In November 2022, ISRO published a report titled, ‘Decadal Change of Night Time Light (NTL) over India from Space (2012–2021)’. To quote from this, “Night Time Lights (NTL) acquired by satellites is nowadays becoming one of the indicators to analyse wide-range man-made activities by correlating NTL with land use patterns, socio-economic parameters (like GDP, poverty, population, electrical consumption, crime rates, etc), environment variables (like climate change, carbon emissions, etc).”
To quote a bit more, about the changes from 2012 to 2021: “Overall at national level, normalised NTL radiance (Cumulative NTL radiance/Geographic Area) is increased by 43% in the year 2021 w.r.t 2012. Significant increase observed in Bihar, Manipur, Ladakh and Kerala. Good increase observed in Arunachal Pradesh, Madhya Pradesh, Uttar Pradesh and Gujarat. Moderate increase observed in Lakshadweep, Maharashtra, Tamil Nadu, Jharkhand, Haryana, Punjab, West Bengal, Uttaranchal, Karnataka, Odisha, Telangana, Andhra Pradesh, Nagaland, Chandigarh, Himachal Pradesh, Rajasthan, Tripura, Goa, Chhattisgarh, Assam, Andaman and Nicobar, Meghalaya, Jammu & Kashmir. In most of the States, a fall in NTL cumulative radiance observed in the year 2020, and this could be the impact of COVID-19 pandemic. Again, in most of the States there is a rise observed in the year 2021 for NTL cumulative radiance.”
This sounds frightfully dense and turgid, as indeed it is. The visual impact of this is much more when you look at all-India maps for 2012 to 2021. Stated simply, India is much better lit at night in 2021, compared to 2012. I said India, but the report zeroes in on states, and within states, on districts. Visually, the spread of light at night is striking. For example, in 2012, the lighting in Bihar centred around Patna and had Muzaffarpur, Darbhanga, Begusarai, Gaya and Bhagalpur. In 2021, the radiance had spread throughout the state. What caused this? Electrification of households under the Saubhagya Scheme and the construction of national highways are obvious answers.
For several years now, NASA has released images of the world at night, including for India, and for the last 20-odd years, there has therefore been interest in using such data as surrogate indicators for socio-economic development.
The Economic Survey 2021–22 had a chapter on tracking development through satellite imagery and cartography. To quote from that, “Night-time luminosity provides an interesting representation of the expansion of electricity supply, the geographical distribution of population and economic activity, urban expansion as well as growth of ribbon developments between urban hubs.” When regular data surfaces with a time lag, NTL, which doesn’t have such a time lag, has natural appeal.
There is some cross-country empirical work correlating NTL (luminosity) with GDP. No one doubts there is correlation. There have also been a few studies correlating NTL with state-level or district-level economic activity within India. The problem is the obvious one—correcting NTL for extraneous factors—phases of the moon, seasonal vegetation, cloud cover, atmospheric emissions, and even the exact position of the satellite at the time. There is a blurring effect that makes urban agglomerations seem larger than they actually are.
In other words, NTL data needs to be processed and cleaned. They have measurement errors and biases. The ISRO report isn’t based on such clean data. Therefore, one shouldn’t jump to conclusions.
The recent Economic Survey 2022–23 reported that 29 million rural households have been electrified under Saubhagya. “The government launched the Pradhan Mantri Sahaj Bijli Har Ghar Yojana—Saubhagya in October 2017 with the objective to achieve universal household electrification by providing electricity connections to all willing un-electrified households in rural areas and all willing poor households in urban areas in the country by March 2019. The Scheme involved the organisation of camps in villages/clusters villages for on-spot registration and the release of connections. The connections were given free for economically poor households and for others Rs 500 was charged after the release of the connection in 10 installments. The Saubhagya scheme has been successfully completed and closed on 31 March 2022.”
So far so good. Other than highways, electrification under Saubhagya is bound to improve NTL, and it is plausible that Saubhagya had a role to play in the improvement reported by ISRO. But consider this. What kind of light connection did Saubhagya provide? One with LED bulbs. I am going to use this as an example of statistical nitpicking. Suppose no new connections had taken place, but the old-fashioned bulbs had been replaced by LED bulbs. NTL would immediately have increased. That’s the nature of NTL and luminescence with LEDs. We can think of cities where street lighting has switched to LED. Immediately, the roads seem brighter.
This is indeed nitpicking, and Saubhagya, successful as it was, must have contributed to the improvement of NTL. NTL is descriptive. Once the data has been cleaned, it describes a factual state of affairs.
However, there must care about interpreting those facts and deducing the causation. The nitpicking is meant to illustrate nothing more than that.
Let me finally quote from a World Bank paper. “This paper estimates the impact of a differential relaxation of COVID-19 containment policies on aggregate economic activity in India. Following a uniform national lockdown, the Government of India classified all districts into three zones with varying containment measures in May 2020. Using a difference-in-differences approach, the paper estimates the impact of these restrictions on the nighttime light intensity, a standard high-frequency proxy for economic activity. … The analysis finds that nighttime light intensity in May was 12.4 per cent lower for districts with the most severe restrictions and 1.7 per cent lower for districts with intermediate restrictions, compared with districts with the least restrictions. The differences were largest in May, when the different policies were in place, and slowly tapered in June and July.”
Such data and analysis are useful. But NTL isn’t quite a substitute for conventional data. It isn’t robust enough.
Chairman, Economic Advisory Council to the PM