Natural born economists?

Natural Born Economists? Giam Pietro Cipriani, Diego Lubian, Angelo Zago∗ Department of Economics University of Verona via dell’Universit`a 37129 Verona - Italy 23rd May 2007

Abstract We carried out a survey among a large group of undergraduate students of different disciplines to test whether the study of economics influences student’s views on profit maximization and the market mechanism. We find that there are significant differences between economics students and the others, suggesting both the presence of a selection bias against the market system in non economics students and a learning effect in economic students only when there is no direct conflict between profit maximization and the welfare of workers. Hence we argue that there is learning but no indoctrination.

Keywords: Economists, fairness, learning, selection. JEL Classification: A13, A20, B40, D63

The authors thank Nunzio Cappuccio and Raffaele Miniaci for helpful suggestions, Federica Barzi and Dolores Rizzotto for capable research assistance. We also thank our collagues at the University of Verona for letting us carry out the survey during their classes. Financial support from the University of Verona and MIUR (Cofin 2005) under contract #2005-132539 (Diego Lubian) is gratefully acknowledged. Corresponding author: Giam Pietro Cipriani, Dipartimento di Scienze Economiche, Universit` a di Verona, Viale dell’Universit` a, 4 Verona 37129, Italy. Email: [email protected], [email protected], [email protected]. ∗

1

Introduction

A number of papers have recently considered how the study of economics influences students’ views on the virtues of the market system and their acceptance of the standard “nonfairness assumption” employed by the traditional economic theory. The general result is that the economics teaching has an influence on those views and tends to make students more selfish. If true, this should be carefully considered by our profession since our teaching methods could have an impact on students’ actual behaviour in situations that require a balance between profit maximization and ethical aspects. Kahneman et al. (1986b) revisited the concept of fairness using household surveys of public opinions to identify the standards of fairness and their possible implications on market outcomes. All questions asked involved a firm making a decision on prices or wages that affects some transactors. They find that many actions which would be implied as a result of the standard profit maximization approach are in fact perceived as unfair exploitation of market power by the public. The authors point out how this result has clear implications in so far as it may influence a company’s reputation or even the legislation. Since then, many other papers focused their attention on the different behaviour of economists, in general economic students, with respect to other groups and they generally found substantial differences. For example, Gorman and Kehr (1992), using the same questions of Kahneman et al. (1986b) found that business executives have a different attitude toward fairness than the general public in that they are less inclined to judge profit maximizing behaviour as unfair. Also, Marwell and Ames (1981) found that economics students were more likely to be free riders. Other papers have studied specifically the issue of self-selection. In this literature, Carter and Irons (1991) use an ultimatum bargaining experiment to find that economists’ behaviour is closer to the game-theoretic solution (i.e., they propose to keep more and the other part is available to keep less) than non-economists. However, they also find that this result is essentially due to a self-selection mechanism, since freshman economists are more inclined to behave according to the rational self-interest model than non economists. They do not find evidence for the learning hypothesis since economics students were no more different from the others at the beginning of their studies than at the end. On the contrary, another paper that considers the same ultimatum game, Stanley and Tran (1978), finds that economics students are less motivated by self-interest than other students. However their evidence is based on a rather small sample. Again on the issue of learning (or indoctrination) versus selection, Frey et al.

1

(1993) begins by discussing Marwell and Ames (1981) result cited above which they find consistent both with the selection hypothesis (an “innate” willingness to apply the price system) and the indoctrination hypothesis (willingness to behave according to the orthodox economic theory after exposure to formal economic education). Their study tries to discriminate between the indoctrination and the selection hypothesis by distinguishing advanced students from beginners and by using, as a control, a sample of the general population randomly drawn from the telephone directory. They conclude against the indoctrination hypothesis and in favor of the selection hypothesis. On the contrary, Haucap and Just (2003) replicated the same experiment as Frey et al. (1993) on their own students and found just the opposite: the indoctrination effect is important and significant. Frank et al. (1993) carry out a survey on charitable giving by mailing questionnaires to college professors from different disciplines asking them to report their charity giving. They find that economists were among the least generous. Also in a prisoner’s dilemma game situation, they found that economists appear to behave less cooperatively than non-economists. By comparing upperclassmen from underclassmen, they found that there was a trend towards more cooperative behaviour toward graduation but not among economics students, which is not inconsistent with the learning hypothesis. Frey and Meyer (2003, 2005), differently from Frank et al. (1993) carry out an analysis of actual charity giving and find no evidence for the indoctrination effect, while there is a selection in that business economics students appear as more selfish than the average student. Although their paper has the advantage of addressing this issue in a natural setting, the sum of money involved is fixed (the only decision is whether to contribute or not) and rather small and the control variables include only gender, nationality and age. Some of these studies did not test students at the very beginning of their courses but a couple of months later, when already some economics had been studied (although some, like Carter and Irons (1991), point out that this was macroeconomics and not microeoconomics). In this study we investigate whether the study of economics affects how students perceive fairness and efficiency of different allocation systems, together with the conflict between profit maximization and other social considerations, aiming at directly disentangling the selection from the learning/indoctrination effects. In addition, contrary to most of the above reported studies, we include many other variables (demographics, type of high school attended, high school performance, parents’ education and jobs) which allows us to control for other factors in explaining results. Finally, while some of these papers use more heterogeneous groups - for example, Frey et al. (1993) use as a control sample a drawn of households who obviously dif2

fer in a variety of ways from undergraduate students - we use other students at the same university attending non-economics courses in order to test for the selection hypothesis. Our work has been inspired by a recent paper by Rubinstein (2006). He criticizes the existing literature which, in his opinion, has not provided decisive evidence in favor or against the indoctrination effect. Then he presents his “main experiment” which we replicate in our study together with other questions explained in the next Section. Differently from Rubinstein, we use a larger sample (about 1.5 thousand students) and we present the questionnaire not only to students from different departments but also to students of different years in order to test for learning or indoctrination. Also, as said, we control for expected income, demographic factors and previous education. Finally, among the economics students we distinguish between those whose degree involves no microeconomics at all from those whose degree involves only a short microeconomics course and from those who study microeconomics at a more intermediate level. On the one hand, we find a significant selection effect, and therefore in favour of a “natural-born economist”. On the other hand however, we also find evidence of a learning effect, since the more the teaching of Microeconomics to which economics students are exposed, the higher they tend to choose market outcome responses. However, when they are confronted with issues presenting more socially and ethically demanding choices, the teaching of economics does not seem to have significant effects. In other words, more trained economists may appreciate more the efficiency virtues of different market mechanisms, but they still choose also according to their social and ethical effects. Hence we will talk of a “learning” effect and not of an “indoctrination” effect, where by the term “indoctrination” we mean a type of teaching which is meant to be taken literally and uncritically by students. The paper proceeds by discussing the data used and the survey design in Section II. Section III presents the analysis and results. Section IV draws conclusion.

2

The Experiment and Survey Design

We aim at disentangling the effects of selection into economics and/or learning about economics and to assess the balance between efficiency and equity considerations after exposure to economics. To do so, in the first place, we need to compare first year students (thus at the beginning of their academic career) from different Departments, to assess whether the students who decide to enroll in Economics are different from those attending other Departments. Secondly, to investigate the relevance of a learning effect among economics students, we need a comparison 3

between students in their third (and last) year of study and students just enrolled to test whether significant differences in responses can be attributed to greater exposure to economics. To this end, we conducted a survey of undergraduate students enrolled at the University of Verona. In October 2006, thus at the beginning of the academic year 2006-2007, we interviewed students from 5 different Departments: Economics (coded as econ), Law (law), Natural Sciences (Mathematics and Computer Science - math), Foreign Languages (lang), and Tourism (tour). All students were asked to fill a questionnaire at the beginning of a lecture during the first week of term, which ensures that first year students had not yet been exposed to any economics training at university level and eliminates a possible teacher effect. This timing of the survey allows to test for the differences between economics and other students, i.e., for the presence of a selection effect. To investigate the possible learning (or indoctrination) effect, i.e., the effect on students’ response determined by their exposure to economics teaching, we take advantage of the three different curricula that economics students can pursue at the University of Verona exploiting the variability in both the quantity and the intensity of exposure to Microeconomics. In fact, after a common first year, Economics students can choose amongst three curricula: Management & Marketing (M&M), Accounting (Acc), Economics & Business (E&B). All students attend a Macroeconomics course (10 ects, the European common unit of measurement for courses length, corresponding to about 90 hours of classes with the teacher or teaching assistant) during the first year, without any distinction amongst the curricula so that, in practice, students from the three curricula mix in the Macro class. In the M&M curriculum students are taught no Microeconomics at all, and they receive probably the least formal training, being exposed to courses that are mostly of a descriptive and, to some extent, “philosophical” nature. In other words, this curriculum is more similar to a Communication than to a Marketing curriculum found in an American or English university. Accounting students, on the other hand, receive a general training which is a bit more formal, spending relatively more time and attention on different accounting practices. They are also taught a short Microeconomics course (5 ects, or about 45 hours) in their second year. Following a mainstream textbook1 , they are introduced to consumer’s behavior, firm’s behavior, aggregate demand and supply, market equilibrium, monopoly, oligopoly, game theory, and pure exchange. Lastly, the E&B curriculum students attend a full course (10 ects, or 90 hours) of Microeconomics in the second year, thus receiving the maximum training in Micro1

The adopted textbook is the Italian translation of Varian (2002), Intermediate Microeconomics.

4

economics that an undergraduate economics student can get at this university2 . Besides the topics that Accounting students cover, they are introduced to uncertainty, inter-temporal choices, exchange with production, price discrimination, asymmetric information, and auctions. All students in the survey were asked four different questions concerning the functioning and efficiency of market mechanisms, where each question presents to a different extent the need to balance efficiency and fairness, either in terms of reference transaction or with more explicit social implications such as workers layoffs. The survey was anonymous, and all students were instructed that there were no right or wrong answers. For comparability reasons, the four questions are taken from previously published articles in the field ((Kahneman et al., 1986a,b; Rubinstein, 2006). The first question concerns a judgment of the fairness of different resource allocation mechanisms (auction, lottery, queue). This is perhaps the most “neutral” question in terms of social implications, and it requires the acknowledgement that waiting on line does not exhaust all gains from trade. In the second question the conflict between fairness and the functioning of a market is more pronounced. This question requires a judgement about the fairness of a price increase following an excess demand without any increase in the cost incurred by a firm in the presence of a reference transaction. The third question is the most demanding, and it presents a dilemma between profit maximization and worker layoffs, asking the student what would their behaviour be as a vice president (VP) of a company when facing an explicit trade-off between the number of worker to be laid off and the profit earned by the firm. Finally, the fourth question is about the students’ view of the behavior of a real VP who faces exactly the previous question. This question is interesting but for a different reason: it allows us to investigate how students differ in their beliefs about the role of a firm’s manager and how often he would choose the profit maximization solution disregarding the social consequences. Table 2 reports response patterns and associated observed frequencies for each Department group. In general, students’ responses to our questions are the outcome of a complex process where family, peers, learning, and ability to evaluate the social implications of private actions interact. The decision to pursue the study of economics or other subjects is also determined, besides unobservable factors, by individual taste for the subject, family background, the educational track attended by the students at high school. To account for these factors, we use several control variables such as educational background,3 ability (as measured by the high school graduation marks), 2 3

See footnote 1. High schools in Italy are essentially of 3 types: professional schools, which should directly prepare for blue

5

family background (parental education and sector of activity). Descriptive statistics on these control variables are reported in Table 3.

2.1

Questions and Preliminary Evidence

Following Kahneman et al. (1986a), our first question (coded as auction) is Question 1. A football team normally sells some tickets on the day of their games. Recently, interest in the next game has increased greatly, and tickets are in great demand. The team owners can distribute the tickets in one of three ways. (1) By auction: the tickets are sold to the highest bidders. (2) By lottery: the tickets are sold to the people whose name are drawn. (3) By queue: the tickets are sold on a first-come firstserved basis. Rank these three in terms of which you feel is the most fair and which is the least fair - the auction, the lottery, and the queue. Table 4 presents the statistics describing the responses to this question. It is clear that economics students judge the auction as the fairest mechanism more often (about 21%) than other students, with the closest being the law students (about 19%) and the least likely those in the humanities (lang, about 9%, and tour, with about 12%) and math, with about 12%. There is also evidence of a possibile composition effect due to the presence of three curricula in economics. This curricula effect is strong and striking among third year students, where E&B3 students (those with the longest exposition to microeconomics) are more favorable to the auction mechanism by almost 16 percentage points with respect to M&M3 students (those with no microeconomics treatment). Thus in the first question we find both the selection and learning effects to be significant4 . We take the second question (coded shovel) from Kahneman et al. (1986b) Question 2. A hardware store has been selling snow shovels for 15 Euro. The morning after a large snowstorm, the store raises the price to 20 Euro. Please rate this action as: Completely Fair

Acceptable

Unfair

Very Unfair

This simple question allows to deal with both the profit-seeking and the fairness aspects. Excess demand in this market would require a price increase to clear collar type jobs, emphasizing also manual skills; technical schools, aimed at both blue and white collar type jobs but with a more technical content; and “liceo”, a more general school which should better prepare for university. High school graduation mark ranges from 60 (lowest) to 100 (highest). The graduation mark depends both on students’ performance during their career at high school and on a nation-wide comprehensive examination produced at a centralized level (Ministry of Education) identical for all students but specific to the school type. 4 As a comparison, in Kahneman et al. (1986a), only 4% of respondents in a telephone survey of 191 households rated auction as the fairest mechanism. This confirms our selection effect.

6

the market exploiting the profit opportunities of the hardware store acting in its self-interest. The fairness concern is related to the dislike of customers to price increases which are not justified in terms of higher production (or sale) costs, especially when customers have developed some entitlement to an experienced reference transaction (as explained in Kahneman et al. (1986b)). In this question thus the dilemma between fairness and efficiency is more pronounced. Our prior is that economics students in general should tend to be less concerned with fairness issues, or equivalently more concerned with efficiency considerations. In addition, we would expect economics students with more training in Microeconomics to see more the equilibrium adjustment effect, i.e., following a demand shock it is natural to see an increase in price to reach the market equilibrium, than the fairness effect, i.e., the fact that a price increase which does not come from a cost increase is considered unfair. The four answers to this second question are grouped to generate the proportion of students rating the action as fair (Completely fair, Acceptable) or unfair (Unfair, Very unfair). Table 5 contain these proportions distinguishing between students enrolled in the Department of Economics and students enrolled in other Departments and also between first-year and third-year economics students. More than 23 percentage points separates first-year econ students to other first-year students: this difference is large and significant as indicated by Pearson’s χ2 test suggesting substantial self-selection into economics and business. Thus, students more prone to the kind of reasoning where it appears to be justifiable to exploit all available profit opportunities regardless of some fairness concern do enroll in economics. Third-year students do not exhibit a significant increase in the proportion considering it unfair to raise prices without cost increases. However, there is a differentiated shift in the attitude towards this issue within economics students: about 70% of E&B3 students believe that the price increase is fair while about 55% of E&B1 students do so. On the other hand, about 60% of M&M1 students believe that the price increase is fair while less than 40% of M&M3 students do so5 . Third year M&M students - those with no Microeconomics teaching - therefore seem to have developed a greater concern for customer’s attitude towards fairness when making 5

As a comparison, in Kahneman et al. (1986b), 82% of respondents in telephone surveys of 107 households

consider it unfair. This we believe is additional evidence of a selection effect.

7

their own fairness judgement and to comply with the possible manager’s reluctancy to apply demand-based pricing for fear of (possible) customer dissatisfaction. In other words, in this second question the selection effect is even stronger than in the first question, while the learning effect appears working as well, so much that, in the absence of Microeconomics teaching, M&M students seem to be concerned much more with the fairness than the efficiency consideration. The third question (coded as profmax) is exactly as in Rubinstein (2006), Question 3. Assume that you are a vice president of ILJK company. The company provides extermination services and employs a certain number of permanent administrative workers and 196 nonpermanent workers who are sent out on extermination jobs. The company was founded five years ago and is owned by three families. The work requires only a low level of skill, with each worker requiring only one week of training. All the company’s employees have been with the company for three to five years. The company pays its workers more than minimum wage. A worker’s salary includes payment for overtime, which varies from 1,200 to 1,500 Euro per month (suppose that the minimum wage is about 1,000 Euro per month). The company makes sure to provide its employees with all the benefits required by law. Until recently, the company was making large profits. As a result of the continuing recession, there has been a significant drop in profits, although the company is still in the black. You will be attending a meeting of the management in which a decision will be made regarding the layoff of some of the workers. ILJK’s finance department has prepared scenarios of annual profits shown in the following. Number of Workers Who Will Continue to Be Employed 0 (All the workers to be laid off) 50 (146 workers to be laid off) 65 (131 workers to be laid off) 100 (96 workers to be laid off) 144 (52 workers to be laid off) 170 (26 workers to be laid off) 196 (0 workers to be laid off)

Expected Annual Profit (Millions of Euro) Loss of 8 Profit of 1 Profit of 1.5 Profit of 2 Profit of 1.6 Profit of 1 Profit of 0.4

Complete the following: I recommend continuing to employ . . . of the 196 workers presently employed by the company. This question (profmax), as clearly explained by Rubinstein (2006), is purposely designed as a dilemma. Students are asked to consider themselves as the executive who had to choose how many workers to dismiss, with a given number of possible answers showing both the number of workers to be kept (either 0, 50, 65, 100, 144, 170 or the whole 196) and the effects on total profits (with the maximum of 2 million euro for the choice of 100 workers kept). It is indeed clear that on one side the manager can achieve some profits, but against the information on how 8

many dismissible workers to retain. In addition, notice that profit maximization is more explicitly associated with ethical or moral issues, since one can think about the personal and family implications of a dismissal. In other words, with this question the respondents’ sympathy feelings toward the fate of dismissed workers are more explicitly called for and this is the least impersonal question of the three. A reasonable prior for this question, which would be consistent with Rubinstein’s findings, is that economics students would tend to choose the profit maximizing answer more often than other students, since these latter would presumably weight more the “social implications” of their choice. In addition, we expect that the more the training in Microeconomics the more frequent the choice of the profit maximizing option, irrespective of the implications for the workers. Table 6 reports the frequency distribution of answers to the profmax question. econ students seem to be more inclined to do the maximizing profit choice (about 50% of them), even though law students chose it even more often (55%). Other students indeed choose it less frequently: 41% for tour students, 35% for lang, and 25% for math. In addition, the learning effect appears to be negative, in the sense that economics students at the beginning of their third and last year choose the profit maximizing option less often (43%) than freshmen econ students. Therefore, while we find evidence of selection for econ (and law) students, economics teaching does not seem to induce students to lean more towards the profit maximizing solution when there is a clear trade off with “ethical” issues (the welfare of workers). Hence we claim that learning does not imply indoctrination (as defined in the Introduction). Our last question, also taken from Rubinstein (2006), explores the students’ view on the behavior of a real VP in the hypothetical settings of the previous question. Question 4. What do you think would be the choice of a real vice president in Question 3? I think that he would recommend continuing to employ . . . of the 196 workers in the company. Thus, students are asked to consider themselves as the executive who had to choose how many workers to dismiss, with a given number of possible answers showing the number of workers to be kept (either 0, 50, 65, 100, 144, 170 or the whole 196) and the effects on total profits (with the maximum of 2 million euro for the choice of 100 workers kept). Proportions are reported in Tables 7 and 8. About 70% of law and tour students think that the real VP would behave as a profit

9

maximizer and less than 60% of math and lang students do so. Surprisingly, first year econ students have the lowest proportion, about 55%, which is also similar for third year econ students. A composition effect is at work among econ students: third year E&B students believe that a real VP would fire more than what thought by first year students. How deep is the matching (identification) of students with a real VP? Table 8 provides some insight on this issue. Responses by math, lang, and tour students have the lowest match with their own idea of what of real VP would choose. They think that a real VP would behave very differently and, in particular, they think that a real VP would fire more than them, probably according to their view that “greedy” managers put more weight on profits.

3

Selection, Learning and Indoctrination

3.1

Is There Selection into Economics?

The first issue addressed in the empirical analysis concerns the existence of a selection effect into economics, whereby students who lean more towards profit maximization or who give more importance to the functioning of the market mechanism self-select into the study of economics. Under the selection hypothesis, students’ stance on profit maximizing behavior should be a predictor of the decision to enroll in economics or in the other Departments. To this end, we consider the sub-sample of first-year students enrolled in the different Departments who, being asked to fill the questionnaire in the first week of classes, should know nothing about economics. We model the dichotomous dependent variable “enrollment in economics” as a function of gender, high school type, graduation mark, family background both in terms of parents education and parents jobs, and as a function of the answers in the four questions auction, shovel, profmax and profmaxVP. Regression results are reported in Table 9. The coefficients on auction, shovel and profmax have the expected sign and are highly significant (1% level) providing strong and neat evidence of selection into economics based on students’ sentiment on efficiency issues. In other words, students that appreciate more the efficiency aspects of markets’ functioning and of firms’ choices, are significantly more likely to study economics. Looking at the marginal effects, students that for instance chose the efficiency aspects of the first 10

two questions, auction and shovel, were almost 31% more likely to enroll into economics. Moreover, the estimated coefficient on profmaxVP has negative sign and is significantly different from zero (at a 10% significance level) indicating that, as we have seen in Table 8, students of economics are less likely (marginal effect of about 5%) to believe that a real VP is a brute profit maximizer who do not consider workers layoffs. High school type and family background are also of some importance in the decision to enroll in economics. In particular, students coming from a technical high school are significantly (10% significance level, with a marginal effect of 12%) more likely to enroll into economics. It is also interesting to notice that the probability of choosing economics increases (by 10%) when the father belongs to the upper middle class - either being a senior manager, a member of the profession or an entrepreneur - suggesting either a more business prone attitude in these families or some inertia in the children’ choice trying to follow suit the father’s career path.

3.2

Learning vs. Indoctrination: Is There Sample Selection?

Having established the relevance of the selection into economics, in this section we restrict our analysis to economics students to study the central issue of learning vs. indoctrination. Any discussion of possible treatment effects due to the exposure to economics must realize that students are not randomly assigned to treatment but rather they self select into economics based on their attitude towards the market mechanisms and other characteristics. To take into account the sample selection issue we estimate a probit model with sample selection for each of the four questions. We assume that there exists a latent variable yj∗ , where j = auction, shovel, profmax, profmaxVP, which reflects the economics students’ stance towards fairness, profit maximization and the market mechanism and a vector of demographic and social factors (xj ) which all affect the student’s answer as in ∗ yij = x′ij β j + uij ,

(1)

where β j is the vector of unknown parameters in equation j. Given that only students’ answers are observable, our observation is ( ∗ >0 1 if yij yij = 0 otherwise. 11

However, we only observe yij for the i-th student if enrolled in economics. Let us assume that the propensity to enroll in economics, say e∗i , comes from the underlying relationship e∗i = z ′i δ j + vij ,

(2)

where δ j is the vector of unknown parameters in the selection equation. z includes the variables in x and additional factors that determine selection so that we observe yij only when ei = 1, where ei =

(

1 if e∗i > 0 0 otherwise.

Finally, we assume that the error terms uij and vij are jointly normally distributed with correlation coefficient ρj . This is a probit model with sample selection which can be estimated by Maximum Likelihood. In our specification, the same vector of regressors xij is present in each equation reflecting information on several observed individual characteristics of the students, such as gender and family background (parental education, work status and occupational sector of parents, number of siblings) which should influence their attitude towards the issues presented in the questionnaire. In the selection equation (2), besides all regressors included in the treatment equation (1), we use high school background (high school graduation mark, high school type) as the factor that determine selection. The reference student in the estimated model is a first year student enrolled in the M&M curriculum, coming from a professional high school with the lowest graduation mark, blue collar father and mother both with junior high school certificate, without siblings. Estimation results are reported in Table 10 for the probit model on the treatment equation and in Table 11 for the underlying selection model. The estimated correlation coefficient between the treatment and the selection equation is never statistically significant suggesting independence between the treatment and the selection. Then, efficiency gains could be achieved by estimating the treatment equations jointly as we do in the next section to which we refer for a discussion of estimation results.

3.3

Learning vs. Indoctrination: A Multivariate Analysis

Given that the correlation coefficient between the error term of the treatment equation and the selection equation is never significantly different from zero (Table 10), 12

and given that students’ responses can be viewed as the joint observed outcome determined by treatment (exposure to economics), we consider the four equations multivariate model ∗ yij = x′ij β j + uij , ∗ is a latent dependent where j = auction, shovel, profmax, profmaxVP, and yij

variable which reflects the students’ stance towards fairness, profit maximization and the market mechanism. Since our observation is ( ∗ >0 1 if yij yij = 0 otherwise our specification gives rise to a multivariate probit model, analogous to the seemingly unrelated regression model but for the presence of a binary dependent variable. The same vector of regressors x′ij is present in each equation. We assume that the error term uj , which embodies unobservable factors affecting students’ answers, is distributed as multivariate normal with zero mean and covariance matrix Σ where, for identifiability reasons, we assume unity on the main diagonal and non-zero correlation coefficients ρrs = ρsr as off-diagonal elements6 . This specification with unrestricted non-zero correlations in Σ allows for correlation amongst the unobservable factors affecting the students’ responses to the four questions posed under the assumption that the same unobservable factors influence the answers. Of course, all equations could be estimated as single equation Probit models but the coefficient would be estimated inefficiently because of the neglected correlation among the error terms. The multivariate Probit model is estimated by Simulated Maximum Likelihood using the Geweke-Hajivassiliou-Keane smooth recursive simulator to evaluate the multivariate normal probability (B¨orsch-Supan and Hajivassiliou, 1993; Keane, 1994; Hajivassiliou and Ruud, 1994)7 . Estimation results are provided in Table 12. Looking at the first question, as in Table 10 the intercept is negative and significant, which matches with the large number of students choosing lottery or queue, and there are no significant effects attributable to family background. Economics freshman do not differ significantly between themselves so that there is no selection 6

The univariate probit model is nested as a special case when the correlation coefficients are zero and the

answer to each of the four questions is given independently of the other answers. 7 We use Stata’s mvprobit procedure developed by Cappellari and Jenkins (2003) who found the GHK simulator to behave well both in a bivariate example with real data and with simulated data as long as the number of draws is sufficiently large. We use 500 draws in the estimation.

13

in the three curricula at the enrollment stage. Third year E&B and Accounting students choose auction more often than M&M students, the respective coefficients are highly significant not only with respect to the reference M&M freshman but also with reference to their own freshmen (the Wald tests are equal to 8.99 and 4.13 for E&B and Accounting students respectively) suggesting the presence of a learning effect. Overall, given that we do not find significant differences between E&B and Accounting students (Wald test equal to 1.92), these results suggest that when it comes to apply the normative suggestions of Microeconomics in a context with no social consequences, treatment matters while its amount does not. As for the second question (shovel), our results suggest also that the degree by which the action is assessed as unfair by third year students depends upon the intensity of the treatment: M&M students, with no treatment at all behave as students from others departments, Accounting students with a medium amount of Microeconomics classes (Acc3) are significantly different from Acc1 students (Wald test=4.38 with a p-value of 3.6%), and E&B3 students who receive a full 10 ects course in Microeconomics differ from E&B freshmen (Wald test equal to 2.74 with a p-value of 9%) and from Acc3 students (Wald test = 2.93 with a p-value of 8%). Again, we find evidence in favor of an important learning effect of economics. This is one of the two questions where an important gender effect can be ascertained (the other one is profmaxVP). Females consider unfair the price increase more often than male students and they seem more reluctant to apply demand-based pricing reasoning. Our evidence suggests the possible presence of differences in opinion on fairness issues due to gender in the presence of a reference transaction. In the light of the absence of evidence of a gender effect in the remaining questions, in particular the one posing the profit maximization vs. layoffs dilemma, we would not claim that our evidence is in favor of some “women are nicer” stereotypical effect. In general, evidence concerning a gender effect is mixed and does not support naive conclusions.8 8

In the context of a punishment game, Eckel and Grossman (1996) found that women tend to punish an

unfair partner more often than males in the case of a high-stake game and less often than men in the opposite case. They argue that while males seem to follow moral principles strictly, females exhibit a kind of situational morality. In an experimental study on a dictator game, Andreoni and Vesterlund (2001) find that when giving is more costly, women are more unselfish than men who, on the contrary, give more as the price of giving decreases. Thus, fairness judgement depends on the price level and not just on sex.

14

We now turn to the most challenging (for the respondents) dilemma: profit maximization vs. workers layoffs. Evidence in favor of some learning/indoctrination effect would be suggested by third year economics students more maximizers than first year economics students. Regression results provides no evidence in the favor of any learning effect among economics students. Surprisingly, third year M&M students maximize less than first year students, since the coefficient is significant at a 1.6% level, and significantly less than third year economics students from the two alternative curricula (the Wald test of equality between M&M3 students and E&B3 students or Acc3 students are equal to 4.16 (p-value: 0.04) and 3.76 (p-value: 0.052), respectively). This seems to indicate a “reverse” learning/indoctrination effect, and we conjecture that this opposite response may be related to the training that M&M3 students receive: less formal, more symphatetic to customers with concepts such as customer loyalty, brand image, corporate social responsability, ethical issues and the like, that can render a simple profit maximization as a rather dry and partial objective to pursue. Thus, when faced with a socially relevant dilemma, treatment about economics does not matter: Economics students in their final year are as concerned of the social implications of their actions as students in their first year are. In a sense, this result is encouraging because it implies that students, when needed, are able to implement the normative content of microeconomics in a broad sense taking into account the social impact of their choices. Therefore we do not find evidence in favor of indoctrination. The last question concerns the students’ view of the actual behavior of a real Vice-President when facing the profit/layoffs dilemma. Economics students, both in the first year and the third year, do not differ significantly, regardless of the curricula chosen. Moreover, a different view on the VP’s decision depending on the sex is also present, given that women think that the VP will maximize profits significantly more than men. Interestingly, there is also a significant effect, at the the 5 and 10% level, of the father professional condition: the probability of predicting a profit maximizing behavior decreases when the father is either an entrepreneur, a senior manager, a professional consultant or a craftsman, respectively. One interpretation of this result is that students with a direct, first-hand experience (as shared in the family) of the behavior of a person who might indeed face the profits/layoffs dilemma 15

of this question, might tend to be less inclined to have the stereotypical perception of the VP as a “head cutter”. The estimated correlations among the error terms of the four equations are reported at the bottom of Table 12. Several of them are significant at a 1% level and the null hypothesis of zero correlations of the six correlations can be rejected by a LR test. This implies that single equation probit models would have produced inefficient estimates of the standard errors of the coefficients. As expected, all significant correlations are positive indicating that, on average, the unobservable factors determining a profit maximizing choice are also affecting the responses on the functioning of market mechanisms in the same direction. The strongest correlations are between shovel and auction, a question looking into the compliance to a market mechanism with little, if any, direct social impact, and between profmax and profmaxVP, suggesting that profit maximizers tend to think that a real VP would be a profit maximizer as well. To conclude, notice that our results could be affected by the high drop out rate observed in the Italian university system. For instance, in our sample in the third year we lose about 1/3 of the first-year students. As long as students drop out randomly our results are still valid, but what if the drop out is determined by some mismatch between the expectations about the program at enrollment and the progress in the actual program of study so that only the more lean towards profit maximization (or the market mechanisms) would remain while the more concerned about fairness issues would drop out. Given the distribution of answers at enrollment, if the profit maximizers only remain in the program and the treatment is effective enhancing their initial attitude we would expect a stronger response by third-year students in the profmax question, which in fact we do not observe. Then, we would argue that our estimation results represent something like an upper bound to the treatment effect. On the other hand, since we do actually observe a more favorable stance for the market mechanisms, as testified by the statistically significant treatment effect in the auction and shovel questions when little social/ethical considerations are involved, we would argue in favor of a learning but no indoctrination effect, as defined in the Introduction.

16

4

Conclusions

We investigate the effects of the study of economics into how students perceive fairness and efficiency of the market mechanism, and into the conflict between profit maximization and social considerations. Directly aiming at disentangling the selection from the learning/indoctrination effects, we control for the previous schools type and for other economic and social factors in explaining results. Comparing economics from other students at the same university, we find evidence of a clear selection effect: from the outset economics students differ from others in their tendency to maximize profits and in their view of the market mechanism. These differences should be taken into account by the literature on experimental economics, which often makes use of economics students for implementing economic experiments. Although recent studies with fields experiments show that representativeness of the environment, rather than that of the sampled population, is the most crucial variable in determining the generalizability of results for a large class of experimental laboratory games (see List (2006)), our results seem to indicate that economics students may be not representative of the population of students. We also find evidence of a learning effect: as found by Rubinstein (2006) it appears that orthodox microeconomics teaching encourage students to lean toward profit maximization and the market mechanism. This effect is mostly evident in the first question, dealing with market mechanisms without much fairness or social considerations: while among fresher economics students there are not significant differences (but still a significant difference with non economic students), significantly more third year economics students with more Microeconomics teaching choose the auction outcome compared to students with less training in Microeconomics. However, we would like to emphasize that this effect is not present when students are asked to choose among options with less neutral social consequences. Indeed, in the second and third questions dealing, respectively, with the fairness of a price increase required to achieve the market equilibrium and with the profit maximizing choice and the trade-off with workers layoffs, there are no significant differences between the choices of first and third year economics students when these latter are exposed to the teaching of microeconomics. In other words, the learning effect seems to be working only in those less ethically and morally demanding questions

17

or situations. Hence, we argue that there is learning but not indoctrination, in the sense that our formal microeconomics teaching is not taken literally and uncritically by students and applied directly to a (representation of a) real world situation. In addition, students with little or none microeconomics training, like our Management & Marketing students, who however follow courses which perhaps stimulate more comprehensive thinking, in the more ethically demanding question lean sensibly less towards efficiency or profit maximizing choices in their third year with respect to their first year. These differences could be the due to the way we teach economics. While our Management & Marketing students are presented with less analytical but more comprehensive firms problems, such as corporate social responsability, ethical issues, customer loyalty, etc. and hence may develop a more socially responsible view, Economics & Business students follow more microeconomics, with its emphasis on the mathematical reasoning and exercises, therefore remaining more inclined towards the standard profit maximization choices they had selected in the first year. To conclude, we believe that the results presented in this study complement and are less worrying than those in Rubinstein (2006). In a sense, they are encouraging. Differently from Rubinstein (2006), given that we observe both first year and third year students, we are able to control for selection into economics and to identify the learning/indoctrination effect, separating it neatly from the selection effect. Moreover, according to our suggested interpretation of the different (and increasing) social consequences the questions presented in the questionnaire, we are able to distinguish between learning and indoctrination. Selection coupled with learning without indoctrination is the conclusion we draw from our empirical analysis. Therefore, our findings do not necessarily indicate that the way we teach economics with its emphasis on equilibrium and efficiency might not imply the unpleasant consequences outlined by Rubinstein (2006), i.e. the creation of a selfish economic man. Clearly, this does not mean that the way we currently teach economics is the best and it cannot be improved by using more articulate models like, for instance, those commonly taught in advanced industrial organization courses which could better represent real world situations.

18

Table 1: Definition of variables Econ1 First-year Economics student Econ3 Third-year Economics student Law First-year Law student Lang First-year Foreign Languages student Tour First-year Tourism student Math First-year Math & Computer Science student M&M1, M&M3 First- and third-year student in the Management & Marketing curriculum (Economics) Acc1, Acc3 First- and third-year student in the Accounting curriculum (Economics) E&B1, E&B3 First- and third-year student in the Economics & Business curriculum (Economics) Gender 1 if female, 0 if male Profess HS 1 if the student attended a Professional High School, 0 otherwise Tech. HS 1 if the student attended a Technical High School (vocational), 0 otherwise Liceo HS 1 if the student attended a Liceo High School (Ancient Latin, Math, Philosophy, Physics), 0 otherwise Marks60-69 if High School graduation mark is between 60 (lowest mark) and 69, 0 otherwise Marks70-79 if High School graduation mark is between 70 and 79, 0 otherwise Marks80-89 if High School graduation mark is between 80 and 89, 0 otherwise Marks90-99 if High School graduation mark is between 90 and 99, 0 otherwise Marks100 if High School graduation mark is 100 (highest mark), 0 otherwise Family background Primary school 1 if highest degree is primary school (5 years of education), 0 otherwise Junior High School 1 if highest degree is junior high school (8 years of education), 0 otherwise High School 1 if highest degree is high school (13 years of education), 0 otherwise College Degree 1 if highest degree is College school (17 years of education), 0 otherwise Blue collar 1 if blue-collar worker, 0 otherwise White collar 1 if office worker or teacher, 0 otherwise UpperMiddleClass 1 if senior manager, member of the profession or entrepreneur, 0 otherwise MiddleClass 1 if craftsman, 0 otherwise Unempl 1 if unemployed, 0 otherwise Mother not working 1 if mother housekeeper or unemployed, 0 otherwise Siblings 1 if siblings, 0 otherwise

19

Shovel Auction Pmax PmaxVP

Table 2: Observed response pattern frequencies

0 0 0 0 1 0 0 0 1 1 1 0 1 1 1 1

0 0 0 0 1 0 0 0 1 1 1 0 1 1 1 1

0 0 0 1 0 0 1 1 0 0 1 1 1 0 1 1

0 0 0 1 0 0 1 1 0 0 1 1 1 0 1 1

0 0 1 0 0 1 0 1 0 1 0 1 0 1 1 1

0 0 1 0 0 1 0 1 0 1 0 1 0 1 1 1

0 1 0 0 0 1 1 0 1 0 0 1 1 1 0 1

Econ1 10.65 10.82 6.82 2.00 10.32 10.82 1.16 1.16 8.82 8.82 3.33 2.50 3.33 12.15 1.66 5.66 100.00

Econ3 11.78 13.35 4.71 2.09 8.38 8.12 2.88 1.83 7.33 6.02 5.76 1.83 5.76 10.99 2.36 6.81 100.00

Law 7.89 17.11 10.53 0.66 5.26 22.37 1.97 0.00 11.84 2.63 0.00 3.29 0.66 5.92 2.63 7.24 100.00

Tour 16.40 24.80 9.40 2.40 5.80 11.20 1.80 0.60 7.40 5.80 1.40 0.80 1.00 9.00 0.80 1.40 100.00

Lang 15.69 33.33 3.92 1.96 1.96 17.65 3.92 0.00 7.84 2.94 0.00 1.96 0.00 7.84 0.98 0.00 100.00

Math 16.58 29.41 6.42 1.07 11.23 6.95 2.67 0.53 10.16 3.74 1.07 0.53 3.21 4.28 2.14 0.00 100.00

Total 12.99 18.45 7.17 1.92 8.00 11.28 2.03 0.94 8.47 6.19 2.65 1.77 2.81 9.62 1.66 4.05 100.00

0 1 0 0 0 1 1 0 1 0 0 1 1 1 0 1

M&M1 11.29 6.45 4.84 0.81 12.10 10.48 0.81 1.61 12.90 11.29 4.84 2.42 3.23 11.29 3.23 2.42 100.00

Acc1 9.86 12.33 7.95 1.64 9.32 10.96 1.64 1.37 8.49 6.30 2.74 3.01 4.11 12.33 0.82 7.12 100.00

E&B1 12.50 10.71 5.36 4.46 11.61 10.71 0.00 0.00 5.36 14.29 3.57 0.89 0.89 12.50 2.68 4.46 100.00

M&M3 17.86 19.64 6.25 2.68 10.71 8.04 3.57 1.79 5.36 4.46 3.57 1.79 4.46 6.25 1.79 1.79 100.00

Acc3 9.84 13.47 3.63 1.55 7.77 8.81 2.59 2.59 8.29 5.70 4.66 2.07 6.74 12.44 3.11 6.74 100.00

E&B3 7.79 3.90 5.19 2.60 6.49 6.49 2.60 0.00 7.79 9.09 11.69 1.30 5.19 14.29 1.30 14.29 100.00

Total 11.09 11.80 6.00 2.03 9.56 9.77 1.83 1.42 8.24 7.73 4.27 2.24 4.27 11.70 1.93 6.10 100.00

Pattern sequence: • • • •

Shovel: 0, unfair; 1, fair Auction: 0, lottery or queue; 1, auction Pmax: 0, no max profit; 1, max profit PmaxVP: 0, no max profit; 1, max profit.

20

Table 3: Descriptive Statistics.

Econ 1 (n = 517) Mean (SD) .504(.500)

Econ 3 (n = 333) Mean (SD) .591(.492)

Law (n = 120) Mean (SD) .691(.463)

Tour (n = 411) Mean (SD) .856(.351)

Lang (n = 84) Mean (SD) .916(.278)

Math (n = 148) Mean (SD) .283(.452)

.056(.230) .500(.500) .442(.497) .226(.418) .280(.449) .220(.414) .168(.374) .104(.306)

.060(.237) .603(.489) .336(.473) .144(.351) .201(.401) .228(.420) .255(.436) .171(.377)

.066(.250) .316(.467) .616(.488) .150(.358) .325(.470) .175(.381) .166(.374) .183(.388)

.065(.248) .369(.483) .564(.496) .216(.412) .236(.425) .221(.415) .192(.394) .133(.340)

.095(.295) .178(.385) .726(.448) .142(.352) .309(.465) .273(.448) .142(.352) .130(.339)

.047(.212) .459(.500) .493(.501) .283(.452) .243(.430) .209(.408) .128(.335) .135(.343)

.075(.264) .288(.453) .464(.499) .172(.377) .172(.377) .239(.427) .491(.500) .096(.295)

.102(.303) .300(.459) .444(.497) .153(.360) .204(.403) .240(.427) .429(.495) .126(.332)

.066(.250) .258(.439) .458(.500) .216(.413) .166(.374) .266(.444) .441(.498) .125(.332)

.082(.275) .343(.475) .450(.498) .124(.330) .240(.428) .238(.426) .403(.491) .116(.321)

.083(.278) .214(.412) .547(.500) .154(.363) .202(.404) .392(.491) .273(.448) .130(.339)

.074(.263) .283(.452) .439(.497) .202(.403) .222(.417) .331(.472) .331(.472) .114(.319)

Mother background Primary school .054(.226) Junior High School .319(.466) High School .512(.500) College Degree .114(.318) Blue collar .098(.298) White collar .394(.489) UpperMiddleClass .131(.338) MiddleClass .029(.168) Unemployed .027(.162) Housekeeper .319(.466)

.120(.325) .381(.486) .393(.489) .105(.307) .129(.335) .312(.464) .126(.332) .048(.214) .015(.121) .369(.483)

.058(.235) .308(.463) .441(.498) .191(.395) .133(.341) .358(.481) .141(.350) .058(.235) .016(.128) .291(.456)

.065(.248) .372(.483) .472(.499) .087(.283) .111(.315) .343(.475) .128(.335) .046( .21) .021(.146) .347(.476)

.059(.238) .321(.469) .464(.501) .154(.363) .095(.295) .464(.501) .023(.153) .035(.186) .011(.109) .369(.485)

.067(.251) .418(.495) .418(.495) .094(.293) .121(.327) .358(.481) .074(.263) .081(.273) .040(.197) .324(.469)

Female Education Prof HS Tech.HS Liceo HS Marks60-69 Marks70-79 Marks80-89 Marks90-99 Marks100

Father background Primary school Junior High School High School College Degree Blue collar White collar UpperMiddleClass MiddleClass

21

Table 4: Fairness of Allocation Mechanisms

Lottery Auction Queue

Econ1 7.64 21.22 71.15

Econ3 8.54 29.89 61.57

Law 7.46 19.40 73.13

Tour 6.65 11.75 81.61

Lang 5.47 9.38 85.16

Math l 6.54 11.68 81.78

Lottery Auction Queue

M&M1 7.14 23.38 69.48

Acc1 7.55 21.46 70.99

E&B1 8.53 17.83 73.64

M&M3 9.35 23.74 66.91

Acc3 8.07 30.49 61.43

E&B3 8.43 38.55 53.01

Table 5: Fairness and Excess Demand

Unfair Fair

Econ 1 45.92 54.08

Econ 3 46.60 53.40

Law 63.82 36.18

Lang 67.40 32.60

Tour 78.43 21.57

Math 64.17 35.83

M&M3 61.61 38.39

Acc3 44.56 55.44

E&B3 29.87 70.13

Pearson’s χ25 = 94.53, p-value = 0.00

Unfair Fair

M&M1 38.71 61.29

Acc1 48.77 51.23

E&B1 44.64 55.36

Pearson’s χ25 = 23.04, p-value = 0.00

Table 6: Profit Maximization

96 layoffs 52 layoffs 26 layoffs 0 layoffs

Econ 1 49.58 26.12 16.31 7.99

Econ 3 42.67 22.25 20.42 14.66

Law 54.61 19.08 17.76 8.55

Lang 39.00 29.60 21.20 10.20

Tour 35.29 24.51 23.53 16.67

Math 24.60 27.27 32.62 15.51

96 layoffs 52 layoffs 26 layoffs 0 layoffs

M&M1 47.58 29.84 15.32 7.26

Acc1 49.86 25.48 17.26 7.40

E&B1 50.89 24.11 14.29 10.71

M&M3 32.14 27.68 21.43 18.75

Acc3 45.08 21.24 19.17 14.51

E&B3 51.95 16.88 22.08 9.09

22

Table 7: My opinion on VP’s Behavior

96 layoffs 52 layoffs 26 layoffs 0 layoffs

Econ 1 55.24 20.13 16.97 7.65

Econ 3 57.07 17.80 12.57 12.57

Law 70.39 15.79 11.84 1.97

Lang 57.40 16.60 15.60 10.40

Tour 72.55 11.76 11.76 3.92

Math 57.22 25.67 14.97 2.14

96 layoffs 52 layoffs 26 layoffs 0 layoffs

M&M1 50.00 26.61 18.55 4.84

Acc1 60.00 15.89 15.34 8.77

E&B1 45.54 26.79 20.54 7.14

M&M3 50.89 25.89 10.71 12.50

Acc3 61.14 12.95 16.06 9.84

E&B3 55.84 18.18 6.49 19.48

Table 8: The VP and myself

VP would choose like me VP would fire more VP would fire less

Econ1 45.92 29.12 24.96

Econ3 43.46 36.13 20.42

Law 46.05 35.53 18.42

Tour 33.60 40.60 25.80

Lang 38.24 50.00 11.76

Math 26.20 58.29 15.51

VP would choose like me VP would fire more VP would fire less

M&M1 43.55 30.65 25.81

Acc1 46.85 30.41 22.74

E&B1 45.54 23.21 31.25

M&M3 34.82 43.75 21.43

Acc3 44.56 36.79 18.65

E&B3 53.25 23.38 23.38

23

Table 9: Selection into Economics

Shovel

OLS 0.168 ∗∗

Probit 0.473 ∗∗

0.116 ∗∗

0.322 ∗∗

(0.028)

Auction

(0.077)

(0.037)

(0.103)

0.281 ∗∗

0.095 ∗∗

ProfMax

(0.076)

(0.026)

Marg. Effects 0.182 ∗∗ (0.0294)

0.126 ∗∗

(0.0412)

0.108 ∗∗

(0.0293)

ProfMaxVP

− 0.047 +

− 0.140 +

− 0.054 +

Gender

− 0.186 ∗∗

− 0.517 ∗∗

− 0.199 ∗∗

0.113 + (0.059)

0.311 + (0.168)

(0.026)

(0.029)

Tech. HS Liceo HS

− 0.018

(0.058)

Marks70-79

(0.076)

(0.082)

− 0.079

(0.168)

0.023

0.059

(0.038)

Marks80-89

(0.109)

0.035

0.099

(0.040)

Marks90-99

(0.115)

0.048

0.135

(0.0294)

(0.0316)

0.120 +

(0.0646)

− 0.030

(0.0646)

0.023

(0.0423)

0.038

(0.0448)

0.052

(0.044)

(0.127)

Marks100

− 0.030

− 0.113

− 0.042

Father HS

− 0.024

− 0.074

− 0.028

Father College

− 0.023

− 0.069

− 0.026

(0.045)

(0.032)

(0.047)

Mother HS

(0.093)

(0.136)

0.051

0.150

(0.031)

Mother College

(0.093)

0.013

0.040

(0.050)

Father WhiteCollar

(0.148)

0.026

0.089

(0.042)

(0.125)

0.262 ∗

0.088 ∗

Father UpperMiddleClass

(0.039)

Father MiddleClass

(0.114)

0.074

0.216

(0.051)

Mother WhiteCollar Mother UpperMiddleClass

(0.135)

(0.149)

0.001

0.000

(0.050)

(0.148)

− 0.008

− 0.033

(0.059)

0.146 ∗ (0.075)

(0.172)

0.462 ∗ (0.237)

(0.0500) (0.0507)

(0.0357)

(0.0518)

0.057

(0.0359)

0.015

(0.0575)

0.034

(0.0485)

0.100 ∗

(0.044)

0.084

(0.0590)

0.000

(0.0570)

− 0.012

(0.0657)

− 0.164 ∗

Mother MiddleClass

−

Mother not working

− 0.015

− 0.048

− 0.018

Siblings

− 0.002

− 0.007

− 0.002

Intercept

(0.047)

(0.035)

0.315 ∗∗ (0.083)

−

(0.139)

(0.102)

−

(0.0737)

(0.0534)

(0.0392)

0.506 ∗∗ (0.237)

n 1278 1278 Correctly predicted 67.29% 67.29% Log-likelihood -776.47 2 Wald χ23 182.42 Pseudo R-squared 0.138 0.11 Binary dependent variable: Enrolled in Economics=1, OLS and Probit Estimates with Robust Standard Erros. The sample consists of firstyear students only. The reference group is given by male Professional HS graduates in the lowest Mark bracket (60-69), blue collar parents 24 at 10%, ∗ : significant at 5%, ∗∗ : with no HS diploma. + : significant significant at 1%.

Table 10: Learning vs. Indoctrination. A Probit Model with sample selection - n = 1611, censored observations: 763. Robust Standard Errors in parenthesis. E&B1 Acc1 E&B3 Acc3 M&M3

Auction 0.051

Shovel − 0.008

(0.201)

(0.180)

0.187

−

(0.157)

0.712 ∗∗

0.244 + (0.143)

0.449 ∗∗

0.023

(0.138)

− 0.000

0.004

− 0.097

(0.172)

(0.158)

0.076

− 0.606 ∗∗

(0.198)

(0.173)

0.335

(0.207)

(0.210)

ProfMax 0.125

(0.177)

(0.155)

− 0.419 ∗∗ (0.176)

0.243 +

(0.138)

0.123

(0.187)

0.244

(0.151)

0.012

(0.167)

− 0.025

Father HS

− 0.072

(0.119)

(0.109)

(0.107)

0.020

− 0.213 ∗

Father College

− 0.086

0.076

0.105

− 0.253 +

Mother College Father WhiteCollar Father UpperMiddleClass

(0.162)

0.027

(0.132)

(0.177)

(0.163)

0.124

0.212 +

− 0.060

0.316 +

− 0.138

(0.124)

(0.111)

0.199

(0.201)

(0.184)

(0.157) (0.111) (0.180)

0.115

(0.148) (0.106)

(0.151)

0.061

(0.108)

0.003

(0.177)

(0.164)

(0.151)

0.109

− 0.253 +

− 0.147

0.189

0.272 + (0.148)

− 0.097

− 0.205

0.080

− 0.067

− 0.235

0.136

(0.164)

Father MiddleClass

− 0.298

Mother WhiteCollar

− 0.003

Mother UpperMiddleClass

− 0.037

Mother MiddleClass

(0.140)

0.002

(0.171)

Gender

Mother HS

− 0.186

(0.194)

ProfMaxVP − 0.127

(0.213)

(0.183)

− 0.382 ∗

(0.191)

(0.173)

0.009

(0.216)

(0.200)

0.060

0.128

(0.149)

(0.144) (0.177)

0.042

(0.172)

0.232

(0.196)

0.600 ∗

(0.151) (0.182)

0.063

(0.172)

0.185

(0.196)

− 0.148

(0.274)

(0.312)

(0.283)

Mother not working

− 0.135

− 0.252

(0.156)

(0.154)

(0.157)

Siblings

− 0.039

− 0.166

− 0.011

− 0.179

Intercept

− 0.846 ∗∗ (0.346)

(0.265)

− 0.152

− 0.002

ρ

− 0.154

− 0.190

(0.174)

(0.127)

(0.119)

0.530 ∗

(0.385)

(0.310)

(0.289)

(0.149)

0.105

(0.115)

(0.271)

0.239

(0.303)

0.077

(0.116)

(0.284)

0.396

(0.306)

Log-pseudolikelihood −1597.82 −1619.926 −1503.852 −1609.135 + : significant at 10%, ∗ : significant at 5%, ∗∗ : significant at 1%. ρ is the correlation coefficient between the selection equation (2) and the treatment equation (1). The reference group is given by male Professional HS graduates in the lowest Mark bracket (60-69), blue collar parents with no HS diploma.

25

Table 11: Learning vs. Indoctrination: Selection Equation. n = 1611, censored observations: 763. Robust Standard Errors in parenthesis. Gender Tech. HS Liceo HS Marks70-79 Marks80-89 Marks90-99 Marks100 Father HS Father College Mother HS Mother College Father WhiteCollar Father UpperMiddleClass Father MiddleClass Mother WhiteCollar Mother UpperMiddleClass

Auction − 0.512 ∗∗ (0.070)

Shovel − 0.512 ∗∗

ProfMax − 0.514 ∗∗

0.263 ∗∗ (0.140)

0.252 + (0.144)

(0.070)

0.267 ∗ (0.139)

(0.070)

− 0.208

− 0.213

− 0.222

0.079

0.079

0.085

(0.140)

(0.140)

(0.103)

(0.099)

0.148

0.146

(0.103)

(0.103)

0.295 ∗∗

0.293 ∗∗

(0.109)

(0.108)

0.089

0.092

(0.143) (0.097)

0.153

(0.101)

0.301 ∗∗

(0.107)

0.087

ProfMaxVP − 0.509 ∗∗ (0.070)

0.256 +

(0.140)

− 0.226

(0.141)

0.069

(0.099)

0.128

(0.106)

0.273 ∗∗

(0.114)

0.087

(0.115)

(0.114)

(0.115)

(0.114)

− 0.025

− 0.025

− 0.026

− 0.024

0.035

0.036

0.034

(0.080)

(0.080)

(0.118)

(0.118)

0.059

0.060

(0.080) (0.118)

0.058

(0.080)

0.039

(0.119)

0.059

(0.080)

(0.080)

(0.080)

(0.080)

− 0.008

− 0.008

− 0.007

− 0.004

0.080

0.078

0.081

(0.131)

(0.131)

(0.107)

(0.108)

0.298 ∗∗

0.296 ∗∗

(0.098)

(0.098)

0.185

0.183

(0.128)

(0.128)

0.001

0.003

(0.126)

(0.126)

0.043

0.045

(0.131) (0.108)

0.297 ∗∗

(0.098)

0.184

(0.128)

0.002

(0.126)

0.044

(0.132)

0.081

(0.108)

0.296 ∗∗

(0.098)

0.190

(0.128)

0.000

(0.126)

0.041

(0.146)

(0.146)

(0.146)

(0.146)

Mother MiddleClass

− 0.293

− 0.289

− 0.289

− 0.296

Mother not working

− 0.025

− 0.022

− 0.025

− 0.025

0.005

0.004

0.005

Siblings Intercept

(0.194)

(0.116)

(0.194)

(0.116)

(0.087)

(0.087)

0.076

0.080

(0.192)

(0.193)

+:

(0.194)

(0.116) (0.087)

0.087

(0.196)

(0.193)

(0.116)

0.002

(0.087)

0.100

(0.197)

significant at 10%, ∗ : significant at 5%, ∗∗ : significant at 1%. The reference group is given by male Professional HS graduates in the lowest Mark bracket (60-69), blue collar parents with no HS diploma.

26

Table 12: Learning vs. Indoctrination: A Multivariate Probit Model.

SML with 500 draws, n = 848. Robust Standard Errors in parenthesis. E&B1 Acc1 E&B3 Acc3

Auction 0.055

Shovel − 0.008

(0.200)

(0.181)

− 0.247 +

0.178

(0.157)

(0.143)

0.712 ∗∗

(0.209)

(0.176)

0.034

(0.138)

0.347 +

− 0.008

(0.208)

0.446 ∗∗

ProfMax 0.135

(0.197)

(0.171)

(0.160)

0.013

− 0.105

0.074

− 0.610 ∗∗

− 0.423 ∗

Gender

− 0.084

− 0.243 ∗∗

Father HS

− 0.074

Father College

− 0.109

M&M3

Mother HS Mother College Father WhiteCollar Father UpperMiddleClass Father MiddleClass Mother WhiteCollar Mother UpperMiddleClass Mother MiddleClass

(0.198)

(0.177)

(0.099)

(0.091)

0.020

(0.118)

(0.109)

0.068

(0.156)

(0.176)

0.079

(0.090)

− 0.059

0.308 +

− 0.123

(0.186)

0.243

(0.158)

0.006

(0.174)

0.255 ∗∗

(0.090)

− 0.238

0.214 +

0.192

0.120

(0.197)

0.122

0.122

(0.200)

0.264 +

(0.142)

− 0.213 ∗

(0.165) (0.112)

(0.180)

0.024

(0.107)

(0.175) (0.124)

ProfMaxVP − 0.124

(0.157) (0.112)

(0.182)

(0.108)

(0.156)

0.067

(0.112)

0.029

(0.183)

(0.164)

(0.152)

0.118

− 0.256 +

− 0.161

0.203

0.305 ∗

− 0.137

− 0.281 ∗

0.105

− 0.096

− 0.299 +

0.120

(0.148)

(0.136)

− 0.288

(0.208)

(0.180)

− 0.390 ∗

0.002

(0.193)

(0.173)

− 0.031

0.013

(0.218)

(0.201)

0.035

0.096

(0.303)

(0.284)

(0.149)

(0.134)

(0.175)

0.043

(0.171)

0.229

(0.197)

0.647 ∗

(0.281)

Mother not working

− 0.137

− 0.262 + (0.156)

(0.154)

Siblings

− 0.031

− 0.164

− 0.017

Intercept

− 0.929 ∗∗

(0.174) (0.127)

(0.120)

0.416 +

(0.227)

Log pseudo likelihood = -3850.24 - Wald   0.086 1 0.244 ∗∗ 0.075 (0.060) (0.059)   (0.058)  1 0.119 ∗ 0.056    b = (0.056)  (0.055) Σ    1 0.268 ∗∗   (0.052)  1 2 LR test χ6 = 47.58 p-value = 0.00 Robust Standard Errors in parenthesis

0.113

(0.116)

− 0.010

(0.206) (0.209) χ2112 = 348.01

+:

(0.152)

(0.137) (0.177)

0.073

(0.175)

0.187

(0.201)

− 0.082

(0.274)

0.097

(0.159)

− 0.198 + (0.118)

0.240

(0.213)

significant at 10%, ∗ : significant at 5%, ∗∗ : significant at 1%. The reference group is given by male Professional HS graduates in the lowest Mark bracket (60-69), blue collar parents with no HS diploma.

27

References Andreoni, J. and L. Vesterlund (2001), ‘Which is the fair sex? Gender differences in altruism’, Quarterly Journal of Economics 116, 293–312. B¨ orsch-Supan, A. and V. Hajivassiliou (1993), ‘Smooth unbiased multivariate probability simulators for maximum likelihood estimation of limited dependent variable models’, Journal of Econometrics 58, 347–368. Cappellari, L. and S.P. Jenkins (2003), ‘Multivariate probit regression using simulated maximum likelihood’, The Stata Journal 3, 278–294. Carter, John R. and Michael D. Irons (1991), ‘Are economists different, and if so, why?’, Journal of Economics Perspectives 5, 171–177. Eckel, C. C. and P. J. Grossman (1996), ‘The relative price of fairness: Gender differences in a punishment game’, Journal of Economic Behavior & Organization 30, 143–158. Frank, R.H., T. Gilovich and D.T. Regan (1993), ‘Does studying economics inhibit cooperation’, Journal of Economic Perspectives 7, 159–171. Frey, B.S. and S. Meyer (2003), ‘Are political economists selfish and indoctrinated? evidence from a natural experiment’, Economic Inquiry 41, 448–462. Frey, B.S. and S. Meyer (2005), Do business students make good citizens?, Working paper, Institute for Empirical Research in Economics. Frey, B.S., W.W. Pommerehne and B. Gygi (1993), ‘Economics indoctrination or selection? Some empirical results’, Journal of Economics Education 24, 271–281. Gorman, Raymond F. and James B. Kehr (1992), ‘Fairness as a constraint on profit seeking: Comment’, American Economic Review 82, 355–358. Hajivassiliou, V. and P. Ruud (1994), Classical estimation methods for LDV models using simulation, in E.Engel and D.McFadden, eds, ‘Handbook of Econometrics’, North-Holland, pp. 2383–2441. Haucap, J. and T. Just (2003), Not guilty? another look at the nature and nurture of economics students, Discussion paper no. 8, University of the Federal Armed Forces Hamburg. 28

Kahneman, D., J.L. Knetsch and R. Thaler (1986a), ‘Fairness and the assumption of economics’, Journal of Business 59, S285–S300. Kahneman, D., J.L. Knetsch and R. Thaler (1986b), ‘Fairness as a constraint on profit seeking’, American Economic Review 76, 728–741. Keane, M. P. (1994), ‘A comuputationally practical simulation estimator for panel data’, Econometrica 62, 95–116. List, John A. (2006), ‘Field experiments: A bridge between lab and naturally occurring data’, Advances in Economic Analysis & Policy 6, Article 8. Marwell, G and R.E. Ames (1981), ‘Economists free ride, does anyone else? experiments on the provision of public goods IV’, Journal of Public Economics 15, 295– 310. Rubinstein, A. (2006), ‘A sceptic’s comment on the study of economics’, The Economic Journal 116, C1–C9. Stanley, T.D. and U. Tran (1978), ‘Economics students need not be greedy: Fairness and the ultimatum game’, Journal of Socio-Economics 27, 657–664. Varian, H. R. (2002), Intermediate Microeconomics: A Modern Approach, W.W. Norton & Company, New York.

29