Live as you learn : are students of economics (getting) closer to the homo economicus?

College of Economics and Business

Live as you learn: are students of

economics (getting) closer to the homo

economicus?

Karin Fura, 10416064

Bachelor Thesis

Semester 2, 2014/2015

Supervisor: Dr A.S. Booij

Acknowledgement

I am grateful to Dr J.F.L.H. van den Broek, ResearchNed, for being granted access to

the dataset Studentenmonitor 2001-2014 via the Data Archiving and Networked

Services (DANS) archives. Without permission to use it, conducting this research

would not have been possible.

Statement of Originality

This document is written by Student Karin Fura who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

Abstract

This research sheds light on the borrowing behaviour of students of economics (economists) and the extent to which they make use of the favourable conditions of the student financial aid system. According to neoclassical economic theory, a rational individual should borrow the maximum amount possible because doing so allows for a higher life-time income and a higher utility. The evidence found shows that economists overall are less likely to borrow than other students but that the amounts borrowed are higher on the intensive margin. The reasoning for the borrowing decision is in line with the standard economic prediction with the exception of working, since economists seem to substitute borrowing with working to some extent. This may however be a conscious choice reflecting a preference, rather than a result of a cognitive limitation. (JEL D83, I22, I28)

Table of Contents

1. Introduction ... 1

2. Background ... 2

2.1 The student finance system ... 3

2.2 Rational choices according to neoclassical economics ... 4

2.3 Previous research on student debt... 6

2.4 Previous research on how economists differ ... 9

3. Methodology and Data ... 9

3.1 Method ... 9

3.2 Data ... 11

4. Empirical results ... 13

4.1 Borrowing behaviour of economists ... 13

4.2 Difference-in-difference comparison ... 13

4.3 Reasoning behind borrowing decisions ... 14

4.4 Additional tests on working behaviour ... 16

5. Limitations ... 17

6. Discussion and conclusion ... 18

7. References ... 20

1

1. Introduction

The student finance system in the Netherlands entitles higher vocational students to financial support. The total funds that they can obtain amount to over €1,000 per month and are accessible to anyone, without conditions such as a sufficient credit rating. The interest rates for the loans are usually far below the market rates (DUO, 2015a). Despite the favourable conditions, the take-up rates are low as a little over a third of the available funds is used (Booij, Leuven & Oosterbeek, 2011). The Dutch Ministry of Education, Culture and Science (OCW) considers this situation undesirable and would prefer students taking loans and focussing on finishing their studies as soon as possible so they can start working and paying back the borrowed amounts (Oosterbeek & Van den Broek, 2008).

The student finance system is due to change from the start of the academic year 2015/2016. Students will not be entitled to a basic grant anymore, but the entire amount that the state provides will have to be paid back (DUO, 2015b). Students will thus be more likely to be required to take on a loan and the magnitude of the loan is likely to increase. In a paper by Callender and Jackson, a similar change in education policy which was conducted in the UK and that led to students being required to borrow more to sustain their living costs is discussed. According to the findings, prospective students from low social classes are deterred from obtaining higher education due to the debt that is connected with it, but it has no effect on prospective students from other classes (Callender & Jackson, 2005). The Dutch Education council (Onderwijsraad, 2009) has equal opportunity as a primary objective which is a second reason why the analysis of borrowing behaviour is relevant.

In view of the upcoming policy change and the desired outcome of having students completing their studies within the designated time frame, it is interesting to investigate possible reasons as to why students refrain from borrowing. A homo economicus, being an individual who is perfectly rational according to economic theory and who aims to maximize her utility and income, would take advantage of the low interest rates of the student loan and complete her studies as soon as possible. Preventing a study delay of one year can prevent an indirect cost of a foregone year salary. The aim of this study is to determine whether students of economics, hereafter economists, make use of the knowledge of neoclassical economic theory that they gain and apply it in practice. The argument is that their information and cognitive ability constraints are lifted during their studies and that this should lead them to approach the behaviour of the homo economicus.

More specifically, the results of a yearly national survey are analysed to compare the borrowing behaviour of economists with that of other students. The paper contributes to existing literature in several ways. The combination of analysing empirical results, borrowing behaviour and the difference between economists and other students is unique. In contrast

2

to an experimental approach on student loan behaviour which has been used before (Booij, Leuven & Oosterbeek, 2011; Eckel et. al, 2007), the fact that these observations are empirical adds to the external validity. The focus on students of economics enables the investigation of the effects caused by lifting information and cognitive constraints. While the students are not randomly selected to the study programmes, the questionnaire includes some qualitative questions where students make the reasoning behind their choices known which makes it possible to draw some inferences concerning causality. Also, the data cover a time-span of eleven years and a large sample of 67,512 relevant responses that provide a representation of the Dutch university student population.

The results show that some groups of economists do make significantly more use of the available funding, but not all. In accordance with a previous study which is aimed at investigating whether economists are “made or born” (Carter & Irons, 1981), the evidence found points towards that they are intrinsically different at the start of their studies. There is insufficient proof for that they change their borrowing as a result of their education, but compared to other students on the intensive margin they borrow greater amounts. The qualitative results also indicate that the reasoning behind their borrowing decisions is closer to what would be predicted by a neoclassical economic framework. This is with the exception of their working behaviour: they work more and study less than students of other fields which can be considered to be suboptimal. However, some attention is brought to that they may anticipate that they face an industry in which working experience is important after they have graduated, since they prioritize working over achieving the best possible study results. If this is a reflection of taste, they may still be acting in accordance with standard economic theory.

The structure of the paper is as follows. In section two, the theoretical predictions are discussed in detail and existing literature is reviewed. Section three contains information about the statistical methods and the data, whereas the empirical results are included in section four. This is followed by taking the limitations into consideration in section five, and the paper is finalized with a discussion and conclusion in section six.

2. Background

This chapter provides a theoretical background and predictions of the results. To begin with, the conditions and entitlements of the available funding will be described. Next, the reasons for why borrowing is rational according to neoclassical economic theory will be assessed, followed by reasons that cause deviations from optimal behaviour. Lastly, previous research on borrowing behaviour among students and differences between economists and other students will be discussed.

3

2.1 The student finance system

According to the governmental institution DUO which is responsible for the provision of student financial aid, those eligible are entitled to four components: a basic grant, a supplementary grant which depends on parents’ income levels, a loan for tuition fees and a general loan. In the academic year of 2014/2015 the total funds available amounted to €1013, out of which a maximum of €727 represented a definite loan (DUO, 2015c).

The funds are available to all higher vocational students who request them before the age of 30 and no credit rating is required. The transaction cost for making adjustments is low since after the initial application has been processed, the composition of the grant and loan can be adjusted by the student on a monthly basis using an online tool. The grant is available for a period of three years for an academic Bachelor and one year for an academic Master. While this matches the nominal time that is designated for most programmes, the loans can be provided for an additional three years. Students who do not obtain their diploma within ten years will be required to pay the grant back with the same interest rates and within the same time frame as those that apply for the loans (DUO, 2015e).

The interest rates for active users of the funding can be adjusted by DUO on a monthly basis, while those that apply after graduation are determined for a five year period at a time. The rates have followed a cyclical pattern since 2003, ranging from 0.12 to 1.5 percent on a yearly basis. The entire amount needs be paid back within 15 years after graduation and there is a maximum amount set that is due every month, being twelve percent of what is earned by an individual in addition to the applicable minimum wage. Additionally, it is possible to temporarily halt the payments due to personal circumstances. Lastly, someone who has yearly earnings below a set amount and who has not obtained their diploma can be exempted from paying the loan back (DUO, 2015a). As mentioned in the introduction, the interest rates are far below the market rates and the finance system offers insurances for unforeseen circumstances. These are two reasons that should lower the threshold for risk averse individuals to take on the loan.

The current conditions and amounts apply until the revision of the financing system, which will take place in conjunction with the start of the academic year of 2015/2016. From then on, the conditions will change for all students who start a new education programme. In short, there will be no basic grant, the loan needs to be paid back within 35 instead of 15 years and unlike until then, there will be no maximum amount that can be earned as salary. While this is of limited relevance for the current students, it may change the borrowing patterns in future years if students want to avoid debt. This possibility will be discussed more in detail in the next section.

In summary, there are enough funds that are offered to students in order for them to sustain a living without working on the side of their studies. The interest rates are favourable

4

and there are insurances available for situations in which a student may not be able to pay the loan back, for instance due to low earnings or not obtaining the diploma. More details on the funds and conditions are included in the appendix.

2.2 Rational choices according to neoclassical economics

Neoclassical economics is based on three fundamental assumptions: firms maximize profits and individuals maximize utility, individuals have rational preferences over outcomes, and agents act independently and based on complete information (Weintraub, n.d.). Concerning the conditions under which decisions are made, it is assumed that there are no time constraints for the decision making and that agents are not restrained by cognitive limitations that prevent them from making optimal decisions, such as risk aversion or preferences which are inconsistent with time (Arrow, 1989). Some implications of these assumptions are discussed next.

Individuals’ utility functions are increasing in their earnings, or rather the commodities that can be purchased with them (Marshall, 1920). Based on their preferences, individuals maximize their utility, subject to the applicable constraints (Becker, 1962). Hence if the objective is to maximize utility, the objective should be to maximize income. A student who faces the possibility of taking an additional year to complete his study ceteris paribus should avoid doing so if the alternative is to work. This is because working directly generates income, while studying does not immediately increase personal wealth but causes an opportunity cost in the foregone wage. Therefore, a student who takes longer to complete the programme is lowering his income and his utility.

Additionally, in accordance with the life-cycle hypothesis mentioned in Davies and Lea, individuals should smooth their lifetime income which means that someone who is temporarily below her level of permanent income should borrow money to be able to increase her consumption. Students have a relatively low income with higher earnings prospects which is thus another reason that they should make use of the funding (Davies & Lea, 1995).

The optimal amount that should be borrowed depends on future income and expected expenditures, as well as current expenditures and income. This is based on the assumption that agents’ decisions are consistent and that they are forward-looking, so they maximize their utility not just for the short-term but also with a view of the optimal decision in the long-term (Ando & Modigliani, 1963). Furthermore, the applicable interest rate and the subjective discount rate are relevant. The market interest rate should be used to discount cash flows and any investment opportunity which could yield a higher payoff can increase consumption, meaning that the option should be utilized (Loewenstein & Thaler, 1989). In case there is uncertainty, risk attitudes will also play a role in the decision to borrow or not (Oosterbeek &

5

Van den Broek, 2009). To enable consumption smoothing over time, there cannot be liquidity constraints. On the market for loans there may be barriers to borrow as a credit rating may be required before an applicant is approved to take a loan, or it could have implications for the interest rate. Theoretically, these barriers should not apply for students since a credit rating for state-provided aid is not a part of the procedure for student loans. Carneiro and Heckman (2002) investigate empirically if there is a credit constraint applicable for US students. They conclude that such a constraint affects about four percent of the American youth but that it is eliminated in the long term, adjusting for other factors. Hence liquidity constraints should not impose notable barriers for consumption smoothing.

As has been made clear above, an individual focussing on maximizing life time earnings would make use of the student financial aid. Not only is it beneficial to take a loan to be able to finish the studies as soon as possible, but for many years it has been possible to make a profit on the loan. This is because the interest rate on the student loan has been lower than the market savings interest rates so that riskless profits could be made by borrowing at the lower rate and placing the amount in a savings account. Furthermore, someone who is planning on taking out a commercial loan at market prices later in life, for example for a mortgage, could use an amount that is borrowed via the student finance system as a part of the down payment. The interest rates applicable for commercial loans are higher than the interest rates for the student loans and the difference is larger as compared to the market savings rates (retrieved from DNB, 2015). In summary, according to neoclassical theory, there are several reasons why students who do not make use of the available funding are not acting in their best self-interest. Yet the empirical evidence shows that students do not behave according to the optimal predictions. In what follows, some reasons as to why that could be are discussed.

First of all, while information about the entitlements and conditions of the loans are easily accessible to the public, students may be unaware of how to find it or have limited cognitive ability to process it. Second of all, there may be emotions involved such as an overestimation of the transaction cost necessary to request the funds, or a general debt aversion. Loewenstein and Thaler investigate intertemporal choices and why individuals may deviate from optimal economic behaviour. For instance, they mention that what may be referred to as the iteration level related to a product a plays a role. People tend to realize that an investment in insulation will save electricity cost, likely because there is a direct connection between the product, insulation, and the desirable outcome, which is to save money. An equally valuable energy-saving refrigerator tends to be valued lower however, seemingly because its primary purpose is not to save costs on electricity (Loewenstein & Thaler, 1989). Similarly, it may be that students do not make optimal use of the funding that

6

is available because they are incapable of making the connection between the debt and the indirect cost of a foregone year salary.

Additionally, the authors discuss variations in internal discount rates. Discounting seems to be inconsistent over time so that preferences are reversed, which contradicts the standard economic assumption of transitivity. First, they mention that decisions are different depending on if the amount is going to be lost or gained. Generally, the absolute value of the utility of gaining an amount is perceived as smaller than the disutility of losing an equivalent amount (Loewenstein & Thaler, 1989). A possible explanation as to why students avoid taking on the debt is that they consider the amount that they can utilize from receiving the extra cash to be worth less than what is needed to compensate for their increasingly negative account balance. In absolute terms and direct cost this is also the case due to the interest rate. If the amount is re-invested, however, it can yield a higher payoff and the direct gain will be positive.

Loewenstein and Thaler use another example related to debt aversion, which is the fact that people tend to pay off mortgages faster than what is necessary or avoid to take on debt, even though they could earn a higher rate on safe investments because they prefer that their utility increases over time. The authors conduct an experiment where subjects can choose between three different payment profiles. The undiscounted hypothetical wages offered are identical, but the developments of the periodical wage payments differ. In one the payments increase, in the next they remain constant and in the third they decrease. Despite the economically optimal one being the one with payments that decrease over time, based on a positive interest rate and net present value, only twelve percent of the subjects prefer that profile. This last result can erroneously be considered an anomaly and the result of a negative rate of time preference in a neoclassical framework. However, if the behaviour is simply a reflection of taste, it is still in accordance with the theory since avoiding debt may increase the utility of an individual (Loewenstein & Thaler, 1989). The same principle would be applicable if students anticipate that they are unable to use the funds in a disciplined manner and therefore prefer to intentionally limit the amounts that they have at their disposal.

2.3 Previous research on student debt

In their 2011 paper, Booij, Leuven and Oosterbeek study the impact of increased information provision on the study financing system using an experimental approach. Their motivation is that usage of the facilities is very limited since 35 percent of the available funds are taken up. At the same time, many students spend more than the nominal number of years to obtain their degrees. It would not just be in the government’s interest to limit the study duration to the nominal amount but also in the interest of the students. Rather than working during the studies and taking one or two years longer to complete the programme, since this entails

7

foregoing one or two year salaries estimated at €45.000 each, students could take out a loan and complete the study faster.

According to Booij, Leuven and Oosterbeek, there is evidence that students do not make use of the available funding because they are not aware of the possibility and rules, as well as that they are prevented by the transaction cost of going through the application process. They test whether the provision of information surrounding the loan conditions, such as interest rates and the repayment period, will increase take-up rates. A survey is conducted a priori in order to investigate the correlation between take-up rates and knowledge concerning the conditions. The results reveal a positive relationship, which could indicate that informational campaigns may increase the use of the funding. However, there could be reversed causality if students look up information concerning the loans as the need for extra funding arises.

While the experiment that they conduct is intended to provide more conclusive evidence in this respect, its outcome is not straightforward. The experiment entails that a treatment group of randomly selected students receives information about student financial aid while a control group of equivalent students does not. Six months after the information is distributed, students are prompted to fill in a questionnaire designed to test how much they borrow and whether they remember the information. The authors do this in order to establish whether there is a causal relationship between a lifted information constraint and increased take-up rates. The results show an increase by half a percentage point in the treatment group and that the difference is insignificant. Booij, Leuven and Oosterbeek conclude that the provision of information does increase knowledge of the loan conditions but that it does not increase the take-up rates significantly. Debt aversion and cognitive constraints are considered as the remaining possible explanations (Booij, Leuven & Oosterbeek, 2011).

Oosterbeek and Van den Broek (2008) also investigate borrowing behaviour of higher vocational students in the Netherlands. Their focus lies primarily on the relationships between debt attitudes and take-up of the loan. In one study that is discussed in their paper, conducted by Davies and Lea (1995), the intention is to determine whether debt attitudes affect having debt, or if having debt influences debt attitudes. They find that first and second year students have similar debt attitudes while second year students’ debts are substantially higher. Third year students are less debt averse than second year students and the authors therefore conclude that it is a change in debt which causes the change in debt attitude. They argue that students are not initially in favour of debt but change their attitudes during their studies as they are required to borrow in order to achieve an acceptable lifestyle (Davies & Lea, 1995). However, Oosterbeek and Van den Broek suggest that there could be selective attrition causing these results to be biased. According to them, students who do not borrow

8

may be more likely to have a job to finance their study, which in turn could make them more likely to drop out.

Another experimental study is conducted by Eckel et. al. (2007) to compare the preferences for two options: a cash alternative versus an education funding alternative. Surveys are used to measure the participants’ risk aversion, debt attitudes and time preferences. The results show that more risk seeking as well as more patient individuals are more likely to choose the education subsidies over cash. Additionally, individuals with a higher level of debt aversion are equally likely to take on subsidized loans as are those with a lower amount of debt aversion, and students who already have a large amount of debt are more likely to make use of the loan scheme. The author’s interpretation of this result is that debt aversion does not impose a barrier to investments in higher education (Eckel et. al., 2007 in Oosterbeek & Van den Broek, 2008). Oosterbeek and Van den Broek disagree based on the argument that not having debt could also be used as a measure of debt aversion and then the conclusion would be the opposite.

The research which they conduct entails analysing the observations from a survey among higher education students in the Netherlands. A factor considered for lowering the take-up rates is debt aversion and the outcomes are in line with this possibility. While standard economic models do not include debt aversion, behavioural economic models indicate that individuals overvalue their debt in the sense that they perceive it as more expensive than it is. In the model that Oosterbeek and Van den Broek estimate, which includes a variable for debt aversion, they find that there is a significant negative causal relationship between debt aversion and borrowing behaviour.

Based on the assumption that economists become equipped with a more complete financial toolkit during their studies compared to other students, the viewpoint of this paper is that their information and cognitive constraints are lifted. As they get educated in the ways to optimize their incomes and thus their utilities, they should approach the rational behaviour of the homo economicus and hence make more use of the student financial aid system later in their studies compared to students of other fields. They gain information on how to maximize their wealth, the consequences of differing interest rates and arbitrage opportunities, the neoclassical economic theories and situations in which individuals are prone to make decisions with limited rationality. The hypothesis is therefore that the utilization of the student financial aid system should be positively associated with studying economics. This could however be caused by two underlying reasons, self-selection or a learning effect. Some research on similar topics has already been done and will be discussed next.

9

2.4 Previous research on how economists differ

Marwell and Ames (1981) conduct an experiment in which they test the free rider hypothesis empirically, using a public goods game. The essence of this game is that all players would be best off if all other players invest in a common good, but it is in each individual’s interest to keep his own amount and thus free ride on other player’s contributions. During the experiment, the strategies of economics graduate students are analysed separately, in comparison to a control group of students of other fields. The outcome displays that economists are different. The average contribution made is just 20 percent of the available funds and economists are significantly more likely to free ride than students in the control group. Two possible explanations are offered for their diverging behaviour: there may be self-selection by individuals who are especially interested in economic incentives to education programmes which deal with such topics. Alternatively, the behaviour of economics students may change as they learn about the axioms of the economic theory (Marwell & Ames, 1981 in Carter & Irons, 1991).

Carter and Irons conduct an experiment to test for both selection effects and learning effects. They use a simple ultimatum bargaining game in which a proposer has an amount to split with another player, who can accept or reject the offer. They find that economists behave more closely to the rationally optimal prediction, in which the proposer offers an incremental amount and the second player accepts anything above zero. In their regressions, freshman economists already behave differently as compared to freshman non-economists. This is an indication of self-selection. When they control for the effects of senior economists the differences do not widen, so they conclude that they find no evidence for the hypothesis that economists learn and adjust their behaviour as a result thereof (Carter & Irons, 1991). In other words, the objective of their paper is similar to the one of this research. While the approach in this paper is empirical and the time-span studied is longer, the intention to determine whether economists are different and if so whether this is due to self-selection or due to a treatment effect underlies both papers.

3. Methodology and Data

In what follows, analyses and statistical tests are conducted to determine the ways in which economists differ compared to other students, and whether this is likely due to self-selection or a learning effect. The data come from the national student monitor survey (Studentenmonitor), which is commissioned by the OCW on a yearly basis. It is conducted by ResearchNed, the institution which provided the data (ResearchNed, 2015a). The survey is intended to provide an appropriate representation of the Dutch student population. Due to

10

overrepresentation among certain respondents, sample weights are used for factors of study year, study sector, type of education (university or university of applied sciences) and gender (ResearchNed, 2015b). The distinction between study programmes is made by means of a coding system (CROHO) which is determined by the authorities. The CROHO code for economics includes all studies in economic fields, such as business economics, business administration, fiscal economics, management and actuarial sciences, and similar (DUO, 2015d).

Based on the data, three questions are analysed in particular. Firstly, whether economists borrow significantly more than other students. This question is investigated both on the intensive and on the extensive margin. If they are different in this aspect, this captures both a causal and self-selection effect. Secondly, in an attempt to isolate a causal relationship underlying the behaviour, a difference-in-difference approach is used to test whether economists’ borrowing changes in a different way compared to that of students of other fields. This question too is analysed on the extensive and on the intensive margin. For this part of the analysis, third year Bachelor students and first year Master students of economics are used as the two treatment groups while students of other fields are the control group. With this approach it is possible to determine what part of a difference between students should be accounted to the learning effect, and what part to the selection effect.

Thirdly, the motives for borrowing or not are compared between the economists and students of other fields. The survey includes qualitative questions concerning why students choose (not) to borrow. Based on this, it can be inferred whether economics students have significantly different motives for their pro- or anti-borrowing attitudes.

In the first question, the fraction of borrowers among economists versus students of other academic fields is compared. Also, a test is conducted to determine whether there is a difference in the mean amount borrowed among the students on the intensive margin. Based on the fact that the standard deviation of the population is unknown and on the central limit theorem, which states that the sampling distribution of the mean of a sufficiently large random sample may be assumed to be normal (Keller, 2012), the first questions are estimated by means of a t-test under the assumption of unequal variances.

The second question using the difference-in-difference approach is dealt with by conducting two tests per treatment group. This allows for the possibility to compare the developments on the extensive and on the intensive margins, throughout the studies. The principle behind the model which is estimated can be described as follows:

Yit= β1+ β2 Economics + β3 Study stage + β4 (Economics * Study stage) + εit

where Yit represents the fraction of borrowers or the borrowed amount in the first and the second regression respectively. The dummy variable “Economics” is 1 for students of

11

economics and 0 otherwise. Similary, the variable “Study stage” is 1 for third year Bachelor students in the regression for Bachelor students and 0 otherwise, while the equivalent specification applies to the regression for Master students.

β

“Economics” captures the selection effect, being the intrinsic difference between economists and other students from the start of the programme. The coefficient on the interaction term,

β

, represents the causal effect from following an economic education.

β

is the constant

and the error term

ε

is zero on average by assumption. Two

per study stage: the first to estimate the effect of economics on the probability of borrowing, and the second to estimate the effect on the mean amount given borrowing. The regressions are executed with weighted observations.

The third question which is used to investigate the reasons why economists borrow or not, is answered by means of some of the qualitative questions that are included in the questionnaires from 2004 onwards. In these questions, the scale is ordinal and reaches from 1, which represents “unimportant” to 5, meaning “very important”. The inferences comparing the two student populations are made based on a Wilcoxon rank sum test, since according to Keller (2012) this is the appropriate way to treat ordinal data.

Lastly, due to the results from the three main questions, a fourth area is investigated in an attempt to clarify the priorities of economists who work. This is to get more information on whether they make choices which are in line with neoclassical economic theory, due to tastes or other considerations. The analysis is made using t-tests for the numerical values and Wilcoxon rank sum test for the ordinal values.

3.2 Data

The data set includes the results of the national student monitor from the period 2001 to 2014 (n = 154,527), with the exception of 2010 when no survey was conducted. Because financial information was added to the survey in 2003, the observations from 2001 and 2002 are excluded. To allow for the analysis of a group as homogeneous as possible, the scope is limited to the observations for full time university students (n = 90,998). Additionally, when one financial variable is invalid, for instance missing or outside of the possible range, then all financial variables are coded as such (ResearchNed, 2015b). These students’ responses have also been removed, leaving n = 67,512 observations with a usable value for the variable of interest. This is “income study loan” which captures the amount that students receive as a loan on a monthly basis. In the years until/including 2009 this amount was reported via the registry of the DUO, and since 2011 onwards the amounts are self-reported by respondents.

12

During the period which is studied, the educational system changed from one in which students followed one programme lasting up to five years nominally (doctoraal), to a separated bachelor-master system. Because the majority of students follow a new style programme, the observations for students of the old style education have been recoded and included in the new style variables. The dummy variable BSc for Bachelor students includes the three first study years of the old system as well as the new system, while the dummy MSc for Master students includes the years four and five of the old system and year one and two of the Master programmes in the new system since it is a separate education programme. Students in their sixth year of the old style education are excluded. The variables of the new style education include the possibility that participants of the new Bachelor programme are in their fourth year, which entails a study delay. In order to prevent that the equivalent of nominal Master students of the old system are mixed with those who have a study delay, these fourth year Bachelor students are specified separately.

Table 1. Descriptive statistics

Fraction of borrowers Mean amount borrowed

Economics Other studies p-value Economics Other studies p-value

First year BSc 0.225 0.237 0.190 346.21 330.99 0.127 (0.013) (0.004) (12.69) (4.07) Second year BSc 0.237 0.264 0.038 356.58 329.65 0.023 (0.014) (0.005) (12.83) (4.01) Third year BSc 0.345 0.325 0.087 404.66 363.50 < 0.001 (0.014) (0.005) (11.48) (3.74) Fourth year BSc* 0.402 0.429 0.233 406.24 363.93 0.065 (0.035) (0.012) (26.47) (8.00) First year MSc 0.376 0.362 0.170 469.87 417.01 < 0.001 (0.013) (0.005) (11.54) (3.98) Second year MSc 0.421 0.407 0.311 505.38 433.63 < 0.001 (0.026) (0.006) (21.29) (4.81)

Standard errors are shown within parentheses

* Only includes students of the new educational system Mean amounts in euros

Table 1 shows an overview of the fraction of borrowers and the mean amount that these students borrow, per study year. Economics is specified by a dummy variable that depends on the CROHO codes, and the p-values are based on one-sided t-tests. The fraction of borrowers is higher among third year Bachelor economists than other third year students and the difference is weakly significant (p = 0.087). This pattern does not appear in other years since on the contrary, the fraction of borrowers in other study years of the Bachelor is lower

13

among economists and in the second year significantly so (p = 0.038). While the fractions are higher in the Master, the difference for both years is insignificant. In all study years though, the amounts borrowed by economists compared to other students and from the second year onwards (weakly) significantly so. In the third year of the Bachelor, as well as in the first and the second year of the Master, there is strong evidence for that economics students borrow a larger amount on the intensive margin (p < 0.001).

4. Empirical results

4.1 Borrowing behaviour of economists

First of all, when comparing the proportions of those who borrow between economics and other fields of study overall, the outcome contradicts the prediction. Rather than the fraction of borrowers being greater among economists, it is smaller at a significance level of α = 0.1. The estimated fraction for economics is 0.315 (s.e. 0.006) and the fraction for other studies is 0.323 (s.e. 0.002). On the other hand, looking at the amounts on the intensive margin, the result is in accordance with the prediction that economics students should borrow more. The average amount borrowed among economists is 437.7 (s.e. 5.2) while for other academic fields it is 388.4 (s.e. 1.6) and the difference is highly significant (p < 0.001). This is largely in line with the picture that is shown in the descriptive statistics, looking at the division per year. The fact that the difference in the amount borrowed spreads in the later study years may be evidence for the hypothesis that students adjust their behaviour as they are exposed to the standard economic theories. Making the distinction between a selection and a causal effect more clear will be dealt with next.

4.2 Difference-in-difference comparison

The second objective is to conduct a difference-in-difference analysis of how the utilization of the student finance system develops during the studies. The two treatment groups of third year Bachelor students and first year Master students of economics are compared with the control group of students of other fields. This way it is possible to distinguish whether economists are different due to a selection effect or due to a learning effect, i.e. whether the differences are captured by the coefficient on “Economics” or by the coefficient on the interaction variable. In both regressions, the independent variables are the dummies for economics, for being in the third year of the Bachelor, as well as an interaction variable between these two. The equivalent specification applies for the second treatment group of first year Master students. The results of the regressions, that were run with weighted observations, are shown in Table 2 below.

14

Table 2. Difference-in-difference regression of borrowers and amounts, per stage of study

Fraction of borrowers Mean amount borrowed

(1) (2) (1) (2) Economics -0.035*** -0.036*** 34.5*** 28.9*** (0.01) (0.01) (6.2) (6.2) Third year BSc 0.025*** -25.3*** (0.01) (5.4) Economics*BSc 0.028 -6.7 (0.02) (13.5) First year MSc 0.050*** 43.9*** (0.01) (5.8) Economics*MSc 0.033 6.8 (0.02) (13.9) Constant 0.338*** 0.335*** 405.7*** 393.9*** (0.00) (0.00) (2.4) (2.3)

See section 3.1 for the specification of the model Standard errors are shown within parentheses Amounts in euros

Significance levels 0.1, 0.05 and < 0.001 are displayed as *, ** and ***

Economists are less likely to borrow than other students are. For both Bachelor and Master students, the fractions of borrowers are significantly lower (p < 0.001). The effect of the interaction term on the fraction of borrowers is positive yet insignificant for Bachelor students as well as for Master students. The effect on the mean amount on the intensive margin is however different between the groups. For Bachelor students it is negative and insignificant, but for Master students the effect is positive and insignificant. The variable for economics is highly significant and approximately €35 for Bachelor students and €29 for Master students in the two regressions on the mean amount borrowed (p < 0.001). In other words, there is insufficient evidence to infer that students change their borrowing behaviour according to neoclassical economic theory as a result of their studies. There is however strong evidence for that economists are different in that they are less likely to borrow, but do borrow a greater amount on the intensive margin.

4.3 Reasoning behind borrowing decisions

The third question is whether the motives and reasoning behind the borrowing decisions are different among economists compared to students of different programmes. Specifically, four variables are investigated in detail, for borrowers and non-borrowers respectively. The statements were only posed to students for which they were relevant, so only those who had indicated that they borrow were asked about their motives for doing so. An English

15

translation of the statements is included in the appendix. According to the theory, economists should score higher in all of the following categories. The results of the responses together with a brief description as to why are shown below. For all results in this section, necon =

2,094 and nother = 19,331 and the z- and p-values are based on Wilcoxon rank sum tests.

• The loan conditions are beneficial: based on the assumption made in the beginning of the paper, economists should be better informed of the loan conditions and aware of the profits that could be made or at least losses that could be avoided. Their answers are consistent with this assumption: they do score higher (z = -5.29, p < 0.001). • To afford more luxury: due to consumption smoothing, making consumption and

investment decisions based on their life-time income rather than their short term income, economists should spend relatively more per month as compared to others if they behave more rationally from a neoclassical economic perspective. Consistent with this prediction, they do have a higher score (z = -8.83, p < 0.001)

• Certain or sufficient future earnings: in connection with the above, if they expect high and/or stable incomes in the future, they have reason to make investments in the short term which are based on incomes that they will earn in the long term. This outcome too is in line with the theoretical reasoning and they score higher (z = -12.68, p < 0.001).

• To have to work less: they should consider it important to work less, make more room available for their studies and increase the likelihood of graduating within the time designated rather than getting a study delay. The outcome is however the contrary: they have a lower score (z = 7.73, p < 0.001).

Next, the question is whether the motives for not borrowing are different among economists as compared to students of other programmes. Again, four of the questions are tested in the same manner. In contrast to the tests above, the theoretical prediction is that economists should score lower in all of these categories.

• Study loan is too expensive: an informed economist should be aware of that the interest rates are favourable in comparison to the market interest rates. While there are direct costs applicable due to the interest rates when taking a loan (and transaction costs for the application), doing so enables making investments with a greater yield. Furthermore, the loan can be used to avoid having incurring the opportunity cost of a missed year salary. Therefore economists should find the conditions more attractive than other students. They do, as they score significantly lower (necon = 3,852, nother = 34,882, z = 9.82, p < 0.001).

• Avoid getting (large) debts: according to standard economic theory, a rational individual should not per definition have an aversion for debt and that is what this

16

question captures. In line with the prediction, the score is indeed lower (necon = 3,855,

nother = 34,900, z = 5.17, p < 0.001).

• Loan conditions DUO not known: as mentioned above, the assumption is that economists are better informed of profit opportunities and the beneficial conditions. Here too, the outcome is a lower score which is consistent with the prediction (necon =

3,851, nother = 34,890, z = 2.70, p = 0.007).

• Working for a salary: the primary objective not to work should be in order to avoid missing out on a year salary. However, here as in the motivations concerning why they do borrow, economists indicate to a greater extent that they focus on working rather than borrowing. In contrast to the theoretical prediction, they score higher than other students (necon = 3,852, nother = 34,900, z = -5.74, p < 0.001).

In sum, the evidence provided by the qualitative reasoning is inconclusive. In most aspects the economists seem to base their decisions on arguments which are in accordance with the theory but their behaviour when it comes to work is different. This will be discussed more in detail below.

4.4 Additional tests on working behaviour

From the results in question three, in which the qualitative reasoning of borrowers is analysed, it is clear that economists reason according to what the theory suggests aside from when it comes to working for pay. Therefore, additional tests are run in order to investigate the reasons of their working and studying behaviour. Using t-tests, the weekly amounts of hours that they spend on studying and working respectively are investigated. Economists work more during their studies: the fraction of workers is 0.695 as compared to 0.642 (p < 0.001). The mean amount of hours worked on the intensive margin is also higher at 11.5 against 9.7 hours (p < 0.001). Furthermore, economists spend less hours studying than students of other fields: 24.1 as opposed to 28.2 hours per week (p < 0.001).

Their reasoning behind working also seems slightly different compared to other students (as before, translations of the statements are included in the appendix). The work that they do is related to their studies to a greater extent and highly significantly so (z = -7.23, p < 0.001). Also, economists find getting experience more important than other students do (z = -8.23, p < 0.001) and they work to increase their chances of a job significantly more (z = -8.76, p < 0.001). For all tests above necon = 6800, nother = 60,712.

Lastly, it is interesting to see whether they sacrifice study results for work: they do. The fraction that indicates that their study results are lower due to working is 0.170 (s.e. 0.008) among economists and 0.132 (s.e. 0.002) among other students so the difference is highly significant (p < 0.001).

17

5. Limitations

There are three limitations to this research. The first concerns financial variables which were reported by the DUO in the first years of the survey while they were self-reported in the later years. The second concerns the definition of the study programmes, since the category economics also includes other related fields. The third relates to the timing of the survey, being the end of the academic year. Each will be discussed below.

Because the financial variables, more specifically “income study loan”, are reported directly from DUO in 2003 to 2009 while they are self-reported in 2011 to 2014, the likelihood of misreporting in the later years may be greater. One could argue that in accordance with the theory if of this paper, economists are more skilled at keeping track of their finances and that the amounts they report therefore are more accurate on average than for other students. There is however little reason to believe that there is systematic over- or underreporting and hence bias in the estimates comparing economists and others. Additionally, if there is, most of these differences are eliminated in the difference-in-difference regression analysis. For that reason, and to increase the power of the tests, it was decided to include the whole period of 2003 to 2014 in the research.

While the fact that the results for economic studies also includes related studies may introduce a measurement error, the percentage of students of general economics can be assumed to be much higher in this category than in other fields. Furthermore, programmes in closely related fields usually include courses in microeconomics and finance. Therefore, the choices by students who are not taking economics as their major should be similar to those studying general economics so they can be included in the treatment group. This was confirmed by checking the coursework of a sample of programmes from five different and universities within the Netherlands. Hence it can reasonably be assumed that the majority of students who are taking a major coded as economics in the CROHO registry at a Dutch university are educated in general economics.

Because the survey is conducted at the end of the academic year, students of all stages have likely already had some coursework in economics when they fill it out. It could therefore be that the differences among first-year students also include the effect of lifting the information and cognitive constraints, rather than just self-selection. This would be the case if the differences between the groups have already arisen before the survey is conducted. However, many students might have had the courses on economic theory late in the year or need time to change their borrowing decision and executing the adjustment via DUO. For that reason the timing of the survey should not have considerable implications on the results.

18

6. Discussion and conclusion

An objective of the Dutch authorities is to trigger students to make use of the student finance system and to avoid taking extra time to obtain their diplomas. The research in this paper is conducted in an attempt to find reasons why take-up rates are low. As suggested by Booij, Leuven and Oosterbeek (2011) possible explanations include information constraints and debt aversion. What is tested in this study is whether lifting the information and cognitive constraints, which occurs to students of economics as they study the theories of neoclassical economics, causes them to make more use of the available funding than students of other fields. By analysing how the borrowing behaviour and reasoning behind the decisions differs among economists compared to other students, inferences can be made concerning the possible causes of the low take-up rates.

In line with previous studies (Marwell & Ames, 1981; Carter & Irons, 1991) the finding is that students of economics are different: there is evidence that supports that students of economics are closer to the homo economicus even though it is not inconclusive. In contradiction to what the theory predicts, the fraction of borrowers it is not generally higher among economists. When taking into account the interaction of the study stage and the study programme in a difference-in-difference setting, there insufficient evidence suggesting that the fraction of borrowers increases, so economists do not seem to start to apply the theories which they are taught. While mean amounts borrowed do also not increase as a result of following an economic education, the study programme does have a significant and positive effect on the amount borrowed. Students of economics who do borrow, borrow amounts that are higher than other students in all stages of their studies and in most years significantly so. In other words, the results indicate self-selection rather than evidence in line with a learning hypothesis. As shown in the descriptive statistics there is also a trend which shows that the amounts that economists borrow increase to a greater extent than other students, but this difference cannot unambiguously be interpreted as a causal effect.

Based on the qualitative reasons that economists indicate as motives for their decision whether (or not) to borrow, they are significantly closer to a neoclassical economic way of thinking than other students. For instance, they consider the loan conditions to be more beneficial and they are less debt averse than other students. The exception to these results concerns their working behaviour as they work more and spend less time on studying than students of other fields. This is not necessarily optimal because of the opportunity cost equal to the salary that they forego from taking longer than the nominal time to complete the study. The results however point towards that they work because they anticipate that they are going to pursue careers in an industry where expectations of students’ working experiences are different than other students upon graduation: economists are more likely to

19

work in a related field, they work to gain experience and they substitute study results to be able to work on the side.

Hence, these results need not represent a cognitive limitation but rather a preference as students experience increased utility from working. An alternative explanation is that they intentionally maximize their life-time incomes because they make the valuation that their future salaries will increase sufficiently as a result of the work experience that they gain, making it profitable to invest a foregone year salary or two that a possible study delay entails.

Because the student finance system will soon be revised, it will be interesting to analyse the borrowing behaviour of students empirically again in a few years. Once the system has changed, students will more likely be required to borrow to obtain higher education. In one paper which was discussed, it was concluded that there is a strong correlation between debt aversion and low take-up rates (Oosterbeek & Van den Broek, 2009). Another study indicated that an increased debt precedes a decreased debt aversion (Davies & Lea, 1995). In a few years the implications of the new system can be investigated since the fact that students will be required to borrow for support may lower their debt aversion and induce them to further increase their loans voluntarily. Another suggestion for a follow-up study is to look at the extent to which employers in economic fields value working experience among graduates and the effects that it has on salaries. This information could be used to determine whether the net value of a study delay caused by extracurricular activities is positive for economists. If that is the case, policy makers could consider alternative measures to increase the incentives for students to complete their studies faster.

20

7. References

Ando, A. & Modigliani, F. (1963). The “Life Cycle” Hypothesis of saving: aggregate implications and tests. The American Economic Review, 53(1). 55-84.

Becker, G.S. (1962). Irrational behavior and economic theory. Journal of Political Economy,

70(1), 1-13.

Booij, A., Leuven, E. & Oosterbeek, H. (2011). The role of information in the take-up of student loans. Economics of Education Review, 31(1), 33-44.

Callender, C. & Jackson, J. (2005). Does the fear of debt deter students from higher education? Journal of Social Policy, 34(4), 509-540.

Carneiro, P. & Heckman, J.J. (2002). The evidence on credit constraints in post-secondary schooling (NBER Working paper No. 9055). Retrieved from http://www.nber.org /papers/w9055

Carter, J.R. & Irons, M. D. (1991). Are economists different, and if so, why? The Journal of

Economic Perspectives, 5(2), 171-177

Centraal Bureau voor de Statistiek (CBS). 2014. Gemiddeld inkomen; personen in

particuliere huishoudens naar kenmerken. Retrieved from http://statline.cbs.nl

/Statweb/

Davies, E. & Lea, S. (1995). Student attitudes to student debt. Journal of Economic

Psychology, 16, 663-679.

De Nederlandsche Bank (DNB). (2015). Rentes van MFI's op leningen en deposito's, met

bijbehorende volumes. Retrieved from https://www.statistics.dnb.nl/financieele-

instellingen/banken/binnenlandse-bankbedrijf-monetair/index.jsp

Dienst Uitvoering Onderwijs (DUO). (2015c). Het oude stelsel van studiefinanciering. Retrieved from http://duo.nl/particulieren/student-hbo-of-universiteit/Het-oude-stelsel- van-studiefinanciering.asp

Dienst Uitvoering Onderwijs (DUO). (2015e). Leenbedrag wizigen. Retrieved from https://duo.nl/particulieren/student-hbo-of-universiteit/studiefinanciering/lenen.asp

21

Dienst Uitvoering Onderwijs (DUO). (2015d). Raadplegen en downloaden CROHO. Retrieved from http://www.duo.nl/zakelijk/ho/croho/raadplegen_en_downloaden _croho.asp

Dienst Uitvoering Onderwijs (DUO). (2015a). Terugbetaler. Retrieved from http://duo.nl /particulieren/terugbetaler/een-studieschuld/rente.asp

Dienst Uitvoering Onderwijs (DUO). (2015b). Weten hoe studiefinanciering werkt. Retrieved from https://duo.nl/particulieren/student-hbo-of-universiteit

/studiefinanciering/weten-hoe-het-werkt.asp

Eckel, C.C., Johnson, C., Montmarquette, C. & Rojas, C. (2007). Debt aversion and the demand for loans for postsecondary education. Public Finance Review 35(2), 233-262.

Keller, G. (2012). Managerial Statistics (9th ed.) (pp. 342, 729) Hampshire, England: South-Western Cengage Learning

Loewenstein, G. & Thaler, R.H. (1989). Anomalies: Intertemporal choice. The Journal of

Economic Perspectives, 3(4), 181-193.

Marshall, A. (1920). Principles of Economics (8th ed.) Book III. London: Macmillan and Co., Ltd.

Marwell, G. & Ames, R. (1981). Economists free ride, does anyone else? Journal of Public

Economics, 15, 295-310

Onderwijsraad. (2009). De Onderwijsraad over toegankelijkheid en gelijke kansen, balans

van tien jaar adviezen. Retrieved from http://www.onderwijsraad.nl/upload

/documents/publicaties/volledig/de-onderwijsraad-over-toegankelijkheid-en-gelijke-kansen.pdf

Oosterbeek, H. & Van den Broek, A. (2009). An empirical analysis of borrowing behaviour of higher education students in the Netherlands. Economics of Education Review, 28(2), 170–177.

22

ResearchNed. (2015a). Studentenmonitor 2001-2014. Retrieved from http://dx.doi.org /10.17026/dans-z8h-ydyt

ResearchNed. (2015b). Onderzoeksverantwoording Studentenmonitor 2001-2014. Retrieved from https://easy.dans.knaw.nl/ui/datasets/id/easy-dataset:60492/tab/2

Weintraub, E. R. (n.d.). Neoclassical economics. The concise encyclopedia of economics. Retrieved on 10 june 2015 from http://www.econlib.org/library/Enc1/Neoclassical Economics.html

23

Appendix

The monthly amounts of the basic grant, supplementary grant, tuition fees and general loan are €286, €267, €159 and €301 respectively in 2014/2015. Generally the amounts available to students who are living with their parents are lower and students can opt for making only partial use of the two amounts that ought to be borrowed. Additionally, the supplementary grant is only available for those whose parents earn up to a set amount, as parents are expected to support their children financially to a certain extent. Students who are not entitled to this grant may borrow the difference, thus increasing their loan component to a maximum of €727.

The provision of the grants has some restrictions. For instance, students may currently earn a maximum of €13,856 per year or they lose the entitlement to the aid. All funding has to be used up within ten years. The limit for the yearly earnings in order to be exempted from paying the borrowed amount back was €38,087 in 2013. In comparison, average year salary the same year was €36,400 among those who work (figures from 2013, CBS).

Statements posed to borrowers:

The loan conditions offered by DUO are beneficial I will earn sufficient money later to pay the loan back

I want to be able to afford more luxury (f.e. going out, vacation, car, expensive phone, hobbies)

I borrow in order to work less or not at all next to my study

Statements posed to non-borrowers:

I find a study loan too expensive I do not want to get (large) debts

I do not know the loan conditions of the DUO I work to borrow little or not at all

Statements posed to workers:

Is your job related to your education?

Reasons for working

To gain working experience

Because it increases my chance to get a job

Does your job influence the time that you spend on your study and the results that you achieve in that study?

The fractions of those who answered “Yes, I spend less time on my study and get lower results” are compared between the two groups.