Faculty of Economics and Business
The predicted economic effects of a track split
A case study for Economics and Business
By Annemijn van Rheden
Student number: 10204512
Bachelor Thesis
2014-2015 Semester 1
Table of contents
1. Introduction……….1
2. Background………...2
2.1 Previous research on tracking ……...………....2
2.2 Characteristics of Economics and Business Students……….4
3. Data and methodology...………....……….…5
3.1 Context………5
3.2 Data……….6
3.3 Historical division of students...……….7
3.4 Method………9
4. Results…..………11
4.1 Simulations ………....………11
4.2 Side effects of a track split ……….……..….…14
5. Discussion and conclusion………...…...17
6. References………...….20
Appendix I………...….22
ABSTRACT
In this paper the predicted effects of a track split of Economics & Business at University of Amsterdam are analyzed. Firstly, the main conclusion is that the predicted impact of a track split on student achievement is positive for Business students. Students in the Economics track perform well, however a large group within the track are students who switched from Business to Economics. In case the switching option is removed by imposing early tracking, the university may lose well-performing students in the most severe scenario. Secondly, the amount of students continuing their studies in Economics is four times higher than for Business, whereas at the start of the year the distribution is 50-50% over Business and Economics. Thirdly, considering the gender effects, women score significantly better than men in study performance and they are more likely to pursue their studies in Economics. Empirical data of two cohorts with 1380 students of the UvA is analyzed in multiple simulations to examine the predicted effects of a track split. By controlling for the specific choice of a broad bachelor and geographical background, little or no differential effects between performance of Economics and Business students were found.
1
1. Introduction
The separation of students after an initial period with one curriculum and according to their ability is called ability tracking. As an indicator of student achievement, many studies have been done on the impact of ability tracking. Worldwide, policies are being made based on the outcomes of these studies, but the value of ability grouping is a subject of much debate in the literature.
The central argument behind ability tracking is that homogeneous classrooms permit a focused curriculum and appropriately paced instruction that leads to the maximum amount of learning by all students (Hanushek and Wössmann, 2006). Supporters of tracking state that detracking comes at the cost of higher ability students, whereas lower ability students would be impacted by a large positive effect on achievement (Argys, Rees, & Brewer, 1996). On the contrary, opponents of tracking, plead that lower ability students in a heterogeneous group need the presence of their higher level peers to stimulate learning and raising their level of achievement (Betts and Shkolnik, 2000a). In addition, Hanushek and Wössmann (2006) conclude that early tracking increases inequality in achievement and state that there is very little evidence that there are efficiency gains associated with this increased inequality. Van Elk et al. (2011) supports this conclusion and did not find an efficiency gain of early tracking in a study of high school tracking on college completion in the Netherlands.
Research on tracking is inconclusive. So far, research on tracking has solely been done in primary and secondary education and the outcomes differ on countries and methods used. Also, in earlier research the main focus is on tracking by ability, whereas this paper focusses on tracking by major. Although the consisting literature focusses on ability tracking, the intuitions give a useful insight on the predicted effects and implications of grouping. This paper presents the results of an empirical study on the predicted effect of tracking by major on student achievement for an academic track by doing simulations. An empirical track split of Economics and Business at the UvA will be simulated in multiple ways, to try and formulate the predictive ability of tracking on student achievement. This research adds to the existing body of literature in a number of ways. Firstly, the main contribution of this paper to existing literature is that it shows the economic consequences of tracking on academic achievement. Secondly, the basis of the Economics and Business bachelor is of a unique kind. Students face the same application requirements for both tracks, as the first year is comprehensive. Therefore, we cannot say upfront that students who take Business Studies are of lower academic ability than the Economics students.
2
The results of this study show firstly that tracking has a predicted positive impact on student achievement of Business students. Students Economics perform well, but consist of a large group of students who switched from Business to Economics during the first year. In case the switching option is ruled out by imposing early tracking, the university might lose these well-performing students. Secondly, the amount of students continuing their studies in Economics is four times higher than for Business, whereas at the start of the year the distribution is 50-50% over Business and Economics. Thirdly, considering the gender effects, women score significantly better than men in study performance and they are more likely to pursue their studies in Economics.
The structure of this paper is as follows. In section 2 the previous relevant literature on tracking is analyzed and the student characteristics of Economics and Business are discussed. In section 3 data used and method are described. In the fourth section the results are presented and those are discussed in section 5, followed by a conclusion.
2. Background
This section gives an overview of the past research done on tracking. This section starts with a brief explanation of tracking, followed by an analysis of the relevant literature on tracking. Secondly, the determinants of choosing an Economics or Business major will be explained by using earlier research.
2.1 Previous research on tracking
During the early schooling of children, there is an initial period of exposure to the same curriculum. After primary or secondary school, the children are separated in different tracks, according to the children’s ability, determined by performance and / or ability tests. This is called tracking, or ability grouping, where higher-achieving students are being separated from lower-achieving students. The key factor in this selection process is perceived ability (Brunello et al., 2004). Consequently, in a world with imperfect information, tracking conveys information about individual ability to the labor market.
Argys et al. (1996) examine the impact of detracking on high school student achievement by estimating a standard education production function using a nationally representative survey from America. They find that students in lower tracks would realize achievement gains from being put together with higher-ability students, but these gains would
3
come at the expense of higher-ability students in higher level tracks. Conversely, Betts and Shkolnik (2000a) find little or no differential effects of grouping for high-achieving, average, or low-achieving students by estimating an education production function. They analyse the differential and overall effects of a formal policy on ability grouping by using data from both grouping and non-grouping schools. The results are less strong than earlier studies suggested. Nor do they find that ability grouping leads to differences in resource allocation. Betts and Shkolnik therefore conclude that ability grouping has no or little effect on both student achievement and resource allocation in secondary schools.
Hanushek and Wössmann (2006) take into account the possible bias of unmeasured factors in the empirical approach of both Argys et al. and Betts and Shkolnik and conduct an international differences-in-differences approach. They identify tracking effects by comparing differences in outcome between primary and secondary schools across tracked and non-tracked systems. A comparison between the level and distribution of performance of younger students and older students across countries with and without tracking is made. The conclusion is that variation in performance between high-ability and low-ability students tends to increase across levels of schooling when a country induces early tracking.
So far, research on ability tracking has only been done in primary and secondary education. The main reason is that in many developed countries this is compulsory, whereas higher education is optional (Brunello et al., 2004). However, some studies (Brunello, & Checchi, 2007; Van Elk et al., 2011; Malamud, Pop-Eleches, 2011) estimate the effects of tracking by ability on higher education completion. The results show that early tracking has a detrimental effect on completion of higher education for students at the margin of the Dutch high and low tracks (Van Elk et al., 2011). The differences in timing of ability tracking have the most negative effect on the low ability pupils. As low and high ability students were grouped together in a comprehensive class, the probability of completing higher education increased by 25%. This result suggests inefficiency of the early tracking system with respect to completion of higher education. A study conducted by Brunello and Checchi (2007) corroborate this. The conclusion of their paper is that early tracking has a detrimental impact on educational attainment, because it prevents some individuals to continue their studies after secondary education. Moreover, they argue that reducing the number of tracks available, might lead to an increase in intergenerational mobility in educational attainment, but that this may induce exclusion of people from disadvantaged backgrounds. Malamud and Pop-Eleches (2011) focus on the effects of a postponement in tracking by ability for disadvantaged students in Romania. They find that students from poor, rural areas with less educated parents
4
are significantly more likely to finish an academic track due to the policy, because these students gain time to catch up with their more advantaged student colleagues. However, the real long-term effect remains unchanged: it does not increase higher education completion of this group.
To summarize, tracking induces students to perform according to the level of the track and are discouraged to deviate from the set route. The effect of ability grouping is ambiguous and depends on the researchmethods used. This paper adds to this literature by analysing the effects of tracking by specialization in an academic program, instead of tracking by ability. The characteristics of the majors Economics and Business are described below.
2.2 Characteristics of Economics and Business students
The number of undergraduate degrees in Economics declined precipitously in the early nineties in Australia (Margo and Siegfried, 1996). These falling numbers have led to concerns of economists and policy makers, and therefore a lot of research is done on the causes of the decline. First of all, Economics and Business Studies are fairly competitive tracks (Lewis and Norris, 1997; Salemi and Eubanks, 1996; Willis and Pieper, 1996). The increase in enrollments for Business Studies the amount of enrollments for Economics degrees decreased significantly. Schools offering undergraduate degrees in Business Studies had only 25% as many economics majors as schools that did not offer a Business major (Willis and Pieper, 1996). However, both Willis and Pieper (1996) and Margo and Siegfried (1996) did not find evidence for a direct substitution effect of Economics and Business majors.
What factors explain the shift in the choice of major from Economics to Business Studies? Worthington and Higgs (2004) researched what personal characteristics would be predictive of a student choosing an economics major. By using a binary probit model, they found that the choice of an Economics major is a function of student personality, interest in the Economics profession and non-Economics secondary studies. Firstly, the study indicates that the students who select an Economics major have a more positive outgoing personality than Business students in general. However, this result stems from the fact that the curriculum tested had a large emphasize on presenting and debating in the introductory Economics courses. The second factor shows that the level of student interest in a specific profession in the field of Economics is seen as an important factor in the choice of the Economics major. The perceptions and interest in the Economics profession is perceived to be much closer for an Accounting and Economics major than that between Economics and a
5
non-Economics or related discipline of the social sciences. Job perspective plays a bigger role in the literature. The most important factors of an increase in Business Studies is that this track was seen as a more career focused track and would lead to more highly paid jobs than Economics (Lewis and Norris, 1997; Calkins and Welki, 2006). The third factor found by Worthington and Higgs is the subject choice in secondary school that influences the choice for an Economics major. Economics is perceived as rigorous, boring and dull (Lewis and Norris, 1997) and overly mathematical (Calkins and Welki, 2006), which reduces interest in secondary school. Students are perceived as being less well prepared in mathematics and prefer ‘easier’ courses, such as Business Studies. Moreover, the importance of the background knowledge of calculus and Economics as determinants for study success in introductory Economics courses was found by Anderson, Benjamin and Fuss (1994).
Finally, gender plays a role in the decision for an Economics or Business major. From the study of Worthington and Higgs (2004), it seemed that female students are much less likely to select a major in Economics than men due to the curriculum that is of less interest to women. Calkins and Welki (2006) found evidence that mathematics is a major determinant for women not to choose Economics. Moreover, the average grade in introductory Economics courses is 2.5 to 3.5- percentage points higher for men than for women (Anderson et al., 1994; Dynan and Rouse, 2010).
3. Data and methodology
3.1 Context
The Economics & Business bachelor program at the UvA has a nominal duration of three years. In the first year, students follow a fixed program with a mix of Economics courses and Business courses. The year is spread in 6 blocks: term 1, 2, 4 and 5 take 8 weeks each: block 3 and 6 consist of 4 weeks. The first year counts 60 European Credits (EC’s) in total and after the first year students can choose from different packages of courses to specialize in Economics or Business oriented tracks. The students are checked during their first year on the amount of credits and scores obtained.
The UvA has set a policy to impose tracking in the first year, therefore students have to choose a major at the start of their studies. All (Business / Fiscal) Economics oriented tracks will be grouped under Economics. Business Studies will be a study on itself. The curricula of both before and after the track split are presented in table 1.
6
Table 1. Curriculum of comprehensive track and tracked by major after a track split
Comprehensive Tracked by major Term Economics & Business Business Economics
1 Mathematics Strategy Introduction in Economics
1 Marketing & Strategy Business Mathematics Mathematics
2 Accounting Marketing Macro-economics
2 Micro-economics Finance Fiscal Economics & Finance
3 Fiscal Economics Academic Skills Research Skills 1
3 Structure of Economical Knowledge
4 Statistics Organization & Management Micro-Economics 4 Organization & Management Business Statistics Statistics
5 Finance Business Operations & Processes Economics of Markets and Organizations
5 Macro-economics Economics Accounting
6 Research Project Business Research Research Skills 2
6 Research Project
1-6 Academic Skills* Academic Skills*
* Academic Skills is a course with no credits, but obligatory attendance. The tutorials enhance both academic skills (writing, presenting, debating) and mentorship.
3.2 Data
The UvA provided data of students who applied for the Dutch bachelor of Economics & Business in 2010 and 2011 (n = 1577). The students filled in an extensive intake survey with background variables, personal information and track choice. Table 2 reports the descriptive statistics of both cohorts. Credits and GPA are continuous variables, the other variables are dummy variables. In 2010, students could choose from six different tracks: Business Studies, Economics, Fiscal Economics, and Business Economics with specializations; Accountancy & Control, Finance and Organizational Economics. In 2011 the curriculum was changed and the following options were available: Accountancy and Control, Business Studies, Economics, Fiscal Economics, Economics & Finance and Finance. In further manipulations the (Business / Fiscal) Economics oriented tracks are grouped under Economics, consistent with the policy of the track split (UvA, 2014). Therefore, only Business Studies is a separate track. The data is controlled for applicants who started their first year (n = 1380). Prior to the start, 762 students filled in a track choice and 618 students are classified as unknown. The unknown group consists of students who did not make a track choice at the intake survey. 724 students passed their first year, which means that these students continued their studies in the second year and passed the first year with a positive binding study advice, which consists of at least 42 credits, including mathematics, or special permission from the examination board. This is indicated by BSA positive (table 2).
7
The identification of students in the track ex post (i.e. at the end of year 1) is done by observing the chosen courses in the second year. As students were allowed to enroll in multiple courses, their definitive choice could not always be observed. These students were most likely to participate in multiple courses from one of the Economics-oriented tracks in both 2010 (n = 52) and 2011 (n = 12), taken into account the overlap in courses, and are therefore added to Economics ex post in further analysis.
Furthermore, two background variables are used as control variables in the simulations. Firstly, in the second simulation there is a control for the motivation of students to choose for the bachelor. At the intake survey, 49% of the students, filled in the broad bachelor as first reason to study Economics & Business. The other reasons (n = 9) are grouped under ‘Other reason than comprehensive track’, and are fragmented (see Appendix I). Secondly, the data has been controlled for the zip code of the students to observe the geographical background for a radius of 100 kilometers around Amsterdam (Geographical Background Amsterdam).
Table 2. Descriptive Statistics
2010 2011
mean s.d. N mean s.d. N
Male 0.72 0.45 727 0.74 0.44 653
Chosen UvA for comprehensive bachelor 0.51 0.50 727 0.47 0.50 653
Other reason for UvA than comprehensive bachelor 0.49 0.50 727 0.53 0.50 653
Geographical background Amsterdam 0.86 0.35 727 - - -
Economics (ex ante) 0.31 0.46 727 0.26 0.44 653
Business (ex ante) 0.29 0.45 727 0.24 0.43 653
Unknown (ex ante) 0.40 0.49 727 0.50 0.50 653
Economics (ex post) 0.42 0.49 727 0.43 0.50 653
Business (ex post) 0.10 0.30 727 0.10 0.30 653
Drop out (ex post) 0.48 0.50 727 0.47 0.50 653
BSA positive 0.52 0.50 727 0.53 0.50 653
Credits 33.62 23.65 727 32.19 24.68 638
GPA 5.57 1.40 661 5.27 1.65 587
Note: s.d. is the standard deviation and N is the sample size. All summary statistics are based on the students that started their studies of cohort 2010 and 2011. There are only observations of geographical background Amsterdam available for 2010. Source: UvA 2014.
3.3 Historical division of students
Table 3 presents the division of the students among the chosen tracks. In the table the first column presents the track choice students have filled in at the intake survey, in percentage of the total amount of starting students. The second column shows the status after the first year: a chosen track or a drop out. The amount of students is presented in the third column, with
8
the matching percentages in the fourth column. In the last two columns of table 3, the mean of obtained credits and mean average grade are presented, indicating student achievement. The scores of the drop outs are omitted as the outcome would not be meaningful, evidently. The statistical difference in study performance of Economics and Business students is measured by a t-test and can be read from the last row of average EC’s and GPA after year 1 in each block. As can be seen from table 3, the difference of student achievement between Economics and Business students are significant for the ex ante Economics and Unknown groups. Although the amount of students that switch to Business Studies is rather small, it is consistent with the literature. Lewis and Norris (1997) found that Business Studies are, on average, of a lower academic ability than the remaining economics students. Also, Anderson et al. (1994) argue that background knowledge of calculus and economics were important determinants for study success in introductory economics courses. Since in the first year of Economics and Business the curriculum consists for 80% of Economics courses (table 1), this statistical difference might be explained by the fact that students who underperform choose Business Studies after a year. Thereby, students were perceived as being less well prepared in mathematics and preferred ‘easier’ courses, such as Business Studies (Lewis and Norris, 1997). In the Unknown group in the last block, we obtain the same significant result and support the intuition stated above.
Table 3. Historical background of students choices, credits and GPA obtained after the first year Choice ex ante Choice ex post Amount % of track Average EC's after year 1 Average GPA after year 1
Economics Economics 189 48% 53.84 (0.66) 6.45 (0.08) Business 17 4% 49.65 (1.68) 5.98 (0.16) Drop out 191 48% 397 29% 4.19 (1.80)** 0.47 (0.18)*** Business Economics 123 34% 54.14 (0.71) 6.40 (0.1) Business 81 22% 54.22 (0.65) 6.48 (0.09) Drop out 161 44% 365 26% -0.08 (0.97) -0.07 (0.13) Unknown Economics 275 44% 53.42 (0.53) 6.36 (0.06) Business 39 6% 50.33 (1.29) 6.05 (0.12) Drop out 304 49% 618 45% 3.08 (1.39)** 0.31 (0.13)**
Standard errors in parentheses. T-test in last row of each block, significance levels are indicated with *α=0.1, **α=0.05, *** α=0.01. Note: total of 1380 observations.
Conversely to Economics and Unknown, in the case of a choice ex ante in Business Studies, there is no statistical difference in performance found between the ex post Economics and
9
Business groups (table 3). This is partly by the latter intuition: Students that choose Economics after the first year perform well. However, in total, 34% of the (also well-performing) students that choose Business Studies at the start of the year switch to Economics (table 3). This result shows that students make use of the first year to orient themselves and explore both Economics and Business tracks. Remarkable is the small group of Business students that remain consistent in their choice for Business Studies, only 22%, whereas an even bigger proportion (34%) switches to Economics.
A marginal difference in the overall performance between Economics students and Business students after the first year is only found in credits obtained, with marginal underperformance of Business students (p = 0.095). However, in the GPA there is no significant difference (p = 0.184). Students who passed their first year, had three decision strategies: students could be consistent in their track choice (37%), switch to another track (19%) or choose later in the year what the track would be (44%). No statistical significant difference in study performance is found between the decision strategies.
To summarize, almost half of the students that start their studies drop out during the first year. The other half can either be consistent, switch or choose their track later during the year, without showing a difference in student achievement. Consistent with the literature, students who do not perform well in Economics tend to switch to Business Studies, but this is a relatively small amount. On the other hand, students who perform well in the introductory Economics courses are likely to choose (or even switch to) Economics.
3.4 Method
To estimate the predicted effect of the track split, three simulations will be done. Here we have to distinguish between two direct effects: the selection effect explains the predicted effect of the track split on study performance. Secondly, every simulation builds in an assumption that is explained by a causal effect: students have different reasons not to participate in the Economics & Business bachelor in case of a track split and therefore cancel out in the simulations. The historical background functions as a base, because it has the unique characteristic that a track can be chosen by the end of the year. Students did not have to fill in a preferred track ex ante nor do they have to stick to their preferred track choice. Now by imposing a track split, this characteristic no longer holds and we therefore assume firstly that students have to choose a track ex ante in all simulations. Secondly, we still take
10
into account the option that students might switch to another track. For convenience, it is assumed that switching at zero cost is still possible. In every simulation the data will be controlled for specific implications of the track split (table 4).
Table 4. Fraction of total student inflow per track choice ex ante in simulations
Background Historical
Simulation 1: People who did not make a
decision upfront (unknown), do not come
to the bachelor
Simulation 2: Unknowns left out and controlling for choice of comprehensive track
Simulation 3: Geographical Background
Choice ex ante 2010 only
Economics 0.29 0.29 0.18 0.27
Business 0.26 0.26 0.14 0.23
Unknown 0.45 - - 0.36
N N=1380 N=762 N=437 N=727
% of students in total still coming in simulation
(N=1380) 100% 55% 32% 86%
N.B. Fractions of choices ex ante in simulations are given in the table. Note that only for 2010 data on geographical background was available (N=727).
First of all, the assumption that students have to choose ex ante which track they would like to do, implies that students who do not have made a decision upfront and are not being served anymore at the UvA. The simulation shows a severe implication of the track split, as it is assumed that these students will not come to the bachelor. We see from table 6 that 45% of the indecisive students at the start of the year, cancel out in the first simulation.
Secondly, in addition to simulation 1 there will also be a control for students’ motivation. Students made known at the intake survey why they would come to the bachelor specifically. From the intake survey it appears that 49% of the students choose the UvA for the broad curriculum of the first year of the bachelor. This fraction is mostly driven by the unknown group (table 5). The UvA is the only university in the Netherlands offering a broad Economics & Business bachelor and will lose its uniqueness in case of tracking. Many students value the broad bachelor, but they are now assumed not to apply, as the faculty does not engage in serving these students anymore, indicating a causal effect of the track split. One could say that only the most determined and motivated students are left over in this second scenario.
11
Table 5. Cross section table of students attaching value to the comprehensive bachelor ex ante
Total
Chosen UvA for comprehensive
bachelor
Other reason for UvA than comprehensive
bachelor
Choice ex ante Freq. Percent Freq. Percent Freq. Percent
Economics 397 29% 154 11% 243 18%
Business 365 26% 171 12% 194 14%
Unknown 618 45% 355 26% 263 19%
N 1380 100% 680 49% 700 51%
In the third simulation, the group that geographically is most likely to come to the UvA has been analyzed (UvA data, 2014). As only for cohort 2010 data was available, the students (n = 727) are controlled for the living area in a radius of 100 kilometers from the university and entails 86% of the students (table 4). In this scenario, it is still assumed that students have to choose ex ante which track they want to do. In addition, we assume that students who live close by will make a decision for a track upfront. Now the undetermined group is not left out, but they are randomly allocated to Economics (52%) and Business (48%). The division is based on the percentages of simulation 1, which is the historical background with omission of the undecided students ex ante. The assignment of unknowns has been done randomly, as we do not know the track choice ex ante. Moreover, the assumption that switching is possible
still holds.
4. Results
4.1 Simulations
Simulation 1. Undecided students ex ante do not come to the UvA in case of tracking
In the first simulation, the undecided students at the start of the year are left out, indicating a causal effect of the track split. The results can be read from table 6, and has the same build up as the table with historical background. Now the division between Economics and Business is 52% and 48%, respectively. Only a significant difference is found in the Economics ex ante group. Students who switch from Economics to Business perform worse, as was already described in the historical background. The division of decision strategy has changed. 34% of the students who passed the first year have switched their track choice, whereas 66% has a consistent track choice. No significant difference in performance is found between these decision strategies (p = 0.33). The statistical significant difference as found in credits from
12
the historical data between the overall performance of Economics and Business ex post disappears in this simulation (p = 0.25). The study performance measured in GPA does not differ for both ex post tracks (p = 0.34).
Table 6. Simulation 1: Undecided students ex ante do not come to the UvA in case of tracking
Choice ex ante Choice ex post Amount % of track Average EC's after year 1 Average GPA after year 1
Economics Economics 189 48% 53.84 (0.66) 6.45 (0.08) Business 17 4% 49.65 (1.68) 5.98 (0.16) Drop out 191 48% 397 52% 4.19 (1.80)** 0.47 (0.18)*** Business Economics 123 34% 54.14 (0.71) 6.40 (0.1) Business 81 22% 54.22 (0.65) 6.48 (0.09) Drop out 161 44% 365 48% -0.08 (0.97) -0.07 (0.13)
Standard errors in parentheses. T-test in last row of each block, significance levels are indicated with *α=0.1, **α=0.05, *** α=0.01. Note: total of 762 observations.
Simulation 2. Unknowns left out and controlling for choice of comprehensive track
Table 7 shows the results of simulation 2, where students who did not fill in a track choice ex ante (unknowns) are left out. In addition, in the simulation the data is controlled for the reason if students come to the comprehensive bachelor specifically: these students are left out. In table 7, the percentages of the historical background are added in order to see the movement in decisions by students in the simulation. Again the track choices can be found in the first 2 columns. In the third, fourth column the amounts and fractions respectively are presented. In the last two columns the study performance is presented. The division for Economics (56%) and Business (44%) has not changed much in comparison to simulation 1. In total, 71 % of the students are consistent in track choice, whereas 29% are switchers. In student achievement, no statistical difference is found between the decision strategies in credits (p = 0.43) or GPA (p = 0.28). In comparison to the historical background, the dropout rate has increased for both groups. Also, the group choosing for Economics has decreased in this scenario. A weak significant difference between Economics and Business students is observed (table 7) in the GPA of students who choose Economics ex ante (p < 0.1). For overall performance (i.e. the difference between Economics and Business in total),
insignificant p-values are obtained for credits (p = 0.41) and GPA (p = 0.45). This indicates a negative selection effect, as in case of tracking the UvA loses some well-performing students
13
who came for the broad bachelor specifically and would have chosen Economics, but now drop out due to the tracking policy.
Simulation 3. Only people from around Amsterdam come Table 8 presents the results of simulation 3. The data is controlled for the living area of the applicants. Thereby, the students who did not make a decision for a track choice ex ante are now divided over either Economics or Business (randomly, 52%-48% respectively). The buildup of the table is the same as the former ones.
There are no statistical significant differences in overall study performance in credits obtained (p = 0.63) or GPA (p = 0.48). However, in comparison to the historical background, the performance seems to increase as a result of the selection effect. No difference is observed between students with consistent (65%) or switching (35%) decision strategies in credits (p = 0.42) or GPA (p = 0.36), nor between students who ex ante chose Economics or Business (table 8).
Table 7. Simulation 2: Unkowns left out and controlling for choice of comprehensive track
Choice ex ante Choice ex post Amount % of track % of track HB Average EC's after year 1 Average GPA after year 1
Economics Economics 110 45% 48% 53.37 (0.9) 6.37 (0.1) Business 8 3% 4% 50.38 (1.63) 5.89 (0.21) Drop out 125 52% 48% 243 56% 2.99 (1.86) 0.48 (0.23)* Business Economics 55 28% 34% 53.71 (0.93) 6.36 (0.14) Business 42 22% 22% 53.76 (1.02) 6.43 (0.12) Drop out 97 50% 44% 194 44% -0.05 (1.38) -0.07 (0.19)
Standard errors in parentheses. T-test in last row of each block, significance levels are indicated with *α=0.1, **α=0.05, *** α=0.01. Note: total of 437 observations. HB= fractions in historical background.
14
Table 8. Simulation 3: Cohort 2010, controlled for geographic background 100 kilometers around Amsterdam
Choice ex ante Choice ex post Amount % of track % of track HB Average EC's after year 1 Average GPA after year 1
Economics Economics 163 49% 48% 54.05 (0.59) 6.43 (0.08) Business 16 5% 4% 53.13 (1.95) 6.25 (0.17) Drop out 156 47% 48% 335 54% 0.92 (2.03) 0.18 (0.19) Business Economics 98 34% 34% 53.60 (0.72) 6.40 (0.1) Business 48 17% 22% 54.98 (0.73) 6.60 (0.12) Drop out 144 50% 44% 290 46% -1.37 (1.03) -0.2 (0.15)
Standard errors in parentheses. T-test in last row of each block, significance levels are indicated with *α=0.1, **α=0.05, ***0.01. Note: total of 625 observations. HB= fractions in historical background.
4.2 Side effects of a track split Distribution of track choice
Figure 1. Distribution of choice ex ante and ex post
From figure 1 can be seen how students are distributed ex ante and ex post over Economics, Business or dropouts in all three simulations. The columns display the division of students over Economics (ECOexante) and Business (BUSexante) at the intake of the students. The division ex ante for historical data is omitted, as it was still assumed that students could be undetermined about their study choice and it would therefore not be meaningful. Moreover, in simulation 1 the distribution of Economics and Business of the historical data is displayed, with omission of the students who did not make a track choice ex ante. The division is stable in the three simulations. The lines reflect the distribution of the students after the first year.
52% 56% 53% 48% 44% 47% 0% 20% 40% 60% 80% 100%
%Historical Sim. 1 Sim. 2 Sim. 3
Amo un t o f s tu de nt s i n per cen ta ges
Type of data manipulation
15
Recall that students could choose for Business (BUS), Economics (ECO) or they could drop out (DROP). The ex post distribution remains also fairly stable in all situations, with the range for Dropouts [46%, 51%], Economics [38%, 43%] and Business [ 10%, 13%]. The enrollments for Economics are found to be approximately three to four times higher than for Business after the first year. However, the inflow shows almost an equal amount of students choosing Economics or Business.
Gender effects
Gender plays a role in the literature and therefore in this sample the differences in performance of men and women are taken into account. From existing research it seems that women are less likely to choose an economics major and perform worse than men. The data from the historical background is divided in gender and can be read from table 9. The students that continue in the Economics or Business track after year 1 are presented in column 2 and 3. A significant difference is found in the dropout rate (p < 0.01), see Appendix II. Of the females, 35% drops out in the first year, whereas for men the amount is 52%. This result also follows from the second and third block by evaluation: females perform significantly better than in males in both credits and GPA. Students need 42 or more credits in order to obtain a positive binding study advice. From the results appears that females are more likely to meet this requirement. Furthermore (see Appendix II), a significant higher amount of the females continued their studies in Economics than men (p < 0.01). For Business, this was only a marginal difference (p < 0.1).
16
Table 9. The gender effect on performance with data on historical background
Economics Business Drop out
Total of females (N=374) 53% 12% 35%
Total of males (N=1006) 39% 9% 52%
Total of students ex post in track 587 137 656
EC's females 54.54 54.73 - (0.64) (0.9) - EC's males 53.28 51.48 - (0.43) (0.74) - T-test on EC's 1.26 (0.77) ** 3.25 (1.16)*** - GPA females 6.56 6.46 - (0.07) (0.12) - GPA males 6.31 6.21 - (0.05) (0.08) - T-test on GPA 0.25 (0.09)*** 0.25 (0.14)** -
Note: Standard errors in parentheses. T-test on the last row with difference between females and males, significance levels are indicated with *α=0.1, **α=0.05, ***α=0.01. In case students dropped out, they are not taken into account in the tracks. That is, of the females (males) starting their studies the percentage that dropped out is 35% (52%).The study succes is not measured, as the results would not be meaningful.
The simulations have a weak effect on the decisions of women: the divisions are again fairly stable. Note that the track choice at the start of the year is higher for Business in the case with women only than when the men are included [48%-54%]. In contrast with the division of all students (figure 1), women are more likely to choose Economics after the first year with a range of [44%, 54%]. The amount of females choosing Business is between [12%, 18%] and can be explained by the lower amount of dropouts in the range [33%, 39%].
Figure 2. Distribution of choice ex ante and ex post for females
46% 52% 50% 54% 48% 50% 0% 20% 40% 60% 80% 100%
%Historical Sim. 1 Sim. 2 Sim. 3
Amo un t o f f ema le st ud en ts in per cen ta ges
Type of data manipulation
17
5 Discussion & Conclusion
Tracking has a predictive impact on study performance of Economics and Business students. This difference in performance between the tracks becomes less significant or even disappears when the data is controlled for student motivation and geographical area in the simulations. This result is mostly driven by the undecided students that start their bachelor at the UvA. From the simulations we conclude that the predicted selection effect is positive on student achievement.
Remarkable is the stable distribution of students over the simulations ex ante and ex post. This result shows that although students did not particularly value the first year because of its broadness (simulation 2), students use the first year to orient on the different tracks and choose different decision strategies. The inflow of students has a rough 50-50% distribution, but after the first year the amount of Economics students is three to four times higher than for Business. This result gives a useful insight on the value students attach to the orientation time on tracks during first year.
If we take a closer look to the predicted effects per track, we conclude that only approximately 10% of the total of students that started their studies continue their track in Business Studies. However, in the simulations 22% of the students stayed consistent in their choice for Business and performed well. In addition, from the historical data it seemed that bad-performing Economics ex ante students tended to switch to Business. This predicted selection effect decreases when controlling for motivation and geographical background. Also, it concerns a small 3-5% of the total ex ante Economics group. We therefore conclude that for Business Studies the track split would have a positive impact on the student achievement.
For Economics, the impact is detrimental. In total, approximately 48% of the students who wanted to study Economics continued in this track, whereas 34% choose Business at the intake survey and switched to Economics. On the one hand, this result contradicts the theory as students see Business Studies as a more career focused track and it is assumed to lead to more highly paid jobs than Economics (Lewis & Norris, 1997; Calkins & Welki, 2006). Since Arcidiacono et al. (2012) found that the expected future earnings play a large role in the major choice for college students, the high amount of switchers to Economics is a striking outcome. However, this contradictory result could be explained by the intuition that students have a vague and boring perception of Economics (Lewis & Norris, 1997) and have a more clear view of the track after the broad first year, as the curriculum consists for 80% of
18
Economics courses. Therefore we could state that if the possibility for switching is cancelled, the faculty could, in the most severe case, lose the well-performing students who choose Business Studies at the intake survey and switched to Economics during the year by imposing tracking due to a causal effect.
Considering the predicted selection effects of a track split on the performances of different genders, the results of this paper are in contrast with research done by Worthington & Higgs (2004) and Calkins & Welki (2006). Women choose Economics more often after the first year and have a significantly lower dropout rate. This isin line with what is found in the comprehensive analysis of males and females, though in a larger extent: 50% of the female students that continued their studies in Economics filled in Business Studies at the intake survey. The selection effect of a track split could be detrimental for women if we include switching costs: these students are well-performing in all simulations. However, as a consequence of the tracking, the causal effect is that especially women could stay in the Business track, instead of switching to Economics. One should bear in mind that by the introduction of the new curricula, that are more Economics or Business oriented, the wish for
switching is likely to decrease.
Women perform better in both credits and GPA in comparison to men and this conclusion is also found in an earlier study about gender peer effects by Oosterbeek and Van Ewijk (2014). However, in contrast to the results of men and women taken together, few marginal differences are found in performance by doing the simulations. Therefore we conclude that the differences in student achievement is driven by men.
The results of the effect of tracking on student achievement should be interpreted with care. First of all, one might doubt the internal validity: can policy implications be measured by historical data? The simulations are done with historical data of two cohorts with no switching costs and a similar curriculum before and after the track split. In reality, the courses for both tracks will change, i.e. will be either more Economics or Business oriented. Therefore, switching from one track to another during the first year will become at a cost for both the university and the student. For instance, the university will have costs in financial terms in order to provide extra courses for switchers. On the other hand, students bear these costs in terms of cognitive investment by doing these extra courses. This might have causal implications for both enrollment ex ante and ex post, i.e. the question is whether students will still consider switching to another track or if they will drop out. Hence, the effect of tracking on switchers is ambiguous at this stage. Secondly, the UvA is the only university in the Netherlands that offers a broad Economics & Business bachelor. In promotion, this is used as
19
a unique selling point. Therefore, one might argue that the motivation of students is biased in simulation 2. Then again, the division ex post is in all four situations more or less the same. This result should thus be interpreted as a valuable determinant of selection. Thirdly, externalities should be taken into account by deciding on imposing tracking. For instance, the Dutch government imposes a new loan policy in 2015, before the track split will be implemented. This could have an influence on the decision makingprocess of future students. Also, competition with other universities might increase with a track split.
This study has shown that tracking in an academic program leads to higher student achievement of Economics and Business students at the UvA. The results have little external validity as other bachelor programs have different content and different characteristics. However, they do contribute to the literature as this study makes the first step towards an analysis of tracking in a college program. Further research could analyze a track split on other academic tracks. Also, in future work the switching strategies should be analyzed more closely in case of tracking. How do we explain the (small group of) bad-performing Economists who switch to Business? And why does the large group, especially driven by women, switch to Economics after the first year?
20
6. Reference list
Alexander, K. L., & McDill, E. L. (1976). Selection and Allocation Within Schools: Some Causes and Consequences of Curriculum Placement. American Sociological Review, 41(6), 963-980. Anderson, G., Benjamin, D., & Fuss, M. A. (1994). The Determinants of Success in University
Introductory Courses. Journal of Economic Education, 25(2), 99-119.
Argys, L. M., Rees, D. I., & Brewer, D. J. (1996). Detracking America's Schools: Equity at Zero Cost?
Journal of Policy Analysis and Management, 15(4), 623-645.
Athey, S., Katz, L. F., Krueger, A. B., Levitt, S., & Poterba, J. (2007). What Does Performance in Graduate School Predict? Graduate Economics Education and Student Outcomes. The
American Economic Review, 97(2), 512-518.
Betts, J. R., & Shkolnik, J. L. (2000). Key difficulties in identifying the effects of ability grouping on student achievement. Economics of Education Review, 19, 21-26.
Betts, J. R., & Shkolnik, J. L. (2000). The effects of ability grouping on student achievement level and resource allocation in secondary schools. Economics of Education Review, 19, 1-15.
Brunello, G., & Checchi, D. (2007). Does school tracking affect equality of opportunity? New international evidence. Economic Policy, (pp. 781-861).
Calkins, L. N., & Welki, A. (2006). Factors that influence choice of major: why some students never consider economics. International Journal of Social Economics, 33(8), 547-564.
Dynan, K. E., & Rouse, C. E. (2010). The Underrepresentation of Women in Economics: A Study of Undergraduate Economics Students. The Journal of Economic Education, 28(4), 350-368. Epple, D., Newlon, E., & Romano, R. (2002). Ability tracking, school competition, and the distribution
of educational benefits. Journal of Public Economics, 83, 1-48.
Hanushek, E. A., & Wössmann, L. (2006). Does educational tracking affect performance and
inequality? Differences-in-differences evidence across countries. The economic journal, 116, 63-76.
Lewis, P., & Norris, K. (1996). Recent Changes in Economics Enrollments. Australia: Department of Economics, Murdoch University.
Malamud, O., & Pop-Eleches, C. (2011). School tracking and access to higher education among disadvantaged groups. Journal of Public Economics, 95(11-12), 1538-1549.
Margo, R. A., & Siegfried, J. J. (1996). Long-Run Trends in Economics Bachelor's Degrees. The Journal
of Economic Education, 27(4), 326-336.
Oosterbeek, H., & van Eeuwijk, R. (2014). Gender peer effects in university: Evidence from a randomized experiment. Economics of Education Review, 38, 51-63.
21 Pekkarinen, T., Uusitalo, R., & Kerr, S. (2009). School tracking and intergenerational income mobility:
Evidence from the Finnish comprehensive school reform. Journal of Public Economics, 93(7-8), 965-973.
Rees, D., Brewer, D., & Argys, L. (2000). How should we measure the effect of ability grouping on student performance? Economics of Education Review, 19, 17-20.
Salemi, M. K., & Eubank, C. (1996). Accounting for the Rise and Fall in the Number of Economics Majors with the Discouraged-Business-Major Hypothesis. The Journal of Economic
Education, 27(4), 350-361.
van Elk, R., van der Steeg, M., & Webbink, D. (2011). Does the timing of tracking affect higher education completion? Economics of Education Review, 1009-1021.
Willis, R. A., & Pieper, P. J. (1996). The Economics Major: A Cross-Sectional View. The Journal of
Economic Education, 27(4), 337-349.
Worthington, A., & Higgs, H. (2004). Factors explaining the choice of an economics major.
22
Appendix 1
Table 1. Cross section table of students motivation to study Economics & Business at the UvA
Reason ex ante Freq. Percent
Recommended 36 3%
Amsterdam 45 3%
Closeby 15 1%
Friends study at the UvA 1 0%
Comprehensive bachelor 680 49%
Prerequisites at other universities 7 1% Quality of education at the UvA 140 10%
For the masters at the UvA 29 2%
Buildup of the bachelor 210 15%
Specializations at the UvA 175 13%
Other reasons 42 3%
23
Appendix II
The difference in division over the tracks after the first year is presented for both females and males in table 1. In table 2-5, the simulations are presented for females only, to examine the impact of a track split.
Table 1. The difference in the percentage of females (males) after the first year based on the historical background
Economics Business Dropout Females (N=374) 0.53 0.12 0.35
(0.5) (0.33) (0.48)
Males (N=1006) 0.39 0.09 0.52
(0.49) (0.29) (0.5)
T-test 0.14 (0.03)*** 0.03 (0.02)** -0.17 (0.03)*** Standard errors in parentheses. T-test in last row, significance levels are indicated with *α=0.1, **α=0.05, ***0.01.
Table 2. Historical Background: females only
Choice ex ante Choice ex post Amount % of track HB Average ECTS after year 1 Average GPA after year 1
Economics Economics 58 57% 48% 55.22 (1.24) 6.72 (0.13) Business 5 5% 4% 50.4 (3.74) 6.09 (0.34) Drop out 39 38% 48% 102 27% 4.82 (3.69) 0.63 (0.37)* Business Economics 52 43% 34% 55.11 (1.18) 6.55 [(0.15) Business 34 28% 22% 55.18 (0.93) 6.55 (0.14) Drop out 34 28% 44% 120 32% -0.07 (1.51) 0 (0.21) Unknown Economics 88 58% 44% 53.74 (0.95) 6.47 (0.1) Business 6 4% 6% 55.83 (2.99) 6.25 (0.3) Drop out 58 38% 49% 152 41% -2.09 (3.13) 0.22 (0.31)
Standard errors in parentheses. T-test in last row of each block, significance levels are indicated with *α=0.1, **α=0.05, ***0.01. Note: total of 374 observations.
24
Table 3. Simulation 1: Historical background with only females, undecided students left out. Choice ex ante Choice ex post Amount % of track
HB with
men Average EC after year 1 Average GPA after year 1
Economics Economics 58 57% 48% 55.22 (1.24) 6.72 (0.13) Business 5 5% 4% 50.4 (3.74) 6.09 (0.34) Drop out 39 38% 48% 102 46% 4.82 (3.69) 0.63 (0.37)* Business Economics 52 43% 34% 55.11 (1.18) 6.55 [(0.15) Business 34 28% 22% 55.18 (0.93) 6.55 (0.14) Drop out 34 28% 44% 120 54% -0.07 (1.51) 0 (0.21)
Standard errors in parentheses. T-test in last row of each block, significance levels are indicated with *α=0.1, **α=0.05, ***0.01. Note: total of 222 observations.
Table 4. Simulation 2: only female students. The undetermined students ex ante are left out and the data is controlled for motivation
Choice ex
ante Choice ex post Amount % of track (HB)
% of track HB Average ECTS after year 1 Average GPA after year 1 Economics Economics 34 55% 57% 56 (1.17) 6.65 (0.15) Business 2 3% 5% 52.5 (1.5) 6.49 (0.47) Drop out 26 42% 38% 62 52% 3.5 (1.9)* 0.16 (0.49) Business Economics 18 31,6% 43% 55.39 (1.58) 6.8 (0.26) Business 18 31,6% 28% 55.28 (1.34) 6.56 (0.18) Drop out 21 36,8% 28% 57 48% 0.11 (2.07) 0.23 (0.32)
Standard errors in parentheses. T-test in last row of each block, significance levels are indicated with *α=0.1, **α=0.05, ***0.01. Note: total of 119 observations.
Table 5. Simulation 3: Cohort 2010, controlled for geographic background 100 kilometers around Amsterdam Choice ex
ante Choice ex post Amount % of track % of track HB Average Credits after year 1 Average GPA after year 1
Economics Economics 46 47% 57% 54.39 (1.45) 6.6 (0.14) Business 5 5% 5% 54 (4.14) 6.36 (0.42) Drop out 47 48% 38% 98 65% 0.39 (1.86) 0.24(0.19) Business Economics 25 47% 43% 55.52 (1.21) 6.61 (0.22) Business 17 32% 28% 55.65 (1.21) 6.84 (0.21) Drop out 11 21% 28% 53 35% -0.13(0.38) -.23 (0.07)
Standard errors in parentheses. T-test in last row of each block, significance levels are indicated with *α=0.1, **α=0.05, ***0.01. Note: Total of 151 observations.
