Output, orders, power prices, payrolls and insolvencies all point at the same split, with one exception
By: Christian Brooks – SeaPRwire – The ifo business climate index reached 89.9 points in September 2026, its third consecutive monthly increase and 4.2 points above the June reading of 85.7. The federal government’s autumn projection, presented on 8 October, puts GDP growth at 1.3% for the year, an upgrade of 0.8 percentage points against the spring forecast. Total industrial output in August stood 0.6% above its level twelve months earlier. None of these numbers describes a boom. All of them describe the end of a decline.
A recovery that stops at the gate of the first process step
My reading of the 2026 data is that Germany is losing the bottom of its supply chain while the top has stopped shrinking. Chemicals and steel are being resized to a smaller permanent footprint. Machinery, electrical equipment and parts of vehicle assembly have flattened out and, in places, recovered. The two layers remain physically joined, by pipelines, process heat and suppliers who make one input for one customer, so the recovery at the top carries a shelf life that the aggregate statistics do not show.
De-integration is a word I use deliberately. It describes a chain losing its intermediate stages while keeping its final ones: the machine tool still gets built in Baden-Württemberg, but the specialty polymer in its housing, the chlorine in its cables and coatings and the steel in its frame increasingly arrive from elsewhere, and the plants that made them locally have closed. The output statistics capture the machine. They do not capture what the machine no longer pulls through the domestic chain.
That is a specific claim, and it can be broken in at least four ways. I list the tests before the evidence, because a judgement that no observation could overturn is a preference dressed as analysis.
The claim I will defend, and what would falsify it
The claim, in one sentence: the German industrial recession of 2022 to 2025 has ended at the top of the value chain and has become permanent attrition at the bottom. The four observations below would each do damage to it.
- If the chemicals production index climbs above 85 (2021 = 100) and holds there for four quarters without another announced plant closure, the attrition reading collapses and the last four years become a very long cycle.
- If machinery and electrical output rises together with domestic order intake and capacity utilisation, the top of the chain is rebuilding capacity rather than working through a backlog, and the shelf-life argument fails.
- If manufacturing headcount holds flat while chemicals stays near 75, firms are re-sourcing the chain abroad rather than shrinking it, and “de-integration” is too strong a word for what is happening.
- If the flat-to-positive output readings since mid-2025 vanish in the next benchmark revision, half of the claim disappears with them.
Two of these tests are already informative, and I return to them at the end. The rest of this piece is the evidence that makes them worth running.
Four indices, one factory floor
Eurostat publishes a monthly German production index, seasonally and calendar adjusted, with average 2021 output set to 100. National statistical offices supply the underlying survey, so the series allows branches to be compared on one scale without the distortions of nominal turnover. Six rows are enough to see the split.
| Branch (NACE Rev. 2) | Aug 2025 | Jun 2026 | Aug 2026 | 2018 average | Aug 2026 vs 2018 |
| Total industry (B–D) | 90.2 | 91.9 | 90.7 | 108.7 | −16.6% |
| Manufacturing (C) | 91.0 | 92.4 | 90.8 | 108.4 | −16.2% |
| Chemicals (C20) | 77.4 | 78.3 | 74.9 | 98.7 | −24.1% |
| Basic metals (C24) | 90.3 | 88.1 | 88.5 | 111.6 | −20.7% |
| Machinery (C28) | 90.2 | 83.5 | 89.7 | 110.3 | −18.7% |
| Motor vehicles (C29) | 98.4 | 105.7 | 91.5 | 133.9 | −31.7% |
The first two rows support the recovery story. Total industry and manufacturing sit within half a point of where they were a year ago, and both have moved sideways since June. The last four rows explain why that average is a poor description of the parts. Chemicals are 24.1% below their 2018 average, basic metals 20.7% below, machinery 18.7% below, vehicles 31.7% below. Gaps of that size are not closed by a good quarter. They are the accumulated result of six years in which the German cost position moved faster than the capital stock could adjust.
August deserves a warning. It is the month of collective holidays in the metal and chemical industries, and month-on-month changes in a single August carry more noise than signal. The year-on-year column and the comparison with 2018 are more robust, which is why the table is built around them. A reader who wants to track the turn should watch three-month averages, not single months.
Why the sources disagree, and why that is the finding
Four published measures of the German industrial economy pointed in different directions this autumn, and the disagreement tells you something about the object being measured. The first involves cars. The VDA reported 392,200 passenger cars built in Germany in September 2026, unchanged from September 2025. Eurostat’s vehicle index for August was 7.0% below its level a year earlier. Units and value-added weights treat the same factories differently, and the association’s series is not seasonally adjusted, so both can be right while describing different things.
The second involves sentiment. ifo’s August release described July as 86.7 points; the July release had reported 86.6. A tenth of a point is noise, yet it is a reminder that the number in a headline is a preliminary estimate revised in the following month’s press release. Anyone building an argument on a single month of survey data is building on sand that moves.
The third involves insolvencies, where the monthly rates are almost useless in isolation. The Federal Statistical Office reported a 2.0% decline in May 2026 against May 2025, a 15.8% increase in June, with 2,266 applications, and an 8.0% increase in July. Filing dates cluster around quarter-ends and the statistic counts applications rather than closures, so the direction of travel over a year is the only usable signal.
The fourth involves employment. The EU labour force survey puts manufacturing employment at 7.59 million in 2025; establishment-based headcounts from the same statistical system give a different number because they count positions rather than people and cover firms above a size threshold. The gap between the two is not an error. It is a definition.
Base years add another layer. Every index in this piece is expressed against 2021 = 100, a year distorted by pandemic restrictions, supply bottlenecks and the beginning of the energy price squeeze. A branch that fell 30% between 2018 and 2021 and has been flat since looks stable on this scale and devastated on a 2018 scale. The choice of base is not a technicality; it decides which story the chart tells.
Chemicals: a new low rather than a cycle
The chemicals index fell to 74.9 in August 2026, down from 78.3 in June and 3.2% below August 2025. In a series that starts in 2015, only one month has been lower: December 2022, at 74.7, when gas prices were at their post-shock peak. The branch is now 24.1% below its 2018 average and a quarter below its 2021 base.
Corporate decisions match the index. BASF announced at the end of August 2024 that it would close its adipic acid, cyclododecanone and cyclopentanone plants at Ludwigshafen, and it is pursuing around €1.1 billion in annual savings by the end of 2026 under a Europe-focused programme, within a wider effort intended to deliver more than €2.1 billion a year from that point. The company’s own explanation names low market growth in Europe, a sluggish recovery in demand, gas prices higher than in other regions and continuing administrative hurdles. None of those causes reverses when the order book improves.
The sector’s scale makes the attrition consequential. The chemical and pharmaceutical industry turned over around €220 billion, employed roughly 476,300 people across more than 2,100 firms, and exported goods worth more than €252 billion in 2025, according to its association. It accounts for close to a tenth of manufacturing turnover and sits at the start of chains that run into plastics, construction, vehicles and health care. A branch of that weight does not shrink by a quarter without the shrinkage being felt downstream.
The downstream arithmetic is worse than the headline. An index at 75 means the crackers are running well below the rate at which their fixed costs were designed to be covered, and the shortfall is passed to every plant that buys steam or an intermediate from them. Closure decisions in this branch are made plant by plant, yet the cost base they respond to is site-wide, so each closure raises the pressure on the next one. That is how a site loses a quarter of its output without any single decision looking large.
Energy prices stopped being the binding constraint
The standard explanation for the German industrial problem is the price of power, and the standard version of that explanation is out of date. Eurostat’s comparison for the first half of 2026, for industrial consumers using between 2,000 and 19,999 megawatt-hours a year and excluding recoverable VAT, puts Germany at 17.42 cents per kilowatt-hour against an EU average of 15.59. France pays 12.39, the Netherlands 18.24, Poland 18.12 and Italy 19.91.
The German figure has also fallen, from 19.32 cents in the first half of 2025 to 17.42 cents a year later, a decline of 9.8%. The premium over the EU average is now roughly 12%, not the doubling that circulates in political speeches. Among large consumers using 150,000 megawatt-hours or more, Germany pays 12.67 cents against an EU average of 11.28, a premium of 12.3%. In the middle band of 500 to 1,999 megawatt-hours the gap widens to 14.3%, which is where the Mittelstand lives.
Three conclusions follow. Power is more expensive in Germany than in France, and cheaper than in Italy, Poland and the Netherlands, so a story in which energy alone explains the location of European industry cannot hold. The burden is distributed unevenly: a plant in the 500 to 1,999 megawatt-hour band pays 20.58 cents while one above 150,000 pays 12.67, which turns electricity into a scale advantage and rewards exactly the plants that are already big. And for chemicals, the binding input was never only electricity; it was gas as feedstock, priced against American shale and Middle Eastern associated gas rather than against the European power exchange.
The distributional detail is where the politics sits. A plant consuming 800 megawatt-hours a year pays 20.58 cents against an EU average of 18.00 for that band. One consuming 10 gigawatt-hours pays 17.42 against 15.59. One above 150 gigawatt-hours pays 12.67 against 11.28. The percentage premium is similar at each step; the absolute bill is not, and the mid-sized supplier has no purchasing department able to negotiate the spread away. Relief schemes that work through consumption thresholds protect the plants that are already competitive and leave that supplier exposed.
The machinery rebound is real and narrower than it looks
Machinery output rose from 83.5 in June to 85.2 in July and 89.7 in August, a gain of 7.4% in two months. Set against August 2025 the index is 0.6% lower, and against 2018 it is 18.7% lower. The rebound is genuine in the monthly data and modest in any longer frame.
The order data complicate the picture. The VDMA reported a 5% fall in order intake in August 2026, with every sub-indicator down, ending two months of growth. Its quarterly survey, which collected around 1,060 responses from 797 companies between 22 September and 7 October, found sentiment brightening while structural problems at the German location remained visible. The association’s chief economist expects 2026 to be the fourth consecutive year of falling machinery production, with growth returning in 2027 on the strength of the better order book.
That is a forecast of recovery from a low base, made by an organisation whose members have every incentive to talk up the second half. It deserves to be read as a statement about direction rather than level. A branch that has lost nearly a fifth of its output since 2018 does not return to that level because sentiment improves by four points; it returns when its customers invest, and German manufacturers have been slow to do so.
What the payroll and the insolvency files already know
Manufacturing employment fell from 7.99 million in 2019 to 7.59 million in 2025, a loss of 409,000 jobs or 5.1%. Measured from the 2020 peak of 8.24 million the decline is 657,000, or 8.0%. The series is annual and survey-based, so the level should be treated with caution and the direction with less.
Two features stand out. The decline began in 2021, before the gas shock of 2022 had fully reached the labour market, which weakens the case that a single price event caused it. And it has continued through 2024 and 2025, years in which output was roughly flat, which is what attrition looks like: firms that expect volumes to return keep people, and firms that have written the volume off do not.
Employment also lags output by more than a year, so the 2026 stabilisation in production has not yet had a chance to appear in these numbers. If the recovery is real, headcount should stop falling in the 2026 or 2027 vintage of the survey. If it keeps falling while output holds, the top of the chain is substituting capital for labour and the jobs will not come back with the volumes.
The insolvency file points the same way, though it is noisier. Company insolvencies rose 10.3% in 2025 against 2024, according to the Federal Statistical Office. The monthly path in 2026 has been erratic: a 2.0% fall in May, a 15.8% rise in June with 2,266 applications, an 8.0% rise in July. The volatility is a property of the statistic, not of the economy. Applications are filed when a filing becomes unavoidable, which clusters around quarter-ends and around the dates on which banks review facilities.
The composition matters more than the count. Insolvency in Germany is dominated by small firms and by sectors tied to construction and consumer demand, and the manufacturing cases that make the news are usually the visible tip of a longer list of suppliers. A rising trend into 2025 is consistent with a chain that is shedding its weakest links, which is precisely what attrition looks like from the inside.
The statistic cannot settle the argument on its own. It can, however, rule out one story: an economy in a sharp cyclical trough would show a spike followed by a fall as demand returned. A sustained upward drift with monthly noise looks more like a structural clearing.
Neither series can say which plants closed and which merely stopped hiring, and that gap is why the argument about German industry keeps recycling the same evidence. The production index reports what was made. The payroll reports who was paid. Neither reports what capacity has been removed for good, which is the only number that separates a bad cycle from a smaller industry.
China changed sides of the table
In 2025 China returned to being Germany’s largest trading partner, with goods trade of €251.8 billion, up 2.1%, while trade with the United States fell 5%. The composition of that relationship has changed. In the first nine months of 2025 German exports to China fell 12.3% to €61.4 billion, while imports from China rose 8.5% to €124.5 billion.
For the machinery branch the implication is uncomfortable. The export surplus that financed German investment for two decades rested on selling capital goods to customers who did not yet make them. Those customers now make them, and the German response has been to defend a shrinking share of the Chinese market rather than to find a new one. The United States, the other large market, has become more expensive to sell into.
This is the mechanism that links the two halves of the story. A chemical plant or a steel mill that closes releases its customers to search for alternatives abroad, and a machinery firm that loses its Chinese market share loses the volume that justified its own capacity. The bottom of the chain loses first, and the top loses later.
Three mechanisms that keep the split open
The split is not self-correcting. Three mechanisms keep it open, and they operate at different speeds, which is why the aggregate can look stable while the distribution underneath it keeps shifting.
- Co-location economies. A cracker supplies steam, gases and intermediates by pipeline to the plants around it. Run it at 75% and the fixed cost per tonne rises for every one of those plants. The 2022 gas shock turned a variable-cost problem into a fixed-cost problem, and the fixed-cost problem outlasts the price spike.
- Qualification stickiness. A carmaker or a machine builder that re-sources a steel grade or a specialty chemical spends a long time qualifying the new supplier, and once qualified it does not switch back on price alone. The demand that leaves during a closure does not return when volumes recover.
- Skills hysteresis. Four hundred thousand fewer people work in manufacturing than in 2019, and the ones leaving are disproportionately older. Apprenticeship intake tracks output with a lag. Restarting a plant requires operators, maintenance crews and process engineers who have already retired or moved.
Each mechanism converts a temporary shock into a permanent level shift. Together they explain why an aggregate index can stabilise while the distribution underneath it keeps moving.
What would change my mind
Of the four tests, the first is the one to watch and the one currently moving the wrong way. Chemicals fell to 74.9 in August, away from the 85 threshold, and no closure has been reversed. The second test is ambiguous: machinery output rose, but order intake fell 5% in August and the survey evidence is about expectations rather than capacity. The third has not been triggered, because employment is still falling. The fourth remains open until the next benchmark revision of the production index, which will tell us whether the stabilisation since mid-2025 was a level or a base effect.
The next two quarters will settle much of it. ifo publishes its October reading on 26 October, the day after this piece goes out, and a fourth consecutive gain would confirm that the turn is more than a one-month artefact. The Federal Statistical Office will publish third-quarter output and the November insolvency count. Chemicals either break 80 or they do not.
One further possibility deserves naming. The stabilisation could be the result of firms working through order backlogs accumulated in 2023 and 2024, in which case output turns down again once the backlog is gone and the improvement in sentiment will look like a lagging indicator rather than a leading one. Machinery orders fell 5% in August, which is consistent with that reading and is the main reason I would not yet call the top of the chain recovered.
If chemicals recover above 85 and hold, I will have been describing a cycle as a structure, and the correction should be loud. If they stay near 75 while machinery and electrical equipment grow, the German problem is not the price of energy or the level of demand. It is that the country is trying to run an export machine whose first stages have been sold, closed or moved, and no amount of sentiment will substitute for the tonne of ammonia that is no longer made.
Sources
- Eurostat — Production in industry, monthly data (sts_inpr_m), Germany, NACE C, C20, C24, C28, C29 and B–D, 2021 = 100, seasonally and calendar adjusted — https://ec.europa.eu/eurostat/databrowser/view/sts_inpr_m/default/table?lang=en
- Eurostat — Electricity prices for non-household consumers, bi-annual data (nrg_pc_205), Germany and EU27, bands IC, ID and IG, excluding VAT, first half of 2026 — https://ec.europa.eu/eurostat/databrowser/view/nrg_pc_205/default/table?lang=en
- Eurostat — Employment by sex, age and economic activity (lfsa_egan2), Germany, manufacturing, thousand persons, 2015–2025 — https://ec.europa.eu/eurostat/databrowser/view/lfsa_egan2/default/table?lang=en
- ifo Institute — ifo Business Climate Index for Germany, September 2026 reading of 89.9 points and the August and July releases — https://www.ifo.de/en/survey/ifo-business-climate-index-germany
- ifo Institute — release calendar and monthly press releases, October 2026 index due 26 October 2026 — https://www.ifo.de/en
- Statistisches Bundesamt (Destatis) — company insolvencies in 2025 up 10.3% on the previous year — https://www.destatis.de/DE/Presse/Pressemitteilungen/2026/03/PD26_085_52411.html
- Statistisches Bundesamt (Destatis) — press release index, company insolvencies in July 2026 up 8.0% on July 2025 — https://www.destatis.de/DE/Home/_inhalt.html
- STB Web — report citing Destatis, company insolvencies in June 2026: 2,266 applications, up 15.8% year on year — https://www.stb-web.de/
- VDMA — order intake in mechanical engineering, August 2026: a slight damper, all indicators down, minus 5% — https://vdma.eu/viewer/-/v2article/render/166300976
- VDMA — economic outlook pages, including the Q3 2026 survey of 797 companies and the chief economist’s forecast of a fourth consecutive year of falling production — https://vdma.eu/konjunktur
- VDA — production and market statistics, German passenger car output of 392,200 units in September 2026 — http://en.vda.de/
- VCI — industry portrait of the German chemical and pharmaceutical industry: turnover, employment, exports — https://www.vci.de/die-branche/zahlen-berichte/branchenportraet-der-deutschen-chemisch-pharmazeutischen-industrie.jsp
- BASF — objectives for the Ludwigshafen site, including plant closures, the €1.1 billion savings target and the reasons given — https://bericht.basf.com/2024/de/zusammengefasster-lagebericht/grundlagen-des-konzerns/die-basf-gruppe/zielsetzung-fuer-den-standort-ludwigshafen.html
- Bundesministerium für Wirtschaft und Energie — autumn projection 2026, GDP growth of 1.3% and the upgrade of 0.8 percentage points — https://www.bundeswirtschaftsministerium.de/Navigation/DE/Home/home.html
- Xinhua / People’s Daily Online — German trade with China in the first nine months of 2025, exports down 12.3% to €61.4 billion — https://en.people.cn/n3/2025/1120/c90000-20392887.html
- Xinhua / ezhejiang — China-Germany trade of €251.8 billion in 2025, up 2.1%, China back as Germany’s largest trading partner — https://www.ezhejiang.gov.cn/2026-02/25/c_1162065.htm