The effect of abolishing the New Schools Policy on the provision of schools and enrolments in Australia

ABSTRACT

Australia has been an enthusiastic embracer of school choice and has one of the highest levels of public funding for non-government schools. The 1996 election of John Howard’s conservative government is a point of interest as it is often framed as a period of political and ideological attacks on public schooling. A key policy move was the abolishment of the New Schools Policy. The New Schools Policy withheld public funds from helping to establish new non-government schools in areas where there was already adequate supply. Using large-scale data sets on supply (number of schools) and demand (enrolment), time-series analysis demonstrates that while the non-government schools have grown (in real and market share terms), the abolishment of the New Schools Policy did not lead to a shock in enrolment numbers. However, the subtle shift from a needs-based to entitlement logic fundamentally altered the landscape of Australian school-based education ever since.

KEYWORDS:

• Australia
• New Schools Policy
• school choice
• non-government schools
• forecasting

Introduction

The February 2023 release of the annual Schools, Australia [Catalogue 4221.0] data set from the Australian Bureau of Statistics (ABS) indicated that enrolments in non-government (independent and Catholic) schools grew while public school enrolments fell for the second successive year (−0.6%, n = 16,929). This led to a series of mainstream (e.g. newspaper, television, online) and social (e.g. blogs, posts) media stories and posts speculating on the reasons. Claims of an exodus from public education are not uncommon in Australia, or globally, where for the past thirty or more years, school choice has been one of the primary policy levers (the others being autonomy and accountability) used by governments in their pursuit of greater effectiveness and efficiencies in the provision of education (e.g. Schütz, West, and Wöbmann 2007). That said, Australia presents an interesting case to investigate the shifting enrolment patterns between public and private schooling because:

• It is unique for its high levels of public funding for non-government schools (Caldwell 2010; Stewart and Russo 2001);

• Between-school segregation along social, economic and academic lines is pronounced (Maire 2021);

• There are high levels of choice, privatisation and competition (Perry and Southwell 2014);

• It has been dominated by policy incoherence (Ball 2019); and

• Sector (public, Catholic, independent) based debate is one of the most corrosive issues in school education (Thompson, Hogan, and Rahimi 2019).

With the increasing availability of large-N datasets on students, schools, and to a lesser extent achievement, there has been an increase in research on the economics of education (Pugh and Foster 2014) and an opportunity to test ideas using a scale and scope not previously possible. Building from Rowe (2020), this paper takes a particular historical moment – the election of John Howard’s conservative government in 1996, as a point of interest for understanding any shifts in the market share of government and non-government schools. The Howard era (1996–2007) is often framed as a period of political and ideological attacks on public schooling leading to what if often framed as the dismantling of public education systems (Smyth 2008; Wilkinson, Niesche, and Eacott 2019). Shortly after coming to power in 1996, the Howard government abolished the New Schools Policy, following through on a campaign promise of promoting greater choice in education. Introduced by the previous Hawke Government in 1985, the New Schools Policy withheld public funds from helping to establish new non-government schools in areas where there was already adequate supply (Connors and McMorrow 2015). Making it easier, whether real or perceived, for non-government providers to establish schools, is seen as foundational to the current stratification and segregation in Australian schooling.

Since the abolishment of the New Schools Policy, Australia has experienced declining performance in international tests (e.g. PISA, TIMSS), and widening disparity gaps between advantaged and disadvantaged students in national testing programmes (Thomson 2021). Using large-scale publicly available datasets on supply (the number of schools) and demand (enrolment) going back to 1956, time-series forecasting analysis (Autoregressive integrated moving average [ARIMA]) demonstrate that while the number of non-government schools has grown (in both real and market share terms), the abolishment of the New Schools Policy did not lead to a shock in enrolments as represented in market share. Instead, the subtle change in policy logic from needs-based to entitlement has fundamentally altered the landscape of Australian school-based education ever since.

Background

The policy architecture of Australian federalism makes education policy-making one of the most complex of government responsibilities (Keating and Klatt 2013; Savage 2020). Constitutionally, education is the responsibility of the states and territories however due to taxation policies, the Commonwealth retains the fiscal capacity to shape policy. Commonwealth interventions and particularly their underlying principles therefore have a considerable impact on the design of Australian school systems.

From parallel provision to dual system

British colonisers arrived in Australia in 1788 without any formally identified ‘teacher’ or ‘schoolmaster’ (Heffernan 2021). This allowed for multiple forms of schooling to emerge, primarily offered by churches (e.g. Church of England, Catholic, Protestant) but also including convicts becoming teachers, and wealthy landholders importing tutors from England. However, in the period 1872–1895 colonies (New South Wales, Queensland, South Australia, Victoria, and Western Australia) legislated free, compulsory, and secular education coupled with the withdrawal of public funding for non-government schools. Non-government schools did not cease to exist but instead operated in parallel to the public system. Access to non-government schools was limited to those with the financial resources to afford it. Government funding, or state-aid, was focused on the provision of school education to all citizens, without denying those with the resources to pursue alternatives.

An initial sign of change occurred in 1962 in what has come to be known as the ‘Goulburn incident’ in the state of New South Wales (Greenwell and Bonnor 2022). A poorly functioning toilet block and the absence of government funding to fix it, led to all Catholic schools in the diocese closing and an influx of students (approx. 640 with an equal number left without a place) into local public schools and overwhelming them. Bringing significant attention to the issue of state-aid, the Goulburn incident brought two principles to the fore: (i) the entitlement of non-government schools to public funding as a matter of non-discrimination and equal treatment of all citizens; and (ii) that the government had not only an obligation to help non-government schools but a financial interest to do so as it reduced government spending through cross-subsidisation of operating costs (Connors and McMorrow 2015). The timing was significant. As the post-WWII baby boom was swelling enrolments in government and Catholic schools, coupled with increased immigration to Australia, Catholic schools could no longer survive on the modest parental contributions (e.g. fees), and the free labour of nuns and brothers. The enduring viability of Catholic schools was being brought into question.

Capturing the political moment, the Commonwealth government saw an opportunity to engage with the current and aspiring middle class by supporting non-government schools through financial contributions. Constitutionally, education is the responsibility of states and territories. Commonwealth intervention in education represented a fundamental shift in the design principles of Australian school systems. Initially ad hoc, through capital grants (e.g. science laboratories, libraries), Commonwealth funding was systematised following the Karmel Report (1973) and the introduction of recurrent funding for government and non-government schools based on the principle of ‘needs based’. This approach was broadly accepted by major stakeholders and the public through until 1996 (Connors and McMorrow 2015). While the Whitlam years (1972–1975) focused on involvement in education as a national priority, the Fraser government (1975–1983) made incremental changes that favoured non-government schools as part of parental choice and market-based approaches to public services. These changes created the conditions for the New Schools Policy.

The New Schools Policy

In response to the 1984 Funding Policies for Australian Schools report the Commonwealth Government acknowledged the dual system of schooling in Australia and the need to plan for its development in a manner which promoted social harmony, educational cooperation, and the planned, efficient use of resources (Commonwealth Schools Commission 1984). It then followed up in the 1985 report Planning and funding policies for new non-government schools, articulating:

The allocation of public funds for new non-government schools should be consistent with the attainment by all schools of the resource standards necessary to ensure provision of a positive and effective education to their students, taking into account those differences between individuals and groups which tend to produce or to reinforce unequal chances of success in schooling in our society. (Commonwealth Schools Commission 1985, 6)

In addition, the States Grants (School Assistance) Act 1984 outlined the commitment of the Commonwealth to offer financial assistance to a new non-government school only after the Minister was satisfied that any new school is not likely to have a significant adverse effect upon the viability of existing government and non-government schools in the area. While difficult to assess the future impact of a new school, it shifted the burden of proof to the new non-government school to meet planning requirements, including being able to demonstrate that the proposed new school will not have a significant negative educational impact on existing government or non-government schools in the same area. Underlying this policy shift was a question of whether government can or should support a free market for the establishment of new non-government schools. At stake was the potential for wasting millions of dollars of tax-payer money through inefficient and ineffective provision of schooling (e.g. duplication, under-utilised assets). The New Schools Policy rejected the idea that any group of citizens had an entitlement to public funding for establish school provision (Connors and McMorrow 2015).

A shift in focus

As noted previously, building on campaign promises the Howard government quickly abolished the New Schools Policy. The Federal budget for 1996 made large concessions to the private school sector, removed the previous waiting period before a new school could be established, and increased subsidies for non-government schools (Potts 1999). Increasing the level of public money to non-government schools meant that economic resources became less of an inhibiting factor for families in pursuing choice in schooling. For some non-government schools, and particularly the Catholic sector, the new funding arrangements made it possible to operate with low fees and high subsidies – shifting their post-1970 approach and allowing for growth in their market share. The Howard government saw increasing public funding for non-government schools as an incentive for parents to move their children from a fully funded government system into a cross-subsidised non-government system reducing the costs of schooling for government (Connors and McMorrow 2015). However, from a systemic design standpoint, there was a cost as a family shifting from government to non-government schooling re-distributed the funding required (even if cross-subsidised) from state and territories to the Commonwealth. Commenting at the time, Marginson (1996) argued:

By abandoning the ‘new schools policy’ and increasing the grants to non-government schools, while taking the money for these off its allocation to Government schools, the Federal Government is creating a deregulated market in private schooling – but one that is heavily subsidised by the government and in a manner designed to induce a big shift of enrolments to non-government schools. (15)

In removing the threshold question of efficient use of public resources, it is fair to ask if abolishing the New Schools Policy led to a proliferation of non-government schools and enrolments. Howard government reforms are frequently cited as a catalyst for the proliferation of the non-government sector and the sustained under-funding of public schooling (e.g. Smyth 2008). For many, a particularly telling moment was when Howard labelled public schools as a ‘safety net’ existing for the purpose of guaranteeing ‘a reasonable quality of education in this country’ for those who cannot afford a non-government school (Armitage 2007, May 16). This was an explicit articulation of the belief that non-government schooling should be the desired form of provision and public schooling as the lesser – although still of reasonable quality – choice.

Arguments about the consequences of various policies are usually rhetorical and highly speculative as even with analytically sophisticated modelling, there is a lack of empirical evidence on how they play out. Capitalising on the time passed since the abolishment of the New Schools Policy, this paper tests the hypothesis that there will be evidence of greater supply of (measured by number of schools) and demand for (evidenced through enrolments) non-government schools since 1996. Expressed more formally:

H₁: The abolishment of the New Schools Policy led to a shock in provision through unprecedented expansion of non-government schools nationally.

H₂: The abolishment of the New Schools Policy led to a shock in enrolments for non-government school.

Taken together these hypotheses will provide crucial evidence as to the effect of abolishing the New Schools Policy. Consequently, the findings of this paper will allow for more targeted and tailored analysis of shifts in provision profiles and enrolment patterns since Howard era reforms and mechanisms that may, or may not, be required to redress them.

Methodology

Like most policy and reform research, this analysis is under-taken post hoc (Ladwig 2014). To test the hypotheses, time-series analysis and forecasting techniques have been employed using longitudinal data sets on the number of schools and enrolments nationally.

Data

The core data for this study were longitudinal Australian Bureau of Statistics (ABS) 4221.0 – Schools, Australia data releases. In some cases, the data were manually extracted from digitised archives or pdf reports (1956–1999) and since 2000, downloadable data cubes / pivot tables. The data was curated into a novel data set using Microsoft Excel, saved as a.csv file and eventually exported to Python 3.11 for analysis. Specific measures were the number of schools and enrolments (full-time equivalent), nationally, by state/territory, and by sector. Similar data (at least enrolments), was previously used by Rowe (2020) for assessing changes in public school enrolment overtime. Significantly, these are the official government records of school and enrolment numbers and with sufficient observations (65 years in total, 40 pre-intervention and 25 post) to allow for meaningful time-series analysis and forecasting. To provide comparability over time, the data was converted from raw numbers to a percentage of market share.

Forecasting

A common approach to time-series analysis in econometrics is an Autoregressive Integrated Moving Average (ARIMA) model. ARIMA is a class of linear models utilising historical values (e.g. number of schools) to forecast future values. Non-seasonal ARIMA models are generally denoted ARIMA(p, d, q) where parameters p, d and q are non-negative integers, p is the order (number of lags) of the autoregressive model (AR), d is the degree of differencing (the number of times the data have had past values subtracted), and q is the order of the moving averages (MA) model. Given that the number of schools change minimally by the seasons (compared to, for example, the availability and sales of fresh produce), there is no requirement to factoring in seasonality.

To establish the appropriate model was a multi-stage process. Initially the data was checked through a sequence chart to identify the trend in the data. There was a positive trend for non-government schools, and it was sufficient to warrant ARIMA modelling (see Appendix). After testing the stationary of data (e.g. Augmented Dickey-Fuller [ADF test]), autocorrelations were then used to make the data stationary. This is achieved by correlating observations against a time lagged version of itself, removing the trend and allowing for a constant mean based on similarity in data points. The order of differencing (d) can then be identified. The Autocorrelation Function (ACF), which plots the correlation coefficient against the lag, can be used to identify the q, and the Partial Autocorrelation Function (PACF), is used to identify the p. Resulting from this procedure, and checked with Python’s auto_arima function, the most appropriate model for the number of schools was an ARIMA(0,1,0) and for enrolments an ARIMA(0,2,0). The principal criteria for section, beyond the above-described process, was the lowest AIC, followed by MAPE and RSME. These indicators are negative orientated, so lower the better accurately for predicting future values. The details of this process can be found in the Appendix.

Structural break test

As the intervention year for the data set was already known (1996), and consistent with the goal of understanding potential changes in trajectory, a Chow test (1960) was used to test for a structural break in the time series data in 1996. Using the chowtest 0.1.4 package in Python, a Chow test determines whether the regression coefficients of each regression line (1956–1996, 1996–2021) are equal. If they are not equal, it means there is significant evidence of a structural break in the series.

Findings

Over the period 1996 to 2021 the total number of schools in Australia contracted from 9631 through to 9581 (−0.52% change). This has been experienced differently by sectors, with the number of government sector reducing by 5.60%, or specifically from 7089 to 6692 schools (n = 397), whereas the non-government sector has grown from 2542 to 2889 or 13.65% (n = 347). With this shift in the number of government and non-government schools there is a strong and negative correlation, r(64) = –.85, p < .001.

The first hypothesis of this project was to estimate the impact of abolishing the New Schools Policy on the provision of education nationally. This was assessed by looking at the number of schools (supply) as a percentage of market share by sector over time. Taking 1996 as the point of intervention, ARIMA forecasting procedures were employed to establish the predicted market share of schools (government and non-government) for the period 1997–2021 which could then be compared with the actual distribution. Figure 1 displays the results for the non-government sector (as this analysis focuses on percentage of market share, it is redundant to include both sectors). A Chow test (1960) indicates a structural break in the time series in 1996 (F = 7.772, p = <.001) between the actual and predicted values (MAE = 0.328, RMSE = 0.430, Percentage Error = 0.779, R²= 0.978). With the gap between the predicted and actual number of schools increasing from pre-intervention (MAE = .162, RSME = .264, Percentage Error = -.020, R²= .966) to post-intervention (MAE = .600, RSME = .611, Percentage Error = 2.090, R²= .532). Though the actual numbers remain within the confidence intervals of the forecast values and only 0.66% above predicted (n = 63 schools).

Figure 1. Number of schools (government v non-government), actual (⸺) and predicted (- -) (including upper and lower confidence bounds), 1956–2021.

Variation in the growth and contraction of government and non-government schools has not been experienced equally across all states and territories post-1996. Only Western Australia (WA, 4.97%) and the Northern Territory (NT, 4.83%) experienced growth in the number of government schools over the period 1996–2021. All other states and territories saw a decline ranging from 21.81% in South Australia (SA) through to New South Wales (NSW) at 1.37% (Tasmania [Tas] = 16.59; Australian Capital Territory [ACT] = 10.10; Victoria [Vic] = 9.00; and Queensland [Qld] = 5.25). The consolidation of public assets, and the efficient use of resources has led to the reduction in government schools in many states and territories. Goals of reducing variance and inefficiencies in the system means that many small schools have shut or merged to achieve economies of scale within systems.

In contrast, all state and territories saw growth in the non-government sector, ranging from the Northern Territory with 44.44% to Tasmania at 4.20% (Qld = 30.98; WA = 21.57; ACT = 17.50; NSW = 10.25; SA = 6.70; Vic = 6.04). It is this growth in non-government schools and decline (in most states and territories) of government schools that legitimises the perceived shock to the system nationally. Growth in the non-government sector is difficult to describe. The percentage is often reflective of the size of operations within a state or territory. There is also substantial heterogeneity in the sector ranging from long established independent schools through to more recently established (and under ongoing viability stresses) schools. Some of the most substantial growth has been in start-up alternative schools such as Steiner and Montessori (Eacott and Wainer 2023)

Visually representing these changes in school numbers in education research has traditionally been limited to charts (line, pie, or column) or superimposing data on to a map of the country. The former simply transforms the data from a table into a chart requiring substantial reader interpretation, and the latter foregrounds physical geography rather than the strength of data. As an alternative to such orthodoxy, Figure 2 represents the growth of market share for non-government schools at the state or territory level based on intensity (size of the change) rather than physical geography. Extending an historical trend, all states and territories experienced a real increase in the market share (as measured in number of schools) of the non-government sector.

Figure 2. Percentage change in market share in the provision of schooling for non-government schools by state / territory, 1996–2021.

Figure 3 displays the ARIMA forecast model for the distribution of enrolments as a percentage of market share for the non-government sector over the period 1956 to 2021. Unlike the number of schools, the predicted enrolment share is below the predicted, although within the upper and lower confidence bounds. The 2021 variance is 6.68% below the predicted. This degree of variance represents a gap of 268,823 students.

Figure 3. Distribution of enrolments as percentage of market share for non-government schools (actual and predicted, including upper and lower confidence bounds), 1956–2021.

A Chow test conducted on the entire sample indicated a structural break between the actual and predicted values in 1996 (F = 120.47, p = <.01, MAE = 0.757, RMSE = 1.795, Percentage Error = −1.908, R²= 0.865). With the gap between the predicted and actual enrolments by sector increasing from pre-intervention (MAE = 0.129, RMSE = 0.176, Percentage Error = −0.013, R²= 0.995) to post-intervention (MAE = 1.787, RSME = 2.907, Percentage Error = −5.060, R²= .350).

Given the diversity of state and territory demographics, it is not surprising to see the distribution of enrolments experienced differently in the various states and territories. Figure 4 visually displays the difference in market share for non-government school enrolments between 1996 and 2021. Over this period, South Australia (−2.27% growth, a loss of 4033 students) and Tasmania (−11.05%, a loss of 6937 students) were the only states where the absolute number of enrolments in government schools contracted. All other states and territories saw growth (Qld = 39.39; WA = 30.14; Vic = 2.60; ACT = 15.35; NT = 8.53; and NSW = 5.54). As with the number of schools, all states and territories have seen an increased (both absolute and relative) enrolment in non-government schools since 1996. This growth ranged from Queensland’s 87.88% to Tasmania’s 19.89% (WA = 75.08; SA = 45.48; NSW = 44.45; Vic = 40.84; ACT = 36.22; and NT = 35.14).

Figure 4. Percentage change in market share of enrolment for non-government schools by state / territory, 1996–2021.

Taken together, while there has been higher than predicted growth in the number of non-government schools following the abolishment of the New Schools Policy, enrolments have not matched predictions.

Discussion

Australia has a lengthy history of government and non-government school provision, and long-held public support for state-aid to non-government schools (Sherman 1982). The analysis presented in this paper has demonstrated that the Howard government abolishing the New Schools Policy tracks an increase – above forecast projections – in non-government schools but had less impact on enrolments. However, the growth of schools and enrolments in real terms and market share legitimises concerns of exponential growth in the non-government sector. OECD (2019) data indicates that 94.4% (1.0 S.E., OECD average = 76.9, 0.4 S.E.) of Australian principals identify as working in a school competing with others for enrolments. Rather than previous parallel or dual systems of school provision, competition for enrolments is evidence that non-government schools are not just serving distinct sub-groups within society but rather competing with government schools for not just enrolments but also the matching state-aid. Analysis of such a system requires nuance of the principles underlying changes and not a priori judgement and imposition of normative assumptions.

Brighouse (2000) argues that the support for school choice is usually over-enthusiastic and the opposition over-critical. Both advocates and critics impose endogenous criteria to assess reforms crushing the possibility of anything other than confirmation of a pre-existing normative orientation. This has been a major constraint to the research and public discourses on school choice throughout the world, and in particular Australia. Taking a systemic lens, it is not the presence of differentiated provision (e.g. multiple school sectors) that is problematic. People sort into schools, communities, suburbs, and places of work in ways that make it difficult to disentangle self-selection effects from actual (causal) effects of school choice (Sacerdote 2014). Critique or advocacy of choice reforms need to identify violations of broader principles defined in ways not limited to ‘choice’.

Returning to the principle of the New Schools Policy, the focus of assessment was the efficient use of public resources. Specifically, the Hawke government’s Planning and funding policies for new government schools cited that distribution of public resources should be ‘consistent with the attainment by all schools of the resource standards necessary to ensure provision of a positive and effective education to their students’ (Commonwealth Schools Commission 1985, 6). In addition to abolishing the New Schools Policy, the Howard government reworked the funding mechanisms for schooling. Following a 2001 change recalibrating base funding, the Howard government guaranteed that any non-government school negatively impacted would have its funding maintained in real terms. By 2011 – a decade later – there were over 1000 (out of 2617) non-government schools funded above their entitlement (Greenwell and Bonnor 2022).

School funding is a principal site of policy reform (Gerrard, Savage, and O’Connor 2017). Currently, and employing consistent language from the past, funding policy draws on the ‘School Resourcing Standard’ or SRS. This value is an estimate of how much public funding a school needs to meet its students’ educational needs and was developed based on recommendations by Gonski et al. (2011). The SRS includes four student loadings (disability, Aboriginal and Torres Strait Islander, socio-educational disadvantage, and low English proficiency) and two school-based loadings (size and location). Most non-government schools are subject to a Capacity-To-Contribute (CTC) reduction in their base funding level (no discount is accrued on loadings) based on the direct measure of income reported by parents and guardians of students at the school. Courtesy of Australia’s unique policy architecture, the Commonwealth provide 80% of the SRS to non-government and 20% to government schools, with the remaining coming from state or territory governments. Recent analysis however shows that government schools are funded to 87.1% of the SRS and non-government schools at 105.9% (Cobbold 2023). Under current arrangements, schools funded below their target will transition to that target by 2023 and those above their target will transition by 2029. Returning to the principle of the Hawke reforms, distortions in funding at the SRS level invites questions regarding the efficient use of public resources.

The public returns on non-government schooling have long been debated (e.g. Williams and Carpenter 1991), and yet, choice and markets have been a simple and compelling message for governments and reformers targeting improved outcomes (MacLeod and Urquiola 2019). Across time, countries, and outcome measures, non-government schools have outperformed government schools in absolute terms (Coulson 2009; Marks 2015; 2017), but less so – if at all – when accounting for composition effects and focused on value-added measures (Larsen et al. 2022; Marks 2015; Miller and Voon 2012; Rodgers, Neri, and Moran 2016). It is worth noting that it was not until the School Assistance Act 2008 (Cth) that the Commonwealth government explicitly conditioned state aid to non-government schools based on compliance with detailed accountability requirements (Zehavi, 2011). Once this data was in place, an analysis of the efficiency of Australian schools (e.g. input-output analysis of growth in national testing scores) found that on average, at the primary level non-government schools are less efficient than public schools, but this efficiency reverses at the upper end of secondary schooling (Nghiem, Nguyen, and Connelly 2016).

If non-government schools are no better than public schools for value-added measures this raises questions regarding the most efficient use of public resources (e.g. taxpayer money). A fundamental shift in the policy positions of the New Schools Policy and Howard reforms was a move in the underlying principle from needs based to entitlement. The former being an outcome variable and the latter an input variable. Policy decisions became focused on individual entitlement rather than systemic outcomes, and the effects of this shift have been felt for decades.

Abolishing the New Schools Policy made it easier for groups to segregate through schooling with the assistance of government. Increasing funding for non-government schools served as a form of subsidised family wealth, deliberately embedding segregation based on socio-economic status into the provision of schooling nationally. Removing public funding from non-government schools would not irradicate the non-government sector (as demonstrated in the late 1800s and early 1900s). Small operations would struggle, particularly those operating as low fee – high subsidy schools, but more established schools would simply become more exclusive, amplifying segregation and inequities of access to differentiated provision. Abolishing the New Schools Policy was symbolic of a shift in policy principles from needs based to entitlement.

Conclusion

The idea that non-government schools would come to dominate the Australian school education system has been an enduring concern. Caldwell (2010) went as far as to suggest that private schooling would become the (at least in secondary) major provider. Based on the evidence available to this point, the greatest provision of schooling has not flipped from government to non-government. However, growth in both real and market share terms across all states and territories legitimises the belief of exponential growth. The greater shift in policy was not so much the abolishment of the New Schools Policy but instead the funding shift from needs based to entitlement. Creating a funding environment that grants primacy to equality (an input variable) and then seeks to achieve equity (an outcome variable) through compensatory loadings. This is a particularly difficult agenda to wind back in a cultural context of individualism, choice, and aspiration. Removing the New Schools Policy meant the principle of the efficient usage of public resources and the threshold of all schools achieving the resourcing standard no longer applied. The shock to the system was therefore not in the number of schools or enrolments but the re-casting of the underlying generative principles of how schooling was to be maintained by governments. It is only by attending to the principles by which policy is generated is it possible to prosecute the case for viable alternatives.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Notes on contributors

Scott Eacott

Professor Scott Eacott leads an interdisciplinary research program concerned with the intersection of the teacher shortage, the housing crisis, and the organisation of education. His work seeks to develop tools for government, stakeholders, systems, and educators to better understand how best to meet legal, social, and cultural expectations in the provision of education.

Appendix. Forecasting models

Non-government schools

Initial analysis was an Augmented Dickey-Fuller (ADF) test, conducted using Python 3.11. Using the original, pre-intervention (1956–1996) data, the ADF was 0.563452 (p = 0.986679). Therefore, we reject the null hypothesis and need to difference the data set. Repeating the ADF for the full (1956–2021) data set returned a 0.449594 (p = .983258).

Testing for stationary fell between 1 and 2. To overcome under- or over-differencing, auto_arima was used to test all models using the principal criteria of the lowest AIC, followed by MAPE and RSME.

Table

ARIMA Model	AIC	Stationary R-squared	R-squared	RMSE	MAPE	MAE	MaxAPE	Normalised BIC	Preferred model
0,1,0	11.928	−4.441E-16	.967	.271	.719	.167	4.912	−2.522	Y
1,1,0	13.577	.009	.968	.273	.713	.165	4.952	−2.413	N
0,1,1	13.467	.012	.968	.272	.712	.165	4.933	−2.416	N
1,1,1	14.757	.029	.968	.274	.715	.166	4.810	−2.315	N
0,2,0	35.822	−6.661E-16	.939	.369	1.027	.238	5.521	−1.902	N
1,2,0	29.841	1.45	.948	.345	1.012	.235	5.553	−1.938	N
0,2,1	16.353	.399	.963	.290	.819	.189	5.238	−2.290	N
2,2,0	26.053	.266	.955	.324	.933	.216	5.086	−1.970	N
3,2,0	26.355	.299	.957	.322	.904	.209	5.105	−1.894	N
2,2,1	19.379	.410	.964	.295	.815	.189	5.120	−2.066	N
1,2,1	–	.396	.963	.294	.817	.189	5.237	−2.165	N
3,2,1	21.379	.410	.964	.299	.814	.189	5.139	−1.942	N
2,2,2	21.045	.414	.964	.298	.819	.190	5.078	−1.950	N
1,2,2	19.074	.413	.964	.294	.816	.189	5.113	−2.070	N
0,2,2	-	.398	.963	.294	.817	.189	5.201	−2.167	N
1,2,3	21.043	.410	.964	.299	.818	.190	5.114	−1.943	N
0,2,1	16.353	.399	.963	.290	.819	.189	5.238	−2.290	N

Non-government enrolments

Initial analysis was an Augmented Dickey-Fuller (ADF) test, conducted using Python 3.11. Using the original, pre-intervention (1956–1996) data, the ADF was −0.852046 (p = 0.803354). Therefore, we reject the null hypothesis and need to difference the data set. Repeating the ADF for the full (1956–2021) data set returned a −0.286042 (p = .927458).

Testing for stationary fell between 1 and 2. To overcome under- or over-differencing, auto_arima was used to test all models using the principal criteria of the lowest AIC, followed by MAPE and RSME.

Table

ARIMA Model	AIC	Stationary R-squared	R-squared	RMSE	MAPE	MAE	MaxAPE	Normalised BIC	Preferred Model
0,2,0	18.504	.000	.995	.184	.581	.137	2.061	−3.295	Y
1,2,0	17.402	.023	.995	.184	.558	.133	2.082	−3.198	N
0,2,1	17.390	.023	.995	.184	.560	.133	2.085	−3.198	N
1,2,1	15.408	.023	.995	.187	.558	.133	2.084	−3.077	N