Energy Prices Explain the Cliff. They Do Not Explain the Plateau.

There are two lazy versions of Germany's industrial crisis.

The first comes from the industry associations. In this version, high energy prices, emissions trading, and regulation have made competitive production in Germany almost impossible. The second is the irritated reply: every European producer faced the same shocks, so BASF and the VCI should stop blaming Berlin and Brussels for their own mistakes.

Both stories are too convenient.

The 2022 energy shock was real, enormous, and causal. It hit gas-based basic chemistry harder than the average factory. BASF's additional energy bill alone was EUR 3.2 billion, 84 percent of it in Europe. Treating this as public-relations theater would be analytically unserious.

But the shock does not explain everything that followed. German chemical production was already falling before COVID-19. It did not recover when gas and electricity prices came down from their peaks. Its relative decline is not uniquely German: chemicals underperformed manufacturing more severely in the euro area as a whole. And the company accounts show outcomes that no single energy-price story can reconcile.

My conclusion is uncomfortable for both camps:

The shock was external. Exposure was inherited. Persistence is the test.

Russia supplied the impulse. The European energy system transmitted it. Yet product mix, feedstock dependence, plant configuration, pricing power, China exposure, and capital allocation determined how much damage the impulse caused, and whether a company recovered.

Related live evidence. This essay asks why the level break persisted. The July 2026 Chemical Industry Radar tracks the current operating environment, company signals, capital allocation, and location decisions month by month. The two views are complements: one identifies the structural mechanism; the other monitors whether it is changing.

The verdict in seven numbers

All production figures below are calendar-year means of Eurostat's seasonally and calendar-adjusted industrial production index. The calculations use the official data snapshot available on 13 August 2026. Chemicals are NACE C20, explicitly excluding pharmaceuticals.

The long-run comparison tells us where not to start. It would be wrong to locate the beginning of the chemical industry's weakness in February 2022. Between 2010 and 2019, chemical output declined 6.1 percent while manufacturing rose 13.6 percent and real economy-wide value added rose strongly. Chemicals briefly rebounded in 2021. The rebound concealed rather than resolved the earlier divergence.

That pre-trend matters. Any estimate that labels the entire post-2021 loss an "energy effect" will be biased upward unless it accounts for the trajectory already visible in basic chemicals, paper, metals, and parts of Germany's capital-goods complex.

Three shocks, not one

COVID-19 was a stress test, not the decisive chemical break

COVID-19 initially hit supply chains, mobility, and capital goods more severely than chemicals. From 2019 to 2020, German manufacturing production fell 9.8 percent. Chemical production fell only 1.4 percent. Motor vehicles and basic metals were hit much harder. Chemical plants benefited from essential demand, long production runs, and the fact that the crisis was not an energy-input shock.

The firm accounts confirm the heterogeneity. In 2020, Covestro reduced core volumes by 5.6 percent but limited the EBITDA decline to 8.2 percent and increased free operating cash flow. Heidelberg Materials' sales fell 6.6 percent, yet its operating result increased 8.1 percent. Aurubis improved operating earnings. The common COVID-19 shock produced radically different cash-flow outcomes because portfolio, cost flexibility, inventories, and pass-through differed.

This is the first warning against treating "energy-intensive industry" as a single technology.

The energy cliff began before the invasion

Industrial gas saving in Germany started in September 2021, not February 2022. Ruhnau and coauthors estimate that adjusted German gas use across sectors was roughly 23 percent below baseline in the second half of 2022. Their timing matters: the reopening-driven gas run-up, storage concerns, and reduced Russian supply preceded the invasion. A single post-February dummy misses part of the treatment. Ruhnau et al., Nature Energy

After the invasion, the price impulse became extreme. In Destatis' current 2021-based producer-price series, the relevant industrial gas index peaked at 400.6 in September 2022 and special-contract electricity at 248.9 in August. By June 2026, electricity had returned to 103.1. Gas remained at 149.1: far below its peak, but still about half above the 2021 level. The gas series covers deliveries of more than 500,000 MWh a year; the electricity series covers special-contract customers of more than 200,000 kWh a year. Neither is a spot-price chart, so both capture pass-through with lags. Destatis methodology and chart data

The production response is exactly what economic incidence predicts. In February 2022, the chemical index stood at 95.9 and total industry at 100.8. By December, chemicals had fallen to 74.7 while industry remained at 98.4. Hutter and Weber's sector-level design attributes a 4.1 percent monthly production loss and a 2.6 percent real-turnover loss to the pre-existing energy-intensity channel after the invasion. Short-time-work notifications rose and vacancies fell, but layoffs did not. That is the closest German causal estimate, and it rejects the claim that energy was merely an excuse. Hutter and Weber, Energy Policy

The macro evidence points in the same direction. A Bundesbank simulation found the largest gross-margin exposure in chemicals, glass and ceramics, basic metals, and paper. An ECB decomposition attributes about a 2 percent euro-area industrial-output loss from September 2021 to September 2022 to adverse energy-supply shocks, alongside a rise in cheaper non-energy imports. Ammonia is the telling example: European chemical companies imported more instead of producing it with expensive gas at home. Bundesbank, ECB

The plateau is the harder question

The chemical index did not meaningfully recover after the price peak. It was 76.2 in December 2023, 78.4 in December 2024, 75.1 in December 2025, and 78.5 in June 2026. Total industry also weakened, but less. From February 2022 to March 2026, Destatis reports a 15.2 percent fall in the five energy-intensive sectors, versus 9.5 percent in total industry; chemicals fell 18.1 percent. Destatis, 15 May 2026

That persistence is not proof of management failure. Low utilization can remain optimal when energy prices fall but global product prices fall faster. Capacity closures, long contracts, inventories, and weak customer demand create lags. Yet persistence changes the burden of explanation. The external-only story must explain why output stayed near its trough as electricity normalized and gas fell by more than half from its peak.

The crucial benchmark is Europe, and then the United States and China

The statement "all European firms were affected" is correct but incomplete. A common shock is not a common treatment. German basic chemistry used more gas and used it not only as fuel but also as feedstock. The Ludwigshafen Verbund carries large fixed loads and produces upstream intermediates that are highly tradable. A cement plant faces the same gas market but sells a heavy, local product into a more concentrated market. The incidence cannot be equal.

Still, the country comparison is devastating to the claim that the chemical decline is primarily a uniquely German event.

From 2019 to 2025, German chemical production fell 17.9 percent. Euro-area chemical production fell 17.7 percent and EU-27 chemical production 16.2 percent. In absolute terms, Germany was not an outlier. What was unusual was the domestic denominator: German manufacturing fell 11.5 percent, while euro-area manufacturing rose 0.6 percent and EU manufacturing 2.7 percent.

So German chemicals underperformed German manufacturing by 6.5 percentage points. Across the euro area, chemicals underperformed manufacturing by 18.3 points. This is a descriptive triple difference, not a causal estimate. Belgium and the Netherlands also have index and composition issues. But the sign is clear: the Germany-specific weakness since 2019 is broad manufacturing weakness; the chemistry-specific weakness is largely European.

The transatlantic comparison makes the industrial geography even clearer. US chemical production rose 5.7 percent from 2019 to 2025, while US manufacturing fell 2.1 percent. China reported 7.8 percent growth in value added for raw chemical materials and chemical products in 2025 alone. These are not perfectly harmonized series, and China's figure is a year-on-year growth rate rather than a 2019-based level comparison. They nevertheless point in the same direction: Europe experienced a chemical shock that advantaged-feedstock and capacity-expanding regions did not. Federal Reserve G.17, FRED chemical production, China National Bureau of Statistics

This is where the VCI argument must become more precise. "Germany is expensive" is not enough. The relevant counterfactual is product by product: German ammonia versus imported ammonia; European steam crackers versus US ethane crackers; commodity polymers versus Chinese capacity; specialties versus local customer willingness to pay.

Gas and electricity are now different stories

At the 2022 peak, German large-user prices were not uniquely high within Europe. For the 70-150 GWh electricity band excluding taxes and levies, Germany paid 17.28 euro cents per kWh in the second half of 2022, below the EU average of 18.92 cents and far below Italy's 33.34 cents. For the 1-4 million GJ gas band, Germany paid 8.65 cents, also below the EU average of 9.76 cents.

By the second half of 2025, the picture had split.

• German large-user electricity cost 12.83 euro cents per kWh, 17 percent above the EU average and 88 percent above France.
• German gas cost 3.53 euro cents per kWh, 6 percent below the EU average and 10 percent below France, although still more than twice its German 2019 level.

The contemporary competitiveness problem is therefore not one generic "energy price." Gas still carries a Europe-wide post-Russian premium. Electricity now contains a clearer German system wedge, even before taxes and levies: generation mix, networks, contract structure, and scarcity all matter. Headline household tariffs tell us little about a chemical plant with self-generation, exemptions, hedges, and bilateral contracts.

A sensible industrial paper must construct realized plant-level exposure. Wholesale TTF or EEX prices alone are not treatment.

Output collapsed more than jobs or profits

The production chart looks apocalyptic. The income accounts are less dramatic, and more interesting.

German chemical nominal output rose from EUR 136.0 billion in 2019 to EUR 189.8 billion in 2022 even as physical production fell. This was price pass-through and mix, not real growth. Net operating surplus fell from EUR 9.9 billion to EUR 8.1 billion. Its share of output dropped from 7.25 to 4.28 percent. In 2023, nominal net operating surplus recovered to EUR 10.1 billion, slightly above 2019 in euros, while the margin remained lower at 6.10 percent. Eurostat national accounts

The Bundesbank reaches the same conclusion from corporate statements. Material and energy costs surged in 2022, yet enterprises passed much of the increase through. The aggregate pre-tax margin fell from 5.1 to 4.5 percent but remained above its long-run average. Chemicals and pharmaceuticals were among the few groups with a visibly larger earnings decline, partly from an exceptional 2021 base. There was no general balance-sheet collapse. Bundesbank corporate accounts for 2022

Employment adjusted even less. National accounts record 352,000 people in German chemicals in 2019 and 372,000 in 2023. Hours rose 2.7 percent while real chemical GVA fell 9.2 percent. Implied real GVA per worker fell about 14 percent; the cruder production-per-worker measure fell more than 20 percent. In the five official energy-intensive sectors, output fell 15.2 percent from February 2022 to March 2026, but employment only 6.3 percent.

This can be read in three ways. Firms hoarded scarce skills in expectation of recovery. Fixed energy loads and plant minimums made low utilization technically inefficient. Or structural productivity and portfolio problems were already present. The data cannot choose among these mechanisms by themselves. They do reject a simple equation of lower domestic tonnage with an equal loss of jobs, value added, or solvency.

The other energy-intensive industries are not a control group with one answer

Destatis defines five energy-intensive branches: paper, refining, chemicals, non-metallic minerals, and basic metals. Together they accounted for 16.9 percent of industrial gross value added but 77 percent of industrial energy use in 2021. Chemicals alone used 36.7 percent of industrial natural gas. Destatis methodology

Their paths differ because energy is only one input into different markets.

Paper demonstrates why chronology matters. Its long-run demand problem was created by digitization, not by the German electricity market. Energy costs determine which marginal machine closes and where; they do not create the decline in newspaper circulation.

Non-metallic minerals demonstrate why tradability matters. Their production fall was marginally worse than chemicals, but Heidelberg Materials recovered energy inflation through price-over-cost management in local cement markets. Basic metals demonstrate why commodity spreads matter: Salzgitter earned one of its best results in 2022 because steel prices rose faster than costs. Refining demonstrates that an energy-intensive producer can benefit from an energy shock when its output price moves with the shock.

And the broader manufacturing benchmark is itself weak. Machinery production fell 15.6 percent from 2019 to 2025 and motor vehicles 16.4 percent. Neither is in Destatis' five-sector energy-intensive composite. The Bundesbank finds that German export-market-share losses have been broad-based since 2017 and increasingly associated with products and markets in which China gains share. Energy matters, but so do product demand, unit labor costs, supply constraints, and the auto transition. Bundesbank, July 2025

The company accounts: one shock, many diagnoses

Consolidated annual reports are not a causal dataset. BASF's group accounts mix Ludwigshafen with Texas and China; Heidelberg Materials sells a less tradable product; company EBITDA definitions differ. But they are excellent mechanism tests. If one common German energy factor dominates, companies with heavy exposure should display broadly similar margin paths. They do not.

BASF: the strongest case for energy, and then a different problem

BASF quantified EUR 3.2 billion of additional energy cost in 2022 versus 2021, EUR 2.2 billion from gas. Ludwigshafen's gas bill rose EUR 1.4 billion despite consumption falling by roughly 35 percent. This is a material earnings shock by any standard. BASF FY2022 presentation

But the later evidence is no longer mainly an energy story. BASF's 2025 segment results diverged: agriculture and surface technologies improved while chemicals, materials, and industrial solutions weakened. Management cited a missing amines recovery, lower volumes, and poor contribution margins. The company closed upstream Ludwigshafen assets while committing capital to Zhanjiang and Geismar. It also maintained a EUR 3.40 dividend through 2023, distributing roughly EUR 3 billion annually before cutting the payout for 2024. Energy explains the initial site shock; portfolio, geography, and capital allocation help explain adaptation.

Covestro and Wacker: the pre-trend and the falsification

Covestro's EBITDA already halved from EUR 3.2 billion in 2018 to EUR 1.6 billion in 2019 because of competitive price pressure. By 2025, its sales decline was driven by price, volume, and currency across Europe, North America, and Asia-Pacific. That is not consistent with a Germany-only energy diagnosis. Covestro 2019, Covestro 2025

Wacker supplies the cleanest within-company falsification. Its energy cost fell from 8.6 percent of sales in 2024 to 5.5 percent in 2025. EBITDA nevertheless collapsed 43 percent. The report points to lower prices and volumes, poor utilization, and enormous Chinese solar-polysilicon capacity. It also records major impairments on Siltronic, German deferred tax assets, and acquired Biosolutions goodwill. High German energy costs add burden. They cannot be the dominant marginal explanation for a year in which their disclosed share of sales fell so sharply. Wacker forecast comparison

The non-chemical falsifications

Heidelberg Materials' cement volumes and operating result both fell in 2022, proving that the shock reached it. Yet pricing recovered most of the cost. Its margin then rose to 20.1 percent in 2023, 21.3 percent in 2024, and 21.8 percent in 2025; ROIC reached 10.4 percent. Cement is heavy and local, so this is not a claim of equal exposure. It is proof that tradability and market structure determine incidence. Heidelberg Materials 2025

Aurubis earned record operating EBT of EUR 532 million in FY2021/22 despite high gas and electricity prices. Its next-year result was hit by a EUR 169 million metals shortfall caused by criminal activity: an unambiguous internal-control failure, not an ETS effect. Aurubis FY2021/22, Aurubis FY2022/23

Salzgitter earned EBITDA of EUR 1.62 billion in 2022 because steel prices dominated energy inflation. Later earnings deteriorated as steel prices, imports, and demand weakened. UPM's German and Austrian paper closures principally reflect secular graphic-paper demand and capacity balancing; energy decides the marginal site, not the existence of the demand decline.

Regulation and emissions trading: the evidence is less dramatic than the rhetoric

The EU ETS is an EU-wide policy. It cannot by itself explain a uniquely German chemical residual. A German effect must come from the power mix, indirect carbon costs, compensation, national implementation, or the interaction with unusually exposed assets.

Industrial installations at carbon-leakage risk continue to receive free allowances under harmonized EU benchmarks. The 2021-2030 leakage list covers 63 sectors representing about 94 percent of industrial ETS emissions. Free allocation does not eliminate the marginal opportunity cost of an allowance, and it does not compensate every plant fully. But any serious cost calculation must use verified emissions minus free allocation, then add indirect electricity-carbon cost net of compensation. Quoting the EUA price against all emissions overstates cash cost. European Commission on industrial free allocation, carbon-leakage list

The causal literature does not support blanket claims of deindustrialization from electricity taxes or the ETS:

• Gerster and Lamp use the German EEG-levy exemption threshold. Exemption increased electricity use but produced no detectable sales, export-share, or employment benefit. Economic Journal
• Flues and Lutz exploit German electricity-tax discontinuities and find no systematic effects on turnover, exports, value added, investment, or employment. ZEW Discussion Paper
• Von Graevenitz and Rottner use exogenous local network-price variation. Firms reduced purchased electricity and increased on-site generation, without significant sales or employment losses. ZEW paper
• Colmer and coauthors find that the EU ETS reduced French regulated-firm emissions 14-16 percent in its early phases without detectable output or employment contraction. Review of Economic Studies

These are not proofs that a 2022-sized gas shock is harmless. The German tax studies identify local policy margins, while much of the ETS evidence covers earlier phases with lower allowance prices and generous free allocation. Tail firms and current Phase IV exposure can differ. The correct conclusion is narrower: there is no demonstrated average competitiveness payoff from blanket electricity-price exemptions, and the best causal evidence does not support broad ETS-deindustrialization claims.

What can reasonably be attributed to what?

"Purely internal" is an appealing phrase and a poor estimand. Even after controlling for energy, a chemical residual can contain unobserved customer demand, product mix, or Chinese capacity. A defensible paper should call it a conditional residual, then test mechanisms within chemicals and plants.

Policy: insure the shock, do not subsidize the plateau

The distinction between impulse and persistence has direct policy consequences.

Temporary liquidity support, gas-security measures, and well-designed short-time work can be justified for an unforeseen supply shock. A permanent subsidized industrial power price is a different instrument. It transfers risk to taxpayers and may preserve assets whose difficulty is global overcapacity, weak demand, or poor capital allocation. The German exemption studies provide no evidence that broad electricity subsidies buy the promised sales, exports, or jobs.

The better policy package is conditional and mechanism-specific:

1. Fix the electricity-system wedge. Expand grids and firm low-carbon supply, accelerate interconnection and procurement, and make network charges reflect system value. The 2025 price data show a real problem, especially against France.
2. Support transformation, not tonnage. Use contracts or tax support tied to verified investment, emissions reduction, operational milestones, and clawbacks, not historical energy consumption.
3. Price transition risk transparently. Report net ETS cost after free allocation and compensation. Do not confuse gross EUA exposure with cash cost.
4. Make exit and reallocation possible. Retraining, site redevelopment, and fast permitting for new investment are more credible than indefinite operating subsidies for structurally uncompetitive lines.
5. Demand better disclosure. Companies asking for public support should publish German-site energy and feedstock costs, realized hedged prices, capacity utilization, domestic capex, free allocation, compensation, and product-level import exposure.
6. Apply a capital-allocation test. Public claims that a site lacks investment deserve scrutiny when the same group maintains large payouts or chooses growth capex elsewhere. That choice may be rational; it should not be socialized without conditions.

The bottom line

German chemicals have a real competitiveness problem. The 2022 gas shock caused a large and identifiable part of it. The current electricity disadvantage deserves policy attention. Europe-wide carbon and energy policy is not free.

But the evidence does not support the stronger industry narrative.

The chemical decline began before Ukraine. Similar or worse production losses occurred in paper and non-metallic minerals. European chemicals declined almost exactly as much as German chemicals, while US chemicals grew and Chinese capacity expanded. Aggregate profitability and employment proved more resilient than tonnage. And individual companies facing the same broad shock produced outcomes ranging from record returns to deep losses driven by overcapacity, acquisitions, site configuration, or internal controls.

Energy prices explain the cliff. They do not explain the plateau.

That plateau is where the serious economic work begins.

Data and interpretation note

The production comparison uses WZ 2008/NACE Rev. 2 consistently: chemicals C20; paper C17; refining C19; non-metallic minerals C23; basic metals C24; manufacturing C. Indices are real; sales, energy purchases, and accounting profits are nominal unless stated otherwise. Eurostat's structural-business-statistics methodology breaks in 2021, so the analysis avoids treating a mechanically stitched 2020-2021 level as an economic event. Simplified energy balances exclude non-energy/feedstock use and therefore understate chemical exposure. Annual-report figures are consolidated and selected; they validate mechanisms but do not identify causal German-site effects. All 2025 and 2026 official statistics remain subject to revision.

Selected references

Official statistics and institutions

• Eurostat, Industrial production index, sts_inpr_m, data snapshot 13 August 2026.
• Eurostat, National accounts by 64 industries, nama_10_a64.
• Eurostat, Employment by industry, nama_10_a64_e.
• Eurostat, Non-household electricity prices, nrg_pc_205.
• Eurostat, Non-household gas prices, nrg_pc_203.
• Destatis, Production in energy-intensive branches.
• Deutsche Bundesbank, Corporate profitability and financing in 2022.
• Deutsche Bundesbank, German export competitiveness, July 2025.
• European Central Bank, Energy shocks, competitiveness, and euro-area production.
• Federal Reserve, Industrial Production and Capacity Utilization, G.17.
• National Bureau of Statistics of China, 2025 industrial production release.
• European Commission, Free allocation for industrial installations.

Causal and empirical research

• Ruhnau, O. et al., Natural gas savings in Germany during the 2022 energy crisis, Nature Energy.
• Hutter, C. and E. Weber, The effect of the energy crisis on German industry, Energy Policy.
• Gerster, A. and S. Lamp, Energy-tax exemptions and firm outcomes, Economic Journal.
• Flues, F. and B. Lutz, The effect of electricity taxation on German manufacturing, ZEW.
• Von Graevenitz, K. and E. Rottner, Network prices and industrial electricity use, ZEW.
• Colmer, J. et al., The EU ETS and regulated-firm outcomes, Review of Economic Studies.

Company evidence

• BASF, FY2022 analyst presentation and 2025 ten-year summary.
• Covestro, 2019 business development and 2025 results.
• Evonik, 2025 reporting.
• Lanxess, 2025 financial summary.
• Wacker, 2025 multiyear overview and forecast comparison.
• Heidelberg Materials, 2025 results.
• Aurubis, FY2021/22 results and FY2022/23 extraordinary effects.