Key Details
The RBI October 2026 Bulletin article Nowcasting State-Level Economic Activity in India with Night-Time Light and Google Trends tests whether more frequently available data can help fill the gap between annual releases of Gross State Value Added (GSVA), a measure of State economic output.
States tested: Maharashtra, Karnataka and Gujarat.
Signals combined: Changes in light recorded by satellites at night, Google searches across economic and social categories, and past GSVA growth.
Result: Compared with a model using past growth alone, the combined model reduced estimation error by approximately 38% in Maharashtra, 54% in Karnataka and 62% in Gujarat.
Different roles for the signals: Night-time lights contained information useful for predicting subsequent activity; Google searches mainly added information about activity in the current period.
Annual State Data Leave a Gap in the Quarterly Picture
Official GSVA figures are released annually, typically with a 12–18-month lag, making it difficult to assess a State economy as conditions change. The researchers test a way to produce earlier quarterly estimates. Satellite images capture changes in night-time illumination, while Google Trends records changes in search activity—for example, searches connected with jobs, shopping or housing.
RBI researchers had previously studied night-time lights as an indicator of Indian economic activity. The contribution here is combining refined satellite data with search behaviour and testing whether the two improve short-term State estimates together.
The Combined Model Performs Better in Three States
The model using both signals produced lower estimation errors than one relying on past GSVA growth alone. It also performed better than a model using night-time lights without Google Trends. The gains varied by State, suggesting that one signal does not work equally well everywhere.
The researchers also tested whether areas with little night-time light should be excluded because sectors such as agriculture are less directly visible from space. Excluding them did not consistently improve estimates of total GSVA. Economic activity in less-lit areas can still be connected to spending and production in illuminated settlements.
The Quarterly Benchmark Is Itself an Estimate
The result needs a careful reading. Because an independent official quarterly GSVA series is not available for these States, the study constructs its quarterly comparison series from annual State GSVA and the national quarterly GDP pattern. The reported reductions in error are measured against that constructed benchmark — not against independently observed quarterly State output.
The evidence supports further use and testing of these data as supplementary early indicators. It does not establish that the model can replace official State accounts or perform equally well in States with different economic structures and data coverage.
Policy Relevance
Earlier State-level signals could help governments and researchers notice changes in activity before annual figures arrive. Their value would depend on testing the method beyond three industrial States and being clear about what its estimates measure. For official statistics, satellite and search data are best treated as complements to survey, administrative and national-accounts data — not substitutes for them.
Follow the Full Piece Here: Nowcasting State-Level Economic Activity in India with Night-Time Light and Google Trends