Herding towards pygmalion: Examining the cultural dimension of market and bank based systems

Abstract

The study aims to not only detect the presence of herd behavior in the countries studied, but also to examine the effect of cultural dimensions and market/bank-based systems on the herding behavior of financial market investors. The study employs the Cross-Sectional Standard Deviation and Cross-Sectional Absolute Deviation methods to analyze daily data from public companies traded in the capital markets in Emerging Seven and Group of Seven economies. The results suggest that being a member of E7-G7, a Future Oriented (FO), and a Performance Oriented (PO) cultures are the most important factors in explaining herd behavior. Additionally, the study found that the Ridge Classifier and CatBoost Classifier algorithms arethe most superior model for estimating herd behavior periods determined by the CSSD and CASD models, respectively. The feature selection results show that the Assertiveness (A) in-group collectivism (GC) are the three most important explanatory factors of the herd behavior.

1. Introduction

The Pygmalion Syndrome depicts situations that bear similarities to the fictional town in John Steinbeck's 'East of Eden', where the residents followed each other blindly, has been observed in the financial markets. The herding behavior of investors, driven by the cultural dimension of a market and bank based system, has been the cause of many market crashes throughout history. In this paper, we will delve into the intricacies of Pygmalion Syndrome and how the cultural dimension of a market and bank based system plays a crucial role in exacerbating herding behavior phenomenon ([1,2,3]).

This study critically examines the herding behavior in financial markets, particularly focusing on the Emerging Seven (E7) and Group of Seven (G7) economies. By exploring the interplay between cultural values and market or bank-based systems, we aim to unravel the complexities behind investor decisions that often lead to significant market fluctuations. Unlike traditional finance theories that assume investor rationality and efficient market pricing, we address the behavioral aspects that often defy these assumptions. Investors, varied in their access to information, risk preferences, and investment horizons, exhibit behaviors that challenge the bedrock of efficient market hypothesis ([4,5,6]).

There have been several instances in financial markets where traditional finance theories have been called into question. Here are a few examples: the Dutch Tulip Mania of the 1630s, the South Sea Bubble of 1720, the Mississippi Bubble of the early 18th century, the Wall Street Crash of 1929, Black Monday of 1987, the Dot-com Bubble of the late 1990s, the Global Financial Crisis of 2008, and the GameStop (GME) stock price surge in January 2021 ([7,8,9,10,[11], [12], [13], [14],15,16,17,18]). These events, marked by irrational exuberance and collective investor behavior, highlight the necessity of understanding the psychological and sociological factors at play.

The study introduces a pioneering approach to the analysis of herding behavior in financial markets, setting it apart from existing literature. Its novelty lies in the integration of comprehensive cross-country data, encompassing over 55 million stocks across 18 stock exchanges in 14 countries over two decades. This expansive dataset provides a global perspective, rarely achieved in prior studies. Moreover, the research uniquely combines traditional herding behavior analysis methods like CSSD and CASD with advanced machine learning algorithms, including Ridge and Catboost Classifiers. This innovative methodological blend allows for a more nuanced prediction and understanding of herding trends, influenced by cultural dimensions and market categories (E7-G7). Furthermore, the study delves into the impact of cultural factors on herding behavior in both bank-based and market-based systems, offering a fresh perspective on how cultural nuances shape financial market dynamics. This aspect of cultural influence represents a significant departure from the conventional focus of herding behavior studies, often limited to specific market conditions or geographic regions. By bridging the gap between cultural studies and financial market behavior, the paper contributes a unique lens to view and analyze global financial market trends. The findings of this research not only complement the existing literature but also pave the way for future investigations into the interplay between cultural elements and financial market dynamics in the era of digitalization and global interconnectedness.

The aim of this study is to provide valuable insights into the dynamics of herd behavior in the capital markets of 14 countries, belonging to the E7 (Emerging 7: China, India, Brazil, Mexico, Russia, Indonesia, Turkiye) and G7 (Group of 7: Germany, USA, UK, France, Italy, Japan, and Canada) economies. The Culture and System Based Approaches Theoretical Framework used in this study aims to elucidate the complex relationship between cultural norms, values, and beliefs and their influence on decision-making and behavior within financial markets. We explore how different cultural dimensions, such as individualism, power distance, and collectivism, along with the structural characteristics of market and bank-based systems, contribute to the phenomena of herding behavior. This approach offers a comprehensive understanding of the factors driving investor behavior in diverse economic contexts ([19,20,21]).

The purpose of this study is to not only detect the presence of herd behavior in the countries studied, but also to examine the effect of cultural dimensions and market/bank-based systems on the herding behavior of financial market investors. The study aims to identify the most important factors that explain herd behavior and to find the best models for estimating herd behavior periods using the CSSD and CASD baseline models.

The study employs the CSSD (Cross-Sectional Standard Deviation) and CASD (Cross-Sectional Absolute Deviation) baseline models to detect herding behaviour. The research will utilize the machine learning algorithms for estimating herd behavior periods determined by the CSSD model and the CASD model. The Culture and System Based Approaches Theoretical Framework aims to understand the interplay between culture and structural activity of markets.

The literature review section delves into previous studies and literature on culture and system based approaches, examining both the studies done in developing and developed countries. The cultural dimensions relevant to the study, including individualism, power distance, and collectivism, are discussed in detail. The models and methods section provides an overview of the various models and methods used in the study, including the Structural Activity Index and data, baseline models, and methods. The results section presents the findings of the study, along with statistical analysis, while the policy implications section discusses the impact of the study on organizational behavior and management practices, including recommendations for organizations and governments. The significance of the study is summarized in the conclusion, highlighting its relevance for future research and policy-making.

1.1. Literature review

Herding behavior in individuals has been recognized as far back as the works of Keynes (1936) [22] and Veblen (1899) [23], although they didn't use the term "herding." The theory of herding behavior gained prominence in the 1990s, driven by Banerjee (1992) [24] and Bikhchandani et al. (1992) [25], however, these works did not explicitly examine herding behavior in the context of capital markets. There have been numerous studies on herding in the capital market, but our understanding of it remains limited, and the current theories do not fully explain it. In recent times, herding behavior has been differentiated from informational cascades. The concept of informational cascades was introduced by Bikhchandani et al. (1992) [25], but research has not yet clearly distinguished between the two and they are often used interchangeably. Informational cascades refer to a situation where individuals make decisions based on the observed actions of others, rather than on their own private information. Although there is overlap between these two concepts, they are not identical. Informational cascades focus on the role of information and its influence on decision-making, while herding behavior highlights the social dynamics of group behavior and imitation ([26]).

Before delving into the literature review on the manifestation of herding behavior across different country groups and sociological variations, it is beneficial to explore the biological and psychological theoretical foundations of herding behavior. In the following sections, the discussion will pivot to the debates surrounding limited rationality in behavioral finance. We will then examine the amalgam effect and the Pygmalion syndrome, shedding light on how they underpin bubbles in financial markets. The interaction of herding behavior with cultural and market-based systems will also be explored.

Subsequently, in alignment with the research focus of this study, we will investigate the occurrences of herding behavior in both developed and developing countries. The potential interactions between the included cultural dimensions and herding behavior will be elucidated, drawing upon existing literature. This comprehensive approach aims to provide a nuanced understanding of the multifaceted nature of herding behavior in diverse economic and cultural contexts.

The exploration of herding behavior's historical roots and its distinction from informational cascades sets the stage for a deeper analysis into the psychological and economic paradigms that shape financial decision-making. Moving from the general concept of herding, we now delve into the evolution of economic thought, particularly focusing on the transition from the traditional view of the rational, informed homoeconomicus to the more nuanced perspectives offered by behavioral finance.

1.1.1. Reviving homoeconomicus: the role of bounded rationality, intuition, and behavioral finance theories

In neoclassical economic theories, people are often portrayed as ideal decision-makers with complete rationality, perfect access to information, and consistent, self-interested goals. The term homoeconomicus originated from an essay by John Stuart Mill in 1836 in which he described a being who desires to possess wealth and is capable of judging the comparative efficacy of means for obtaining that end. The defining traits of homoeconomicus include maximizing profit, flawless rationality, unlimited cognitive capacity, perfect information, narrow self-interest, and preference consistency ([27,28]).

Bounded rationality is a concept in economics and decision theory that suggests that individuals and organizations do not always make perfectly rational decisions. Instead, they may make decisions based on limited information and cognitive biases. The concept was first introduced by Herbert Simon in the 1950s ([[29], [30],31,32]).

Behavioral finance uses psychology and sociology to understand how investors make decisions and affect market prices, challenging the traditional economic assumption of homo oeconomicus as the ideal model of human behavior. Leibenstein's revolutionary theory in 1976 shook the foundations of traditional economics by proposing that homooeconomicus is not the ideal model of human behavior, but rather a rare and exceptional manifestation that only surfaces under extraordinary circumstances. Bounded rationality seeks to bridge traditional and behavioral finance theories by acknowledging that investors' rationality is limited by cognitive and emotional constraints ([33,[34], [35]]).

According to De Bondt et al. (2008) behavioral finance is divided into three main categories: a catalog of biases, the speculative dynamics of asset prices in global financial markets, and how decision processes shape decision outcomes. They believed that behavioral finance is based on three main building blocks: sentiment, behavioral preferences, and limits to arbitrage ([36,37,38]).

Having discussed the limitations of traditional economic models and the emergence of behavioral finance, we now turn our attention to specific phenomena that exemplify these theoretical shifts. The Amalgam Effect represents a complex interplay of biases, anomalies, and syndromes, including the Pygmalion Syndrome, that contribute to financial market fluctuations. This section will explore how these elements collectively influence market dynamics, often leading to financial bubbles and crashes.

1.1.2. The amalgam effect: how bias, anomalies, Pygmalion Syndromes, and a blend of events fuel financial crashes or bubbles

Anomalies and biases are both important concepts in the study of financial markets and investor behavior. Anomalies refer to deviations from what is expected or predicted by a particular theory or model, such as the efficient market hypothesis (EMH), which states that financial markets are always perfectly efficient, but which has been shown to be inaccurate in some instances. Bias, on the other hand, refers to a systematic error or deviation from the truth in a measurement, sample, or experiment. Behavioral biases refer to the systematic ways in which people deviate from rational decision-making, such as overconfidence, confirmation bias, and herding behavior. Herding behavior has been linked to several cognitive biases, such as availability bias and confirmation bias, as well as social influences such as groupthink and information cascades. These biases and behaviors can lead to anomalies in financial markets, as investors deviate from rational decision-making and follow the actions of others in the market. This can create market inefficiencies and opportunities for profit, which is at odds with the efficient market hypothesis ([39,40,4,41,5,42,43,6,44]).

Bias, anomalies, syndromes, amalgam of events can all contribute to the formation of bubbles or collapse and price bubbles in 4C financial markets, i.e. commodity markets, cash (money markets), crypto markets, and capital markets. Bias can lead to irrational exuberance or overconfidence in certain investments, which can fuel a bubble. Anomalies, or deviation from normal market behavior, can also indicate the presence of a bubble or impending crash. Syndromes (pygmalion) refer to recurring patterns of behavior that lead to financial crises, such as excessive leverage or inadequate risk management. Amalgam of events refers to the combination of various events, from economic indicators to political changes, that can contribute to the formation of a bubble or crash. These factors can all interact and exacerbate each other, leading to a crash or bubble in the financial market. Understanding the relationships between these factors can help to identify potential warning signs and mitigate the impact of future financial crises. Herding behavior is at the heart of our understanding of mass movements of rational and irrational senses ([45,46])

Some important psychological and biological components of herding behaviours: mirror neurons, risk and return optimization, less cognitive effort, group belonging, mirror neurons, empathy, imitation behavior, swarming, stigmergy, and herd intelligence, belief in group purposes, less data collection efforts, market sentiment, market news, risk and return optimization, easy money, following a leader or group, common purposes over risk return trade off, imitation of others investment behaviours, Pygmalion Syndrome, risk and uncertainty, cut of trade off between price and intrinsic value, and market crashes and bubbles. The phenomenon in which people's expectations of others influence their behavior, which can contribute to herding behavior is known as Pygmalion Syndrome. People may follow others in order to minimize risk and uncertainty, this can lead to herding behavior, the risk and uncertainty play a crucial role in herding behavior. Intrinsic value is the inherent value of an asset based on its underlying fundamentals, such as earnings, assets, and growth prospects. When investors engage in herding behavior, they may ignore intrinsic value and instead focus on the actions of other investors. This can lead to a mispricing of assets, as market prices become disconnected from their intrinsic value. As more and more investors flock to a particular investment or market, the potential for a bubble to form increases ([47,48,4,1,2,49,50,51,52,53,54,6,55,56,57,58,59,60])

The echoes of bubbles and crashes history remind us that herding behavior can lead to market bubbles and subsequent crashes. It is important to note that herding is not always negative and can also lead to undervalued stocks rally.

Recognizing the influence of psychological and biological factors on herding behavior leads us to consider the broader cultural and systemic contexts in which these behaviors occur. The following section examines how cultural norms and systemic structures shape, and are shaped by, herding tendencies. This exploration reveals the intricate interplay between individual behaviors and the larger societal and economic systems.

1.1.3. Culture and system based approaches: theorotical framework

Culture and system based approaches are theoretical frameworks that attempt to understand the relationship between cultural values, norms and practices and the systems, institutions and organizations in which they exist. In this framework, the focus is on how cultural norms, values and beliefs influence decision-making and behavior within societies, and how societies can promote cultural diversity and inclusion.

According to Kahn (2004), the individual exhibits herding behavior when he avoids acting independently of the group with whom he is socially interacting ([19,20,21]). According to Scharfstein and Stein (1990), on the basis of herding behavior, the belief of "ordinary" investors that the buys and sells of major investors, who are thought to have more valuable information, and that stock prices are the consequence of these movements. This irrational trading scenario can be explained by psychological biases such as overreacting or underreacting to information, loss and risk aversion, and misjudgments by other investors ([61]). These group members' social life and aspirations are commonly similar. To offer an example from daily life, herding behavior is exhibiting the same consumption patterns as individuals with identical demands in our surroundings or exhibiting a behavior that attracts our attention through various social media platforms. In reality, all of these behaviors may be linked to the third and fourth levels of Maslow's Hierarchy of Needs. The demand for respect at the fourth level, as well as a sense of belonging at the third level, play a part in the establishment of herd behaviors ([62,63]).

The bank-based view highlights the benefits of banks in terms of acquiring information about firms, managing risk and mobilizing capital. On the other hand, it also points out the shortcomings of market-based systems, such as lack of incentives for investors to acquire information and reduced corporate control. On the other hand, the market-based view emphasizes the growth-enhancing role of functioning markets in research incentives, corporate governance, and risk management. However, it also highlights the problems of banks such as stifling innovation and impeding efficient corporate governance. The relationship between financial structure and systemic risk and finds that market-based financial structures are more resilient to systemic risk compared to bank-based financial structures ([64,65,[66], [67]]). There has been a long-standing debate on the merits of bank-based and market-based financial structures, with the earlier literature finding that the degree of financial development and liberalization is more important than the type of financial structure. In a market-based system, participants in the stock market, such as investors and traders, make decisions based on the information they have and the expectations they form about future market conditions. This can lead to herding behavior, where participants follow the actions of others rather than making independent decisions, which can result in market inefficiencies and heightened volatility ([68,69,[70], [71], [72]]).

Building on the understanding of cultural and systemic influences on herding behavior, we now shift our focus to the empirical evidence of herding in different economic contexts. This section compares herding behavior in emerging and developed markets, highlighting the varying degrees and manifestations of such behavior in response to diverse market structures and cultural backgrounds.

1.1.4. Emerging and developed countries

Studies suggest that herding behavior is observed in various financial markets, with more occurrences in emerging markets with low trading volumes and high volatility ([73])Herding behavior affects perceived market efficiency and increases market volatility, leading to a negative impact on individual investors but a positive impact on mutual funds ([74]). Institutional herding behavior is investigative and unintentional, with higher instances in developed markets compared to developing markets and more pronounced in informationally transparent markets ([75]).

Lobão and Serra (2007) [76] suggest that incomplete regulatory frameworks, specifically poor information transparency, cause herding behavior in developing capital markets. Herding is more common in less mature markets and decreases as markets become more mature. The study found higher levels of herding on the buy side and that herding is less common in funds with diverse portfolios. Balcilar et al. (2014) [77] found evidence of herding in all Gulf Arab stock markets, with market volatility being the most impactful factor governing switches between non-herding and herding states. The study also established a link between volatility and herd behavior, and showed that global variables such as US stock market performance, oil price, and interest rates play a significant role in herding. Litimi et al. (2016) [78] investigated the impact of herding on market volatility and bubble formation in the US stock market at a sectoral level. The study revealed that herding affects bubbles in some sectors but not in all, and that herding behavior leads to higher trading in specific stocks and therefore higher stock volatility while decreasing overall market volatility in large markets.

Multiple studies have investigated the existence of herding behavior in various stock markets. Some of the models used in these studies include Christie and Huang (1995) [79], Hwang and Salmon (2004) [80], and Chang et al. (2000). These studies analyzed from various time periods and regions, including the Chinese, Vietnamese, Taiwanese, Greek, Spanish, Portuguese, Italian, Saudi Arabia, Russian, and Turkish stock markets, as well as the commodity market. The results of these studies show that herding behavior is present in some markets, but not in others. Additionally, demographic factors, market sentiment and corruption perception can also play a role in herding behavior ([81,82,83,84,85,86,87,88,89,90]).

The examination of herding behavior across various market types naturally leads to a discussion on the role of specific cultural dimensions in influencing these behaviors. This final section delves into the nuanced ways in which cultural traits, such as power distance and uncertainty avoidance, shape the propensity for herding, offering a comprehensive view of the multifaceted nature of this phenomenon in the global financial landscape.

1.1.5. Cultural dimensions

The GLOBE study (Global Leadership and Organizational Behavior Effectiveness) is a large-scale cross-cultural research project that has investigated the relationship between culture and organizational behavior. The study included data from over 17,000 managers in over 951 organizations and 62 societies. The study has identified nine cultural dimensions that are believed to be universal and relevant to organizations ([91]).

Cultural dimensions refer to the ways in which culture can influence behavior, and herding behavior refers to the tendency for individuals to follow the actions of others, particularly when it comes to making decisions.

Assertiveness (A): refers to the degree to which individuals express their opinions and feelings directly and confidently in social interactions. According to French et al. (1959) [92], assertiveness is related to expert power, which involves influencing others through one's knowledge or skills. The article suggests that assertiveness can be developed through training and practice and is a positive quality for individuals. It is also a key component of effective leadership and is positively related to social influence ([93,94]).

Institutional Collectivism (IC): Refers to the degree to which individuals in a culture value and support the institutions and organizations that make up society. Cultures high in IC tend to have a strong sense of loyalty to organizations and a belief in following rules and procedures. Institutional collectivism motivates individuals to make sacrifices for group benefit ([74,57,58,59])

In-Group Collectivism (GC): Refers to the degree to which individuals in a culture value and support their in-groups, such as family and close friends. Cultures high in GC tend to prioritize group harmony and cooperation. In-group collectivism is positively associated with herding behavior, the tendency to follow others' decisions over independent choices. This behavior may occur more frequently in financial markets as individuals conform to group norms for the benefit of the group. In-group collectivism is linked to 'ingroup bias,' the preference for one's own group over others, which can also contribute to herding behavior ([95,96,97,98])

Future Orientation (FO): Refers to the degree to which individuals in a culture value and plan for the future. Future orientation refers to individuals' motivation to achieve future goals, sacrificing present gains to achieve them. Herding behavior refers to individuals' tendency to follow the actions of others instead of making independent choices. Individuals with higher future orientation are less likely to conform and engage in herding behavior as they tend to make decisions based on their own goals ([99,100,101])

Gender Egalitarianism (GE): Refers to the degree to which individuals in a culture believe in equal rights and opportunities for men and women. Cultures high in GE tend to have more equality between the sexes. Traditional gender roles and stereotypes may impact herding behavior, with men being more likely to engage in herding due to societal expectations of risk-taking, while femininity may be associated with greater conformity to group norms ([19,102,100,103,104,105,106,107,90]).

Humane Orientation (HO): Refers to the degree to which individuals in a culture value and support the well-being of others. Cultures high in HO tend to be more caring and compassionate.

Performance Orientation (PO): Refers to the degree to which individuals in a culture value and strive for success and achievement. Cultures high in PO tend to be more competitive and driven to succeed. Performance orientation can have both positive and negative effects on herding behavior. While those with a high performance orientation are less likely to engage in herding behavior, they may rely on the decisions of others in situations where they perceive a lack of information or expertise ([48,108,109]).

Power Distance (PD): Refers to the degree to which individuals in a culture accept and expect unequal distribution of power and status. Cultures high in PD tend to have a more hierarchical social structure. Power distance refers to the unequal distribution of power within a culture, and high power distance cultures tend to have more herding behavior in financial markets. Individuals in these cultures are more conformist and less likely to question authority. However, the relationship between power distance and herding behavior may be moderated by other cultural dimensions such as long-term orientation ([110,111])

Uncertainty Avoidance (UA): Refers to the degree to which individuals in a culture feel uncomfortable with uncertainty and ambiguity. Cultures high in UA tend to have more rigid rules and regulations and a lower tolerance for risk. High uncertainty avoidance is characterized by a need for structure, stability, and predictability, and individuals with high uncertainty avoidance tend to engage in more herding behavior. The relationship between uncertainty avoidance and stock market participation is mediated by loss aversion ([112,102,113,114]).

In summary, cultural dimensions may play a role in herding behavior, but it's important to consider other factors and to be cautious when making generalizations about cultural groups. Also, cultural dimensions are not fixed and may change over time, and are not limited to a geographic region. It's important to note that there are other factors that may influence herding behavior as well, such as market conditions, investor sentiment, and information availability. Researchers have also identified various antecedents that can increase herding behavior such as, information overload experience, social pressure, lack of confidence in one's own abilities, transparency, accurate information, critical thinking and independent decision-making.

Table 1 provides a comprehensive overview of three distinct studies on herding behavior in financial markets. The compilation covers varied aspects including herding in a small European market, the development of a herding model to explain market dynamics, and gender differences in herding behavior. Each study is analyzed based on its authors, title, objectives, methodology employed, key findings, and the broader implications of these findings. This systematic literature review table is designed to offer a concise yet thorough understanding of the evolving research in herding behavior within financial contexts, highlighting diverse methodologies, focal points, and conclusions drawn in recent studies.

Table 1.

Systematic literature review on herding behavior in finance.

Authors	Title	Objective	Methodology	Key Findings	Implications
Chang and Lin (2015)	The effects of national culture and behavioral pitfalls on investors' decision-making: Herding behavior in international stock markets	The impact of national culture and behavioral errors on investor decision-making processes in international stock markets	Data analysis, experimental design	Herding behaviors are common in countries with Confucian culture and less developed stock markets. Certain national culture indices are closely related to the display of herding behaviors. Investors' behavioral errors dominate their herding tendencies.	Demonstrates the significant impact of national cultural factors and behavioral errors on investor decision-making processes and market efficiencies. Offers a new perspective in understanding behavioral models in financial markets.
Ahmad and Wu (2022)	Does Herding Behavior Matter in Investment Management and Perceived Market Efficiency? Evidence from an Emerging Market	The impact of herding behavior on investment decisions, performance, and perceived market efficiency among individual investors actively trading in the PSX	Survey, cross-sectional design, data analysis using SEM	Herding behavior has a significant negative impact on perceived market efficiency and positively affects the decision-making processes of individual investors.	Enhances understanding of the role of herding behavior in investment management and perceived market efficiency. Makes a significant contribution to the literature in this area, particularly for emerging markets.
Baddeley et al. (2012)	Herding in Financial Behaviour: A Behavioural and Neuroeconomic Analysis of Individual Differences	The herding behavior in financial decisions, analysis of individual differences, and neuroeconomic aspects	Behavioral experiments and neuroimaging techniques	It was found that social information influences herding behavior, and individuals tend to follow others' actions despite their private information. Herding behavior varies among individuals and is associated with personality traits such as gender, age, impulsivity, adventurousness, empathy, and extroversion. Herding decisions show a strong correlation with activations in the amygdala, a brain region associated with social decision-making and emotional processing.	Emphasizes the role of individual differences in financial decision-making processes and their impact on herding behaviors. Points to the role of the amygdala in modulating the impact of social information on financial decisions. Demonstrates the importance of neuroeconomic approaches in understanding financial behaviors.
Chang, Cheng, Khorana (1999)	An examination of herd behavior in equity markets: An international perspective	Investigating investor behaviors and herd behavior in different international markets (USA, Hong Kong, Japan, South Korea, Taiwan)	Analysis of stock return distributions based on market returns using nonlinear regression analysis	No evidence of herd behavior was found in the USA and Hong Kong. Partial herd behavior in Japan, and significant herd behavior in South Korea and Taiwan were documented. The rate of increase in return dispersion during market rises is higher than during market falls.	Indicates that systemic risk is more considered in emerging markets (South Korea, Taiwan) than firm-specific macroeconomic information. Impacts on investment behaviors and market dynamics in international markets. Emphasizes the effects of herd behavior on investment decisions and market efficiency.
Chen et al. (2014)	Contrarian Strategy and Herding Behaviour in the Chinese Stock Market	To investigate the presence and impact of contrarian strategy and herding behaviour in the Chinese stock market	Analysis of 300 stocks in the Chinese stock market from 2007 to 2012, regression models	Widespread presence of contrarian strategy and herding behaviour; contrarian strategy increases market returns, while herding decreases market volatility and liquidity	Emphasizes the importance of promoting contrarian strategies and limiting herding behaviour to enhance market efficiency and stability in China
Choi & Skiba (2015)	Institutional Herding in International Markets	To investigate the presence, reasons, and consequences of institutional herding behavior in international markets.	Analyzing stock ownership data of institutional investors across 41 countries to measure herding tendencies. Testing the impact of herding behavior on price stability. Examining the relationship between herding behavior and information asymmetry in target countries. Testing whether prices adjust more quickly in countries with lower information asymmetry.	Found widespread herding behavior in international markets. Herding behavior stabilizes prices and is based on fundamental information. More prevalent herding in countries with lower information asymmetry, where price adjustments are faster.	Suggests that institutional herding in international markets is based on fundamental information, is unintentional, and makes markets more efficient. Highlights the importance of understanding the decision-making processes of institutional investors who hold a significant share in global portfolio investments.
Clements, Hurn, & Shi (2017)	An empirical investigation of herding in the U.S. stock market	To investigate the presence of herding behavior in the U.S. stock market, expanding traditional regression approaches to a vector autoregressive framework using a Granger causality test.	Uses daily data of 30 components of the Dow Jones Industrial Average from 2003 to 2016. The method incorporates a Granger causality test to examine changes in causality relationships and accounts for macroeconomic news announcements and firm-level news flow.	No evidence of herding for the entire sample using traditional methods. Time-varying Granger causality test shows presence of herding at the onset of economic crises like the subprime mortgage crisis, European debt, U.S. debt ceiling crises, and the 2015 Chinese stock market crash.	Suggests that herding behavior in the stock market is a periodic phenomenon, particularly evident at the onset of economic crises. Highlights the importance of considering temporal variations in market dynamics when studying herding behavior.
Economou et al. (2011)	Cross-country effects in herding behaviour: Evidence from four south European markets	To test the existence of herding effects in the Portuguese, Italian, Spanish, and Greek markets and examine potential asymmetries of herding effects with respect to market return sign, trading activity, and volatility.	Using a survivor-bias-free dataset of daily stock returns from January 1998 to December 2008 for analysis.	Comprehensive evidence of herding effects in all four South European markets. Found that cross-sectional dispersion of returns in one market is affected by the cross-sectional dispersion of returns in the other three markets.	Significant implications for financial stability in the Eurozone and international portfolio diversification. Herding behavior became more intense during the 2007–2008 global financial crisis.
Ferreruela & Mallor (2021)	Herding in the bad times: The 2008 and COVID-19 crises	To investigate herding behavior in Spanish and Portuguese markets during the 2008 crisis and COVID-19 pandemic, and under different market conditions.	Employed CSSD and CSAD measures to assess herding intensity in rising vs. declining markets and during high vs. normal volatility days. Also, examined cross-country herding effects between Spain and Portugal.	Found evidence of herding in both markets before and after the 2008 crisis, but not during the crisis. In the COVID-19 period, herding was not generally detected in Spain but was observed in Portugal. Asymmetries in herding behavior were seen in rising and falling markets, with stronger herding during bearish days.	Highlights the different reactions of investors to various types of crises. Indicates that herding is influenced by market conditions and volatility. Shows the importance of considering geographical proximity and market interdependence in understanding herding behavior.
Gavrilakis & Floros (2021)	The impact of heuristic and herding biases on portfolio construction and performance: the case of Greece	To examine the influence of heuristic and herding biases on portfolio construction and performance among investors in Greece.	A structured questionnaire distributed to active private investors and finance professionals; analyzed using multiple regression analysis.	- Heuristic biases positively influence private investors' portfolio satisfaction. Overconfidence significantly impacts portfolio satisfaction.Herding biases do not significantly affect private investors. In finance professionals, heuristic biases positively affect, while herding biases negatively affect portfolio satisfaction.	Highlights the importance of understanding behavioral biases in investment decision-making Useful for investors and finance professionals in improving portfolio management strategies.Suggests the need for further research in behavioral finance, particularly in volatile markets.
Júnior et al. (2019),	Analyzing herding behavior in commodities markets – an empirical approach	To examine beta herding in the commodities market using a state-space model methodology.	Utilizing the Hwang and Salmon (2004) methodology and a standardized beta adaptation by Hwang, Rubesam, and Salmon (2018). Analysis of fifteen commodities from 2000 to 2018, focusing separately on food commodities.	- Betas may deviate from fundamentals in commodities.Food commodity betas tend to revert faster to stability, leading to long-run equilibrium in risk-return factors.	Indicates significant herding behavior in commodities markets. Suggests the potential for market inefficiencies and price distortions due to herding. Proposes further research on low-beta anomaly in commodities markets.
Klein (2013)	Time-variations in herding behavior: Evidence from a Markov switching SUR model	To test for time-variations in herd behavior in stock markets and analyze investor behavior during market turmoil and tranquil periods.	Employed a Markov switching seemingly unrelated regressions (MSSUR) model to analyze US and Euroarea stock markets from July 2001 to June 2011.	- Herding behavior is more pronounced and persistent during periods of high volatility. There are significant herding spillovers between the US and Euroarea markets during times of crisis.	Highlights the increased influence of behavioral effects on stock prices during financial crises. Suggests the need for a deeper understanding of market dynamics and investor behavior during turbulent times.
Liang (2011)	A Neural Basis of Herd Behavior in Stock Market: An Experimental Design	To investigate the neural basis of herd behavior in stock trading under both normal and extreme market conditions.	Proposes an experimental design using neuroimaging to study brain activity during stock trading, particularly focusing on the anterior insula and medial prefrontal cortex.	Predicts significant activation in the anterior insula and medial prefrontal cortex during dramatic and smooth price changes, respectively. Also, anticipates different brain region activations based on market conditions and investor behavior.	Offers insights into the neural substrates underlying investors' strategic reasoning and herd behaviors in different market conditions, contributing to the understanding of investor sentiments and irrational behaviors in finance.
Lobão & Maio (2021)	Herding around the World: Do Cultural Differences Influence Investors' Behavior?	Explore the impact of cultural differences on investors' decision to herd	Empirical analysis using Hofstede's cultural dimensions across 39 countries (2001–2013)	Masculine cultures and cultures with higher power distance are less prone to herd. Individualism, uncertainty avoidance, and long-term orientation have non-significant impact	Highlights the importance of cultural factors in financial decision-making and herding behavior in stock markets
Nasarudin et al. (2017)	Investigation of Herding Behaviour in Developed and Developing Countries: Does Country Governance Factor Matters?	Examine the influence of countries' governance on herding decisions among investors	Cross-sectional absolute deviation (CSAD) method applied to data from 60 countries	Strict governance minimizes herding activity. Herding is more prevalent in countries with moderate or weak governance. Information dissemination efficiency due to governance level affects herding behavior	Highlights the importance of governance in financial markets, suggesting that better governance can reduce herding behavior and improve market stability. Suggests the need for improved transparency and information dissemination to minimize herding in financial markets.
Schmitt & Westerhoff (2017)	Herding behaviour and volatility clustering in financial markets	To explore the impact of herding behavior on the dynamics of financial markets, specifically how it contributes to volatility clustering	Development of a financial market model where speculators follow a mix of technical and fundamental trading rules influenced by herding behavior; the method of simulated moments is used for model estimation	Herding behavior intensifies during periods of heightened market uncertainty.	The study highlights the crucial role of herding behavior in influencing market dynamics, especially in terms of volatility.
				Increased synchronization of speculators' trading behavior leads to significant price changes and prolonged periods of volatility.	The findings suggest that understanding and modeling herding behavior can provide deeper insights into market fluctuations and potential instability.
				The model effectively replicates various stylized facts of financial markets, including bubbles, crashes, excess volatility, fat-tailed return distributions, and volatility clustering.	The model's robustness to various modifications indicates its potential utility in analyzing financial market behavior under different scenarios.
Vieira & Pereira	Herding Behaviour and Sentiment: Evidence in a Small European Market	The study aims to analyze herding behavior in the Portuguese stock market (PSI-20) from 2003 to 2011. It focuses on understanding the intensity of herding behavior and its relationship with investor sentiment in a small and illiquid capital market.	Herding Intensity Measurement: Two approaches were used to measure herding intensity. The first, based on Patterson and Sharma (2006), examines intraday order sequences. The second, following Chang et al. (2000) and Christie and Huang (1995) [79], uses cross-sectional standard deviations (CSSD).	Herding Intensity: The study found negative and significant herding intensity across all types of sequences (up, down, and zero runs), indicating systematic imitation among investors.	The study highlights the need for a methodological rethink in measuring herding behavior as different approaches yield different conclusions. It also opens avenues for future research to explore herding behavior during different economic cycles, the influence of other factors beyond sentiment on herding, and contagion of herding behavior at an international level.
			Investor Sentiment Measurement: European Economic Sentiment Indicator (ESI) was used to represent investor sentiment.	Method Sensitivity: Results indicated that herding evidence is sensitive to the measurement method. While Patterson and Sharma's approach suggested systematic herding, the CSSD approach did not find evidence supporting herding behavior.
			Data Source: Data on PSI-20 prices and constituent securities from NYSE Euronext Lisbon for 2003–2011 were used, comprising 2308 daily closing prices.	Sentiment and Herding Relationship: The regression model showed a negative relationship between investor sentiment and herding. However, the Granger causality test suggested causality from sentiment to herding only in neutral position scenarios.
			Analytical Approach: The study conducted statistical analyses, including regression models and Granger causality tests, to investigate the relationship between herding intensity and investor sentiment.
Zheng et al. (2021)	Gender and Herding	To examine the herding behavior of Chinese individual investors with a focus on gender differences.	- Analysis of data from a large anonymous Chinese brokerage firm.LSV method for daily herding measurement. Individual-level herding measurement constructed following Merli and Roger (2013). Use of buy–sell imbalance as an alternative measure for correlated trading behavior.	Females are more inclined to follow 'same-sex' investors. Both genders herd more intensively in bull markets and on stocks with better liquidity and larger market capitalization. Female investors generally yield lower returns than males when herding intensively, especially during bull markets.	Highlights the influence of gender on investment behavior. Suggests the need for tailored investment strategies considering gender differences in herding behavior. Indicates the impact of market conditions on herding behavior and the resultant investment returns, particularly among female investors.

2. Models and methods

There are several models for detecting herding behavior in stock data:. Those are correlation analysis ([115,12]), volatility clustering ([116,117]), Granger causality test ([118,119]), Hirschman Herfindahl Index (HHI) ([118,119]), agent-based Modeling ([120,121]).

Our research is primarily motivated by the intriguing yet unexplored facets of herd behavior dynamics in financial markets that extend beyond the scope of predator-prey models. In contrast to the ecological focus of studies such as those by Ghanbari and Djilali (2020) [122] [123] and Djilali (2018, 2019) [124] [125], which explore herd behavior within predator-prey interactions and spatial diffusion dynamics, our investigation pivots this concept into the realm of financial markets. We employ a multidisciplinary approach, integrating principles from behavioral finance, cultural studies, and economic theory. This shift from biological models to socio-economic frameworks is driven by our motivation to understand the underlying cultural and economic factors that influence herd behavior in financial markets.

Our methodology comprises an extensive literature review and comparative analysis, aiming to elucidate how various cultural dimensions interact with market-based systems to shape investor behavior. This approach diverges significantly from the ecological and biological perspectives of previous studies. By bridging the gap between cultural psychology and economic behavior, our study seeks to offer novel insights into the complex dynamics of financial decision-making, thereby contributing to the field of behavioral finance.

In summary, our research stands apart by focusing on the socio-economic implications of herd behavior, exploring how cultural values and market structures impact collective behaviors in economic settings. This represents a relatively uncharted territory in behavioral finance, distinct from the biological perspectives. We aim to fill a gap in the current understanding of how non-biological factors, such as cultural norms and market dynamics, influence herd behavior in financial contexts.

2.1. Model

2.1.1. Data, baseline Models and methods

The daily data between January 1st, 2000 and December 31st, 2020 of over 40,000 stocks, totalling over 55 million, from the stock markets in 14 countries belonging to the G7 and E7 group (which includes Germany with 2 markets, England with 2 markets, Italy with 1 market, France with 1 market, Canada with 1 market, the USA with 3 markets, Mexico with 1 market, Indonesia with 1 market, Russia with 1 market, Turkey with 1 market, China with 2 markets, Japan with 1 market, India with 2 markets, and Brazil with 1 market), was analyzed to determine herd behavior motives.

The Cross-sectional Absolute Deviation (CASD) and Cross-sectional Standard Deviation (CSSD) are two techniques that help to determine the existence of herding behavior in stock market data.

CASD involves the calculation of the average deviation of individual stock returns from the mean return of all stocks in a specific time frame. If the CASD is high, it suggests that investors are following different investment strategies and not herding. However, if the CASD is low, it could indicate that investors are following similar investment strategies and hence, herding.

CSSD involves the calculation of the standard deviation of individual stock returns from the average return of all stocks in a specific time frame. If the CSSD is high, it suggests that investors are following different investment strategies and not herding. On the other hand, if the CSSD is low, it could indicate that investors are following similar investment strategies and herding.

CASD and CSSD can both be used to determine herding behavior in stock market data by analyzing the deviation of individual stock returns from the mean return of all stocks. However, CSSD is more widely used since it is more sensitive to changes in the distribution of returns and can provide more accurate results.

It's important to keep in mind that both CASD and CSSD are built on the premise that herding behavior can be identified through the similarity of returns among stocks. If stock returns are similar, it could suggest the presence of herding behavior.

Christie and Huang (1995) [79] Cross Sectional Standard Deviation (CSSD):

$$\text{CSSD}=\mathrm{a}+{\mathrm{b}}_1\text{UP}+{\mathrm{b}}_2\text{DOWN}+\mathrm{\epsilon }$$

Separate extreme returns defined as 1 %, 5 %, or 10 % for the right and left tails of the distribution of returns are calculated, and the UP and DOWN variable takes the value of "1″ when the market return has extreme return defined as 1 %, 5 %, or 10 %, while it takes the value of "0″ in the case of normal returns that are outside of the extreme returns defined as 1 %, 5 %, or 10 % of the distribution of returns. A statistically significant and negative value of the b₁ and b₂ coefficients, in other words, an increase in the volatility of the market returns clearly up and down, means a decrease in the CSSD value indicates the presence of rational movements in the market, while the presence of an opposite relationship, i.e. a positive and significant value of b₁ and b₂, indicates the presence of herd behavior ([126,127]).

Chang et al. (2000) [128] Cross-sectional Absolute Deviation (CASD):

$$\text{CASD}=\mathrm{a}+{\mathrm{b}}_1{\mathrm{r}}_{\mathrm{m}}+\mathrm{b}2\left|{\mathrm{r}}_{\mathrm{m}}\right|+\mathrm{b}3{\mathrm{r}}_{\mathrm{m}}+\mathrm{\epsilon }$$

In the model, r_m represents the market return, |r_m| represents the absolute value of the market return, and r_m² represents the square of the market return. A statistically significant and negative value of the coefficient of the non-linear return b₃ indicates the presence of herd behavior, while the opposite sign indicates rational behavior in an efficient market, just like in the CSSD model ([82,129]).

2.1.2. structural activity index

The Structural Activity Index (SAI) is a measure of the level of economic activity in a country or region. It is calculated by measuring a variety amount of indicators such as of new construction, building permits, and other indicators of economic activity. There are two main types of SAI: market-based and bank-based.

A market-based SAI is calculated by measuring the level of economic activity in the stock market, such as stock prices and trading volume. A bank-based SAI is calculated by measuring the level of economic activity in the banking sector, such as loan activity and deposit growth.

The SAI is a composite index and is calculated based on several economic indicators that are selected and weighted by the organization that calculates the index. In general, the SAI can be a useful tool for assessing the level of economic activity in a country or region and can be used to predict future economic trends. However, it is important to note that the SAI is a composite index and should be used in conjunction with other economic indicators and analysis.

It measures the activity volume of capital markets relative to banks. To measure the activity volume of the capital markets, the ratio of the total value of the shares cleared in the stock markets to GDP is used. This ratio measures the volume of market changes over total economic activity. In our study, the market capitalization values of the year consisting of the sum of the relevant trading days were used. The activity level of banks is measured by the banking credits ratio, calculated as the credits provided by commercial and specialist banks to the private sector divided by GDP. This Index does not take into account credits given to the government ([130]).

$$\mathrm{Y}-\mathrm{F}\text{Index}=\ln ((\text{Market}\text{Cap}/\text{GDP})/(\text{Private}\text{Sector}\text{Credit}/\text{GDP}))$$

The Index is determined by the logarithm of the ratio of traded shares to banking credits. If the Index value is positive, it indicates that the financial system is market-based, with the fraction of traded shares to banking credits exceeding 1.

2.2. Methods

Our study's methodology stands out for its comprehensive approach, utilizing CSSD and CASD models on an extensive dataset covering more than 55 million stocks in 18 stock exchanges across 14 countries over a two-decade period (2000–2020). This broad dataset allowed for an in-depth analysis of herding behavior across various market systems and cultural dimensions. Significantly, our study delved into the nuances of herding behavior during different market conditions, particularly highlighting years of high herding incidence such as during the global financial crisis and European debt crisis. The application of machine learning algorithms, notably the Ridge and Catboost Classifiers, to predict herding behavior based on cultural dimensions and market categories (E7-G7) is a novel approach. This methodology not only identified patterns of herding behavior but also unraveled the influence of cultural factors on such behaviors in both bank-based and market-based systems.

In contrast, other studies in the field often employed a more focused approach. Some studies concentrated on investor behaviors and herd behavior in specific international markets like the USA, Hong Kong, Japan, South Korea, and Taiwan, using nonlinear regression analysis to examine stock return distributions. Their approach was in geographical scope and didn't extensively incorporate cultural dimensions as a variable. Similarly, some other studies focused on herding effects in specific European markets, using datasets limited to certain time frames and regions. These studies typically employed traditional regression models and didn't integrate advanced predictive analytics or machine learning tools.

The key distinction lies in the scope and technological integration of our methodology. While other studies provide valuable insights within their specific contexts or regions, our research offers a more global and technologically advanced perspective. By incorporating a vast array of data and employing machine learning algorithms, our study not only complements the existing literature but also expands the understanding of herding behavior's complexity and its relation to cultural and systemic factors. This approach allows for a nuanced understanding of herding behavior that transcends traditional market analysis, offering valuable implications for investors and policymakers in diverse financial landscapes.

Although machine learning algorithms have introduced some more comprehensive and innovative estimation techniques that are alternative to econometric models, the prediction applications of econometric models for parameters have lost their place in machine learning model prediction applications. In addition, it has been shown that machine learning models outperform econometric models with wider prediction horizons ([38,131]).

Table 2 presents the key features, advantages, and disadvantages of the 15 different machine learning algortihms used in the study. All algorithms have certain advantages and disadvantages in terms of factors such as sample size, statistical distribution, linear and non-linear relationships. In this regard, the performance analysis results for all algorithms in terms of comprehensiveness are shared in the results section.

Table 2.

Key features of Random Forest, Gradient Boosting, CatBoost, XGBoost, and LightGBM.

Model	Key Features	Advantages	Disadvantages
Random Forest	Ensemble of decision trees	Handle non-linear relationships, handle large datasets, reduces overfitting	Slower training time, more complex interpretation
Gradient Boosting	Ensemble of decision trees using boosting	Handle numerical and categorical features, handle non-linear relationships, high prediction accuracy	Slow training time, prone to overfitting, complex interpretation
CatBoost	Gradient boosting with categorical feature support	Handling categorical features without one-hot encoding, handling missing values, automated model tuning, fast parallel processing	Steep learning curve, slower training time
XGBoost	Gradient boosting for decision trees	Fast training speed and high prediction accuracy, handles missing values and non-linear features, parallel and GPU processing, many tuning parameters	Difficult interpretation, prone to overfitting, high memory usage
LightGBM	Gradient boosting with selective sampling of high gradient instances	Fast training performance, good prediction accuracy, low memory usage	Not suitable for complex models
Decision Tree Classifier	Uses a tree-like structure to represent decisions and their possible consequences. Each node represents a feature, and the branches represent the outcomes of that feature.	Easy to interpret and visualize. Can handle both numerical and categorical data. Can handle non-linear relationships between features and class labels. Can handle missing values and irrelevant features.	Prone to overfitting if the tree is too deep and has too many branches. Can be computationally expensive for large datasets. Can be biased towards features with a large number of outcomes. Can suffer from unstable performance if the data is noisy or has small variations.
AdaBoost Classifier	Combines multiple simple models (also known as weak learners) to make a strong model. Assigns weights to instances based on their difficulty to predict correctly.	Simple and easy to implement. Can handle both linear and non-linear problems. Can handle noisy and imbalanced data. Can improve the performance of the weak learners by combining them.	Can be sensitive to noisy data and outliers. Can be computationally expensive for large datasets. Can be vulnerable to overfitting if the weak learners are too complex.
Extra Trees Classifier	An ensemble method based on decision trees. Creates multiple random trees and combines their predictions.	Can handle high-dimensional data and noisy data. Can handle missing values and irrelevant features. Can be used for both classification and regression problems.	Can be computationally expensive for large datasets. Can suffer from overfitting if the number of trees is too large. Can be sensitive to the choice of hyperparameters.
Logistic Regression	A statistical method for solving classification problems. Fits a linear model to the relationship between the input features and the class labels.	Simple and easy to implement. Can handle large datasets efficiently. Can provide an estimate of the probability of the class labels. Can handle both linear and non-linear relationships between the input features and class labels.	Can be sensitive to irrelevant features. Can suffer from overfitting if the model is too complex. Can be vulnerable to outliers and noisy data.
Ridge Classifier	Linear regression with L2 regularization penalty term	Handles multicollinearity, provides stability to coefficients	Difficult to find optimal alpha value, not well-suited for large number of predictors or non-linear relationships
K-Nearest Neighbor (KNN)	Non-parametric supervised learning algorithm	Simple, efficient, sensitive to local structure of data	Can be computationally expensive, may suffer from the curse of dimensionality
Linear Discriminant Analysis (LDA)	Method for finding a linear combination of features that separates classes	Good for dimensionality reduction and classification	Assumes normality and equal covariance matrix across classes, may not perform well if classes are not well-separated
SVM with Linear Kernel	Support Vector Machine with linear kernel	Robust and efficient classifier, can handle non-linearly separable data with feature transformation	Sensitive to choice of kernel, may not perform well if classes are not linearly separable
Naive Bayes	Simple and easy to implement, fast and efficient for large datasets, works well with high-dimensional data, can handle both continuous and categorical data.	Simple and easy to implement, fast and efficient for large datasets, works well with high-dimensional data, can handle both continuous and categorical data.	Independence assumption is often not true, can be sensitive to irrelevant features, may not perform well for data with complex relationships between features.
QDA	A statistical linear discriminant analysis method used for binary and multi-class classification, allows each class to have its own covariance matrix.	Can capture non-linear relationships between features and classes, can model the different covariance structures of the classes, effective when the number of features is small compared to the number of samples.	Can be computationally expensive, requires a large number of samples to avoid overfitting, may perform poorly when the number of features is large compared to the number of samples.

Notes: Models evaluation: Bentéjac and Martínez-Muñoz (2021) [132]; Böhning (1992) [133]; Breiman (1996:2001) [134] [135],Chen and Guestrin (2016) [136]; Cortes and Vapnik (1995) [137]; Cover and Hart (1967) [138]; Freund and Abe (1999) [139]; Friedman (2001) [140]; Geurts et al. (2006) [141]; Hodges (1950) [142]; Ke et al. (2017) [143]; Leung (2007) [144]; Peng and Cheng (2007) [145]; Prokhorenkova et al. (2018) [146]; Rish (2001) [147]; Schapire (2003) [148]; Swain and Hauska (1977) [149]; Tharwat (2016) [150]; Vezhnevets and Vezhnevets (2005) [151].

3. Results

The herd behavior tendencies detected by using CSSD and CASD models on daily data between January 1st, 2000 and December 31st, 2020 for more than 55 million stocks in 18 stock exchanges of 14 countries are as follows.

Graph 1 shows that in almost all of the graphs that present the stock price charts and the years in which herd behavior was detected in parallel, it has been observed that there were significant corrections in the stock exchanges during the herd behavior periods. One of the observations is that the upward or downward trends continued to strengthen after the corrections.

Graph 1.

CSSD and CASD Herding Behavior Detected Years

Herding behavior was observed in many stock markets over 21 years and was particularly high in the years 2008–2009, 2014–2015. These years coincide with the global financial crisis and the European debt crisis, suggesting that market uncertainty and heightened risk perception may have driven herding behavior (see Table 3).

Table 3.

Stock markets and herding behavior detected years.

NAME	YEAR
Türkiye BIST 100	2004-2010-2011-2014-2015
USA Nasdaq	2000-2001-2002-2004-2005-2008-2009-2013-2016-2018-2020
USA NYSE	2006
USA S&P	2000-2002-2004-2007-2008-2009-2010-2011-2014-2016
UK FTSE	None
Germany Frankfurt	2000-2002-2003-2005-2007-2011-2014-2016-2018-2019
France CAC 40	2000-2005-2006-2007-2008-2011-2016
China Shenzen	None
China Shangai	None
Brazil Bovespa	2003-2006-2007-2012-2014-2017-2018
Mexico Bolsa Mexıcana De Valores	2002-2005-2006-2007-2012-2014-2015-2016-2017-2018
Japan Nikkei 225	2000-2001-2002-2003-2006-2017
Canada S&P/TSX	2001-2002-2004-2005-2006-2007-2008-2009-2010-2011-2012-2013-2014-2015-2016-2017-2018-2019-2020
Indonesia Jakarta Composite	2000-2002-2004-2011-2016-2019
Italy FTSE_All	2006-2009-2010-2016
India NSE	2000-2001-2002-2005-2007-2009-2011-2012-2013-2014-2015-2016-2020
India NIFTY	2001-2002-2003-2004-2008-2009-2010-2011-2012-2013-2014-2015-2016-2017-2018-2019-2020
Russia Moex	2000-2004-2006-2011-2014-2017-2018

Table 3 offers a comprehensive view of herding behavior across various global stock markets over specific years. This table is pivotal in understanding the temporal and geographical spread of herding behavior in significant financial markets.

Diverse Temporal Patterns: The table reveals that herding behavior is not confined to specific global events but rather occurs sporadically across different time frames. For instance, the USA Nasdaq showed herding behavior in a range of years from 2000 to 2020, indicating that herding is a persistent phenomenon in this market, potentially driven by various economic, political, and technological changes over the years.

Geographic Variability: The presence or absence of herding behavior across different markets highlights the geographic variability in investor behavior. Notably, the UK FTSE and China's Shenzhen and Shanghai markets showed no herding behavior during the study period, suggesting either more individualistic investment decisions or differing market dynamics compared to other regions where herding was more prevalent.

Economic Crises and Herding: The occurrence of herding behavior in certain years aligns with global and regional economic crises, such as the 2008–2009 global financial crisis. This alignment suggests that herding behavior may be a response to market uncertainty and volatility, where investors tend to follow market trends or the actions of perceived successful investors.

Emerging vs. Developed Markets: The table also provides insights into herding behavior in emerging markets (e.g., Brazil's Bovespa, India's NSE and NIFTY, Russia's Moex) compared to developed markets. The varied years of herding behavior in these markets could reflect the different stages of market development, regulatory changes, and the entry of new investors with diverse risk appetites.

Longitudinal Trends: Some markets, like Canada's S&P/TSX and India's NIFTY, show a more continuous pattern of herding behavior over the years. This continuity could be indicative of underlying structural factors in these markets that consistently drive investors towards collective behavior.

Market-Specific Factors: Each market's herding years could be reflective of specific local or regional events, economic policies, or investor sentiment shifts unique to those markets. For example, the herding behavior in Germany's Frankfurt market in the early 2000s might be connected to the economic conditions in the Eurozone during that period.

In conclusion, Table 3 offers invaluable insights into the dynamics of herding behavior across major global stock markets. It underscores the importance of contextual factors, including economic conditions, market maturity, and cultural influences, in shaping investor behavior. This table is a significant contribution to understanding the global patterns of herding behavior in stock markets, providing a foundation for future research to delve deeper into the causes and implications of such behavior in different market environments.

The countries with a significant number of occurrences of herding behavior include the USA (Nasdaq and NYSE), France (CAC 40), Brazil (Bovespa), Japan (Nikkei 225), Canada (S&P/TSX), India (NSE and NIFTY), and Russia (Moex). On the other hand, countries such as Turkey (BIST 100), the UK (FTSE), Germany (Frankfurt), China (Shenzen and Shanghai), Indonesia (Jakarta Composite), and Italy (FTSE All) have had fewer occurrences or none at all.

Fig. 1 provides the “Simultaneous Herding Among Stock Markets": this term refers to the occurrence of similar investment behaviors and decisions among investors in multiple stock markets at the same time. This phenomenon is often driven by a variety of factors such as market rumors, news, and media reports, as well as emotional and psychological biases. Simultaneous herding can lead to market inefficiencies and even market crashes, as investors collectively drive up the prices of certain assets while selling others. It is important for market participants to understand the potential impact of simultaneous herding and to make investment decisions based on their own thorough research and analysis.

Fig. 1.

Simultaneous of herding among stock markets.

Additionally, factors such as increased access to information and communication technology, which allows for rapid dissemination of information, and the growing influence of social media, which can amplify market sentiment, may have also contributed to simultaneous herding behavior. Furthermore, psychological biases, such as the tendency to follow the crowd and a lack of confidence in one's own opinions, could also play a role in herding behavior in financial markets.

In the context of the research, this network graph strengthens the argument that herding behavior is not an isolated event but a global phenomenon influenced by cultural, economic, and systemic factors. It also suggests the need for a holistic approach in studying financial markets, taking into account the complex interplay between cultural dimensions, structural activities, and categorizations such as E7 and G7, which can all be critical in shaping investment behaviors and patterns.

Inter-market Connectivity: The dense web of connections indicates a high level of interdependency among international stock markets. This suggests that investor behavior in one market could have a ripple effect across others, potentially leading to simultaneous herding behavior.

Market Influence: The network reflects the varying degrees of influence certain markets have on others. Markets with a larger number of connections, such as the United States' Nasdaq or S&P, may serve as barometers or trendsetters for global herding behavior, given their significant impact on global financial dynamics.

Crisis Propagation: During times of financial uncertainty or crisis, this network could facilitate the rapid spread of herding behavior, where fear or optimism in one market quickly transmits to others, amplifying market movements and volatility.

Investor Sentiment Transmission: The network may also indicate the transmission of investor sentiment, where positive or negative news in one market can sway investor perceptions and actions in others, resulting in a cascade of herding behavior that transcends borders.

Information Flow: The interconnectedness highlighted in the figure underscores the role of information flow in herding behavior. With the advent of global news and social media, information can disseminate swiftly, leading to synchronized investment decisions across different markets.

Policy Implications: Such a network can have significant policy implications. Regulators and policymakers may need to consider the global implications of domestic financial policies, as decisions in one market can have unforeseen consequences due to the interconnected nature of global finance.

Strategic Investment: For investors, the visualization underscores the importance of considering global market trends in investment strategies. An event in one country can affect market sentiment elsewhere, and thus, a well-informed global perspective is crucial for effective portfolio diversification and risk management.

We found that the cultural dimensions have an impact on herd behavior in stock markets across different countries and market systems. The effective cultural dimensions in the matching of years and stock markets where herd behavior is not detected are Performance Orientation, In-Group Collectivism, Future Orientation and Humane Orientation. On the other hand, where herd behavior was detected, the effective cultural dimensions were found to be Performance Orientation, Future Orientation, In-Group Collectivism, and Humane Orientation. The findings were consistent across countries that belong to categories E7 and G7, and across bank-based and market-based market systems (see Fig. 2, Fig. 3, Table 4, Table 5) (Trommsdorff et al., 1982; Yamagishi et al., 1998; Bikhchandani and Sharma, 2000; Ro and Gallimore, 2014; Baddeley et al., 2014; Chang and Lin, 2015; Chen et al., 2018; Rhode, 2018; Ahmad and Wu, 2022; Ahmed et al., 2022; Jannati et al., 2020; Chan et al., 2022) (see Table 6).

Fig. 2.

CSSD Model: Cultural Values, Country Category, and Economic Activity

Fig. 3.

CASD Model: Cultural Values, Country Category, and Economic Activity

Table 4.

Effective cultural dimensions, country category and economic activity.

Configurations/Cultural Values	A	IC	GC	FO	GE	HO	PO	PD	UA
E7, BANK, CSSD=None	3,98	4,79	5,45	5,36	4,21	5,34	5,71	2,71	5,11
E7, BANK, CSSD = 1+	3,98	4,79	5,45	5,36	4,21	5,34	5,71	2,71	5,11
E7, MARKET, CSSD=None	4,13	4,75	5,40	5,38	4,25	5,31	5,75	2,68	4,99
E7, MARKET, CSSD = 1+	3,73	4,81	5,48	5,57	4,40	5,30	5,75	2,59	4,89
G7, BANK, CSSD=None	4,05	4,53	5,68	5,38	4,95	5,57	6,06	2,79	4,15
G7,BANK, CSSD = 1+	4,22	4,36	5,64	5,36	4,93	5,54	6,06	2,80	4,02
G7,MARKET, CSSD=None	3,82	4,69	5,78	5,40	5,01	5,62	6,08	2,77	4,25
G7,MARKET, CSSD = 1+	3,86	4,74	5,91	5,35	4,88	5,75	6,12	2,85	4,19
E7, BANK, CASD=None	3,98	4,79	5,45	5,36	4,21	5,34	5,71	2,71	5,11
E7, BANK, CASD = 1+	4,03	4,79	5,40	5,30	4,15	5,38	5,66	2,71	5,10
E7,MARKET, CASD=None	3,99	4,79	5,36	5,52	4,42	5,27	5,69	2,56	4,59
E7,MARKET, CASD = 1+	3,99	4,79	5,36	5,52	4,42	5,27	5,69	2,56	4,59
G7, BANK, CASD=None	4,07	4,50	5,69	5,38	4,96	5,57	6,07	2,80	4,14
G7,BANK, CASD = 1+	4,08	4,54	5,68	5,40	4,93	5,58	6,06	2,79	4,15
G7,MARKET, CASD=None	3,84	4,77	5,81	5,46	4,96	5,65	6,09	2,76	4,27
G7,MARKET, CASD = 1+	3,86	4,74	5,91	5,35	4,88	5,75	6,12	2,85	4,19

Notes: Cultural values are presented based on average values from the year and country match.

Table 5.

Summarizing of findings for effective cultural dimensions, country category and economic activity.

Country Category	Market System	Herd Behavior (CSSD)	Average Cultural Dimensions	Effective Cultural Dimensions
E7	Bank-based	Not detected	A = 3.97, IC = 4.79, GC = 5.45, FO = 5.36, GE = 4.20, HO = 5.34, PO = 5.71, PD = 2.71, UA = 5.11	Performance Orientation, In-Group Collectivism, Future Orientation, Humane Orientation
E7	Bank-based	Detected	A = 3.97, IC = 4.79, GC = 5.45, FO = 5.36, GE = 4.20, HO = 5.34, PO = 5.71, PD = 2.71, UA = 5.11	Performance Orientation, In-Group Collectivism, Humane Orientation, Future Orientation
E7	Market-based	Not detected	A = 4.12, IC = 4.74, GC = 5.40, FO = 5.37, GE = 4.25, HO = 5.30, PO = 5.74, PD = 2.78, UA = 4.19	Performance Orientation, In-Group Collectivism, Future Orientation, Humane Orientation
E7	Market-based	Detected	A = 3.73, IC = 4.81, GC = 5.48, FO = 5.57, GE = 4.39, HO = 5.30, PO = 5.75, PD = 2.59, UA = 4.89	Performance Orientation, Future Orientation, In-Group Collectivism, Humane Orientation
G7	Bank-based	Not detected	A = 4.04, IC = 4.52, GC = 5.67, FO = 5.37, GE = 4.95, HO = 5.57, PO = 6.06, PD = 2.79, UA = 4.15	In-Group Collectivism, Humane Orientation, Performance Orientation, Gender Egalitarianism, Future Orientation
G7	Bank-based	Detected	A = 4.22, IC = 4.36, GC = 5.64, FO = 5.35, GE = 4.92, HO = 5.53, PO = 6.05, PD = 2.80, UA = 4.02	Performance Orientation, Humane Orientation, In-Group Collectivism, Future Orientation
G7	Market-based	Not detected	A = 3.82, IC = 4.69, GC = 5.78, FO = 5.40, GE = 5.00, HO = 5.62, PO = 6.08, PD = 2.76, UA = 4.24	Performance Orientation, In-Group Collectivism, Future Orientation, Gender Egalitarianism
G7	Market-based	Detected	A = 3.86, IC = 4.73, GC = 5.91, FO = 5.34, GE = 4.87, HO = 5.74, PO = 6.11, PD = 2.84, UA = 4.19	Performance Orientation, In-Group Collectivism, Humane Orientation, Future Orientation

Table 6.

CSSD model algortihms information criteria and ridge Algo 100 iteration results.

Model	Accuracy	AUC	Recall	Prec.	F1	Kappa	MCC
Logistic Regression	0.8126	0,6542	0,269	0,6667	0,3656	0,2898	0,3389
Ridge Calssifier	0.8126	0	0,269	0,6667	0,3656	0,2898	0,3389
Linear Discriminant Analysis	0.8126	0,6743	0,269	0,6667	0,3656	0,2898	0,3389
K Neighbors Classifier	0.8091	0,7271	0,3143	0,6633	0,4049	0,3119	0,3506
CatBoost Classifier	0.8091	0,7418	0,3619	0,615	0,4435	0,341	0,3631
Random Forest Classifier	0.7890	0,7252	0,381	0,5314	0,4332	0,3119	0,3226
Extreme Gradient Boosting	0.7749	0,7227	0,4262	0,5386	0,4627	0,3248	0,3357
Ada Boost Classifier	0.7682	0,6725	0,2381	0,4775	0,2828	0,1788	0,2078
Gradient Boosting Classifier	0.7682	0,6774	0,2857	0,3642	0,309	0,1999	0,2061
Light Gradient Boosting Machine	0.7649	0,7174	0,3	0,4292	0,3357	0,2114	0,2199
Extra Trees Classifier	0.7616	0,6768	0,3786	0,4267	0,3938	0,2525	0,2542
Naive Bayes	0.7379	0,6623	0,4405	0,4422	0,4221	0,2598	0,2693
Decision Tree Classifier	0.7370	0,6182	0,3762	0,4202	0,3797	0,2184	0,2272
SVM-Linear Kernel	0.6751	0	0,2	0,467	0,757	0	0
Quadratic Discriminant Analysis	0.5697	0,6945	0,6595	0,2936	0,4015	0,1434	0,1716
Ridge Classifier 1000 Iteration Result
	Accuracy	AUC	Recall	Prec.	F1	Kappa	MCC
Mean	0,8126	0	0,269	0,6667	0,3656	0,2898	0,3389
Standard Deviation	0,0421	0	0,1517	0,3249	0,1812	0,1739	0,1925

Table 4 can be invaluable for policymakers, economists, and market analysts seeking to understand the interplay between culture, economic systems, and investor behavior in global financial markets. It suggests that cultural norms and values should be considered when analyzing market movements and investor behavior, particularly in the context of globalized financial markets where behaviors in one region can influence outcomes in another. The table is organized into configurations that show the average values of these cultural dimensions within certain contexts—specifically, whether the countries fall under E7 or G7 categories, whether their markets are predominantly bank-based or market-based, and whether herding behavior is present.

Consistency Across Configurations: The cultural values seem to remain consistent within the same country category and economic activity, regardless of whether herding behavior is detected or not. This suggests that cultural dimensions are inherent to specific countries and economic systems and are not solely a reaction to herding behavior.

Impact of Economic Activity: The differences in cultural dimensions between bank-based and market-based systems are subtle yet noticeable. This might indicate that the type of economic activity influences how cultural dimensions manifest themselves in the context of herding behavior.

E7 vs. G7 Countries: There are variations in cultural dimensions between E7 and G7 countries. For instance, G7 countries tend to show higher assertiveness and performance orientation across both bank and market-based systems when herding is present, which might reflect a more aggressive economic behavior or a higher value placed on performance in these countries.

Detection of Herding Behavior: Interestingly, the detection of herding behavior does not drastically alter the cultural dimensions. This could imply that herding behavior is influenced by a complex interplay of factors, of which cultural values are only one part.

Cultural Values and Herding Behavior: Specific cultural dimensions such as Performance Orientation and Future Orientation seem to be relevant in the context of herding behavior. These dimensions may influence how investors anticipate future market trends and their performance, which could lead to herding under certain conditions.

Subtle Differences: Small differences in cultural dimensions between when herding is detected and when it is not could be essential. For instance, a slight increase in Assertiveness or a decrease in Uncertainty Avoidance might contribute to a more conducive environment for herding.

Implications for Market Dynamics: Understanding these cultural dimensions could have practical implications for market prediction, risk management, and crafting of investment strategies that account for the possibility of herding behavior.

Table 5 presents a synthesized view of the findings related to cultural dimensions, country categories, market systems, and the detection of herd behavior. It compares these aspects across E7 and G7 countries within bank-based and market-based systems, identifying which cultural dimensions are effective in each scenario. Table 5 provides valuable insights into how cultural values interact with economic systems and how this interplay can potentially affect market behaviors like herding. This table serves as a useful resource for financial analysts, investors, and policymakers to consider cultural factors when evaluating market dynamics and crafting strategies for international investment and cooperation.

E7 Countries - Bank-based Systems: Herd behavior's presence or absence doesn't appear to alter the average cultural dimensions, which suggests a stable cultural influence regardless of market behavior. The effective cultural dimensions in these scenarios emphasize performance, collectivism, future orientation, and humane orientation, indicating a strong inclination towards group harmony, long-term planning, and compassionate engagement within these cultures.

E7 Countries - Market-based Systems: There is a slight shift in cultural dimensions when herd behavior is detected, notably in Assertiveness (A) and Uncertainty Avoidance (UA). This could imply that when markets exhibit herd behavior, individuals from E7 countries may become less assertive and more accepting of uncertainty. The consistency of Performance Orientation and Future Orientation across both detected and undetected scenarios underscores the importance of achievement and forward-looking attitudes in E7 markets.

G7 Countries - Bank-based Systems: In G7 countries with bank-based systems, detected herd behavior corresponds with a slight increase in Performance Orientation and a decrease in In-Group Collectivism and Uncertainty Avoidance. This may reflect a shift towards individual performance during times of collective market movements. The presence of Gender Egalitarianism among the effective cultural dimensions in non-herding scenarios could suggest that equality between genders may contribute to a more stable market environment.

G7 Countries - Market-based Systems: The cultural dimensions shift slightly in the presence of herd behavior, with Performance Orientation and Future Orientation remaining consistently effective. This suggests that even in market-based systems, G7 countries prioritize performance and long-term perspectives during periods of herding.

Comparing E7 and G7: Between E7 and G7 countries, there's a noticeable difference in the effective cultural dimensions. G7 countries seem to place a higher value on Gender Egalitarianism in non-herding scenarios, which is not as prominent in E7 countries.

Cultural Dimensions Influence: The effective cultural dimensions such as Performance Orientation and Future Orientation are common across most scenarios, implying that these values significantly influence market behavior and investor decision-making.

Policy Implications: Understanding these cultural dimensions can be critical for policymakers and investors. For example, fostering environments that encourage long-term planning and performance may reduce susceptibility to herd behavior. Similarly, promoting gender egalitarianism could potentially contribute to market stability.

According to the results of the machine learning algorithm that included cultural dimensions, structural activity index, and E7-G7 categories for the years in which herd behavior was predicted with the CSSD baseline model, the Ridge Classifier algorithm was able to predict herd behavior with an accuracy of over 80 % compared to other algorithms. According to the feature importance results of the Ridge Classifier algorithm, the market's E7-G7 country categorical feature, cultural dimensions such as Future oriented, Performance oriented, Power distance, and Gender Egalitarianism, are highly important in predicting herd behavior (see Fig. 4, Fig. 6, Table 7, Table 8).

Fig. 4.

CSSD model algortihms information criteria and ridge algo 100 iteration results.

Fig. 6.

CASD model algortihms information criteria and catboost algo 100 iteration results.

Table 7.

CSSD model feature importances.

Variables	Feature Importance Value
G7E7	0,35
FO	0,29
PO	0,28
PD	0,23
GE	0,22
LNATCINDEX	0,12
UA	0,07
IC	0,05
MARKBANK	0,05
GC	0,05
A	0,05
HO	0,04

Table 8.

CASD model algortihms information criteria and catboost Algo 100 iteration results.

Model	Accuracy	AUC	Recall	Prec.	F1	Kappa	MCC
CatBoost Classifier	0,757	0,7179	0,4306	0,676	0,5183	0,3646	0,3866
Random Forest Classifier	0,7537	0,6985	0,4778	0,6583	0,5438	0,3792	0,3952
Ridge Calssifier	0,7402	0	0,3069	0,6783	0,4139	0,275	0,3149
Linear Discriminant Analysis	0,7368	0,6583	0,3625	0,6417	0,4553	0,2959	0,3228
Ada Boost Classifier	0,7367	0,6768	0,4264	0,6063	0,4803	0,3148	0,3323
Logistic Regression	0,7333	0,6726	0,2736	0,635	0,3758	0,2422	0,2775
Gradient Boosting Classifier	0,723	0,673	0,4042	0,5725	0,4653	0,2856	0,2983
Extra Trees Classifier	0,7092	0,6573	0,4764	0,5324	0,4981	0,2948	0,2989
Light Gradient Boosting Machine	0,7055	0,6632	0,3958	0,5601	0,4471	0,2554	0,2711
Naive Bayes	0,692	0,6231	0,4181	0,4975	0,4468	0,2362	0,243
K Neighbors Classifier	0,6889	0,6004	0,2833	0,4951	0,3518	0,1668	0,1823
Decision Tree Classifier	0,6785	0,6295	0,4875	0,4732	0,4755	0,2455	0,2487
Extreme Gradient Boosting	0,6617	0,6389	0,4181	0,4442	0,4247	0,1872	0,1906
Quadratic Discriminant Analysis	0,6071	0,7359	0,6917	0,4169	0,5185	0,2251	0,2422
SVM-Linear Kernel	0,4252	0	0,7	0,2128	0,3262	0	0
CatBoost Classifier 1000 Iteration Result
	Accuracy	AUC	Recall	Prec.	F1	Kappa	MCC
Mean	0,774	0,7061	0,4292	0,7444	0,5331	0,3972	0,4295
Standard Deviation	0,0773	0,093	0,1183	0,2355	0,1507	0,1889	0,2069

According to the results of the machine learning algorithm knowledge criteria which include cultural dimensions, structural activity index, and E7-G7 categories, the Catboost Classifier algorithm was able to predict the herd behavior in the relevant years with a accuracy above 75 % compared to other algorithms when using the CASD baseline model to predict herd behavior. The feature importance results of the Catboost Classifier algorithm show that the cultural dimensions of Assertiveness, Year Value, In-group Collectivism, Uncertainty Avoidance, and Human Oriented are highly important in predicting herd behavior in a market.

The findings (see Fig. 5, Fig. 7, Table 5, Table 9) show that cultural dimensions play a crucial role in shaping the market system and herd behavior in different countries. The countries studied can be broadly divided into two groups, the E7 and G7 countries. Both groups have both bank-based and market-based systems, and the findings indicate that the market system has a greater influence on cultural dimensions than the bank-based system. The effective cultural dimensions that have the greatest impact on the market system and herd behavior include Assrrtiveness, Performance Orientation, In-Group Collectivism, Humane Orientation, and Future Orientation. The study also found that the presence of herd behavior has a significant impact on cultural dimensions, with the effective cultural dimensions being different between countries where herd behavior is detected and where it is not. The findings reveal that cultural values and beliefs can have a profound impact on economic systems and behavior and therefore, should be considered when evaluating and understanding economic performance.

Fig. 5.

CSSD model feature importances.

Fig. 7.

CASD model feature importances.

Table 9.

CASD model feature importances.

Variables	Feature Importance Value
A	0,19
date	0,18
GC	0,14
UA	0,085
HO	0,075
GE	0,06
LNATCINDEX	0,06
IC	0,04
PO	0,04
PD	0,035
FO	0,024
G7E7	0,015
MARKBANK	0,01

4. Conclusion with policy implications

Our study offers a multifaceted exploration of herding behavior in financial markets, drawing parallels and establishing distinctions from the existing literature. One of the primary convergences between our research and previous studies, such as those by Economou et al. (2011) and Ferreruela & Mallor (2021), is the pronounced prevalence of herding behavior during periods of financial instability, particularly highlighted during the 2008–2009 global financial crisis and the European debt crisis. This collective observation emphasizes the critical influence of market uncertainty and heightened risk perception on investor behavior, affirming a recurring theme in the literature that financial crises act as catalysts for intensified herding.

Moreover, our research extends the discussion on the cultural impact on herd behavior, aligning with the insights provided by Chang and Lin (2015). Both our study and theirs underscore the nuanced interaction between national cultural dimensions and investor behavior, thereby affecting market dynamics. This cultural perspective enriches the understanding of herding behavior, demonstrating its variability across different geographic and cultural landscapes. However, while Chang et al. (2000) report an absence of herding in specific markets like the USA and Hong Kong, our research documents significant occurrences of herding across a diverse array of markets, including the USA. This discrepancy highlights the evolving nature of herding behavior across different temporal and spatial contexts.

A distinctive aspect of our study is the incorporation of machine learning algorithms and the emphasis on the influence of information technology on herding behavior. This modern approach marks a departure from traditional methodologies and offers a contemporary lens through which the effects of digitalization and information dissemination on investor decisions can be understood. Our findings suggest that technological advancements, particularly in information and trading platforms, significantly shape herding tendencies in today's financial markets.

In discussing the dual nature of herding behavior, our study provides a more comprehensive understanding than the predominantly risk-focused narrative prevalent in the literature. We acknowledge both the potential benefits, such as increased market participation and liquidity, and the risks, including market inefficiencies and volatility, associated with herding. This balanced view underscores the complexity of herding behavior, highlighting its multifaceted impact on financial markets.

Furthermore, our exploration into the roles of algorithmic trading and social media platforms in exacerbating herding behavior introduces a novel element to the discourse. This aspect reflects the changing landscape of financial markets, where digital platforms increasingly influence investor behavior, leading to new dynamics such as the amplification of herding through collective actions on social media and automated trading systems.

In conclusion, our study contributes a significant and contemporary perspective to the literature on herding behavior in financial markets. By integrating cultural aspects, technological advancements, and a balanced view of the effects of herding, our research not only resonates with existing findings but also pioneers new avenues of understanding in this ever-evolving field.

4.1. Policy implications

From the bustling markets of the unstable markets to the sprawling crisis of the stable markets, culture and systems play a critical role in shaping the investment environment we trade in. With an eye towards the future, researchers have delved into the intricate relationship between culture and systems, exploring the impact these forces have on everything from economic stability to individual behavior. Through rigorous study and a deep understanding of the cultural dimensions that define our markets, they have constructed theoretical frameworks and developed innovative models and methods to better understand this complex relationship. From the baseline models, like such as the CASD, CSSD, Structural Activity Index, that form the foundation of our understanding to cutting-edge techniques, like Machine Learning Methods, these researchers are dedicated to uncovering the secrets of our financial markets and charting a course towards a better future investment environment. The results of their labors hold the key to a more stable and equitable markets, where the forces of culture and systems can be harnessed for the betterment of us all.

Herding behavior in financial markets can have both positive and negative effects. One potential positive side of herding behavior is that it can provide a sense of security for investors. When investors see others making similar investments, they may feel more confident in their own investment decisions. This could lead to increased market participation and liquidity, which can benefit the overall market. Herding behavior can also lead to the efficient functioning of financial markets. In an efficient market, prices reflect all available information, and herding behavior can help to ensure that prices adjust quickly to new information. On the negative side, herding behavior can lead to market inefficiencies and bubbles. When investors blindly follow the actions of others rather than making independent decisions based on their own analysis, it can cause prices to deviate from their fundamental values. This can lead to market bubbles, where prices become unsustainable and eventually collapse. Herding behavior can also lead to increased volatility in financial markets. When a large number of investors all make the same trades at the same time, it can lead to sudden price movements, which can create uncertainty and instability in the market.

In addition, herding behavior can also lead to the rise of misinformation and manipulation in the market, as seen in the GME stock case. Social media platforms, such as Reddit and Twitter, have made it easier for investors to share information and coordinate their actions, which can lead to increased herding behavior. The GameStop (GME) stock price surge in January 2021 is an example of how social media can be used to coordinate buying and drive up stock prices. Algorithmic trading, which uses computer programs to execute trades based on pre-set rules, has also contributed to herding behavior in financial markets. Algorithmic trading systems can quickly and simultaneously execute trades based on the same set of rules, which can lead to sudden and large price movements. This can create a self-fulfilling cycle of buying and price increases, further fueling herding behavior.

Furthermore, algorithmic trading can also contribute to market inefficiencies and volatility, as the algorithms can be designed to react to certain market conditions, such as momentum or volatility, which can lead to increased herding behavior. In summary, social media and algorithmic trading have greatly increased the speed and ease of coordination and execution of trades, which has led to an increase in herding behavior in financial markets. This can lead to market inefficiencies, volatility and at times market manipulation as seen in the GME stock case.

It's important to note that herding behavior is often influenced by a variety of factors, including market sentiment, investor psychology, and the ease of access to information and trading platforms.

4.2. Limitations

This study, while providing valuable insights into herding behavior in financial markets, is subject to several limitations that merit consideration. Firstly, the analysis predominantly relies on historical data from 2000 to 2020, which, though extensive, may not fully capture the rapidly evolving dynamics of current financial markets. The rapid advancement of technology and changing investor demographics post-2020 could introduce new behavioral patterns not encapsulated in this timeframe. Additionally, the use of machine learning algorithms, while innovative, depends heavily on the quality and completeness of the input data. Any inherent biases or gaps in the dataset could potentially skew the results. Furthermore, the study's focus on specific cultural dimensions, though informative, may not encompass all the socio-economic factors that influence herding behavior. Aspects such as political climate, regulatory changes, and global economic events, which can also significantly impact market behavior, were not explicitly incorporated into the analysis. Moreover, while the study spans a broad range of markets, including E7 and G7 countries, it may not fully represent the nuances of smaller or less-developed financial markets, which could exhibit distinct herding patterns. Finally, the theoretical frameworks and models used, though robust, might not fully account for the complex and often irrational nature of human decision-making in financial contexts. Addressing these limitations in future research could provide a more comprehensive and nuanced understanding of herding behavior in global financial markets.

Data availability

Data associated with our study has not been deposited into a publicly available repository. However, data will be available up on request.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Disclosure statement

The authors report there are no competing interests to declare.

CRediT authorship contribution statement

Ömür Saltik: Writing – original draft, Methodology, Data curation, Conceptualization. Wasim Ul Rehman: Writing – review & editing, Resources, Investigation, Conceptualization. Türker Kaymaz: Visualization, Software, Investigation, Data curation. Suleyman Degirmen: Writing – review & editing, Supervision, Software, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.