What Good Policy Actually Looks Like

Every country has an abundance of good policy analysis, and India's digital public infrastructure is one of the rare cases where that analysis actually got executed at scale. UPI processed 172 billion transactions in 2024; Aadhaar-based direct benefit transfers have, by most evaluations, reduced welfare delivery leakage by changing the transaction structure from multi-intermediary to direct, though findings vary across programs and inclusion errors have also been documented. These are the measurable outcomes of policy that simplified, aligned incentives, and generated feedback. Think tanks, academics, consultants, and former officials produce an enormous quantity of well-researched recommendations for how governments should address problems. Almost none of it gets implemented. Of the small fraction that does, a smaller fraction still gets implemented in ways that actually produce the intended outcomes.

The bottleneck in development is not the quality of policy thinking. It is the capacity to turn thinking into outcomes through real institutions with real constraints.

The implementation gap

There is a concept sometimes called the "implementation gap", the distance between what a policy says it will do and what it actually does. This gap exists everywhere, but it is particularly consequential in developing country contexts where the institutional infrastructure for execution is thinner and the problems being addressed are more complex.

The implementation gap is not just about corruption or incompetence, though both contribute. It is about the fundamental difficulty of designing policy in a context and then executing it in a different context. The people who write policy are typically educated, urban, and operating in environments with strong institutional support. The people who implement policy are often working in contexts that are none of those things, with much less information, much higher variance in conditions, and accountability structures that often reward compliance with rules rather than achievement of outcomes.

A policy that works in the environment where it was designed may fail in the environment where it is deployed, not because it is wrong but because the assumptions embedded in its design do not hold in the implementation context.

What good looks like in practice

The policies that have worked best in India in recent decades share some characteristics that are worth naming.

They simplify. The most successful interventions have typically taken complex, multi-step processes and reduced the number of steps, handoffs, and discretionary decisions. GST replaced a patchwork of state and central taxes with a single national system. Aadhaar-based direct benefit transfers replaced a multi-intermediary delivery chain with a direct transfer. Simplification reduces the surface area for failure, reduces the opportunities for rent-seeking, and reduces the cognitive load on both implementers and beneficiaries.

They align incentives. Good policy tends to produce the desired outcome as the path of least resistance for the people involved in its execution, rather than requiring people to resist their own interests for the policy to work. Schemes that require frontline workers to exert effort with no accountability or reward structure attached to outcomes tend to fail. Schemes that make the desired behavior the default, or tie some form of accountability to it, tend to do better.

They generate feedback. The policies that have scaled successfully have built in mechanisms for learning, data collection, outcome monitoring, the ability to identify what is not working and adjust. The failure mode of many development programs is that they execute for years without generating clear evidence of whether they are working, because the evaluation infrastructure was not built alongside the program.

The political economy constraint

The hardest thing about good policy is that the characteristics that make it effective are often in tension with the political characteristics that allow it to get made and sustained.

Simplification often threatens the interests of intermediaries whose position depends on complexity. Incentive alignment sometimes requires acknowledging that current incentives are misaligned in ways that implicate powerful interests. Feedback mechanisms generate data that can be politically inconvenient.

Good policy analysis that ignores these constraints is incomplete. The design problem is not just "what would work?" but "what would work and can actually be implemented given the political economy?" These are different questions, and conflating them produces recommendations that are technically elegant and practically inert.

The most interesting policy entrepreneurs are the ones who have figured out how to navigate both questions, who can design interventions that would actually work and build the political coalition to get them implemented. These people are rare and they are doing more development-relevant work than most people who write about development appreciate.

The hardest thing about good policy is that the characteristics that make it effective are often in tension with the political characteristics that allow it to get made and sustained. Simplification threatens intermediaries. Incentive alignment implicates powerful interests. Feedback generates inconvenient data. Good policy analysis that ignores these constraints is incomplete.

Good policy is not the smartest analysis. It is the analysis that gets executed well enough to change outcomes for actual people.

Manas Majhi grew up in Junagarh, Kalahandi, Odisha. He writes about opportunity, development, and the systems that fail to distribute either equitably. He is the founder of Majhi Group and Majhi OS.

See also: State Capacity Is the Ultimate Competitive Advantage, Building for a Billion People, Digital Public Infrastructure

Sources

NPCI UPI Product Statistics

Direct Benefit Transfer: Ministry of Finance, Government of India

IDinsight: Aadhaar and Direct Benefit Transfer evaluation evidence