Notes

Why governments need external partners for reform

Government systems face an inherent challenge when trying to improve their own operations. They must be confident that new approaches will work before implementing them, but gaining that confidence requires experimentation and risk-taking that goes against institutional culture.

Nagpal explains why external partners become essential:

"Government systems, when government itself is doing something, it needs to be sure that that thing will work. Whereas for it to be sure that it works, it needs some degree of risk capital, risk appetite. And that typically is provided by external partners who can then do that experimentation, learn from that, and then scale that to business as usual only when it works."

This creates a natural division of labor. External organizations can pilot performance management systems, test outcomes-focused tracking approaches, and experiment with different measurement frameworks. Only after proving effectiveness do these innovations become part of regular government operations.

The resistance isn't just about risk aversion. Performance measurement and outcomes tracking inherently challenge existing ways of operating. There's "an intrinsic defensiveness to changing the way we operate or having our performance be measured." External partners can help build the case for change by demonstrating concrete improvements in citizen outcomes.

Flipping from evidence-based policy to policy-driven evidence

Traditional approaches to government reform often assume that generating good evidence will naturally lead to policy changes. Nagpal's experience suggests this framework is backwards and often counterproductive.

The standard model treats evidence as objective and universal - academic researchers generate findings and expect governments to implement them. But this ignores fundamental questions about what constitutes legitimate evidence and who decides what counts as proof.

Nagpal argues for reversing this relationship:

"A lot of my approach for the last seven and a half years has been to flip that around and sort of to think about policy-driven evidence rather than evidence-based policy. People who constitute the Indian state, be they political leadership, ministers, their advisors, be it bureaucrats, middle management, frontline workers, they are incredibly hardworking people who are optimizing within very difficult constraints."

This reframe requires understanding the specific constraints that government officials face daily. India's complexity means that relationships between state and society vary dramatically across regions. What works in one context may fail completely in another, not because the evidence was wrong, but because it didn't account for local political economy factors.

The policy-driven approach starts by understanding what government officials need to know to make better decisions within their existing constraints, then generates evidence to meet those specific needs. This requires "a lot of humility" from external partners and deep appreciation for why current approaches exist.

The tension between evaluation and learning

Using data to judge programs versus using data to improve them creates fundamentally different dynamics. Nagpal has observed this tension across his work with both government and nonprofit partners.

Evaluation tends to trigger defensive responses, especially when results are negative. This isn't unique to government - it's a basic human reaction to having performance questioned. When someone tells us our work isn't achieving its goals, "I feel unsafe, I feel sort of hurt, I want to fight back, you know, because my ego is important in that moment."

Learning requires overcoming this initial defensiveness. Even when evaluations show programs aren't working, partners don't immediately respond with "oh, that's excellent, great, let's shut down this program." Instead, they become defensive: "no, but you haven't thought through this aspect, that aspect... this context matters."

The solution involves treating evaluation and learning as distinct functions that may need to happen in different parts of government. For learning to occur, external partners must carefully consider "what language and what vocabulary should we use, how should we position it, how should we time and sequence it."

Nagpal explains his prioritization:

"We definitely prioritize the learning elements of it, because it is learning which will lead to improvements over time. As we learn new things, we need to process it ourselves and we need to co-process it with our sort of partners in government and non-profits."

This means accepting that immediate program changes may not happen after negative evaluations. Instead, the goal becomes building systems that help government officials make better decisions over time as they incorporate new information into their mental models.

Sources of optimism about state capacity

Despite the challenges of government reform, Nagpal maintains optimism about the trajectory of the Indian state. This optimism rests on structural changes in political incentives and technological capabilities.

Political leaders have increasingly learned that effective service delivery translates into electoral success. This creates a virtuous cycle where "the political leadership knows that if they deliver, they get reelected, they care about delivery." This represents a shift from previous eras where other factors may have dominated electoral calculations.

The commitment to delivery requires improving state capacity. Politicians who promise better services to voters need bureaucratic systems capable of actually delivering those improvements. This creates top-down pressure for the kinds of performance management and outcomes tracking that external partners can help implement.

Technological improvements also support this trajectory. India has built "a lot of digital systems that are making the quality of the data better, more reliable." Better data enables better decision-making and makes it easier to track whether programs are achieving their intended results.

Nagpal frames the work as carrying moral weight beyond technical considerations:

"It's kind of like a moral imperative to work with the state. The state, the Indian state serves one point four billion people, and if we're able to make even a tiny, tiny dent in its capacity, in helping it improve, I think it would be a really important landmark for human welfare."

The scale means that small improvements in effectiveness can have enormous aggregate impact. This makes the patient work of building measurement systems, supporting capacity building efforts, and helping government officials learn from data worthwhile despite the slow pace of institutional change.

Supplementary Resources

• Four Challenges to Building Capacity in Government by Madhav Seth (IDinsight article)
• Strengthening State Capacity to Improve Service Delivery for Citizens in India by Karan Nagpal et al. (IDinsight project)
• Enhancing Evidence Use in Government Social Programs by T. Arthur Chibwana (IDinsight article)

The complete transcript file is available to download below.