Publication - Publication

# Scottish Budget - Draft Budget 2017-2018 devolved taxes methodology report

**15 Dec 2016**

Report detailing the methodologies used for the devolved tax forecasts.

58 page PDF

5.1 MB

58 page PDF

5.1 MB

### Chapter 5 - Land and Buildings Transaction Tax

**Introduction**

The Land and Buildings Transaction Tax ( LBTT) was introduced as a fully devolved tax in Scotland from 1 April 2015. LBTT is a tax applied to residential and commercial land and buildings transactions (including commercial purchases and commercial leases) where a chargeable interest is acquired. In 2015-16, LBTT raised a total of £425 million and, so far in 2016-17 (7 months to end-October 2016), it has raised £269 million. ^{ [24] } Different LBTT rates and thresholds apply for residential, non-residential and lease transactions. Draft Budget 2017‑18 does not propose any change to the 2016-17 rates and thresholds which apply to the various elements of LBTT. Therefore the rates and thresholds will continue to be as set out below in Table 20.

**Table 20: Proposed LBTT rates and bands for residential and non-residential property transactions, 2017-18**

Purchase price (Residential transactions) | LBTT Rate | Purchase price (Non-residential transactions) | LBTT Rate | Net present value of rental payments (Non-residential leases) | LBTT Rate |
---|---|---|---|---|---|

Up to £145,000 | 0% | Up to £150,000 | 0% | Up to £150,000 | 0% |

Above £145,000 to £250,000 | 2% | Above £150,000 to £350,000 | 3% | Above £150,000 | 1% |

Above £250,000 to £325,000 | 5% | Above £350,000 | 4.5% | ||

Above £325,000 to £750,000 | 10% | ||||

Above £750,000 | 12% |

In Draft Budget 2016-17, the Scottish Government introduced the Additional Dwelling Supplement ( ADS) on the purchase of additional residential properties, such as buy-to-let properties or second homes from 1 April 2016. This supplement, is levied at 3 percentage points of the total price of the property for all relevant transactions of £40,000 and above, and is levied in addition to the LBTT rates in Table 20. Net ADS revenues in 2016-17 to date are £51 million consisting of £57 million of gross ADS revenue less about £6 million of repayment. ^{ [25] }

The policy intentions of the residential LBTT structure and ADS are related. Residential LBTT is structured in such a way that it most benefits those buying at the lower end of the housing market, which typically includes first-time buyers. ADS was introduced to allow first-time buyers to compete more effectively in the market with (for example) buy-to-let landlords or second home owners. For non-residential LBTT, the structure is again aimed at protecting smaller businesses with the lower rate of 3% charged between £150,000 and £350,000 and no LBTT charged at £150,000 or below. ^{ [26] }

**Revenue forecasts**

Five-year forecasts for the various elements of LBTT as contained in Draft Budget 2017-18 are set out below in Table 21.

**Table 21: Five-year LBTT forecasts for Draft Budget 2017-18 to 2021-22 (£ million)**

2017-18 | 2018-19 | 2019-20 | 2020-21 | 2021-22 | ||
---|---|---|---|---|---|---|

Residential LBTT | Standard | 211 | 235 | 251 | 265 | 280 |

ADS* | 72 | 75 | 78 | 80 | 82 | |

Total | 283 | 310 | 329 | 345 | 362 | |

Non-residential LBTT | 224 | 233 | 242 | 252 | 262 | |

Total LBTT | 507 |
543 |
571 |
597 |
624 |

* ADS forecasts presented on a net basis, i.e. once all repayments have been made ^{ [27] }

In overall terms, LBTT revenues are forecast to rise from £507 million in 2017-18 to £624 million in 2021-22 an increase of 23 per cent. In 2017-18, we forecast that about 42 per cent of total LBTT revenue will come from residential LBTT, about 14 per cent from ADS and the remaining 44 per cent from non-residential LBTT. This distribution is not expected to shift significantly over the 5-year forecast period.

**Methodology**

**Residential LBTT - standard element**

The house-price distribution is modelled using a log-normal distribution, which can be fully characterised by its mean and median. Therefore, forecasts of mean and median house prices are required, as well as the volume of housing transactions.

Autoregressive integrated moving average ( ARIMA) models are used to model each of these three housing-market determinants. ^{ [28] } Mean house prices are modelled using their annual growth rate as the dependent variable, median house prices using the ratio of median to mean house prices, and volumes using the turnover ratio (the proportion of the private housing stock that is sold each period). Forecasts for mean house prices are derived by applying the forecast growth rates to base-year prices, forecasts for median house prices by multiplying forecast mean house price by the forecast median-mean ratio, and forecasts for transactions by using National Records of Scotland household-growth projections to estimate future levels of the private housing stock, and then multiplying these by the forecast turnover ratio.

The log-normal house price distribution is then parameterised using these forecast mean and median house prices. Revenues are calculated by applying the tax rates to this distribution and multiplying by forecast volumes. Since the log-normal distribution does not match the house-price distribution exactly - for example, in 2015‑16 there were somewhat more sales in the highest LBTT band (£750k and above) and somewhat fewer in the band just below this (£350k to £750k) than the log-normal distribution predicted - correction factors (calculated from the 2015-16 data) are applied in order to improve the fit. While these factors increase the projected share in some bands, and reduce them in others, the overall impact on revenues is small. ^{ [29] }

It is important to note that a small degree of over/underprediction by the log-normal distribution for different price ranges was present in the historical data as well, and has not increased since the introduction of LBTT. As Figure 13 illustrates, outside the quarters affected by forestalling behaviour, the share of transactions in the £325k-£750k band is at similar levels in the post- LBTT period to the pre- LBTT period - there is no evidence in the data so far of a sustained underperformance of this segment of the market. ^{ [30] }

**Chart 13 Transactions with a purchase price between £325,000 and £750,000, as a share of total transactions**

**Residential LBTT - ADS**

Robust data on the sale of homes not for owner occupation was unavailable prior to the introduction of ADS on 1 April 2016, and at this stage there is only limited data for the period of its operation. Analysis is further complicated by the fact that the early months are likely to have been affected by forestalling behaviour, and also because the supplement can be reclaimed if the buyer is able to dispose of their previous main residence within an 18-month period, meaning the final value of ADS liabilities relating to transactions in 2016-17 will only be known in October 2018.

The approach followed has therefore been to use data from the July-September quarter to estimate the size of the market, since this excludes the period most likely to have been affected by forestalling. Furthermore, when ADS is paid, the taxpayer can indicate whether they intend to reclaim the supplement. For the period April-September 2016, taxpayers indicated an intention to reclaim 25% of transactions, amounting to 34% of ADS liabilities. Given that the pattern of repayments so far is not inconsistent with these expressed intentions, we consider it prudent to assume for budgeting purposes that repayment will be in line with expressed intentions. This produces an estimate that the transactions permanently liable for ADS constitute 16% of total residential transactions.

With respect to prices, the limited outturn data available, once corrected for those transactions where the taxpayer has reclaimed or intends to reclaim ADS, are consistent with our previous assumption that the mean purchase price for an ADS transaction is approximately 10% lower than for all residential transactions. This assumption is therefore maintained.

In the absence of robust time-series data for this segment of the market, it is assumed that the average price and volume for ADS transactions will grow at the same rate as the overall market. Since the ADS tax rate is a fixed 3%, it is sufficient to use the forecast mean price and volumes to calculate the forecast revenue without having to first fit a house-price distribution.

**Non-residential LBTT**

Historically, trends in non-residential tax receipts in Scotland have closely tracked those of the UK. As Chart 14 shows, this pattern continued in 2015‑16 after the introduction of LBTT, which has a similar rate structure for non-residential transactions to the UK Government's Stamp Duty Land Tax. Accordingly, as in previous forecasting rounds, we have used the latest Office for Budget Responsibility forecasts for commercial price and transactions at the UK level as an input to the non-residential LBTT forecasts.

**Chart 14: Historic non-residential SDLT/ LBTT receipts in Scotland and UK, £ million**

Chart 14 also illustrates the volatility of non-residential revenues, with the volatility in the overall level of transactions exacerbated by the fact that much of the revenue is generated by a small proportion of high-value transactions. Therefore, in order to smooth out some of the volatility in the historical revenue series, multi-year averages are used as a baseline. In particular, the latest three years of non-residential revenue outturns for Scotland, i.e. 2013‑14, 2014-15 and 2015-16, are averaged to form a base from which to project future revenues. In doing this, the first two years of revenue outturns are uplifted by the average increase in non-residential prices. This resultant base, which is expressed in 2015-16 non-residential prices, is then grown by the product of the most recent OBR price and volume forecasts from 2016-17 onwards.

**Behavioural effects**

In Draft Budget 2016‑17, it was assumed that there would be a long-run behavioural response to the introduction of ADS, which would reduce the number of transactions in this segment of the market either because these transactions switched into the owner-occupied market or because they did not take place at all - which in the latter case implied a reduction in revenues from the standard element of residential LBTT as well as reduced ADS revenues. Forecast revenues for both the ADS and standard elements were also adjusted to take into account forestalling behaviour. Since the residential LBTT forecasts set out in the 2017‑18 Draft Budget use the latest available outturn data from 2016-17, it is assumed that these long-run behavioural impacts are already reflected in the baseline, and so no further adjustment is made. Since no change in rates and thresholds is proposed for 2017‑18, there is also no need to adjust forecasts for residential and non-residential LBTT revenues for forestalling behaviour.

**Evolution of the LBTT forecasts**

Table 22 shows how the LBTT forecasts have evolved from Draft Budget 2016-17 to Draft Budget 2017-18. There is a significant downward revision in the standard element of residential LBTT revenues, a significant upward revision in the ADS element relative to its size, while non-residential revenue forecasts are little changed.

**Table 22: Forecast comparison in Draft Budget 2017-18 and Draft Budget 2016-17, £ million**

Draft Budget 2017-18 |
|||||||
---|---|---|---|---|---|---|---|

2016-17 | 2017-18 | 2018-19 | 2019-20 | 2020-21 | 2021-22 | ||

Residential | Standard | 211 | 235 | 251 | 265 | 280 | |

ADS | 72 | 75 | 78 | 80 | 82 | ||

Total | 283 | 310 | 329 | 345 | 362 | ||

Non-residential | 224 | 233 | 242 | 252 | 262 | ||

Total LBTT |
507 |
543 |
571 |
597 |
624 |
||

Draft Budget 2016-17 |
|||||||

Residential ^{ [31] } |
Standard | 282 | 347 | 406 | 469 | 533 | |

ADS | 36 | 51 | 56 | 62 | 66 | ||

Total | 318 | 398 | 462 | 531 | 599 | ||

Non-residential | 220 | 230 | 240 | 250 | 260 | ||

Total LBTT |
538 |
628 |
702 |
781 |
859 |
||

Change |
|||||||

Residential | Standard | -136 | -171 | -218 | -268 | ||

ADS | 21 | 20 | 17 | 15 | |||

Total | -115 | -152 | -202 | -254 | |||

Non-residential | -6 | -7 | -8 | -8 | |||

Total LBTT |
-121 |
-158 |
-209 |
-262 |

**Residential LBTT - standard element**

The downward revision to the standard element of residential LBTT is due to weaker forecasts for both prices and transactions. This is the result of outturn data in 2015‑16 and 2016‑17 being weaker than was forecast at the time of Draft Budget 2016‑17, as well as changes to the forecast methodology, which has been developed in the light of feedback from the Scottish Fiscal Commission. It is important to note that these downward revisions relate to the market as whole, rather than particular segments of the market. ^{ [32] }

Previously, mean house prices were forecast using an ARIMA model for the short run and then smoothing back to an assumed long-run growth rate of 4.5%, comprising 2.5% real house-price growth and an assumption that inflation would be in line with the 2% target. Mean house prices are now forecast using an ARIMA model throughout the five-year period. Suggestions from the Commission also led to further development of the ARIMA model, with structural breaks added so as to divide the period into pre-, during- and post-financial-crisis periods, and dummy variables added to capture the two episodes of forestalling behaviour around the introduction of LBTT and ADS.

Median house prices were previously assumed to grow at the same rate as mean house prices, implying a constant median-mean ratio. Given the considerations raised by the Commission as to the impact that changes in this ratio could have on the house-price distribution, this ratio is now modelled separately, allowing its value to vary over time.

Transactions were previously forecast assuming that their growth rate would fall linearly until the turnover ratio reached its long-run average. Transactions are now modelled using an ARIMA model, using the turnover ratio as the dependent variable so that its long-run value is generated by the ARIMA model itself. As in the case of the ARIMA model for mean house prices, the ARIMA model for transactions includes structural breaks to take account of the impact of the financial crisis.

The forecasts for mean and median prices and the growth rates of transactions are set out below, together with the corresponding forecasts from Draft Budget 2016‑17.

**Table 23: Comparison of forecast growth rates in residential prices and transactions in Draft Budget 2017-18 and Draft Budget 2016-17**

2015-16 | 2016-17 | 2017-18 | 2018-19 | 2019-20 | 2020-21 | 2021-22 | |
---|---|---|---|---|---|---|---|

Draft Budget 2017-18 | |||||||

Mean | -0.5% | -0.1% | 1.8% | 1.4% | 1.3% | 1.3% | 1.3% |

Median | 0.5% | 0.9% | 0.1% | 0.6% | 1.2% | 1.2% | 1.2% |

Transactions | 7.5% | -2.3% | 0.8% | 2.4% | 1.9% | 1.5% | 1.2% |

Draft Budget 2016-17 | |||||||

Mean | 5.9% | 5.6% | 5.3% | 5.1% | 4.8% | 4.5% | |

Median | 5.9% | 5.6% | 5.3% | 5.1% | 4.8% | 4.5% | |

Transactions | 5.4% | 4.6% | 3.8% | 2.9% | 2.1% | 1.3% |

**Residential LBTT - ADS**

Since transactions liable for ADS are forecast to grow at the same rate as the overall residential market, the downward revisions to prices and forecasts described above also reduce forecast ADS revenues. However, this is more than offset by the fact that the outturn data available so far indicate that the number of transactions in this segment of the market is larger than was expected at the time of Draft Budget 2016‑17.

**Non-residential LBTT**

The forecasting approach to non-residential LBTT remains largely the same as in Draft-Budget 2016‑17. The fact that there was only a small downward revision in revenues was as a result of a downward revision to the OBR forecast for price growth being largely offset by an upward revision to their forecast for transactions growth. As a result, the overall growth factor applied to produce the Scottish Government forecasts did not change significantly.

**Table 24: Office for Budget Responsibility forecasts for non-residential price and transactions growth**

2015-16 | 2016-17 | 2017-18 | 2018-19 | 2019-20 | 2020-21 | 2021-22 | |
---|---|---|---|---|---|---|---|

November 2016 OBR Economic and Fiscal Outlook | |||||||

Prices | 11.9% | -5.1% | -3.2% | 2.1% | 1.8% | 1.9% | 2.0% |

Transactions | 4.7% | 6.7% | 1.3% | 1.9% | 2.1% | 2.1% | 2.0% |

November 2015 OBR Economic and Fiscal Outlook | |||||||

Prices | 7.4% | 3.1% | 1.8% | 1.9% | 2.1% | 2.1% | |

Transactions | 3.7% | 2.4% | 2.5% | 2.5% | 2.3% | 2.3% |

**Risks and sensitivities**

**Residential LBTT - standard element**

With respect to both prices and transactions, a significant element of the downward revision is due to the change in methodology described above. In particular, the addition of structural breaks means that the models allow the long-run growth rate in the post-financial crisis period to differ from the earlier periods. This means that the high level of price and transactions growth in the early 2000s housing boom, which was arguably unsustainable, has much less influence on the modelled future long-term average growth rates. In particular, house-price growth is now modelled to converge on a long-run average of 1.3% rather than 4.5%, while the turnover ratio is modelled to converge on 5.1% rather than 6%.

The advantage of new modelling approach is that it uses a consistent approach to all the housing-market determinants, by applying an ARIMA approach to each variable, using data from the same source. It also minimises the amount of subjective judgement applied, since forecasts for all five years are now produced from the same model and same dataset, rather than using one approach for the earlier years of the forecast and then converging, in a somewhat ad hoc manner, on a long-run average calculated from another dataset. A risk, however, is that the downward adjustment in forecast price and transaction growth due to the adoption of this new modelling approach is too large. In particular, the introduction of structural breaks related to the financial crisis implies that the post-crisis period will be different, even in the long run, to the pre-crisis period. On the other hand, it can be argued that the boom years of the early 2000s were characterised by unsustainably high transaction and house-price growth, and it is therefore desirable that the model minimises the impact that data from this period has on long-run forecasts. ^{ [33] }

**Residential LBTT - ADS**

The ADS segment of the forecast is assumed to follow the same trends as the overall market; therefore, sensitivities identified above apply to the ADS forecasts as well. In addition, any change in price and transaction growth relating specifically to this segment of the market could also lead to revenues diverging from forecasts. Another key risk relates to repayments, especially as, with an 18-month window to reclaim repayment, it will take some time to build up a robust profile of repayment behaviour.

**Non-residential LBTT**

As Chart 14 illustrates, non-residential forecasts are particularly volatile, due to the fact that the timing of major business investment decisions can be strongly influenced by economic conditions. Given the uncertainties in the current economic environment, including but not limited to the final form of Brexit, there is a significant risk that any adverse shocks could increase uncertainty and cause investment decisions to be delayed.

**Scottish Fiscal Commission view of the forecast**

The final LBTT forecast, published as part of the Draft Budget 2017-18, is set out in Table 21 above. The Commission has endorsed the final forecast as reasonable. Further details on their views together with additional analysis and commentary are set out in the Commission's Final Report.