Banerjee-Duflo’s models don’t always work, but the Nobel is an affirmation of the need for evidence-based policies
Unlike many Nobel prizes where the practical applications of the winning theory is not immediately clear, it is easy to understand why Abhijit Banerjee, Esther Duflo and Michael Kremer won the economics award this year. Tackling poverty is a global goal and governments such as those in India spend upwards of Rs 300,000 crore – that’s two per cent of GDP by the centre alone, the states spend a similar amount – every year on some form of welfare activity, so it is important to make sure the country gets the biggest bang for its buck.
Over time, the trio’s research – through the Abdul Latif Jameel Poverty Action Lab in the case of Banerjee-Duflo – has pointed to what solutions work and what don’t; on how to tweak delivery systems to make them better. Over time, the trio set up two groups of people in what is called a Randomised Control Trial (RCT) – one that get the benefits and one that don’t – to see just how well the action/intervention works. In the case of schools in Vadodara and Mumbai, in association with the NGO Pratham, Banerjee-Duflo’s RCT established that providing tuitions for individual students improved overall test scores dramatically. The reason, it turns out, is that most government schools have classrooms with children of different abilities – some have Maths skills of Class 2, some 3 and some 6 – so the class teacher doesn’t know quite where to pitch the lesson; the remedial intervention takes care of this problem by bringing all the kids to the same level. The Nobel citation talks of how “more than five million Indian children have benefitted from effective programmes of remedial tutoring in schools”.
In the case of Punjab, The Indian Express reports the Pani Bachao Paisa Kamao scheme is based on a JPAL recommendation where farmers are paid a fixed amount for buying electricity which is used to pump out water; if farmers use less electricity – which means they use less water – they get to keep the money they save; one farmer, Karamjit Singh Rai, earned Rs 12,904 through this scheme. If such solutions were to be used across the board, the Nobel committee must have felt, the impact could be quite dramatic.
While it is heartening to see governments listening to the advice of economists like Banerjee-Duflo-Kremer, it is not as if RCTs don’t have their fair share of critics. While RCTs try and eliminate regional and other biases by selecting two groups within the same area, it is not as if solutions that work in one geography necessarily work in other geographies; nor is it clear why a particular solution works. In Udaipur, when it was found that not enough women were bringing their children in for immunization even though it was free, Banerjee-Duflo’s team gave those who came in some free daal when they came in; this seemed to work and immunization levels in the area rose fairly quickly. So, the obvious question is, would giving cash have worked just as well, and didn’t the mothers think that free immunization in itself was worth it? Indeed, if it is not, why is Swachh Bharat working as well as it is?
Some of this is also intuitive. Agriculture economist Ashok Gulati, without the benefit of RCTs, has been recommending that the government give farmers a fixed amount of money each year to spend on fertilizers; once farmers know they will get the subsidy regardless of how much fertilizer they buy, Gulati argues they will consume less fertilizer and use the rest of the money for something else. As a result, the damage to the soil will also reduce. That is also why, more than five years ago, Gulati recommended the government scrap most of its agriculture schemes like MSP-based procurement and, instead, just opt for a flat annual payment to farmers.
Nor do RCT-type of experiments always give you a clear answer on the preferable policy option. Banerjee, for instance, helped the Congress party design its NYAY scheme where the poor were to get Rs 2,500 per month as dole – the party raised this to Rs 6,000 when announcing the scheme. So is NYAY a better option to individual govt schemes in various fields like education or health or are the poor not smart enough to use the money well? For whatever reason, Banerjee never explicitly recommended NYAY in place of other welfare-spend, or at least not publicly.
Yet, SEWA’s pilot studies in Madhya Pradesh in 2012-13 found that villages that got untied cash transfers tended to open more bank accounts, build toilets (this was before Swacch Bharat began), used more modern fuel for cooking/lighting, sent more children to school, etc. And Gulati’s study with Shenggen Fan and Sukhadeo Thorat found that, while every million rupees spent on fertiliser subsidies reduced the number of poor by 24, the reduction was 335 if the same money was invested in building roads and 323 if it was spent on agricultural R&D.
While the policy lesson from the Nobel prize is that all policy actions must be based on evidence – and must be junked if there is no proof of their impact after a reasonable period of time – it is important that policy-makers also keep in mind the limitations of micro-actions of the Banerjee-Duflo-Kremer kind. Certainly, they will help alleviate poverty or suffering in the short-run and they will make health and education spending more efficient – this is a very big positive – but they have to be used in conjunction with other policies to enhance growth. That is why, for instance, Banerjee opposed demonetization since this reduced economic growth and the biggest sufferers from this were the poor; no amount of welfare intervention can make up for the loss of livelihood. Also, larger government spending on building roads or creating Aadhaar-like solutions create their own virtuous cycles that help the poor. Eventually, good governance creates good economics for the poor; given how little economic thought has gone into most pro-poor schemes in India, Banerjee-Duflo’s defining work on fighting poverty is called Poor Economics.