Missing the Poor, Big Time: A Critical Assessment of the Consumer Pyramids Household Survey

India’s statistical system is in bad shape with a near absence of regular publicly funded household surveys in recent years. All eyes have now turned to the Consumer Pyramids Household Survey (CPHS), a panel survey of over 170,000 households, privately executed by the Centre for Monitoring Indian Economy (CMIE) since 2014. Given its breadth and high frequency, CPHS has become a widely referenced barometer of the Indan economy. Research papers using CPHS have also mushroomed. However, there has been little validation of the nature and quality of CPHS data. Most crucially, is it true that CPHS is an “all-India representative survey” as claimed by CMIE and echoed by multiple articles in prestigious journals? Comparing CPHS with various national surveys on a set of key demographic and economic indicators, this paper argues that, far from being nationally representative, CPHS under-represents women and young children, over-represents well-educated households and under-represents the poor. A possible source for these biases (among others) is the strange, unorthodox sampling design adopted by CMIE, which differs from standard sampling approaches on various counts. Further, the bias in the CPHS sample appears to be growing in recent years, posing a serious challenge when using the data to study trends over time