This page runs your figures through two engines at once: the States of Guernsey 2026 Tax Reform Package Calculator, lifted verbatim from the live page, and beside it a version built to match the tax proposals the States are being asked to approve. Wherever the live calculator departs from those proposals, its figure is struck through and the matching figure is shown beside it.
Read the accompanying article: Check their sums →
Figures update as you type; the ledger and the breakdown beneath it recalculate with every change, so there is no button to press. Loading a preset triggers one difference and lights up the matching step in the breakdown headed “Where the live calculator departs from the proposals”. A plain salary triggers none of them, which is why that breakdown opens by reporting that the calculator matches the proposals for this household; the presets are the quickest way to see a single difference on its own.
| Line | Current system | 2026 package (States) | 2026 package (corrected) |
|---|
Starting from the States figure and applying each correction in turn, until the figure that matches the proposals. A red step marks a point where the calculator overstated the benefit of the package; a blue step marks one where it overstated the cost.
Ordered by materiality. The single-person arithmetic is sound; the differences from the proposals sit around it. Throughout, the benchmark is the 2026 Tax Reform Package policy letter of 8 June 2026 and its Executive Summary, not independent ground truth.
The capture contains four requests: the page (twice) and a missing favicon. No external scripts, no fonts fetched, no cookies, no data sent on completion or on leaving the page; the content-security-policy is restricted to the site itself. The privacy claim holds as deployed. Two caveats: the policy needlessly permits fonts.googleapis.com although nothing is loaded from it, and the page request itself still produces ordinary server access logs, so “no data collection” is accurate for your figures but not for request metadata.
Matched to the penny against the policy letter at £25,000, £50,000, £100,000 and £200,000. This also settles an ambiguity in the drafting: the 15% band is anchored at £28,000 of income, not of taxable income; withdrawn allowance falls into the 15% band, giving an effective marginal rate of 33.2% between £85,000 and £140,610. The abolition of the lower earnings limit, both allowance tapers, the pensioner classes, mortgage relief as a deduction and the 75% treatment of childcare and healthcare costs are all implemented correctly.
Each is reproduced live above and in the downloadable test suite.
| Ref | Departure from the proposals | Illustrative effect |
|---|---|---|
| F1 | Mortgage interest removed from the GST base twice (it is already inside the mortgage-payments field) and without the £3,500 cap. | +3% of interest entered; £20k of interest strips £16.5k from the base |
| F2 | An employed person earning below the £11,122 contributions allowance who also has investment income is reclassified as non-employed and charged 8.5% on the lot. The policy letter states on page 8 that unearned income “will not be brought into scope for employed and self-employed people”, and its rate table gives 0%; the 8.5% charged is the abandoned 2024 package’s rate. | ≈ £2,455/yr over-charge on a 10k+30k case |
| F3 | The self-employed 2.5% band between the upper earnings limit and £300,000 is missing; the future self-employed settings also silently borrow the employee cap and allowance. | up to £2,586/yr under-charge above £196,560 |
| F4 | The unused-allowance transfer is granted to every two-adult household. The calculator never asks about marriage at all: its data model carries a marital-status field, but no question on the form sets it and no formula reads it. | ≈ £490 to £570/yr mis-stated for an unmarried couple |
| F5 | The annual vehicle-tax matrix runs £40 to £300, outside the published £25 to £280 schedule. | large petrol vehicle charged £20 over the stated maximum |
The Essential Costs Relief Payment is hard-coded at £310 single and £520 couple, with a cliff-edge eligibility ceiling of £32,400 and nil for a household recording zero income. The policy letter itself promises only payments proportionally lower than the previous package’s, to reflect the lower GST rate, and states no amounts for the reduced payment; the £520 and £860 it recites, in a rasterised paragraph, are the February figures. The November 2024 package set the payments at £520 and £860 as the estimated annual GST incurred by a household on Income Support at 5%; scaled to a 3% rate those become £312 and £516, which matches neither figure in the code, and the code’s couple payment of £520 is exactly the previous package’s single-adult payment. The £32,400 ceiling is the previous package’s higher-rate tax threshold, recited as such in the February 2026 resolutions and at page 74 of the policy letter bundle, and appears nowhere in the current package. The corrected engine leaves these untouched and flags them rather than inventing replacements.
Unless the single savings-and-abroad field is completed, the tool treats every remaining pound as standard-rated consumption. Against the archetypes in the policy letter’s own distributional model this moves the answer by roughly £400 to £900 a year, in both directions. Taken together the differences do not push in one direction: weighted by who will actually use the tool they lean slightly against the package, because the two largest (F2 and the wrongful transfer in F4) overstate the cost of the reform, while the one that flatters it (F1) is small and the one that flatters it most (F3) reaches only self-employed people earning above £196,560. A boundary inconsistency at £24,960 (the current side charges a pensioner at exactly the limit, the package side exempts) and the crediting of the 1.9% uplifts gross and untaxed are lesser items; the corrected engine harmonises the former and flags the latter.
The calculator compares the two systems at one moment. Income is entered once and used unchanged on both sides, so there is no wage growth and no fiscal drag; the bands, allowances and thresholds are held at present values. The only price movements written into the engine are two policy parameters rather than any measure of inflation: a fuel duty reduction that reprices fuel from £1.70 to £1.47375 a litre, a fall of 13.3% with the volume held constant, and a single 1.9% uplift to States pension and benefits as GST compensation. The Essential Costs payments are fixed at £310 and £520. No consumer price index appears anywhere.
Rent is entered once and never inflated. It is correctly treated as exempt: it is deducted from income to reach disposable income and is left out of the GST base, so the tool adds no GST to rent directly, which is right because residential rent does not attract GST. What is absent is the second-round effect. A landlord’s taxable costs, being repairs, maintenance, materials and agent fees, do carry 3% GST, and a share of that tends to reach rents over time, while a general rise in the price level can feed into rent reviews. None of this is modelled, so a renter’s cost may be understated to the extent that such pass-through occurs. The magnitude is uncertain, and the point is the omission of the channel rather than a figure.
The same snapshot character hides how the balance moves in later years. GST applies to rising nominal prices and consumption, whereas the compensating measures are fixed or one-off; the £310 and £520 payments and the 1.9% uplift are level amounts that inflation erodes, while the 3% keeps applying to a growing base. A household shown as roughly neutral today may drift worse off over subsequent years, and that drift does not appear here. Whether allowances and bands are uprated in future budgets is a policy choice the tool cannot reflect. The calculator’s own disclaimer is consistent with this, noting that it makes assumptions about how GST affects prices and that the real impact depends on the pricing decisions of businesses.