Assessing the performance of Spanish secondary education institutions: Distinguishing between transient and persistent inefficiency, separated from heterogeneity

This paper evaluated the performance of Spanish secondary schools whose 15‐year‐old students were assessed in mathematical competencies by the OECD (PISA program) in 2003 and 2012. The technique employed was the stochastic frontier analysis for panel data using a sample of schools which participated simultaneously in both waves. First, the parametric measurement of time‐varying technical inefficiency was done in this paper using three standard models. Second, we used the four random component stochastic frontier model proposed by Kumbhakar, Lien, and Hardaker [2014. Journal of Productivity Analysis, 41(2), 321–337] that distinguishes between residual or transient technical inefficiency and persistent technical inefficiency, separated from heterogeneity. Persistent (time invariant) inefficiency was a larger problem than residual (time varying) inefficiency when evaluating the educational performance of Spanish secondary schools over time. Finally, we introduced the recent model recommended by Badunenko and Kumbhakar [2017. European Journal of Operational Research, 260(2), 789–803] to accommodate heteroscedasticity associated with both heterogeneity and the noise terms, incorporating at the same time determinants of both persistent and time‐varying inefficiency. School inefficiency was presumably not caused by something unexpected within each year such as greater difficulty in hiring teachers, but rather by persistent factors such as classroom management—schools with better disciplinary climate tend to be less inefficient in educational production. In addition, we identified the motivation of the students of each school (interest in and enjoyment of mathematics) as the effect of heterogeneity on learning outcomes.