In the recent literature, consensus has emerged that poverty is a multidimensional phenomenon; see Alkire and Santos (2010) for a review of the major arguments. Nonetheless, the most widely used measures of poverty remain unidimensional, being based on income or caloric intake cutoffs. The logic for the use of income based measures was that it was only lack of income which led to deprivation—with sufficient income; rational agents would automatically eliminate deprivations in all dimensions in the right sequence of priorities. However, careful studies like Thorbecke (2005) and Banerjee and Duflo (2006) show that this does not happen. Even while malnourished and underfed, the poor spend significant portions of their budgets on festivals, weddings, alcohol, tobacco and other non-essential items. The move from abstract theoretical speculation based on mathematical models of human behaviour to experiments and observations of actual behaviour has led to dramatic changes in the understanding of poverty and how to alleviate it. Some of these insights are encapsulated in a new approach to poverty advocated by Banerjee and Duflo (2011).