Why Use Consumer Expenditure Surveys for Analysis of the RSBY?

This is relating to the recent debate on the so-called “rigorous evaluation-based robust findings” based on the National Sample Survey Office (NSSO) consumer expenditure survey (CES) data available for 2004-05 and 2009-10 (Selvaraj and Karan 2012a), on the inability of publicly-financed health insurance schemes to reduce out-of-pocket expenditures in India. My comment (Dilip 2012) on the original article (“Why Publicly-Financed Health Insurance Schemes Are Ineffective in Providing Financial Risk Protection” (EPW, 17 March 2012)) was critical about the Rashtriya Swasthya Bima Yojana (RSBY) coverage-related sample sizes for performing such an analysis and also for not making healthcare utilisation rate standardised comparisons between the two-time points. These are two fundamental things that need to be addressed before comparing expenditures on sparingly consumed commodities (the proportion of households incurring expenditure on hospitalisation care in a one-year reference period was about 9% in 2004-05 and 13% in 2009-10) in the two CESs. Due to these limitations, there was no scope/merit in my earlier note to discuss the relevance of the application of difference in difference (DID) method or ­results of their evaluation.

The response from the authors (Selvaraj and Karan 2012b) is without any justification on the feasibility of their analysis. This could be either out of sheer ignorance of the adequacy of the sample size while testing such a hypothesis using survey data sets, or a deliberate attempt to create a smokescreen in order to avoid answering the methodological flaws pointed out earlier.