Estimating poverty in India without expenditure data: a survey-to-survey imputation approach

This paper applies an innovative method to estimate poverty in India in the absence of recent expenditure data. The method utilizes expenditure data from 2004-05, 2009-10, and 2011-12 to impute household expenditure into a survey of durable goods expenditure conducted in 2014-15. At the $1.90 per day international poverty line, the preferred model predicts a 2014-15 head- count poverty rate of 10 percent in urban areas and 16.4 percent in rural areas, implying a poverty rate of 14.6 percent nationally. The implied poverty elasticity with respect to growth in per capita Gross Domestic Product (GDP) is within the range of past experience, and states with higher gross domestic product growth saw greater predicted poverty reductions. In validation tests, the model's predictions perform comparably to the World Bank's current adjustment method when predicting for 2011-12 but they are far more accurate when predicting for 2004-05. Three alternative specifications give moderately higher estimates of poverty. The results indicate that survey-to-survey imputation, when feasible, is a preferable alternative to the current method used to adjust survey-based poverty estimates to later years.