Stamp Duty Land Tax (SDLT) is a tax imposed on the purchase and rental of properties or land with values over a certain threshold in England and Northern Ireland. It applies to both residential and non-residential (commercial) properties.
Stamp Duty Land Tax parameters can be found in policyengine_uk/parameters/gov/hmrc/stamp_duty and logic in policyengine_uk/variables/gov/hmrc/stamp_duty_land_tax.py.
Legislation¶
The regulation of SDLT is defined in part 4 of the Finance Act 2003.
Residential purchase schedules¶
The model covers recent ordinary purchases by individuals in England and Northern Ireland. A user enters the price of a purchase whose buyer pays the higher rates as an additional purchase; the model does not use ownership history to decide which buyers pay them.
| Effective date | Standard nil band | First-time-buyer nil band | Inclusive first-time-buyer maximum | Additional-dwelling surcharge |
|---|---|---|---|---|
| 8 July 2020–30 June 2021 | £500,000 | No relief | No relief | 3 percentage points |
| 1 July–30 September 2021 | £250,000 | £300,000 | £500,000 | 3 percentage points |
| 1 October 2021–22 September 2022 | £125,000 | £300,000 | £500,000 | 3 percentage points |
| 23 September 2022–30 October 2024 | £250,000 | £425,000 | £625,000 | 3 percentage points |
| 31 October 2024–31 March 2025 | £250,000 | £425,000 | £625,000 | 5 percentage points |
| From 1 April 2025 | £125,000 | £300,000 | £500,000 | 5 percentage points |
From 8 July 2020 to 30 June 2021, buyers pay nothing on the first £500,000, then 5% up to £925,000, 10% up to £1.5 million and 12% above that. Parliament omitted first-time buyers’ relief over those dates, so first-time buyers pay the same rates as other buyers. From 1 July to 30 September 2021, buyers pay nothing on the first £250,000, and first-time buyers can claim their relief again.
Above the temporary £250,000 standard nil band, buyers pay 5% up to £925,000, 10% up to £1.5 million and 12% above that. From 1 April 2025 buyers pay 2% again from £125,000 to £250,000, and the same rates above £250,000. First-time buyers pay 5% above their nil band up to and including the maximum purchase price. A buyer who pays more than that maximum loses first-time buyers’ relief on the whole transaction.
A buyer of an additional dwelling pays the higher rates from the first pound once the price reaches £40,000, and the model charges nothing on a cheaper additional purchase. During both 2020–21 relief periods and from 23 September 2022 to 30 October 2024, that buyer pays 3% up to the standard nil-band threshold, then 8%, 13% and 15% in the higher bands. From 31 October 2024 Parliament raised those rates to 5%, 10%, 15% and 17% in the same bands. From 1 April 2025 that buyer pays 5% up to £125,000, 7% up to £250,000, then 10%, 15% and 17%.
On a residential lease, the tenant pays 1% of the rent’s net present value above a nil band of:
£500,000 from 8 July 2020 to 30 June 2021;
£250,000 from 1 July to 30 September 2021 and from 23 September 2022 to 31 March 2025;
£125,000 at other times.
Sources: SDLT (Temporary Relief) Act 2020, sections 1 and 1A, Finance Act 2021, section 87, SDLT (Temporary Relief) Act 2023, section 1, Finance Act 2025, sections 50–52, Finance Act 2003, Schedule 5 and Finance Act 2003, Schedule 6ZA. Parliament extended the section 1 relief period to 30 June 2021 and inserted section 1A through Finance Act 2021 section 87.
Annual simulation convention and limits¶
The model stores each parameter change on its legal effective date, and for an annual simulation it reads each parameter’s value on 30 April of the model year. For model year 2020 it uses the standard bands, because 30 April 2020 falls before the relief began on 8 July 2020. For model year 2021 it applies the £500,000 nil band without first-time buyers’ relief to the whole year, including purchases in the £250,000 period from 1 July 2021 and under the standard bands from 1 October 2021. For model year 2022 it uses the bands in force before 23 September 2022, for 2023 and 2024 the temporary bands with the three-point surcharge, and for 2025 the restored bands with the five-point surcharge. A user enters each purchase for a whole year, so the model cannot place it on a transaction date within the year.
Under Finance Act 2025 sections 50 and 51, a buyer who entered into a contract before 31 October 2024 keeps the earlier higher rates. A buyer loses that protection after a variation or assignment of rights from that date, if the purchase follows an option or similar right exercised from that date, or if a subsale or other transaction from that date gives someone else the right to call for conveyance. In section 52 Parliament added rules for contracts that the parties substantially performed before a rate change. The model has no input for these contract protections, so it treats every purchase as an ordinary unprotected purchase. The model omits some changes before 8 July 2020. It applies the lease rent table to the household’s rent inputs, not to a lease’s net present value. It has no rates for purchases by companies or non-residents, and it does not round the liability.
Read the purchase curves below as the 2023 marginal purchase rates, and the non-residential curves as the model’s current non-residential purchase and rent tables. Read the residential rent curve as the standard lease rent table, 1% above £125,000; in 2023 the model applies a £250,000 nil band. A first-time buyer who pays more than the maximum loses the relief on the whole price, so read the 2023 main standard curve for that buyer.
Source
import pandas as pd
import plotly.express as px
from policyengine_core.charts import format_fig
residential_rent = pd.DataFrame(
{
"Threshold": [0, 125000, 5200000],
"Rate": [0, 0.01, 0.01],
"Label": ["Residential rent"] * 3,
}
)
residential_purchase_main_first = pd.DataFrame(
{
"Threshold": [0, 425000, 625000, 925000, 1500000, 5200000],
"Rate": [0, 0.05, 0.05, 0.1, 0.12, 0.12],
"Label": ["2023 residential purchase main (first home)"] * 6,
}
)
residential_purchase_main_standard = pd.DataFrame(
{
"Threshold": [0, 250000, 925000, 1500000, 5200000],
"Rate": [0, 0.05, 0.1, 0.12, 0.12],
"Label": ["2023 residential purchase main standard"] * 5,
}
)
residential_purchase_additional = pd.DataFrame(
{
"Threshold": [0, 40000, 250000, 925000, 1500000, 5200000],
"Rate": [0, 0.03, 0.08, 0.13, 0.15, 0.15],
"Label": ["2023 residential purchase additional"] * 6,
}
)
non_residential_rent = pd.DataFrame(
{
"Threshold": [0, 150000, 5000000, 5200000],
"Rate": [0, 0.01, 0.02, 0.02],
"Label": ["Non-residential rent"] * 4,
}
)
non_residential_purchase = pd.DataFrame(
{
"Threshold": [0, 150000, 250000, 5200000],
"Rate": [0, 0.02, 0.05, 0.05],
"Label": ["Non-residential purchase"] * 4,
}
)
df_list = [
residential_rent,
residential_purchase_main_first,
residential_purchase_main_standard,
residential_purchase_additional,
non_residential_rent,
non_residential_purchase,
]
df = pd.concat(df_list)
# plot
fig = (
px.line(
df,
x="Threshold",
y="Rate",
color="Label",
title="Stamp duty land tax rates over property price thresholds",
)
.update_layout(
yaxis_tickformat=",.0%",
xaxis_tickprefix="£",
legend=dict(
orientation="h",
yanchor="bottom",
y=-0.3,
xanchor="right",
x=0.9,
title="",
),
)
.update_traces(line_shape="hv")
)
fig = format_fig(fig)
figInclusive cap and loss of relief¶
In the 2023 annual simulation, an otherwise qualifying first-time buyer pays £10,000 on a £625,000 purchase: (£625,000 − £425,000) × 5%. At £625,020 that buyer exceeds the inclusive maximum and pays ordinary SDLT of £18,751: (£625,020 − £250,000) × 5%. Assume the buyer makes a simple unprotected purchase and meets all other relief conditions.
Source
from policyengine_uk import Simulation
prices = [625000, 625020]
situation = {"people": {}, "benunits": {}, "households": {}}
for i, price in enumerate(prices):
person = f"buyer{i}"
situation["people"][person] = {"age": {2023: 30}}
situation["benunits"][f"family{i}"] = {"members": [person]}
situation["households"][f"home{i}"] = {
"members": [person],
"region": {2023: "LONDON"},
"main_residential_property_purchased": {2023: price},
"main_residential_property_purchased_is_first_home": {2023: True},
}
sim = Simulation(situation=situation)
pd.DataFrame(
{
"Purchase price": prices,
"SDLT": sim.calculate("stamp_duty_land_tax", 2023),
}
)SDLT Statistics for 2019 and 2020¶
HM Revenue & Customs has collected statistics for revenues generated from Stamp Duty Land Tax (SDLT). The SDLT statistics were collected separately for residential and non-residental properties, and for each property type, HMRC divided the data into two categories: corporate and households. The graph below shows SDLT statistics for 2019 and 2020:
Source
from policyengine_uk.system import system
from policyengine_core.charts import format_fig
sdlt = system.parameters.gov.hmrc.stamp_duty.statistics
stats = {"Year": [], "Revenue (£m)": [], "Type": []}
# residential corporate
for parameter in sdlt.residential.corporate.revenue.values_list:
stats["Year"].append(parameter.instant_str[:4])
stats["Revenue (£m)"].append(parameter.value / 1000000)
stats["Type"].append("Residental corporate")
# residential household
for parameter in sdlt.residential.household.revenue.values_list:
stats["Year"].append(parameter.instant_str[:4])
stats["Revenue (£m)"].append(parameter.value / 1000000)
stats["Type"].append("Residental household")
# non-residential corporate
for parameter in sdlt.non_residential.corporate.revenue.values_list:
stats["Year"].append(parameter.instant_str[:4])
stats["Revenue (£m)"].append(parameter.value / 1000000)
stats["Type"].append("Non-residental corporate")
# non-residential household
for parameter in sdlt.non_residential.household.revenue.values_list:
stats["Year"].append(parameter.instant_str[:4])
stats["Revenue (£m)"].append(parameter.value / 1000000)
stats["Type"].append("Non-residental household")
df = pd.DataFrame(stats)
df.sort_values("Year", inplace=True)
fig = (
px.bar(
df,
x="Type",
y="Revenue (£m)",
color="Year",
barmode="group",
text_auto=True,
title="SDLT statistics for 2019 and 2020",
height=500,
)
.update_traces(textposition="outside")
.update_layout(xaxis_title="Property type")
)
fig = format_fig(fig)
fig