Ley de Barrios asks for a neighbourhood diagnosis. Here is what the public data can currently answer

The Catalan neighbourhood regeneration instrument commonly known as Ley de Barrios requires a documented diagnosis at neighbourhood level before it will fund an intervention. So do ESF+ and AMIF applications, in their own formats. The requirement is reasonable: public money for a targeted intervention should rest on evidence that the target was correctly identified.

We wanted to know how much of that diagnosis a Catalan municipality can actually produce today from published data, without a bespoke study. So we worked through the requirements one at a time.

This is a desk exercise against public sources. No municipality is involved and none is named. It is a case study of an instrument, not of a client.

Method

We took the diagnostic dimensions the instrument and its comparable funding programmes ask a municipality to evidence, and for each one asked three questions:

1. Is there a published source at neighbourhood resolution?
2. How old is the most recent release?
3. Can it show change over a period short enough to justify an intervention now rather than describe one that already happened?

A dimension counts as answerable only if all three hold.

What we found

Answerable from published sources.

• Resident population and its composition by origin, at sub-municipal level, annually, from padró-derived series. Age is at best a year old and often more, but the resolution is there.
• Housing stock characteristics and tenure, from cadastral and census-derived sources. Slow-moving, so the lag matters less.
• Registered unemployment by municipality, monthly. Good frequency, but municipal rather than sub-municipal, which is the first place the resolution breaks.
• Educational attainment of the resident population, from census-derived series. Long release cycles, structural rather than operational.

Not answerable from published sources.

• Service demand at catchment level. What a specific social services point, health centre, or school is actually experiencing this quarter. The data exists inside the municipality's own systems and is not aggregated, not compared to a baseline, and not published.
• Change over a period shorter than a year at sub-municipal resolution. Almost every published neighbourhood series is annual. An intervention justified on a movement that happened at some point over the last twelve months is not a targeted intervention.
• Labour market participation at neighbourhood level. Registered unemployment stops at the municipality. Below that, it is inference.
• Integration pathway progression. Whether cohorts are moving through language acquisition, credential recognition, and employment entry, or stalling, and where. Not published anywhere at any resolution.
• Anything about competence rather than status. Published series record what people are, not what they can do.

The pattern

The split is not random. It falls almost exactly along one line: published data describes stocks, and the operational questions are about flows.

A municipality can evidence that a neighbourhood has a given composition. It cannot evidence, from published sources, that the composition changed last quarter, that a service in that catchment is over capacity now, or that a cohort stopped progressing six months ago.

That matters for funding applications specifically, because the instruments are asking for a justification to act. A stock figure justifies a description. A flow figure justifies an intervention. Most municipal applications end up arguing from stock figures and a narrative, because the flow figures are not available to them, and the evaluation panel can tell.

What closes the gap, and what does not

It is worth being clear that this is not a data availability problem in the ordinary sense. Nobody is withholding the missing half. Most of it sits inside the municipality already.

What is missing is aggregation, comparison to a baseline, and a resolution that is both fine enough to be operational and coarse enough to be safe to publish. Specifically:

• Aggregation across service systems. The caseload data is real time and lives in separate systems that do not talk to each other.
• A pinned neighbourhood boundary definition, versioned, so a series does not move because a line moved.
• A minimum cell size applied at query time, so sub-municipal reporting at short intervals does not create disclosure risk.
• A defined confidence method, so an indicator that is producing noise is reported as noise rather than as a movement.

Those four are the difference between the answerable and the unanswerable columns. They are exactly what the Diagnostic Sprint sets up, and exactly what the Observatory then runs on.

What does not close the gap: more national statistics, faster. A more frequent national series is still a national series.

What we would tell a municipality preparing an application

Three things, and none of them require a vendor.

Audit your own systems first. The missing half of your diagnosis is more likely to be in your own service records than in a dataset you have not found. Establish what you hold before you buy anything.

Fix the boundary definition before you fix anything else. If your neighbourhood boundaries are not pinned and versioned, every series you build on them will contain movements that are not real, and you will not be able to tell which ones.

Write the limitation into the application. Evaluation panels are not surprised that operational data is missing. They are unimpressed by applications that do not appear to know it. An application that says plainly which dimensions it can evidence, which it cannot, and what it will do about it reads as competent. One that presents a stock figure as though it were a flow reads as the opposite.

The honest summary

Roughly half the diagnostic requirements of a live Catalan funding instrument are answerable from published sources today. The unanswerable half is consistently the operational half: what changed recently, and what is under pressure now. That gap is not a data shortage. It is a missing layer, and it is the layer we build.