Test whether a scheme's mix of studios, one-beds, two-beds and three-beds can be built at a 15% margin, by region and tenure: first with no State support, then with the interventions you switch on. Costs and values come from the SCSI Real Cost reports, rebased to mid-2026.
Outside Dublin, values and land use the regional proxies described in the Method.
Run on the SCSI inputs, as published (Q2 2025, Dublin, case-study specification), the model should land on the SCSI results. Average costs and margins match closely. Cost rents come out about 3% low, and the split between unit types is a Spatialis derivation because SCSI publishes averages only.
SCSI figures: Real Cost of New Apartment Delivery 2025, Tables 6, 7, 8 and 11, pp22–32.
The model rebuilds the SCSI per-home development appraisal for each unit type, then adds them up across the mix. Nothing is typed in from outside the sources listed at the foot of this page. Where a figure is derived rather than published, the derivation is set out below. Where a figure was needed and could not be found in a source, it is marked Source not verified in the tool and listed here.
SCSI publishes average costs per home, not costs by unit type. The case study gives two designs on the same 4,623 m² building: 48 homes at an average build cost of €307k, and 54 homes at €284k (Table 10, p30). Holding floor area fixed and adding six homes added about €0.6m of build cost, which separates a cost per dwelling from a cost per square metre:
This is a Spatialis derivation. Because both figures are rounded to €1k in the SCSI table, the split carries an error of roughly ±€8k per dwelling. It is the reason small units struggle in the model: the per-dwelling cost does not shrink with the unit, but the sale price does.
The building-type options scale both parts by the SCSI €/m² for that category against the case study's €/m² (Figure 2, p12, total construction incl. parking and siteworks). Gross floor area per unit type comes from Table 9 (p29). The studio has no pre-2025 size in the case study, so the 2025 size is used in both options.
Other cost lines follow the SCSI assumptions (pp20–21): land at €52k per home for sale and rent and €30k for cost rental, spread by floor area; development contributions and utility connections at €24k, €26k and €27k per home; professional fees at 8% of build cost; sales and letting at 2.5% of value (€10k per home for cost rental); contingency at 3%; finance on 75% debt at 7% and 25% equity at 15%, with land held 2.5 years and 0.5% due diligence. SCSI does not publish its drawdown profile. A build-period factor of 0.7 years reproduces its €38k finance cost and is used throughout.
Sizes are entered as net internal area and converted to gross floor area with the SCSI two-bed ratio, 78.5 m² net to 91 m² gross (2025 p11), so 86.3% unless changed. Gross area drives the per-m² part of build cost and the share of land. The per-dwelling part of build cost does not change with size, so shrinking a unit saves roughly the per-m² rate on the area removed, plus the land, fees, contingency and finance that go with it.
The SCSI block sizes are well above the minimums because the scheme was designed with oversized units (half of units at least 10% above minimum, 2025 p28). The minimum-standards preset uses the figures in SCSI's table of the 2025 guideline changes (p28): studio 32 m², one-bed 45 m², two-bed three-person 63 m², two-bed four-person 73 m², three-bed five-person 90 m².
What a smaller unit sells or lets for is the open question. Two settings are offered. Price per home holds keeps the SCSI price and rent for the unit type, which is what SCSI did when it reduced unit sizes in its redesign (p31). Price falls with size moves price and rent along a line drawn through the five SCSI case-study price points against gross area, scaled so each type keeps its SCSI price at its case-study size. The second is a Spatialis derivation; the true answer for a given location lies between the two and needs local sales evidence.
Site area is measured from the uploaded KML polygon on a sphere with the WGS84 equatorial radius, with any holes subtracted. Net developable area is the gross area times the share you set. Homes = net area × the density you enter in dwellings per hectare. Neither the share nor the density has a default from a source; both are for the user to set from the plan and their knowledge of the site.
The optimiser tries every mix in 5% steps (10,626 combinations), runs each through the same appraisal with the selected tenure, region, sizes and interventions, and keeps the mix with the highest share of studios, one- and two-beds that still clears the hurdle. Ties go to the mix with the smaller average home, then to the more profitable one. If no mix clears the hurdle, it shows the mix with the highest margin and the support still needed. Optional caps follow the pre-2025 guideline levels SCSI cites (studios about 20–25%, studios and one-beds up to 50%, 2025 p28); the lower studio figure is used. Land cost stays on the SCSI per-m² basis, so the site's actual price does not enter the appraisal.
Viability for sale and rent is the SCSI test: net value against total cost, with a 15% gross margin on cost as the hurdle (p11). For cost rental it is whether the cost rent fits under both rent rules.
| Lever | How it is modelled | Source |
|---|---|---|
| VAT at 9% | Net value = gross ÷ 1.09 instead of ÷ 1.135. For cost rental SCSI finds only €5 a month, which is applied as stated. | SCSI 2025 p23; deck |
| Croí Cónaithe (Cities) | Build to sell only. Each apartment receives its gap to the hurdle, capped. Dublin cap €120k. Caps for Cork, Galway, Limerick and Waterford were not found and default to zero until entered. | SCSI 2025 pp23–25; deck. Non-Dublin caps: Source not verified. |
| STAR / cost rental equity | State equity per home reduces the capital the rent must repay, sized to bring the rent under both caps, up to €175k Dublin, €150k elsewhere, +€25k sustainability. | SCSI 2025 p26; deck |
| Cost rental CT exemption | €13 a month off the cost rent. | SCSI 2025 Table 7, p23 |
| 125% CT deduction | min(25% of build cost, €50k) × tax rate, for schemes of 10 or more apartments. The tax rate that applies to a developer's profit is not in the sources loaded; the default is an assumption to check with a tax adviser. | Chartered Accountants Ireland, Budget 2026; deck. Tax rate: Source not verified. |
| Levy and Uisce Éireann waiver | Removes the region's planning levy and the €5,600 Uisce Éireann charge from the SCSI 2023 tables. In Dublin that is €17.5k a home, inside SCSI's €15–20k range for the waiver. | SCSI 2023 pp10–16; SCSI 2025 Fig 8 p24 |
| Reduce or omit car parking | Cuts build cost by (base spaces a home − chosen spaces a home) × cost per space. Cost per space = SCSI parking cost per m² for the category × 91 m² two-bed ÷ the spaces per home SCSI assumes (0.8, 0.6, 0.4). The case study uses its 36% ratio and the Category 3 midpoint. Fees, contingency and finance fall with it. No price or rent discount for homes without a space is applied, as no source was loaded for one. The planning policy basis for omitting parking near public transport was not in the sources loaded. | SCSI 2025 Appendix 2 p45; p21. Policy basis: Source not verified. |
| Public land discount | Cuts the land cost by the chosen share. SCSI's cost rental case uses €30k against €52k on the open market. | SCSI 2025 p21 |
| State-rate finance | Replaces the debt and equity blend with one rate, and drops due diligence fees. | SCSI 2025 p21 (HFA 3.75%) |
| Lower margin | Lowers the hurdle. SCSI notes forward-funded or forward-commit schemes may accept less than 15%. | SCSI 2025 p25 |
Part V is not a separate lever. Both SCSI reports treat it as already reflected in land values (2023 pp10–16; 2025 p18). CREL, the Affordable Housing Fund, Project Tosaigh and Home Building Finance Ireland are not modelled separately; their terms in the deck are not specific enough to put into an appraisal, though the land, finance and cost rental equity levers cover the same ground.
The apartment report is priced at Q2 2025 and the housing report, used here only to set regional values and land, at the end of 2023. Both are moved forward with published indices:
| Item | Index | Factor |
|---|---|---|
| Apartment build costs, Q2 2025 → H1 2026 | SCSI TPI 1H25 220.4 → 1H26 229.3 (2H25 222.6 plus 3% for H1 2026) | |
| Apartment sale prices, June 2025 → June 2026 | CSO RPPI, Dublin, +4.6% in the 12 months to June 2026 | |
| House build costs, fees and finance, end-2023 → H1 2026 (regional proxy only) | SCSI TPI 2H23 210.9 → 1H26 229.3 | |
| House values, Dublin / GDA, Dec 2023 → June 2026 (regional proxy only) | CSO RPPI Dublin +8.3% (2024), +5.6% (2025), then national 204.6 → 206.9 to June 2026 | |
| House values outside Dublin (regional proxy only) | CSO RPPI outside Dublin +9.0% (2024), +8.1% (2025), then national to June 2026 |
Limits of this: the TPI is a commercial tender index, used by SCSI itself as a cross-check (2025 p13), not a residential cost index. The GDA region includes Kildare, Meath and Wicklow but is indexed with Dublin prices. CSO's regional house series could not be read in full for every region, so all non-Dublin regions share the outside-Dublin figure. Land, levies and rents are not indexed because no source for their movement was loaded; rents can be adjusted in the overrides.
The 2025 apartment report covers Dublin only, and states that its 1,643 non-Dublin units (Cork, Limerick, Galway) "did not diverge significantly from the general cost trends across the Dublin sites" (p11). So build costs are held at Dublin levels by default. Apartment values, rents and land outside Dublin are not published by SCSI. The model uses the 2023 3-bed semi as a proxy: each region's rebased house value and land cost as a share of the GDA's. This is a Spatialis assumption. It is likely to flatter regional apartments if apartment values outside Dublin sit further below Dublin's than house values do. Use the overrides with local evidence where you have it.
The 33 local authority areas use CSO data rather than the SCSI house proxy, picked from the CSO's own area breakdown (the former city council areas of Cork, Galway, Limerick and Waterford are listed separately from their counties):
The cost rental rent cap outside Dublin applies the same 35% rule to the €59k net income limit (SCSI 2025 p26), giving € a month. SCSI does not state this figure.