This month on CPR Perspectives – our flagship interview series commemorating the Centre for Policy Research’s 50th anniversary – we bring you a conversation with Avani Kapur, a senior fellow at CPR, where she also leads the Accountability Initiative.

The Accountability Initiative focuses on conducting cutting-edge research on India’s public service delivery systems and leveraging this information by ensuring it reaches government officials, academics and citizens with the aim of promoting administrative reforms at the frontlines of service delivery.

Kapur has been at CPR since 2008, beginning as a Research Associate at the Accountability Initiative and working her way up to leading the research group today. Along the way, she has led process- and fund-tracking surveys on vital social sector schemes as well as anchored an annual budget brief series analysing the performance of the Indian government’s major welfare programmes – including, this year, a major lookback at the past 15 years of welfare spending and outcomes to mark AI’s 15th anniversary.

In addition to leading AI, Kapur also set up the PULSE for Development platform in 2020, which brings together more than 90 organisations within the development community dedicated to citizen-centric policies and implementation. Kapur is a Tech4Good Fellow and part of the WICCI Council of Ethic, and is on the editorial board of the Journal of Development Policy and Practice.

In the first part of our conversation, I spoke to Kapur about starting at CPR just as the Accountability Initiative was taking shape, the stunning examples of inefficiency she discovered while looking for bottlenecks in public spending in the field and getting positive feedback from the state – including how one government official described AI’s work as being that of ‘physician’ tracking the flow of blood through the body, searching for blockages.

In the second part of the conversation, which you will receive later this month, we spoke about why the initiative has moved from talking about accountability to ‘Responsive Governance’, how AI does much more grassroots capacity building work beyond its flagship PAISA public expenditure tracking, and what advice she has for young scholars entering this field.

If you prefer audio, this conversation is also available as a podcast here.

And if you missed our previous interviews, read our conversations with Partha Mukhopadhyay (Part 1 & 2) and with Navroz Dubash (Part 1 & 2).

–

(This transcript has been edited for length and clarity).

I’d like to get a sense of how you entered policy in the first place. Did you always know you were going to be going down this road? Could you tell us a little bit about your background well before CPR?

So I was studying history at the Lady Sriram College in New Delhi and I was really loving it. I had great teachers for history who didn’t just focus on the memorization of dates or names but instead about the narratives and stories. All my teachers promoted this sense of inquisitiveness and curiosity and asking questions, not just taking anything on at face value. However, when I got the opportunity to go to a liberal arts college in America, after I had completed two years at LSR, I decided to take that up just because I wanted exposure to different topics. I think I felt that I was a bit too naive and young to decide on a stream and one of the advantages of a liberal arts degree is that I could take up courses on art history, on biology, philosophy, astronomy, even dance or music. The exposure that I got at Smith College definitely paved the way for the journey that I’ve taken in the policy space.
It was during this time when I was at Smith that I took a course on economic policy, and I realised that I had an interest in this field and I wanted to do more in the development space. In fact, during the summer vacations I applied for a research grant and managed to do internships with a micro-finance organization on one break, as well as a research organization on another break. When I completed college, unfortunately in India, during that time there were no real public policy schools and most of the development policy courses in the US were focused much more on international relations or at best international development. But I knew that I wanted to work on India-related issues and so I decided to pursue development studies at the London School of Economics. The course work there, but more importantly, just being in the heart of London where you have numerous scholars coming to visit and give lectures, cemented my decision to work in the public policy space. When I returned to India, I got my first working opportunity at a small research organisation started by Dr Arjun Sengupta, who was then a member of Parliament as well, called the Center for Development and Human Rights. I got to understand some of the rights and welfare literature.
I spent about two years there and then I felt strongly that a thorough understanding of welfare required not just sitting in front of a desk in Delhi, but actually making my way to the realities of the field. I was starting to get a little frustrated about not being able to actually go to the field in that job. And I think luckily at that time I connected with [CPR President] Yamini [Aiyar] and decided to join her in the new initiative she was in the process of conceptualising, called the Accountability Initiative. And 15 years later, it feels almost surreal that I’m now leading that initiative.

That brings us to 2008. As you say that the Accountability initiative was just starting to be conceptualised at the time. So tell us a little bit about that. What was the idea?

When I first joined, India was at the cusp of building the foundations of the rights-based welfare state. You saw an expansion of legislation in key areas of social policy: the right to work, MGNREGA, being a prime example. And you also saw a lot of enhancement in government investment in social policy. What was interesting about this moment was that this expansion in some ways was driven through this unique alliance between civil society organisations on one hand, the judiciary on the other, and members of the political elite. What was also very different about this moment and in some ways, at the heart of this rights-based welfare project was this vision of transforming the dynamics of the citizen state engagement. The idea of empowering citizens to place claims on the state and demand accountability for implementation.
You saw a lot of the new legislation and many of the new anti-poverty schemes that were launched by the government in this phase [had] built into them procedural requirements that asked for greater transparency and created spaces for citizen engagement. For example the use of accountability mechanisms like social accountability, social audits were made mandatory under NREGA. Many of the schemes also talked about the need and importance of participatory bottom up planning.
Yet despite this proliferation of social accountability efforts, there was no analytical framework for understanding and evaluating their effectiveness. It is against that backdrop that AI began this work. The focus of the work was split into two aspects. One strategy was to build a research base and the focus here was to study the unfolding and actual implementation of these new accountability mechanisms, particularly the RTI and social audit. As Yamini would often say, it’s about unpacking the implications of this accountability moment and documenting this new accountability landscape. This was research that tried to understand the citizen-state interactions through these accountability mechanisms, and Yamini was mainly leading that work.
But the second strategy, which is what interested me, was to develop partnerships, to actively engage with the nuts and bolts of implementing social accountability. We started off thinking about education and to do this, we partnered with Pratham and ASER Centre. We initiated an intervention which aimed at strengthening parental participation in elementary schools in a district in Madhya Pradesh through participatory planning. Our work in MP, and in particular that direct engagement with community participation monitoring, meant that every month for a year, I would end up spending time in the field, talking to people, trying to understand what the roles and responsibilities of these parent communities are and trying to develop collectively this ideal plan.
Through that exercise we soon realised that the nuances of government functioning at the local level, whether it’s resource allocation structures, decision making structures, fund flows, end up being a critical bottleneck for effectiveness of these social accountability instruments. In one of the villages that I was working in, the delays in fund flows to schools severely constrained the school’s ability to plan. They never got the money that they had planned for. And as a result, these delays ended up creating a sharp disincentive for citizen engagement and planning. But what was also interesting is that it also undermined the capability of the local administration to respond to the pressures of citizens laying accountability claims. Most of the officials at the local level that we were speaking to, even if they were willing to help, they simply lacked the information or the decision making authority to address the problem and facilitate actions.
This allowed me to get a first hand experience of how implementation works. And in some ways, it was this experience that led us to launch one of our flagship interventions called PAISA.

Yes, and we’ll get to PAISA that in a moment. We’ll also talk about the term accountability. I think the idea that people have in their minds is that money is bottlenecked not by procedural matters but by bureaucratic corruption. What were you finding?

First of all, when the government is making plans or budgets, they’re often making it with incomplete information. Even within government structures, they don’t always have information on what is actually utilised. For the Union government, the second they release money, for them it’s expenditure. But what actually happens with that money is something that particularly then was not captured. There have been some innovations now, but I’m speaking about 2009-2010 where there was very little information about how much of that ₹100 was actually spent and on what?
There was one bottleneck. When you’re trying to create these need-based plans that are done through a bottom-up process, you’re functioning without actually knowing how much money you are going to get in that year and how much you will be able to spend. The problem definitely starts at the planning stage. Once you’ve made the plan, there’s an entire approval process and it’s often the case that what you asked for is not approved. If I’ve asked for ₹100 and ₹60 gets approved, I need to figure out how am I going to divide that 60. If I’ve done my planning right, I would have been able to know exactly where that 40 rupees was meant to go. But if that 40 doesn’t come, how do I rationalise and how do I decide to allocate that money.
We’ve also seen that there are administrative bottlenecks. In one state where we were studying, the sarkari file had to move through 32 desks at the state level alone all seeking approval and trying to capture the corruption question. You’ve got a system of inefficiency in an attempt to capture leakage, but it’s actually leading to a lot more problems in terms of service delivery on the ground.
Some of the decision-making is highly centralised. We have visited schools in Bihar which didn’t have a school building, but there was some order that was passed that every school had to have fire extinguishers. Imagine – a teacher had to lug this fire extinguisher every day back and forth from the school just because someone had passed this order saying that every single school irrespective of need had to get a fire extinguisher. Then you have a situation where money doesn’t come or comes really late. We’ve seen schools where there were annual grants that were meant to come, but either some lag in paperwork or somewhere the bank account numbers didn’t match, so the money just never came in that fiscal year. Or when it came, I’ve seen a number of schools that got money, on the 30th of March or 31st of March, literally one or two days before the financial year is meant to end.
What does the school do? If it doesn’t spend the money, then you’ll be rapped on the knuckles. So the school ends up making decisions of trying to spend, not necessarily on what they need, but on what is faster. I remember going and visiting a large number of schools and what was amazing was that pretty clean looking buildings were regularly getting whitewashed. And when we spoke to people, we realised that that’s one of the easiest ways to spend that money. You whitewash the building so it looks nice and at the same time you are able to show your utilisation. But it still may not be actually what you need because maybe that school didn’t have a boundary wall or didn’t have a toilet facility, but all of those take time.
The other big thing we’ve seen in terms of bottlenecks is that the person who is being held accountable for an action doesn’t always have the authority to undertake that action. We call this the spider web of governance, where you have different people who are on paper meant to be accountable for different activities. But when you start looking at whether they have the authority, the autonomy to make decisions that would enable that accountability, you find that’s not the case. There are a lot of informal networks that are sometimes stronger than what’s written on paper, which impact how services get delivered.

Maybe an outsider’s question here. Why weren’t we – government, or other organisations – tracking this spending already? I don’t know if it’s a naive question, but was it an outcome of the rights based framework that showed up in the late 200s, or is it a matter of capacity?

I won’t say everything is there even now. The fact that we still have governments asking us to help find these bottlenecks and understand what is happening on the ground means there is still a lacuna of information. Part of the reason is that any data below the national level isnt standardised. We haven’t fully valued what all data can be used for, how it can be helped in service delivery, how it can help in planning.
It is, to a large degree, functional capacity. Even now we’ve been studying, for instance data entry operators that collect a lot of data and are responsible for filling information and management information systems which have a lot of ground up granular data. But they don’t have the infrastructure in place. They don’t really fully have computer training in place. So there is a strong capacity gap at the local level, both in terms of data collection, but also data interpretation and data use.
At the national level it’s a question of what we are measuring and what we want to measure. What I’m always surprised about is that the financial side of the story and the physical output and outcome side of the story is rarely analysed together. You’ll have people looking at just the fund flow part and you will have people looking at how many schools were built, how many toilets were constructed. But very few people are actually looking at the process through which money translates into service delivery.
There weren’t too many examples of actual process tracking studies. It was something that was relatively new and it’s been evolving. There’s also a sense that too much information can be dangerous even within the government. So an entire debate in terms of who’s collecting it and for what comes into question as well. Some states are sometimes hesitant to share too much information with the Union government. When we are talking about what’s happening with that ₹100, who’s asking that question? And in some ways, where does that accountability lie?
There were probably some people who would know a little bit about what happened to that ₹100, but that information was never collated nor ever really made publicly available. So you never actually realised that you had the power in terms of using that information about what happens with that money that you have released onwards.

AI has grown tremendously over the years since 2008. And your own role within has evolved quite a bit over that time. Could you give us a sense of how it grew?

It kind of started for me when Yamini asked me to do some research, saying: “Can you tell me how much money actually reaches schools and what is the total education budget? What proportion is going to school bank accounts?” As a young RA, I was of course eager to please my new boss and tried very, very hard to find answers to that question. And I realised that not only was there no work around that, it was impossible to answer. I got training on trying to read budgets. We had a colleague Anit Mukherjee who used to work in the National Institute of Public Finance and Policy who was also partnering with us in the PAISA project and he taught me how to look at budgets and public finance a lot more carefully. I would spend hours in the NIPFP library. Now at least a lot of the state budgets are online. At that time there were 7-8 volumes or maybe even more – thick documents that you have to go through line by line and try to copy onto your laptop. It was very clear that that information just didn’t exist and that’s when we realised that we could fill that gap and try and collect the data.
PAISA started work from when money leaves the government treasury. And to answer this question, we realise that we need to understand government functioning – the links between plans, allocations, expenditures, institutions. How do funds actually flow through the system? What are the links between the needs at the school level and increased expenditures? It was a really exciting time, because, at least in India, public expenditure tracking surveys weren’t very common. Even then, we were very clear that we have to go beyond just expenditure tracking to process tracking.
Expenditure tracking will tell you how much of that ₹100 reaches the ground. Process tracking will also tell you a little bit about the how and the why. The other new aspect we started working on was that we were actually looking at and collecting and analysing district-level data. At that time, the district was never really studied as an implementation unit.
Data, of course, was really, really hard to get. I remember once visiting a government office and collecting some data from them from their computer and then my colleague went a week later and recollected that data, and the two data sets were completely different. This was definitely not a question of corruption. It was just a question of capacity. There were always adjustments that needed to be made and very genuine ones. If money doesn’t come, you can’t stop paying someone’s salary. Money is fungible. So you take it out of a different budget head and then you have to adjust it later. What was really unique for me is we realised that the government itself is not just a monolithic creature. There are people who are very, very real. Who have their own incentives, motivations, and many of them are trying their level best to get things done, but they are also struggling with lack of information, lack of capacity and their own struggles.
Also, for information to be a catalyst for change, people need to be involved in the process of collecting that information. So we did not rely on survey firms. In fact, there were very few survey firms at that time as well, but we would go to NGOs, students and citizens from the same district or the same block and actually get them to collect this information. We were able to get citizens in that sense to participate in governance processes.
One of the best things that we did was partner with ASER Centre in our initial days where we learned a lot. It also allowed us to mobilise a lot of people in the data collection process. In fact, when we were partnering with ASER and coming up with the PAISA national reports, which were part of the ASER report, PAISA became India’s first large-scale citizen-led expenditure tracking survey or process tracking survey. The other thing that we learned through this process was that we hadn’t seen efforts to transfer knowledge on how to do this – how do you create the tools and instruments to generate regular reliable information on service delivery? So we spent a lot of time in our early days making posters, creating ‘how to’ toolkits so that anyone could use this methodology and try their hand at collecting this information bottom up.

For context, for the reader, PAISA is today the flagship research program under AI. It stands for Planning, Allocations and Expenditures, Institutions: Studies in Accountability.

Thank you for remembering the full form. Even now when our state colleagues go to government offices we have to spend some time telling people that we’re not ‘Paisa’ people. Paisa doesn’t mean that we’re going to give you money or take money from you. Since 2010, when PAISA started, we now have full time colleagues in five states of India. And while we started in the education sector, we’ve managed to use that methodology now in different sectors like health, nutrition, sanitation as well as in local governments, at the gram panchayat level.
There were roughly 3 steps: The first was building evidence and research so that was a lot of the desk-based work where we would study the three P’s – the public finance aspect, the processes and the people. Next was going a little deeper into the administrative data that was available, but also coming up with these tactical scalable tools where we could actually collect information on gaps that we saw. This was also accompanied by the capacity building initiative to empower citizens, by getting them to be part of this tracking exercise and use that data collected themselves as well. And then finally, we shared that information not just with the highest levels of bureaucracy, but through the government hierarchy. I think again that was something that was a little unique. Earlier, there was this feeling that a lot of the policy research rests in Delhi or at the state level. But we used to go and speak to district magistrates to cluster resource coordinators to block level officials to school management committees, sharing with them findings.
I remember in December 2015 we had conducted this large-scale PAISA survey where we had combined different schemes – ICDS, Sarva Shiksha Abhiyan and Swachh Bharat Mission. This is after 2014 when the Finance Commission had talked about more untied resources. So we wanted to understand what’s happening with schemes in particular. But when we went to the ground where it found very pretty much similar things that we had seen before, which was the structure challenges of the public finance system, who’s not receiving their money or not receiving it on time. Instead of presenting a report to the government, we ran something that we called PAISA dialogues.
The idea of a dialogue was that instead of circulating reports, we would try to share research findings through a dialogue with certain block-level implementing officials. The overall objective was to leverage research findings to catalyse the discussion on how to improve implementation. So between May and December 2016 we held a total of 40 such PAISA dialogues across ten districts in which we were working. And it was amazing. The kind of response that we got, it forced us also to ask ourselves: Where does our role end? Does it end in just identifying the problem? Or does it also mean that we need to force ourselves to really think about what possible solutions could be? That was a big journey for PAISA. We have since then moved towards articulating the Accountability Initiative’s vision from just accountability to responsive governance.

One of your key insights is that it is not just a matter of us outside the government not fully understanding these processes, but many people within government not having the information either. How have governments at different levels responded to PAISA and your work?

I think we’ve been lucky that there’s been limited scepticism. A senior official from the Elementary Education Department in Purnia in Bihar summarised what we do better than we could:
“The flow of funds through various levels of the government is very similar to the flow of blood from the heart to various parts of the body. If there’s a blockage somewhere, it affects the entire body. So in that regard, PAISA studies do the work of a physician.”

That is what we had hoped would happen. To diagnose where that blockage is and share that with the government because at the end of the day, all of us are trying to ensure that service delivery improves. The intent was never to point fingers, but literally emphasising that this is our problem, not just yours, and so that we can have a meaningful discussion on solutions as well.
I remember relatively early on in 2013, we got a request from the Ministry of Human Resource Development, now the Ministry of Education. They came and asked us to undertake a PAISA study for the midday meal scheme. And I remember being really confused. I had gained confidence by then to actually ask government officers questions. So I remember asking him, ‘Sir, you already have this information or even if you don’t, it’s so easy for you sitting in the ministry to collect this information. You don’t need us. You could literally pick up the phone and try and get all the other officers to collect it for you.’ And I remember so clearly that he turned around and said to me, ‘as an administrator and a researcher, we have very, very different lenses and we’re able to gather different sets of information.’ And I remember him also saying that ‘I don’t just want to hear information that people think I want to hear, but I want to be proved right or proved wrong in my own assumptions.’
That has been really heartening. Even when governments often have access to this information, they have seen value in the kind of work that we’ve been doing. I remember again a district administration coming to us when there was a lot of pressure to meet the Swachh Bharat Mission target of October 2019. There were a lot of construction activities happening on the ground. While the SBM approach talked about it being a lot more demand-led focus on behaviour change, at that time, it was the last year of the mission, a lot of the focus was just supply and infrastructure. This official asked us to use our techniques to ground truth through actual observations whether toilets are there, whether they are being used, but also to understand from the service providers what they thought about this campaign. I really liked the fact that he was curious to know actually what the CEOs (Chief Executive Officer), the gram panchayat sarpanch to the block development officer, what they actually thought about the campaign. To what extent did they believe in it and how it had impacted their work.
It’s much easier for people to open up when it’s not your boss asking you the question, but someone who is independent, going in as a researcher, having conversations with people about what they do and how they feel and their motivations, their incentive, their means, their opportunities. It’s amazing how receptive the government has been for us to kind of come in and help them either think through their own research questions, but also validate, verify and question.

This is part 1 of 2 for our CPR Perspectives conversation with Avani Kapur. We’ll be back in 2 weeks with part 2.