Albert Duodu

Manufacturing in rich countries is getting cleaner—on paper. Recent firm-level studies find little evidence that offshoring to pollution havens drives this clean-up. Linking plant-level fuel use for Swedish manufacturers (2005–2014) to imports weighted by their embodied carbon, I instrument firms’ carbon imports with a shift-share of foreign export-supply shocks. A 10 percent rise in carbon imports lowers production-based CO₂ intensity by 6.7 percent. The decline comes from scale and product mix, not cleaner technique. The pollution-haven margin, sourcing the same products from dirtier origins, is weak on average, as earlier studies find, but significant among firms that bought many distinct products before the shock, the only ones with room to reallocate. Yet the carbon does not disappear. Domestic emissions, which the national inventory counts, are flat in levels, while firms’ own transport emissions and the international freight carrying their imports rise. The imported goods embody 64 percent more carbon than producing them in Sweden would. Only 2 percent of imported carbon flows through foreign affiliates, so leakage policy aimed at multinationals misses almost all of it, and the EU’s carbon border adjustment, which excludes freight, misses the transport leg. Where emissions are counted can matter as much as whether they fall.

We examine how local import competition – measured across different spatial dimensions within Sweden – affects the CO₂ emission intensity of manufacturing firms. Using detailed geographic data on Swedish manufacturers, we show that increased local import competition leads to lower firm-level emissions, with the effect diminishing as the distance between producers and importers grows. We identify two mechanisms: (i) a pro-competitive efficiency effect, with gains in TFP, higher value added, lower marginal costs, and higher markups; and (ii) a product-mix effect, with reallocation away from emission-intensive goods. We also document evidence of carbon offshoring and investment in pollution abatement as additional firm responses.

Carbon pricing is a cornerstone of climate policy, yet its firm-level incidence remains poorly understood. Combining matched employer–employee and firm-level data from Sweden (2006–2014) with a shift-share IV and a dynamic difference-in-differences design that exploits variation in exposure to energy tax reforms, I show that higher energy prices generate substantial reductions in energy use and CO₂ emissions. These environmental gains come with economically meaningful adjustments: declines in productivity and employment, and incomplete cost pass-through. The incidence is highly heterogeneous – low-productivity firms and high-skilled workers bear disproportionate costs, while more productive firms absorb shocks more easily.

How do citizens respond when the state repeatedly fails at a visible task? Using geocoded data on 201,286 respondents across 42 African countries and Ghana’s 2012–2016 dumsor electricity crisis, we show that sustained local outages lower presidential approval while simultaneously raising contentious participation—contacting officials, organising, and protesting. This divergence rules out generalised disaffection and points instead to a targeted political response: citizens punish the incumbent and act, rather than withdraw. Exploiting the dumsor crisis in a difference-in-differences design, we find that exposure sharply erodes beliefs about state competence, with a clear dose-response in outage intensity. Chronic failure to deliver basic services thus reshapes how citizens evaluate and engage the state, with direct consequences for political accountability.

Artisanal gold mining ("galamsey") employs roughly a million Ghanaians and produces a third of national gold output, but its effects on the formal economy are undocumented. We combine a firm census, geocoded household surveys, and night-lights with a shift-share measure of exposure — predetermined Birimian geology interacted with the world gold price. Fully gold-suitable districts host 19 percent fewer formal firms than non-suitable ones, and the survivors are larger; the contraction falls on agriculture-linked sectors, which thin four times faster than mining-adjacent ones, and on wetter, agriculturally productive districts. Household data identify the mechanism: men in exposed districts became markedly more likely to work, and to work in physical and agricultural occupations, after the 2008 boom, with no differential pre-trend over the preceding fifteen years. When the 2017 Operation Vanguard crackdown disrupted mining, night-lights in exposed districts fell about 9 percent. Informal booms reshape local economies by pulling labour into untaxed, unmeasured extraction, not through the fiscal and price channels of large-scale mining.

Where do the local effects of a resource boom land? Using a global panel of half-degree grid cells from 1989 to 2014, I compare petroleum and non-petroleum cells within the same country and year, validating identification with high-frequency oil supply news shocks around OPEC announcements. Petroleum cells brighten by roughly eight percent over the 1998–2008 boom, and organised conflict does not systematically escalate; instead, violence shifts modestly toward lower-intensity, non-state and civilian-targeted events. The development gains depend on institutions: cells in high-capacity states with politically included populations show the full response, cells where either margin is unfavourable show sharp attenuation, and cells where both bind show none. The familiar cross-country link between oil and conflict reappears only when within-country variation is removed – the local response and the cross-country correlation are distinct objects.