The Efficiency of Australian Schools: Evidence from the NAPLAN Data 2009-2011

This study examines the technical efficiency of schools in Australia and its determinants using NAPLAN test results of about 6,800 schools in 2009-2011 and other information from the “My School” website. For each school, we use the average growth of test scores for the same students between 2009 and 2011 as the measure of the school's output and four input measures: the student-teacher ratios, student-non-teaching staff ratios, recurrent income per student and (averaged) capital expenditure per student. We are also able to compare schools by type: including whether or not the school is a public school or a private school, a single sex or co-educational schools, a primary or secondary school, or a school that provides both primary and secondary schooling. In addition we control for several other environmental indicators for each school including: an index of social and educational advantage, the proportion of school children who identify as an Aborigine or Torres Strait Islander, the proportion of students from a non English-speaking background, the proportion of students female, as well as the region, state and territory in which the school is located. We estimate that the average technical efficiency score of Australian schools is 59 per cent and find evidence of input congestion for all of the inputs studied. On average, the growth target for schools in the sample to reach the efficiency frontier is 100 NAPLAN points. Our results suggest that eliminating inputs congestion could, in theory, reduce expenditure per school student by A$2,000. At the primary level, Catholic and independent schools are less efficient than public schools, but this story is reversed at the secondary level. We also find that schools with students from more advantageous social and economic backgrounds and schools with higher ratios of students from non-English speaking backgrounds tend to be more efficient. The results are robust to the choices about how to construct the frontier (e.g., in aggregate or for disaggregates by school type) and to our treatments of output and super-efficiency