1
Is there a link between college quality and
future earnings?
Bachelor thesis 2016/2017
Abstract:
Choosing the correct university is a difficult task for high-school students, one thing to consider when making a decision can be the quality of the university, because a higher quality might increase the wage that they will earn in the future. Since studies in the United States reported that there is a link, it might be interesting to see whether this also applies to other countries. In this thesis the goal therefore will be to find whether there is a link between college quality and future earnings in The Netherlands. College quality is measured by comparing international rankings, then the logarithm of wages per hour is regressed on the college quality. The results of the regression show that there is no significant effect for college quality on earnings. Thus there can be concluded that there is no link between college quality and future earnings in The Netherlands, in contradiction to previous studies conducted in the United States.
Aiko Cerutti
Student number: 10399747 Supervisor: Sabina Albrecht
2
Statement of originality
This document is written by Student Aiko Cerutti, 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.
3
Table of contents
1 Introduction 4
2 Literature review 5
2.1 Theoretical background 5
2.2 Measuring college quality 6
2.3 Previous results 7
2.4 Hypothesis 8
3 Methodology & Data 8
3.1 Data 8
3.2 College quality 10
3.3 Methodology 13
4 Results 14
4.1 College quality and future earnings 14
4.2 College quality and job possibilities 16
5 Conclusion & Discussion 18
4
1. Introduction
During the final stage of high school, students have to decide at which university to study. This is often a difficult decision, with a lot of things to take note of. One important thing to keep in mind, while making a decision, is to think of your future earnings. Usually, future earnings are explained by variables such as innate ability, labor force experience, and years of education (James et al., 1989). In this paper the variable college quality is examined, so that it can be tested whether college quality has a relation with future earnings.
The purpose of this paper is to see whether there is a link between college quality and future earnings for alumni in the Netherlands. So far only research in the United States has been conducted on this relation. This research will try to find out whether the same link can be found in the Netherlands. Furthermore, the possible cause of this relation will be researched.
“For most students, college attendance involves three sequential choices. First, a student decides which set of colleges to apply to for admission. Second, colleges independently decide whether to admit or reject the student. Third, the student and her parents decide which college the student will attend from the subset of colleges that admitted her.” (Dale and Krueger, 2002, p. 1494). This how it usually goes in the United States. In the Netherlands however, students are able to apply to any of the thirteen universities as long as they have obtained a high school degree. Only for a select couple of study courses there are extra terms that have to be met. But generally speaking, every high school student in the Netherlands can go to each one of the universities in the Netherlands.
In the late 1960s and early 1970s the modern literature on the effect of college quality originated. In the 1990s more research was conducted. The results were not only important for theoretical and academic purposes, but also for parents and students (Zhang, 2005). Since the literature on the relation between college quality and future earnings is not unambiguous, it is difficult to see what causes this relation. Dale & Krueger (2002) found no significant effect between college quality and future earnings, but Brewer et al. (1999) and Behrman et al. (1996) did.
In the following chapter the previous research that has been done on this subject will be discussed. That chapter will also discuss different ways of calculating college quality. After chapter two, it will be explained how the data is obtained. Moreover, the methodology that will be used to answer the research question, is discussed. In chapter four the results will be presented and discussed. Finally, a conclusion will be given along with a short discussion of this paper.
5
2. Literature review
In this section a theoretical background will be provided and the relevant literature will be studied in detail. Paragraph 2.1 gives a theoretical background on the returns to college. Paragraph 2.2 discusses measurements of college quality. Afterwards, paragraph 2.3 reviews prior studies and methods to measure the link between college quality and future earnings. Finally the hypothesis is defined in paragraph 2.4. Note that in this paper the terms ‘college’ and ‘university’ will be used interchangeable.
2.1 Theoretical background
There have been multiple analyses on the relation between college quality and earnings and they use the same basic methodology. The logarithm of an individual's earnings or hourly wage rate (Ln Wi)is regressed on a set individual characteristics(Xi) and set of college characteristics for the university he or she attended (Cij). Furthermore, (µi) denotes a normally distributed error term: Ln Wi = β0 + β1Xi + β2Cij + µi (Brewer et al., 1999)
In C college quality is measured and β2 can be interpreted as the relation between college quality and earnings. This model can be further extended by specifying the individual characteristics into family background (Fi ), academic background (Ai), adding demographic characteristics (Di) and job market conditions (Ji):
Ln Wi = α0 + α 1Di+ α 2Fi + α3Ai + α4Ji + α5Cij + µi (Zhang, 2005)
When trying to measure the effect of college quality it is important that the unobserved individual characteristics do not correlate with college quality, as they would cause omitted variable bias. A potential weakness of this methodology is that students self-select themselves in high-quality universities, because the expected economic payoff is higher for these universities. If more able students apply to higher quality universities, the earnings may be caused by the students themselves and not by the higher quality of a university (Weisbrod & Karpoff, 1968). In this paper this will be called 'student selectivity'. Another weakness of this methodology is that universities admit students based on student ability, measured by SAT scores and high school grades. Therefore more selective universities admit only those students with a higher SAT score. The earnings of the alumni of those universities might then be higher because of students who have more ability and not because of higher college quality (Dale & Krueger, 2002). In this paper this will be called 'college selectivity'.
6
2.2 Measuring college quality
The literature shows that there are different ways to measure college quality. A popular way to measure college quality is to look at SAT scores of entering freshmen. Dale and Krueger (2002) argue that an admissions committee bases decisions on two sets of variables: an observed set by researchers and an unobserved set by researchers. The observable set consists of students' SAT scores and high school grade point average (GPA) and the unobservable set includes student’s motivation, ambition and perceived maturity. According to Dale and Krueger (2002) a higher selective college admits students with higher average SAT scores. However, it may be a poor proxy for college quality. In the Netherlands graduating from high school is good enough to be admitted to every university, only for certain studies higher average grades at high school increase the possibility of being admitted.
Another measurement of college quality is college tuition and expenditure per student. If universities are efficient users of resources, then expenditure can be a reasonable index for quality of schooling (Wachtel, 1976). College tuition may have a significant effect on future earnings, because universities with a higher tuition provide their students with more, or higher quality, resources (Dale & Krueger, 2002). Dale and Krueger (2002) find a correlation of less than 0.30 between college tuition and student expenditure. There are limitations for measuring expenditures: undergraduate and graduate students expenditures are combined, expenditures are irregular over time and there are difficulties classifying expenditures that contribute to teaching quality and expenditures that do not contribute to teaching quality. However, Behrman et al. (1996) argue that alumni from colleges where the faculty salaries are higher, earn significantly higher wages. The rankings that are used in this study do look at faculty salaries, however Dale and Krueger (2002) argue that expenditure per student can provide mixed results. Also, since college tuition in the Netherlands is more or less the same for each university, this will not be used to measure college quality.
To measure college quality Brewer et al. (1999) used the Barron’s rating. The Barron’s Profiles of American Colleges groups colleges based on the percentage of applicants admitted in four different categories and so measures college selectivity. In combination with SAT scores Brewer et al. (1999) divided the universities in different quality ranks. When measuring college quality in the Netherlands it is not possible to use the Barron’s rating, since it is for the United States only.
Black and Smith (2004) also add the variable freshman retention rate to measure college quality. The retention rate is the fraction of freshmen that return to the same university in their next year. This
7
rate represents a ”voting with your feet” measurement and measures the perceived quality by students and their parents. In this thesis rankings will be used that also processed this rate in their ranking.
Dill and Soo (2005) have examined college rankings in four different countries: Australia, Canada, the United Kingdom and the United States. The rankings of those countries use among other things academic reputation, the quality of the faculty and the financial resources. Dill and Soo (2005) argue that the rankings in those countries are able to represent college quality adequately. In this paper the international rankings will be used to measure college quality.
2.3 Previous results on college quality and future earnings
In previous studies there have been mixed results with regards to the effect of college quality on earnings. Brewer et al. (1999) found that students who attended higher quality colleges earned a larger wage, after controlling for ethnicity, gender, family size, parental education, test scores and having a part-time job. In addition Behrman et al. (1996) found similar results, but instead of using a structural model to control for selection bias, they conducted their research using monozygotic (identical) and dizygotic (non-identical) twins.
In contrast, Dale and Krueger (2002) found that there was no significant effect on earnings. As mentioned before, Dale and Krueger used SAT scores to determine college quality. They argue that students with higher SAT scores are more likely to attend a college of higher quality, suggesting that not the quality of a university causes a higher wage, but the ability of the students themselves. Furthermore, Dale and Krueger (2002) state that students with a disadvantageous background benefit more from attending a higher quality college.
In the paper of James et al. (1989), it is tested which college characteristics create the quality that increases the future earnings potential. In the model they used, they included students characteristics, institutional characteristics, a set of education experience variables and a set of labor market variables. They found that going to a more selective college increased the future earnings potential, however they did not find an impact for expenditure per student. Furthermore, James et al. (1989) state that the college does not have the biggest impact, but the study course students follow has the biggest impact. They also note that students do not base their study decisions solely on future earnings, but also on other factors, such as learning, research and value of information. Hoxby (2009) states that students selectivity has increased in the past years. Where students used to go to a local college regardless of their abilities and its characteristics, nowadays their choices are more based on a college’s resources and student body. She also argues that the rise in student
8
selectivity caused college selectivity to rise in some colleges and fall in others. For the universities in the Netherlands this means that students do not make their decision based on distance as much as they used to. However, the distances between universities in the Netherlands are smaller than those in the United States. For this research it means that there is some sort of students selectivity present, but perhaps not as much as in the United States. The research conducted by Light and Strayer (2000) suggests that students tend to sort themselves by ability: low ability students tend to go to lower quality colleges and more abler students tend to go to higher quality colleges. They also argue that a “match” between student ability and college quality increases the possibility of graduating. According to the rankings that will be used in this paper the differences in quality between the lower and higher ranked colleges in the Netherlands are smaller than the differences in the United States. Therefore, the student selectivity will likely be smaller.
Zhang (2005) has examined whether different college quality measures give different outcomes. Zhang states that the estimated effect of college quality is affected by the measure of college quality. Although all measurements showed a positive relation between college quality and earnings, the magnitude of the effect differed between the measurement types.
2.4 Hypothesis
The goal of this research is to try to find whether there is a relation between college quality and future earnings in the Netherlands. In line with previous studies it is expected that the relation between college quality and future earnings is positive, meaning that those who graduated from a college with a higher perceived quality, earn on average more than those who attended a college with a lower quality. It is however expected that the effect of this relation will be very small, because the differences in college quality are relatively small compared to the United States.
3 Methodology & data
In this section it will be explained how the data is obtained and how college quality will be measured. Furthermore, the model and methodology will be discussed. Paragraph 3.1 explains what data will be used and where the data is conducted from. Then the measurement of college quality will be mentioned in paragraph 3.2. At last, the model and methodology will be discussed in paragraph 3.3.
3.1 Data
The data on alumni earnings is obtained from SEO Economisch Onderzoek (SEO), which is a research facility founded by the economic faculty of the University of Amsterdam. Since 1997 SEO
9
Economisch Onderzoek has annually been bringing out reports on the market position of recently graduated students, commissioned by Elsevier. In this report SEO conducts a survey among alumni from Dutch universities. Data on study choice and wages is earned, but also personal statistics such as ethnicity, age and gender are gathered (January 23, 2017). Retrieved from:
http://www.seo.nl/over-ons/introductie/.
The data set contains 32,637 observations from annual surveys taken since 2007 up to 2016. In the table below the descriptive statistics are shown. The first three variables measure the year of graduation, the year in which the survey was conducted and the year in which the alumnus started working at the current job. The graduation years vary from 2004 to 2014 and the start of job years vary from 2001 to 2016. The variable ethnicity denotes a 0 if both parents and the alumnus were born in the Netherlands and 1 if otherwise. In this dataset 15.1% of the alumni either is born abroad or has at least one parent that is born abroad. Furthermore, hours worked weekly and monthly wage before tax were observed, from those two variables I am able to calculate hourly wages. The same formula is used as the SEO to calculate wages per hour: (12*monthly wage before tax) / (52*hours worked weekly). The average hours worked per week is 35.67 and the average monthly wage before tax is €2548,-. Furthermore, the average wage per hour is €16.53. At last, there is a variable for the study course.
(1)
(2)
(3)
(4)
(5)
VARIABLES
N
mean
sd
min
max
Graduation year
32,637
2009
2.935
2004
2014
Survey year
32,637
2011
2.877
2007
2016
Job start year
29,469
2010
3.028
2001
2016
Ethnicity
32,637
0.151
0.358
0
1
Weekly hours
28,277
35.67
7.548
0
168
Monthly wage b. tax 29,662
2,548
997.2
90
15,847
Hourly wage
27,639
16.53
8.358
0.568
473.4
Study course
32,637
27.04
15.39
1
54
Table 1
In the survey of the SEO the alumni were asked to include their study course. The study courses were grouped in 54 different categories, for this research those 54 categories are combined into ten different categories. The ten different categories are: Education and Nurturing studies (1), Language and Communication studies (2), Art and Cultural studies (3), Law and Administration (4), Economics and Business (5), Social Sciences (6), Medical and Health Science (7), Earth and Environment (8),
10
Computer Science, Mathematics (9) and Technical studies (10). This has been done in order to control for different wage effects of study choice. As James et al. (1989) mentioned, the choice of study course has an effect on future earnings.
3.2 College Quality
In this paper the quality of college is measured using international rankings. The rankings that are used are: The Times Higher Education World University Rankings, the Academic Ranking of World Universities (ARWU) and the QS World University Rankings. According to Dill and Soo (2005) these rankings are able to represent college quality adequately. To measure college quality the rankings for Dutch universities for the last five years have been used. Firstly, an average ranking for each of the three international rankings has been made. Afterwards the three average rankings have been combined to measure the rank for each university. In the right column of each ranking the relative rank compared to universities in The Netherlands is mentioned in between brackets.
The Times Higher Education World University Rankings uses thirteen performance indicators grouped into five areas. These areas are: teaching (30%), research (30%), citations (30%), international outlook (7.5%) and Industry income (2.5%). The teaching area consists of the perceived prestige of the institution in teaching, a staff to student ratio, the proportion of postgraduate research and institutional income. Research captures the university’s reputation for research, the income of research scaled against academic staff and adjusted for purchasing-power parity (PPP), and productivity to measure a university’s ability to publish papers in quality journals. Citations measures how often a university’s public is cited by scholars. The citations show how much a university is contributing to human knowledge. This indicator is obtained from Elsevier’s Scopus database. International outlook is an indicator of the ability of a university to attract students and staff from abroad. Finally, industry income is used to determine the ability to help an industry with innovations consultancy and inventions. In table 2 the rankings of the Times Higher Education World University Rankings are displayed (January 30, 2017). Retrieved from:
https://www.timeshighereducation.com/world-university-rankings/methodology-world-university-rankings-2016-2017.
Times Higher Education World University Rankings Average Ranking University Year 2012 2013 2014 2015 2016 University of Amsterdam 83 83 77 58 63 72.8 (5) VU University of Amsterdam 140 144 136 154 156 146 (11) Utrecht University 67 74 79 62 86 73.6 (6) University of Groningen 89 98 117 74 80 91.6 (7)
11
Leiden University 64 67 64 67 77 67.8 (2)
Erasmus University 72 73 72 71 69 71.4 (4)
Radboud University Nijmegen 127 131 140 125 121 128.8 (9)
University of Wageningen 70 77 73 47 65 66.4 (1)
Delft University of Technology 77 69 71 65 59 68.2 (3)
Eindhoven University of Technology 114 107 144 176 177 143.6 (10)
Maastricht University 115 98 101 88 94 99.2 (8)
University of Twente 187 170 201-225 149 153 174.4 (12)
Tilburg University 201-225 226-250 276-300 201-250 198 232.5 (13) Table 2
The Academic Ranking of World Universities uses six indicators to rank universities. Quality of education (10%) is measured by the alumni that won a Nobel Prize or a Fields Medal. Quality of faculty is measured by two different indicators: the total number of staff that won a Nobel Prize or a Fields Medal (20%) and the number of highly cited researchers selected by Thomson Reuters (20%). Papers published in Nature and Science (20%) and papers in the Social Science Citation Index and the Science Citation Index-expanded (20%) capture research output. At last, per capita performance (10%) is measured by dividing the above five by the number of full-time equivalent academic staff. In table 3 the rankings for the ARWU are shown. The ARWU does not provide an exact ranking for universities ranked over 100, but gives an index range. For the Tilburg University a rank of over 500 was given in the years it did not rank in the top 500 (January 30, 2017). Retrieved from:
http://www.shanghairanking.com/ARWU-Methodology-2016.html. Academic Ranking of World
Universities Average Ranking University Year 2012 2013 2014 2015 2016 University of Amsterdam 101-150 101-150 101-150 101-150 101-150 125.5 (5-7) VU University of Amsterdam 101-150 101-150 100 98 101-150 114.9 (4) Utrecht University 53 52 57 56 65 56.6 (1) University of Groningen 101-150 92 82 75 72 89.3 (3) Leiden University 73 74 77 82 93 79.8 (2) Erasmus University 151-200 151-200 151-200 151-200 101-150 165.5 (8) Radboud University Nijmegen 101-150 101-150 101-150 101-150 101-150 125.5 (5-7) University of Wageningen 101-150 101-150 101-150 101-150 101-150 125.5 (5-7) Delft University of Technology 201-300 201-300 201-300 201-300 151-200 235.5 (9) Eindhoven University of Technology 301-400 301-400 301-400 301-400 201-300 330.5 (11) Maastricht University 201-300 201-300 201-300 201-300 201-300 250.5 (10) University of Twente 301-400 301-400 301-400 301-400 301-400 350.5 (12) Tilburg University 401-500 >500 401-500 >500 >500 500 (13) Table 3
12
The last international ranking that is used is the QS World University Rankings. Similar to the previous rankings the QS World University Rankings also measures academic reputation (40%). It is measured by a survey conducted among academics in which they identify the universities where they believe the best work is currently taking place within their own field of expertise. Secondly, they look at employer reputation (10%), which is also measured by a survey. The survey asks employers to identify the universities that produce the best graduates. The ranking also has a student-to-faculty ranking (20%), capturing the number of academic staff to the number of students enrolled. Just like the other rankings, this ranking also has an indicator for the citations per faculty (20%). Like the Times Higher Education World University Rankings, they also use the Scopus database to collect the information about the citations. Finally, this ranking has two ratios to measure the attractiveness of the university for staff and students from other nations. An international faculty ratio (5%) and an international student ratio (5%) are used to measure these. The rankings for the past five years of the QS World University Rankings are shown below in table 4 (January 30, 2017). Retrieved from: http://www.topuniversities.com/qs-world-university-rankings/methodology.
QS World University Rankings Average
Ranking University Year 2012 2013 2014 2015 2016 University of Amsterdam 62 58 50 55 57 56.4 (1) VU University of Amsterdam 177 181 171 176 199 180.8 (11) Utrecht University 85 81 80 94 104 88.8 (4) University of Groningen 109 97 90 100 113 101.8 (5) Leiden University 75 74 75 95 102 84.2 (3) Erasmus University 99 92 90 126 144 110.2 (6)
Radboud University Nijmegen 136 143 156 177 190 160.4 (10)
University of Wageningen 161 150 151 135 119 143.2 (9)
Delft University of Technology 103 95 86 64 62 82 (2)
Eindhoven University of Technology 158 157 147 117 121 140 (8)
Maastricht University 107 121 118 169 173 137.6 (7)
University of Twente 224 228 212 188 177 205.8 (12)
Tilburg University 401-450 373 367 293 330 357.7 (13)
Table 4
Now that the average ranking for each of the three rankings has been calculated, the final international ranking can be determined by combining the three previous average ranks. The final average rank is displayed in table 5. As can be seen in the table below, there is some consistency between the three rankings. The Universities of Leiden and Utrecht perform good in all three
13
rankings, while Tilburg University and the University of Twente perform poorly in the rankings. There are some small variations in rankings however.
University Rankings Average Times Average ARWU
Average QS Average all 3 University of Amsterdam 72.8 (5) 125.5 (5-7) 56.4 (1) 84.90 (3) VU University of Amsterdam 146 (11) 114.9 (4) 180.8 (11) 147.23 (9) Utrecht University 73.6 (6) 56.6 (1) 88.8 (4) 73.00 (1) University of Groningen 91.6 (7) 89.3 (3) 101.8 (5) 94.23 (4) Leiden University 67.8 (2) 79.8 (2) 84.2 (3) 77.27 (2) Erasmus University 71.4 (4) 165.5 (8) 110.2 (6) 115.70 (6)
Radboud University Nijmegen 128.8 (9) 125.5 (5-7) 160.4 (10) 138.23 (8) University of Wageningen 66.4 (1) 125.5 (5-7) 143.2 (9) 111.70 (5) Delft University of Technology 68.2 (3) 235.5 (9) 82 (2) 128.57 (7) Eindhoven University of Technology 143.6 (10) 330.5 (11) 140 (8) 204.70 (11)
Maastrich University 99.2 (8) 250.5 (10) 137.6 (7) 162.43 (10)
University of Twente 174.4 (12) 350.5 (12) 205.8 (12) 243.57 (12)
Tilburg University 232.5 (13) 500 (13) 357.7 (13) 363.40 (13)
Table 5
3.3 Methodology
In order to determine if there is a link between college quality and earnings the following formula is used:
Ln Wi = α0 + α 1Si+ α 2Ei + α3Fi + α4Wi + α5Ci + µi (1)
The logarithm of an individual’s hourly wage rate (Ln Wi) is regressed on the college major (Si), the ethnic background (Ei), whether the individual works part time (Fi), the work experience (Wi) and the quality of college (Ci). The college major is a dummy variable for each of the ten different categories of college majors. Ethnic background is a dummy variable as well, it is 0 if both parents and the alumnus were born in the Netherlands and 1 if otherwise. The variable for part time is 1 if the alumnus works less than 36 hours and 0 if the hours worked weekly are at least 36. Work experience is the continuous variable measuring the time in years between starting a job and the time of the survey. The alumni in the data set were surveyed two to three years after graduating. However, some of them already had a job before graduating and therefore have more work experience. The last variable is college quality measured as mentioned in the previous paragraph.
In order to answer the research question, the following hypotheses have been formulated: H0 : α5 = 0
14 H1 : α5 < 0
If α5 is not significantly different from zero, it means there is no link between college quality and future earnings in the Netherlands. If α5 is negative, it would indicate that a higher college quality correlates with higher future earnings.
As a robustness check, this regression will be run a second time to see if there is a difference when alumni who are self-employed are excluded. The reason hereof is that those who are self-employed have reported very low hours or very high hours worked weekly. This caused relatively high and low wages per hour.
To see if the possibility of getting a job differs between universities another regression will be run: Li = β0 + β1Si + β2Ei + β3Ci + µi (2)
Here Li measures the labor market position of the alumnus – is someone currently unemployed or not? This is measured by a dummy variable, being 0 if the alumnus has a job and 1 when the alumnus is unemployed. Because the dependent variable is a dummy, a linear probability model will be used. Since students probably also care about finding a job, it is interesting to test whether the possibility of finding a job correlates with college quality. In order to see this it will be tested whether β3 is significantly higher than zero, meaning, would a higher rank (lower college quality) lead to a higher possibility of being unemployed.
H0 : β3 = 0 H1 : β3 > 0 4 Results
In this section the results will be presented and examined. In paragraph 4.1 the first regression will be presented, afterwards the results for the probability of obtaining a job will be demonstrated. 4.1 College quality and future earnings
After running regression (1) of the methodology section, first with self-employed alumni included (Model 1) and secondly without self-employed alumni (Model 2), the results are presented in the table below. There were 604 self-employed alumni that reported a monthly wage and their weekly hours worked.
(1)
(2)
15
2.Language &
-0.0771***
-0.0807***
Communication
(0.00939)
(0.00847)
3.
Art & Culture
-0.0615***
-0.0674***
(0.00911)
(0.00824)
4.
Law &
0.0622***
0.0429***
Administration
(0.00888)
(0.00797)
5.
Economics &
0.111***
0.0915***
Business
(0.00938)
(0.00843)
6.
Social Sciences
-0.0251***
-0.0361***
(0.00953)
(0.00856)
7.
Medical & Health
0.190***
0.114***
Science
(0.00868)
(0.00788)
8.
Earth &
-0.0119
-0.0308***
Environment
(0.00972)
(0.00872)
9.
Computer Science
0.0599***
0.0399***
& Mathematics
(0.00891)
(0.00800)
10. Technical studies
0.0446***
0.0259***
(0.00962)
(0.00865)
Ethnicity
-0.0115**
-0.0124***
(0.00523)
(0.00473)
Part-time
-0.0976***
-0.0625***
(0.00473)
(0.00434)
Work experience
0.0479***
0.0431***
(0.00209)
(0.00189)
College rank
-6.06e-06
-2.38e-05
(2.70e-05)
(2.42e-05)
Constant
2.704***
2.702***
(0.00850)
(0.00765)
Observations
27,121
26,517
R-squared
0.089
0.070
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 6
As we can see in table 6, the effect of college quality, represented by college rank, is negative, however it is not significantly different from zero. This means that there is no link between college quality and earnings in the Netherlands. This result differs from those found in the United States (Brewer et al., 1999 & Behrman et al., 1996). This may be caused because differences in the Netherlands between universities are smaller than the differences between universities in the United States. Every university, apart for the University of Tilburg, is ranked in the top 500 for all
16
three rankings, but not in the top 50. Universities from the United States rank from top 10 to out of the top 800. In other words, universities in the Netherlands are all good but are not the very best. Another possible reason for finding no link is that college quality was measured by comparing international rankings. Zhang (2005) argues that different findings may be caused by different measurements of college quality.
Since there are differences for the college majors that are significant, the choice of study does influence the earnings. This result was also found by James et al. (1989). Those who studied Language and Communication studies (7.71%), Art and Cultural studies (6.15%) and Social Sciences (2.51%)have on average lower earnings compared to Education and Nurturing Studies. Law and Administration (6.22%), Economics and Business (11.1%), Medical and Health Science (19%), Computer Science and Mathematics (5.99%) and Technical Studies (4.46%) earn on average more than those who graduated with a degree in Education and Nurturing studies. Only for Earth and Environment studies there was no significant effect found. But after running the regression again and excluding self-employed alumni, a significant effect can be found that alumni from this study course earn on average less.
Furthermore, being born abroad or having at least one parent not being Dutch decreases the earnings by 1.15% and after excluding the self-employed, a decrease of 1.24% in earnings was found. This finding may be caused by a discrimination on the labor market.
From the results in the table, it can be seen that alumni that do not work full-time earn on average 9.76% less than those that work full-time. However, after running the regression a second time when excluding the self-employed only a 6.25% lower wage was found for not working full-time. As mentioned in the methodology section, self-employed alumni have either reported very high or very low hours worked per week. When excluding this group, the difference between those who work at least 36 hours and those who do not, becomes smaller.
In line with economic theory, an increase in work experience also leads to higher earnings, because it increases productivity. In table 6, it is shown that for every extra year of work experience an increase in earnings of 4.79% was found and for model 2 an increase of 4.31%. The work experience of the participants ranged from zero to six years, therefore it cannot be concluded that this rise in earnings will continue after working for six years. Also, work experience is expected to have diminishing effects over time.
17
In order to earn a wage, one must have a job. Since finding a job can be difficult depending on the situation of the labor market, it might be interesting to see if the possibilities of finding a job are linked with college quality. As can be seen in table 7, college rank is significantly lower than zero, meaning that there is a link between college quality and obtaining a job. This is a contradiction to what was expected, since this means that students that graduated from universities with a lower college quality, have a higher chance to find a job. However, it is still very close to zero and thus has a very small effect.
In line with the findings in the previous regression, being born abroad or having at least one parent not being Dutch decreases the chance of finding a job. This may again be caused by discrimination on the labor market.
Just as the college majors influence wages, they also influence the probability of getting a job. Compared to Education and Nurturing studies only Computer science and Mathematics has a significantly higher chance of finding a job. Those who studied Language and Communication studies, Art and Cultural studies, Law and Administration, Social Sciences and Earth and Environment have a lower chance of finding a job. Economics and Business, Medical and Health Science and Technical studies do not have a significantly different chance of finding a job, compared to those who graduated with an Education and Nurturing degree. The dataset contains surveys taken from 2007 until 2016, this means that there has been a financial crisis that may have caused a disturbance on the labor market, therefore these results may have been different if this would not have been the case.
(1)
VARIABLES
Model 3
2.Language &
0.0713***
Communication
(0.00745)
3.
Art & Culture
0.0901***
(0.00714)
4.
Law &
0.0310***
Administration
(0.00703)
5.
Economics &
-0.00337
Business
(0.00747)
6.
Social Sciences
0.0682***
(0.00760)
7.
Medical & Health
-0.00267
Science
(0.00695)
18
Environment
(0.00775)
9.
Computer Science
-0.0149**
& Mathematics
(0.00705)
10. Technical studies
-0.0107
(0.00768)
Ethnicity
0.0222***
(0.00411)
College rank
-7.25e-05***
(2.20e-05)
Constant
0.0579***
(0.00609)
Observations
32,376
R-squared
0.021
Standard errors in parentheses
*** p<0.01, ** p<0.05, * p<0.1
Table 7
5. Conclusion & Discussion
In this thesis a link between college quality and future earnings in the Netherlands was tried to be proven. College quality was measured by calculating the average ranking of the universities in the Netherlands using three different international rankings. Data on earnings was obtained from SEO Economisch Onderzoek. After running a regression, no link was found between college quality and future earnings in the Netherlands. The cause of this result may have been that the differences in quality of the universities are relatively small compared to the United States, where most of the previous studies were conducted.
However, a relation between the study course and earnings was found, meaning that a student earns on average more after studying a medicinal study rather than a social study. This is in line with the findings of James et al. (1989). Furthermore, it was found that alumni that are born abroad or have at least one parent born abroad, have on average lower earnings and have a lower chance of getting a job. This can be explained by discrimination on the labor market.
In this paper college quality was measured using international rankings. Even though Dill and Soo (2005) reported that these do adequately reflect college quality, it might not be the best way to measure this. For a student the quality of their study course matters and not the overall quality of the university. Since the rankings do not account for different faculty qualities, this may provide skewed results. When conducting further research, looking at faculties individually and comparing
19
them with other universities may provide a better answer as to where a high school student should study.
Furthermore, Dale & Krueger (2002) argue that differences in earnings are not caused by differences in college quality, but are caused by differences in student ability. In this thesis student ability was not controlled for, therefore it might be interesting to do so when conducting further research. Since the alumni in the dataset were surveyed two to three years after graduating, the results only apply for two to three years after graduating. It might be the case that there are differences after five or ten years. Finally, most of the alumni were surveyed during the financial crisis,. Therefore it may have been difficult for some of them to find a job and to get a decent wage. In the future it may be good to conduct another research in a time when there is no crisis, in order to see whether the results differ.
References
Behrman, J. R., Rosenzweig, M. R., & Taubman, P. (1996). College choice and wages: Estimates using data on female twins. The Review of Economics and Statistics, 672-685.
Black, D. A., & Smith, J. A. (2004). How robust is the evidence on the effects of college quality? Evidence from matching. Journal of Econometrics, 121(1), 99-124.
Brewer, D. J., Eide, E. R., & Ehrenberg, R. G. (1999). Does it pay to attend an elite private college? Cross-cohort evidence on the effects of college type on earnings. Journal of Human
resources, 104-123.
Dale, S. B., & Krueger, A. B. (2002). Estimating the Payoff to Attending a More Selective College: An Application of Selection on Observables and Unobservables. The Quarterly journal of
economics, 117(4), 1491-1527.
Dill, D. D., & Soo, M. (2005). Academic quality, league tables, and public policy: A cross-national analysis of university ranking systems. Higher education, 49(4), 495-533.
Hoxby, C. M. (2009). The changing selectivity of American colleges. The Journal of Economic Perspectives, 23(4), 95-118
James, E., Alsalam, N., Conaty, J. C., & To, D. L. (1989). College quality and future earnings: where should you send your child to college?. The American Economic Review, 79(2), 247-252. Light, A., & Strayer, W. (2000). Determinants of college completion: School quality or student
20
Wachtel, P. (1976). The effect on earnings of school and college investment expenditures.
Review of Economics and Statistics, 58, 326–331.
Weisbrod, B. A., & Karpoff, P. (1968). Monetary returns to college education, student ability, and college quality. The Review of Economics and Statistics, 50(4), 491-497.
Zhang, L. (2005). Do measures of college quality matter? The effect of college quality on graduates' earnings. The Review of Higher Education, 28(4), 571-596.
http://www.seo.nl/over-ons/introductie/, January 23, 2017
https://www.timeshighereducation.com/world-university-rankings/methodology-world-university-rankings-2016-2017, January 30, 2017
http://www.shanghairanking.com/ARWU-Methodology-2016.html, January 30, 2017
