Living Arrangement and University Choice of Dutch Prospective Students

Abstract

Sá C., Florax R. J. G. M. and Rietveld P. Living arrangement and university choice of Dutch prospective students, Regional Studies. The nature of university choice and living-arrangement decisions of Dutch prospective students is analysed. A random utility maximization nested logit model of living-arrangement and university-choice decisions is estimated. Estimation results show distance deters both prospective students living at the parental home and those living elsewhere. Dutch youngsters appear not to be guided by investment motives. They attend the same university as their high school peers. Tight housing markets lower the probability of choosing a given university. Male and low-income students stay longer with their parents in comparison with prospective students with non-Dutch parents.

Sá C., Florax R. J. G. M. and Rietveld P. 丹麦准学生的生活安排及院校选择，区域研究。 本文对丹麦国内准学生生活安排与学院选择的特征进行了分析。我们针对生活安排以及院校选择的决策进行了一个随机的效用最大化嵌套LOGIT模型估测。估测结果表明，对于那些生活在父母家中以及其他地方的准学生而言，距离是一个重要的考量因素。投资动机似乎并未引导丹麦的年轻人。他们更青睐那些他们的高中前辈所选择的大学院校。住房市场紧张这一事实降低了学生选择特定院校的可能性。较之那些双亲为非丹麦人的准学生而言，成年及低收入学生与父母共同居住的时间相对较长。

生活安排 院校选择 随机效用最大化 嵌套Logit

Sá C., Florax R. J. G. M. et Rietveld P. Le logement et le choix d’université des futurs étudiants néerlandais, Regional Studies. Cet article cherche à analyser les déterminants du choix d’université et de logement des futurs étudiants en Hollande. On estime un modèle logit hiérarchique au hasard de la maximisation de l’utilité qui traite du logement et du choix d’université. Les résultats estimés laissent voir que la distance dissuade à la fois les futurs étudiants qui habitent chez leurs parents et ceux qui habitent ailleurs. Il semble que les jeunes néerlandais ne soient pas influés par les possibilités d’investissement. Ils assistent à la même université que ne le font leurs pairs des établissements secondaires. Des marchés du logement serrés réduit la probabilité de choisir une université donnée. Les étudiants masculins et à faible revenu restent plus longtemps chez leurs parents que ne le font les futurs étudiants dont les parents ne sont pas de nationalité néerlandaise.

Logement Choix d’université Maximisation de l’utilité au hasard Logit hiérarchique

Sá C., Florax R. J. G. M. und Rietveld P. Universitäts- und Wohnungswahl von potenziellen holländischen Studenten, Regional Studies. Wir analysieren die Art der Entscheidungen von potenziellen holländischen Studenten hinsichtlich ihrer Universitäts- und Wohnungswahl. Hierfür nehmen wir eine Schätzung für ein Nested-Logit-Modell der zufälligen Nutzenmaximierung hinsichtlich der Entscheidungen bei der Universitäts- und Wohnungswahl vor. Aus der Schätzung geht hervor, dass Entfernung für potenzielle Studenten, die bei ihren Eltern wohnen, ebenso abschreckend wirkt wie für Studenten, die woanders wohnen. Junge Holländer scheinen sich nicht von Investitionsmotiven leiten zu lassen. Sie besuchen dieselbe Universität wie ihre Klassenkameraden in der Sekundarstufe. Aufgrund der Wohnungsnot verringert sich die Wahrscheinlichkeit der Auswahl einer bestimmten Universität. Männliche und einkommensschwache Studenten leben im Vergleich zu potenziellen Studenten mit nicht holländischen Eltern länger bei ihren Eltern.

Wohnungswahl Wahl der Universität Zufällige Nutzenmaximierung Nested-Logit-Modell

Sá C., Florax R. J. G. M. y Rietveld P. Opciones de vivienda y universidad para los futuros estudiantes holandeses, Regional Studies. En este artículo analizamos qué tipos de decisiones toman los futuros estudiantes holandeses con respecto a las opciones de universidad y vivienda. Para ello calculamos en un modelo logit anidado la maximización aleatoria de utilidad en lo que respecta a elegir una universidad y vivienda. Los resultados de las estimaciones muestran que la distancia disuade a los futuros estudiantes, tanto los que viven en el hogar familiar como los que viven en otros lugares. No parece que los holandeses jóvenes se guíen por motivos de inversión. Asisten a la misma universidad que sus compañeros de instituto. La escasez de vivienda reduce la probabilidad de elegir una determinada universidad. Los estudiantes masculinos con bajos ingresos viven más tiempo con sus padres en comparación con los futuros estudiantes de padres no holandeses.

Opciones de vivienda Elección de universidad Maximización aleatoria de utilidad Logit anidado

Keywords:

• Living arrangements
• University choice
• Random utility maximization
• Nested logit

JEL classifications:

• C25
• D85
• I2
• R00

Acknowledgements

The first author gratefully acknowledges the financial support of the Portuguese Foundation for Science and Technology, FCT (Grant Number SFRH/BD/5054/2001). The authors wish to thank Marcel Spruit of the Landelijke Studenten Vakbond (LSVb) for supplying the data on rents in student accommodation, and Erik Wijnen of the Informatie Beheer Group (IBG) for his efforts to explain the data. The authors also benefited from the comments and suggestions of two anonymous reviewers, as well as discussions with participants of the 45th Conference of the European Regional Science Association (Amsterdam, the Netherlands) and the Eureka seminar (VU University, Amsterdam, the Netherlands).

Notes

This parameterization is suggested by Heiss (2002). Hensher and Greene (2002) propose different, but equivalent, parameterizations.

Following Heiss (2002), the designation ‘non-normalized nested logit (NNNL) model’ is used to refer to the nested logit formulation that might not be compatible with a RUM formulation. That formulation is the most frequently presented in the literature (Greene, 2003), and the default in most econometric packages.

For a detailed discussion on the RUM formulation of the nested logit model, see, for instance, Heiss (2002), Hensher and Greene (2002), and Koppelman and Wen (1998).

Börsch-Supan (1990) argues, however, that this condition, pointed out by McFadden, may be unnecessarily restrictive. The conditional logit model can be seen as a special case of the nested logit model, when τ _m  = 1, . It can easily be tested against the nested logit model by means of a likelihood ratio test.

An extra reason pointed out by Heiss (2002) has to do with the simplicity of command syntax.

According to Centraal Bureau voor de Statistiek (CBS) (2005), there were 38 890 university applicants in the school year 2003/2004, of whom 22 770 had a VWO diploma.

A composite overall index for educational quality of the university is used to combine the scores on those attributes in only one measure. The composite index has the advantage of positioning each study programme within the classification of each attribute by comparing it with the mean classification. For each university and attribute, the average score over the different study programmes is computed, weighted by the number of students in that specific programme compared with the total number of students of the programmes evaluated in the 2003 exercise at a specific university.

The quality index is strictly positive and varies between 0.5 and 1.5 per attribute (Portela, 2001).

However, the model is estimated with the student–teacher ratio, and the fourth section discusses how those results compare with those for the model with the quality index.

Some authors would argue that the urbanization index would be a good proxy for leisure. The Dutch context is a bit specific in this regard as it is possible to think of several Dutch cities with quite low values for the urbanization index, but still very attractive for students because of their tradition as party cities.

The survey covers 253 offers in all twelve university cities in the Netherlands. Room adverts were randomly picked from http://www.kamernet.nl/. Specifically, the first ten advertisements on a certain number of days were taken and both the rent and the surface area of the room were registered.

This index is used by Brose (2003), who took it from Magurran in his work of 1988 on ecological diversity measurement.

The Dutch Ministry of Education, Culture and Science groups all university study programmes into nine main areas, namely: Behaviour and Society; Economics; Education; Health Care; Land and Natural Environment; Language and Culture; Law; Nature; and Technical. For areas not present in a given university:

which follows from the L'Hôpital rule.

The general formulation for this measure is:

The special case is used when δ = 1, which is standard in the literature of hierarchical destination choice (Fotheringham et al., 2001).

In order to avoid scale problems, the intra-zonal distance is defined, which is relevant when a high school and a university location coincide in the same region, as:

where s _i is the area of region i (m2) (Rietveld and Bruinsma, 1998). The formula assumes that regions are circular, and all zones are equally intensively used. For a discussion on the use of geographical distance as a spatial separation measure, see Sá et al. (2004), pp. 381–382.

In the particular case of Dutch students, the monetary costs of travelling are not relevant as they are all entitled to a transportation permit, allowing them to travel for free.

Each ρ _m can be obtained using the expression:

The authors believe there is no unique reason for this fact. It might have to do with: personal preferences (for example, better students want to concentrate on their studies and give less priority to living independently); the fact that in the Netherlands there is no numerus clausus for most studies, and hence both low- and high-ability students have the same university opportunities; and the parental residential location choice, which makes it more likely that better students will live close to the best schools, and they therefore do not need to move to attend university.

In line with the remarks made in the Introduction, which are reinforced in the fourth section, this does not imply a causal relationship. It just represents the observed data.