There is general consensus amongst scholars and the wider public that education plays an essential role in modern societies and this shared appreciation of the importance of education has resulted in a high level of investment in systems of schooling. Economists point to the important role that education plays in the accumulation of human capital, which is a critical component for economic growth. Indeed, some economists have asserted that the health of a nation's education system is akin to an early warning device for the future health of the nation's economy (Hanushek 2006). Sociologists, on the other hand, often point to the critical role that education plays in transmitting social capital (Teese 2005). To monitor the effectiveness of educational institutions, many jurisdictions conduct or participate in regular testing for key domains of learning (for instance, the Programme for International Student Assessment (PISA) conducted by the Organisation for Economic Co-operation and Development (Abalde 2014)). Jurisdiction-wide standardised testing and reporting of student and school academic achievement are often controversial and criticised by teachers, in particular, but they do 'tend to capture what parents and voters in a school district ultimately care about' (Andrews et al. 2002).
In principle, it should be possible to employ this data on school performance (both economic and academic) in empirical research to identify the determinants of school success, and thus point the way to efficacious public policy. However, in practice, problems with the data, empirical models and disparate conceptual approaches of scholars have largely thwarted evidence-informed public policy (Marks 2015).
The issue of optimal school size is a good illustration of the problems that are encountered by those wishing to formulate evidence-informed public policy. In many jurisdictions, school-level data are not available, which means that analysts either neglect the issue of size entirely or are forced to work at a level for which data can be obtained (often the district level) (Andrews et al. 2002; Marks 2014). Other times, the model specification has been unsatisfactory--for instance, some of the extant empirical work fails to allow for likely parabolic associations and is therefore incapable of accurately reflecting the theoretical expectations regarding the association between size and performance (both economic and academic) (e.g., Leithwood and Jantzi 2009). Indeed, much of the actual policies implemented have been formed and executed in an evidence vacuum and have, not unsurprisingly, proven largely ineffective (Hanushek 2003). Even when data availability and model specification have been appropriate, very disparate conceptual approaches to the question of optimal school size have produced quite contrary evidence.
The two main approaches employed by scholars to investigate the optimal size of schools are the economic and ecological perspectives. The economic approach is generally concerned with the association between school size and unit cost and is pursued for the purpose of arriving at the most technically efficient (defined as the optimal conversion of inputs into outputs) size of education institutions (Lee and Loeb 2000). The evidence arising from studies of this type has led to a policy preference for relatively large schools capable of capturing economies of scale in production process (economies of scale refer to an economic concept whereby the average total costs of producing some goods is expected to initially decrease in response to expanded output) (Crosnoe et al. 2004). The ecological perspective, on the other hand, focusses on how the size of schools affect the number and types of interactions between students, teachers and communities (Lee et al. 2001). The conclusion of this stream of research has generally been that smaller educational institutions are less bureaucratic, foster more and a better quality of interactions and hence tend to be conducive to improved education outcomes (Lee and Loeb 2000; Leithwood and Jantzi 2009). Rarely have scholarly studies employed both approaches and hence attempted to discover whether there is a school size (or indeed a range of school sizes) that might strike a balance between minimising economic inputs while maximising academic achievement outputs (Crosnoe et al. 2004). Demonstrating how the two disparate approaches yield quite different recommendations for optimal size, and hence public policy prescriptions, is the principal aim of this paper.
It is easy to see how failure to examine simultaneously the question of school size from both the economic and ecological perspectives can easily result in multiple public policies working at cross-purposes. The risk of this sort of undesirable outcome occurring would seem to be further magnified where different tiers of government are involved in formulating and implementing public policy. For instance, one can imagine how federal governments might implement public policy aimed at increasing educational standards in order to improve the nation's economic welfare but be thwarted in their objective by state governments, which may avoid spending on new schools in order to alleviate their own budgetary pressures. Indeed, if the respective public policymakers are not even aware that school size might be a determinant of performance (both economic and academic)--because there is a vacuum of empirical evidence relating to the association--then it may not even be apparent to stakeholders that the tiers of government are working at cross-purposes. As we will detail later, this is essentially the set of circumstances in play in Australia, which is the context for our empirical work that follows.
To investigate the disparate public policy recommendations arising from economic and ecological approaches respectively, we first outline the theoretical foundations for why school size might matter. Thereafter we describe the system of secondary public schooling in Australia and the National Assessment Program in Literacy and Numeracy that forms the context for our empirical work. We also outline the panel regression strategy that we employ to determine optimal school size for urban public schools from both economic and ecological perspectives. Following this, we present the results from our various empirical estimations and outline the range of optimal school sizes suggested by each of the conceptual lenses. We conclude with an investigation of the public policy levers that might be used to optimise school size.
WHY SIZE MATTERS
Student population size has particular salience for urban schools, because this is where population pressures are most keenly felt (Lee and Loeb 2000) and also where public policy architects have greatest flexibility to introduce interventions such as school consolidations (due to relatively small distances between existing education campuses). Moreover, it is important to carefully define the level of schooling that one is focussed on investigating because the effects of size are likely to be different for primary and secondary schools, respectively. In the discussion that follows we concentrate on the implications of size for secondary schools because these institutions are more likely to be affected by matters such as breadth of curriculum, teacher specialisation and the like that are detailed in the extant literature. Public (government) schools are the focus for our study given that public policymakers have greater leverage in this portion of the sector for significant reforms in connection with optimising school size (such as school consolidation which is mostly not an option for private education providers given the relatively less dense footprint of their campuses). We first review the theoretical foundation for the economic approach (which deals with the question of how school size is related to per student expenditure) to urban public secondary school size, before turning to the literature on the ecological approach (which looks at the association between school size and educational achievement).
Economic Approach (School Size and Per Student Expenditure)
There is a large literature that examines the economics of education production in a general sense--how inputs are related to outputs, and the sort of incentives that might be introduced to improve performance (such as merit based pay, privatisation of some education functions, and fostering choice for parents) (Hanushek 2003). However, the literature regarding the association between school size and unit cost is much narrower and focussed on American studies (Abalde 2014). In a thorough survey of the literature since 1990, Leithwood and Jantzi (2009, p. 480) report mixed evidence from five studies where 'two of these studies report results favouring large schools, two favouring small schools and one favouring mid-size schools'. In an earlier review of the literature, Andrews et al., (2002) surveyed ten studies of which eight presented evidence consistent with economies of scale (and one presented evidence suggesting constant returns to scale). The relative scarcity of economies of scale studies in education is due to problems of data availability (often only district enrolment data is available; Leithwood and Jantzi, 2009). Moreover, a good proportion of the extant studies have been criticised for perceived specification problems--principally the failure to include a squared term for school size required to identify the turning point for 'U-shaped' functions (Andrews et al., 2002). In sum, the existing literature is not conclusive given the narrow context of studies and perceived specification problems and hence it is fair to say that a consensus does not exist in the literature regarding the association between school size and efficiency.
In the language of neo-classical economics, the term 'economies of scale' refers to the situation whereby...