Imagine you lived in a world where night lights from satellite images tell you instantly about the distribution and growth in economic activity and the extent and evolution in poverty. While such a world is probably still far off, night lights as observed from space are increasingly being used as a proxy of human economic activity to measure economic growth and poverty. In a fascinating 2012 paper in the American Economic Review, Henderson and colleagues found a strong correlation between growth in night lights as observed from space and growth in GDP, basedon data on 188 countries spanning 17 years. They use their estimates for two main purposes: (i) to improve estimates of “true” GDP growth in countries with weak statistical capacity and (ii) to estimate GDP growth at levels where national accounts are typically non-existent (sub-national or regional levels; coastal areas;,…).
The added value of such an approach for Africa is obvious. Most African countries rank low on the World Bank’s Statistical Capacity Indicators , with some countries lacking national accounts altogether. Some African countries are huge (in size), and having sub-national estimates of GDP growth would help identifying leading and lagging areas, and why. For a country such as Kenya, which is starting an ambitious decentralization project, the approach could estimate GDP growth for its 47 newly formed counties to help in their economic planning. Nightlights can even be used to show where the Pirates of Somalia are spending their ransom money. 
Satellite data on night lights are converted into yearly datasets on light intensity per “grid” (a small geographical area corresponding to approximately 0.9 square kilometer at the equator). Light intensity is expressed by a digital number from 0 to 63, with higher values representing higher intensity of light. The result is grid-level panel data on light intensity, which can be used to shed light on growth of relatively small geographical units for which national accounts GDP data is not available. This opens up the possibility of estimating economic growth of cities, secondary towns, rural hinterlands, border areas, etc
Can this approach be used in Africa? Is Africa enough lit, and is there enough variation in night lights as seen from space to gain meaningful insights in growth and poverty and their spatial distribution? Several examples show promise. Take for instance Kenya’s growth story over the past two decades as seen from outer space. Figure 1 shows night lights in Kenya in 1992. Basically we observe a big bright spot in Nairobi, a bright spot down at the coast in Mombasa, and two smaller bright spots at Kisumu and Eldoret. A number of smaller spots are scattered between Nairobi and Kisumu (The two bright spots south of the figure are Arusha and Moshi in Tanzania). By 2000, visual inspection shows little had changed (see Figure 2). While Nairobi seemed to have expanded a bit, the brightness of Kisumu and Eldoret has faded. This coincided with anemic GDP growth and declining per capita GDP. Now look at
Figure 3, which shows Kenya’s night lights in 2010. What happened? First, all the existing bright spots are bigger and brighter (Nairobi, Mombasa, Kisumu and Eldoret); Second, many more lights are now visible from space (especially the area to the north and northwest of Nairobi). This coincided with the brightening (pun intended) of Kenya’s economic fortunes: The economy grew at more than 4 percent per year between 2000 and 2010 and real GDP per capita was 15 percent higher in 2010 than in 2000.
An even starker example of the relation between night lights and countries’ economic fortunes comes from Rwanda and Eastern DRC. In 1992, before the region collapsed into chaos, Kigali and the secondary town of Butare (now Huye) are visible from space, as well as the Congolese border towns of Goma and Bukavu (Figure 4). By 2000, the DRC has collapsed and the bright spots of Goma and Bukavu have died down (Figure 5). Rwanda is still recovering from the devastating genocide, witnessed by the faded lights of Kigali and Huye (its GDP per capita has not yet caught up with 1993 levels).
Between 2000 and 2010, Rwanda’s economy boomed, growing at more than 8 percent per year, and this is clearly seen from outer space. Kigali’s night lights have become brighter and bigger, and many more secondary towns can now be picked up from orbit (Figure 6). Bukavu and Goma are back too, although their lights are now confounded with the increased lighting from the towns at the other side of the border, the Rwandan centers of Cyangugu (next to Bukavu) and Rubavu (next to Goma).
These two simple examples suggest an Africa-specific model for the relation between night lights and GDP growth could well work. Another question relates to poverty. Only a handful of African countries have a systematic and relatively high frequency system of surveys to measure and monitor poverty. As a result, poverty data in many countries are archaic, and a satellite-based estimation of poverty, if reliable, would be most welcome. While modeling the extent of poverty based on the brightness of lights as picked up from space might seem quite a stretch, it is exactly what this paper  by Elvidge et al. does. And it works rather well. Their night lights-based poverty index correlates strongly with poverty as measured from surveys (correlation of 0.85). There is however substantial variation in precision across countries. For Rwanda, they estimate, based on lights, that 86 percent of the population lives on less than US$ 2 a day. The actual survey-based proportion is 84 percent. For Zambia however, they estimate 67 percent, more than 25 percentage points lower than the survey estimate (94 percent). Which is "correct"? You decide.
The Elvidge et al model is based on 233 countries. The question is whether an Africa-specific model of night lights and poverty makes sense, especially given the large variation in poverty and living standards across unlit areas. Stay tuned…we are about to find out.