School output measures in the Australian National Accounts: experimental estimates, 2007-08 to 2017-18

This paper describes new experimental indexes for the volume of output of Australian schools

Released
28/09/2020

Nirmala Narayan (Economic Statistics Research Section)¹ ²

Introduction

The output of school education services in Australia is substantial. In 2016/17, educational services provided by schools (primary, secondary, pre-schools and special needs education) contributed 52% of education industry output.³ The vast majority of this output was produced by primary and secondary schools, which constitute the scope of the ‘school education’ services in this paper. The total number of enrolments in school education (primary and secondary) was 3.9 million in 2018, with 65.7% of students enrolled in public schools.⁴

The ABS currently estimates the volume of school activity by weighting together changes in Full Time Equivalent (FTE) student numbers for preschools, primary, secondary and special schools sourced from Schools, Australia (ABS Cat.no. 4221.0). The weights are derived from Australian National Accounts: Input-Output Tables (ABS Cat.no. 5209.0.55.001) and are updated periodically.

The new experimental method offers two enhancements to this approach. First, the weights were derived from directly observed data, which can be updated annually and applied directly to enrolments data to implicitly capture some aspects of quality variation and change. Second, the calculation of volume measures has been undertaken at a lower level of disaggregation than is currently used in the National Accounts. This enables the production of estimates for public and private schools, and schools in different states and territories. These enhancements have the potential to materially improve the output volumes and current price estimates for education statistics in the National Accounts.

The experimental indexes in this paper are based on a range of datasets, including data from the Australian Curriculum and Reporting Authority (ACARA).⁵ ABS data is also used, including data from: Schools, Australia (ABS Cat. no. 4221.0); unpublished data relating to Government Finance Statistics, Australia (ABS Cat. no. 5512.0) and data from the Economic Activity Survey (EAS) which is published as part of Australian Industry (ABS Cat. no. 8155.0) estimates.

The ABS invites feedback on the experimental methods proposed in this paper. For further discussion of these methods please contact the ABS at economic.research@abs.gov.au.

Footnotes

  1. I am extremely grateful to Kristen Stone, Jason Annabel, the Education and Training, Annual Industry Statistics, and Government Finance sections, and the National Accounts team within the ABS for their ready guidance, support and cooperation.
  2. The experimental indexes proposed in this paper would not have been possible without the data and assistance provided by Australian Curriculum and Reporting Authority - ACARA. The author thanks them for this assistance.
  3. These estimates are contained in Australian National Accounts: Input-Output Tables (Product Details), 5215.0.55.001
  4. Schools Australia publication 4221.0
  5. https://www.acara.edu.au/reporting/national-report-on-schooling-in-australia/national-report-on-schooling-in-australia-data-portal
     

Overview of education services in Australia

All states and territories provide 13 years of formal school education. Typically, schooling commences at age five, is compulsory from age six until age 16 (with provision for alternative study or work arrangements in the senior secondary years), which translates to school grades pre-Year 1 to Year 10, and is completed for most students at age 17 or 18. Age requirements vary across the states and territories.

In most states and territories, primary education comprises a pre-year 1 grade followed by Years 1 to 6, and secondary education comprises Years 7 to 12. Until 2015, South Australia, Queensland and Western Australia classified Year 7 as part of primary school education. In 2015 Queensland and Western Australia reclassified their Year 7 students into secondary school; South Australia commenced this reclassification in 2019, and it is expected to be completed by 2023.

As school education is compulsory from pre-Year 1 to Year 10, the number of students in the school system in a given year is primarily driven by demographic changes in each state/territory.

The regulation and funding for schools in Australia is primarily the responsibility of States and Territories. However, the Australian (Commonwealth) government also plays a funding role. Fees, charges and other parental and private contributions provide the final component of school funding.

Education services are provided through public and private schools, with the latter receiving a greater proportion of their funding through parental and private contributions. Government recurrent funding for school education services in 2018 was $61.5 billion, with 75.8% provided to public schools.

A number of private schools provide primary and secondary school education services in combined schools, covering all 13 years of school education.

The Australian Curriculum and Reporting Authority (ACARA) is a regulatory authority involved in school education services. ACARA is a statutory body established in 2009 to develop school curriculum, national assessment and reporting on school education activities.¹

Footnotes

  1. Kindergarten in New South Wales and the Australian Capital Territory, Preparatory in Victoria, Queensland and Tasmania, Reception in South Australia, Pre-primary in Western Australia, Transition in the Northern Territory, and Foundation Year in the Australian Curriculum (Report on Government Services 2018 - PART B CHAPTER 4, page 4.34)
  2. https://www.acara.edu.au/docs/default-source/default-document-library/national-report-on-schooling-in-australia-2018.pdf
  3. https://www.abs.gov.au/AUSSTATS/abs@.nsf/allprimarymainfeatures/E1759367BE2FD4AACA2570AF003E058F?opendocument
  4. Ibid footnote 7
  5. Ibid footnote 7

Scope and classification of education services

The classifications considered in this paper have been determined based on National Accounts requirements, hence variables such as ‘remoteness’, although of high relevance and interest, have not been included in this analysis.

Industry classification of education services

The indexes proposed in this paper cover primary, secondary and combined school education only. They correspond to the Australian and New Zealand Standard Industrial Classification (ANZSIC) classes 8021, 8022 and 8023.

Education services provided in primary, secondary or combined schools are jointly referred to as ‘School education’ services in this paper.

Preschool education and special school education are not included in the experimental indexes presented in this paper, but are being considered for future extensions to this work.

Market and non-market classification of schools

Non-market output occurs when goods and services are provided free of charge, or at prices which are not economically significant. These are prices that have no significant effect on the amount that producers are willing to supply and the amounts purchasers wish to buy.¹¹ The ABS examines each producing unit in the economy to determine if it charges economically significant prices for its output. Units that predominantly charge prices that are not economically significant are classified as non-market producers.

Private schools, comprised of Independent and Catholic schools, are considered to be market-based entities, and are treated as non-financial corporations. For the purposes of this paper they are considered to be market producers.

Public schools deliver education services at prices which are not deemed to be economically significant and hence are considered to be non-market in nature. As public schools are entities under the control of state/territory governments, they are classified to the general government sector in macroeconomic statistics.

More information on the institutional sector classification used in Australia can be found in Standard Economic Sector Classifications of Australia (ABS Cat. no. 1218.0).¹²

Measuring non-market activity is difficult because of a lack of meaningful prices. In the absence of meaningful prices, it is difficult to weight the output of different types of products appropriately. This in turn creates difficulties in measuring productivity.

For further information about the conceptual framework which underpins the measurement of non-market output, please see Non-market output measures in the Australian National Accounts: a conceptual framework for enhancements.

As a result, the ABS currently excludes industries that have considerable non-market economic activity within them, including school education, from its published multifactor productivity statistics.

Footnotes

  1. 2008 SNA, paragraph 4.18, see https://unstats.un.org/unsd/nationalaccount/docs/SNA2008.pdf.
  2. ibid footnote 8

Data sources and adjustments

To compile a cost-weighted volume index, two forms of data are required; a quantity indicator and a corresponding indicator to capture the cost of production.

A quantity indicator

The ideal quantity indicator would be total hours of education provided by a teacher to the group of students receiving that service, which is equivalent to the total hours of service received by students for a particular product. In this case the ideal product type would be individual year levels of education, so that services are most homogenous within product types and heterogenous across them. Owing to data limitations, the closest approximation available to the ideal quantity indicator is the number of students measured on an FTE basis.

FTE school enrolments data are available from Table 43a of Schools, Australia publication. These data are collated from administrative school enrolment databases, and available by level of school education (primary or secondary) in each state/territory, split into public and private schools, and available from 2007 to 2018.

A cost of production indicator

Market prices are available for private school education services provided in Australia, and these could be used in private school output index compilation. However, as there are no equivalent prices available for non-market education services, cost weights must be used to aggregate public school education output. To enable comparability across indexes for private and public schools in this first phase of analysis, cost weights were also used to aggregate private school education output. The cost data used was the operating expenditure incurred in providing the services: wages and salaries, use of materials, energy and other services, and depreciation.

The costs incurred in the delivery of primary and secondary education in public schools within each state/territory are available from the ABS Government Finance Statistics dataset. This dataset is based on information contained within the public accounting databases of each state/territory government. Data covering the financial years 2007-08 to 2017-18 were available.

Cost estimates for primary, secondary and combined school education for private schools are available from the Economic Activity Survey (EAS) dataset also covering 2007-08 to 2017-18. These data are derived from a sample-based survey and are robust at the national Industry subdivision level, but not at the primary/secondary or state/territory level of disaggregation.

To derive primary/secondary by state/territory cost splits for private schools, ACARA data was used to proportion the national estimates from the EAS dataset. The ACARA dataset did not include operating expenditure for private schools, but income was available. Income for private schools included revenue from government as well as parental and private contributions. Income was used as an auxiliary variable to derive lower level estimates from EAS expenditure data as these variables were highly correlated at the National level.

Stratification of school services

Separating the public and private systems of education and primary from secondary education into different strata enables some of the differences between these strata to be captured in the output measure.

It can cost more to deliver an education service to a student in a private school than in a public school.¹³ It also costs more, on average, to deliver an education service to students in secondary schools compared to primary schools, where there is a need for a wider range of facilities, such as science laboratories. This has the effect, on average, of (a) increasing the weight of private school enrolments versus public school enrolments in the aggregate index and (b) increasing the weight of secondary school enrolments versus primary school enrolments, compared to using a simple FTE or headcount of total enrolments.

Deriving lower level cost estimates from EAS data

Using ACARA enrolments data, which is available by individual school year level, it was possible to estimate ACARA income data splits for combined schools (Catholic and Independent), for each state/territory. Adjustment factors based on average income per student were then applied to these estimates to account for known cost differences between primary and secondary education. The adjusted estimates were then added to the non-combined primary and secondary income totals for each stratum. Next, Catholic and Independent school income totals were collapsed and used to derive income proportions for private primary and secondary schools in each state/territory. These proportions were then applied to total operating expenses in the EAS dataset to derive primary/secondary school by state/territory estimates of expenditure for private schools. This process is illustrated in Figure 1.

    Figure 1: Disaggregating EAS Expenditure data

    Figure 1: Disaggregating EAS Expenditure data

    Figure 1: Disaggregating EAS Expenditure data

    This image is a flow chart diagram beginning on the left with ACARA income data (Catholic/independent by state/territory) which is grouped by Non-combined Primary Schools, Non-Combined Secondary Schools and Combined School.

    This then flows into Derived ACARA income estimates (State/territory) which is grouped by Primary schools and Secondary schools.

    This then flows into Derived ACARA income proportions (Cell proportions) which shows a table style graphic for Primary and Secondary schools and a % placeholder for their percentage for State/territory.

    This then flows into Total EAS Australian private school education expenditure.

    The final flow is into Disaggregated EAS private school expenditure (State/territory) which is combined with another table style graphic for Primary school costs and Secondary school costs and a placeholder for their State/territory.

    A similar adjustment was not required for public schools, as the cost data obtained from the Government Finance Statistics dataset already contained the required level of detail.

    Aggregate estimates of school income compiled by ACARA at the national level align closely with the EAS estimates of operating Total expenses and hence were considered to be suitable for this purpose.

    However, ACARA income data was only available for financial years 2009-10 to 2016-17 (i.e., calendar years 2009 to 2017), so the ACARA proportions above for 2009-10 were applied back to 2007-08 EAS data, and the 2016-17 ratios were applied to the corresponding 2017-18 data to achieve the required time series.

    Adjustments applied to calendar year data

    Enrolments data and the ACARA income data for the analysis are available on a calendar year basis, whereas EAS data are available on financial year basis; Australian National Accounts also compile their data on a financial year basis. To meld the data from these different sources and to suit National Accounts compilation the ACARA estimates were converted to a financial year basis by averaging the data for each consecutive ‘pair’ of calendar years. A similar treatment was applied to the FTE data from ABS Schools Australia publication which is also available on calendar year basis only.

    The data sources used to generate the experimental indexes in this paper are summarised in Figure 2.

      Figure 2: Data sources

      Figure 2: Data sources

      Figure 2: Data sources

      This image is a table diagram that explains which data source was used to generate the experimental estimates in this paper.

      For Public and Quantity (FTE) (school level by jurisdiction) - ABS School Australia publication* was used.

      For Private and Quantity (FTE) (school level by jurisdiction) - ABS School Australia publication* was used.

      For Public and Expenditure (School level by jurisdiction) - ABS Government Finance Statistics (unpublished data) was used.

      For Private and Expenditure (School level by jurisdiction) - ABD Economic Activity Survey (unpublished data) and ACARA* were used.

      * Data converted to financial year

      Footnote

      1. This is subject to variation across school level, school sector (Catholic, Independent), time periods and jurisdictions.

      Volume index calculated method

      Variables used to stratify student enrolments and related expenditure data were:

      • school affiliation (i.e. public or private);
      • school level (i.e. primary or secondary); and
      • jurisdiction (i.e. state/territory).

      Combining the number of students enrolled (FTE) in each school level within each jurisdiction with the corresponding cost incurred in providing those education services yields ‘Level 1’ cost-weighted indexes. These are:

      • public schools – primary education;
      • public schools – secondary education;
      • private schools – primary education; and
      • private schools – secondary education.

      Aggregating the ‘Level 1’ primary and secondary indexes within each school affiliation with their corresponding cost weights yields ‘Level 2’ indexes. These are:

      • public schools – total; and
      • private schools – total.

      Aggregating the ‘Level 2’ indexes for public and private schools with their corresponding cost weights yields ‘Level 3’ index which describes growth in the volume of education output provided by all primary and secondary schools in Australia.

      The indexes are chained progressively at each level. This process is illustrated in Figure 3.

        Figure 3: Compilation process

        Figure 3: Compilation process

        Figure 3: Compilation process

        This image is a flowchart diagram.

        The first box indicated as Input data and contains Student Quantity and Expenditure Data (for aggregation weights): Affiliation by school level by jurisdiction and lists NSW, VIC, QLD, SA, WA, TAS, NT, ACT.

        This flows down into the Level 1 indexes which is spilt into two boxes, one containing Public Primary schools and Public Secondary schools and the other containing Private primary schools and Private secondary schools.

        These two then flow down into the Level 2 indexes, with one box labelled Public Schools and the other Private schools.

        Finally these both flow down into the final Level 3 index which contains overall school education.

        The method also supports aggregation by state/territory. For example, in each state/territory, it is possible to calculate cost-weighted indexes for primary and secondary school education. These can then be aggregated to form a cost-weighted index for total school services in each state.

        Technical details of index calculation methods can be found in the Appendix.

        Results

        Comparisons between unweighted and weighted indexes were undertaken to assess the impact of the application of cost weights. Given that cost shares at various levels of disaggregation in the input data sets are largely time invariant, no noticeable differences were observed between the two sets of indexes. However, by way of future proofing, it is recommended that weighted indexes be employed to measure changes in volumes of education output.

        Figure 4 compares the aggregate cost-weighted volume index for primary and secondary school education to an ‘unweighted’ index of student enrolments (on an FTE basis) for those school levels. The unweighted indexes are compiled using the same FTE data as the cost-weighted indexes. Should there be sudden changes in costs in the lower level aggregates, or significant compositional change due to policy or demographic shifts, a cost-weighted volume index would reflect these events whereas an unweighted index would not.

        Figure 5 shows a disaggregation of the weighted index in Figure 4 into indexes for public schools and private schools. The output of private school education has grown slightly faster than public school output over the time span. The private primary and secondary school series show some divergence from 2014-15 onwards. This is likely to be the combined effect of at least two factors. First, the proportion of students in secondary school rose over this period due to the reclassification of Year 7 from primary to secondary school in Queensland and Western Australia; this was more pronounced in private schools. Second, unit costs for secondary education are generally higher than those for primary education (and more so for private schools). This implies that the increase in average costs per student resulting from the reclassification was greater in the private school system, which resulted in the divergence between the weighted and unweighted series in Figure 5. Similar patterns are not as easily observed in the public schools indexes as total enrolments are much larger in comparison to the number of reclassified students.

        Figure 5 also demonstrates the potential for the subtler compositional changes to be captured by the application of cost weights to more highly disaggregated data.

        Figure 6 shows a disaggregation of the volume indexes in Figure 5 into primary and secondary education services. As mentioned earlier in the paper, Year 7 education in Queensland and Western Australia was reclassified from primary to secondary school education in 2015, which creates a compositional shift in the indexes. The sharp turning points in the primary and secondary (weighted and unweighted) indexes in 2014-15 and 2015-16 relate to these reclassifications.¹

        While these indexes have the ability to pick up quality changes associated with such reclassifications, the use of ‘ideal’ product (year level) data, as described earlier in this paper, could result in indexes which are more robust to the reclassification of year groups into primary or secondary streams.

        The layers of stratification used to construct the experimental indexes make it possible to generate indexes for each state/territory. The generation of state indexes presents potential for future enhancements to volume estimates in the State Accounts (cat. no. 5220.0).

        Footnote

        1. Note: The reclassification mainly occurred in calendar year 2015, but on a financial year basis the impact spreads over two consecutive years.

        Next steps - possibilities for future work

        Possibilities for future work to enhance measures of school education output include further refinements to the stratification of data, expanding the project to include other levels of schools, and the incorporation of outcomes measures.

        Using private school market price data to compile the private school indexes

        One logical extension to the methods proposed in this paper would be to weight the private school quantity indexes by the appropriate market price data.

        Exploring the possibility of further refining the stratification

        The literature recommends stratifying the quantity and cost data as finely as possible. A finer level of stratification (year level) than that used in this paper has been adopted by the Office for National Statistics (ONS, UK) in compilation of output volume measures.¹ While enrolments data for Australia are available by year of enrolment the costs data are not available at this level of granularity. This is because schools are funded on the basis of primary and secondary students in the first instance with additional amounts allocated at a school level to address disadvantage in various forms. The only significant cost differential between grades at a school is therefore based on the number of students in each grade. In any case, the size of the potential benefits from this approach are unclear. As a result, the analysis in this paper focuses on differentiating between primary and secondary education, and not by individual year of enrolment.

        There may, however, be value in subdividing secondary education into a junior secondary component (Years 7 to 10) and a senior secondary component (Years 11 and 12), given that Years 11 and 12 are not compulsory and there are options to engage in a ‘blended’ education that combines classroom teaching with work experience, which may also change the cost profile for these years of education.

        Participation rates of the student population are noticeably lower for Years 11 and 12, resulting in different FTE and cost parameters than for the rest of the secondary education stratum. Teacher-student ratios may also be different in Years 11 and 12, which would have implications for the quality of services. Refining stratification for these year groups may enhance the accuracy of volume measures.

        Expanding the project to include pre-school and special needs school education

        A complete measure of school education output would be achieved if volume indexes for pre-school and special needs education were developed, and incorporated into the indexes presented in this paper. Some initial exploratory work has been done to consider these levels of schools, but this work is in its early stages.

        Incorporating outcome measures into the indexes

        There is no international consensus about the strength of the link between education services provided and outcomes for students, and there is no suggestion that the ABS would include outcomes-based estimates in the national accounts. However, measures of student progress based on student test scores have been included in productivity statistics published by the ONS.¹⁶ ¹ Other factors that might impact on student outcomes including parental support, innate ability and motivation, and socio-economic status, are not directly accounted for in the ONS analysis.

        The ABS could potentially explore a similar option using the National Assessment Program – Literacy and Numeracy (NAPLAN) test results for grades 3, 5, 7 and 9. This could be further enhanced by incorporating metrics capturing rates of Year 12 completion and scores in Year 12 examinations.

        References

        Show all

        Atkinson, T., 2005, “Atkinson Review: Final Report – Measurement of Government Output and Productivity for the National Accounts”, United Kingdom.

        Annabel J (2019), “Enhancing measures of non-market output in economic statistics: Progress paper”, Australian Bureau of Statistics.

        Cornell-Farrow S (2019), “Improving Measures of School Education Output and Productivity in Queensland”, Queensland Productivity Commission.

        David Gonski (2018), “Through Growth to Achievement”, Report of the Review to Achieve Educational Excellence in Australian Schools.

        Diewert W E (2008), “The Measurement of Non-market Sector Outputs and Inputs Using Cost Weights”, University of British Columbia.

        Diewert W E (2010), “Measuring Productivity in the public Sector: Some Conceptual Problems”, University of British Columbia.

        Eurostat manual and guidelines (2016): “Handbook on prices and volume measures in national accounts”.

        Gibbs Timm (2017), “Sources and Methods: Public service productivity estimates: Education”, United Kingdom.

        Martin Josh (2019), “Improved methods for total public service productivity: total, UK, 2017”.

        Productivity Commission (2018), Report on Government Services – School education.

        Schreyer P (2010), “Towards Measuring the Volume Output of Education and Health Services: A Handbook”, OECD.

        Schreyer P (2012), “Output, outcome and quality adjustment in measuring health and education services”. Review of Income and Wealth, 58(2), 257-278.

        Statistics New Zealand (2010), “Measuring government sector productivity in New Zealand: a feasibility study. Wellington: Statistics New Zealand”.

        York, J. 2010, “The Devil Is In The Detail: Demonstrating the Impact of Measurement Choices on Inputs to Government Sector Productivity”, Statistics New Zealand.

        Technical appendix

        Show all

        The compilation method involves building annually-reweighted chain linked Laspeyres volume indexes for school education output at various levels of stratification and aggregating them up using appropriate cost weights at each step as indicated below. The weights used in aggregation are based on operating costs.

        Variables and symbols used in the equations

        Affiliation (index k) 

        • public
        • private

        School level (index j)

        • primary
        • secondary

        Jurisdiction - state/territory (index i)

        • NSW, Vic, Qld, SA, WA, Tas, NT, ACT

        Quantity measure (f)

        • Enrolments on a full-time student equivalent basis (f)

        Finance item for aggregation weights

        • Expenditure (e)

        The compilation is performed in three steps.

        1. Create chain linked Laspeyres volume of output indexes for each school level and affiliation at time t:

         \(L^t_{jk} = L^{t-1}_{jk}*\big\{ \sum^8_{i=1} \big[\big(\frac{f^t_{ijk}}{f^{t-1}_{ijk}}\big) *(e^{t-1}_{ijk}/\sum_ie^{t-1}_{ijk}\big)\big]\big\}\)    (1)

        Where j=1,2 correspond to volume indexes for primary and secondary education, k=1,2 correspond to volume indexes for public and private schools and the Laspeyres volume index for the first year in the data series is set to 100.

        2. Create chain linked Laspeyres volume of output indexes for total public and total private schools at time t:

        Two overall Laspeyres school education volume indexes (across all school levels) are estimated, one for each of the subgroups in the ‘affiliation’ stratification variable, i.e., private, public, where

        \(L^t_{k} = L^{t-1}_{k}*\big\{ \sum^2_{j=1} \big[\big(\frac{L^t_{jk}}{L^{t-1}_{jk}}\big) *(e^{t-1}_{jk}/\sum_je^{t-1}_{jk}\big)\big]\big\}\)    (2)

        Where \(e^{t-1}_{jk} = \sum_{i=1}^8e^{t-1}_{ijk}\) and the Laspeyres volume index for the first year in the data series is set to 100.

        3. Create chain linked Laspeyres volume of output index for total school education at time t:

        \(L^t=L^{t-1}*\big\{\sum_{k=1}^2\big[\big(\frac{L_k^t}{L^{t-1}_k}\big)*\big(e^{t-1}_k/\sum_ke_k^{t-1}\big)\big]\big\}\)    (3)

        Where public and private expenditure on education at time t -1=\(e^{t-1}_k=\sum^2_{j=1}e^{t-1}_{jk}\) for k=1 and 2 and the Laspeyres volume index for the first year in the data series is set to 100.

        Previous catalogue number

        This release previously used catalogue number 5900.0.00.001