In this interview with The Policy Edge, Dr Gokarn reflects on how state capacity is built and exercised in practice: from decision-making under uncertainty and the limits of paper systems, to designing interventions that endure beyond individual tenures. Drawing on experiences across policing and administration, he offers insights into how governance operates on the ground, and what it takes to move from individual action to system-level impact.
Indian cities are governed by master plans that are often comprehensive on paper but weak in shaping actual urban growth. Is this primarily an implementation failure, or does it reflect deeper flaws in how planning itself is conceptualised?
If you look at most master plans, they are quite detailed: land use, zoning, road networks, everything is laid out. The issue is not that planning is absent, but that it often assumes a static city, whereas cities grow dynamically.
When we started reviewing master plans across cities in Uttar Pradesh, one of the first things we did was to look at how the city had actually grown over time using satellite imagery and even night-time light maps. That pattern is rarely aligned with what the master plan had envisaged.
The problem, therefore, is not just implementation; it is also conceptual. Plans are often framed as regulatory documents that define what is allowed, rather than frameworks that reflect how the city is evolving.
Plans are also disconnected from incentives. If the market is pushing growth in one direction, but the plan restricts it without providing viable alternatives, development will still happen, but informally or in ways that bypass the plan.
Planning, therefore, should guide and channel growth, not assume that growth will follow the plan.
Your introduction of ecological layers, such as mapping all water bodies irrespective of current land use, challenges existing property and development patterns. How should policymakers navigate conflicts between ecological integrity, legal ownership, and market incentives?
When we introduced what we called the “blue layer”, the idea was very simple: every water body, including ponds, tanks, drainage channels, must be mapped as it originally existed, irrespective of its current status.
In many cases, these water bodies no longer existed: having been encroached upon, filled up, or built over. But in the land records, they are still recorded as ponds or tanks. The question is: how did this transition happen?
If you ignore that and treat the current status as the baseline, you are effectively legitimising the loss of ecological assets. Over time, this leads to flooding, waterlogging, and depletion of groundwater.
Our approach was to first establish ecological reality as a layer in the master plan. This did not immediately resolve the conflict with existing ownership or development, but it ensured that future decisions are informed by that reality.
Similarly, with the green layer, we mandated that a certain proportion of the city must remain as green space. These are non-negotiables, because without them, the city becomes unlivable.
There will always be tension between ecology, law, and markets. Policy must ensure that ecological considerations are not completely overridden.
Your work involved moving building approvals towards a trust-based, automated system. It shifts the state’s role from ex-ante control to ex-post enforcement. Under what conditions can such a model sustain compliance without increasing the risk of evasion or capture?
Traditionally, building approvals were highly control-driven. Files would move through multiple layers, and approvals would take time. This created delays and discretion.
We introduced a system where plans submitted by architects are approved automatically, provided they are fully compliant with the master plan, with no violation of blue or green layers and no encroachment on restricted areas.
We called it a “fast pass.” The idea was to shift from the government checking every file in advance, to a system where compliance is built into the process itself.
For this to work, two things are important. First, the master plan must be clear and unambiguous. If the rules are unclear, automation will not work. Second, accountability must shift to the professional or the architect submitting the plan.
Any violation is traceable, as the digital system automatically generates red flags where there is an intersection with restricted layers.
It is not that monitoring disappears; it changes form. Instead of delays and discretion upfront, you have transparency and traceability built into the system.
You have argued that urban planning should enable, rather than constrain, economic growth. In practice, how can planners reconcile the organic expansion of cities with infrastructure needs and long-term sustainability?
Cities grow because economic activity drives them. People move where there are opportunities, and development follows that movement. If planning does not recognise this, it will always lag behind.
We first focused on understanding the engines of growth in each city: industrial clusters, commercial hubs, connectivity corridors. The question then became:: how do you support that growth with infrastructure, rather than restrict it?
At the same time, unstructured growth cannot be allowed. That leads to congestion, lack of services, and long-term inefficiencies. The approach, therefore, is to combine flexibility with structure. For example, by allowing township development with integrated infrastructure such as roads, sewage, and open spaces, one can channel growth into planned formats.
We also ensured that certain constraints like ecological layers are non-negotiable; so that growth happens within a framework.
In essence, planning should not try to control where growth happens, but prepare for it by providing infrastructure, ensuring connectivity, and setting boundaries where necessary.
In GST administration, you identified leakages arising not from policy design but from weak compliance linkages within government spending itself. What does this suggest about the limits of tax reform without corresponding administrative reform?
When GST was introduced, a lot of focus was on the design aspects such as rates, structure, compliance requirements. But what we found in practice was that even within government systems, there were gaps.
For example, when a department pays a contractor, it includes GST in that payment. But there was no systematic check to ensure that the contractor had actually deposited that GST with the government. Effectively, the government was paying GST, but not necessarily receiving it back. This is a system issue, not a policy one.
We linked payments to compliance. Before releasing subsequent payments, departments had to verify that the GST from previous payments had been deposited. This created a direct incentive for compliance. Once this was enforced across departments, collections improved significantly.
Policy reform alone is not sufficient. Without administrative systems ensuring implementation, leakages will persist.
The integration of multiple datasets and the use of AI allowed for large-scale detection of discrepancies in tax filings. To what extent can data-driven enforcement substitute for traditional administrative capacity?
One of the challenges in tax administration is the volume of data. For any given entity, there are multiple datasets: returns filed, e-way bills, electricity consumption, and so on. Manually analysing this is not feasible.
We worked with institutions like IIIT to develop systems where these datasets could be integrated and analysed automatically. The system would identify discrepancies where reported activity did not match other indicators and generate notices.
This allowed us to cover a much larger base than would be possible manually. In that sense, data-driven systems can significantly enhance administrative capacity; but they do not replace judgment. Data can indicate discrepancies, but interpretation and action still require human intervention. There are also cases where data may not capture the full context.
Technology is an enabler. It expands the reach of the system, but it does not eliminate the need for institutional capacity.
Across your work, a recurring theme is the shift from individual-driven governance to system-driven outcomes. What are the minimum conditions required for a governance system to become self-enforcing rather than dependent on continuous administrative intervention?
A system becomes self-enforcing when it aligns incentives, reduces discretion where unnecessary, and builds in feedback loops.
If compliance depends on constant monitoring by an individual, it is not a system but an effort. The moment that effort is withdrawn, outcomes change.
The objective is to design processes where the desired behaviour is the easiest behaviour. For example, if approvals are automatic when rules are followed, people are more likely to comply.
Clarity is also important. If rules are ambiguous, systems become dependent on interpretation, which brings back discretion.
The minimum conditions are: clear rules, aligned incentives, traceability, and feedback. Once these are in place, the system can function with less reliance on individual intervention.