In India, as elsewhere, crafting good macroeconomic policy is difficult. Measurement of aggregates such as output and inflation is fraught with challenges, and it is hard to build reliable models of the entire economy. A few years ago, there was a somewhat heated debate among prominent economists about the merits of GDP growth as a prime objective of economic policy in India, with claims made for promoting “inclusive” growth and even non-GDP measures of well-being such as health and education outcomes. Implicit in disagreements on these topics were different ideas of how the economic growth process functions in India, and the merits of trade-offs between the welfare of different slices of society at a point in time, or across time periods. I did see a formalisation of inclusive growth by Kaushik Basu when he was Chief Economic Adviser, and Jagdish Bhagwati and Arvind Panagariya tried to capture different growth impacts of policies with the concept of Type I and Type II reforms, but as far as I can tell, there was not a sustained intellectual legacy of this debate in policy making. And we still seem to have little idea of the detailed process of India’s economic growth. Policy for controlling inflation is a hotly debated topic in India right now. The shift to targeting a desired band for inflation, changes in what inflation measure should be used, and the institution of a formal committee process have represented major shifts in the process of policy-making. But there still seems to be little concrete knowledge on the process that generates inflation in India, as well as the process by which changes in monetary policy (chiefly, the policy interest rate controlled by the Reserve Bank of India) affect inflation.
The tradeoff between inflation and growth is also not well understood. One unavoidable problem is that what people expect future inflation to be is a key variable, and measuring this expectation, along with how it is formed and evolves, is not something that has a solid history in India. Random shocks like demonetisation do not help in figuring out what is optimal for monetary policy. It is also the case that economic theory is an unreliable guide: the RBI Governor has expressed concerns about high fiscal deficits fuelling inflation, but the link between such deficits and inflation is not solidly established, either theoretically or empirically. Problems of measuring output add a further layer of uncertainty to the process. Macro policy-cannot be based on controlled experiments. Recently, however, economists have been showing that such experiments can be a pathway to better micro policy. The work of Karthik Muralidharan in education is a good example.
Here I offer my own take on some of his work. Over a decade ago, Amarjeet Sinha of the IAS tackled the problem of low school enrolment in India with the massive Sarva Shikshya Abhiyan (SSA). The SSA did increase enrolment, but learning outcomes stayed poor, as NGO Pratham and several economists pointed out. Attention turned to the process of teaching, and the absence of teachers from classrooms because of lack of proper monitoring and incentives. Experiments showed that this could be addressed, and some were scaled up into quasi-policy changes. Muralidharan, in particular, explored the use of lower-cost “teaching assistants,” but to my mind, this risked being “more of the same.” Muralidharan’s latest work is both simple and stunning. An experiment in a single school in a poor area of Delhi showed that the right software can aid learning dramatically. The reason is that students who are not taught effectively early on get moved on to the next grade, where they are taught that level of curriculum, but without the base to build it on.
As they move through grades, what the teacher teaches diverges more and more from their level of comprehension: this is particularly true for mathematics. Software can be customised and adaptive to a student’s level and even transitory states. Humans are present to ensure the students stay on task, but don’t need to do more. The more the software is used, the more refined it can become. The next step is to expand this experiment to multiple locations, and to shift it from a separate, after-school location (which it was) to being done within regular school times and locations. Certainly unexpected nuances and obstacles may arise, but the ability of economists to design good policy experiments and, in this case, use software to generate large amounts of refined data, is striking.
Making the delivery of basic skills to large numbers of India’s children more effective, and in as rapid a manner as possible, can be a huge win for the country. It can also validate the modern economic approach of paying close attention to details of behaviour, technology and institutional settings. Of course, software and economists are useful, but not essential. Several years ago, another IAS officer, Amarjit Singh, figured out how to make a micro intervention in health successful, with the Chiranjeevi scheme for institutional maternal deliveries in Gujarat. He noted at the time the difficulties of replicating and scaling up that effort. But while macro-policy making continues to struggle, at least there can be progress at the micro level.