Count homes, not permissions: why Cork ranks 19th on one planning measure and 1st on the other

Ireland's planning statistics have a counting problem, and it changes the story about which counties are approving the most homes to be built.

The CSO's planning permissions data (BHQ17) counts each permission once, no matter what it covers. A one-off rural house granted permission counts the same as a 200-unit apartment scheme: one line, one permission. Over the four quarters from 2025Q3 to 2026Q2, one-off houses made up 77.8% of all permissions granted nationally -- but only 13.5% of the actual homes those permissions cover. The other measure in the same CSO release, "units for which permission granted," counts dwellings instead of paperwork. Nationally, each permission covers 5.76 homes on average, but that average is doing a lot of work hiding a very lopsided distribution: most permissions are for a single house, and a small number of permissions are for large schemes carrying most of the units.

That distinction flips the county-level story.

Counting permissions, counties that grant more of them tend to have lower sale prices -- a correlation of -0.52 across 26 counties. That fits an easy, familiar narrative: rural counties with cheaper land wave through lots of one-off housing, expensive counties don't. But count homes instead of permissions, and the relationship reverses to +0.48. Counties with higher sale prices are, on this measure, granting permission for more homes per resident, not fewer. And homes permitted track homes actually completed at the council level too: -0.43 for permission counts against completions, +0.49 for units against completions, across 29 local authorities.

The clearest illustration is Cork. Ranked by raw permission count per 1,000 residents, Cork sits 19th out of 26 counties -- hardly a big licenser of planning applications, with 1.37 permissions per 1,000 people. Ranked by homes per 1,000 residents, Cork is 1st, at 12.48 -- a jump of 18 places. Cork isn't granting many permissions, but the ones it grants cover large numbers of units: apartment blocks and estates rather than individual houses. Wicklow shows the same pattern (16th on permissions, 2nd on homes, up 14 places), as does Meath (22nd to 5th, up 17).

The pattern runs the other way too. Mayo is 2nd nationally on raw permission count (2.75 per 1,000) but falls to 19th on homes per 1,000 -- a drop of 17 places, the joint-largest fall alongside Kilkenny, which slides from 6th on permissions to 23rd on homes. Donegal, top of the permissions table at 3.7 per 1,000, drops five places once homes are counted instead. These are counties where planning activity is dominated by one-off rural houses: a lot of permissions, each covering one dwelling.

None of this proves that expensive counties cause more building, or that cheap land discourages density. These are cross-county correlations over a single 12-month window, and correlation is not causation. Controlling for land area, general affluence, and age structure shrinks some of these relationships and even flips the permissions-count correlation's sign in one partial (affluence, +0.17), while the homes-based correlation holds up more consistently across all three controls (+0.38 to +0.49). That asymmetry is itself informative -- it suggests the raw permissions-price relationship may be substantially an artefact of general county affluence, while the homes-based relationship isn't simply that -- but it isn't proof of a causal channel, and a single window can't rule out one-off effects like a handful of large schemes landing in a particular quarter. The same pattern turns up at council level: the permissions-completions relationship (-0.43) nearly vanishes once affluence is controlled for (partial r = -0.02), while the homes-completions relationship stays comparatively robust across all three controls (+0.40 to +0.50).

The practical upshot for anyone using planning data -- journalists covering a local authority's housing record, councillors comparing performance against neighbours, or researchers building a housing supply model -- is to check which of the two CSO measures they're using before drawing conclusions, and to remember that both measures count permissions, not completed construction. "Permissions granted" and "units for which permission granted" tell different, sometimes opposite, stories about the same underlying approvals activity, and the gap between them is largest precisely where one-off housing dominates. A county topping a permissions ranking may simply be processing a lot of small individual applications; a county near the bottom may be quietly approving one very large scheme.

The full county-by-county figures, alongside the completions and sale price data used for these comparisons, are available to browse or query on statsmapped.