver only. Finally, there is unidirectional causality from futures to spot price for gold, agricultural, livestock, and precious metals, and an independent relationship was found for nickel that suggests a neutrality effect. The author concludes that aluminum and copper are more information efficient than other commodity markets in the study, and thus the market participants will not be able to generate additional profit due to the significant market efficiencies; however, vice versa for the case of crude oil, gold, nickel, and silver that the market is less efficient. Pradhan et al. (2021) provide their feedback based on the evidence that futures prices respond to the new information faster than spot prices because of flexibility in short selling, lower transaction costs and lower capital outlay, faster execution time, an additional advantage in the futures market for leverage mechanism.

Shao et al. (2019) paper investigates the time-varying lead-lag structure between the crude oil spot and futures markets by employing the nonparametric and non-linear symmetric thermal optimal path (TOPS) model, which is the improved previous TOP model that was proposed as the novel method in quantifying the dynamic evolution of lead-lag structures between the two-time series. This study’s main objective was to overcome the previous research difficulties that the conventional methods are static, linear, and parametric. The TOPS method was applied to determine the time-dependent lead-lag relative to the time series based on daily (23^rd April 1987 to 10^th October 2017), weekly (17^th April 1987 to 6^th October 2017), monthly (15^th March 1987 to 15^th September 2017) spot and futures returns for the maturities of one, two, three and four months of the West Texas Intermediate (WTI) crude oil benchmark from 1987 to 2017 with different frequencies. The research results for the daily data indicate that the lead-lag relationship between the spot and futures market only exists for one day and some specific days. In contrast, on average, the lead-lag relationship between the two markets diminished when the frequency increased. Besides, the solid lead-lag relationship results witness an overlapped with most of the main influential event changes that occurred in the oil and stock markets, including the geopolitical regime changes. Finally, the author concluded that the lead-lag relationship relative to the two spot and futures oil markets only exists temporarily within the overall observation period.

In summary, even though most of the previous literature recognized the futures market as the price discovery function, however, due to the differences in research mechanisms for each study, the result turns out to be different, especially in comparable literature focusing on the Malaysia market by Sifat et al. (2021) and Tauson et al. (2018) that only observe the 15s high-frequency data interval.  This research paper differs from the rest of the works of literature from the data interval viewpoint, where the interval will be in 1-minute and 5 minutes intervals to increase the robustness of Malaysian research empirical evidence. Based on the proposed research parameters, this would later offer a significant value for domestic and foreign key industry players from the asset management company, investors, traders, speculators, arbitrageurs, and academicians and regulators.

 

Table 2.1: Summarized research from the above lead-lag relationship literature:

Literature	Data Interval	Main conclusion	Research Method
Ren et al. (2022)	2015–2020 (Minute)	Futures lead spot	TOP, VECM IS
Pradhan et al. (2021)	2009-2020 (Daily)	Mixed	ARDL, VECM
Sifat et al. (2021)	Sep 2017-Aug 2018 (15s)	Spot leads futures	Wavelet, MODWT
Alemany et al. (2020)	Jan 2014-Sep 2015 (5 Minutes)	Futures lead spot	Regime switching models
Shao et al. (2019)	1987-2017 (Daily, Weekly, Monthly)	Neither market dominant	TOPS
Jiang et al. (2019)	Feb 2017-Feb 2018 (1 Minute, 5 Minutes, 1 Hour)	Futures market leads the spot market	TOP, Granger Causality
Demir et al. (2018)	Jan 2005-April 2017 (Daily)	Futures lead spot	Granger Causality
Zhang et al. (2018)	Jan 1997-Feb 2016 (Daily)	Mixed	MF-DCCA
Tauson et al. (2018)	Jan 2018-Feb 2018 (15 seconds)	Spot leads futures	Cross correlations
Chen et al. (2017)	1979-2013 (Daily)	Mixed	ADF, VECM
Wang et al. (2017)	2010-2014 (Daily and Minute)	Neither market dominant	TOP
Kharbanda et al. (2017)	Feb 2010-Jun 2016 (Daily)	Futures lead spot	Granger Causality, VECM
Bhat and Suresh V.N. (2017)	Nov 2001-Mar 2014 (Daily)	Mixed	Granger Causality, VECM
Judge and Reancharoen. (2014)	2006-2012 (Daily)	Futures lead spot	Granger Causality

 

Chapter Three

Data and Methodology

 

3.1 Introduction

This research study utilizes the Malaysia capital market financial time series data equity market, FTSE Bursa Malaysia KLCI Index spot (FBM KLCI) and futures market, FTSE Bursa Malaysia KLCI Futures (FKLI) prices. The sampling collection involves the process of gathering information from various kinds of resources, and this paper only will capture the quantitative data gathered from the Bloomberg terminal.

The daily observations obtained for this study started before the pandemic period from 10^th April 2018 to 24^th January 2020, which is a day before the first covid case in Malaysia, and backward 469 no of trading days align with the no of observation during the pandemic period observation. The period of observation during the pandemic period is starting from 18^th March 2020 until 3^rd January 2022. The starting date was selected based on the first day of implementing the Movement Control Order in Malaysia until the end of movement restrictions in Malaysia.

The spot and futures market prices were derived from the daily closing price of 4 futures prices being utilized (Spot month, the next month, and the next two calendar quarterly months (March, June, September, and December) based on the FKLI contract specifications. The period of observation caters to the Malaysia stock exchange Monday to Friday, except on Malaysia public holidays and other market holidays for both spot and futures prices. In total, there are 469 observations for both before pandemic and during pandemic gathered for this research examination.

First, to get some brief informational summary of the dataset, descriptive statistics will be presented that summarize the structure of the variables. Next, a unit root test will be conducted to check the stability of the data because it is crucial to ensure the data to be in the order of integration I(1), means the first order difference of two non-stationary time series will be converted to stationary time series for later cointegration test. This is to test the long-run equilibrium between the variables. Furthermore, Error Correction Model will be executed if cointegration is existed to correct the short-term and long-term steadiness. Finally, the lead-lag relationship will be analyzed using the Granger Causality Test, and further particulars of each test will be discussed in the next section.

 

 

3.2 Methodology

3.2.1 Descriptive Statistics

Descriptive statistics will be carried out to analyze the behavior of the data structures for both markets. The common components are mean, median, maximum, and minimum statistics of the overall data. Usually, it describes the basic data features in a study, including the standard deviation, skewness, and kurtosis for market prices.

3.2.2 Unit root test

The time series data consist of timely observation with some stochastic processes considering the time series as the realization from the random variables and related to the stochastic processes. The variances and covariance from the time series are independent of time rather than the entire distribution. When the data comprise no constant mean, variance and covariance or all of the component, the time series is considered non-stationary. When it is non-stationary, the problem will arise, and over time, spurious regression phenomenon may lead to an exceptionally significant t-statistics value and high  value, this is called a unit root problem in the series. Newbold and Granger (1974) pointed out that the spurious regression phenomenon is caused by the non-stationarity in the time series. Therefore, the stationarity of the series is important because correlation could continue in non-stationary time series even if the sample is large, and the result is called nonsense regression Yule. (1989).

On the other hand, stationary refers to the process of mean, variance, and covariance staying constant, and probability distribution for the entire series remains steady. The unit root non-stationary issue in time series is able to resolve by data set differencing. Wei. (2006), Phillips and Perron (1988) revealed that by conducting a unit root test, we would know whether the time series trend is stochastic or not, and determining the moment where the non-stationarity commences will help us during the non-stationary elimination process.

For example, the first-differencing method can be used to eliminate the non-stationarity drift in the random walk drift model; meanwhile, the detrending approach removes the non-stationarity in a deterministic trend process. Furthermore, before running the cointegration test, it is required to ensure that both variables are integrated in the same order because if both series are cointegrated or order n, the respective series will need to be differenced by n times to restore stationarity. A unit root test will be performed on all data series with the objective of to establish the stationary on the time series and avoiding any misleading results caused by the regression.

In testing the unit root test, this paper will use the Augmented Dickey-Fuller (ADF) test, another variation from the Dickey-Fuller (DF) test in which the previous test assumes that the residuals are not correlated. Which for our case, both series might be correlated with one another; thus, the ADF capable of overcoming this assumption and resulting from it will be much more appropriate. Following are the three forms of ADF test:

• Pure random walk model
• Random walk with drift or intercept
• Random walk with drift and linear time trend

Finally, to get some assurance from the main ADF unit root test result, Phillips-Perron (PP) test developed by Phillips and Perron in 1988 will be conducted to ensure that both results are consistent. PP methodology is selected to ensure the unit root accuracy because it utilizes nonparametric statistical methods to address serial correlation errors. Moreover, the ADF test’s t-statistics tend to generate better finite sample properties. This highlights Campbell and Perron (1991) also one of the main reasons for using PP methodology to support the ADF test assessment. The hypothesis involves in this test are as follows:

 

 

 

3.2.2.1 Lag Order Selection Criteria

To estimate the Error Correction Model (ECM) for inferential and interpretive purposes requires the selection of the most appropriate number of differenced, lagged terms in the model.

This paper will identify the lagged difference terms using the Schwarz Information Criterion (SIC) to ensure that the errors are not correlated among the variables. This would later increase the ability to identify a unit root when conducting the test, Enders. (2004). The justification behind choosing SIC instead of Akaike Information Criterion (AIC) is because of the large sample possessions which fit with our study parameter, imposes a more significant penalty for additional coefficients, and because AIC usually tends to pick an overparameterized model supported by the study conducted by Hoxha. (2010).

3.2.3 Cointegration test

When the time series data are happened to be non-stationary, but the combination between both data sets is stationary and moves together over time, this sequence is considered to have a cointegration relationship. This is as per defined by Engle and Granger (1987) that the linear combination between variables is cointegrated if both are stationary. This study will utilize the Engle and Granger (1987) method to test out the cointegration relationship between the KLCI Index and FKLI index futures. When the time series data are cointegrated, the residuals are stationary and will have a long-run equilibrium relationship between the variables. Because the cointegration test requires all series to be integrated in the same order, the following equation is derived for this study:

The hypothesis for this test is as follows:

 

 

 

3.2.4 Vector Error Correction Model (VECM)

Once the cointegration relationship has been confirmed to be present between the KLCI index and FKLI index futures, there is a long-run equilibrium relationship between both. The next step is to use the VECM to indicate the speed of adjustment from the short-run equilibrium to the long-run equilibrium state. The higher the  value, the faster it will be for the model to rebuild equilibrium following a shock. The VECM is considered a restricted VAR, which was designed to use for non-stationary series that are known to be cointegrated. Even though the cointegration test is able to identify the long-run equilibrium relationship, however, the presence of disequilibrium within the short run is not guaranteed. Therefore, the VECM method is employed to trace the short-run linkage between the variables, and the following equation is established where  is the error correction term with  has to be a statistically significant and negative value.

3.2.4 Granger Causality test

The Granger Causality test developed by Granger (1969) was conducted to assess further the causal relationship between the KLCI spot market and FKLI futures market in the short run. The three measures of causality are unilateral, bilateral, and no causal link. The first unilateral is when there is only a one-way direction of a causal relationship in the middle of the two variables. Secondly, the bilateral relationship will only exist when there is a two-way causal relationship between variables. Meanwhile, there will be no presence of a relationship among the variables. The three measures summarize in the table below:

Unilateral relationship	One-way causal relationship	KLCI causes FKLI or FKLI causes KLCI
Bilateral relationship	Two-way causal relationship	KLCI causes FKLI AND FKLI causes KLCI
No causal relationship	No causal relationship	KLCI not causes FKLI AND FKLI not causes

 

 

The causal relationship can be estimated as follows:

The rejection to accept the null hypothesis in equation 1 signifies that futures prices lead and turn the spot market prices into lagging. This means the investor could use futures prices to predict the future movement of KLCI spot prices. Meanwhile, the rejection of equation 2 indicates vice versa, where spot prices lead to futures prices and futures prices are lagging; hence, the spot market plays a significant price discovery function; ultimately, the futures market’s role is rejected. Nevertheless, if both equations (1) and (2) are rejected, there will be no lead-lag variable relationship between both markets, which denotes a bilateral causality relationship. This means that there is a two-way causing one another relationship between each variable.

The statistical Chi-square  test will be used to verify the causal relationship. The granger causality test is a statistical hypothesis test to determine if one variable affects the other with the null “X does not Granger cause y” or “y does not Granger cause x” with i in the equation representing the number of lag lengths following the lag length selection criteria. The alternate hypothesis to test the causality can be expressed as below:

3.3 Summary

This chapter explained the overall research design for this paper with the main intention to study the lead-lag relationship between FTSE Bursa Malaysia KLCI Index (FBM KLCI) and FBM KLCI Futures (FKLI) in the Malaysia market during the Covid 19 crisis period precisely from MCO phase towards Recovery phase. Various kinds of tests such as the Unit Root test, Lag order selection criteria, Vector Error Correction Model, Johansen’s cointegration test, and Eagle will be conducted to investigate the lead-lag relationship.