The aim of this paper is to propose a model for cultural finance and to develop theory-based hypotheses on stock market investing. In contrast to the inductive financial research, a deductive approach is offered here to connect widely-accepted behavioral hypotheses on the stock market.
Understanding behavioral influences on an investor’s decision-making surprisingly has not been driven much by the acknowledgment of the mediating role of culture. While behavioral finance criticizes excessive simplifications regarding an investor’s behavior according to the homo oeconomicus, it makes the unrealistic assumption that actors exhibit universal biases and equally apply heuristics when facing different choices of action.
This paper addresses cultural finance as an important background variable and suggests a conjoint effect with behavioral finance. This means that the culture can enhance, decrease or reverse biases and heuristics which are still mostly examined in the United States and only replicated in western countries. The paper is encouraged to implement cultural finance as a future research field.
Table of Contents
1 Introduction
1.1 Overview on Topic and Motivation
1.2 Approach and Structure
2 Cultural finance as Research Field
2.1 Cultural Dimensions
2.2 Behavioral Biases and Heuristics
2.3 Relevance and Framework of cultural finance
3 Culture and Behavioral Finance
3.1 Risk
3.1.1 Attitudes towards risk
3.1.2 Risk perception and preference
3.1.3 Co-movement
3.2 Cross-sectional Returns
3.3 Style Investing
3.3.1 Momentum
3.3.2 Size
3.3.3 Value
3.3.4 Liquidity
3.3.5 Defensive
3.3.6 Carry
3.3.7 Resume and Remarks
4 A Paradigm Shift? - Discussion on cultural finance
5 Conclusion
List of References
List of Figures
Figure 1: Ten Fields consistent with Cumulative Prospect Theory
Figure 2: Iceberg concept of Culture
Figure 3: Framework for cultural finance
Figure 4: Descriptive Statistics on IPO
Figure 5: Results for the accuracy of forecasts
Figure 6: Map of Investment Styles and its Cultural Background
Figure 7: Dimension maps of the world
1 Introduction
1.1 Overview on Topic and Motivation
Understanding behavioral influences on an investor’s decision-making surprisingly has not been driven much by the acknowledgment of the mediating role of culture. While behavioral finance criticizes excessive simplifications regarding an investor’s behavior according to the homo oeconomicus, it makes the unrealistic assumption that actors exhibit universal biases and equally apply heuristics when facing different choices of action. This paper addresses cultural finance as an important background variable and suggests a conjoint effect with behavioral finance. This means that the culture can enhance, decrease or reverse biases and heuristics which are still mostly examined in the United States and only replicated in western countries. The paper is encouraged to implement cultural finance as a future research field.
“It is naive to assume that such [financial] goals are culture-free. (...) The finance function has been the last stronghold in business administration to escape crosscultural analysis”. Hofstede (2001, p. 385)
As stated by Hofstede himself, the impact of research on culture has been weak for financial issues although it has become established in other business disciplines such as organizational theory, management & leadership, accounting, marketing, entrepreneurship or economic development. What makes this kind of research dominant in business fields, lies in the nature of culture itself. Hofstede (1983, p. 75) defines culture as a collective mental programming that shapes thinking, feeling, and acting in such way that it distinguishes the members of one group or from others. As culture is based on several fields like anthropology, sociology, and cross-cultural psychology, its application generates insights which still are not fully covered by traditional research. Also, collective mental programming means that cultural measures exhibit strong explanatory power between members of different groups1 (Aggarwal et al., 2016, p. 467). While behavioral finance primarily examines how cognitive and social factors influence financial decision-making, cultural finance analyzes how values and attitudes of (national) cultures do. Since culture shapes the way of thinking, feeling and acting, one might argue that culture forms the base of cognitive and social factors.
The application of culture on finance faces some challenges which might explain its underpart. First, unlike financial measures like stock returns, which are based on numerical observations, culture is a difficult-to-define construct based on subjective answers mainly driven by interviews or questionnaires. Second, the combination of culture and finance might face methodological problems such as endogeneity. While reverse causality is a minor problem due to the fact that cultures remain stable even when economic conditions are changing, spurious relationships or autocorrelation and errors in measurement remain problematic. Spurious relationships can be addressed by including country-level controls, autocorrelation by watchful choice of variables and errors in measurement by the right execution of the study.2 The most common challenge of cultural finance lies in establishing the theoretical link. Most of current studies are empirical and, if they have a theoretical framework, it is focused on direct effects of culture, but ignore the question about the background through which culture affects financial decisions3 (Aggarwal et al., 2016, p. 467). Breuer/Quinten (2009, p. 13f.) criticize as well besides the small amount of contributions and the missing theoretical background the argumentation of most papers. They argue that there is less argumentation on the way culture works (direct influence channels), but more on the results which means that there is no orientation to implement and, thus, no justification to see cultural finance as fully developed research field. Williamson (2000) formed a model of economic institutions incorporating culture as an informal institution. In his model culture does not play a big role, but still gives a theoretical link for further empirical work.
1.2 Approach and Structure
The aim of this paper is to propose a model for cultural finance and to develop theory-based hypotheses on stock market investing. In contrast to the inductive financial research, a deductive approach is offered here to connect widely-accepted behavioral hypotheses on the stock market. The paper aims to identify possible cultural proxies for the strength and sign of behavioral biases. The next step is to propose plausible (causal) relationships to robust styles. Based on this, the last step focuses on the generation of hypotheses and the development of a model incorporating cultural finance in the current finance research.
Cultural finance can be explained in every discipline of finance; however, the focus lies on portfolio management and asset pricing while existing literature of culture on corporate finance is partly implemented. In this context, the paper is dedicated to identifying different trading patterns depending on cultural dimensions.
The remaining parts of this paper are organized as follows. Section 2.1 briefly describes the definition and application of widely known cultural dimensions, whereas 2.2 recapitulates a selection of behavioral biases and heuristics used by investors to gather and process information as well as to decide between different financial alternatives. Afterwards, in 2.3 the framework of cultural finance within current portfolio management and asset pricing is modelled. Section 3.1 deals with the findings on cross-cultural variations in risk attitudes, perception and preferences and additionally on the tendency for stocks to co-move. In 3.2 the selected cultural impacts on cross-sectional returns are presented to lead over to the core of this paper dealing with different investment styles in 3.3. Within chapter 3 many findings are connected to other sections to synthesize the discussion in chapter 4 whether cultural finance introduces a paradigm shift. Lastly, chapter 5 offers concluding remarks.
2 Cultural finance as Research Field 2.1 Cultural Dimensions
This paper examines several cultural dimensions by different authors. Among the authors in cultural sciences the best-known is Geert Hofstede who first conducted surveys, collected and analyzed data in 40 countries with over 116,000 participants in total (Hofstede, 1983, p. 4). Unless stated otherwise, the explanations of Hofstede’s cultural dimensions in this paper rely on his books “Cultures Consequences” (2001) and “Software of a mind” (2010). His maps on the cultural dimensions of the world are presented in the appendix.
Most subsequent research dealt especially with the cultural dimensions of individualism and collectivism which he defines the following way:
“Individualism pertains to societies in which the ties between individuals are loose: everyone is expected to look after him- or herself and his or her immediate family. Collectivism as its opposite pertains to societies in which people from birth onward are integrated into strong, cohesive in-groups, which throughout people’s lifetime continue to protect them in exchange for unquestioning loyaltyHofstede (2001b, p. 92)
Cultures of western countries like the United States (U.S.), Australia or Great Britain are prone to individualism, cultures of eastern countries like China or South Korea rather to collectivism. Although replications have different statistical approaches, this dimension is originally determined by an index of 14 statements regarding work and work environment (Hofstede et al., 2010, p. 35).4 Factor analysis leads to scores which are multiplied by 25 and added to 50 to generate a standardized value range from 0 to 100 (Hofstede et al., 2010, p. 93).
Next, uncertainty avoidance has proven itself to be an important cultural dimension. Hofstede (2001, p. 146) argues that societies have adapted in different ways to coping with basic uncertainty in every day’s life. The three manifestations of these ways include technology (against uncertainties caused by nature), law (caused by humankind) and religion (caused by other powers). Thus, uncertainty avoidance is defined as the extent to which the members of a culture feel threatened by ambiguous or unknown situations (Hofstede et al., 2010, p. 191). The index originally consists of three questions based on rule orientation (Q1), employment stability (Q2) and stress (Q3). The calculation of the index equals:
Abbildung in dieser Leseprobe nicht enthalten
While individualism/collectivism expresses the cohesion of a group and uncertainty avoidance the attitude towards uncertainties, power distance expresses the treatment of social inequalities. More precisely, the attitude to authorities and its manifestation in hierarchical social structures expressed among others by status, prestige, wealth distribution, law rights and rule depends on the degree to which a culture accepts an individual to be directed by interpersonal power. While in societies the treatment with inequalities can be expressed by the concentration of power (Hofstede, 1984, p. 72) - held by few (monolithism) or distributed competitively (pluralism) - in working environment this is expressed by the discrepancy between real and desired working environment. Monolithism and high discrepancy both show that people accept that power is distributed unequally, which equals the exhibition of high power distance (Hofstede et al., 2010, p. 61). Power distance is expected to affect asset pricing and portfolio management due to principal-agent relationships between corporate managers and shareholders, institutions and their employees and/or portfolio managers and their patrons. As indicated, Hofstede’s application in the working environment consisted of two questions regarding the perceived working environment and one question regarding working environment preference; all of them based on the leadership of one’s manager.
As last “original” cultural dimension5 Hofstede distinguishes between masculine and feminine cultures. Derived from studies regarding sex roles in organizations, he sums up that men put more emphasis on career goals including earnings, recognition, advancement and women more on the working environment and relationships including variables like congenial associates, leisure time and variety. Hofstede (1984, p. 188f.) defines men to be more assertive and women to be more modest and measures to what extent cultures enhance this relationship. A masculine culture is described as tough when gender roles and emotions are clearly separated as demonstrated above. This means that a high gender gap regarding “original sex roles” defines masculinity (although both women and men generally score higher values on masculinity in countries with this dimension). In financial environment, masculinity expresses how aggressive investors might trade or how reward-orientated corporate/portfolio managers and share-/stakeholders might be.
Lastly, time-orientation is discussed by several authors like Trompenaars (1997), but still mostly linked to Hofstede et al. since they were the first to give empirical rather than theoretical evidence. It describes to what extent individuals foster “virtues towards future rewards - in particular perseverance and thrift” (Hofstede et al., 2010, p. 239). Patience and timing are keys variables in stock market strategies which makes them important on investor- level. However, also corporate managers are expected to act towards future rewards and think in long-term relations.
Besides, the cultural dimensions of Hofstede, there are more findings on cultural value of which there lies the focus only on the most important ones for the sake of this paper. Moreover, in the young field of cultural finance, there is scarce literature dealing with cultural indices besides those of Hofstede.
Beginning in 1992 Schwartz discussed Hofstede’s findings and its limitations and, afterwards, conducted a survey arraying 49 nations with validated instruments and derives seven types of cultural values (Schwartz, 1999, p. 29). Schwartz (1992, p. 17) creates a questionnaire on the guiding principles of life with 56 values motivated by Rokeach value survey (1973). Answers are given on a nine-point scale ranging from 7 (supreme importance) to -1 (opposed to my values) from participants in 20 countries. In his subsequent study, Schwartz used a similar approach but excluded some values due to non-equivalent meanings. National cultures are compared in the relative importance they ascribe to the values in the questionnaire by a coplot technique which sums the absolute differences between all pairs of samples (national cultures) producing a matrix which is put in a twodimensional plot (Schwartz, 1999, p. 34f). Most relevant cultural values for this paper are mastery and harmony. Mastery emphasizes active self-distortion trough changing the natural and social environment and permits actions to increase personal power and wealth while harmony describes quite the opposite which is to fit in a given environment avoiding changes and accepting the world as it is (Schwartz, 1999, p. 40f.).
The origin of mastery and harmony lies in the work of Rotter (1966;1990) who introduces the concept of locus of control. This describes the degree to which persons expect that a reinforcement or an outcome depends on their own behavior and characteristics or solely on chance, luck, fate or other sources (Rotter, 1990, p. 489). Smith et al. (1995, p. 380) used the internal-external locus of control scale (I-E scale) by Rotter (1966) in a survey across 43 countries to identify its global validity and the dimensions it captures. They analyze the data by three-dimensional scaling where a matrix of dissimilarities between each pair of nations is computed and by calculating the correlations of the three dimensions, and present cultural indices. The first dimension in the paper links Schwartz’s culture values mastery and harmony in the context of politics with the answers to the (I-E scale) and finds strong correlations and explanatory power (Smith et al., 1995, 395). Moreover, the scale also captures individualism and collectivism. In this case, this means that believing in an internal locus of control makes a culture more individualistic and vice versa (Hamden- Turner/Trompenaars, 1997, p. 152). Since mastery combines both, a will to strive for own goals and values for masculinity such as ambition, independence and reward/success, Hofstede (2010, p.101;145) concludes high correlations between mastery and individualism as well as between mastery and masculinity and between individualism and locus of control.
Lastly, the cultural dimensions will be applied mostly on cultural clusters, not on countries, regions or continents. Thereto, ten clusters are proposed and generally supported by research (Gupta et al., 2002). However, most of the literature presented will deal with four to five clusters, namely Confucian (South-East) Asia, Southern Asia, Anglo Cultures, Europe (Nordic, Germanic, Latin) and to some extent Latin and Central America.
2.2 Behavioral Biases and Heuristics
Finance experienced a paradigm shift from neoclassical to psychologically based behavioral Finance. Major novelty of behavioral was to accept that agents fail firstly to receive and process additional information and update their beliefs correctly and secondly to make decisions that are compatible with subjective expected utility (Barberis/Thaler, 2003, p. 1053). Thus, the normative theory was replaced by a descriptive one. Although at first behavioral models seem to be ad hoc, additional research begins to establish this inductive approach (Subrahmanyam, 2007, p.13). De Bondt et al. (2008, p. 9) argue that behavioral research has become convincing because results are stable even with replications, many studies rely on surveys or observations of individual behavior in a natural environment and, lastly, results match conventional market-level price and volume data. The main reason for this are the findings from the psychology literature which explain why people deviate from being fully rational in their actions (Shefrin, 2010, p. 6). This paper will divide heuristics of behavioral finance which result in biases into three categories which are cognitive biases in a) gathering information (by selective attention and information overload), b) processing and evaluating information (by limited attention and processing capacities) and c) decision-making (by cognitive dissonance).6 Next, the most relevant heuristics in order to establish a model of cultural finance will be selected and briefly explained.
Research in cognitive biases was the starting point for what behavioral finance has become later. Tversky/Kahnemann (1981, p. 453) explain biases in perception in the context of framing. Framing means that the course of action, the possible outcomes and consequences as well as the contingencies (decision-frame) depend on the formulation, thus, on the presentation of the problem. The common pattern of this bias is that choices involving gains are rather risk-averse while choices of lost are rather risk-seeking. Another related bias involves selective attention or confirmation bias which is motivated by Wason’s study (1960). It means that people rather tend to overweight information which match the expectations and underweight those which do not. In combination with the primacy effect this means that information acquired in the early stage of processes have more weight on the decision than later information which is received and processed in a way that is partial to that opinion (Nickerson, 1998, p. 187). Combined with the recency effect, the later information would be overestimated. In the most extreme form, this leads to biases not only in gathering, but also processing information, namely conservatism (or Status-Quo bias) and representativeness bias.
The idea of conservatism7 lies in the concept that individuals once they formed an impression are slow to change that impression despite of new evidence (Barberis ,1998, p. 15). According to the studies of Griffin and Tversky (1992), cases of coin tossing show that people are more biased in the sample proportion (strength) than sample size (weight) which means that traders stick to the Status-Quo as long as the losses are not too extreme. For the stock market, this means that several small loses might not change expected returns of investors, while a large loss does.
Barberis et al. (1998) collected and created evidence on over- and underreaction of investors where models of under/overweighting new information have been successfully implemented. Following Barberis et al. (1998, p. 318ff) investors do not assume earnings to follow a random walk and think in a stationary process with two regimes determining earnings. Both regimes are Markov processes while in regime one earning shocks are likely to be reversed and in regime two shocks are followed by shocks of the same sign (trend). The regime-switching process is also a Markov process.8 Earnings in “Regime 1“ react too little to new information and as long as the investor assumes “Regime 1“ he will not react until he grudgingly changes his expectation on future returns when evidence has become too strong. However, in reality earnings follow a random walk, so that a positive shock is as likely to be reverted by a negative as to be followed by a positive shock. Since average realized returns will be positive for positive and negative for negative shocks, returns do not revert to the mean or follow a trend but depend from the random shocks.
This means, the bias discussed above leads to drifts in stock prices because the investor is not adjusting his decision frame correctly. These findings are consistent with post-earnings announcement drift and short-term momentum (Barberis et al. ,1998, p. 321).
Psychological evidence shows that gathering and processing information is the costlier the more abstract it is (Hirschleifer, 2001, p. 1546). Therefore, one rather assumes that the information gathered will hold true for the future as well. However, reducing complex tasks also leads to opposite effects such as representativeness bias. This is modelled in “Regime 2“ in the study of Barberis et al. (1998). The base line of this bias is to reduce complexity by attribute substitution. Subconsciously, people tend to substitute difficult-to-answer questions by easier ones. Instead of “How likely is it that the candidate will be hired?” the recruiter will answer a question like “How high was his test results in the Assessment Centre?” (Kahnemann/Frederick ,2002, p. 4). The same applies for a complicated task like predicting stock prices. In agreement with the availability bias, the most convenient substitution are the past stock prices violating Bayes’ rule (De Bondt, 1993, p. 359ff). Tversky/Kahnemann (1974) have shown that representativeness occurs when prior probabilities and/or sample size are neglected, and/or the evaluation of conjunctive events is biased. Lakonishok et al. (1994, p. 1564) empirically give evidence for differences between investors forecasted “tied to past growth rates” and actual stock returns (precisely representativeness bias in form of extrapolation). However, following the reasoning of Tversky/Kahnemann (1974) this bias might also be interpreted as giving an excessive overweight to recent outcomes leading to a substitution of the most recent with the most probable outcome, in the form of gambler’s fallacy. Extrapolation is more associated with primacy effect and its continuation while gambler’s fallacy is more connected with recency effect following Ayton/Fischer (2004, p. 1376). They conclude that probabilities attached to outcomes get subjective and are not interpreted as pure randomness; however, it is not responsible for the different manifestations of the representativeness bias. Culture might fill a gap here.
Next, when processing and evaluating the information heuristics play a key role to solve the problem of limited attention and cognitive capacity. Thus, another relevant bias is herding behavior which is the phenomena of groups of investors trading in the same direction over a period (Nofsinger/Sias, 1999, p. 2263). Wermers (1999, p. 582) compromised four possible explanations for this bias. First, managers might fear reputational risk of acting differently than the consensus, second, they might trade together simply because they receive strongly correlated information, third, they might infer the same information from past traders or, fourth, share aversion to specific characteristics of stocks. After Grinblatt, Titman and Wermers (1995) analysis on the behavior of mutual funds’ evidence of momentum strategies and herding, Wermers (1999) showed highest levels of this bias in trades of small stocks and in trading of growth-oriented funds.
Second prominent bias when processing and evaluating information is overconfidence which in financial models usually describes the tendency to overestimate the precision of one’s knowledge and is stronger the more complex the question/task is (Odean, 1998a, p. 1892). Moreover, this is true when fast and clear causal feedback lacks (Barber/Odean, 2001, p. 263). Since the stock investing is rather complex and clear feedbacks lack, overconfidence in the form of a belief, that the trader’s information is more precise than it is, serves as one straight-forward explanation for topics on stock participation (e.g. high trading volume), pricing and returns .9 Moreover, since participants in financial markets tend to be confident on their abilities (otherwise, they would not trade) and unsuccessful traders either lose their job, drop out of the market or just trade less on average, eventually overconfidence is highly existent (selection and survivorship bias as suggested by Odean (1998a, p. 1896)). Wang (2001, p. 150ff.) shows in his evolutionary game that moderate overconfidence survives and even dominates - when the share of overconfidence traders is higher than a threshold level and fundamental risk is large. Also, underconfident traders never survive in the long run. Regarding style investing, Daniel et al. (1998, p. 1856) early argue that overconfidence leads to momentum. When overreaction is confirmed by a market signal, traders become more confident (short disconfirms in form of losses decreases their confidence little or it remains constant) which triggers more overreaction due to self-attribution bias leading to a stock price momentum. In the long run, since more information are available, prices gradually move to fundamentals reversing the overreaction. This assumption will be critically reviewed later in this paper.
Most of literature assumes overconfidence to be a homogeneous behavioral bias (Glaser et al., 2004); however, in the following this paper will show its variations. Moore and Healy (2008) examine three types of overconfidence known in present research and their impact. While roughly 64% of empirical papers deal with the archetype of overconfidence, namely, overestimation, that is the evaluation of one’s actual ability, only 5% of empirical papers do with overplacement leaving 31% of empirical papers which examine overprecision. Overplacement bias was first discovered by Svenson (1981) when he conducted a study on the competence of U.S.-American and Swedish car drivers relative to the other participants in the study. Using scales of personality traits Larrick et al. (2007) find out that those people who rank themselves in a higher percentile also show higher overconfidence. It is also referred to as referential overconfidence. Overprecision describes the tendency to be overly confident of the precision of a private belief. Soll/Klaymann (2004) show that interval estimates, that is the definition of a range with a nearly 100% certainty to include the right value, are more prone to overconfidence than two-point binary choices, which are the definition of the lowest and highest possible value with a nearly 100% certainty. Moore and Healy (2008) aggregate previous works. conduct own studies and examine that the results suggest that the three types of overconfidence are conceptually, and empirically distinct, but not only different manifestations of the same underlying construct as financial research has assumed in the past. For this paper, this means that it is necessary to hypothesize which type of overconfidence does evoke which effects. Moreover, its different manifestations might have different exposure to culture.
Third component of biases in behavioral finance are the so called preferences which deal with the process of decision-making of people. The neoclassical approach suggests that people base their decisions on a principle of expected utility. It argues that the decisionmakers can measure the utility of events by a utility function where outcomes and their chances are included. Since studies show that several axioms of this model are violated in reality - such as the transitivity axiom among others shown by Tversky (1969) or the common consequence effect by papers of Allais (1953/1979) and Kahneman/Tversky (1979) and, moreover, the biases mentioned above violating the implicit principle of invariance, that is the independence between outcome and description of the same problem - the theory of non-beautiful people is added to that of beautiful people (Tversky/Kahnemann, 1986, p. 253). Modifications face the same common problems since people do not approach a problem as holistically as theories of decision-making under expected utility suggest (Shoemaker, 1982, p. 548). Overall, the biggest impact is made by the (cumulative) prospect theory by Kahnemann/Tversky (1979, 1992) which describes the majority of present biases.10
Abbildung in dieser Leseprobe nicht enthalten
Figure 1: Ten Fields consistent with Cumulative Prospect Theory
Source: Camerer (1998)
It combines a probability weighting function and a value function which bears in mind that the utility not depends on the absolute outcome, but more whether it is a loss (risk-taking to avoid) or a gain (risk-averse to not lose) that is involved (Camerer, 2005, p. 8). Both the gain and the loss value function represent diminishing sensitivity. This distinction covers the loss aversion of individuals which provides a basis for a range of bias in the decisionmaking process.
Status-Quo effect reveals the bias that individuals tend to remain at their current situation when another option is defined as they fear more disadvantages than advantages of leaving (Kahnemann et al., 1991, p. 198). This effect holds especially true under ambiguity. Samuelson/Zeckhauser (1988) show in a consumer test that the participants resist to change their current electric power provider regardless whether their reliability was good or bad. While in a group of consumers with high reliability 60% have chosen their current option, 58% of the consumers with low reliability remained in their provided option. The endowment effect has a similar effect and argues that people ascribe more value to a good merely because they possess it (Kahnemann et al., 1991, p. 194). Loewenstein and Kahnemann (1991) attribute this bias to loss aversion. They compare a group of students which were given a pen and another group which got a coupon for an unspecified gift. After all participants were asked to rank gifts based on attractivity, they are a given a choice to trade their current gift by chocolate. Those, who possessed the pen (a specified gift instead of a coupon) are less willing to trade although both groups have ranked the attractivity of a pen equally.
The triggers and effects of disposition effect, that is the tendency to ride losses and realize gains, are analyzed by Frazzini (2006, p. 2018) who concludes that mental accounting besides loss aversion generates the disposition effect. Thaler (1999, p. 187) has shown that individuals in accordance to the disposition effect account for relative losses or gains, but that it is piecemeal and topical. Since sensitivity is diminishing with higher gains/losses, gains are usually segregated and losses integrated. Moreover, this means closing an account with losses takes longer since already a small loss is more painful for an individual to accept than a gain of the same size. In contrast to rational decision-making the initial investment is used as measure for decision making (sunk cost).11 In summary, this means that individuals rather sell winners and keep losers, also rationally it might be more effective to do the opposite. Frazzini (2006, p. 2038) sees that the disposition effect expresses itself by the underreaction to news of the whole market. With mental accounting and loss aversion, investors sell winners resulting in a price depreciation which means that (from that lower base) relative future returns will be higher. Hence, disposition effect generates positive price drifts for winners and negative drifts for losing stocks. Tracking the data of 10,000 customers of an US-American brokerage firm Odean (1998b) checked whether winners are sold too fast and losers kept too long. Summing the number of stocks sold for a gain and a loss correlates with the movement of the market leads to manipulated results.12 Thus, he sets up a portfolio for each date in the period from 1987 - 1993 and if in the portfolio the stock’s daily high and low are above its average purchase price, it is counted as a paper gain and vice versa and, otherwise, it is not counted for one or another. Odean (1998b, p. 1783ff.) finds that the proportion of gains realized to the sum of realized and paper gains is significantly higher than the same proportion for losses providing convincing evidence for a disposition effect.
Regarding asset pricing behavioral finance argues that these are deviations from its fundamental value caused by irrationality. While the efficient markets hypothesis argues that mispricing is removed by rational traders who seek investment opportunities, behavioral finance argues that mispricing does not always show an investment opportunity (Barberis, 2003, p. 1057). First, limits to arbitrage arise through noise trader risk of irrational investors. This deals with the risk that current mispricing, which the trader is exploiting, might grow further in the short run which might result in the need to liquidate the asset to raise cash. This is shown by Froot/Debora (1999) between twin companies which merged their interests on a predetermined basis. The deviations on the theoretical parity are constantly high. In the example provided, Royal Dutch was up to 35% underpriced and up to 15% overpriced due to noise trading (Froot/Debora, 1999, p. 192f.). Several hedge funds tried to bet on this development, but still failed in the short run resulting in liquidations like that of LCTM. This risk is mostly dominant with low-frequency events (Ritter, 2003, p. 436). The second risk of arbitrage deals with fundamental demand shocks which are nonrepeating and long-term in nature like M&A, stock’s addition to or deletion from prominent market indices (Gromb/Vayanos, 2010, p. 254). A third risk are financial constraints regarding short selling costs, leverage and equity constraints. Those constraints might be based on law or simply by the cost of resources needed to arbitrage. Keeping these three types of risks in mind, the analysis of asset pricing will be focused on the first question. The leading question is what an impact culture has on noise trading, specifically the risk of mispricing by the biased perception, processing and evaluation of information (Daniel/Hirshleifer, 2015, p. 80). The answer to this question might solve issues like the equity premium puzzle.
The heuristics for information perception, information processing and evaluation as well as decision-making enjoy strong evidence. As mentioned before, explaining stock market investing with behavioral finance seems to be ad hoc. Heuristics contradict each other and moreover, they are selectively used as explanation without a unifying structure to show why other contradicting biases do have less impact (Shefrin, 2010, p. 6). But also, the opposite is true, that investment styles like momentum are explained by mutually inconsistent heuristics (De Bondt et al., 2015, p. 15). The lack of a coherent system of behavioral finance prevents this discipline to be able to provide a precise interpretation of economic and financial events and, even more, to predict events such as stock price crash risks.
This paper attempts to incorporate the main strength of behavioral finance, that are the findings on how people act differently from fully rational behavior, into a unifying foundation and forming a coherent system with new evidence on cultural finance. Main goals are to give a push to further research on the impact of Culture on Stock Markets by offering connecting points of cultural sciences and finance research and linking theories.
2.3 Relevance and Framework of cultural finance
Trading activities and diversification across cultures have been examined in the past leading to the establishment home and foreign biases. In their pioneering work for cultural finance, Grinblatt/Keloharju (2001) examine investors’ preference for stocks of Finnish firms that are close to the investor in terms of distance, language and culture. Although this effect is less dominant for well informed and perceptive institutions, it remains prominent for a majority consisting of less savvy institutions and households. Grinblatt/Keloharju (2001, p. 1072) proposed familiarity as strong reason but could not agree whether it is within rational actions (less costly and more available information) or an irrational bias (less/inefficient diversification due to familiarity). In addition, Kilka/Weber (2000, p. 186) show that investors are generally more optimistic in domestic stocks since they feel more competent in judging them as foreign stocks. This would mean that due to the investor’s confidence in his chosen domestic stocks, he rejects to diversify more with international stocks. In their study, Massa and Simonov (2006) model an index of familiarity consisting of professional and geographical proximity as well as the holding period of stocks (since investors are better informed on stocks they hold and trade with). To test whether familiarity is behavioral or information-based, they include shocks in their study, namely professional change, relocation and unemployment, as well as the wealth as a proxy for the degree of informativeness of an investor to test how available information reflect familiarity (Massa/Simonov, 2006, p. 642). Indeed, these shocks decrease the sensitivity to familiarity meaning that traders replace the cheap but outdated information of familiarity with other.13 This evidence indicates that common culture serves as a competitive advantage to fight ambiguity - due to information overload, limited capabilities and cognitive dissonance - which encourages to assess its importance and build a framework in which cultural finance can be pointed out as future research field. Recent research by Beugelsdijk/ Frijns (2010) and Anderson et al. (2010) attach specific cultures a tendency for home bias arguing that culture directly influences investor behavior. This will be made more clear in the next chapter. However, these presented findings encourage the following framework for cultural finance.
The first straight-forward assumption when modelling cultural finance deals with the hierarchical impact chain. Culture cannot be treated like a code of Law since as Hofstede et al. (2010) state, it is the “software of mind” which makes it much more subtle and hard-to- define in written words but works in an all-embracing way. Relating to the iceberg concept of Hall (1976), the top of culture - called behavior - is primarily in awareness, while the more profound invisible bottom of culture - values and thought pattern and to some extent beliefs14 - are out of awareness.
Abbildung in dieser Leseprobe nicht enthalten
Figure 2: Iceberg concept of Culture
Source: Own representation, adapted from Hall (1976)
First, this model of cultural finance assumes just like Hall (1976) that the mentioned values determine the behavior. Second, these values are somehow “programmed” in the members of a culture making it a “software of mind” and mostly stable during the life of the members. Next, behavioral finance identifies empirically established rationality defects, but fails to explain its relevance and validity. To fill this gap, culture will serve as an explaining background variable to account for “the diverging relevance of certain behavioral patterns between countries” (Breuer/Quinten, 2009, p. 14). In this context, Culture attacks another weak spot of behavioral finance which is the lacking acknowledgement of social components, more precisely how individual non-rational actors aggregate on a market level. Levinson/Peng (2007) incorporate values of different cultures to test for the validity and strength of behavioral biases with supporting results for the presented assumptions. Lastly, the model assumes that the cultural dimensions have high explanatory power and validity for national culture. Figure 3 shows an insight into the model. The length of an arrow symbolizes the extent of significance of the component while the white boxes contain further explanations. The goal of chapter 3 is to capture existing literature and critically embed the findings in a cultural perspective, before chapter 4 will discuss and interpret these findings.
Abbildung in dieser Leseprobe nicht enthalten
Figure 3: Framework for cultural finance
Source: Own representation
3 Culture and Behavioral Finance
3.1 Risk
3.1.1 Attitudes towards risk
Assessing the degree of risk of an asset depends on the perception of its uncertainties and the notion of their probabilities. Tversky/Kahnemann (1974, p. 1124) explain that "people rely on a limited number of heuristic principles which reduce the complex tasks of assessing probabilities and predicting values to simpler judgmental operations”. Although these heuristics make it possible to manage risk, they usually lead to systematic errors from rational decision-making. In the next sections, cross-cultural differences in risk management are explained from different perspectives. The results serve as foundation of the next sections, dealing with the cross-section of returns and style investing. Where applicable only for specific styles, differences in risk management will be explained in the section 3.3. This section in particular deals with the attitude towards risk.
The attitude towards risk in cultural environment is evident in social sciences. Renn (1992) explains that culture structures the mind-set of individuals and organizations, and models prototypes of risk managements. Referring to his studies, Vasvari (2015, p. 34) also concludes that these prototypes can be explained by social inequalities (e.g. power distance) and group cohesiveness (collectivism/individualism) and suggests masculinity/femininity as well as the methods of managing uncertainties (uncertainty avoidance) as further variables.
Based on their survey, Breuer et al. (2014a, p. 456) show that households from individualistic cultures rather agree to participate in financial risk to earn above average returns. They argue that overprecision and overoptimism are the triggers for the willingness to take additional risk. Although overoptimism is linked to overconfidence in several papers,15 it is worth taking a deeper look on its foundation. Puri/Robinson (2007) assessed that overoptimism (proxied by the differences between believed and statistical life expectancy) enhances the propensity to include riskier assets in one’s portfolio such as individual stocks instead of mutual funds. Excessive optimism might be traced to a lack of self-reflection leading to a degree of optimism beyond a “wise portion”. It argues that higher probability is assessed to good than to bad events. While overoptimism can be weakened mostly in principal-Agent relationships for agents by frameworks and penalties, individual investors need to assess their degree of optimism by excluding the tendency of selective perception. As it will be shown, individualistic cultures are more prone to primacy effect and extrapolation bias causing overoptimism. In this context, overprecision makes the trader believe that his judgement is more precise than it actually is.
It is important to summarize the findings yet presented. They focus on the attitude towards risk, not on its perception or tolerance. Both, Puri/Robinson (2007) as well as Breuer et al. (2014a) conduct surveys and follow a sociological approach in which individualistic cultures by definition are expected to see risk as something out of control which needs to be made the best of. The same limitation holds true for the associations made for uncertainty avoidance and power distance. Hillson/Murray-Webster (2005, p. 80) summarize that high power distance rather hinders the identification and eliminations of risk due to their attitude towards authorities and the acceptance of external power. This means that cultures with high power distance will rather have a passive attitude thinking that risk is out of one’s control while low power distance would identify it as something that can be avoided or accepted in change for benefits. Moreover, Hofstede (2001, p. 218) states that although uncertainty avoidance explains the desire of certainty, the actions undertaken to reduce uncertainty might involve risk-seeking activities. Again, high uncertainty avoidance also stands for a measure of attitude towards risk, which exposes the anxiety towards risk. The manifestation of this cultural dimension in risk perception and preference lies in actions reducing ambiguity, not necessarily risk (Hillson/Murray-Webster, 2005, p. 80). Lastly, masculine cultures will face risk proactive as a step to fulfill their achievements whereas feminine cultures carefully think out and react to possible risks of own decision (Kreiser et al. (2010), 963ff.).
Therefore, dissonances between the attitude towards risk and actual risk management are expected. While culture works as software of mind in forming believes and attitudes at the first step as shown in this section, the next step is to assess cross-cultural differences in risk perception and tolerance in practice.
3.1.2 Risk perception and preference
Campbell et al. (2004) measured the overconfidence of participants in their study as the difference between confidence in one’s choice and its accuracy. Overconfidence is modelled as overestimation, that is the tendency to overestimate one’s abilities. In a betting game (with a maximum zero average value) the participants were asked to bet on their previous answers with 100 points when the bet is won, and points lost according to the following equation:
Abbildung in dieser Leseprobe nicht enthalten
The style of the betting game would propose to compensate for overconfidence by rejecting risk-taking since the average outcome is negative otherwise. Still, final average points were far below zero (Campbell et al., 2004, p. 303). These findings correspond to those of Goodie (2003). Furthermore, the authors linked the findings on overconfidence and risk-taking to narcissism. Next, cross-cultural studies were conducted to compare the results. Compared to U.S. participants, the Chinese were much more overconfident and risk-seeking in the study of Meisel et al. (2016, p. 391) which is explained by lower scores on independent (individualism) and higher scores on the interdependent self-construal scales (collectivism). However, previous research has shown that narcissism is low among Asian ethnicity (Foster et al., 2003, p. 481). In conclusion, this means that overestimation is one trigger of risk-taking not because people tend to experience narcissism, but due to their mindset, which is largely influenced by culture. In this case, surprisingly collectivism seems to enhance the underestimation of risk. Assad (2015, p. 103) adds to these findings that overconfident traders do not exhibit a risk-taking preference per se, but are less aware of risk, even when high incentives are given for risk-averse / underconfident behavior.
While the term risk-averse/seeking describes preference towards risk, the trader himself individually perceives the risk first. At the early stages of cultural finance, Bontempo et al. (1997) or Weber/Hsse (1998) have examined cross-cultural differences in the perception of risk. Perceived risk consists on the one hand of the linear weighted combination of the probabilities for a positive, break-even and negative outcome and on the other hand the weighted combination of the conditional expectations of gains and losses indicating the magnitude of the conditional expectation of positive and negative outcomes (conjoint expected risk (CER) model):
[...]
1 Groups in the sense of cultures
2 An example for autocorrelation in cultural economics is provided by Williamson/Mathers (2010). When examining country’s growth rates besides economic freedom and culture control variables where included. As a result, endogeneity occurred due to autocorrelations of economic freedom and investment rate as well as education and growth rates. Errors in measurement for example may occur when questionnaires are misleading for foreign cultures.
3 Indirect or background effects of culture
4 Participants rate on a scale (range of 5) every statement individually (high values symbolizing disagreement). Afterwards, results are aggregated.
5 In his first publication, Hofstede presented power distance, uncertainty avoidance, individualism/collectivism and masculinity as the four cultural dimensions. Time orientation and indulgence/restraint were found in a later study.
6 These three categories were motivated by the paper of Barberis/Thaler (2003) where heuristics are assigned to beliefs (information process) or preferences (decision-making). Since many heuristics cannot be assigned precisely in one category, indeed comparing heuristics with each other is less helpful than comparing categories as a whole with each other. These categories will also help to see where culture has its biggest impact.
7 Also described as Status-Quo-Bias
8 Barberis et al. (1998) assume that investor put more weight on one of the two regimes. The explanation provided is based on underreaction to explain conservatism. However, overreaction leading to representativeness bias is displayed in “Regime 2“. The regime-switching process assumes that the investor does not learn and change the model to expect future returns but ony switches from one to another regime. They assume that underreaction is present for news in isolation and overreaction for a series of news which means that for underreaction the strength and for overreaction the weight of the information is neglected.
9 Among others Barber/Odean (2001) and Gervis/Odean (2001)
10 Camerer (1998) compares the cumulative prospect theory with Expected Utility theory and shows ten biases which are consistent with the first, but not with the second theory (cf. Figure 1).
11 In the case of stock investing, the initial investment by purchasing the stocks is realized while losses/gains when holding those stocks are not realized or declared to e.g. tax authorities. When an investor needs to raise cash, he will rather choose an account of a stock which has generated a gain despite of evidence that the stock will subsequently overperform the market. Mental Accounting also prevents high diversification.
12 This means that in an upward trending market with investors indifferent to sell winners or losers, more stocks would be sold for gains
13 Barberis et al. (2005) use a related approach to examine how co-movement of stocks can be explained. The findings in familiarity can be connected to the category and habitat view which is further presented in section 3.1.3.
14 For the sake of simplicity: Only called values from now on.
15 Among others: Odean 1998, Daniel et al. 1998, Camerer and Lovallo 1999, Van den Steen 2004
-
-
-
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X. -
-
Téléchargez vos propres textes! Gagnez de l'argent et un iPhone X.