24 Million Jobs at Risk: Employment Effects of a Global Energy Price Shock

A 12 percent increase in global refined petroleum prices eliminates $2.69 trillion in gross output across 163 countries and 120 sectors. But output is only one side of the ledger. The labor market consequences of the same shock are both larger in human terms and structurally different in their distribution — and the relationship between the two reveals something important about the speed at which economies adjust.

These losses include both the first-round effects of higher energy costs on firms and households and the subsequent propagation through supply chains, income contraction, and investment decline — the same cascading dynamics that drive the output losses, now expressed in human terms.

The Adjustment Window: Why Employment Falls Slower Than Output

This gap is a central feature of the analysis and reflects how economies actually adjust in the short run. Standard equilibrium frameworks assume that markets clear instantaneously — output contracts, workers are released, and the economy arrives at a new steady state. In practice, that is not what happens. Firms facing higher energy costs do not immediately shed workers at the same rate they cut output. They absorb some cost pressure through margin compression, reduce hours and new hiring before cutting headcount, and — where possible — reallocate labor within and across sectors rather than expelling it from the labor market entirely.

This creates a disequilibrium adjustment window: output contracts faster than employment, and the gap between the two defines the space in which labor market policy can intervene before job losses fully materialize. Credit support for firms absorbing margin pressure, retraining programs for workers in contracting sectors, demand-side stimulus to sustain employment in consumption-dependent industries — all of these operate within this window. Once it closes — once firms exhaust their capacity to absorb costs and begin shedding workers at the rate implied by the output decline — the employment losses become structural and much harder to reverse.

How wide that window is depends on the depth and persistence of the shock, and critically on uncertainty. When firms and households cannot plan — when they do not know whether the price shock is temporary or permanent, whether policy support is forthcoming, whether supply chains will stabilize — the buffer between output contraction and employment loss narrows. Firms shed workers faster, investment collapses more rapidly, and the stabilizing effects of labor market adjustment diminish. The figures presented here should accordingly be read as a scenario under stable expectations; a genuine energy crisis, with its attendant uncertainty, would compress this window considerably.

Regional Employment Breakdown

East Asia and the Pacific loses the most jobs in absolute terms — 7.0 million workers — but records the smallest percentage decline among the six regions at −0.55 percent. The moderating force is structural: a large and relatively insulated labor force in the region’s dominant economy contains the percentage impact, while EAP’s highly integrated manufacturing supply chains allow for partial labor reallocation within the industrial base rather than outright displacement. The density of East Asian supply chain linkages means that the shock propagates rapidly through tightly coupled production networks, reaching sectors with minimal direct petroleum exposure — the same feature that makes these supply chains efficient also makes them efficient transmitters of shocks.

South Asia loses 5.5 million jobs (−0.83 percent), the second largest absolute figure. Here the income channel is the dominant mechanism: the shock compresses household purchasing power in a region with large numbers of workers in consumption-sensitive sectors — retail, food services, small-scale manufacturing — who face employment pressure not from direct energy cost exposure but from the demand contraction that propagates as incomes fall.

Sub-Saharan Africa (−3.8 million, −0.87 percent) and Latin America and the Caribbean (−3.6 million, −1.17 percent) show broadly similar percentage declines, but through different structural pathways. SSA’s employment losses are driven significantly by income and investment contraction operating through the region’s relatively informal, consumption-driven labor markets. LAC’s higher percentage figure is consistent with its output result: energy-intensive sectors in commodity-exporting economies shed workers whose jobs are tied directly to the production activities most exposed to the shock.

Middle East and North Africa records the steepest employment percentage decline in the sample at −1.22 percent, despite having the second-smallest absolute number (−1.9 million). The anomaly lies in the structure of MEN economies that do not export petroleum: they depend heavily on oil-revenue-driven government expenditure and investment as employment-sustaining forces. When the energy shock simultaneously raises production costs and compresses these fiscal and investment flows — even while benefiting the exporting nations within the region — the labor market impact is amplified through both the cost and the income channels at once, producing a disproportionate percentage loss relative to the region’s economic size.

Europe and Central Asia records the most contained employment decline (−2.3 million, −0.52 percent). The presence of energy-exporting nations within the region generates employment gains in extraction and processing that partially offset losses in importing economies, and ECA’s more diversified industrial structure limits the extent of shock propagation through the broader labor market.

The Sectoral Employment Picture

Color intensity reflects percentage employment change: darkest = >2.2% loss, medium = 1.5–2.2%, lighter = 1.0–1.5%, lightest = <1.0%. Hover for both absolute and percentage figures.

Wholesale and retail trade loses the most jobs in absolute terms globally — 3.3 million — despite recording only a 0.70 percent decline relative to its enormous employment base. The sector’s losses are driven almost entirely by income contraction: as households face higher energy costs and lower real purchasing power, discretionary retail spending falls, and the sector that mediates the bulk of consumer demand follows.

Energy-intensive sectors show the sharpest percentage losses. Coal mining falls 2.6 percent, electricity generation 2.5 percent, and telecommunications 2.2 percent — the latter reflecting the sector’s dependence on energy-intensive infrastructure and data centers. Petroleum extraction itself loses 1.2 million jobs globally, a result that might seem counterintuitive given that producers benefit from the price increase. The explanation lies in the distinction between value gains at the extraction stage and cost pressures in refining, distribution, and processing: even in petroleum-producing regions, the downstream refining and distribution workforce contracts as the price increase compresses refinery margins and reduces petroleum product demand across the economy.

Civil construction (−1.75 million) and real estate (−1.62 million) are both highly sensitive to the investment channel: as the shock compresses firm cash flows and reduces household wealth, construction and real estate activity contracts — not because building materials become more expensive (though they do), but primarily because the financing and demand for new construction evaporates.

The sectoral pattern mirrors a consistent finding from the output analysis: the sectors hit hardest by percentage are those directly in the energy transmission chain, but the sectors that lose the most workers in absolute terms are those that serve final demand — the infrastructure through which a cost shock becomes a demand shock and ultimately a jobs shock.

How MINDSET Works

The results in this post are generated using MINDSET (Model of Innovation in Dynamic Low-Carbon Structural Economic and Employment Transformations), a demand-driven multi-region input-output simulation model covering 163 countries and 120 sectors, built on the GLORIA global supply chain database.

MINDSET is designed for the analysis of short-run structural shocks — energy price changes, supply disruptions, policy interventions — in settings where the assumption of instantaneous market equilibrium would produce misleading results. The model’s core design allows for disequilibrium dynamics: rather than solving to a new equilibrium, MINDSET simulates the structural adjustment path under which output, employment, trade, and investment respond at different speeds and through different mechanisms. Firms absorb costs before cutting output; workers are reallocated before being displaced; investment responds with a lag. These features allow the model to trace the divergence between output and employment responses that standard equilibrium frameworks collapse into a single adjustment.

In the scenario analyzed here, the shock takes the form of a 12 percent increase in refined petroleum prices (GLORIA sector 63) calibrated by country using each economy’s actual petroleum import intensity from the GLORIA trade database. The effective pass-through to each country reflects how much of its refined petroleum consumption is imported versus domestically produced — a data-grounded approach that allows the model to distinguish between net importers, partial producers, and exporters without imposing a uniform shock. Gas distribution (sector 94) receives a correlated pass-through reflecting the co-movement of petroleum and gas prices. No supply constraint is imposed: this is a pure price shock.

Sector-level output changes are derived from Leontief input-output linkages across the full 163-country, 120-sector matrix. Trade adjusts through calibrated elasticities. Government revenue and expenditure effects are modeled endogenously. Employment effects are derived from sector-level output changes using employment-output coefficients estimated from national labor force surveys, which capture the labor intensity of production in each country-sector cell. Critically, these coefficients allow employment to respond at a different rate than output — the mechanism that generates the disequilibrium adjustment window discussed above.

For the full methodological framework, simulation architecture, and validation results, see the MINDSET working paper on SSRN.