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

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.

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.