Research and analysis

Responsiveness of commercial transactions to Stamp Duty Land Tax

Published 23 November 2021

Applies to England and Northern Ireland

Abstract

This paper evaluates the impact of Stamp Duty Land Tax (SDLT) on commercial transactions. It uses the change in tax, as a result of the policy at Budget 2016 which moved commercial SDLT from a slab basis to a slice basis, to estimate the responsiveness of commercial transactions to a change in SDLT.

Our preferred regression estimates that a 1% change in the Effective Tax Rate (ETR) (tax paid ÷ value of transaction) leads to a 11.7% change in commercial transactions. This is higher than the equivalent semi-elasticities for residential property of 5% to 7% and much higher than the previous commercial semi-elasticities of 5.4%.

Acknowledgements and disclaimer

The work presented in this report would not have been possible without the input from colleagues both in and outside HM Revenue and Customs (HMRC). We are particularly grateful to Dr. Emma Gorman and Prof. Peter Urwin from the University of Westminster. We also thank officials from HM Treasury (HMT), HMRC and the Office for Budget Responsibility (OBR) for their input.

The views expressed in this paper are those of the authors and do not necessarily represent those of HMRC. Any remaining errors are the authors’ own responsibility.

Authors: J Owen: J Fradley and F Claridge

1. Introduction

1.1. Transaction taxes on property are a widely discussed field. However, most discussion and research has focussed on residential property. This paper adds to the literature material by looking at the impact of SDLT on commercial transactions.

1.2. In this paper we look at the responsiveness of commercial transactions to changes in the tax rate. We exploit the change in the tax rate due to the move from a slab system to a slice system at Budget 2016. Using HMRC returns data before and after the change, we estimate that the semi-elasticity for commercial transactions is -11.7. This means a 1% increase in the Effective Tax Rate (ETR) leads to a -11.7% change in commercial transactions.

1.3. The ETR is a measurement for tax paid on a transaction. It is the tax paid divided by value of the transaction as shown in the formula below.

Effective Tax Rate (ETR) = Tax paid ÷ Value of transaction

For example, if a transaction paid £1,000 tax on a transaction of £100,000 the ETR would be 1%.

ETR = £1,000 ÷ £100,000 = 1%.

1.4. All elasticities mentioned in this paper are semi-elasticities showing the percentage change in transactions to a 1% change in the ETR.

1.5. In this paper we refer to commercial transactions for ease of understanding. However technically our evaluation looks at non-residential transactions rather than just commercial transactions. Under SDLT, all property that is not classified as residential is categorised as non-residential and pays SDLT on a separate tax schedule.

Non-residential property includes:

a. commercial property, for example shops or offices

b. property that isn’t suitable to be lived in

c. forests

d. agricultural land that’s part of a working farm or used for agricultural reasons

e. any other land or property that is not part of a dwelling’s garden or grounds

f. 6 or more residential properties bought in a single transaction

g. a ‘mixed’ property is one that has both residential and non-residential elements, for example a flat connected to a shop, doctor’s surgery or office

1.6. This evaluation only looks at the effect on transactions which pay SDLT on the premium. This is the one-off payment made for purchases of freehold, the assignment of an existing lease and for the premium on a new leasehold transaction. SDLT paid on rent on leaseholds (known as the Net Present Value) is outside the scope of this evaluation.

1.7. From HMRC published statistics, commercial SDLT raised £2.5 billion in the financial year 2020 to 2021 compared to £6.1 billion for residential SDLT ^{[footnote 1]}.

2. Background to the policy change

2.1. At Budget 2016 commercial SDLT was changed to a slice system whereby SDLT was chargeable on each portion of the purchase price which falls within each rate band ^{[footnote 2]}. Previously SDLT was chargeable at a single rate which depended on the purchase price, this was referred to as a ‘slab’ system. These changes made commercial SDLT consistent with residential SDLT which undertook a similar reform in 2014. The current and previous rates are shown in table 1 below.

Table 1: Tax rate before and after change

Chargeable consideration	Rate
Up to £150,000	0%
Over £150,000 and up to £250,000	2%
Over £250,000	5%

Purchase price	Rate
Up to £150,000	0%
Over £150,000 and up to £250,000	1%
Over £250,000 and up to £500,000	3%
Over £500,000	4%

This measure decreased the tax paid for transactions below £1.05 million and increased the total tax paid above £1.05 million. This is shown for example transactions within table 2 below.

Table 2: Change in tax paid

Price of transaction	Tax paid under slab regime	Tax paid under slice regime
£150,000	£0	£0
£250,000	£2,500	£2,000
£500,000	£15,000	£14,500
£1,000,000	£40,000	£39,500
£3,000,000	£120,000	£139,500
£5,000,000	£200,000	£239,500

2.2. At the time the costing was assumed to cost £400 million to £600 million per year ^{[footnote 3]}. In Budget 2016 the assumed elasticities for commercial were published by the OBR (August 2017) and shown in table 3 ^{[footnote 4]}.

2.3. The elasticities used in the costing were agreed with the OBR. In the absence of evidence specific to the commercial market, the OBR made judgements about how elasticities in the commercial market were likely to differ from those in the residential market, for which econometric evidence was available.

2.4. The OBR believed that the transaction semi-elasticities should be more elastic than those used for the high end of residential at the time, which were assumed to be -4.0. The OBR’s judgement reflected 3 considerations for commercial transactions:

1. there is more scope for avoidance than on the residential side

2. the type of buyer is probably more price sensitive than on the residential side

3. elasticity assumed for residential reform looks to have been an underestimate

2.5. The results presented in this evaluation suggest that the OBR’s reasoning was correct in its direction, but that the extent to which elasticities in the commercial market exceed those in the residential market is even greater than was assumed.

Table 3: Elasticities used in commercial slab slice change

Commercial elasticities	Year 1	Year 2	Year 3 and steady state
Transactions semi-elasticity	-5.4	-5.2	-5.0

3. Literature review

3.1. As mentioned in the introduction, current research surrounding the elasticity of property transaction taxes appears to be focused solely on the residential market. This is opposed to the commercial market, which is the focus of this paper.

3.2. Whilst papers that discuss the effect of transaction semi-elasticities in response to commercial property taxes are not available, we are able to explore and contrast 3 different areas of research relating to tax policies in the residential market, which provide a basis to compare our results from the commercial market.

3.3. The first area of residential property tax research was undertaken by Best and Kleven (2016), who investigated housing market responses to the Stamp Duty Land Tax (SDLT) holiday between 3 September 2008 and 31 December 2009. This change moved the nil rate band from £125,000 to £175,000. They employed a difference-in-difference approach using data on all property transactions in the UK from 2004 to 2012.

3.4. They conclude that, transaction taxes are highly distortionary across a range of margins, causing large distortions to the price, volume and timing of property transactions.

3.5. The second area of residential property tax research was undertaken by Bolster (2011), who employed both difference‐in‐difference and time‐series regression analysis to estimate the impact of SDLT first time buyers’ relief. This provided temporary SDLT relief for first time buyers of residential property up to £250,000, which came into force on the 25 March 2010.

3.6. Bolster concludes that the tax relief had not at that time had a significant impact on improving affordability for first time buyers, estimating that most of the people who benefitted would have purchased property in the absence of the relief anyway.

3.7. The third area of residential property tax research looked at in this paper was published by the OBR (2017), who published the behavioural elasticities based on the evaluation conducted by HMRC into the change in the residential market from an SDLT ‘slab’ system (where a single tax rate is paid on the entire purchase price) to a ‘slice’ system (where successive bands of the purchase price are taxed at increasing rates). This change was announced and implemented in 2014.

3.8. Residential data from January 2013 to December 2015 was studied using a first difference Ordinary Least Squares (OLS) methodology. The approach compared the change in transactions and change in ETR due to the reforms from one time period to the next for identical price groups, across the price spectrum.

3.9. Best and Kleven (2016) found that temporary transaction tax cuts are an enormously effective form of fiscal stimulus. They found that a temporary elimination of a 1% transaction tax increased housing market transactions by 20% in the short run due to both the timing response (intertemporal substitution by those who would have purchased a house anyway) and the extensive response (house purchases that would not have taken place absent the tax holiday).

3.10. Their findings represent a transaction semi-elasticity of -20.

3.11. Bolster (2011) estimates that first time buyer transactions were between 0‐2 percent higher than they would have been in the absence of the relief after controlling for wider economic and credit conditions.

3.12. The final estimated impact of the relief on first time buyer transactions is however lower at 1 per cent. This is because analysis of transactions across the different price ranges suggests that first time buyers who would have otherwise bought at just below the £125,001 threshold due to the SDLT liability were tending to buy above the threshold after the relief was introduced.

3.13. These findings represent a transaction semi-elasticity of -1. This is a considerably lower figure than Best and Kleven (2016).

3.14. The OBR (2017) published the below semi-elasticities for the residential market.

Table 4: Residential Transaction semi-elasticities, OBR (August 2017)

Transactions	Under £250,000	Between £250,000 and £1,000,000	Above £1,000,000
Year 1	-7	-5	-6
Year 2	-6.5	-4.75	-6
Year 3	-6.5	-4.75	-6

3.15. The OBR residential transaction semi-elasticity for year 1 ranges between -5 and -7, figures which are lower than the Best and Kleven interpretation but higher than Bolster’s finding of -1 for first time buyers.

3.16 Table 5 below summaries the transaction semi-elasticities reported by the existing literature.

Table 5: Residential transaction semi-elasticities

Category	Transaction semi-elasticity
Best and Kleven	-20
Bolster	-1
OBR published residential	-5 to -7

3.17. The reasons for these differences are the nature of the question which is being studied. The papers cover different sectors of the market, separate tax events and in separate years, hence yield different results. For example, Best and Kleven includes a timing effect which increases the elasticity. Bolster’s paper only looks at a sub-section of the residential market.

3.18. Both also only look at certain price bands in the market, £125,000 to £175,000 for Best and Kleven and £125,000 to £250,000 for Bolster.

3.19. In contrast the figures published by the OBR are based on an evaluation that affected the whole residential market. For this reason, the OBR figures provide better estimates to compare to our paper. Secondly the elasticities were based on an evaluation of the equivalent change in policy in the residential market as analysed in this paper.

3.20. Therefore, when comparing our results from this paper to existing literature we will compare against the elasticities the OBR published.

4. Data

4.1. For our data we use SDLT data on commercial transactions. We compare one year prior to the implementation date, 17 March 2016, to the year following the implementation date to measure the change in transactions. All transactions that claimed a relief and with a value over £4 million are removed.

4.2. We remove transactions with a value in excess of £4 million as there are too few of these to divide into price bands with a large enough sample size and narrow enough price band.

4.3. Transactions below the value of £40,000 are also removed as they are not required to submit an SDLT return. All transactions relating to Scotland are removed due to SDLT being devolved to Scotland in April 2015.

4.4. We have broken the SDLT return data down into price bands which increase in width the higher the price due to less transactions occurring at the high end. We have aimed for a minimum of 40 transactions in each price band.

The price bands are as follows:

   a. £10,000 price bands up to £250,000

   b. £25,000 price bands up to £300,000

   c. £50,000 price bands up to £500,000
 
   d. £100,000 price bands up to £1,000,000 

   e. £1,000,000 price bands up to £4,000,000

4.5. We have completed a first difference regression, in order to compare the change in ETR with the change in transactions. We compare one year before the reform with one year following. This approach allows us to measure the change in transactions against the change in the ETR.

4.6. The time period used are identical 28-day periods in the 2015 to 2016 financial year vs 2016 to 2017 financial year. 28-day periods are used to compare periods of equal length with the same number of working days and weekends.

4.7. Each corresponding 28-day period and price band in the 2015 to 2016 and 2016 to 2017 financial years are then joined together to create a single observation for each month and price band combination across 2015 to 2016 and 2016 to 2017. This is so that the change in transactions and the change in ETR can be calculated for each observation, in order to run a first difference regression.

4.8. We have investigated whether some price bands are distortive under the old ‘slab’ system which may have caused bunching just below a threshold due to an increase in SDLT for properties sold just above the threshold. Distortive thresholds were removed from the residential evaluation into the change in the residential market from slab to slice.

4.9. However, we find that for this evaluation bunching around thresholds is not an obvious problem.

This is due to 2 reasons:

• commercial rates are lower than residential rates and therefore the tax saving through bunching is not as big
• the price bands used in the commercial dataset are wider due to fewer transactions

For price bands that are around the previous thresholds, table 6 below shows the change in tax due and the number of transactions in each year.

Table 6: Potential bunching around thresholds

Price band	Midpoint	Tax due before	Tax due after	Change	Transactions 2015 to 2016	Transactions 2016 to 2017
£150,000 to £160,000	£155,000	£1,550	£100	£1,450	1,806	667
£240,000 to £250,000	£245,000	£2,450	£1,900	£550	935	857
£250,000 to £275,000	£262,500	£7,875	£2,625	£5,250	857	1,606
£450,000 to £500,000	£475,000	£14,250	£13,250	£1,000	1,130	1,405
£500,000 to £600,000	£550,000	22,000	£17,000	£5,000	1,370	1,600

4.10. From looking at the breakdown of transactions in each price band the only price band where bunching looks clear to have taken place is in the £250,000 to £275,000 price band. This price band is therefore removed from the dataset, whilst we choose not to remove the others.

4.11. A summary of our final dataset is shown within table 7 below.

Table 7: Summary statistics

Price band	Number of bands	Number of transactions	Average change in transactions	Average % change
£40,000 to £150,000	120	17,996	8	6
£150,000 to £250,000	120	9,286	5	8
£250,000 to £500,000	84	£9,195	14	19
£500,000 to £1,000,000	60	5,222	9	11
£1,000,000+	48	4,423	-15	-5
Total	432	46,122	-	-

5. Methodology

5.1. A first difference regression approach, using the data above, is chosen because it allows us to both measure a change in ETR which is how semi-elasticities with respect to a tax rate are calculated, and it also allows us to difference away several annually invariant variables such as observed seasonality of the property market, the volume of transactions in different parts of the market and the size of bands chosen in different parts of the market.

5.2. This is a similar methodology used for evaluating the response of residential property to the change from a slab schedule to a slice schedule as discussed in the literature review.

5.3. The model used is shown below:

transactions percentage change = i.constant + ETR change + ɛ

5.4. We have used a simple model as we are not trying to explain the overall growth rate in the commercial market but are using the continuous nature of the change in ETR to explain the difference in the growth rate between different price bands.

For example, if all price bands were to grow at 10% our ETR determinant would be insignificant as there would be no difference in the growth rate of different bands despite the change in ETR. However, as the change in ETR differed across the price spectrum the model effectively weights certain price bands, for example those around thresholds which saw a significant change in ETR, more than other price bands.

5.5. We also have several bands with no ETR change or very small change which acts as an untreated group. We would not expect Gross Domestic Product (GDP) to be significant, as the effect of GDP would broadly increase the growth rates of all bands evenly. We test this assumption in the robustness section.

5.6. This regression assumes that the difference in price bands grow at similar rates and this is consistent with how we treat commercial property in our forecast. We also discuss this assumption further in the robustness section of the paper.

6. Results

6.1. We have tested several variations of regressions, all with the same structure as detailed above, which are shown within table 8 below.

Table 8: Summary of results

Category	Full sample (1)	6 months (October to March) (2)	Transactions > £1 million (3)	Transactions <£1 million (4)
ETR Change	-13.569***, (1.846)	-11.662***, (2.741)	-25,304***, (7.506)	-10.106***, (2.172)
Constant	1.307, (1.257)	2.437, (1.866)	-1.789, (3.614)	3.880, (1.524)
Observations	432	216	48	384
R2	0.112	0.078	0.198	0.054
Adjusted R2	0.110	0.074	0.181	0.051
Residual Std Error	21.642, (df = 430)	22.720, (df = 214)	16.107, (df = 46)	-21.985, (df = 382)
F statistic	54.015***, (df = 1; 430)	18.101*** , (df = 1; 214)	11.366,*** (df = 1; 46)	21.645***, (df = 1; 382)

* p<0.1;  ** p<0.05; *** p<0.01

6.2. The transactions percentage change represents the percentage change in transactions from one year before the reform to the year after the reform. The coefficient of ETR change represents the estimate for the transaction semi-elasticity. For example, in the full sample the regression suggests a 1% change in ETR leads to a 13.6% change in commercial transactions.

6.3. However, using regression (1) we were concerned that the impact of Brexit referendum and the uncertainty generated affected the commercial property market during this time and biased the coefficient on the ETR variable upwards. Therefore, we looked at a 6-month period (October to March) which we believe reduces the effect of Brexit on the coefficient. Therefore, we chose the (2) regression as a preferred estimate.

6.4. We also looked at splitting the market for price bands above £1 million or below £1 million, regressions (3) and (4). This appears to show that the upper end of the market was more responsive to the change in tax rate than the lower end. However due to concerns over the impact of Brexit on transactions and small sample size we did not choose these regressions as our preferred approach.

6.5. Our results are higher than the evaluation completed for the change in residential rates as shown in the literature review (OBR, 2017), this suggests that commercial property is more responsive to changes in tax than residential property. This is consistent with the OBR’s judgement for the change in commercial SDLT for Budget 2016, that commercial transactions were more responsive to changes in SDLT than residential, but the extent of that difference appears to be greater than was assumed.

There are several potential reasons for this result such as:

• as most commercial transactions are made by companies, it means that the decision to buy (or not buy) is likely to be driven much more by a profit motive than residential property

• companies are also more likely to be able to choose their locations for investment internationally compared to home movers, which would also make them more responsive to tax changes

7. Robustness checks

7.1. There are several different areas we explored to check our results were robust. We looked at running a placebo regression comparing the 2014 to 2015 to 2015 to 2016 financial years to make sure there was no bias in our results. We also looked at including macroeconomic variables such as GDP and we discuss further uncertainties in this section.

8. Placebo regression

8.1. It is possible that some unobservable market movement might be correlated with the change in ETR and therefore non policy changes are attributed to the policy change. This could bias our results. We have produced a placebo regression for the time period one year earlier as a type of control test.

8.2. We compare transactions in the 2014 to 2015 to 2015 to 2016 financial years, which are both before the reform but treat the 2015 to 2016 financial year as if it was after the reform, to test if there is any statistical significance in the results.

8.3. As the regression below shows the coefficient for ETR change has a large standard error and is insignificant. Therefore, we have more confidence that our evaluation result in the main regression is unlikely to be unbiased by any variables we are not able to control for.

Table 9: Placebo regression

Category	Full sample (1)
ETR Change	-3.923, (2.655)
Constant	10.703***, (1.576)
Observations	432
R2	0.01
Adjusted R2	0.00
Residual Std Error	27.381, (df = 430)
F statistic	2.18**, (df = 1; 430

* p<0.1;  ** p<0.05; *** p<0.01

9. Additional control variables

9.1. We have tested adding additional variables to the regression such as GDP but have been unable to identify any suitable variables that produce statistically significant results. The regression output shows that GDP is found to be statistically insignificant as suggested in the methodology section.

Table 10: Adding GDP to model

Category	Full sample (1)
ETR Change	-13.579***, (1.835)
Constant	3.080, (1.436)
GDP	-139.680
Observations	432
R2	0.124
Adjusted R2	0.120
Residual Std Error	21.51, (df = 430)
F statistic	30.49, (df = 1; 430)

* p<0.1;  ** p<0.05; *** p<0.01

10. Timing and price effects

10.1. Although the policy was implemented the day following the announcement, we have considered whether taxpayers were expecting it due to a similar reform taking place for residential properties in December 2014. This would bias our results upward as transactions before the change would be artificially low before the change and higher after the change.

10.2. However, we do not believe this would have been the case as the changes to commercial SDLT were 3 fiscal events later than the reforms to residential property rates.

10.3. We are unable to control for any biasing of the results that is caused by price responses to the reform. Increases in price (or decreases for properties in excess of £1 million) caused by a lowering of SDLT (/increasing of SDLT) could lead to transactions shifting price bands and therefore increasing (/decreasing) the estimated transaction semi-elasticity.

10.4. We are unable to investigate this effect as SDLT returns in only a snapshot of transactions and not a longitudinal database we’d need to estimate the impact on prices.

11. Conclusion

11.1. In this paper we investigate the responsiveness of commercial transactions to changes in SDLT. As a result of our evaluation, we find that a 1% change in ETR leads to a 11.7% change in commercial transactions. We attempt to control for the impact of Brexit on the evaluation by looking at a time period 3 months after the vote.

11.2. The analysis and conclusions drawn in this paper have been scrutinised by the OBR and the new transaction elasticities have been certified by the OBR’s Budget Responsibility Committee, meaning they will provide the starting point for policy costings of future measures affecting commercial SDLT. The new elasticities are shown within table 11 below.

Table 11: Final commercial transaction elasticities

Commercial elasticities	Year 1	Year 2	Year 3 and steady state
Transactions semi-elasticity	-11.7	-11.3	-10.8
Price semi-elasticity	-2	-2	-2
Total elasticity (rounded 0.5)	-13.5	-13.5	-13.0

11.3. These elasticities are higher than those assumed in the original costing (-5.4 to -5) and a subsequent evaluation of residential elasticities (-7 to -5).

11.4. To our knowledge this paper is the first to look at the effect of SDLT on commercial transactions. It shows that commercial transactions are more responsive than previously thought and more responsive than residential transactions.

References

Best M. & Kleven H., 2016, Housing Market Responses to Transaction Taxes: Evidence from Notches and Stimulus in the UK

Bolster A, 2011, Evaluating the Impact of Stamp Duty Land Tax First Time Buyer’s Relief

HMRC (2021).Quarterly Stamp Duty Statistics retrieved from https://www.gov.uk/government/statistics/quarterly-stamp-duty-statistics

OBR (November 2017), Residential SDLT elasticities, retrieved from < https://obr.uk/download/residential-stamp-duty-land-tax-elasticities-forecast-evaluation-report-october-2017/>

OBR (August 2017), Non-residential SDLT elasticities, retrieved from https://obr.uk/docs/dlm_uploads/SDLT-elasticities.pdf

1. https://www.gov.uk/government/collections/stamp-duties-statistics ↩

2. https://www.gov.uk/government/publications/stamp-duty-land-tax-reform-of-charging-provisions-for-non-residential-property/stamp-duty-land-tax-reform-of-charging-provisions-for-non-residential-property ↩

3. https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/508147/PU1912_Policy_Costings_FINAL3.pdf ↩

4. https://obr.uk/docs/dlm_uploads/SDLT-elasticities.pdf ↩