Substack

Tuesday, August 7, 2018

Are RCTs crowding out serious research on India?

Devesh Kapur writes on India focused scholarship emanating from the US. He has this assessment of the RCT research,
Suffice it to say the effects have been much more positive for the careers of US-based researchers than for India. Ironically, some of the very strengths, such as the stress on identification and causal inference, have been a source of weakness. The stress on these methods as “the gold standard” comes at the cost of relevance and timeliness. Only certain types of questions can be addressed by these methodologies. This is not to say there aren’t excellent studies that address important policy questions. But more often than not, even if they can address them, the costs and duration of these studies means they are more useful as citations than policy.
When asked how many of these expensive RCTs had moved the policy needle in India, Arvind Subramanian, Chief Economic Advisor, GOI, was hard pressed to find a single one that had been helpful to him in addressing the dozens of pressing policy questions that came across his table.
Kudos to him for having called this out!

This proliferation of RCTs in India may have also had the unintended effect of crowding out other more relevant research on issues concerning India. It should be a matter of great concern that even as many of the RCT papers have become popular, I cannot recall a single serious work by a reputed US-based researcher on India's ongoing banking crisis (say, a comparative study), the infrastructure boom and bust of the last decade, heterodox monetary policy success during the global financial crisis, financial market architecture and capital markets, economic growth trends and drivers over the past two decades, experience with PPPs, comparison of India's SEZs and industrial policy with China and others, public finance reforms, informal markets, fiscal federalism, urbanisation and so on. 

In all these areas, apart from some high quality work by local Indian researchers, serious exploration is confined to either reports by global Consultancies or one-off papers by researchers from IMF, WB, and ADB. It is a pity that even ethnic Indian researchers in these Universities prefer either to do the RCT stuff or focus on US and other countries.

The contrast with works on macroeconomy, infrastructure, financial markets, public finance and so on concerning China, Latin America, Italy, Spain etc coming out of places like Harvard is very stark. It may be no exaggeration to suggest that the only area of interest concerning India in top US Universities revolves around poverty studies of the kind involving randomised control trial (RCTs) and romanticised visions of entrepreneurship and social enterprises that serve the Bottom of Pyramid!

This is not the only negative externality. The RCT movement has also displaced good old ethnographies and other insightful qualitative studies. Consider this. Devesh Kapur again, has a very good paper which goes beyond the IAS bureaucracy and shines light on the dark underbelly of the Indian state - the "chronically under-resourced" frontline bureaucracy and the constraints they face. It is an indictment of the development research that this may be the only such ethnographic study of Indian bureaucracy by a foreign based academic. The vast majority of RCT researchers themselves would be enormously enriched by internalising the field conditions described by Devesh and his co-author.

In order to catalyse serious research on India, institutions like the RBI and NITI Aayog may have to take active interest. In co-ordination with the Ministry of Statistics and Program Implementation, they  could guide research on some of the aforementioned areas by making available with appropriate safeguards and conditions the country's very rich set of databases from Economic and ASI census, Companies Registration (MCA), labour related data (Labour Bureau, EPFO, NPS etc), taxation (CBDT, CBEC, GST, States etc), PPPs (MoRTH, DEA etc), stalled projects, and so on.

No comments: