A study by NCAER and the University of Maryland on MGNREGA throws up odd conclusions. The programme, it says, reduced poverty levels by up to 32%, improved the use of formal credit and served the cause of gender justice by providing more employment to women. And yet, as the study itself points out, 70% of the poor have not been able to get the work promised under the programme; at another place, it says since 20.6% of rural households were poor in FY12 and the coverage of such households was 31%, poor in MGNREGA comprised just 6% of rural households. And while talking of financial inclusion, the study also acknowledges that financial inclusion has been improving anyway with the dependence on money-lenders falling across the country. It is important to be cautious while interpreting the study. Its main conclusion is that while MGNREGA is a good programme, it hasn’t done well because in the poorer states like Bihar and Odisha, the implementation has been poor. In other words, to improve MGNREGA, get it to work in these states and increase the amount of funds available.
That would be a mistake for, as FE columnist Surjit Bhalla points out, as opposed to the rural development ministry’s claim that 284 crore man-days of work were provided under MGNREGA to 5.26 crore households in FY10, the NSS survey found 3.86 crore households worked for 147 crore man-days. In other words, MGNREGA is as leaky as any other government scheme like PDS. Also, it is not clear how the study isolated the impact of growth-led poverty reduction—while poverty fell 0.74% a year between FY94 and FY05, it fell 2.18% per year between FY05 and FY12. Since MGNREGA has not been much better than the previous avatars of food-for-work programmes, chances are the poverty impact of MGNREGA has been overestimated—a look at the small number of jobs created by MGNREGA and those created by normal GDP growth also makes this clear. Given that over 40% of MGNREGA spending is in non-wages, and that no great capital formation is created by the scheme, pure cash transfers for the full amount would in any case deliver a better result.