Do increased resources increase educational attainment during a period of rising expenditure? Evidence from English secondary schools using a dynamic panel analysis

Abstract

This article estimates the effects of school expenditure on school performance at Key Stage 4 in England, over the period 2003–07 during which real per pupil expenditure increased rapidly. It adds to previous investigations by using dynamic panel analysis to: exploit time series data on individual schools that only recently has become available; adjust for the potential endogeneity not only of expenditure but also of other determinants of performance; and differentiate the short‐run and the (higher) long‐run attainment effects of spending changes. Consistent with other recent work, the article reports a generally significant but small effect of expenditure on school performance, but it also finds that the effect varies between specialist and non‐specialist schools, with the effect on the latter being larger. Further, the article identifies significant dynamics in the school improvement process, quantifies the long‐run effect of expenditure changes, suggests that spending effects increase with socio‐economic disadvantage, and quantifies absence effects.

Educational Achievement and the Allocation of School Resources

“…Our approach hinges on controlling for prior test scores, an extensive set of other (student‐ and school‐level) characteristics and school fixed effects. Other approaches include the use of instrumental variables and boundary discontinuities (Gibbons, McNally and Viarengo ) or Gaussian mixture model estimation of dynamic models on school‐level panels (Pugh, Mangan and Gray ; Pugh et al ). Instrumental variable estimates are informative about the local treatment effects and are difficult to generalise, while school‐level analysis cannot take into account student heterogeneity.…”

What Drives Bank Lending in Domestic and Foreign Currency Loans in a Small Open Transition Economy with Fixed Exchange Rate? The Case of Macedonia

“…The results indicated that at a 10% level of significance we cannot reject the null hypothesis of their joint validity, supporting the choice of system over difference GMM (Tables 2 and 3). Moreover, as a rough test for the presence of cross-sectional dependence in the model, we conducted the difference-in-Hansen test for the validity of the instruments for the two lagged values of the dependent variable (Sarafidis et al, 2006;Pugh et al, 2010). At a 10% level of significance, we cannot reject the null hypothesis for the joint validity of the instruments for the lagged values of the dependent variable (Tables 2 and 3).…”

School expenditure and school performance: evidence from New South Wales schools using a dynamic panel analysis

“…A more recent study by Pugh et al . (), using English data at Key Stage 4, uses a dynamic panel model estimated on five years of data. It reports a generally significant but small effect of expenditure on school performance in the short run, with larger effects in the long run, but it also finds that the effect varies between specialist and non‐specialist schools, with the effect on the latter being larger.…”

“…The recent dynamic panel data study (Pugh et al ., ) covered a five‐year period (2003–2007) in England when real (inflation‐adjusted) expenditure per student rose by more than 20% and there also was a strong upward trend in the performance measure. The latter cannot be the case here, given that we have a relative performance measure.…”

“…However, there is a lack of studies that examine the effects of school and non‐school inputs such as financial resources, teacher characteristics, family socio‐economic status and student composition on student outcomes in the context of Australian schools. The focus of this paper is on the analysis of the relative performance of New South Wales (NSW) secondary schools, where performance is measured by schools’ median Year 12 university entrance Australian Tertiary Admission Rank (ATAR) scores, utilising dynamic panel econometric methods similar to a recent study on English secondary schools (Pugh et al ., ). This article further contributes to establishing the case for investigating school performance by using panel data and a modelling strategy that allows for the influence of past performance on current performance.…”