Key messages
Generated automatically from the results. The badge shows the confidence level and the reason for it (sample size, significance, sensitivity to thresholds).
National light activity, month by month
The solid line removes seasonal variation (snow, vegetation, holidays): read this line to judge whether activity is rising or falling.
Source: VIIRS DNB monthly (NOAA/EOG), cleaning and STL seasonal adjustment.
The map could not be loaded (mapping library
unavailable). The tables on the other pages remain available.
National series: observed, seasonally adjusted, trend
Grey bands mark known events; purple lines, statistically detected structural breaks; orange dots, anomalous months (deviation of at least 5% after removing seasonality).
Source: VIIRS DNB monthly (VCMSLCFG); robust STL on log(SoL); breaks from piecewise-linear segmentation.
Year-on-year growth (same month)
Comparing a month with the same month of the previous year neutralises seasonality.
Source: VIIRS DNB monthly; own calculations.
Regions (ADM1), seasonally adjusted series
Index 100 = first-year average: this compares growth paths, not sizes.
Source: VIIRS DNB monthly; STL by region.
Lights and GDP
Relationship between log(lights) and log(real GDP)
The slope β is the elasticity: on average, +1% GDP is associated with +β% light.
Sources: VIIRS VNL V2 annual; World Bank (NY.GDP.MKTP.KD); OLS, HAC standard errors.
GDP growth: official and estimated from lights
Out-of-sample estimate: each year is predicted without using it. The last point, if any, estimates a year not yet covered by official statistics.
Sources: World Bank; VIIRS monthly aggregated by year; growth-rate model.
Who is growing, stagnating or declining?
Robust annual growth (Theil–Sen slope) with its 95% interval. "Growth" and "Decline" require a statistically significant trend of at least ±1%/year.
Source: VIIRS VNL V2 annual, gas flares masked.
Where is activity concentrated?
Share of national light emitted by the brightest unit and by the brightest 10% of units.
Convergence
Each point is a unit: growth of light per capita against its initial level. A negative slope means the least-lit units are catching up.
Source: VIIRS VNL V2; population GHSL/WorldPop.
Shocks: depth and recovery time
Trough = maximum deviation of the seasonally adjusted level from its average over the 12 months before the event. Recovery = first month in which that level is regained.
Daily data (Black Marble)
Source: NASA Black Marble VNP46A2, high-quality pixels free of cloud and snow.
Impact by region
Source: VIIRS DNB monthly, seasonally adjusted by region.
Detected structural breaks
A break with no known event within ±6 months warrants investigation: sensor change, outage, unrecorded local event.
Lorenz curve of light
The further the curve lies from the diagonal, the more light (and the activity it reflects) is concentrated in a small share of the population.
Source: VIIRS VNL V2; population GHSL/WorldPop; light per capita of units.
Gini and Theil over time
The Theil index decomposes into inequality between regions and within regions.
Source: calculations on VIIRS VNL V2 and population.
Populations "in the dark"
People living in areas with no detectable light
The shaded band shows the range obtained with different radiance thresholds: it is the uncertainty margin of the indicator.
Sources: VIIRS VNL V2; GHSL/WorldPop; World Bank (EG.ELC.ACCS.ZS).
Where are they?
Source: VIIRS VNL V2; GHSL/WorldPop.
Urban / rural gap
Light per capita in urban areas divided by that in rural areas (GHSL degree of urbanisation).
Download the data
All tables shown can be downloaded. Licences: VIIRS and Black Marble (public domain, attribution requested), geoBoundaries and WorldPop (CC BY 4.0), GHSL (CC BY 4.0), World Bank (CC BY 4.0).