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

Report detailing the methodologies used for the devolved tax forecasts.

Chapter 4 - Income Tax Forecasts


This chapter provides an overview of the methodology used to prepare the forecasts of non-savings non-dividend ( NSND) income tax liabilities published in the 2017-18 Draft Budget.

The Scotland Act 2016 provides the Scottish Parliament with the power to set the rates of income tax as well as the thresholds at which these are paid for all NSND [12] income tax paid by Scottish taxpayers, in addition to devolving the power to create new bands. However, certain elements of income tax will continue to be determined by the UK Government. This includes the definition of taxable income, the Personal Allowance, the ability to introduce or change tax reliefs or exemptions and the taxation of income from savings and dividends.

Implications for the Draft Budget

Table 11 provides the forecast of Scottish NSND income tax liabilities based on the income tax policy announced in the Draft Budget for the five year period 2017-18 to 2021-22.

Table 11: Forecast of Scottish NSND Income Tax Liabilities, 2017-18 to 2021-22, £ million

2017‑18 2018‑19 2019‑20 2020‑21 2021‑22
NSND Liabilities 11,829 12,290 12,912 13,647 14,559

NSND income tax is forecast to raise £11.8 billion in Scotland in 2017-18, with liabilities increasing to reach £14.6 billion in 2021-22. NSND liabilities are forecast to grow by 2.6% between 2016-17 and 2017-18, with growth rising to 3.9% in the following year. Thereafter, growth rates are forecast to accelerate as nominal wage growth is projected to return to its long term trend, with NSND liabilities expanding by 6.7% in 2021-22. Chart 10 illustrates the path of Scottish NSND liabilities over the period 2002-03 until 2021-22. The gaps in the data reflect the years for which detailed income tax statistics for Scotland are not currently available.

Over the period 2002-03 to 2010-11, NSND liabilities grew at an average annual rate of 6.7%. Chart 10 also shows that annual growth in NSND income tax liabilities has been quite variable, with growth ranging from 1.3% in 2003-04 to 11.8% in 2005-06. This reflects a myriad of factors, such as variation in economic growth, and hence income tax liabilities, across years, changes in the income distribution and changes in income tax policy.

Chart 10: Scottish NSND Income Tax Liabilities, 2002-03 to 2021-22, £ million [13]

Chart 10: Scottish NSND Income Tax Liabilities, 2002-03 to 2021-22, £ million

Modelling approach


The Scottish Government has developed a process for forecasting income tax that includes multiple building blocks, models and data sources. The steps involved in producing the income tax forecast are summarised in Chart 11 below.

The sections below discuss each of these building blocks in turn. The details and results of the economic forecasts underpinning the income tax forecasts are discussed in Chapter 3.

Chart 11: Schematic Overview of Forecasting NSND Income Tax Liabilities in Scotland

Chart 11: Schematic Overview of Forecasting NSND Income Tax Liabilities in Scotland

The Income Tax Simulation Model: creating the static income tax forecast

The starting point for forecasting income tax in Scotland is to project forward the Scottish tax base, i.e. the total value of all income in Scotland which is subject to NSND income tax. The Scottish NSND tax base is estimated using data on taxable incomes taken from the Survey of Personal Incomes ( SPI). This is the primary resource used by HM Revenue and Customs ( HMRC) and the Office for Budget Responsibility ( OBR) for UK income tax analysis and for estimating the revenue effects of proposed policies. The SPI comprises a detailed sample of over 40,000 anonymised Scottish tax records, weighted to be representative of all Scottish taxpayers. For each record, there is detailed information on sources and level of income, age group, and a range of other relevant variables. The latest available data is for the financial year 2013-14. The SPI data is then rolled forward, using assumptions about future growth in NSND income and the number of taxpayers, to provide forecasts of the Scottish income tax base in future years, as described in detail in the next section. Box 3 provides further detail on the current income distribution in Scotland.

Box 3: The Income Distribution in Scotland

Table 12 illustrates the number of projected taxpayers in Scotland according to the highest marginal tax rate they face in 2016-17. [14] These figures relate to total income tax, i.e. they include tax paid on savings and/or dividends, but the trends are broadly similar for NSND income.

Table 12: Number of Taxpayers in Scotland by Marginal Rate, 2016-17

Number of Taxpayers Basic Rate Higher Rate Additional Rate All Taxpayers
in '000s 2,150 356 18 2,560
as % of Taxpayers 84.0% 13.9% 0.7% 100.0%
as % of 16+ Population 48.0% 8.0% 0.4% 57.2%

The key points are:

  • There are estimated to be 4.5 million adults in Scotland in 2016-17.
  • Over 40% of Scottish adults - around 1.9 million individuals - pay no income tax.
  • The large majority of taxpayers - around 2.2 million adults - only pay the basic rate of income tax.
  • Less than 10% of Scottish adults, or around 360,000 individuals, are subject to the higher 40p rate.

Less than 0.5% of Scottish adults, or around 20,000 individuals, are Additional Rate taxpayers.

Forecasting growth in NSND income

Each source of NSND income, such as income from employment, pensions and property income, is forecast separately. Public and private sector earnings are also forecast separately.

Table 13 shows the level and share of income arising from each source in Scotland in 2013-14, the latest year for which such detailed SPI outturn data is available. Earnings from private sector employment constitute the largest source, accounting for 59% of all NSND income, followed by earnings from the public sector.

Table 13: Level and Share of Different Sources of NSND income in Scotland in 2013-14, £ million

State pensions Non-state pensions Income from employment - public sector Income from employment - private sector Income - other Total
Levels (£m) 4,025 7,161 17,132 41,817 788 70,923
Share of total (%) 5.7 10.1 24.2 59.0 1.1 -

The remainder of this section sets out the growth assumptions for each of these five sources of income which feed into the income tax forecasts.

The State Pension is set for the UK as a whole and, under current UK Government policy, will increase in line with the triple lock. This means that the State Pension will increase each year by the greatest of growth in average earnings, CPI inflation or 2.5%. The Office for Budget Responsibility publish a table of their forecast for the triple lock guarantee in their Economic and Fiscal Outlook [15] and this is used to grow State Pension income in the income tax forecasting model.

There are a range of private pension products available which grow in different ways over time. Moreover, there will be an existing stock of pensions, with that stock growing in the income it delivers from year to year, plus an on-flow and off-flow of new and ceased pensions each year. The SPI shows that private pensions have grown by an average 3.1% per year between 2002-03 and 2013-14. For simplicity and transparency, the income tax model projects this annual growth rate forward across the forecast period.

For 2014-15 and 2015-16, earnings from private and public sector employment are updated using outturn data from the Annual Survey of Hours and Earnings ( ASHE) for annual average pay for employees in the public and private sector respectively.

From 2016-17 onwards, a forecast of earnings growth is needed. In reality, earnings in the public sector will be determined by a range of factors including the public sector pay arrangements of the Scottish and UK Governments and Local Government. The historic SPI data shows that, over the period 2009-10 to 2013-14, during which the UK Government was implementing its fiscal consolidation programme, average employee earnings from public sector employment have grown by 2.2% each year in Scotland. With the UK Government continuing its policy of fiscal consolidation in future years, public sector earnings are therefore assumed in the model to continue growing at their average rate between 2009-10 and 2013-14.

From 2016-17 onwards, the forecast for average earnings from private sector employment in the model is driven by the Scottish Government's forecast of wages. These are for earnings growth in the economy as a whole, i.e. both the public and private sectors. Since public sector wages are forecast separately in the income tax model, as outlined above, a further adjustment is therefore made before the Scottish Government's wage forecasts are applied to private sector earnings.

Property income and other income not shown elsewhere are grouped under the heading ' all other income' and grown in line with the growth in average earnings, consistent with the Scottish Government's economic forecast. The assumptions for each category of NSND income growth are summarised in Table 14.

Table 14: Assumptions for % Growth in NSND Incomes, 2014-15 to 2021-22

NSND Income Year Pensions Employment income All other Income
State pension [16] All other Pensions Private sector Public sector
Outturn 2014-15 2.7 3.1 0.9 1.8 1.3
2015-16 2.5 3.1 2.6 3.3 2.8
Forecast 2016-17 2.9 3.1 2.3 2.2 2.3
2017-18 2.5 3.1 2.3 2.2 2.3
2018-19 2.5 3.1 3.2 2.2 2.9
2019-20 2.7 3.1 4.2 2.2 3.6
2020-21 3.3 3.1 4.8 2.2 4.1
2021-22 3.6 3.1 4.9 2.2 4.1

Forecasting the number of taxpayers

The number of taxpayers is projected separately for different age bands to factor in demographic shifts in the Scottish population. This approach has the advantage of exploiting the fact that average earnings, and hence NSND liabilities, vary significantly with age. According to the SPI dataset, average incomes peak for middle aged taxpayers and are lower for older taxpayers who primarily receive their incomes through pensions. Relatively faster, or slower, growth in the number of taxpayers in an age group with high average incomes, for example, may therefore boost Scotland's revenue raising capacity, and vice versa.

The SPI dataset allows us to split taxpayers into seven different age groups: under 25, 25-34, 35-44, 45-54, 55-64, 65-74 and 75+. For the two oldest groups (65-74 and 75+), the number of taxpayers is projected in line with the latest 2014-based principal population projections for these age categories, which are produced by the ONS. For those of working age (16-64), the projections are consistent with the Scottish Government's economic forecasts of growth in total employment, reflecting the fact that taxpayers of working age tend to be employed. However, the Scottish Government's economic forecasts only provide a projection of the total employment level in Scotland. In order to project trends in employment growth by age, two additional data sources are therefore incorporated into the analysis: the latest 2014-based ONS principal population projections and labour market data on participation and unemployment in Scotland from the Annual Population Survey ( APS).

Chart 12 overleaf presents the ONS population projections by SPI age groups indexed to 2013-14.

The ONS population projections are then combined with forecasts of labour market participation by age group from the Scottish Government's core economic forecast. From this, the number of people in employment in each age group can be calculated for each year, using disaggregated labour market statistics by age group. A final adjustment is made to these projections to ensure that the resulting growth in total employment matches that from the core economic forecast.

Chart 12: ONS Principal Population Projections by SPI Age Groups, 2013 = 100

Chart 12: ONS Principal Population Projections by SPI Age Groups, 2013 = 100

The resulting projections for the growth in the number of taxpayers by age group are summarised in Table 15.

Table 15: Assumptions for Percentage Growth in the Number of Taxpayers, by Age Group

16 - 24 25 - 34 35 - 44 45 - 54 55 - 64 65 - 74 75+ Employment Growth
(%) [17]
Outturn 2014-15 2.6 3.3 -0.6 2.0 3.4 2.5 2.0 2.0
2015-16 -0.1 1.7 -1.4 -0.1 2.1 1.9 1.0 0.4
Forecast 2016-17 -3.2 0.4 -1.3 0.4 3.9 2.1 1.3 0.2
2017-18 -3.0 0.8 -0.3 -0.3 3.3 1.5 1.8 0.3
2018-19 -3.0 0.8 0.2 -1.1 3.2 1.2 2.4 0.2
2019-20 -3.0 1.0 0.8 -1.6 3.1 1.2 2.5 0.3
2020-21 -2.8 0.3 1.0 -2.2 2.8 1.5 2.0 0.0
2021-22 -1.8 0.6 1.7 -1.4 2.6 1.6 2.4 0.5

Tax parameters

The income tax simulation model then combines the forecasts of the Scottish tax base with the assumed tax parameters for a given year, as detailed in Table 16 , to provide a static forecast of NSND liabilities.

Table 16: Tax Parameters, based on the Proposal in the Draft Budget 2017-18

Effective Personal Allowance Basic rate
Basic rate limit Higher rate threshold Higher rate
Personal allowance limit Additional Rate threshold Additional rate
2017-18 11,500 20 31,930 43,430 40 100,000 150,000 45
2018‑19 11,833 20 32,682 44,516 40 100,000 150,000 45
2019‑20 12,167 20 33,462 45,629 40 100,000 150,000 45
2020‑21 12,500 20 34,051 46,551 40 100,000 150,000 45
2021‑22 12,750 20 34,732 47,482 40 100,000 150,000 45

The Draft Budget 2017-18 confirms the Scottish Government's proposal from 22 March 2016 to freeze the Higher Rate Threshold in real terms in 2017-18 and increase it by no more than inflation until 2021-22. The exact level of the Higher Rate Threshold will be set out each year by the Scottish Government at the Draft Budget. For modelling purposes, it is assumed that the Higher Rate Threshold is increased in line with inflation each year, using September CPI inflation outturn data from the ONS and inflation forecasts published by the OBR [18] .

The power to set the Personal Allowance remains with the UK Government. The Autumn Statement confirmed that the UK Government plans to gradually increase the Personal Allowance to £12,500 by the end of the UK Parliament. The proposals published in March stated that the Scottish Government will implement an effective Personal Allowance of £12,750 in 2021-22 by creating a zero rate band to ensure that this is delivered if necessary. In order to model this aspect of the policy, it is assumed that the personal allowance increases in a straight line from its current level to £12,500 by 2020-21 and reaches £12,750 in 2021-22.

Behavioural analysis

The static forecasts generated by the income tax simulation model do not incorporate behavioural change in response to changes in policy. This means that a further adjustment has to be made to the static forecasts. The methodology used for calculating the behavioural response broadly follows the approach adopted by HMRC for UK-wide tax policy changes. It examines each group of taxpayers affected by the policy change separately and applies Taxable Income Elasticities ( TIEs) to estimate the changes in their taxable income as a result of their behaviour changing in response to a policy change. A broad range of TIEs have been employed to capture individuals' responses to a change in their marginal or average tax rate as a result of a policy change. This is discussed in more detail in Box 5 (on page 33).

Box 4: Comparing the OBR and Scottish Government Income Tax Forecasts

The Office for Budget Responsibility ( OBR) also produces forecasts of Scottish NSND liabilities. However, these do not impact upon the forecasts of Scottish income tax liabilities used in the Scottish budget. The OBR and Scottish Government forecasting approaches differ and this section briefly summarises the OBR's methodology and compares the two sets of forecasts.

The OBR employs a 'top down' approach which uses the forecasts of UK-wide NSND liabilities as a starting point and then derives a Scottish 'share' based on the proportion of UK NSND liabilities historically raised in Scotland. The OBR assume that these historic shares will continue at recent levels unless evidence suggests otherwise. For example, adjustments are made to these historic shares if an announced income tax policy is estimated to have a disproportionate impact on Scottish taxpayers or if outturn data points to different trends in the underlying Scottish tax base. [19]

The OBR forecasts published alongside the Autumn Statement 2016 also assume that Scotland matches the UK's income tax policy. Moreover, they do not model the UK Government's commitment to raise the Personal Allowance to £12,500 and the Higher Rate Threshold to £50,000 by the end of the UK Parliament. The OBR instead assume that all tax thresholds and allowances are uprated in line with inflation after 2017-18. More details on the OBR's methodology can be found in its latest forecast report for the devolved taxes. [20]

The Scottish Government takes a 'bottom up' approach, forecasting Scotland-specific trends in employment and earnings growth to build up the forecast of NSND liabilities. In addition, the Scottish Government's forecasts incorporate announced changes to income tax policy across the entire forecast period.

The Scottish Government and OBR forecasts of Scottish NSND income tax liabilities are therefore not directly comparable. They are based on different modelling techniques and make different assumptions about the future path of income tax policy.

Table 17 summarises the OBR's and the Scottish Government's forecasts for Scottish NSND liabilities for the five year period from 2017-18.

Table 17: Comparison of the OBR's and Scottish Government's Forecasts, £ million

2017-18 2018-19 2019-20 2020-21 2021-22
Scottish Government 11,829 12,290 12,912 13,647 14,559
OBR 11,768 12,220 12,770 13,432 14,181
Difference ( SG- OBR) 61 70 142 215 378

Box 5: Accounting for Taxpayer Behaviour in Response to a Change in Scottish Income Tax

Estimates of the behavioural effects are determined by the Taxable Income Elasticity ( TIE) used, the Marginal Retention Rate ( MRR) and the taxable income of the individual. TIEs measure how responsive total taxable income is to a 1% change in the MRR, which is the amount of each additional pound earned and received by the individual after tax. The behavioural effect is calculated as follows:

Behavioural Effect = TIE x (% Change in MRR) x Taxable Income

The revenue implications in terms of tax lost, or gained, is then derived by multiplying the behavioural effect with the average tax rate in that group. Whilst there is a general consensus in the academic literature that TIEs are much higher for those on higher incomes, the impact of behaviour change is uncertain and is influenced by a range of factors.

There are a series of TIEs previously applied by HMRC. However, evidence from historical tax policy changes in the UK may not always provide the most appropriate indication of the behaviour response which can be expected to occur as a result of future income tax policy changes in Scotland.

  • Scottish income tax powers only apply to NSND income. Since this is largely income from employment, there may be fewer opportunities for individuals to artificially minimise their tax liabilities compared to income from savings and dividends. This could reduce the TIE.
  • Behavioural effects between Scottish and the rest of the UK taxpayers may be more significant than estimated in the existing literature because labour mobility between Scotland and the rest of the UK could be larger than between the UK and other countries. This means that TIEs in Scotland could potentially be much higher, particularly for the highest earners.

Given these uncertainties, a set of low-high TIEs have been adopted, as set out in Table 18.

Table 18: TIEs, Response to Changes in the Marginal Rate of Income Tax

Applied TIE Basic Rate Higher Rate Additional Rate
Low 0.015 0.1 0.35
High 0.015 0.1 0.75

These TIEs only apply where a taxpayer sees a change in their marginal rate of tax. However, taxpayers will not only react to a change in their marginal rate but may also respond to a change in their average rate of tax, for example when tax thresholds are changed. Evidence suggests that there is a much lower response to changes in average than marginal tax rate changes. Therefore, lower TIEs were applied to calculate this aspect of the behavioural response.

These TIEs are kept under review as new information on the behaviour of Scottish taxpayers becomes available.

When considering the behavioural response from the Scottish Government's income tax proposal announced in the Draft Budget, the forecasts under the Scottish Government's policy are compared with a 'do nothing' scenario where the key income tax thresholds remain unchanged in real terms. The Personal Allowance is higher under the Draft Budget proposal, thus reducing NSND liabilities whilst the Basic Rate Limit is marginally lower [21] , resulting in a small boost to NSND liabilities. Since the changes to income tax policy are small, relative to the counterfactual, and the impact is concentrated amongst basic rate taxpayers who tend to be less sensitive to policy changes, the behavioural response is estimated to have little impact on the headline numbers.

Off model adjustments to the income tax forecasts

In addition to accounting for behavioural change, a number of further adjustments have been applied to the forecasts of NSND liabilities for Scotland.

i. In its Economic and Fiscal Outlook, the OBR highlighted the growing cost to the public finances of the recent rise in incorporations as more people set themselves up as a company to minimise their tax bill. The OBR expects UK incorporations to rise by 5% per annum over the forecast period, much faster than the 0.4% increase in total employment. It forecasts that this could cut total UK income tax receipts by £3.1 billion in 2021-22, compared with a situation where incorporations increased in line with employment. [22] The rising trend in incorporations therefore implies that relatively more taxpayers are expected to pay tax on dividends and profits rather than employment income which would depress NSND liabilities. A downward adjustment has been made to the final Scottish income tax forecast to account for this revenue risk. It is estimated that incorporations reduce Scottish NSND liabilities by around £200 million in 2021-22.

ii. A deduction is made to the income tax forecasts to reflect the gift aid that charities claim from HMRC on charitable donations made by Scottish taxpayers. This reduces Scottish NSND liabilities by over £100 million each year.

iii. An upwards adjustment has been made to the forecasts to take into account new income tax policies announced at the Autumn Statement which apply in Scotland and are expected to boost liabilities. This includes the removal of tax and National Insurance advantages in salary sacrifice schemes, the reduction in the money purchase annual allowance and the extension of disguised remuneration targeting the self-employed as well as some other smaller measures. The revenue implications of these tax policies in Scotland is taken directly from the OBR's Devolved Taxes Forecast and the tax measures are estimated to boost NSND liabilities by £26 million in 2021-22 [23] .

Final forecast

Taking all the different components together results in the forecast of Scottish income tax liabilities presented in Table 19.

Table 19: Forecast for Scottish NSND Income Tax Liabilities, 2017-18 to 2021-22, £ million

2017‑18 2018‑19 2019‑20 2020‑21 2021‑22
NSND Liabilities 11,829 12,290 12,912 13,647 14,559

NSND income tax is forecast to raise £11.8 billion in Scotland in 2017-18, with liabilities increasing each year, albeit at varying rates, to reach £14.6 billion in 2021-22.


Back to top