The figure displays regression coefficient estimates βt of the interaction between the gold exposure and the year on (a) the total urban population of the cell, and (b) the probability of having a town or city in the cell (see Equation A.4). Error bars represent 90% confidence intervals. Urban population is taken in logarithm transformation. The estimation includes country×year fixed effects and controls for temperature,
precipitation, SPEI, the initial log population, the ever-presence of industrial mines at the ring distance interacting with the corresponding mineral price, and crop suitability interacting with the corresponding crop price. Standard errors are estimated allowing for infinite serial correlation and spatial correlation within a 500km radius (Conley 2008; Hsiang 2016).
This article finds that artisanal #goldmining in Sub-Saharan Africa over 45 years has spurred new small towns in remote areas, causing fragmented #urban growth without significant industrial development.
Read: spkl.io/63324AFoNr
Subscribe: spkl.io/63327AFoNV