E     I     P
		Short-Term Forecasting of the Demand for University Places
		97/21 Philip Lewis Geoff Crockett Centre for Labour Market Research March 1998
		Evaluations and Investigations Program Higher Education Division Department of Employment, Education, Training and Youth Affairs

©Commonwealth of Australia 1998

ISBN 0 642 23755 7

This work is copyright. Apart from any use as permitted under the Copyright Act 1968, no part may be reproduced by any process without prior written permission from the Australian Government Publishing Service. Requests and inquiries concerning reproduction and rights should be addressed to the manager, Commonwealth Information Services, Australian Government Publishing Service, GPO Box 84, Canberra ACT 2601.

Executive Summary

The project, Short-Term Forecasting of the Demand for University Places, has been carried out by the Centre for Labour Market Research (CLMR) for the Department of Employment, Education, Training and Youth Affairs (DEETYA) as part of the Department’s Evaluations and Investigations Program. The aims of the project are to:

• conduct an extensive search of the literature concerned with university enrolments in order to identify previous modelling and assist with the selection of appropriate variables to be used in an empirical analysis of the demand for university places;
• identify possible factors affecting the demand for university places, including

• an analysis of the link between school and university participation,
• changes in employment conditions,
• fee structures for university courses,
• the rate of return to university education, and
• a state by state analysis of the demand for education;

• identify factors affecting demand for specific disciplines

• this involves the collation of discipline specific data relating to the 10 major fields of study under DEETYA classifications (Agriculture, Architecture, Arts, Business, Education, Engineering, Health, Law, Science and Veterinary Science), and
• some analysis of the link between the decision to remain at school and entering university will be made; and

• the development of a quantitative model of the demand for university education

• this will involve modelling the decision to attend university using a wide range of variables,
• this has been undertaken for both males and females, and
• testing of the models includes tests for significance, diagnostic statistical tests and tests for model and forecast stability.

The Introduction outlines the rationale for the report and also assesses the importance of short-term forecasting in the higher education sector. It sets out the structure of the report and acknowledgments.

An Overview of Trends in Secondary Education and Youth Employment (in Chapter 1) focuses on the most significant component of demand for university places, namely teenagers in upper secondary education. This chapter outlines and examines the trends in the teenage labour market and school retention in Australia over the past 25 years. It covers a broad range of associated issues, such as school participation rates, youth unemployment and apprenticeships. The chapter is designed to identify for the reader the important movements in young peoples’ participation in post-compulsory education and the labour force and the changing interaction between the two in recent decades. The findings from this chapter are essential in establishing the existence of a link between the teenage labour market, school attendance and future university participation.

Recent Trends in Higher Education (in Chapter 2) presents an historical overview of changes in the Australian higher education system. This chapter documents increases in total and commencing student numbers by gender, age group, field of study, level of course and enrolment type, while also looking at overseas student enrolments in Australia. A discussion of the labour conditions faced by 20–24-year-olds is also undertaken, as labour force participation is the main alternative for the majority of students under 24 years of age. Important aspects of the demand for higher education are outlined, with an emphasis being placed upon the rate of return to education, and in particular an analysis of graduate work opportunities based on data from the Graduate Destination Survey. The importance of these factors in relation to education over the past two decades is examined, as is the introduction of the Higher Education Contribution Scheme (HECS) in 1989, and changes in the level of student assistance.

Review of the Education Participation Literature (in Chapter 3) provides an overview of the literature in this area, especially literature dealing with the estimation of demand for university places. Focus is placed on recent Australian studies which model rising school participation rates as a ‘discouraged worker’ effect. For this reason, the literature included emphasises the importance of economic factors in the decision to attend school or university. Recent studies analysing the demand for university places are also investigated with a view to using them as the basis for modelling in this report.

In Modelling and Forecasting the Demand for Secondary Education (in Chapter 4) theory and previous work from other researchers are used to develop a model of secondary education participation. This model follows closely recent work undertaken in Australia, in that it uses economic variables to explain movements in various definitions of post-compulsory school enrolment. The results from this modelling are presented in full, and the model is used to generate forecasts for two observation points (two years). The forecasting accuracy of the model is assessed using standard statistical measures.

Modelling and Forecasting the Demand for University Places from School Leavers (in Chapter 5) records the development of a model that explains university commencements by school leavers. The most important explanatory variable in this model is that of a measure of upper secondary education participation. Other explanatory variables include starting salaries and unemployment rates for graduates under the age of 24. Models are developed for male, female and all school leaver university commencements. Findings from this modelling and an assessment of the forecasting performance of the model are all reported.

Modelling and Forecasting the Demand for Higher Education (in Chapter 6) reports the findings from the estimation of a model of higher education commencements (as a proxy for total demand) among Australian students in terms of total numbers and by field of study for both sexes. Data limitations leading to the pooling of data by sex, as well as the lack of a suitable data set for mature age and state modelling, is discussed, along with the problems of defining total demand for university places. The modelling of total higher eduction demand is confined to total commencements as a result of these problems, although the importance of this variable in determining overall final demand for higher education is recognised. Full results for the modelling of commencements in total and by field of study are presented. The forecasting performance of the model is tested in a series of out of sample forecasts over the short-term. An assessment is made of the applicability of the model to forecasting.

The Summary of Findings (Chapter 7) summarises points the authors wish to emphasise:

• The overview of the youth labour market and school educational participation isolates the coincident increase in school participation and decline in full-time employment among 15–19-year-olds. It is noted, that this relationship is not inconsistent with the accepted ‘discouraged worker’ explanation for the increase in school enrolment in Australia since the early 1980s.
• Given the presence of two strong and consistent trends in the youth labour market and in school participation, the findings from the literature review of previous studies into school participation focus on the link between economic conditions and schooling. A number of studies undertaken in Australia over the past two decades confirm that youth labour market conditions have had a large impact on the decision to remain at school.
• A brief summary of the findings from the modelling of upper secondary participation is provided. The preferred specification is one where a measure of school enrolment as a proportion of the relevant population is regressed against a group of economic variables, such as full-time and part-time employment population ratios, real junior wages, the level of government assistance to students and real household disposable income. The results from this modelling indicate that declining full-time employment and increasing income support for students have been the most important determinants of school participation in Australia over the last 25 years. Forecasting using this model suggests that it is suitable for making short-term projections of school participation;
• Trends in higher education show that the total number of students in Australian universities has increased by 148 per cent since 1974. A summary of trends in total and commencing student numbers by sex, age, level of course, state, type of enrolment, and overseas student numbers is provided;
• A brief discussion of previous literature in the demand for higher education places is undertaken. Emphasis is placed upon recent Australian research into the demand for higher education, including studies using scenario modelling and a Markov probability model. The previous literature is discussed in the context of the identified trends in higher eduction and general economic conditions.
• The findings from the modelling of university commencements by school leavers are discussed here. The most important finding is that school enrolment, or more precisely upper secondary participation is the most important determinant of university commencements among school leavers. Given the results from the school enrolment modelling, these findings suggest that changes in the youth labour market can have a significant impact on the number of teenagers commencing university in later years. Graduate unemployment rates are also significant in determining university commencements by school leavers, although graduate starting salaries are not.
• An assessment of the forecasting performance of the school leaver model indicates that estimated equations with the school participation or Year 12 retention rates are the models most suited to forecasting. The model for university commencements by all school leavers is the most accurate in this respect, with Mean Absolute Percentage Errors (MAPEs) over a two-year forecast horizon of .55 and .66 per cent in the two best forecasting specifications of that model.
• A model explaining university demand is estimated. Given the problems involved in identifying a measure of university demand, total commencements and commencements by field of study are modelled as a function of graduate starting salaries, unemployment rates and previous movements in commencements (total commencement model) or total commencements (commencements by field of study model). The findings from this research contend that the economic variables are only marginally significant in explaining changes in university commencements.
• The forecasting performance of the model over the short-term is assessed. The total commencements model forecasts reasonably well, as indicated by a test for predictive failure and MAPEs. From this, the model can be assumed to be suitable for short-term forecasting of the demand for university places. The field of study (FOS) commencements model forecasts are quite sound for disciplines with relatively large enrolments, such as Business and Arts. Smaller disciplines have larger MAPEs, and the model also does not forecast well for both sexes in disciplines where they are considerably under-represented, such as Engineering for females and Health for males.

Suggestions for future research arising from this study are outlined. They fall into two categories. First, there is a paucity of good data on what constitutes total demand for university places in Australia. The Australian Vice-Chancellors’ Committee (AVCC) has collected a measure of total demand since 1985; however, this data has several problems as it is not broken down by gender for the entire period and there are marked differences between each of the state education systems (a more complete analysis of the AVCC data set can be found in Appendix 2). However, it is worth noting that several data sources have been of use in the construction of variables. The GCCA survey of graduate destinations has been utilised in the development of variables measuring graduate employment prospects. The Australian Youth Survey provides some insight into youth employment and education patterns. A similar survey analysing factors affecting the decision to enrol in specific university courses and covering issues such as the Higher Education Contribution Scheme would be invaluable. The other aspect of analysing the demand for university places over the short-term is that of the choice of model. Future modelling of university commencements should include a number of socio-economic variables, especially when forecasting by field of study. This would, it is hoped, improve the accuracy of the modelling in these areas. Data collection in recent years by the Department of Employment, Education, Training and Youth Affairs will provide researchers with a larger set of observations for university commencements and enrolments by age, sex and course level in the near future.

[return to top] [Download Full Report]