The Importance of Individual and Environment Characteristics in Analysing Students’ Year 10 Academic Achievements

Using a rich longitudinal data set, two issues regarding academic performance analysis are examined. First, we address the endogeneity of pre-exam school-leaving choices to (potential) performance in grade 10 national examinations and error correlation between the two outcomes. Second, we evaluate the sensitivity of academic performance analysis to excluding variables often unavailable in datasets (such as childhood IQ, early family resources, and teenage peer effects). An important finding of this paper is that inclusion of these (often unobserved) characteristics may alter policy implications drawn from academic performance research. Our results show further that in addition to childhood ability and family background characteristics, later family, school and peer resources also have major and significant effects on academic achievement. JEL codes: I21, J24, J13, J18 ∗ This study was made possible through a research grant from the New Zealand Treasury. Special thanks go to David Fergusson and John Horwood for making the data set available and for insightful information on the data set; Francine Blau, Lawrence Kahn, Stephen Jenkins, Anna Vignoles, Geeta Kingdon, Eric Isenberg, Bob Haveman, Barbara Wolfe, Karl Scholz, Dean Hyslop, David Mare, seminar participants at IZA, the 2007 SOLE meetings, and the 2006 Australasian Labour Econometrics Workshop for insightful comments. The authors alone are responsible for the views expressed.