Has India’s learning crisis really worsened?

Despite continuing improvements in educational inputs, the main source of learning metrics in India – the Annual Status of Education Report (ASER) surveys – show a large and permanent decline in basic math and reading ability among children aged 5–16 between 2007 and 2018. I argue that this decline is partly an artifact of changing measurement error. Using an external benchmark, I show that pre-2011 learning estimates are biased upwards, potentially due to undersampling of poorer households. This positive bias dissipated over time, appearing to coincide with the introduction of improved data quality assurance procedures in the 2011 survey round. While the true decline was likely smaller than previously thought, concerns about the declining productivity of India's education system remain. • Learning decline in India's ASER surveys (2007–2018) partly reflects changing measurement error over time. • External benchmark shows that pre-2011 levels are biased upward, potentially due to undersampling of poorer households. • Bias was mitigated and data quality improved after 2011, coinciding with improvements in data quality assurance procedures. • Findings challenge existing research attributing the learning decline to the Right to Education Act 2010.