Satellite-based night-time lights data and Google search trends can help estimate quarterly economic growth in Gujarat, Maharashtra and Karnataka, reducing errors in growth estimates by up to 62 per cent, according to a report in the Reserve Bank of India’s (RBI’s) monthly bulletin.
The report was authored by RBI staffers, and the views are those of the authors, not the RBI, it was clarified.
The study, titled Nowcasting State-Level Economic Activity in India with Night-Time Light and Google Trends, combines satellite imagery and digital search behaviour to estimate quarterly growth in gross state value added (GSVA).
It seeks to address gaps in the availability of timely state-level economic data, as official GSVA figures are released annually, typically with a lag of 12-18 months.
The report found that combining the two indicators improved the accuracy of estimates compared with a benchmark model that relied on past GSVA growth.
The reduction in the root mean squared error (RMSE), a measure of estimation error, was around 38 per cent for Maharashtra, 54 per cent for Karnataka and nearly 62 per cent for Gujarat.
The framework uses night-time light data from NASA’s Black Marble satellite dataset and Google Trends indices across 25 search categories. These include searches related to jobs and education, real estate, shopping, finance, travel and automobiles. The model also incorporates past GSVA growth to generate estimates for the current quarter.
Night-time light data captures changes in illumination across a state, which can reflect activity in urban settlements, industrial areas and other locations. Google search trends, meanwhile, provide a measure of digital behaviour and demand-related activity. Together, the indicators offer a more frequent view of economic conditions than annual state income data.
The study found a strong relationship between year-on-year (Y-o-Y) growth in night-time light intensity and overall GSVA growth. The correlation was highest in Maharashtra at 0.90, followed by Karnataka at 0.79 and Gujarat at 0.66. The relationship remained strong and statistically significant even after the researchers excluded quarters affected by Covid.
The findings suggest that satellite-based illumination can capture broad changes in economic activity, while search trends provide additional information that improves estimates. The study also found that night-time light data carried information useful for predicting subsequent GSVA growth, while Google Trends indicators contributed mainly through signals available in the same period.
For the analysis, the researchers processed monthly satellite observations to remove low-quality pixels and background noise before aggregating the data into quarterly measures. They then compared changes in night-time light intensity with official state-level economic growth figures and incorporated the indicators into a statistical model to produce quarterly estimates.
The study covered quarterly observations from 2013-14 to the third quarter of 2024-25 for Maharashtra and Karnataka. Gujarat’s series extended up to the fourth quarter of 2023-24 because of data constraints. The model’s performance was tested against a benchmark using an expanding-window approach, in which estimates were generated using information available up to each period.
The authors also examined a modified measure of GSVA that excluded sectors such as agriculture, forestry and mining, where economic activity may not be fully reflected in night-time illumination.
However, estimates for aggregate GSVA generally performed more consistently, indicating that excluding sectors with lower light intensity did not systematically improve short-term estimates.
The study focuses on Gujarat, Maharashtra and Karnataka, selected as major industrial states. The authors said high-frequency regional estimates could help identify changes in economic activity and provide additional information for short-term economic assessment.
