What have we done?! The impact of economics on the beliefs and values of business students

We examine the effect of studying business on the beliefs and values of students and study the role of economics in influencing this effect. We observe significant differences between business students and students from other disciplines, among which economics, at the start of their first year. We also discover that some of these differences persist or are reinforced at the end of the year. Furthermore, we find changes in beliefs and values of business students that take only one year to manifest. Above all, while we observe that some values and beliefs of business students change over time, we observe no such changes for economics students. This suggests that the effect of studying business is not entirely driven by exposure to economics in business studies.


Introduction
The spread of economics in business schools and its influence on management theory have raised the concern that it may sap the values and norms of business students, resulting in undesirable management practices. In an influential paper, Ghoshal (2005) argues that economics was likely responsible for various cases of corporate misconduct. He traces that influence back to Milton Friedman's statement that the only responsibility of managers is to make as much money for their stockholders as possible. Ghoshal (2005) more generally claims that the reliance of economics on agency theory, transaction cost theory, and game theory has reduced business students' sense of moral responsibility, resulting in ruthless managers.
His paper initiated a strand of literature emphasizing the ethical implications for students at business schools of what has commonly been referred to as the "dismal science" since Thomas Carlyle (1853) anonymously coined the expression in 1849. Huehn (2008), for instance, argues that economics describes self-interested and brutish individuals and, therefore, legitimizes selfishness and fosters a "gloomy vision" of human behavior. Wang et al. (2011) argue that economics blurs the distinction between self-interest and greed, resulting in more positive attitudes toward the latter. Fotaki and Prasad (2015) claim that economic concepts promote values of individualism and profit maximization, which are treated as unavoidable human traits. Racko (2019) argues that economics' assumption of utility maximization legitimizes hedonism and delegitimizes universalism. Moreover, its emphasis on efficiency legitimizes the seeking of power both as a means and as an end. A final argument, put forward by Rubinstein (2006) and Fotaki and Prasad (2015), is that the reliance of economics on mathematics tends to treat human problems as mathematical ones. As a result, profit maximization becomes a legitimate and sufficient objective.
Maybe unsurprisingly, the beliefs, values and attitudes of business and economics students have been under scrutiny and research has intended to also empirically determine whether these students are indeed less ethical or social than others. The results of these investigations are mixed. On the one hand, business students have been found to be less ethical or social than other students, see e.g. McCabe et al. (1994), Smyth and Davis (2004), Wang et al. (2011), Arieli et al. (2016), andRosengart et al. (2020). Similar findings have been reported for economics students, for instance by Marwell and Ames (1981), Rubinstein (2006), and Racko (2019). These findings are in line with the presumptions of the literature spurred by Ghoshal (2005). On the other hand, other studies report no significant differences between business and/or economics students on the one hand and other students on the other, see for example Malinowski and Berger (1996), Hummel et al. (2018), Neubaum et al. (2009), or Meiseberg et al. (2017. Moreover, the role of economics in generating observed differences between business students and other students is difficult to identify. To do so, one should track changes in students' beliefs and values over time and relate them to their degree of exposure to economics. Delis et al. (2019) are able to track individual students over time in their study into political preferences of business and economics students. However, they pool economics and business students together, and, as such, cannot comment on exposure to economics. Racko et al. (2017) use a longitudinal survey of students who follow different curricula. They observe that students registered at a more homogenous economic program display larger value changes than students in a more heterogeneous economic program. However, it is difficult to rule out that the results of this study are not affected by unobserved heterogeneity, since the programs that Racko et al. (2017) consider not only differ in the exposure to economics, but also in terms of teaching language and methods.
In this paper, we address these concerns by taking advantage of a survey among students registered at a leading Belgian business school. In this survey, a series of questions about students' beliefs and values were asked, both in their first week and at the end of their first year at university. The key characteristic that makes the Belgian business school particularly suitable to study the role of economics in affecting business students is that it offers both a bachelor in business and a bachelor in economics. The two bachelors differ in their names, but more importantly in the share of economics in their curricula. By contrast, their fees are the same and studies follow the same structure. Accordingly, they cannot generate differences similar to those observed by Froelich (2016), who observed that the structure of college degrees affects the propensity to enrol in doctoral studies. By studying how the answers of the students of the two bachelors change over time, we can therefore gauge the role of exposure to economics. We refer to changes in beliefs and values over time as socialization effects. Socialization thus refers to what makes a typical business student or a typical economics student different from a typical student from another discipline during the course of studying. However, socialization operates on students who have self-selected in a discipline. So, to sift the effect of socialization from the effect of self-selection, we also use surveys conducted among students of three additional fields of study in the same university: psychology, law, and social sciences.
A second desirable characteristic of the business school that we study, is that it is a department of a public university, like the departments to which it is compared. It is therefore organized in the same way, ensuring greater comparability. Moreover, all courses are given in the same language, French, in all departments. Furthermore, the fees paid by students are exactly the same for all students and there is no prerequisite or official selection to choose a bachelor, provided one has completed high school. As such, self-selection effects we find are driven by the proclivity of students for a given discipline, and socialization effects by the influence of studying that discipline. Since our sample contains students from the same university, we focus on self-selection and socialization effects of studying in different fields compared to studying in general. For findings on the latter, one may refer to Hastie (2007) and Mayhew et al. (2016).
A useful feature of the survey is that it carefully distinguishes beliefs and values. We consider beliefs to be 'any simple proposition, conscious or unconscious, inferred from what a person says or does ….' (Rokeach 1968, p. 113). We define values as a special type of belief about what ought to be or how someone ought to behave. Values have a normative character (Rokeach 1968(Rokeach , 1973Schwartz 1992). For conciseness's sake, we refer to the beliefs that are not values simply as beliefs.
To the best of our knowledge, we are the first to look at how economics and business studies affect both beliefs and values.
Previewing our results, we find significant differences in beliefs between business students and economics students at the start of their academic career. We also find differences in both beliefs and values between business students and students from disciplines other than economics. There is, thus, a self-selection of students into business studies that is different from self-selection into economics or other disciplines. We also find that some of the differences in beliefs and values persist beyond a year of socialization. However, others disappear over the course of the first year. These findings suggest that there are socialization effects. Moreover, we find that business students change their beliefs more than students from other disciplines. This suggests that it is not just higher education, but also disciplines that influence beliefs and values, which provides additional support for socialization effects. Above all, while we observe that beliefs and values of business students change over time, we find no similar trend among economics students. This finding contrasts with the contention that the effect of studying business is driven by the role of exposure to economics in business studies.
The rest of the paper is organized as follows. The next section surveys the existing literature to provide the framework in which this study is embedded. Section 3 describes the survey and our empirical strategy. Section 4 reports our findings and Sect. 5 concludes.

The difference of business studies
In this section, we discuss how studying business may make students different and why students who choose business studies may be different from other students.

How business studies make students different
Most teachers and professors likely consider, or hope, that they foster the way their students understand how the world works by providing them with new information. Accordingly, most disciplines officially take a positive, as opposed to normative, perspective. This is also the case for many subjects taught in business, see for example the classic works from Keynes (1891), Robbins (1932) and Friedman (1953). Studying a certain discipline should therefore affect the beliefs of students, and in many instances, it is also found that it does (see Hastie (2007;Mayhew et al. 2016); ; for comprehensive reviews of the literature).
However, some disciplines have ethical implications and may, as such, also affect the values of students. The discussion in the introduction emphasized the ethical implications of economics (Pitelis 2002;Ghoshal 2005;Huehn 2008Huehn , 2014Wang et al. 2011;Fotaki and Prasad 2015). Additionally, some subjects taught in business schools are value-laden, and either explicitly or implicitly contain value judgments. The reference to ethical values is, by definition, explicit in the teaching of business ethics and corporate social responsibility, in which teaching has expanded worldwide (Matten and Moon 2004;Koehn 2005;Christensen et al. 2007;Neubaum et al. 2009;van Liedekerke and Demuijnck 2011). Other disciplines taught in business schools are also ultimately prescriptive, because they are meant to train decisionmakers, see e.g. Hunt (1976). Etzioni (1991) even goes as far as claiming that moral messages, whether implicit or not, are contained not only in many of the subjects taught at business schools, but also in matters such as admission criteria, grading rules and prescribed literature.
At first pass, one may argue that values are stable, as Rokeach (1968Rokeach ( , 1973 and Schwartz (1992) point out, and should not change over the course of studying. Nevertheless, there are three channels through which studying a discipline may affect students' values: teaching, social interaction with peers and staff, and congruence through differential attrition.
Teaching is the most direct way whereby studying business may affect values. As Neubaum et al. (2009) recall, ethics can be taught and the effects of it have been observed (Krawczyk 1997;Menzel 1997). More generally, many disciplines carry implicit values. Ghoshal (2005), for instance, emphasizes that no social theory can be value-free. Racko (2019) argues that students tend to align their values with the normative priorities of their field of study. As such, business students will align their values with those that the teaching of business carries. Desplaces et al. (2007) and Cantoni et al. (2017) find evidence in line with this contention.
Social interaction is the second channel through which students may change their values. During higher education students interact with faculty members and peers. This will prompt them to internalize the values of their field (Weidman 1989;Hastie 2007;Racko et al. 2017). Mayhew et al. (2016) even conclude that the effect of peers is larger than exposure to curriculum or faculty members, while Dey (1996) finds the two effects to be of similar magnitudes. Algan et al. (2015) illustrate the role of interaction with peers using the random allocation of first-year students to tutorial groups as a quasi-experiment. These authors observe that the political preferences of peers tend to converge as soon as 6 months after the beginning of the academic year.
The third channel affecting the values of students is differential attrition. Institutions can evaluate individuals more positively when there is a stronger correspondence in ideologies, which in turn can result in the resigning or dropping out of those individuals with diverging views. To substantiate this claim, it has been found that a higher level of congruence between university students and their educational institution is associated with better academic performance and more satisfaction (Nafziger et al. 1975;van Laar et al. 1999;Kemmelmeier et al. 2005). This would imply that students, who do not adhere to the values of business studies, might be more likely to drop out or fail. As a result, the ideologies of the students who graduate or complete an extra year of study will be more aligned with those of their discipline. While differential attrition does not prompt the values of individual students to change, it selects the pool of students that continue with their studies. This, in turn, increases the congruence between the values of the pool of students that remain with those of their field.
Throughout the paper, we use the term 'socialization' to refer to the total effect of the three channels that can affect values and beliefs at the group level over the course of studying a certain discipline. As such, we refer to socialization as what makes a typical business student different from a typical student who studies another discipline.

3
The empirical evidence on the effect of socialization on values is mixed. McCabe et al. (1994) show that management students' values remain relatively stable over the course of 2 years, while the law school program increases the importance of intrinsic values and decreases the importance of accomplishments. Gandal et al. (2005) could observe little change over the course of the first year in the importance that students give to values of self-enhancement. Likewise, Hummel et al. (2018) observe no change over time of cognitive moral development in business students, and Delis et al. (2019) report no evidence of a causal effect of majors in business or economics on students' political ideology. Yet, other studies do observe changes over time. Racko et al. (2017) find that studying economics leads to larger value changes and report a significant role for interaction with peers, and Racko (2019) finds that studying economics results in increases in the values of power, hedonism, and decreases in the values of self-direction. Furthermore, Allgood et al. (2012) find that graduating from business and economics majors influences civic behaviours, such as donating to a political party or party membership and O'Roark (2012) finds that members of Congress that have majored in economics are more likely to vote for free trade outcomes. On the other hand, van Laar et al. (1999) find reductions in anti-egalitarian values after studying what they refer to as hierarchy-enhancing disciplines, which include both business and economics. In line with the latter findings, Arieli et al. (2016) observe a decrease in benevolence values of business students over the course of their first year of study.

How different students choose business studies
If business students are different from students of other disciplines when they start their studies, any differences observed at the end of their studies may simply reflect initial differences. In other words, the impact of studying business must be weighed against the self-selection bias of students, a point initially made by Yezer et al. (1996) and substantiated by later studies. Sidanius et al. (2003), for example, find a lower propensity among students from disciplines other than business to support the belief that group-based inequality in society should be maintained or promoted and show that this is due to self-selection. Frey and Meier (2005) find that students of economics and business tend to donate less than students of other disciplines in the real-world to funds supporting needy and foreign students. They argue that this finding is driven by self-selection.
Whereas the above cited studies focus on self-selection, others report both socialization and self-selection effects. When both are considered, the latter is typically stronger and more systematic. Looking at the political attitudes of students, Elchardus and Spruyt (2009) report evidence of strong self-selection of students across fields, but also find small discipline-specific effects of socialization on attitudes relating to ethnocentrism, authoritarianism, and individualism. Additionally, Cipriani et al. (2009) find that students, who prioritize profit maximization, are more likely to enroll in economics. Surprisingly, they find that students who specialize in management prioritize profit-maximization less in their third year than in their first year, while there is no variation observable among students specializing in What have we done?! The impact of economics on the beliefs and… accounting or economics. Furthermore, comparing business school students with students of social work, Arieli et al. (2016) find substantial selection effects into both fields, and observe some socialization effects for the business school students that arise in the first year of their academic education. Moreover, Fischer et al. (2017) observe both a self-selection and a socialization effect of studying business and economics. The same holds for Hammock et al. (2016), however, they focus solely on economics students. Contrasting these findings, Frey et al. (1993) only find a selection-effect for economics students when studying fairness beliefs, but do not find any evidence for socialization. Finally, Haucap and Just (2010) find that students of business and economics have a higher propensity to prefer market mechanisms than students of other disciplines. Moreover, this propensity increases during the course of their studies.
Whereas the findings of Haucap and Just (2010) suggest that self-selection and socialization influence beliefs, the results of the other studies cited above apply to values or the implication of certain values. And to be fair, while self-selection has been repeatedly observed, it is not systematic. Hummel et al. (2018), for instance, observe no selection effects when studying cognitive moral development. Whether beliefs and values are subject to either self-selection or socialization is, therefore, an empirical matter.
Following from our review of the literature, we test a series of four hypotheses related to self-selection into a discipline and subsequent socialization effects of studying that discipline. To uncover the existence of self-selection into studying business, we test the following hypothesis: H1 (self-selection): The beliefs and values of the typical business student are different from those of other students at the beginning of their studies.
Basically, we test whether business students initially differ from students in other fields when they start their studies. Not rejecting this hypothesis implies that there is a self-selection effect.
To test socialization, we compare the beliefs and values of students of the same discipline at the end of their first year with their beliefs and values at the beginning of the year. We therefore test: H2 (socialization): The beliefs and values of the typical business student change over time during their first year at the university.
Not rejecting this hypothesis implies a socialization effect of studying business. Furthermore, changes in the beliefs and values of students of other disciplines may contribute to business students being different from other students. Thus, we also test this hypothesis on students from other disciplines.
Doing so additionally allows us to test the influence of exposure to economics. If economics is the driver of business students being different, then the changes observed for business students should also be observed for economics students. We therefore test the following hypothesis: H3 (the role of economics): The beliefs and values of the typical economics student change over time during their first year at the university in the same direction as those of the typical business students.
Not rejecting this hypothesis would lend credence to the contention that the teaching of economics drives the changes in the beliefs and values of business students.
Conversely, not finding such evidence would make it hard to reconcile our results with that contention.

The survey
In this section, we first describe the questions that were used to measure the beliefs and values of students and then how the survey was administered. We lastly specify our empirical strategy.

The questionnaire
To address our research questions, we need to prompt students to report their beliefs and values. 1 As we want to measure the impact of studying business and economics, we focus on beliefs that are likely affected by those studies. To measure these beliefs, we rely on a survey that asks both standard questions from large crossnational surveys such as the World Values Survey and the General Social Survey, and questions specifically designed for the survey. 2 Introductory courses in economics routinely discuss the welfare loss of monopolies after they have argued that competitive markets maximize social welfare. To capture beliefs about the existence of deadweight losses created by the market power of firms, we relied on the following question: 'Generally speaking, do you think that when a firm grows it abuses its size?' This question allows us to test to what extent students are exposed to theories and models of monopolistic firms. Business and economics students are likely exposed more to these theories and models, whereas this might not be the case for students from the other disciplines.
Related is the belief that exchange is mutually beneficial. This notion lies at the heart of standard economics courses (Blaug 1997;Goossens and Méon 2015). To measure this belief, we rely on the following question: 'Generally speaking, do you think that when two individuals exchange a good or a service for money, it is that it makes them both better off?' This question was designed by Goossens and Méon (2015) for their paper on differences between economics versus other students regarding views on market transactions. 3 For this and the previous question, the answer options ranged from 1 (disagree) to 5 (agree).

3
What have we done?! The impact of economics on the beliefs and… Economics courses also discuss the relative merits of laissez-faire versus public intervention. As such, it discusses market and government failure. Students of these courses can therefore be expected to update their beliefs about the extent to which they can trust the state to implement efficient policies, as the level of trust is an indication of government performance and support for the state (Easton 1975;Keele 2007). We assess students' trust in the state by using the following question: 'Generally speaking, would you say that the state can be trusted to achieve the missions that it has been given?' The answer options ranged from 1 to 5, where 1 corresponds to disagreeing with the question and 5 with agreeing with it. A version of this question appears, for instance, in the World Values Survey.
Furthermore, we rely on a question that appears in the same form in the World Values Survey and the General Social Survey to measure beliefs about the origin of someone's lot in life: 'Generally speaking, do you think that whatever one's lot in life, it has always been deserved?' Answer options ranged from 1 (disagree) to 5 (agree). Introductory courses in economics emphasize that in a competitive economy, all factors of production, including labour, are remunerated at their marginal productivity. Accordingly, differences in wages and incomes are presented as resulting from differences in productivity. We therefore expect students exposed to economics to agree more with the statement. 4 We also capture beliefs regarding students' intergenerational mobility expectations. Business and economics are perceived as studies that improve students' prospects on the labour market resulting in upward mobility. We expect students who have chosen those studies to hold different beliefs in that respect, possibly revising them as they learn the workings of the labour market. We measured these beliefs with a question asking whether respondents think their standard of living in twenty years will be better, equivalent or worse than that of their parents. The answer categories for this variable ranged from 1 (worse) to 3 (better).
The previous section recalls that business and economics studies have been accused of fostering selfishness, self-enhancement, and materialistic values. We therefore rely on questions that capture such values.
We use two questions that capture whether students value equality. The first question originates in Schwarz's (1992) value questionnaire and is posed as a statement: 'As a guiding principle in my life, equality (equal chances for all) is …' Answer categories ranged from 1 (against values) to 7 (fundamentally important). We complement this by a hypothetical scenario used to gauge how averse students are to inequality. In the survey, students were asked to choose between an unequal pay rise for themselves and an equally capable colleague and an equal but lower pay rise for them both. A student's choice for the equal pay rise reveals an aversion to inequality. This question originates in the work of Bazerman et al. (1992).
To capture whether students value wealth, we rely on a question also originating in Schwarz (1992) asking students to react to the following statement: 'As a guiding principle in my life, wealth (material possessions, money) is …' Answer categories ranged from 1 (against values) to 7 (fundamentally important). 5 As stated by Schwarz (1992), wealth is considered a power type value. Those that value wealth likely aim for status, prestige and control. In line with the literature, we expect business and economics studies to foster this value.
Economics is not the only discipline that may affect other students' beliefs or attract students with specific beliefs. Accordingly, the answers of students of other disciplines may also stand out. For instance, exposure to competition law may make law students more suspicious of large firms. Also, because psychology emphasizes unconscious decisions and various cognitive biases, students of that discipline may be less confident in the notion that voluntary transactions make participants better off. Finally, the emphasis of sociology on social reproduction may make students of social science less skeptical about the statement that one's lot in life is deserved. Although we cannot discuss the possible effect of each of the disciplines that we consider on the answers of respondents to all the questions of the survey, we test for the effect of each discipline with specific dummies in all regressions.
Finally, the questionnaire included questions on students' demographics: age, gender, and which faculty they belonged to.

Administration of the survey
The survey was administered at a leading Belgian business school, either during lectures or exams. Before handing out the questionnaires, a standardized introduction was provided emphasizing that the survey was designed by scientists for scientific purposes only; that answers were completely anonymous; that the survey was not an exam; and that there were no 'good' or 'bad' answers, to avoid prompting socially (or academically) desirable answers.
The survey was administered in five distinct bachelor's degrees. In addition to the bachelors in business and in economics, the survey was also set out in the bachelors in psychology, law, and social sciences. An important feature of the curricula of the bachelors in business and in economics is that they are very similar. They share five topics, some of which are taught jointly to the two groups of students (e.g. mathematics, statistics, economic geography, and economic history). In terms of the European Credit Transfer and Accumulation System (ECTS) those common courses are worth 25 out of 60 credits. What matters to our study is that, although both groups take an introductory course in microeconomics and macroeconomics, the courses are 50 percent larger in the bachelor in economics than in the bachelor in business. In terms of the European credit system, these economics courses are worth 15 credits in the bachelor in economics and only 10 in the bachelor in business. 6 The two waves of the survey were conducted in the academic year 2006-2007. The first took place amongst first year students during the first two weeks of the academic year to ensure they had had as little education and socialization as possible. 7 The first wave therefore allows the identification of initial differences between business students and students from other faculties. The second wave of the survey was administered in the same cohort but at the end of students' first academic year. 8 Due to university regulations, identifying who filled in the questionnaires was not possible. We can therefore not track individual students over time. We, thus, examine if and how the typical business or economics student differs from the typical student of other disciplines at the beginning and at the end of the year using the pseudo-panel structure of our dataset. By virtue of the structure of our data, any identified changes in beliefs and values of the average student of different disciplines can be the result of a combination of social interactions, learning and/or differential attrition.

Empirical specification
Firstly, we look at differences across disciplines in beliefs and values at the start of the year, i.e., the self-selection effect. We also estimate if and how beliefs and values still differ at the end of the first year, i.e. whether differences are persistent or not. To do so, we adopt the following specification: In this specification, Attitude ij is the score of individual i on belief or value j. Our independent variables of interest, Discipline ik , are discipline dummies indicating in which faculty k individual i is enrolled. Our reference category consists of business students. As controls we include a gender dummy ( Female i ) that is 1 for female (1) Secondly, when considering changes in beliefs and values over time, we focus on whether and how beliefs and values change over the course of the first year within a discipline. For this purpose, we specify the following model: We capture changes over time with the dummy variable Endoftheyear i , which is equal to 1 if individual i filled in the survey at the end of his/her first year. As such, our reference category consists of students who participated in the survey at the start of their studies. We control for gender with the dummy variable Female i . 9 We estimate the above model for each faculty separately.
The majority of our dependent variables are categorical, ranging from 1 to 5 or 7, and one is binary (i.e. inequality aversion). Consequently, we estimate ordered probit models for the former and a probit model for the latter. 10

The results
Before discussing the results of estimating the models specified in the previous section, we take a first look at the data and examine the characteristics of our sample.

A first look at the data
There are about 2300 observations in our dataset: approximately 1600 at the start of the first year of study and 700 at the end of the first year. If we pool the observations over the first year, 26% study business, 15% economics, 16% psychology, 23% law, and 21% social sciences. Table 1 shows the average age of our respondents, as well as the share of female students and the number of observations split according to discipline and beginning versus the end of the first year. 11 This table shows that students of Business and Law are slightly younger when they start their studies where i = 1, … , n; and j = 1, … , 8 1 3 What have we done?! The impact of economics on the beliefs and… Table 1 Characteristics of respondents-split between begin and end of the first year and according to disciplines For the age of the respondents the mean and standard deviation (in parentheses) is given. For gender, the percentage of females in the sample is given. The summary statistics are split between the beginning of the first year and the end of the first year. The t-test, p-values in parentheses, indicate whether there is a significant difference between age and gender in the two time periods the survey was administered. compared to the other three disciplines. In terms of gender, Psychology stands out with 80% female students, followed by Law and Social Sciences. In Economics and Business, the opposite holds; there are more male students than female students. We also conduct t-tests (see Table 1) to test whether the students differ significantly in terms of age and gender at the two points in time the survey was administered. Unsurprisingly, in all disciplines students differ significantly in terms of age at the end of their first year, compared to the beginning of the first year, as the second wave of the survey was carried out approximately eight months after the first. Regarding gender, there are no significant differences for all disciplines under consideration. This is reassuring in light of concerns regarding non-random attrition. Figure 1 shows histograms of business and economics students' beliefs at the beginning of the first year and at the end of the first year. Figure 2 shows the same for values. Similar figures for the other three disciplines can be found in the appendix (Figs. 3,4,5). Figures 1 and 2 allow us to examine how opinions changed over the course of the first year for business students and how this compares to economics students. A few things stand out. Figure 1 and 2 show that some beliefs and values of business students did not seem to change much between the beginning and the end of the first year, specifically the belief that exchange is mutually beneficial and valuing wealth in one's life. On the other hand, business students seem to agree more that someone's lot in life is deserved and seem to value equality less at the end than at the start of the first year. These students are also less inclined to believe that firms abuse their size. Moreover, a larger part of business students expect to fare better than their parents at the end of the year than at the start of the year. For economics students, there seem to be similar trends in beliefs and values over time. 12

Differences across disciplines: selection and persistence
Estimating Eqs. 1 and 2 allows us to refine previous results by teasing out the role of demographics. Table 2 shows the results of estimating Eq. 1 for a sample restricted to the beginning of the first year in the upper panel and a sample restricted to the end of the first year in the bottom panel. From the table we see that there are many differences in beliefs and values between business and other students, both at the start and at the end of the year. As we test our hypothesis on a set of eight beliefs 1 3 What have we done?! The impact of economics on the beliefs and…  and values, we calculated standard p-values as well as q-values based on the false discovery rate and, thus, adjusted for multiple hypothesis testing. 13 If we compare business and economics students, there is a significant difference at the 10%-level based on the p-value regarding expectations of intergenerational income mobility at the start of the academic year (Table 2, column 5). This seems to suggest that economics students are less likely than business students to believe that they will earn more than their parents in the future. However, this estimate loses significance, when correcting for the false discovery rate. We also find that this difference is not persistent over time. This would suggest that there are changes in the expectations of business students, economics students or both. Moreover, at the end of the year, there is a marginally significant difference between business and economics students with regard to the belief that the state can be trusted ( Table 2, column 4). This would imply that economics students are less likely to believe in the trustworthiness of the state at the end of the year than business students. However, also here we find that this estimate is not significant based on the adjusted q-values. Regarding the other beliefs and values under investigation, there are no significant differences between business and economics students at the beginning of the year, nor at the end of the year.
We then compare business students and students from the other three disciplines under consideration. At the start of the year, business students on the one hand and psychology and social sciences students on the other hand differ with regards to all but one belief, whereas law students differ on six (five based on adjusted q-values) out of eight beliefs and values. However, some of these differences are persistent over time while others are not. For example, psychology students are less likely to trust the state than business students (Table 2, column 4), however, this difference for psychology and business students is not persistent (based on the q-value). Psychology, law and social sciences students are also more likely to believe that firms abuse their size compared to business students (Table 2, column 1). However, the difference with business students is only persistent for social sciences students. Furthermore, social sciences students are less likely to believe that exchange is mutually beneficial compared to business students at the beginning of the year, while at the end of the year this difference between social sciences and business students is not statistically significant anymore (Table 2, column 2). It also holds that students from the three other disciplines are more likely to value equality and less likely to value wealth than business students (Table 2, columns 6, 7 and 8). Some of these differences in values are persistent, such as the difference in values between business students and social sciences students. However, the significant differences between 13 Q-values are based on the false discovery rate approach to multiple hypothesis testing. Where a p-value of 5% indicates that 5% of conducted tests are the result of a Type I error, a q-value of 5% indicates that 5% of all relations that are statistically significant are the result of a Type I error. This procedure therefore greatly reduces the chance of rejecting the null hypothesis, where it should be accepted. Q-values are calculated using the Simes-method (Benjamini and Hochberg 1995;Benjamini and Yekutieli 2001;Simes 1986). We use the STATA qqvalue command (Newson 2010)). We thank an anonymous referee for suggesting to correct for multiple hypothesis testing. business students and law and psychology students on, for example, valuing equality disappear over time. 14 We can therefore not reject hypothesis 1 for many beliefs and values under investigation. Our results show that, from the outset, the beliefs and values of business students are different from students of other disciplines. This shows that business students self-select into the field on the basis of certain beliefs and values. We also find that some differences in beliefs and values between business students and students from other disciplines persist over the course of the first year, while others do not. This tells us that, over the course of the year, changes in beliefs or values occurred for either business students, students of the other discipline, or both. Although we are unable to distinguish between these three scenarios based on the results in Table 2, this finding is indicative of socialization for students of either one or of both disciplines. 15

Differences within disciplines: is it economics?
In this section, we present the results of estimating the model specified in Eq. 2. We examine whether and how beliefs and values change within a discipline over the course of the year. The reference category, thus, consists of students of the relevant discipline at the start of the first year. As we cannot track individual students over time, the changes that we observe are changes in the typical student and can be due to a combination of social interactions, learning and differential attrition effects. The results are reported in Table 3. As with the results in Table 2, we calculated standard p-values as well as q-values based on the False Discovery Rate and, thus, adjusted for multiple hypothesis testing. 16 15 To check whether differences increased over time, we estimated a model pooling all observations together and including an interaction term between disciplines and end-of-year dummies. To be interpretable the model was estimated as a linear probability model. Table 25 shows the estimated coefficients of the LPM with interactions and Table 26 shows marginal effects. We find three cases in which beliefs at the end of the year are significantly different from the beginning of the year, i.e. where the interaction term is significant (compared to the reference group of business students at the beginning of the first year). This is evidence in favour of socialization. However, we also find that, for some beliefs and disciplines, differences between business and other students are persistent over time, i.e., the marginal effects at the beginning of the year are not significantly different from those at the end of the year. This is evidence against socialization. Regarding values, the estimation results show no significant interaction terms, i.e., no evidence for the effect of discipline on values being different at the end of the year compared to the beginning of the year.

3
What have we done?! The impact of economics on the beliefs and… We find that business students have the most changes in beliefs and values over time. At the end of the first year, business students are more likely than at the beginning of the year to believe that someone's lot in life is deserved (Table 3, column 3) and that they will earn more than their parents in the future (Table 3, column 5). Moreover, business students are the only students for whom these beliefs significantly change over time. Conversely, they are less likely to value equality after their first academic year (Table 3, column 6). The evidence reported in Table 3 does not allow us to reject hypothesis 4; there are changes in beliefs and values of the typical business student over the course of the first year. 17 By studying marginal effects, which can be found in the appendix, we can comment on the effect sizes we find regarding changes in beliefs and values of business students over time. Taking the belief that someone's lot in life is deserved as an example, we find that male business students are about 7 percentage points less likely at the end of the year to disagree with the statement that someone's lot in life is deserved (i.e., to choose answer option 1) than at the beginning of the year. They are also about 5.5 percentage points more likely to believe that someone's lot in life is deserved (i.e., choose answer option 4 or 5) at the end of the year (see Appendix, Table 19). Regarding valuing equality, male business students are 6 percentage points less likely to view equality as fundamentally important in their life at the end of the year than at the start of the year. These students are 3 percentage points more likely to find equality 'important' (i.e., answer option 4, the middle of the scale) and 4 percentage points more likely to find it either not or not at all important (see Appendix, Table 22).
An important finding is that we find no significant changes over the course of the first year for economics students in Table 3, unlike what we observe for business students. We therefore reject hypothesis 3, regarding the role of economics in making business students different. Accordingly, the changes over time that we observe for business students are unlikely driven by their exposure to the teaching of economics. If economics courses were the culprit, the trends in the values and beliefs of economics students should be at least qualitatively similar to those observed for business students, since economics students are exposed to more economics than business students. Instead, we observe that the values and beliefs of economics students are inert. Exposure to economics can therefore not be driving the trends in the beliefs and values of business students.
The take-away of this section is that we cannot reject hypothesis 2 in several instances: some beliefs and values of business students change over time during their studies. We, therefore, find that socialization affects both the beliefs and values of the typical business student. For other disciplines and for some of the beliefs and values under consideration, there are also instances where hypothesis 2 is not rejected. These changes are likely to contribute to the difference between business students on the one hand and students from other disciplines on the other. On the  In all models, we control for age and gender. Columns 1-5 show estimation output with beliefs as dependent variable and columns 6-8 for values as dependent variable.
Main independent variables are discipline dummies equal to 1 if respondent is a student of that discipline and 0 otherwise. The reference group consists of business students. Coefficients are reported in bold. Robust standard errors are in parentheses. For the models in column 1-6 and 8, the coefficients are estimated using an ordered probit model. For those in column 7, coefficients are estimated with a probit model. Models are estimated for two samples: one sample obtained at the beginning of the first year of study and one obtained at the end of that year. The table shows standard p-values as well as q-values (adjusted for multiple hypothesis testing and calculated using the Simes-method, STATA qqvalue command (Newson 2010)). Significance is indicated as follows: ***p < 0.01, **p < 0.05, *p < 0.1  contrary, we do not observe socialization effects for the typical economics student over time and, therefore, reject hypothesis 3. 18

Conclusion
We empirically study how the beliefs and values of business students differ from those of students from other disciplines, and how they vary over the course of their first year. We find that business studies attract students that are less concerned about the market power of firms and more confident in the benefits of exchange than other students. Business students are also more inclined to believe that individuals' lot in life is deserved. This suggests that business students have more trust in market mechanisms than other students. Additionally, they are more optimistic about their future prospects. Our results also show that business students tend to value equality less, and material possessions and wealth more. It thus seems that business students tend to be more self-interested than students from other disciplines from the outset of their studies. Over time, some of the initial differences in beliefs and values are reinforced. Firstly, initial differences in beliefs and values between business students and students from other disciplines are, in all but one case, persistent over time. Secondly, when we consider changes within the group of business students, we find that they become more optimistic about their future over the course of the year and are more likely to believe in the role of effort in determining someone's outcome in life. Furthermore, they value equality less over time, again suggesting that business students become somewhat more selfish.
In line with previous findings (van Laar et al. 1999;Delis et al. 2019), we also provide evidence that the beliefs and values of business students are the closest to those of economics students. Nevertheless, we find some indicative evidence that business students are more optimistic about their future standard of living at the beginning of their study program than are economics students, thereby suggesting that business students might self-select into the field on the basis of different beliefs. Importantly, however, while beliefs and values of business students change over time, we observe no evidence that those of economics students do. This finding contradicts the contention that the harmful effects of business schools originate in exposure to economics (Ghoshal 2005;Rubinstein 2006;Fotaki and Prasad 2015;Racko 2019). If they did, any changes in the values and beliefs of economics students should be qualitatively similar to those of business students. Since we find that it is not the case, there must be additional effects of studying business that prompt 18 One may argue that part of the changes in values and beliefs are due to initial differences being amplified by interactions between peers, as we point out in Sect. 2.1. This mechanism is a dimension of socialization and we cannot rule it out. However, the large number of insignificant results in Table 3, in particular for economics, suggests that that mechanism is not massive. We thank an anonymous referee for raising this point.
What have we done?! The impact of economics on the beliefs and… students to select it as their field of study and to become even more different over time.
Even though it is standard in the literature to focus on one country and one educational institution (see e.g. McCabe et al. (1994) or Racko (2019)), our results might be sample-specific. Performing a study as ours in several countries and institutions at the same time would allow drawing more general conclusions. Additionally, our findings show changes in beliefs and values due to selection and socialization of the typical business student. We thus observe a combination of the effect of social interaction, learning and differential attrition. Future research could improve upon this by following individual students throughout their academic education. Research avenues might even open up by following individuals not only during their studies, but also during their careers, so as to evaluate the longitudinal impact of studying business versus other disciplines.
Additionally, our finding that not only the beliefs, but also the values of students are affected by studying might imply that values are less stable over time than previously assumed (Rokeach 1973;Schwartz 1992). However, we need to follow individual students over the course of their education to be able to make hard claims about this, which paves the way for future research trying to disentangle the variation in individuals from the variation of the composition of students. Our findings nonetheless show that it takes just one year for learning and socialization to affect the values of students. A natural question to pose here is: should we, as teachers, be concerned about how we affect the values of students and maybe even about changing them at all? We leave this open for further academic and ethical debate.
As a final comment, our results show that the beliefs and values of business students change over time but suggest that economics is not the main driver behind those changes. The key questions thus become: what determines those changes? Are other disciplines to blame? Or is it more generally the interaction of students with their peers? Further investigations are necessary before other suspects take the place that economics used to have. Tables 4, 5 , 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25 and 26.   In all models, we control for gender. Columns 1-5 show estimation output with beliefs as dependent variable and columns 6-8 for values as dependent variable. The main independent variable is a time dummy equal to 1 if the survey is administered at the end of the first year and 0 otherwise. The reference group are the students that responded to the survey at the beginning of the first year. Coefficients are reported in bold. Robust standard errors are in parentheses. For the models in column 1-6 and 8, the coefficients are estimated using an ordered probit model. For those in column 7, coefficients are estimated with a probit model. The table shows standard p-values as well as q-values (adjusted for multiple hypothesis testing and calculated using the Simes-method, STATA qqvalue command (Newson 2010)). Significance is indicated as follows: ***p < 0.01, **p < 0.05, *p < 0.1. Models are estimated for the pooled sample of the surveys administered at the beginning of the first year of study and one obtained at the end of that year   1 3

See Figs. 3, 4 and 5 and
What have we done?! The impact of economics on the beliefs and…  Table 7 T-tests of difference in means at the beginning and end of year by discipline For the beliefs/values and disciplines under consideration, this table shows a t-test of the difference in means between answers at the beginning and the end of the year. In the rows 'H1: mean(Begin) ≠ mean(End)', it is tested whether the mean answer of students at the beginning of the year is significantly different from the mean answer of the students at the end of the year. P-values are shown. The rows 'Test-statistic' shows the corresponding test-statistic (3)

3
What have we done?! The impact of economics on the beliefs and… For all disciplines and beliefs/values under consideration, this table shows the variance in responses at the beginning and at the end of the year. It also shows the F-statistic calculated as the ration between the two variance on which equality of variance test or based. In the row 'H1: var(begin) > var(end)', it is tested whether the variance at the end of the year is smaller than that at the beginning of the year, i.e. whether beliefs/values are becoming less dispersed over time. P-values are shown. In the row 'H1: var(begin) < var(end)' the opposite is tested

3
What have we done?! The impact of economics on the beliefs and… Table 9 Marginal effects of discipline on belief Firms Abuse their Size at the beginning and end of 1st year-calculated for a male student (based on (ordered) probit models in Table 2) This table shows marginal effects of disciplines for a male student per answer category (in columns) for the belief: Firms abuse their size. The scale for this question ranges from 1 (disagree) to 5 (agree). Columns show the (marginal) likelihood that a male student of a certain discipline chooses a certain answer. The reference group consists of male business students.  Table 10 Marginal effects of discipline on belief Exchange is Mutually Beneficial at the beginning and end of 1st year-calculated for a male student (based on (ordered) probit models in Table 2) This table shows marginal effects of disciplines for a male student per answer category (in columns) for the belief: Exchange is mutually beneficial. The scale for this question ranges from 1 (disagree) to 5 (agree). Columns show the (marginal) likelihood that a male student of a certain discipline chooses a certain answer. The reference group consists of male business students. Marginal effects are calculated for the beginning of the year sample and the end of the year sample separately What have we done?! The impact of economics on the beliefs and…

Table 11
Marginal effects of discipline on belief Lot in Life is Deserved at the beginning and end of 1st year-calculated for a male student (based on (ordered) probit models in Table 2 Table 13 Marginal effects of discipline on the belief Mobility at the beginning and end of 1st year-calculated for a male student (based on (ordered) probit models in Table 2) This table shows marginal effects of disciplines for a male student per answer category (in columns) for the belief of a students that they will fare worse (1), similar (2)  What have we done?! The impact of economics on the beliefs and…

Table 14
Marginal effects of discipline on value Equality at beginning and end of 1st year-calculated for a male student (based on (ordered) probit models in Table 2) This table shows marginal effects of disciplines for a male student per answer category (in columns) for the value: Equality. The scale for this question ranges from 1 (This is against my values) to 7 (This is a fundamental value to me). Columns show the (marginal) likelihood that a male student of a certain discipline chooses a certain answer.
The reference group consists of male business students. Marginal effects are calculated for the beginning of the year sample and the end of the year sample separately Equality  Table 15 Marginal effects of discipline on value inequality aversion at beginning and end of 1st year-calculated for a male student (based on (ordered) probit models in Table 2) This table shows marginal effects of disciplines for a male student per answer category (in column 1) for the value: inequality aversion. This variable is a dummy that is 1 for people that are inequality averse. The column shows the (marginal) likelihood that a male student of a certain discipline is inequality averse. The reference group consists of male business students. Marginal effects are calculated for the beginning of the year sample and the end of the year sample separately

Table 16
Marginal effects of discipline on value Wealth at beginning and end of 1 st year -calculated for a male student (based on (ordered) probit models in Table 2) This table shows marginal effects of disciplines for a male student per answer category (in columns) for the value: Wealth. The scale for this question ranges from 1 (This is against my values) to 7 (This is a fundamental value to me). Columns shows the (marginal) likelihood that a male student of a certain discipline chooses a certain answer. The reference group consists of male business students. Marginal effects are calculated for the beginning of the year sample and the end of the year sample separately Wealth   Table 3) This table shows marginal effects of the end-of-year dummy for a male student per answer category (in columns) for the belief: Exchange is mutually beneficial. The scale for this question ranges from 1 (disagree) to 5 (agree). Columns shows the marginal likelihood that a male student at the end of the first year chooses a certain answer. The reference group consists of male business students. Separate models are calculated for each discipline   year-calculated for a male student (based on (ordered) probit models in Table 3) This table shows marginal effects of the end-of-year dummy for a male student per answer category (in columns) for the belief of a students that they will fare worse (1), similar (2) or better (3)

Table 22
Marginal effect of end-of-year dummy on value Equality at beginning and end of 1st year-calculated for a male student (based on (ordered) probit models in Table 3) This table shows marginal effects of the end-of-year dummy for a male student per answer category (in columns) for the value: Equality. The scale for this question ranges from 1 (This is against my values) to 7 (This is a fundamental value to me). Columns shows the marginal likelihood that a male student at the end of the first year chooses a certain answer. The reference group consists of male business students. Separate models are calculated for each discipline Equality  Table 23 Marginal effect of end-of-year dummy value inequality aversion at beginning and end of 1st year-calculated for a male student (based on (ordered) probit models in Table 3) This table shows marginal effects of the end-of-year dummy for a male student per answer category (in columns) for the value: inequality aversion. This variable is a dummy that is 1 for people that are inequality averse. Columns shows the marginal likelihood that a male student at the end of the first year chooses a certain answer. The reference group consists of male business students. Separate models are calculated for each discipline

Table 24
Marginal effect of end-of-year dummy value Wealth at beginning and end of 1st year-calculated for a male student (based on (ordered) probit models in Table 3) This table shows marginal effects of the end-of-year dummy for a male student per answer category (in columns) for the value: Wealth. The scale for this question ranges from 1 (This is against my values) to 7 (This is a fundamental value to me). Columns shows the marginal likelihood that a male student at the end of the first year chooses a certain answer. The reference group consists of male business students. Separate models are calculated for each discipline Wealth  Table 25 Effect of disciplines on beliefs/values over time (OLS) The models are estimated using OLS and we control for gender. Columns 1-5 show estimation output with beliefs as dependent variable and columns 6-8 for values as dependent variable. Main independent variables are discipline dummies equal to 1 if respondent is a student of that discipline and 0 otherwise, an end-of-year dummy that is 1 at the end of the year and the interaction between these variables. The reference group consists of business students. Robust standard errors are in parentheses. Significance is indicated as follows: ***p < 0.01, **p < 0.05, *p < 0.1 Dependent variable (1) (3)

Table 26
Marginal effects per discipline at beginning and end of 1st year (based on OLS models with time-dummy interaction in Appendix Table 25) This table shows the marginal effect of studying a certain discipline on having a certain belief/value. The reference group consists of business students. Marginal effects are calculated for the begin of the year and the end of the year sample. The rows 'Sign. different?' shows whether the marginal effects at the start of the year are statistically different from the ones at the end of the year at the 10% level, i.e. whether beliefs/values changed significantly over time. This is based on the significance of the interaction terms in Table 25 Dependent variable (5) (1) (3)