University of Groningen Uppsala Universitet
Faculty of Economics and Faculty of social sciences
Business
Master Thesis - MSc International Financial Management
Dividend policy in the banking sector; the
influence of culture
Matthijs Hollander
Supervisor: Dr. J.H. von Eije
Co-assessor: Dr. H. Gonenc
2
Abstract
This thesis investigates the influence of culture on the dividend policy of banks. Dividend
policy is an extensively discussed topic. Many theories and empirical studies exist.
However, none of them has been able to resolve why firms pay dividend and why this
amount differs among firms. Recently, researchers attempted to get more insight in this
issue by researching the influence of culture on the dividend policy. They found
significant results, indicating that culture matters. However, they excluded the banking
industry in their research. Though it plays an essential role in a country‟s economy. In
addition, insight into what moves the investor preference for the amount of dividends
they receive is highly relevant for banks, as they have the additional complexity of
complying to strict regulation on capital requirements.
Tobit panel regressions embracing 1600 banks in 65 countries are conducted. Significant
negative relations are found between both the level of individualism and masculinity and
the amount of dividends paid by banks.
Keywords
3
Table of contents
1. Introduction ... 4
2. Theoretical background ... 6
2.1 The dividend paradox ... 6
2.2 Explaining the dividend paradox ... 7
2.2.1 Signaling theory ... 8
2.2.2 Bird in the hand theory ... 9
2.2.3 Pecking order theory ... 9
2.2.4 Catering theory... 9
2.2.5 Agency theory ... 10
2.3 A cross-cultural perspective... 10
2.4 Dividends and the Banking industry ... 13
2.4.1 Relevance of the banking industry ... 13
2.4.2 Difference of the banking industry... 14
2.5 Literature review summary ... 16
2.6 Hypothesis building... 17
2.6.1 Individualism vs. collectivism ... 18
2.6.2 Power distance ... 19
2.6.3 Uncertainty avoidance ... 20
2.6.4 Masculinity vs. femininity ... 21
3. Data and methodology ... 23
3.1 Sample ... 23
3.2 Dependent variable ... 24
3.3 Independent variables ... 25
3.4 Control variables ... 25
3.4.1 Firm level control variables: ... 25
3.4.2 Country level control variables: ... 27
3.5 Methodology and Equation ... 29
4. Results ... 29 4.1 Descriptive statistics ... 29 4.2 Regression results... 30 4.3 Robustness tests ... 32 5. Discussion ... 33 5.1 Individualism ... 34
5.2 Power distance and uncertainty avoidance ... 35
5.4 Masculinity ... 35
6. Conclusion and implications ... 35
6.1 Conclusion ... 36
6.2 Practical and theoretical implications ... 36
6.3 Limitations and future options ... 37
Literature list ... 39
4
1.
Introduction
“The harder we look at the dividend picture, the more it seems like a puzzle, with pieces that just don‟t fit together” (Black, 1976: 5). These words were written by Black in 1976 referring to the effect of taxes on the dividend policy. It is an example of the struggle scholars have with the question why it is that firms pay dividends and why it is that investors care. It is a paradox that has not been solved so far (Bhattacharyya, 2007; Kinkki, 2001). One of the articles on this topic that has received a large amount of attention is written by Miller and Mogliani (1961), who suggested the dividend irrelevance. According to this irrelevance the amount of dividends paid has no influence on the firm value, and thus it should not be relevant how much dividends a firm pays. However, the dividend irrelevance is based on a number of assumptions, making it questionable whether it is applicable in real life. These assumptions include for example that firms operate in an efficient market and that actors behave based on rationality. Several theories have been suggested to explain the dividend policy behavior, such as the agency theory (Jensen, 1986), signaling theory (Ambarish and Williams, 1987), bird in hand theory (Bhattacharya, 1979), and the catering theory (Baker and Wurgler 2004). Although these theories have given some insight into the dividend behavior, none of them has been able to solve the question why firms pay dividends (Bhattacharyya, 2007; Kinkki, 2001).
5
whether the determinants for the dividend policy differs. Baker, Veit, and Powell (2001) for example found that in nine out of the 22 researched factors financial firms differ from non-financial firms. Casey and Dickens, and Newman (2002) found agency problems to be a significant determinant in the banking industry whereas Theis and Dutta (2009) did not find it to be significant factor. Another contradiction is found in the influence of signaling effects. Whereas Nnadi and Akpomi (2008) found them to be significant Casey, Dickens, and Newnam (2002) did not consider it to be a significant factor. An important realization when reviewing the contradictions in the suggested determinants for the dividend policy is that different samples were used on which they based their decisions. It is possible that their different findings can be explained by cross-border differences and by cross-border cultural differences.
In this thesis I research whether culture is a significant factor for the dividend policy of banks. My motivation is based on several factors. Firstly, previous literature showed culture to have a significant influence on the dividend policy, but excluded the banking sector. However, I consider the banking sector to be one of the most important industries. Banks play a essential role in a country‟s economy. Moreover, the dividend policy is highly relevant for banks as they not only have to satisfy the investors, but also have to comply with regulation related to capital requirements. Second, investigating the banking industry can be interpreted as a robustness test for previous researches on the influence of culture on the dividend policy. Thirdly, research on whether the banking industry differs from other industries in terms of their dividend policy and their determinants is inconclusive so far. Samples of these studying are restricted within different regions, thus limited to specific cultures. By investigating if culture has an effect on the dividend policy an answer might be given to the inconclusiveness found so far in previous research.
Based on the above motivation I perform empirical research in order to answer the following research question:
“Is culture a significant factor for dividend policies in the banking industry?”
6
addition, determined is which additional control variables are required in order to ensure that the influence of culture is distilled. Moreover the appropriate type of methodology in order to test the sample is determined. Section three describes these aspects. Thirdly, the findings of the analysis are presented. This starts with descriptive statistics, which is followed by regression results and robustness tests. These can be found in section four. In section five I will discuss the acquired results. This will be done based on the four hypotheses as formulated in the literature review. Finally, in section six I will present my conclusions, discuss the relevance of the results, both for the academic and practical field, and give the limitations of the research and ventures for future research.
2.
Theoretical background
As mentioned in the introduction the dividend policy of firms is an extensively studied topic in finance. In this section I give an overview of these studies. The paradox of the dividend policy and the main theories explaining the dividend policy will be described. This will form the first part of this section, which consists of four parts with each their own sub-sections. In the second sub-section an overview of the field of culture will be given and the contribution of cross-cultural studies to the dividend puzzle will be discussed. The third sub-section describes how the banking industry deviates from other industries in terms of dividend policy. Finally, in the fourth sub-section I combine the previous three parts in order to formulate the hypothesis.
2.1 The dividend paradox
7
case investors prefer dividends while a company does not issue any, they can simply create homemade dividends by selling part of the stocks they have.
The question then arises, if theory suggests that it does not matter whether companies pay dividends or not, why then do companies bother paying dividends and why do they have deliberate payout strategies as suggested by Lintner (1956)? Moreover, why do investors care about the level of dividends they receive if it does not influence their return? It seems there is a difference between what theory suggests and what happens in practice. An answer may lie in the point that the dividend irrelevance proposition of Miller and Mogliani (1961) is based on several assumption, such as that firms operate in an efficient market, that the investment policy is decided independently of the dividend decision, that the discount rate is the same among shareholders and that the dividend decision is based on a world without taxes. However, is the dividend irrelevance proposition still reliable considering all these assumptions? Perhaps the answer to the question why firms pay dividends in varying amounts while it is suggested that this decision does not affect the firm value lies in that the assumptions used do not hold in real life. Based on this notion many scholars have studied dividend policies.
For example, the dividend irrelevance proposition does not take into consideration that taxes have to be paid on dividends. These taxes are usually higher than the taxes on capital gains (Bhattacharyya, 2007). This would lead to the assumption that investors prefer capital gains instead of receiving dividends. However, Miller and Scholes (1978) showed that even under taxes investors could be indifferent towards the level of dividends issued. They based their suggestion upon the idea that investors can make their own tax-shield, which prevents them from having to pay taxes. However, creating a tax-shield involves transactions, which in real-life come along with costs. Allen and Michaely (2002) found in their literature review that theory and empirical findings indicate that taxes do matter when transactions are costly and in case risks cannot be hedged.
They also state that transaction costs creates costs for investors in case a firm decides to retain its earnings and an investors wants to use home-made dividend. This might be a motive for investors to prefer receiving dividends.
2.2 Explaining the dividend paradox
8
information inside the firm. Moreover, the dividend irrelevance is based on the assumption that investors behave rationally. However, it is questionable if markets are efficient and if humans always base their decisions on rationality. Several theories exist, of which some are based on behavioral finance, which is based on the assumption that instead of basing their decisions on purely rationality, actor‟s decisions are influenced by cognitive factors such as beliefs and perceptions (Miller, 1986). The following theories will be discussed: signaling (Ambarish and Williams, 1987), bird in hand (Bhattacharya, 1979), pecking order (Myers, 1984), catering (Baker and Wurgler 2004) and agency theory (Jensen, 1986).
2.2.1 Signaling theory
In the dividend irrelevance proposition of Mogliani and Miller (1961) it is assumed that outside and inside investors have the same amount of information. The signaling theory is based on the idea that this is not the case; there is information asymmetry between insiders and outsiders. Since managers work inside the company and have access to internal information it is hypothesized that they know more about the real state of earnings compared to outsiders. Their actions are based on information, which the market does not have and therefore might be interpreted as a signal to others. This also holds for the dividend policy. According to the signaling theory managers‟ use dividends to signal private information to outside investors (Ambarish and Williams, 1987; Bhattacharya, 1979; Miller and Rock, 1985; John and Williams, 1985). Denis, Denis and Sarin, (1994); Ofer and Siegel, (1987); and Carroll, (1995) find that analysts forecasts are changed after dividend announcements, indicating that analysts consider dividends to provide information. In his literature review, Kinkki (2001) concludes that signaling effects have been found in empirical studies widely.
9
2.2.2 Bird in the hand theoryThe bird in the hand theory suggests that dividends increase the shareholders wealth because retained earnings bring uncertain future cash flows, whereas dividends are guaranteed returns (Bhattacharya, 1979). Firm value can be created by issuing a high level of dividends, which will attract risk adverse investors (Shao, et. al. 2009). However, the bird in the hand theory is also strongly criticized. Mogliani and Miller (1961) showed that retained earnings increase the share value as long as the money is invested in positive NPV projects, and thus also resulting in a gain for investors. Moreover, Easterbrook (1984) states that investors usually do not use dividends for consumption or to invest in risk-free projects, but reinvest it. Thus, even when dividends are received, an investor will still bear risk on it.
2.2.3 Pecking order theory
The pecking order theory suggests that managers prefer to finance their investments with internal funds (retained earnings) instead of external funds in order to limit the amount of external monitoring by the market. In case the managers needs external financing, a clear order of preference exists, based on what is the safest choice. The first choice will be debt, second a hybrid form such as convertible debt and the last choice will be equity. According to the pecking order theory, a company adapts its dividend policy on its investment opportunities (Myers, 1984).
2.2.4 Catering theory
A relatively new model is the catering model as suggested by Baker and Wurgler (2004). This model is based on the assumption that the market is not fully efficient. It is suggested that managers issue dividends based on what the shareholders want. In other words, they cater for their wishes. As for the question why investors prefer dividends, they suggest that sentiment is the main determinant. Important to note is that the catering theory only explains the propensity to pay and not the amount. According to Baker and Wurgler this depends more on profitability of the individual firms.
10
2.2.5 Agency theoryThe agency model is used in this theory to explain the influence of culture on the dividend policy. The model involves the separation of the management (agent) and the ownership (principal) of the firm. The agent is expected to act in the best interest of the principal. However, often the agent thinks first of its own interest and then on the interest of the principal. The interests of the two parties diverge. It is therefore necessary to take actions in order to align the interests of the agent with the interests of the principal (Jensen, 1986). In terms of the dividend policy this has several complications. According to Easterbrook (1984) dividends are a relatively cheap measure to solve agency problems. By issuing dividends, the earnings are not available for financing anymore. Thus, managers need to attract external capital instead, making them subject to external monitoring. Jensen (1986) reasons that dividends can be used to reduce the amount of free cash flows. With a high amount of free cash flow available to managers they might be tempted to invest in projects that offer a high personal benefit, but also with a lower NPV. By issuing dividends the amount of free cash flow is reduced and thus limits the risk of overinvestment. La Porta, Lopez-de-Silanes, Schleifer, Vishny (2000) propose two different agency models. The first model, which they call the outcome model, is similar to Jensen‟s way of reasoning. Managers spend dividends more in countries that have good investor protection and less if investor protection is bad. The second model, which they call the substitute model, suggests that firms need a good reputation in order to be able to attract external capital. A good reputation is created by paying dividends. They empirically test these two models on a cross country basis and conclude that the agency theory is highly relevant for explaining dividend policies. They especially found support for the outcome agency model.
2.3 A cross-cultural perspective
Even though the dividend policy has been studied so extensively and many theories tried to explain why different firms pay different amounts of dividend, it still is a question that is unanswered (Bhattacharyya, 2007; Kinkki, 2001). The question why corporations pay dividends and why investors find it important keeps questioning scholars.
11
and Wiilliamson, 2006) and have found various factors explaining dividend differences among countries. La Porta et. al. (2000) for example found that the level of agency problems differs among countries. Adaoglu (2000) finds large differences between companies located in emerging versus developed markets and states that the emerging markets adds more pieces to the paradox of why firms pay dividends. Nevertheless, realizing why dividend policies differ among countries might give scholars a better understanding of the determinants for these policies.
The fact that there are differences found among countries might be explained by cultural differences. Cross-cultural studies are gaining popularity in the field of finance and might answer questions unanswered before (Breuer and Quinten, 2009; Chang, Lee, and Yi, 2009). According to Chang et. al. (2009) culture is mostly ignored in the field of finance due to the idea it is impossible to measure culture and the difficulty to interpret the concept, but at the same time empirical studies have proven that culture is a relevant factor in many topics of finance. Whereas traditional theories in finance are based on rationality and market efficiency, culture, as part of behavioral finances, does not hold these assumptions as granted, but instead focuses more on cognitive factors such as beliefs and perceptions (Miller, 1986). Breuer and Quenten (2009) believe that a new field of finance, which they call cultural finance, has the potential to solve the weak points in the established fields of finance.
According to Fidrmuc and Jacob (2010) the popularity of culture in the field of economics has been increasing. This trend can also be observed for the topic of the dividend policy. Recently, two studies by Shao, Kwok and Guedhami (2009) and Fidrmuc and Jacob (2010) tested the relation between cultural differences and the level of dividend payments and both came to the conclusion that culture is a highly relevant factor.
12
order to compare cultures in an empirical study it is necessary to quantify it. Different studies exist in which this has been done. In these studies assigned values on multiple cultural dimensions are used as a proxy for culture.
Of these studies, the cultural framework of Hofstede (1980) is by far the most popular and most often used. The framework is based on 5 different cultural dimensions. These are the level of power distance, masculinity versus femininity, collectivism versus individualism, the level of uncertainty avoidance, and short term versus long term orientation. Of these dimensions the last one was added later (Hofstede and Bond, 1988). Individualism vs. collectivism captures the degree to which people believe they should take care of themselves and their close surrounding, versus being part of a collective which takes care of them in exchange for loyalty. Small vs. large
power distance refers to the degree to which members of a culture accept that power is divided
unequally. In a low power distance culture the difference in hierarchy needs to be justified, whereas in a high power distance culture the differential is accepted by the difference in power and traditions. Masculinity vs. femininity; in a masculine culture success and career, heroism, ambition and independence are dominant values whereas in a feminine culture taking care of each other, mentoring, and the quality of life are more important. Uncertainty avoidance refers to the degree in which people feel uncomfortable with uncertain situations. In high uncertainty avoidance cultures people will prefer to achieve stability and reduce uncertainty. Short term
versus long-term orientation is inspired by confusion dynamism. Long-term orientation is
associated with thrift and perseverance, whereas a short-term orientation is based on respecting both tradition and social obligations and protecting one‟s face.
13
Besides Hofstede, there are several different frameworks, such as the framework of Kluckhohn and Strodtbeck (1961), Schwartz (1992), Trompenaar (1993) and the Globe project (House, Hanges, Javidan, Dorfman, and Gupta, 2004)
All these models have used different ways to research the cultures with different samples and were conducted at different times. They all produce their own cultural dimensions, varying from six dimensions in Kluckhohn and Strodtbeck‟s study to nine in the Globe study. However, according to Thomas (2002) the models show a strong similarity with Hofstede‟s (1980) study. Cultural frameworks offer a tool to systematically compare national cultures in a quantitative way. However, it is also important to realize that they have their limitations. The most important issue of these frameworks is that they might lead to overgeneralizations. Although members within a country share the same national country they still remain unique individuals with all their own norms and values. Cultural frameworks might therefore result in oversimplification and stereotyping (Thomas, 2008). Moreover, does every country have its own national culture or do multiple cultures exist? Do for example in Canada the French speaking inhabitants have the same culture as the English speaking inhabitants, or does the nation contain multiple, different cultures. However, although it is important to realize these limitations, the cultural frameworks still offers a good tool for empirical studies. First there aren‟t really any alternatives to measure and compare cultures. Moreover, these cultural dimensions focus of those points shared within cultures, and many studies have proven them to be significant in explaining various topics.
2.4 Dividends and the Banking industry
So far the literature review included theories on the dividend policy and culture, both key factors of this research. However, this literature has not taken into consideration yet that the research is specified on the banking industry. This will be done in this sub-section. Two aspects will be discussed. Firstly, I will discuss why the banking industry is a relevant industry for the influence of culture on dividends. Secondly, I will discuss how determinants for dividend policies differ in the banking industry compared to other industries.
2.4.1 Relevance of the banking industry
14
also excluded in the literature on the relation between culture and dividends (Shao, et. al., 2009; Fidrmuc and Jacob, 2010). However, I believe that because banks are different from other industries they are so important to research. They have an essential role in a society as one of the main institutions that is responsible for the capital allocation of an economy. Because of this key role in society it important to understand how their dividend policy is formed and how national culture affects this. Moreover, banks are subject to strict regulations related to their capital requirements in order to control their risk profile. However, under these restrictions banks still pay dividends. Why is it that they issue these dividends, considering that issuing dividends reduces the level of retained earnings and thus their level of capital? A final issue that makes banks an interesting industry to research is the relative importance of a good reputation and image in the industry. This makes the banking industry more dependent on how investors perceive the dividend policies. As I will hypothesize in the next sub-section, this perception is influenced by cultural differences.
2.4.2 Difference of the banking industry
As mentioned before, articles often exclude banks because they are different. However, are banks really different or are they just excluded to avoid possible implications while the sample is already large enough? In 1985 Fama wrote an article with the title: “What‟s different about banks?”. This question is still debated and far from conclusive (Al-Ajmi, 2010; Andrea, 2007). This debate and inconclusiveness also holds for the dividend policy of banks versus other industries. When looking at the actual dividend policies Eriotis, Vasiliou, and Zisis, (2007) state that profitability is more volatile for banks because most of their assets are mainly financial assets which are reported at fair value. Fair value is a dynamic value that frequently changes as the market changes. They suggest that because of this volatility the current year dividends are not based on the previous year dividends.
When looking at the determinants of dividend policy current research is conflicting. Moreover, the research done so far is rather limited (Al-Ajmi, 2010), and none of them are based on a cross-country comparison.
15
in the banking industry has a negative relation with the level of agency problems, which is similar to other industries.
Although these two articles report similarities, other researches shows that there are differences. Casey and Dickens (2000) tested the model of Rozeff (1982) in the banking industry. Previous studies showed that the five determinants of Rozeffs model, the firm‟s growth rate, the beta, the level of insider ownership, the expected future growth, and the the number of common stockholdeers, all hold in other industries. However, Casey and Dickens study found that only the expected future growth and the level of insider ownership have a significant influence, whereas the other determinants show no relationship. Moreover, they suggest that the regulatory capital requirements may play a role in the dividend payout level.
In a later study, Casey and Dickens, together with Newman (2002) tested the determinants in the banking industry again. This time they tested seven factors, which are the investment opportunities, size, agency problems, dividend history, risk, signaling and capital adequacy. Of these determinants the first five are found to be significant. Theis and Dutta (2009) revisited the model of Casey, et. al. (2002) and concluded that the original model is robust, but that only 4 of the five suggested determinants hold their significance. Agency factors did not prove to be significant. Important to note is however is that the studies of Casey and Dickens (2000) and Casey et. al. (2002) are limited to the US market, whereas the study of Theis and Dutta (2009) is based on a sample from the Southeast and Mid-Atlantic regions. Al-Ajmi (2009) tested the dividend policy in a different market, namely in the countries of the Gulf Cooperation Council. He found that the model of Lintner (1956) also holds for the banking industry in all countries, except for Qatar. The results suggest that banks smooth their dividend payments. Both the agency and signaling theory are tested. However, these gave both mixed results. These mixed results might indicate that the determinants differ per country, which is consistent with Laporte et. Al. (2000), who suggested that the level of agency problems differ among countries. This might also explain the difference between Casey et. al. (2002) and Theis and Dutta (2009) findings.
16
of Raghavan (2005). He tested whether share repurchases are used as signals for future profits in the banking industry and, in contrary to other studies on non-financial markets, he did not find significant results. Bodla, Pal and Sura (2007) however did find signaling a relevant factor for banks in the Indian market. Boldin and Leggett (1995) found signaling to be a relevant factor for banks in the US to issue dividends. These conflicting results make it inconclusive whether signaling is or is not a significant factor.
The above mentioned findings show that the answer to the question whether banks are different in terms of dividend policy is inconclusive. Most researchers found both differences and similarities for the determinants that affect the dividend policy. As for theories explaining the policy, both the agency and signaling theory shows mixed results, which can likely be explained by the differences in the samples used, indicating that different markets are affected to a different extent by these theories. Possibly, cultural differences between the regions researched play a role in how banks are influenced by determinants for their dividend policy.
2.5 Literature review summary
17
2.6 Hypothesis building
Dividend policies in general, their theories, national cultures and differences of the banking industry have been discussed. Based on these points the theory of how culture influences the level of dividend payments will be discussed and the hypotheses will be formulated. This will be done based on the cultural dimensions of Hofstede. I include all four of the original dimensions. The fifth dimension, long term vs. short term orientation will be excluded due to the limited amount of data available for this dimension. For each of the four cultural dimension I will first discuss what the theory suggests for the dividend policy in general. After this I will indicate if this also applies to the banking industry or if something else should be expected.
However, before building the hypothesize it is important to have an insight in who decides the dividend policy. Two major parties play a role in this decision, which are the managers and the investors. The manager, as the legal body who is in control of the daily activities of the firm ultimately sets the dividend policy. The investor on the other hand, as the owner of the firm owns the earnings from which the dividend is issued.
Agency theory, on which the hypothesis are based, suggests that the manager‟s interest can differ from the investors‟ interest. Possibly they set the dividend policy based on the most favorable terms for themselves. Also, in their position as insider in the organization they have access to more information than outside shareholders which might give them a different opinion on what the best divided policy for the bank is. This is possibly stronger in the banking industry due to the high level of opaqueness. Both are reasons why the dividend policy managers set might be different from the preference of the investors.
18
negative stock market response. Based on the above I believe that both the managers as the investors play a role in the dividend policy decision, but that the investors have the decisive influence.
2.6.1 Individualism vs. collectivism
Previous literature suggests contradicting relationships between the level of dividends and the level of individualism. Fidrmuc and Jacob (2010) hypothesize that agency conflicts are higher in high individualism cultures because individualists pursue their own needs more rather then follow someone else his needs and preferences. Managers feel a need for autonomy in such cultures. On the other hand in collectivist cultures the need for autonomy is less, there is an increased level of identification with the group and a higher feeling of shared responsibility and obligations. In order to control managers in a higher individualism culture outside shareholders are expected to demand a higher amount of dividends in order to keep the managers more restricted. This is similar to Johnson and Droege (2004), who theorize that a lower level of individualism leads to a lower level of agency problems due to using social relations instead of market relations for exchange and a stronger focus on group relations instead of personal concerns. Thus implying that in cultures with a lower level of individualism manager have less need to demand dividends as a tool to resolve agency problems.
However, Shao, Kwok and Guedhami (2009) theorize an opposite relation by using the cultural dimension of conservatism from Schwartz‟s study (1992). This cultural dimension show strong similarities to the individualism versus collectivism dimension of Hofstede (Shao et. al., 2009). A high level of conservatism is similar to a low level of individualism. They theorize that in conservative cultures people are more involved in groups and want to foster positive interactions with other members of these groups. This creates a higher incentive to protect the agent-principal relationship and thus lowers agency problems. Managers prefer to show self-discipline and issue dividends. At the same time shareholders welcome this self-discipline because now they don‟t have to make demands which might harm their relationship. Thus, Shao, et. al. (2009) hypothesize that there is a positive relation between the level of dividends paid and the level of conservatism, which would be similar to a positive relation between collectivism and the level of dividends or a negative relation between the level of individualism and dividends.
19
When looking at the banking industry I believe that several factors need to be taken into consideration. Fidrmuc and Jacob (2010) base their arguments on the level of freedom or autonomy within a culture. However, the banking industry is characterized by a high level of regulation, limiting their level of autonomy. Especially regulation on the capital requirements limit the freedom of managers to issue and of the investors to demand a large amount of dividends, as retained earnings may be required to maintain the solvability. This makes this theory less strong for the banking industry. Shao et. al. (2009) their argument is based on the importance of positive interactions and group relationships within a society. This importance may be stronger in the banking industry considering the reputation of a bank is a key aspect to attract and maintain customers. If a bank does not comply with a cultures standard of fostering positive interactions they may hurt their reputation. Because of this I believe that positive interaction between the various groups of participants is more important in the banking industry. This makes dividend payments become more important for investors from collectivist societiesand less important for investors from individualistic societies. I therefore hypothesize:
H1: There is a negative relation between the level of individualism and the amount of dividends paid.
2.6.2 Power distance
20
For the banking industry no arguments were found why the above suggested theory would be different. The suggested effect of power distance on the dividend policy is based on equality and tolerance of differences. These are factors that like any other industry are also relevant in the banking industry. I therefore follow the above discussed reasoning and hypothesize:
H2: There is a negative relationship between the level of power distance and the amount of dividends paid.
2.6.3 Uncertainty avoidance
Probably the most obvious relation between dividends and uncertainty avoidance relates to the bird-in-hand theory. Dividends offer a guaranteed return for investors, whereas retained earnings have yet to prove a higher return. Therefore, a positive relation between the level of uncertainty avoidance and preference of dividends by investors would be expected. However, according to Fidrmuc and Jacob (2010) this argument fails to recognize that a low level of dividends is more predictable, which after all relates to uncertainty avoidance (Hofstede, 1991). It is relatively easy to being able to pay a low level of dividends, whereas paying a high level of dividends is more difficult due to higher requirements for sufficient funds. Fidrmuc and Jacob (2010) argue that retained earnings are a cash resource for the firm which can be used against unforeseen financial distress, offering a more stable and predictable profit. Both managers and investors might therefore prefer a lower level of dividend in high uncertainty avoidance cultures. Moreover, with a higher level of uncertainty avoidance, corporate insiders put heavier emphasis on ensuring that dividends will be paid, which investors welcome. High uncertainty avoidance will therefore be accompanied with lower dividends. Fidrmuc and Jaboc (2010) found empirical evidence that supports this line of thinking. This is in line with empirical evidence from Ramirez and Tadesse (2009), and Chang and Noorbakhsh (2009), who found that uncertainty avoiding firms have a higher level of cash in order to hedge against possible unforeseen future events. In contrary to these results however, Shao et. al. (2009) found a positive relation between the level of uncertainty avoidance and the amount of dividends paid. They did however not give any argumentation for this finding.
21
are already bound by regulation on capital requirements. It is possible that the decision of how much retained earnings to issue is for a large part the result of regulation instead of the bank‟s own preference. This would thus make the suggested theory weaker for the banking industry. Moreover, due to the critical role banks play in a society there is a too-big-to-fail factor, which significantly reduces the bankruptcy costs of many banks (Dickens, et. al. 2002). This reduces the risk profile of banks, having a reduced need of investors for retained earnings as a tool to reduce the risk. However, as discussed before, banks are characterized by more volatile profits compared to other industries due to the high level of assets that are reported at fair value. This volatility makes it harder to predict the performance of a bank, which might strengthen the suggested effect of the level of uncertainty avoidance on the dividend policy.
Based on the above there are contradicting factors that influences the suggested theory. However, although the first two arguments give reasons that the relationship is weaker they do not give any reason that the relation does not exist at all. I therefore follow the suggested theory with while noting that the effect might be less strong. The hypothesis is:
H3: There is a negative relation between the level of uncertainty avoidance and amount of dividends paid.
2.6.4 Masculinity vs. femininity
Shao, et. al. (2009) reason that in cultures with a high level of mastery (one of the dimensions of Schwartz‟s study, which shows a high level of similarity with Hofstede‟s dimension of masculinity) independence is considered important and therefore both managers and investors prefer a low level of dividend. This results in a higher level of cash available to the managers, giving them room to make their own decisions and undertake new projects more easily. A high level of dividends on the other hand might result in the necessity to attract outside capital, resulting in a lower level of freedom for both the managers and shareholders. The higher level of independence also results in a higher level of trust from the investors in the managers, limiting their need for agency measures such as high dividends. Shao, et. al. (2009) also found empirical support for their theory. Moreover, Chui, Lloyd, and Kwok (2002) found empirical evidence that there is a negative relation between mastery and leverage, showing that in a high mastery culture shareholders don‟t want to limit their and the managers freedom by debt holders.
22
in case of failure. This strengthens the theorized positive relation between masculinity and dividends for managers, as they prefer to retain profits in order to exploit investing opportunities.
As with the previous cultural dimension, I believe that regulation is an important aspect that will limit the suggested relation for the banking industry. The above discussed theory is based on independence and opportunity. However, banks are to some extent limited in their independence by regulation. On the other hand, when looking back at the economic crisis of 2007 and the amount of risks banks took in the period prior to this it is questionable how much regulation limits a banks desire for independence.
Berger and Di Patti (2003) state that the banking industry is characterized by information opaqueness. This might make it difficult for investors to have a clear understanding of how the bank is managed and thus gives managers a higher level of independence. I therefore believe that the suggested relation of Shao, et. al. (2009) also holds for the banking industry and hypothesize:
H4: There is a negative relation between the level of masculinity and the amount of dividends paid.
Figure 1 summarizes the relations suggested by the theory.
-
-
-
-
- -
Level of dividend payments IndividualismPower distance Uncertainty avoidance
Masculinity
23
3.
Data and methodology
In this section, the data and used methodology is discussed. First I will present the sample and the selection criteria used. This is followed by presenting and giving argumentation for the variables. These are divided into three groups; the dependent, independent and control variables. An overview of all the variables can be found at the end of this section in table 1. In the final sub-section the methodology used in this research is explained.
3.1 Sample
In this thesis, I research the influence of culture on the amount of dividends paid by firms in the banking sector. This leads to two main criteria for selecting the sample. First, the selected companies need to banks. Company data is acquired from Bankscope (from Bureau Van Dijk). Bankscope offers a large range of financial data for banks worldwide. Because the database is restricted to only banks the first criteria of only selecting banks is automatically selected.
The second main criterion relates to the nationality of the banks. The influence of national culture on the banks is researched, thus I select banks from countries for which cultural values are available. As discussed in the theoretical background, there are various national culture frameworks. The framework of Hofstede is used. This framework consists of cultural scores for 65 countries located in all parts of the world. The sample is therefore restricted to firms with one of these nationalities, which are included in table A1 (appendix).
Besides these two main criteria the sample is restricted by several other factors. Only firms that are listed on a stock market are selected, as non-listed firms often do not publish sufficient financial information. Secondly, only selected firms that have a known value for dividend paid for at least one year within the time range are selected. Both active and inactive banks have been included in order to prevent survivorship bias.
24
A final issue for the sample is the time range. The sample is spanning the years 2004 until 2009. This period is mainly based on the available years Bankscope offers. Although Bankscope offers the option to acquire data from multiple years, the data for before 2003 is very limited and most of the time not available. The reason to start the time range at 2004 instead of 2003 is due to lags in control variables. Considering accounting data is usually reported with a delay, the year 2010 is too recent at the time of writing this thesis in order to include. The time range includes a major economic crisis, to which especially banks were affected. As will be discussed in subsection 3.4 a variable is used to control for this.
Based on the mentioned criteria above the sample consists of 1681 banks of which 1578 are left after applying the discussed filters. This results in a unbalanced panel of 4267 firm years for the period 2004-2009.
3.2 Dependent variable
The research contains one dependent variable, which relates to the amount of dividend paid. Dividend paid is defined as the total amount of dividends paid to investors in a year, including both dividend to common and to preferred shareholders. In order to make the value comparable a ratio as proxy should be calculated. Previous literature suggests multiple ways to do this. For example Fidrmuc and Jacob (2010), and La Porta, et. al. (2000) discuss the ratios dividend to net earnings, dividend to asset ratio, and dividend to cash flow. Shao, et. al. (2009) also use the ratio dividend to assets and includes the ratio dividend to sales.
Dividend to net earnings might be the most obvious ratio as dividends is a part of earnings, which flows back to the investors. However, this ratio has several strong limitations. First of all the ratio is impossible to use in the sample, as I did not exclude firms with negative net income. Secondly, the level of net income can vary strongly among years, especially in my time range which includes a major economic crisis. Thirdly, the net earnings value is strongly influenced by the accounting standards, which, in a cross-country study, may be too different to be compared. Lastly, net earnings are relatively easily manipulated as accounting standards usually give room to make different choices. This is also expressed by the saying: “Cash is a fact, profit is an
opinion”. This statement brings us to the second ratio, dividend to cash flow, which could
25
accounting practices than net income, are relatively stable, and available for all firms in the sample.
Similar to Fidrmuc and Jacob (2010), and Shao, et. al. (2009) I do not take share repurchases and stock dividend into consideration as a factor for the dividend policy.
3.3 Independent variables
The independent variables are the cultural dimensions of Hofstede; the level of power distance, individualism vs. collectivism, the level of uncertainty avoidance, and masculinity vs. femininity. As mentioned during the hypothesis building, the fifth dimension, short term versus long term orientation is excluded in this research because data for this dimension is only available for a limited amount of countries (26). The argument for choosing Hofstede‟s dimensions and the meaning of them have been discussed in the literature review. For every country, a score varying between 1 and 100 on each dimension is given. These scores are assumed constant over time as discussed in the literature review and thus only one score for all the years is given.
3.4 Control variables
Many factors can potentially influence the dividend policy of firms. In order to more accurately calculate the influence of culture, various control variables are included in the calculation. As discussed in the literature review, there does not exist a conclusive opinion on whether the factors affecting the dividend policy in the banking industry differ from other industries. Therefore, I assume no differences in choosing the variables. The control variables are categorized in two groups; variables at the firm level and variables at the country level.
3.4.1 Firm level control variables:
On the firm level various variables are included based on previous research related to dividend. They include leverage, firm size, growth rate, profitability, and the firm‟s lifecycle.
26
leverage pay a lower amount of dividends because they have higher costs on their loans and because debt holders will try to prevent the transfer of wealth to shareholders. A third reason is given by Megginson and Von Eije (2008), who theorize that a higher level of leverage might be a proxy for companies that are larger, more stable and with a higher level of profitability.
In the banking sector the percentage of leverage is often bound by regulators. Financial firms, which include banks, are characterized by a relatively high level of leverage. However, they are affected by debt in the same way as other firms (Flannery, 1994). I therefore assume that the above-suggested relation between leverage and amount of dividends paid also holds in the banking industry. The variable is calculated as total debts to total assets.
Firm size and growth rate: Both Fama and French (2001), and Dennis and Osobov (2008) found empirical proof that small firms with a high growth rate have a lower propensity to pay dividends. This can be explained by the fact that these companies, in general, have better investment opportunities and therefore prefer to use their profits for reinvestments instead of returning it to their shareholders. Besides this, agency issues might be higher in larger and lower growth firms as they have a larger surplus of free cash flow available to the managers (Jensen, 1986). Thus, a lower amount of dividends is requirement as a tool to reduce the agency problems. Finally, Dickens, et. al. (2002) theorize that due to the to-big-to-fail effect larger banks will have lower bankruptcy cost. Because of lower bankruptcy costs, banks can acquire external financing at lower costs making it less interesting to retain earnings. All three factors lead to an assumption of a positive relation between firm size and dividends paid and a negative relation between growth rate and dividend paid. Firm size is calculated as the natural log of the firm size of last year. The growth rate is calculated as the assets divided by the assets of last year.
Profitability: Fama and French (2001) found that companies paying dividend have a higher level of profitability then non-payers. According to Fidrmuc and Jacob (2010) it is an important indicator of the funds available for a firm to pay to its investors. Moreover, Eriotis, et. al. (2007) state that the profitability in the banking industry is relatively volatile compared to other industries. Besides this, they found that current year dividends are strongly related to current year earnings and not to previous year‟s dividend policy. I therefore assume a strong positive relation between the current year profitability and current year amount of dividends.
27
Lifecycle: DeAngelo, DeAngelo and Stulz (2005) find that there is a positive relation between the amount of dividend paid and the lifecycle of the company. Companies that are further in the life cycle and thus are older tend to issue a higher level of dividend. Moreover, Baker, Dutta and Saadi (2008) surveyed firms in both the financial and non-financial sector that are listed on the Toronto stock exchange. They found that managers of both type of firms considered the firm lifecycle as an important determinant of their dividend policy. I assume that the positive relation between the firm‟s life cycle and the amount of dividends found by DeAngelo et. al. (2005) also holds in the banking industry. The stage in the lifecycle is measured by the cumulative retained earnings to the total equity of the previous year.
3.4.2 Country level control variables:
Besides firm characteristics influencing the dividend policy there are also two country level variables to control for legal and institutional determinants that might affect the dividend policy. I include the stock market development and the country income level. Both variables are proxies for investor‟s protection. Fidrmuc and Jacob (2010) argue that there is a positive relation between the level of dividends and the level of shareholders protection because minority shareholders have more tools available to influence the managers and to protect them from managers diverting wealth to them. Like Shao, et. al. (2009), I assume that in more developed stock markets the level of investor‟s protection is higher. Similar to them, I measure stock market development as the share price times the number of shares outstanding of all shares listed per country. This value is then divided by the country‟s GDP.
According to Shao, et. al. (2009) more developed countries normally have more mature financial markets and better institutions. Like them, I made a dummy indicating either a high or low income level based on the ranking giving by the World Bank.
28
Table 1: Overview of the variables
Name Symbol Type Measurement Description Expected relation
Dividend per asset
Div Ratio Dividend paid divided by assets
Ratio to measure the amount of dividends paid n.a., dependent variable Masculinity vs femininity
Masc Interval Given (from 1-100, feminine to masculine)
Main variable to test, proxy for culture
Negative
Individualism vs collectivism
Ind Interval Given (from 1-100, 1= collective,
100=individualistic)
Main variable to test, proxy for culture
Negative
Power distance PD Interval Given (from 1-100) Main variable to test, proxy for culture
Negative
Uncertainty avoidance
Unc Interval Given (from 1-100) Main variable to test, proxy for culture
Negative
Leverage Lev Ratio Total debts year-1/total
assets year-1
Leverage indicates the level of equity compared to liabilities.
Negative
Firm size Siz Interval Natural log of total assets year-1
Proxy for the size of the company
Positive
Growth rate Grow Ratio Total asset / total asset year-1
Proxy for how quickly the company grows
Negative
Profitability Prof Ratio Net income/ total assets year-1
Indicates how profitable the company in the given year was
Positive
Life cycle Lif Interval Retained earnings/ total equity year-1
Indicate the stage the company is in.
Positive
Stock market development
Smd Ratio Share price times number of shares outstanding. Accumulation of all listed companies per country. This divided by the country‟s GDP
Indicates how developed a stock market is A high level of stock market development is assumed to give better investors protection
Positive
Mandatory tax Mand dummy 0=no
1=yes
Some countries have laws that force companies to issue dividend. Might bias results
Positive
Country income level
Inc Ordinal 0= low income
1= high income
Higher income countries are assumed to have better institutions are more mature financial markets
29
3.5 Methodology and Equation
The dataset includes both longitude and cross-sectional data. Therefore, a panel-dataset is created. Due to several missing data, the panel is an unbalanced panel. The next step is to decide which kind of estimation technique to use. Chosen is to perform a Tobit analysis. Tobit is an often-used technique when the dependent variables are censored (Brooks, 2008). The model assumes that a number of observations are clustered at a limiting or censored value and makes an estimation of how these values would be if they were not censored. The advantage of TOBIT is that because of this not only the values above the limit, but also at the limit can be included without leading to biased results. This is in contrary to a standard Ordinairy Least Squares (OLS) regression. With an OLS regression the values at the limit can be included, but would lead to a biased result. They are therefore often excluded (McDonald and Moffit, 1980). The exclusion of the zero values however is inefficient and also results in biased and inconsistent values (Brooks, 2008).
In my case, the depended variable involves dividends, which is also a censored value. It is very uncommon to issue a negative dividend. Moreover, the few cases in which negative dividends were issued have been excluded in this research. Therefore, my sample has a large amount of values at zero (12%) and a large amount of values just above the zero, making a TOBIT analysis appropriate. Using TOBIT for regression analysis has also been done in other articles involving dividends, see for example Casey et. al. (2002), and Barclay, Smith, and Watts (1995).
4.
Results
This section will show the results based on the previously mentioned data. There are three parts. In the first part the descriptive statistics of the sample and the control variables is given. In the second part, the results of the regression analysis are presented. In the third and last part the results of various robustness analyses are shown and discussed and the limitations of the model are explained.
4.1 Descriptive statistics
30
On the firm level there are five variables, the dependent variable dividends/assets and four control variables. The dividend to asset ratio shows a minimum value of zero as it is a censored variable and a few negative values (12) are excluded. The low mean relative to the range of the data shows that a large amount of observations have a value close to zero. On the country level table 3 shows three variables of which two are dummies with a value of either zero or one.
Table A1 (appendix) shows the cultural values of the four researched dimensions and the number of observed companies per country. In total there are 65 countries and 1681 companies. For six countries there are no banks that fit the criteria. They are therefore excluded. A few countries have a low number of companies. Chosen is to include them as not the countries them self, but their cultural values that are measured. Excluding them might reduce the variance of the sample.
Table 2: Descriptive statistics
Mean Median Maximum Minimum Std. dev.
Div 0.0048 0.0028 0.2808 0.0000 0.0106 Prof 0.0104 0.0095 0.5261 -0.2205 0.0300 Grow 1.7195 1.1151 17.0698 0.10077 18.7094 Lev 0.8806 0.9102 1.4431 0.0000 0.1306 Lif 0.4653 0.4098 101.8410 -57.15530 2.4353 Smd 103.61 105.76 323.92 0.0000 48.066 Inc 0.9428 1.0000 1.0000 0.0000 0.2322 Mand 0.0527 0.0000 1.0000 0.000 0.2235
Notes: This table shows the descriptive statistics both on the firm and country level. For each variable the mean, median, maximum, minimum and standard deviation is given.
4.2 Regression results
In this sub-section the results of the Tobit regression will be shown. However, before any regressions are performed the issue of multicolleniarity should be considered. Previous studies, such as the study of Fidrmuc and Jaboc (2010) found strong correlation between the cultural dimensions of Hofstede. I therefore first calculate the correlation between the various variables. The results of this calculation can be found in the correlation matrix in table A2 (Appendix). It shows that there is correlation between the variables. Especially a strong negative correlation between power distance and individualism can be observed. This strong correlation would bias the result of a regression that includes all variables. I therefore will also perform separate multivariate regressions for each cultural variable.
31
Individualism shows a significant negative result with a p-value of 0.0003. This suggests that a higher level of individuality is accompanied by a lower level of dividends. Masculinity shows a significant negative relation with a p-value of 0.0000. This implies that a higher level of masculinity has a negative influence on the dividends paid by banks. However, the other two cultural variables, power distance and uncertainty avoidance, appear to give insignificant results. The model thus does not find any proof that power distance and uncertainty avoidance influence the amount of dividends banks pay.
When looking at the control variables most of them are consistent with the theory. As expected, profitability has a very strong positive influence on the amount of dividends paid. Moreover, the degree of leverage and firm size have a significant negative and positive relation, respectively. No proof however is found for the firm‟s growth rate, life cycle, nor the variables on the country level. On the country level the stock market development appears to have a significant positive relation. However, I do not find proof for significant relations with the firms growth rate, the country income level and whether the country has a mandatory tax regulation or not.
A Wald test is performed to test the goodness of fit for this model. When all variables are included the test gives a P-value of 0.000, indicating the model has a high explanatory power for the level of dividends paid.
Besides separate regressions for each cultural variable, I also perform a multivariate regression including all cultural variables. When interpreting the results of this regression it is however important to keep in mind the correlation among the cultural variables, which is especially strong between power distance and individualism. The results of the regression can also be found in table 3.
Both the level of individualism and masculinity hold their significance negative relation in this regression. Consistent with the previous regression, both power distance and uncertainty avoidance remain insignificant.
32
Table 3: Tobit panel regression results, including both regression with all cultural variables and regressions with only one cultural variable.
(1) (2) (3) (4) (5)
diva diva diva diva diva
individuality -7.52e-07 -6.53e-07
(0.000)** (0.003)**
power distance 2.17e-07 -2.68e-07
(0.360) (0.343)
masculinity -9.92e-07 -8.41e-07
(0.000)** (0.000)**
uncertainty av. 2.90e-07 2.38e-07
(0.143) (0.261)
leverage -0.0412 --0.0425 -0.0423 -0.0426 -0.0413
(0.000)** (0.000)** (0.000)** (0.000)** (0.000)**
firm size 0.0002 0.0004 0.0005 0.0004 0.0003
(0.035)* (0.000)** (0.000)** (0.001)** (0.030)*
growth rate -2.50e-06 -2.41e-06 -1.99e-06 -2.45e-06 -2.18e-06
(0.664) (0.677) (0.730) (0.671) (0.705)
profitability 0.0362 0.0377 0.0377 0.0378 0.0364
(0.000)** (0.000)** (0.000)** (0.000)** (0.000)**
life cycle 0.0001 0.0001 0.0001 0.0001 0.0001
(0.673) (0.777) (0.789) (0.758) (0.673)
st. mark. dev. -3.78e-06 -5.83e-06 -4.97e-06 -4.67e-06 -2.27e-06
(0.324) (0.126) (0.191) (0.233) (0.563) mand. tax -0.0004 -0.0016 0.0020 0.0014 -0.0001 (0.756) (0.261) (0.136) (0.300) (0.920) country inc. lvl 0.0011 0.0016 -0.0017 -0.0025 0.0001 (0.400) (0.210) (0.107) (0.023)* (0.962) constant 0.0396 0.0338 0.0403 0.0347 0.0446 (0.000)** (0.000)** (0.000)** (0.000)** (0.000)** wald chi 701.66 703.62 708.48 677.89 729.09 (0.000)** (0.000)** (0.000)** (0.000)** (0.000)** observations 4267 4267 4267 4267 4267
Notes: Shows the results of the regression including all cultural variables (5) and of the regressions
including one cultural variable per time (1 to 4). P-value is given between brackets. ** indicate significance at 1%. Goodness-of-fit determined based on Wald test. All regressions has a p-value for this of 0,0000
4.3 Robustness tests
33
individualism and masculinity, respectively (Shao, et. al. 2009). Shao, et..al. (2009), in their empirical research about the level of dividend payments and national culture used the framework of Schwartz as their main measure for culture and found significant results for both dimensions. Moreover, Fidrmuc and Jaboc (2010), who researched the same topic, used the framework of Schwartz as a robustness test as I do and found significant results.
By using the cultural values of Schwartz, it is necessarily to change the dataset. Cultural values are only available for 31 countries of which some countries are different from the countries included in Hofstede‟s study. The list of countries can be found in table A3 (appendix). Besides this difference, all other data in the sample remain the same.
Table 4: Tobit regression results using Schwartz as cultural measure
(1) (2) (3) Mastery -0.0102 -0.0103 (0.0001)** (0.0001)** Conservatism 0.0011 0.0012 (0.0784)* (0.0895)* Firm Size 0.00049 0.00063 0.0004 (0.0000)** (0.0000)** (0.0000)*
Growth rate 6.88E-06 1.43E-06 6.17E-06
(0.5059) (0.8747) (0.5464) Leverage -0.03526 -0.03899 -0.03520 (0.0000)** (0.0000)** (0.0000)** Life Cycle 0.00037 0.00037 0.00038 (0.0934)* (0.0891)* (0.0845)* Profitability 0.16474 0.17070 0.16432 (0.0000)** (0.0000)** (0.0000)**
Country income level -0.00596 -0.00210 -0.00567
(0.0057)** (0.2332) (0.0102)*
Mandatory Dividends 0.00079 0.00088 0.00073
(0.3339) (0.2859) (0.3809)
Stock Market Development -9.71E-07 -4.38E-07 -8.44E-07
(0.0021) (0.1124) (0.0209)*
C 0.07941 0.03283 -0.01031
(0.0000)** (0.0000)** (0.0001)**
Notes: Shows the results of the regression including both cultural variables (1) and of the regressions including one cultural variable per time (2 and 3). P-value is given between brackets. ** indicate significance at 1%. ** indicates a significance at 10%. Goodness-of-fit determined based on Wald test. All regressions has a p-value for this of 0,0000
34
a significant negative relation at a 10% significance level. For conservatism a significant positive relation is found at 1%. This is similar to the previous regression with Hofstede as the dimension of individualism has a significant negative relation. In addition, the result of conservatism is similar to the result for masculinity in Hofstede‟s regression. The results of both cultural values of Schwartz‟s study are consistent with the previous results using Hofstede. This indicates that the results are robust to other cultural measures.
5.
Discussion
In this section the acquired results will be analyzed and discussed. The analysis is structured based on the hypothesis and discussed for each cultural dimension.
5.1 Individualism
