Evidence auditGerman manufacturing · 2015–2026

Orders / backlogs / production

The June divergence is real. The structural break is not established.

German electronics and machinery accumulated orders faster than factories converted them into output. Yet the full 2015–2026 record does not show that backlogs became less predictive after 2021.

Follow the evidence
Electronics / June orders
+22.7%

Production +0.1% month on month

Machinery / June orders
+12.7%

Production −3.9% month on month

Aggregate major-order wedge
+3.6 pp

Total orders minus orders excluding large-scale orders

01 / The question

Did German manufacturing backlogs lose their predictive value after 2021?

The latest release invites a strong story. In June 2026, new orders rose by +22.7% in electronics and +12.7% in machinery. Production barely moved in electronics and fell −3.9% in machinery. Order books lengthened to 9.5 and 11.5 months. Orders arrived; output did not follow.

That observation is correct. The interpretation is not yet determined. A one-month divergence can reflect lumpy major orders, delivery schedules, capacity constraints, product mix, revisions, or timing. A structural conversion failure requires a different test: whether today's backlog contains less information about future production after 2021 than it did before the pandemic.

The current gap is a fact. A permanently broken order-to-output mechanism is a hypothesis.

The answer from two matched two-digit manufacturing sectors is deliberately less dramatic. Raw six-month correlations become more positive after 2021, not less. Once new orders and an energy-supply producer-price control enter the model, the post-2021 backlog coefficients are too imprecise to separate from zero. The evidence therefore rejects a clear deterioration, but it does not establish a stable causal conversion channel.

02 / June 2026

The release contains two divergences, not one.

June 2026 monthly manufacturing indicators by sector
SectorNew ordersOrder stockProductionTurnoverOrder range
Electronics+22.7%+4.6%+0.1%−2.3%9.5 months
Machinery+12.7%+0.9%−3.9%−0.3%11.5 months

Sector conversion gap

Annual backlog growth exceeded annual production growth in both sectors: by 16.9 points in electronics and 6.3 points in machinery on the smoothed series.

Aggregate major-order wedge

Manufacturing orders rose +3.1%, but fell −0.5% when large-scale orders were excluded. The headline was unusually lumpy.

Different units of analysis

The major-order check is aggregate manufacturing. The conversion analysis uses matched sector series. It diagnoses the release; it does not mechanically explain either sector.

Interactive 01 / matched sector paths

Backlogs and production separated sharply, but not permanently.

Annual growth rates use three-month moving averages. The shaded interval is the post-2021 comparison period, not a presumed structural break.

Computer, electronic and optical products order-stock and production growth, March 2016 to June 2026In Jun 2026, annual order-stock growth was +16.6% and production growth was −0.4%. The two series diverged during the pandemic and in 2022, then partially converged.POST-2021 TEST PERIOD-20%-10%0%10%20%30%40%201620182020202120232026Move across the chart to inspect a month
MonthJun 2026
Order stock / year+16.6%
Production / year−0.4%
Standardized gap+1.18σ

Move across the chart or use the month slider.

Text table for benchmark months
MonthOrder-stock growthProduction growthConversion gapOrder range
Mar 2016+2.8%+0.5%−0.36σ4.0 months
Dec 2019−4.6%+0.3%−1.09σ4.9 months
Jan 2021+12.0%+0.7%+0.53σ6.0 months
May 2022+28.7%+3.0%+1.80σ8.0 months
Jun 2026+16.6%−0.4%+1.18σ9.5 months

The gap is a descriptive index: standardized annual order-stock growth minus standardized annual production growth within each sector. It is not an estimate of unfilled demand or forgone output.

Interactive 02 / predictive test

The raw signal improves. The controlled estimate remains uncertain.

The left panel shows the simple correlation between today's backlog growth and production growth over the next 6 months. The right panel shows the post-2021 backlog coefficient after controlling for new orders and energy-supply producer prices.

Raw correlation / −1 to +1

Electronics

2016–2019−0.1472021–2026+0.607

Machinery

2016–2019−0.1262021–2026+0.548

Controlled post-2021 coefficient / 95% HAC interval

Electronics

-4-20+2+4

β +0.905 · 95% [−0.681, +2.491] · n = 60

Machinery

-4-20+2+4

β −0.341 · 95% [−3.177, +2.495] · n = 60

Forecast evidence at the selected horizon
SectorHorizonPre raw correlationPost raw correlationPost controlled coefficient95% interval
Electronics6 months−0.147+0.607+0.905[−0.681, +2.491]
Machinery6 months−0.126+0.548−0.341[−3.177, +2.495]

Models predict future production growth; they do not identify a causal backlog effect. Newey-West intervals use the forecast horizon as the lag length. The controlled post-2021 intervals cross zero for both sectors at six months.

05 / Interpretation

Three claims survive. The strongest one does not.

First, the backlog-output gap widened materially in specific episodes. Electronics reached its largest standardized gap during the 2020 production collapse. Machinery reached its maximum in spring 2022, when annual order-stock growth exceeded 20% while production contracted. A visible gap is therefore not unique to June 2026, and its meaning changes across episodes.

Second, the post-2021 backlog is not obviously less informative. At a six-month horizon, the raw stock correlation changes from −0.147 to +0.607 for electronics and from −0.126 to +0.548 for machinery. Both move upward. That pattern runs against the proposed structural deterioration.

Third, raw predictive content is not the same as an independent backlog effect. In the controlled six-month model, the electronics coefficient is +0.905, with a 95% interval from −0.681 to +2.491. Machinery is −0.341, with an interval from −3.177 to +2.495. Both intervals cross zero by a wide margin.

The stronger conclusion is not that conversion works. It is that the available evidence cannot support a claim that it clearly stopped working after 2021.

For an industrial manager, that distinction changes the diagnostic. A long order book remains a reason to inspect future production, but not a production forecast by itself. The useful operating question is narrower: which constraints determine whether a particular order becomes scheduled, supplied, produced, and recognized as turnover? Aggregate backlog data point toward that question. They do not answer it.

06 / Method and limits

A reproducible predictive audit, with a deliberately narrow claim.

Design

  1. 1. Match sectors. Electronics (WZ08-26) and machinery (WZ08-28) are observed across new orders, production, turnover, order stock, and order range.
  2. 2. Align adjustment. Monthly index series use X13 JDemetra+ calendar- and seasonally adjusted variants with 2021 = 100.
  3. 3. Reduce monthly noise. Three-month moving averages precede annual log-growth transformations.
  4. 4. Split the test. Models compare March 2016–December 2019 with January 2021–June 2026; 2020 is excluded from both regimes.
  5. 5. Preserve uncertainty. One- through six-month forecasts use Newey-West HAC intervals with lag length equal to the horizon.

Limits

  • Two sectors cannot establish a manufacturing-wide structural break.
  • Predictive regressions do not identify a causal conversion mechanism.
  • The GP19-35 series is an energy-supply producer-price control, not a direct sector energy-cost measure.
  • Major orders explain the aggregate June headline but are not allocated to the two sector panels here.
  • Destatis series are revised; this publication freezes the release vintage retrieved on 19 August 2026.
  • The January 2026 GP classification change affects detailed production allocation; the analysis remains at two-digit WZ level.

07 / Data and provenance

The conclusion travels with the panel, models, and hashes.

Eight official GENESIS snapshots are preserved as immutable raw inputs. The generator checks their byte counts and SHA-256 hashes, selects explicit variable codes, rebuilds every public artifact, and records output hashes in the publication manifest.

Release contract: 276 panel rows, 24 forecast models, data through June 2026, generated from frozen inputs retrieved on 19 August 2026.

Matched sector panel

CSV · 276 monthly observations across electronics and machinery

Download panel CSV

Forecast models

CSV · coefficients, HAC intervals, correlations, horizons, and samples

Download model CSV
Frozen official inputs
TableMeasureBytesSHA-256
42151-0004 Monthly manufacturing new-orders volume index for aggregate manufacturing and main groups1,432,035b33baf1426ec2c1e9d79a02cf9032ba00b3a72c4a9ce6219d288f8fdca0e9e3c
42151-0005 Monthly manufacturing new-orders volume index by WZ 2008 sector and market direction2,878,08930e67c67175c0577228ffa8964a502d0ea94b323aab5cc2379a6a9cfb2dbf64d
42151-0008 Monthly manufacturing new-orders volume index excluding large-scale orders1,489,631417bbcd38847d61085693688049f829879ccc5ce5b425df712ee23216d56677e
42152-0005 Monthly manufacturing turnover volume index by WZ 2008 sector and market direction2,807,976d67c302eae252f9995e99540089994e3dda7d349ba4847d557823f9e6c29396b
42153-0002 Monthly manufacturing production index by WZ 2008 sector494,543b8482ce82a0bb3929e032207764e481cb6fb49e87e0aac3032d3ccb4cf93c1e9
42113-0002 Monthly range of unfilled manufacturing orders in months by WZ 2008 sector82,499ca3ef6c80f4a0ad0c34c1ba5e8d89cea471ee11ed28e9a6799189406964aa437
42155-0005 Monthly manufacturing order-stock volume index by WZ 2008 sector and market direction358,25734183349e340d9c8d3af8c55b56b8e0a1f52ac41999a6e64498f2728e568bf11
61241-0004 Monthly producer price index for industrial products by GP 2019 product group40,320b1aee289e7863fb7973a2258e35a821725030fea9ce1dfd3343e4f9396048419

08 / Sources

References

Source data are licensed under Data licence Germany – attribution – Version 2.0. Release values are cross-checked against the cited Destatis publications.

Eight source tables · one reviewed vintage
  1. 01

    Stock of orders in manufacturing in June 2026: +0.8% on the previous month. Destatis press release No. 294, 19 August 2026.

  2. 02

    New orders in manufacturing in June 2026: +3.1% on the previous month. Destatis press release No. 276, 6 August 2026.

  3. 03

    Production in June 2026: +0.2% on the previous month. Destatis press release No. 279, 7 August 2026.

  4. 04

    GENESIS-Online manufacturing and producer-price tables. Tables 42113, 42151, 42152, 42153, 42155 and 61241.

  5. 05

    Data licence Germany – attribution – Version 2.0. Official German open-data licence.