What Satellites Reveal About India's Unmeasured Economy

Decisions that matter for growth and welfare are taken at the level of districts, blocks, peri-urban corridors and villages. Yet India’s measurement systems still speak largely in the language of states and national averages. That mismatch creates uncertainty for both governments and firms. Public programmes are designed with imperfect information and private investors make location choices with partial visibility.

India’s official statistics have improved meaningfully. The Ministry of Statistics and Programme Implementation (MOSPI) and the National Statistical Office (NSO) have modernised operations, increased transparency and expanded the availability of high-frequency indicators – data that is updated more often. For instance, the redesigned Periodic Labour Force Survey (PLFS) now produces monthly urban labour market indicators and quarterly estimates, giving policy makers a much more up-to-date picture of jobs and employment. This is a major step for economic debates based on current data rather than assumptions.

The remaining gaps concern granularity and local frequency, i.e., how detailed the data is and how often it is updated locally. Most official data observe the economy at the level of states or districts, while implementation happens far below that scale. Investments in physical infrastructure, welfare programmes and logistics networks depend on information that is sharper than a district average and more frequent than a major survey round.

In short, India’s macroeconomic story is well measured, but its microeconomic geography remains harder to see. Put another way, we understand India’s economy at the national level, but still struggle to see how development varies within districts and regions.

How satellite signals add complementary intelligence

Satellite sensors generate persistent proxies for economic activity. In other words, they provide consistent indirect indicators of economic activity. For example, night-time lights can help assess market intensity or built-up density, which in turn help understand the levels of urbanisation or land use. Similarly, how surfaces reflect light can signal construction activity or flood damage. Machine learning methods can translate these signals into estimates mapped onto very small geographic areas that can be updated continuously.

Also read: Govt Data On Farm Fires ‘Does Not Paint Entire Picture,’ Says New Report

Importantly, satellite information is a complement to household surveys and administrative records rather than a substitute for them. As Stanford economist Marshall Burke has observed, “In combating poverty, like any fight, it is good to know the locations of your targets.” The message is straightforward. Satellites do not replace ground truth – that is, field-verified data. They indicate where to look and when to verify.

What the figures show: a case from Uttar Pradesh

The proof of concept developed for Uttar Pradesh shows how this works in practice [see Note below]. The state was divided into small geographic grids of approximately two and a half kilometres each. All the grid cells were enriched with night-time light intensity and population density. A convolutional neural network then identified patterns in the satellite images and an algorithm  grouped similar areas together without pre-defined categories. Different types of development – dense urban corridors, semi-urban zones and sparsely developed areas were thus identified.

The resulting maps show a striking pattern:

• Dark blue areas represent high night-light intensity, dense built-up zones and relatively stronger infrastructure.
• White or paler areas show low night-light intensity, dispersed settlement, and thinner economic activity.
• Green dots indicate clusters with better infrastructure or development, often concentrated within the dark-blue corridors.

Map 1. Urban activity hotspots overlaid on nighttime lights in Uttar Pradesh (2020)

Advertisement

Map 2. Multidimensional Poverty Index (MPI) score, district-wise, from NFHS, 2019-21. The colour represents MPI score, with intensity rising from left to right. Data: Niti Aayog

When read together, these two visualisations (maps 1 and 2) reveal spatial gradients that district averages hide. In many cases, a district classified as “average” on poverty has pockets of concentrated infrastructure and bright night lights that coexist with areas that are significantly less developed. This is exactly the information that a state planning department, welfare ministry or infrastructure agency needs to plan and allocate more effectively.

The validation step is equally important. When the clusters were compared to multidimensional poverty indices (MPI) derived from National Family Health Survey (NFHS)-based work, the patterns aligned intuitively, matching what we would have expected based on known patterns of poverty.

In Map 2, the encircled region corresponds to a bright and dense corridor in the satellite output seen in Map 1, as well as to lower MPI in the published district-level reports. This is early but promising evidence that development patterns inferred from satellite data can meaningfully track welfare indicators, despite being generated without a dedicated survey.

Satellite-derived spatial signals therefore help prioritise where limited survey teams and funds should be deployed.

A policy application in Assam

A second application involves integrating these spatial proxies with ongoing work to produce District Domestic Product (DDP) estimates in Assam. Linking satellite signals on infrastructure and built environment with DDP estimates could provide planning departments with a more continuous and a more local, area-by-area picture of economic activity.

Such intelligence is actionable, because scheme design, capital expenditure approvals and welfare implementation happen at sub-district scales.

Global precedents show viability

Several countries already use satellite information for policy-relevant measurement. Researchers have used night-time lights to estimate subnational Gross Domestic Product (GDP) in Kenya and Rwanda, especially where survey coverage is thin. Brazil’s National Institute for Space Research has used satellite monitoring for decades to enforce environmental regulation and track deforestation.

Also read: Rethinking Poverty: Why India's MPI Needs an Update

In the United States, remote sensing is part of the workflow for crop insurance adjustments and disaster relief verification. These examples show that spatial intelligence can be institutionalised when validation and operational procedures are in place.

Policy asks for India

Three practical reforms would allow India to harness these tools responsibly. First, statistical authorities could publish validation protocols explaining how satellite-derived estimates will be compared with household surveys, administrative records and national accounts. Validation makes the methods transparent and builds trust in policy decisions.

Second, spatial information should be embedded into routine workflows. Planning departments, welfare ministries and infrastructure agencies can use spatial intelligence to prioritise survey resources, adjust bidding timelines and monitor project progress.

Third, there is a need for interoperable geospatial layers – maps and datasets that can be easily combined with each other. Satellite signals become far more valuable when aligned with road networks, land records and beneficiary databases.

Measurement matters for growth and welfare

Measurement innovations have shaped India’s development trajectory before. National accounts enabled macroeconomic management; the National Sample Survey and the NFHS improved welfare and health planning; digital administrative databases now enable direct benefit transfers.

Frequently updated and highly local geographic data is the next logical step in this progression. It reduces waste, strengthens targeting and lowers uncertainty for private investment. Satellite data and machine learning will not solve every measurement challenge. However, when validated and integrated intelligently, they help governments act more precisely, in the right places, and help India allocate public money more efficiently and fairly.

Note: The boundaries used in these figures are based on OpenStreetMap and have not been standardised for official administrative use. The exercise was replicated as part of an Asian Development Bank course and should be viewed as an academic proof-of-concept. Outputs can be further improved through higher resolution remote sensing and additional satellite sensors.

Uday Khanna is a data scientist at the Center of Data for Economic Decision-making (CoDED), Pahlé India Foundation. Payal Seth is head of CoDED and fellow at Pahlé India Foundation.

The Wire is now on WhatsApp. Follow our channel for sharp analysis and opinions on the latest developments.