How Randomised Evaluations Can Galvanise Policymaking In Myanmar
Over the past two months, much ink has been spilled regarding the exploits of Nobel laureates Abhijit Banerjee, Esther Duflo, and Michael Kremer and their pioneering work with randomised controlled trials (RCTs).
Most have been positive, but in spite of the trio’s achievements, naysayers persist.
For the uninitiated, an RCT is an experiment in which people are randomly assigned to treatment and control groups. As differences between the two groups are cancelled out by randomisation, any changes at the end can be attributed to the intervention itself.
Last month, The Economist ran an article titled “ This year’s Nobel prizes prompt soul-searching among economists “. The article highlighted some of the criticisms of RCTs, citing the ungeneralizable nature of its findings and its potential to dull the ambition of development research.
These much-rehearsed criticisms are all valid. Bill Easterly is right in that RCTs cannot substitute for larger macroeconomic and institutional reform. South Korea, Singapore, and most recently Rwanda’s transition from backwater country to economic engine can be traced to the presence of good institutions and strong governance.
The reality, though, is that most developing countries are trapped in a catch-22; the tools needed to escape their malaise are locked in the very future they are trying to materialise. The presence of path dependency also renders it difficult to make sweeping governance reforms.
Despite Myanmar’s transition to democratic governance, the military retains control over large swathes of the government apparatus and is designated 25 per cent of parliament seats under the terms of the 2008 constitution. Changing the constitution would no doubt accelerate economic development, but doing so requires more than 75 per cent of votes in parliament, rendering the task effectively impossible.
In addition, over half of Myanmar is employed in the agriculture sector, and it is unclear how much macroeconomic reform can benefit poor farmers who are more greatly affected by the vagaries of weather patterns than the interest rates of central banks. While Indonesian farmers were benefiting from increases in world food prices in 2008, Myanmar farmers were left reeling after Cyclone Nargis decimated large areas of paddy fields.
In this vein, RCTs may offer us much more in alleviating poverty, by providing an intimate understanding of poverty closer to its beneficiaries.
In Myanmar, state support to farmers is poor. The Myanma Agricultural Development Bank provides limited financial services and also has limited geographical reach, resulting in many of the country’s farmers remaining underbanked. Farmers also do not benefit from a nationwide policy like the Common Agricultural Policy in the EU or the US farm bill.
Short of that, encouraging farmers to take the steps they can to maximise yield is paramount. Duflo, Kremer, and Jonathan Robinson’s study on fertiliser use in Africa do just that, by uncovering why farmers choose not to buy fertiliser even when they know it greatly enhances yield. They found that the presence of time-inconsistent preferences means that most farmers end up spending fertiliser money by planting time, even if they say they want to use it at harvest time. Their solution was thus to design a treatment that offered free delivery of fertiliser at harvest time, which significantly increased take up.
Currently, the use of RCTs in Myanmar is few and far between, but it carries huge potential for energising a country with a lacklustre record in policymaking. This is a situation that has not seen much improvement since the NLD government took the reins.
Last year, the government revised its social pension policy by reducing the pension age from 90 to 85. The wisdom of providing MMK 10,000 a month — a little over US$6.50 — to 85-year-olds in a country with a life expectancy of 67 is not immediately apparent. This is compounded by the fact that few elderly have access to banking systems, making distribution of pensions a highly personal undertaking.
This is a clear example of government failure when designing policies. Conducting RCTs in this space could generate insights into the effect of cash transfers in shaping elderly consumption habits, and whether it might be more appropriate to distribute household essentials or health services instead.
For instance, IPA recently concluded a study in Myanmar on how maternal cash transfers to pregnant mothers alone do not reduce child stunting on average, unless it is coupled with social and behavioural change communication (SBCC). These findings underline the importance of complementing cash transfer programs with SBCC and the value of ensuring full coverage for children in their first three years of life.
Such discoveries give us an insight into the psyche of Myanmar’s poor and help formulate interventions that nudge them in the right direction in small increments. Indeed, the RCT approach should not be viewed as a retreat from the biggest questions in development.
To suggest that RCTs lack ambition is to imply that there is a silver bullet in the developmental toolbox we should be looking for. As seen from the well-documented failures of structural adjustment programmes, silver bullets do not exist in development.
Intellectual jousting in academia is a good barometer of the health of the discipline, but critics of RCTs should be wary of creating the same aid versus trade debate that consumed development economics in the 2000s. Proponents of RCTs and applied microeconomics do not deny the importance of top-down reform, only that we should not limit the tools at our disposal.
And for a country that has spent too long in the shadows, Myanmar is in need of a well-stocked toolbox, even as it attempts to fight rising inflation rates and a declining international reputation.
Using randomised trials to evaluate interventions will greatly energise the work of non-governmental organisations and policymakers working in Myanmar. In ameliorating poverty, economists ought not to lose sight of the forest for the trees.