Report on the revised forecast of States income for Spring 2021

STATES OF JERSEY

REPORT ON THE REVISED FORECAST OF STATES INCOME FOR SPRING 2021

Presented to the States on 21st September 2021 by the Minister for Treasury and Resources

STATES GREFFE

2021  R.151

Income Forecasting Group

Report on the revised forecast of States income from taxation and duties for Spring 2021

Contents

1. Executive Summary....................................................................................................... 2
2. Uncertainties around the forecast ............................................................................... 3
3. Base case economic assumptions .............................................................................. 4
4. Summary of forecasts ................................................................................................... 5
5. IFG central forecast for Spring 2021 ............................................................................ 7
6. Range of forecast .......................................................................................................... 9 Appendix A – Personal income tax forecast .................................................................... 10 Appendix B – Corporate income tax forecast .................................................................. 35 Appendix C – GST forecast ................................................................................................ 42 Appendix D – Impôts forecast ........................................................................................... 49 Appendix E – Stamp duty forecast .................................................................................... 54 Appendix F – Other income forecast ................................................................................ 59 Appendix G – Terms of reference ..................................................................................... 67

1. Executive Summary

1. The Income Forecasting Group (IFG) has increased each year of the forecast when compared to the autumn 2020 forecast.[1] In total, the forecast for general tax revenue in 2021 has increased by £77.6m, or 9.8%. Over half of this increase is due to improvements in the forecast for income tax.

Figure 1 – Year 2021 forecast variation compared to autumn 2020

Year 2021 £77.6m Forecast variation compared to autumn 2020 £'000

Other Income, £5,570 , 7%

Impots, £5,666 , 7%

Stamp Duty, £12,291 , 16%

Income Tax, £41,000 , 53%

GST,  £13,090 , 17%

2. The IFG's forecast has been informed by the updated economic forecast[2] produced by the independent Fiscal Policy Panel (FPP) in April 2021. The updated forecast suggests a more gradual economic recovery than previously, however this is against a backdrop of considerable uncertainty.
3. The IFG has considered 2020 data outturn compared to the adjustments made to the autumn 2020 forecast in respect of anticipated Covid-19 restrictions. These have been reviewed with further individual adjustments made for the Spring 2021 forecast. These are detailed in each of the respective component reports appended.

4. The Spring 2021 forecast (so called because it is based on the FPP assumptions of April 2021) has been developed as a central forecast' to represent the IFG's view of the most likely outcome. In view of the uncertainties around the forecast it should be considered within a range which is described in section 5.

2. Uncertainties around the forecast

1. Significant uncertainty remains around the IFG forecast, and emphasis should be on the illustrative range presented by the IFG, rather than solely on the central forecast.
2. Much of the risk to the previous forecast was around the uncertainty over the likely impact of the Covid-19 pandemic and the public health measures put in place in response. However, considerably more data is now available on the economic impacts and on the impact on tax. Therefore, IFG's view is that the level of uncertainty on the forecast has reduced and is now more symmetric. The remaining uncertainty regarding the Covid-19 pandemic is primarily around how quickly the economy will recover, and the impact of any ongoing restrictions in 2021.
3. Similarly, more is now known about the UK and Jersey's trading relationship with the EU as the Brexit transition period came to an end in December 2020. The remaining uncertainty is therefore around how much impact this might have on Jersey's economic prospects in the medium term.
4. The financial services sector contributes a large part of Jersey tax take, either directly or indirectly. The sector has proven resilient against a number of challenges including the pandemic, though the impact of low interest rates will act as a drag on banking profitability and therefore corporate tax.
5. There continue to be some uncertainties around external regulation. The OECD Inclusive Framework on Base Erosion and Profit Shifting has agreed the Building Blocks of a proposal to reform international taxation of multinational corporations. The detailed design features of this proposal, together with an implementation plan, should be produced later in the year and this should address the uncertainty somewhat.
6. In the longer term, Jersey's economy faces similar risks to other advanced economies – including the impact of ageing demographics and challenges with low productivity growth. These continue to contribute uncertainty around the income forecast.

3. Base case economic assumptions

1. The FPP economic assumptions have been updated based on the latest local and international developments to April 2021. (Figure 2)
2. The main variations to the economic assumptions used in autumn 2020 reflect the additional releases of local data and include:

• A steeper fall in financial services profits in 2020, reflecting data on banking profits produced by the Jersey Financial Services Commission. However, profits are expected to grow more quickly in the later years of the forecast; as market expectations for interest rates have improved.

• Stronger inflation in the early years of the forecast, and a more robust housing market.

• Stronger growth in house prices.

3. The IFG has considered the economic assumptions from the FPP and have agreed that these assumptions be used as the basis of the income forecast modelling for Spring 2021.

Figure 2 – FPP Revised economic assumptions as at April 2021 (% change except where stated)

Figure 3 – Variations in FPP economic assumptions between October 2020 and April 2021

4. Summary of forecasts

1. The individual forecasts are appended, and these provide further detail of the assumptions and adjustments made to each component of the forecast.
2. Personal income tax (Appendix A) – The forecast has increased for all years of the forecast, from £41m in 2021 decreasing to £31m in 2024. This is primarily driven by the higher than expected tax take for 2019, and data available from the Income Tax Instalment System (ITIS) for 2020. The IFG has reconsidered the adjustments made for the impact of Covid-19, which has increased the forecast for 2020 to 2023. A slight reduction has been made in respect of the updated FPP economic assumptions.
3. Corporate income tax (Appendix B) – The latest FPP economic assumptions suggest a sharper fall in profits from financial services which has decreased each year of the forecast. However, this has been partially offset by the IFG's judgement about an increase in the forecast for tax from property income and large corporate retailers. Overall, the forecast has decreased by £1m in 2021, £4m in 2022 and £3m in 2023 and is unchanged for 2024.
4. GST (Appendix C) – Initial outturn data from 2020 have provided an insight into how the Covid-19 restrictions have translated into receipts. Incorporating this into the assumptions used to estimate the disruption from the pandemic has resulted in an increase in each year of the forecast, from £6.4m in 2021 tapering to £1.8m in 2024. An increase of c.£2.0m has been made in respect of the changes to the FPP economic assumptions and re-estimation of the statistical relationships used to forecast GST excluding ISE fees. An additional £3.7m annual increase has been included in respect of the change to ISE fees agreed in the Government Plan 2021-24.

5. Impôts duty (Appendix D) – Travel restrictions arising from the Covid-19 pandemic reduced the availability of duty-free purchases which translated to an increase in impôts duty. However restrictions arising from Covid-19 led to a decrease in excise receipts from fuel. The adjustments for Covid-19 have been reviewed and the latest FPP assumptions for RPI have been incorporated into the forecast, which has increased by 8% (£5.7m) in 2021 but remains consistent with the autumn 2020 forecast for the other years of the forecast.
6. Stamp duty (Appendix E) – The property market in 2020 proved stronger than anticipated in autumn 2020, despite Covid-19 restrictions. Together with several high-value transactions expected in 2021, this strong outturn has materially improved the forecast in each year, from c.40% (£12.3m) in 2021 decreasing to c.32% (£10.3m) in 2024.
7. Other Government income (Appendix F) – The outturn in 2020 was c.£6m above forecast, largely in respect of dividend income and currency returns. The additional adjustments made in respect of the anticipated effects of Covid-19 have been reconsidered and applied to the forecast. Overall the autumn 2020 forecast has been increased in each year of the forecast, from £5.6m in 2021 decreasing to £2.2m in 2024.

IFG central forecast for Spring 2021

Figure 4 – IFG central income tax forecast

5. Range of forecast

1. The central forecast has been prepared based upon the FPP economic assumptions with additional consideration by IFG, as outlined in the separate reports.
2. There are uncertainties that may be expected around the forecast, including those described in section 2. Therefore the IFG advise that the central forecast should be considered within an illustrative range, as shown below (Figure 5).

Figure 5 – Range of Spring 2021 IFG forecast compared to forecasts from autumn 2020 and autumn 2019

£'000 Range of total IFG forecast Spring 2021  1,150,000

 1,100,000

 1,050,000

 1,000,000

 950,000

 900,000

 850,000

 800,000

 750,000

2020 2021 2022 2023 2024 2025

Upper Central Lower Autumn 2020 Autumn 2019

Appendix A – Personal income tax forecast

Personal income tax forecast

The IFG's autumn 2020 forecast has been updated to incorporate the latest outturn tax data and the FPP's April 2021 economic assumptions. This should be considered provisional, as work is still being undertaken within Revenue Jersey to verify the consistency of the data. An update on this work will be provided during the next forecasting round.

The updated personal income tax forecast is summarised in Figure 1. The forecast has increased in all years. Around half of the aggregate increase over 2019-24 is from the 2019 outturn. The other main drivers are stronger than expected earnings growth reported through ITIS in 2020, and changes to the IFG adjustments.

Figure 1: Changes to personal income tax forecast since October 2020

£m 2019 2020 2021 2022 2023 2024 2025 Personal tax

October 2020 forecast 467 434 461 500 534 561

Tax outturn +15 +14 +14 +15 +16 +18

2020 ITIS outturn 0 +7 +8 +8 +8 +8

Economic data/assumptions 0 0 -1 -3 -2 +1

New relationships 0 0 +1 0 +2 +4

Updated HVR forecast  +1 0 0 0 0 0

Changes to IFG adjustments 0 +14 +20 +10 +5 0

July 2021 forecast 483 470 503 530 562 593 628

Change since October 2020 forecast +16 +36 +41 +30 +28 +31

October 2019 forecast 475 497 524 550 576

Change since October 2019 forecast +8 -27 -21 -21 -14

The remainder of the note is set out as follows:

• Section 1 describes how the forecast is carried out.

• Section 2 sets out the new economic assumptions and updates to tax outturn data.

• Section 3 explains the IFG's adjustments to the forecast to account for the COVID-19 crisis.

• Section 4 sets out the forecast.

• Annex 1 sets out in detail some improvements made to the statistical equations used in the forecast for key taxable income types.

1. How the forecast is carried out

An overview of the personal income tax forecasting model is shown in Figure 2. There are two main elements – forecasting taxable income and then forecasting the likely average effective tax rate (i.e. tax collectable per £1 of taxable income) based on forecasts of the value of deductions (including exemption thresholds for marginal rate taxpayers, and the various reliefs, credits and allowances claimed by taxpayers). The forecast of tax collectable is therefore the product of the forecasts for taxable income and the average effective tax rate.

Taxable personal income is estimated over the forecast period by taking outturn data provided by Revenue Jersey and projecting it forward, largely using statistical relationships between taxable income and various economic variables. The economic variables include compensation of employees (CoE), company profits, employment, average earnings, inflation and interest rates. Forecasts of these variables are overseen by the independent Fiscal Policy Panel (FPP).

The average effective tax rate is then forecast by taking the baseline data for the value of deductions and forecasting changes in these in line with assumptions about future taxpayer numbers, inflation, interest rates and policy changes announced in previous Budgets and Government Plans. So, for example, the aggregate value of the basic exemption thresholds might be assumed to rise in line with the lower of RPI inflation and earnings (to represent the anticipated annual increase in the threshold), and employment growth (to represent the increase in taxpayer numbers claiming this threshold).

Figure 2: Model overview

BASELINE FORECASTS

• statistical relationship

Income Forecast income

• economic assumptions

minus minus

• taxpayer number assumptions

Exemptions, reliefs and allowances . Forecast reliefs

• known and future policy assumptions

• economic assumptions

multiplied by multiplied by

Tax rates Tax rates

= =

Net tax collectable Forecast net tax collectable

2. New economic assumptions and updated tax data

Updated economic assumptions

The Fiscal Policy Panel's (FPP) updated economic assumptions have been used in the model to update the income tax forecast. The economic assumptions were published in April 2021. The economic assumptions letter can be found on the FPP web-site www.gov.je/fiscalpolicypanel

Updated information from Revenue Jersey Personal tax assessments

Outturn tax data now include the 2018 and 2019 years of assessments. However, considerable care needs to be taken over interpreting these two years. 2018 was the first year in which the new Revenue Management System (RMS) was in place; and 2019 was the first year for online filing. The impact of both means there is potentially a structural break in some of the data[1].

The outturn for 2018 was around £7m higher than at the time of the IFG's last forecast, while the outturn for 2019 has come in around £16m higher than forecast. There have been some changes to the structure of the underlying dataset that mean it is not possible to fully disaggregate the difference between the forecast and the outturn. For example, RMS categorises taxed at source income according to the income type - rather than categorising it separately as in ITAX. Further, RMS makes it possible to split out income from trusts and estates for example – which can now be forecast separately to individuals. Therefore, the growth rates below are not entirely comparable, and there may be a structural break over the period for which the growth rates have been calculated.

Further, the effective tax rate was forecast to remain at 13.7% but the new data show it has grown from 13.5% in 2017 to 14.2% in 2019.

On the face of it, taxable income grew more slowly than forecast but the effective tax rate was higher than forecast. However, this is susceptible to any structural break in the data due to either the move to RMS or the move to online filing. It is possible that some income that was reported in ITAX no longer appears in RMS – meaning income (and importantly income growth) is lower but the effective tax rate is higher.

Taxpayers on the high-value residency regime

Assessments for taxpayers on the high-value residency (HVR) regime show that around £23m was assessed for these taxpayers in 2019. Tax from these taxpayers is forecast separately, and the outturn of £23m has been incorporated into the forecast as a new base. This is around £½m higher than forecast.

Other entities

Revenue Jersey has improved the data set by separating out other entities that appear in the personal income tax total. This includes tax from clubs/associations, estates, and pension schemes. This tax amounted to a net £2.8m in 2019. The forecast assumes this amount remains flat in nominal terms.

ITIS data

Revenue Jersey has provided a revised figure for growth in employment income reported through the Income Tax Instalment System (ITIS) for 2020. This indicates that employment income grew by 1.1 per cent in 2020.

The 1.1 per cent for the full year is the combination of relatively strong year-on-year growth in quarter 1, followed by a dip in quarter 2 during the lockdown. The second half of the year was relatively flat, with the exception of strong performance in both September and December.

While the annual growth rate is much smaller than in recent years, it is strong compared to expectations for a fall in 2020. For information the average growth over 2016-2019 was over 4%.

Figure 3: Income reported through ITIS in 2020: Growth on the same month in 2019

8% 6% 4% 2% 0% -2% -4% -6%

This has been incorporated into the forecast, rather than the fall in earnings that would be estimated by the equation that underpins the tax forecasting model.

ITIS data for the first five months of 2021 are available, but recent months generally should not be considered complete data due to the lag in receiving employer returns. Earnings reported through ITIS in the first quarter of 2021 were 4.0% higher than earnings reported in the first quarter of 2020. However, this includes some payments to HVRs that are considerably higher than 2020 – after stripping these out earnings reported through ITIS in quarter 1 of 2021 is relatively flat compared to the same quarter of 2020. However, it would not be appropriate to extrapolate this across the full year – given the highly volatile nature of earnings in 2020 due to the pandemic.

3. Adjustments to personal income tax forecast due to the COVID-19 crisis

The Income Forecasting Group has made additional adjustments to account for the impact of the global pandemic and the resultant restrictions on economic activity. These adjustments are to the smaller income lines, i.e. income other than employment income. This is because the impact on employment income was already included in the FPP forecast. The approach in normal times' is to forecast many of these smaller income lines to be flat in real terms, or to grow in line with recent averages, and IFG took the view that adjustments were needed to these assumptions to reflect the impact of the pandemic not only on the labour market but on other income streams.

These adjustments are based on the judgement of the IFG, using available local and international evidence and their knowledge of the Jersey tax base.

The IFG discussed and agreed changes to their adjustments when they met in April 2021. These have been incorporated into the new forecast:

• The adjustments previously made to pension income and to the average effective tax rate have been removed from the forecast.

• The adjustment to business income is unchanged.

• The adjustments to bank, dividend and other unearned income and to distributions continues to assume that these income types fall by 20% in 2020.

• The adjustment to property income has been amended so that property income is flat in 2020 (rather than the fall previously assumed).

In total, these four types of taxable income represent less than 20% of total taxable income. They have been adjusted using the IFG's judgement as the forecasts are not otherwise directly related to economic performance reflected in the FPP forecast. The exception is bank, dividend and other income where IFG has considered the temporary impact on dividends and other investment income to be greater than the impact solely of falling interest rates[2].

The impact of the remaining IFG adjustment remains significantly negative, though less so than in the previous forecast. The net effect is to reduce the forecast by £26m in 2020, £14m in 2021 and £7m in 2022.

1. The downward adjustment to business profits reduces the forecast by £12m in 2020.
2. The downward adjustment to distributions reduces the forecast by £9m in 2020.
3. The downward adjustment to investment income reduces the forecast by £4m in 2020.
4. The downward adjustment to property income reduces the forecast by £2m in 2020.

While the IFG adjustments are significant, they are just one of a number of uncertainties around the impact of the Covid-19 pandemic, or uncertainties more broadly. For example, a variation in earned income of just 1% would result in a variation of £5m-£6m in each year of the forecast.

Figure 4 shows the impact of each of the four remaining adjustments. This results in a forecast of sharp falls in 2020 for investment income, profits and distributions.

Figure 4: Forecasts for taxable income types after IFG adjustments (£m)

600

Forecast 500

400

300

200

100

0

Profit Property income

Investment income Shareholder income / distributions

4. Updated personal income tax forecast

The forecast has increased in all years, with the main drivers being the 2019 tax outturn, ITIS data for 2020 and changes to the IFG's adjustments.

Figure 5: Changes to personal income tax forecast since October 2020

£m 2019 2020 2021 2022 2023 2024 2025 Personal tax

October 2020 forecast 467 434 461 500 534 561

Tax outturn +15 +14 +14 +15 +16 +18

2020 ITIS outturn 0 +7 +8 +8 +8 +8

Economic data/assumptions 0 0 -1 -3 -2 +1

New relationships 0 0 +1 0 +2 +4

Updated HVR forecast  +1 0 0 0 0 0

Changes to IFG adjustments 0 +14 +20 +10 +5 0

July 2021 forecast 483 470 503 530 562 593 628

Change since October 2020 forecast +16 +36 +41 +30 +28 +31

October 2019 forecast 475 497 524 550 576

Change since October 2019 forecast +8 -27 -21 -21 -14

For information, Figure 6 breaks down the drivers of the aggregate change over the six years 2019- 24. Total revenue from personal tax over these six years is now forecast to be £183m (6%) greater than the October 2020 forecast. Around half of this increase is due to the higher than expected tax take for 2019, with the remainder largely due to data available from ITIS for 2020 and changes to the IFG adjustments.

Figure 6: Changes in total revenue 2019-24

A more detailed disaggregation to the changes to the forecast is set out in the remainder of this section.

Tax outturn

As stated in section 2, the outturn for the 2019 year-of-assessment was £16m higher than forecast, including £1m for taxpayers on the HVR regime. The £15m for non-HVR taxpayers is incorporated into the base and therefore recurs in future years. Given that this update includes two new years of tax data (2018 and 2019) and there have been changes to the historical data from ITAX to make it consistent with RMS; the impact of the outturn in future years varies a little. Different types of taxable income are forecast to grow at different rates so if there is a change in the balance between the types of income, the aggregate growth rate will change.

ITIS outturn

As stated in section 2, data from ITIS for 2020 suggest annual growth in earnings of 1.1%. This is considerably higher than the previous forecast for earnings to fall by ½%. Therefore, using growth in ITIS earnings to inform the estimate for growth in assessed earnings adds around £7m-£8m to the forecast in each year.

ITIS data for the first quarter of 2021 suggest relatively flat earnings when compared to the first quarter of 2020. This is significantly lower than the forecast for full-year growth of 2.1%, which is based on the equation used to forecast earnings. However, given the volatility of monthly earnings in 2020, it is not considered appropriate to extrapolate 2021's first quarter growth rates across the full year.

Economic data/assumptions

The FPP's April economic assumptions have been incorporated into the forecast:

• The higher forecast for inflation results in a reduction in the tax forecast of £1m-£2m each year from 2022. This is primarily due to the impact of higher inflation on projections for allowance growth.

• The lower forecast for financial services profits in the early years of the FPP forecast results in a reduction to the tax forecast; as financial services profits is included in the equation to forecast taxable earnings. In later years, the FPP forecast assumes a stronger recovery in financial services profits – due to a change in market expectations for interest rates. The net impact is to reduce the forecast by £2m in 2021 and £1m in 2022 but increase the forecast by £2m in 2023 and £5m in 2024.

• The lower forecast for compensation of employees reduces the forecast by £1m in 2022, £3m in 2023 and £4m in 2024.

• The higher forecast for interest rates increases the forecast by around £1m in 2023 and 2024. This is due to the impact on bank interest, dividend and other unearned income.

• The increased house price forecast results in a small reduction to the forecast, due to a small increase in the projections for mortgage interest tax relief. It is worth noting that mortgage interest tax relief will be phased out by 2026.

New relationships

This relates to changes in the relationships used to forecast individual types of taxable income.

• Taxable earnings are forecast using an equation that estimates the relationship between earnings and the components of GVA. This equation has been re-estimated to include the latest data, and the model has been respecified to use growth rates in earnings reported through ITIS in recent years, rather than assessed earnings. The changes to the equation are explained in Annex 1 and are designed to overcome any potential structural break from the move to RMS and the move to online filing. The new equation increases the tax forecast by £3m in 2021, £4m in 2022, £8m in 2023 and £12m in 2024, when compared to the October 2020 forecast.

• The equations used to forecast pensions and investment income (bank, dividend and other unearned income) have also been re-estimated with the latest tax outturn. Annex 1 sets out some small improvements to the equations, to make them more robust. The new equations results in an decrease to the forecast of around £½m in 2020 and £2m from 2021 onward.

• Business profits are forecast to grow in line with their average growth over the last five years. In the new data set this growth rate is higher and incorporating this adds £0.8m to the forecast in 2020, which gradually increases each year to reach a £6m increase to 2024.

• Property income is forecast to grow in line with average growth over the last ten years. This growth rate is now smaller, and this reduces the forecast by £100k in 2020, gradually increasing to a £1m reduction to 2024.

Updated HVR forecast

Tax from HVRs is forecast separately to the main taxpayer population. This is because the marginal tax rates faced by these taxpayers differ from those for other taxpayers. Therefore, the forecast for HVR tax is based on expectations for the number of HVR taxpayers arriving and departing.

The outturn for 2019 was around £600k higher than forecast, and this has been incorporated into the base. The number of arrivals in 2020 was a little lower than assumed in the forecast, while the full year expectation for 2021 is greater than assumed in the previous forecast. The number of departures in 2020 and expectations for 2021 are also a little higher than assumed in the forecast.

The net effect of these adjustments is relatively small, less than £¼m in each year.

Updated IFG adjustments

The updated IFG adjustments are set out in section 3. The impact of each individual change is set out below:

1. The removal of the adjustment to the average effective tax rate results in an additional £8m in 2020, £17m in 2021 and £9m in 2022.
2. The removal of the pension adjustment results in an additional £4m in 2020, £2m in 2021 and £1m in 2022.
3. The reduction in the adjustment to property income results in an additional £2m in 2020, and £1m in 2021 and 2022.
4. The IFG's view on investment income (bank interest, dividends and other unearned income) is that it should be adjusted to result in a 20% fall in this income in 2020. A small change to the adjustment has been made to ensure this continues to be the effect of the adjustment. This results in an increase to the forecast of around £270k in 2020, £140k in 2021 and £70k in 2022.

Annex 1: Improvements to statistical relationships

The statistical relationships used to forecast taxable income have evolved and adapted over time. Review and improvement are done regularly to ensure that it can be demonstrated that the relationships used accurately explain the past data and therefore maximise the potential for a robust forecast. Currently equations are used to forecast earned income, pensions and investment income. Together these three represent around 85% of aggregate taxable income.

The equations currently used for earnings and pensions were developed by Oxera in 2017. One minor change has been made in recent years – to account for the fact that finance sector bonuses being included in economic data in the calendar year before they are actually paid. The equation for investment income was also tested extensively in 2017 but no change was made.

The three equations currently used are:

1. Growth in earnings is forecast in line with aggregate earnings in the finance and non-finance sectors, and profits in the finance sector.
2. Growth in pensions is forecast in line with average earnings and growth in the over-65 population.
3. Growth in investment income is forecast in line with changes to the Bank of England Bank Rate.

These equations are regularly reviewed with any potential improvements or amendments investigated. However, the recent changes to the way that tax data are structured and compiled (due to the move to RMS) mean that there is a greater than usual need to give consideration to ensuring the equations continue to be as robust as possible. This annex sets out work that has been carried out to improve each of the equations.

Note that the comparisons set out here relate to the monetary impact of using these improved equations, as compared to the alternative of re-estimating the existing equations using the previous structure but with the latest data. This differs from the comparisons set out in the main forecast document, which are the impact of the new equations when compared to the last IFG forecast.

High-level summary

While the three existing equations remain broadly robust and explain a large degree of the past variation in taxable income, the new data series mean that without making any changes the equations would be less robust than before. Improvements have therefore been implemented as follows:

1. The earned income equation has been improved by using data on growth in earnings reported through ITIS from 2016-19, rather than growth in assessed earnings - as there is some uncertainty around the consistency of recent data from assessments.
2. The pension equation has been improved by adjusting for a potential change in the relationship between pension income and the economic variables.
3. The investment income equation has been improved by including a variable to account for the change in interest rates in the previous year – to account for delay in passing on policy interest rate changes.

The small changes to the three equations result in new estimated relationships that are more robust, and therefore they have been incorporated into the forecast. As stated above, reviewing the relationships and making incremental changes is not unusual and would be expected of any econometric model. However, the addition of two new years of data (rather than the usual one year at a time) and the revisions to the historical data (to make them consistent with the output of RMS for 2018 and 2019) mean that there is a greater than normal need to consider improvements.

For context, the new equations would statistically be expected to improve the range around the forecast by around +/-£1.3m in the first year of the forecast. This will increase over future years, as the statistical errors in the equations are compounded.

Earned income equation

As stated above, the earned income equation seeks to explain variations in taxable earnings with respect to variations in financial services profits (FSGOS), financial services aggregate wages (FSCOE) and non-finance sector aggregate wages (NFCOE).

The equation used is of the following form:

= 0 + 1 + 2 + 3 + 4 (1)

+  4 03 08

Where:

earnings is the annual growth rate in taxable earnings 0 is a constant term;

1 is a coefficient representing the effect that the growth rate in financial services gross operating surplus (FSGOS) have on taxable earnings

2 is a coefficient representing the effect that the growth rate in financial services compensation of employees (FSCOE) have on taxable earnings

3 is a coefficient representing the effect that the growth rate in non-finance sectors compensation of employees (NFCOE) have on taxable earnings

3 is a coefficient representing the effect that the growth rate in financial services compensation of employees in the previous year (FSCOE(-1)) have on taxable earnings

4 represents the impact of a dummy variable (D03to08), which is set to 1 for years up to 2008, to account for a structural break in the series[3]

Re-estimating this equation with assessed earnings data from 2004 to 2019 gives the following results. This suggests that the equation explains almost 90% of the past variation in taxable earnings.

Ordinary Least Squares Estimation ********************************************************************************* Dependent variable is EMP_INC

 16 observations used for estimation from 2004 to 2019 ********************************************************************************* Regressor Coefficient Standard Error T-Ratio[Prob]

INPT  0.95674  0.49831  1.9200[.084]

COE_FS  0.20057  0.08245  2.4327[.035]

COE_NF  0.16313  0.15678  1.0405[.323]

GOS_FS  0.07470  0.02747  2.7196[.022] COE_FS(-1)  0.11904  0.06432  1.8508[.094]

D03TO08  2.22130  0.68818  3.2278[.009] ********************************************************************************* R-Squared .89582 R-Bar-Squared .84374

 S.E. of Regression .99583 F-Stat. F(5,10) 17.1982[.000]

 Mean of Dependent Variable 3.7281 S.D. of Dependent Variable 2.5192

 Residual Sum of Squares 9.9168 Equation Log-likelihood -18.8762

 Akaike Info. Criterion -24.8762 Schwarz Bayesian Criterion -27.1939

 DW-statistic 1.8433

The overall fit is reasonable, and all of the variables used are justifiable and have the correct signs. However, the results for two of the individual variables suggest a need for further investigation - the prob' column shows that the coefficient on the variable COE_NF is not significant at a 10% level, meaning there is a greater than 10% probability that it is not statistically significant. The variable COE_FS(-1) is not significant at a 5% level.

Figure 7 compares actual growth rates in assessed earnings (the blue line) with the values that are consistent with the equation. This shows that the equation is a good fit in earlier years but in recent years is less of a close fit.

Figure 7: Existing earnings equation - actual and fitted values

Plot of Actual and Fitted Values of EMP_INC

10 8 6 4 2

0

2004 2008 2012 2016 2019

EMP_INC Fitted

Looking in more detail at the series for assessed earnings from the latest tax data suggest that this has diverged from the series used in earlier forecasts. Figure 8 compares the new (June 2021) and old (August 2020) series for assessed earnings with data on earnings reported by employers through ITIS. The switch to RMS in 2018 and to online filing in 2019 may mean that the 2018 and 2019 growth rates do not reflect the underlying growth in earnings – i.e. they represent a structural break. The divergence in 2016 has not been explained.

Figure 8: Growth rates in old and new series for assessed earnings, and ITIS earnings

10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0%

Old series growth rates ITIS growth rates Latest series growth rates

Therefore, the Economics Unit has considered how to overcome the potential difficulties with recent assessed earnings data. An alternative equation has been developed, which uses growth rates in assessed earnings from 2004 to 2015, then uses growth rates in ITIS earnings from 2016 to 2019. This assumes that growth in ITIS earnings more accurately reflects the underlying growth rate in earnings in recent years, due to potential structural breaks in the assessed earnings data, and therefore is more appropriate to use in order to establish a statistical relationship.

Ordinary Least Squares Estimation ******************************************************************************* Dependent variable is EMPWITIS

 16 observations used for estimation from 2004 to 2019 ******************************************************************************* Regressor Coefficient Standard Error T-Ratio[Prob]

INPT  1.03170   0.36436  2.8316[.018] COE_FS  0.17652   0.06029  2.9281[.015] COE_NF  0.27967   0.11464  2.4395[.035] GOS_FS  0.08581   0.02008  4.2723[.002] COE_FS(-1)  0.15492   0.04703  3.2942[.008] D03TO08  1.45290   0.50320  2.8872[.016]

******************************************************************************* R-Squared .94093 R-Bar-Squared .91139 S.E. of Regression .72816 F-Stat. F(5,10) 31.8558[.000]

 Mean of Dependent Variable 4.0289 S.D. of Dependent Variable 2.4461 Residual Sum of Squares 5.3021 Equation Log-likelihood -13.8671 Akaike Info. Criterion -19.8671 Schwarz Bayesian Criterion -22.1849 DW-statistic 2.0521

This equation explains more than 94% of the past variation in earnings (assessed earnings to 2015 and ITIS earnings from 2016). Further, all the coefficients are significant at least at a 5% level i.e. there is less than 5% probability that they are not significant. The fit in recent years is improved compared to the existing equation.

Figure 9: Alternative earnings equation - actual and fitted values

Plot of Actual and Fitted Values of EMPWITIS

10 8 6 4 2

0

2004 2008 2012 2016 2019

EMPWITIS Fitted

The impact of switching to the new equation would be to increase the personal income tax forecast by £2m in 2021, £5m in 2022, £8m in 2023, £11m in 2024 and £15m in 2025 – when compared to re- estimating the existing equation using the new assessed earnings data.

Figure 10: Comparison of outturn and forecast

a)  Existing equation (re-estimated)  b) Alternative equation

Plot of Actual and Single Equation Static Forecast(s) of EMP_INC

10 8 6 4 2 0

-2

2004 2010 2016 2022 2025

EMP_INC Forecast

Plot of Actual and Single Equation Static Forecast(s) of EMPWITIS

10 8 6 4 2 0

-2

2004 2010 2016 2022 2025

EMPWITIS Forecast

To give an indication of the degree of improvement in the new equation, it can be observed that the standard error' of the alternative equation is 0.728, compared to 0.996 for the re-estimated existing equation. This suggests that there is statistically a 2/3 likelihood that the range around the results of the existing equation represents around +/-£3.2m of tax, whereas the range around the alternative equation represents around +/-£2.3m of tax. This is therefore a significant improvement, which will increase over the forecast horizon by compounding of the errors.

It is proposed that more work is undertaken to establish the appropriate growth rates in past earned income. This will be improved when the default assessment process is completed for 2019. In the meantime, it is proposed to switch to the alternative equation.

Pension equation

IFG has in the past discussed whether the strong growth rates in pension income in recent years may start to ease off as the impact of demographic patterns change. Part of the reason for the strong growth rates was the retirement of baby-boomers', the first of which will have reached the age of 65 in 2011. The impact of these demographic trends will have resulted in strong growth in percentage terms, due to two reasons:

1. These cohorts will be a proportionally large number of people compared to the existing base of retired people.
2. The pensions these people have will be significantly larger than the average pension among the existing base of retired people.

Both these effects will start to fade as the existing population of pension claimants gets larger and the average pension of existing claimants gets bigger. So, while the absolute growth may remain strong, the percentage growth rate may slow.

However, this change is likely to be gradual and spread over a much greater period than the six years of the personal income tax forecast. This means it is difficult to model. The equation developed in 2017 attempted to account for this by considering the growth rate of the over-65 population and earnings growth in the last two years.

In addition to the demographic trends, there will also be an impact from changes in aggregate investment returns over the working lives of those who start receiving a pension. This is also likely to be a gradual effect and is not explicitly captured by the equation.

The results from the existing regression model are set out below, this has been re-estimated with the latest data but has the same specification as the equation used in recent years. This shows that the equation explains almost 77% of the past variation in pension income growth. However, the coefficient for the over65 population growth variable is not significant at a 10% level, i.e. there is a greater than 10% probability that it is not statistically significant.

Ordinary Least Squares Estimation ******************************************************************************* Dependent variable is PEN_INC

 18 observations used for estimation from 2002 to 2019 ******************************************************************************* Regressor Coefficient Standard Error T-Ratio[Prob] INPT   1.3805   1.5455   .89327[.387] EARN   0.8369   0.26683   3.1365[.007] EARN(-1)   0.71048   0.18302   3.8819[.002] OVER65   0.40476   0.37247   1.0867[.296] ******************************************************************************* R-Squared .76807 R-Bar-Squared .71837 S.E. of Regression 1.0345 F-Stat. F(3,14) 15.4545[.000]

 Mean of Dependent Variable 7.2706 S.D. of Dependent Variable 1.9493 Residual Sum of Squares 14.9815 Equation Log-likelihood -23.8889 Akaike Info. Criterion -27.8889 Schwarz Bayesian Criterion -29.6696 DW-statistic 1.3275 *******************************************************************************

Figure 11 looks at how well the equation explains growth rates in individual years. This suggests a relatively good fit up to 2011, with the relationship being less clear from 2012 onward. This may be due to the issues discussed above, that the impact of demographics may be changing over time and that cumulative past investment returns on pension funds may be changing over time. It is not clear what specifically would have changed in 2020 but the statistical analysis suggests this is when the structural break occurred.

Figure 11: Existing pension equation - actual and fitted values

Plot of Actual and Fitted Values of PEN_INC

12 11 10 9 8 7 6 5

4

2002 2007 2012 2017 2019

PEN_INC Fitted

Therefore, an alternative equation has been developed to account for the fact that the relationship since 2012 appears to be different. This has been achieved through including a dummy variable set to 1 for years up to 2012. This equation is more robust, explaining almost 90% of the past variation in pension income growth. Further, the coefficients on all variables are now significant at the 1% level - except for the constant term. It is considered appropriate to continue to include the constant term as to exclude it would be to implicitly assume a constant of zero.

Ordinary Least Squares Estimation ******************************************************************************* Dependent variable is PEN_INC

 18 observations used for estimation from 2002 to 2019 ******************************************************************************* Regressor Coefficient Standard Error T-Ratio[Prob] INPT  -1.0206  1.2346   -.82667[.423] EARN  0.87218   0.18616   4.6851[.000] EARN(-1)  0.56624  0.1326   4.2703[.001] OVER65  1.0948   0.31219   3.5068[.004] D02TO12  1.9851   0.49898   3.9783[.002] ******************************************************************************* R-Squared .89541 R-Bar-Squared .86323 S.E. of Regression .72090 F-Stat. F(4,13) 27.8235[.000]

 Mean of Dependent Variable 7.2706 S.D. of Dependent Variable 1.9493 Residual Sum of Squares 6.7561 Equation Log-likelihood -16.7215 Akaike Info. Criterion -21.7215 Schwarz Bayesian Criterion -23.9475 DW-statistic 2.8345 *******************************************************************************

The positive coefficient on the new dummy variable (D02to12) means that after accounting for growth in earnings and the over65 population, the growth rate in pension income during 2002 to 2012 was faster than from 2013 to 2019.

Figure 12 demonstrates that the new equation still has a reasonably good fit with the data before 2012 but now has a much closer fit with the data from 2012 onward.

Figure 12: Alternative pension equation - actual and fitted values

Plot of Actual and Fitted Values of PEN_INC

12 11 10 9 8 7 6

5

4

2002 2007 2012 2017 2019

PEN_INC Fitted

The impact of switching to the new equation would be to reduce the personal income tax forecast by £1m in 2020 and 2021, £2m in 2022 and 2023, £3m in 2024 and £4m in 2025 – when compared to re-estimating the existing equation with the latest data. It is proposed to switch to this new equation, but again this can be considered further when more verification can be done on the 2019 outturn.

To give an indication of the degree of improvement in the new equation, it can be observed that the standard error' of the alternative equation is 0.721, compared to 1.035 for the existing equation. This suggests that there is statistically a 2/3 likelihood that the range around the results of the existing equation represents around +/-£0.5m of tax, whereas the range around the alternative equation represents around +/-£0.3m of tax. The impact of the difference between the two equations will increase over the forecast horizon by compounding of the errors.

Investment income equation

Investment income is a broad category of income, including all income that does not fit into the earnings, distributions, pensions, property or business income categories. A large element of this is bank interest and dividends, though it is not possible to separate this category into consistent separate series due to changes in the way that certain types of income were recorded in the past.

Growth in this type of income can be quite volatile. Combined with the range of income types included, it can make it more difficult to fit a robust equation to explain the past data. However, an equation has been used for several years that can explain a large part of the variation in past growth rates, using variations in the Bank of England Bank Rate.

Re-estimating this equation with the latest data produces the following results:

Ordinary Least Squares Estimation *******************************************************************************

 Dependent variable is BDO_INC

 18 observations used for estimation from 2002 to 2019

******************************************************************************* Regressor Coefficient Standard Error T-Ratio[Prob] INPT 2.2101 2.6347 .83883[.414] BRPP 12.3587 2.5090 4.9257[.000] ******************************************************************************* R-Squared .60260 R-Bar-Squared .57777 S.E. of Regression 10.8746 F-Stat. F(1,16) 24.2621[.000]

 Mean of Dependent Variable -.79378 S.D. of Dependent Variable 16.7354 Residual Sum of Squares 1892.1 Equation Log-likelihood -67.4365 Akaike Info. Criterion -69.4365 Schwarz Bayesian Criterion -70.3269 DW-statistic 1.9695 *******************************************************************************

This equation explains around 60% of the past variation in investment income. The coefficient on the explanatory variable is statistically significant at a 1% level, the constant is not significant but is included as to exclude it would mean implicitly assuming a constant of zero.

Figure 13 shows that the equation is a relatively good fit in most years. However, while investment income growth is volatile, there have been very few changes in policy interest rates since 2009 i.e. the red line shows a constant rate of growth while actual growth rates have varied as shown by the blue line.

Figure 13: Existing investment income equation: actual and fitted values

Plot of Actual and Fitted Values of BDO_INC

40 20 0 -20 -40 -60

2002 2007 2012 2017 2019

BDO_INC Fitted

Further work has been carried out to identify any potential to develop an equation that fits more closely with the outturn data. An equation has been developed that includes a lagged term on the change in Bank Rate, i.e. the change in Bank Rate in the previous year. This improves the ability of the equation to explain past variations in investment income. This may be due to a lag in deposit- taking institutions in passing on interest rate changes – for example on savings products where the interest rate is fixed in the short term.

The results of fitting the improved equation are below. This shows that the new equation explains almost 75% of the past variation in investment income. The coefficient on the unlagged variable is still significant at a 1% level, and the coefficient on the lagged variable is just outside of a 1% level of significance.

Ordinary Least Squares Estimation *******************************************************************************

 Dependent variable is BDO_INC

 18 observations used for estimation from 2002 to 2019

******************************************************************************* Regressor Coefficient Standard Error T-Ratio[Prob] INPT  3.7776  2.2389   1.6873[.112] BRPP  11.3278  2.0997   5.3950[.000] BRPP(-1)  6.1011  2.0911   2.9176[.011] ******************************************************************************* R-Squared .74648 R-Bar-Squared .71267 S.E. of Regression 8.9707 F-Stat. F(2,15) 22.0829[.000]

 Mean of Dependent Variable -.79378 S.D. of Dependent Variable 16.7354 Residual Sum of Squares 1207.1 Equation Log-likelihood -63.3913 Akaike Info. Criterion -66.3913 Schwarz Bayesian Criterion -67.7268 DW-statistic 1.9137 *******************************************************************************

Figure 14 shows the closer fit to the past data. The fit is improved particular in periods when rates have been changing – but the equation still fails to explain the volatility of investment income during the long period of flat interest rates.

Figure 14: Alternative investment income equation: actual and fitted values

Plot of Actual and Fitted Values of BDO_INC

40 20 0 -20 -40 -60

2002 2007 2012 2017 2019

BDO_INC Fitted

The impact of switching to the new equation would be to increase the personal income tax forecast by £1m in 2020, £½m in 2021 and 2022, £1m in 2023, £2m in 2024 and £3m in 2025 – when compared to re-estimating the existing equation using the latest tax data. It is proposed to switch to this new equation, but again this can be considered further when more verification can be done on the 2019 outturn.

To give an indication of the degree of improvement in the new equation, it can be observed that the standard error' of the alternative equation is 8.971, compared to 10.875 for the existing equation. This suggests that there is statistically a 2/3 likelihood that the range around the results of the existing equation represents around +/-£1.9m of tax, whereas the range around the alternative equation represents around +/-£1.6m of tax. The impact of the difference between the two equations will increase over the forecast horizon by compounding of the errors.

Appendix B – Corporate income tax forecast

Corporate tax forecast

Figure 1 summarises the corporate tax forecast. The forecast has decreased since September, due to a slower recovery in financial services profits – though this is partly offset by an increase in the forecast for tax on property income and large corporate retailers.

Figure 15: Changes to corporate tax forecast since September 2020 forecast

Some columns may not sum due to rounding

2020 outturn from Revenue Jersey

The aggregate outturn for 2020 was largely in line with the IFG forecast. Tax from financial services firms grew strongly, by around 5% in 2020- though this was less than the 10% growth in gross operating surplus (GOS – the national accounts definition of profits) for the sector in 2019. Note that tax is paid one year in arrears, so tax in 2020 relates to profits in 2019.

Tax from property activities (property development and rents) grew by around £½m in 2020, while tax from large corporate retailers (LCRs) fell by around £½m.

Tax from utilities fell by around £1m, as did tax from all other sectors. However, the fall in tax from all other sectors is driven by a reclassification of some taxpayers from corporate income tax to personal income tax.

Overall, corporate tax grew by 1%. However, adjusting for the reclassification of around £1m from corporate tax to personal tax, underlying growth in corporate tax was 2%.

Updated forecast

Financial services

The forecast has been updated with the new FPP economic assumptions, which were published in April 2021. The new forecast has a steeper fall in financial services gross operating surplus (FSGOS – a measure of profits) in 2020, but a strong recovery is now forecast toward the end of the period – reflecting the increase in market expectations for interest rates rises in those later years and the impact this is likely to have on banking profits.

Figure 16: Forecast corporate tax from financial services (£m - budget year)

90 80 70 60 50 40 30 20 10 0

2020 2021 2022 2023 2024 2025

April 2021 forecast September 2020 forecast May 2020 forecast

Note: tax is collected one year in arrears, so tax in 2020 relates to profits in 2019 Large corporate retailers

The September IFG forecast assumed a 50% fall in tax in 2021, based on 2020 profits. A gradual recovery was forecast, but tax revenues in 2024 were forecast to be around 20% lower than 2020. This was consistent with the GST forecast at the time, which was for a permanent reduction in GST due to increased household importation through the internet.

However, two sets of new data are available:

1. GST data for 2020 show that total GST largely held up – outperforming the forecast, which was for a significant fall. GST revenues from retailers in particular held up strongly, with the falls in GST coming mainly from the hospitality sector.
2. New experimental quarterly GVA data published by Statistics Jersey in November showed that profits for the wholesale and retail sector fell by only 1% in the first half of 2020, compared to the first half of 2019. The large increase of 25% in the first quarter largely offsets the fall of 26% in the second quarter. Profits in individual quarters are highly volatile over the four years data are available, even after accounting for seasonal affects[1].

The September forecast of a 50% fall was based on an assumption that around 1/3 of LCRs were in sectors that would not make any profits in 2020, 1/3 were in sectors that would see profits halved and 1/3 were in sectors that would see profits unchanged. This was constructed to reflect the difference between essential retail such as food and medicine when compared to non-essential retailers who were closed for part of 2020 and are likely to be more affected by physical distancing and consumers avoiding public places.

In light of the quarterly GVA data received, it now appears likely that the retail sector as a whole will not see a significant fall in profits for the year. Partly this will reflect that consumers will have substituted on-island hospitality spend and expenditure on holidays/trips outside Jersey with local retail expenditure. This is reflected in sectoral GST data that show GST collected from retail has increased; while GST collected from accommodation and restaurants/bars has fallen. Vehicle Excise Duties for 2020 fell by 22%, suggesting that relatively high-ticket items may have been affected – though this will include a fall in personal imports of vehicles, which will have been much more difficult in 2020 than in previous years.

Therefore, IFG has revised its expected fall in profits for LCRs. Tax is now expected to be flat in 2021 (based on profits in 2020), and in 2022 (based on profits in 2021) as some restrictions continued into the early part of 2021. Thereafter, tax is assumed to grow in line with RPIY inflation. This means that the real-terms fall in tax from LCRs in 2021 and 2022 is not recovered – implying a structural hit of around 3%. This reflects a smaller economy in the long term.

Figure 17: Forecast corporate tax from large corporate retailers (£m - budget year)

8 7 6 5 4 3 2 1 0

2020 2021 2022 2023 2024 2025

May 2020 forecast April 2021 forecast September 2020 forecast

Note: tax is collected one year in arrears, so tax in 2020 relates to profits in 2019

Property

The previous IFG forecast assumed a sharp fall of 9 per cent in 2021, following the forecast for rental income in the personal tax forecast in 2020 – i.e. a fall of 9 per cent. Thereafter the forecast assumed no growth in 2022, and growth in line with inflation for the remainder of the forecast.

The factors driving the base case forecast from May remain valid and both office and retail rents and occupancy will be under pressure throughout the forecast period. Data from the Parish of St Helier / Jersey Business suggest that town centre retail vacancies were 7.5% in the second quarter of 2020, and fell to 6.7% by quarter 4 – comparable data before quarter 2 of 2020 are not available. This contrasts with the experience of the UK where the vacancy rate increased from 12.4% in the second quarter to 13.7% in the fourth quarter. The failure of some UK retail chains in the last twelve months does not appear to have translated into a significant increase in St Helier's overall vacancy rate.

Outside of retail, offices make up the majority of the rest of corporate property income. A survey carried out by D2 Real Estate showed that only 12% of respondents in the Channel Islands expected their occupational requirements to fall, with 25% expecting requirements to rise – 46% anticipated no change (the requirements of the remaining c 17% are under review)[2]. The report from D2 also indicated that rental collection was at 100%, ahead of the UK, and take-up of new property in St Helier was slightly higher in 2020 than in 2019.

Therefore, IFG has revised its forecast for tax from property rental and development for 2021 – based on profits in 2020. Profits are now expected to be largely flat in nominal terms in 2020, and therefore tax is forecast to be flat in 2021. This is in line with the assumption for rental income in the personal tax forecast, which is also 0% in 2020. No change has been made to the approach for the remainder of the forecast – flat in 2022 and growing in line with RPIY inflation thereafter. As with LCR, this suggests a structural hit of around 3% in real terms.

Figure 18: Forecast corporate tax from property rental and development (£m - budget year)

30 25 20 15 10 5 0

2020 2021 2022 2023 2024

May 2020 forecast April 2021 forecast September 2020 forecast

Note: tax is collected one year in arrears, so tax in 2020 relates to profits in 2019

Other corporate tax

Other corporate tax is from utilities, oil, mining and a few other small areas. It represented around 5 per cent of corporate tax revenues in 2019 (£6m) and therefore the overall corporate tax forecast is not particularly sensitive to changes in the growth rate applied to income for these firms.

This forecast assumes that tax from these firms followed the FPP assumption for growth in the overall level of gross operating surplus in the economy (a measure of total profits in finance and non-finance, including the rental income of private households). The FPP's April 2021 forecast for GOS is a larger fall in 2020, but a stronger recovery in later years – when compared to the August 2020 FPP forecast.

Figure 19: Forecast corporate tax from other sectors (£m - budget year)

10 9 8 7 6 5 4 3 2 1 0

2020 2021 2022 2023 2024 2025

May 2020 forecast April 2021 forecast September 2020 forecast

Note: tax is collected one year in arrears, so tax in 2020 relates to profits in 2019

Figure 20: Assumptions used in corporate tax forecast

% growth rate  2020  2021  2022  2023  2024  2025

Financial services  outt4.8urn  -27.5  4.0  8.0  10.0  12.0 Property  2.3  0.0  0.0  2.5  2.4  2.4

Large corporate retailers  -7.5  0.0  0.0  2.5  2.4  2.4 Other  -26.7*  -17.4  7.3  7.2  6.5  6.8

* around half of the fall in other' was due to a reclassification of some tax from corporate to personal. On an underlying basis, tax fell by 11%

Overall forecast

The overall corporate tax forecast has fallen compared to the previous forecast. The FPP forecast for a sharper downturn in banking profits means that the forecast for tax from financial services has fallen, particularly in the early years of the forecast. This is partly offset by the IFG's judgement that tax from other sectors will not see a sharp fall.

Figure 21: Changes to corporate tax forecast since September 2020 forecast

Some columns may not sum due to rounding

Figure 22: Forecast of corporate tax in key sectors (£m)

90 80 70 60 50 40 30 20 10 0

2020 2021 2022 2023 2024 2025

Financial Institution Property Large corporate retailers

Appendix C – GST forecast GST forecast note

This note sets out the Income Forecasting Group's (IFG) forecast of Goods & Services Tax (GST).

The methodology can be divided into two sections: the estimation of "normal" GST receipts according to historic relationships with Compensation of Employees (CoE) totals and the adjustment for the likely  disruption  caused  by  the  Covid-19  pandemic  and  the  consequences  of  the  public  health response.

Base forecast

The IFG uses an equation to forecast GST excluding International Service Entity fees (ISE fees), denoted as GSTx. With an extra year's data to hand (2019), it was possible to re-estimate the equation that seeks to explain growth in GST receipts by growth in CoE and update the forward estimates for GSTx using the Fiscal Policy Panel's (FPP's) assumptions for CoE growth.

******************************************************************************* Dependent variable is GSTX

 10 observations used for estimation from 2010 to 2019 ******************************************************************************* Regressor Coefficient Standard Error T-Ratio[Prob] GSTX(-1) -0.14555 0.039214 -3.7118[.010] COE 0.86242 0.061623 13.9950[.000] GST_RATE 1.2548 0.049313 25.4457[.000] INPT 2.0544 0.39393 5.2150[.002] ******************************************************************************* R-Squared .99795 R-Bar-Squared .99692 S.E. of Regression .013022 F-Stat. F(3,6) 971.8780[.000] *******************************************************************************

Based on the FPP's latest economic assumptions, this relationship results in an estimate of £87.1m for GSTx receipts in 2021. This relationship includes the impact of the pandemic on incomes but does not include any impact of the pandemic on behaviour.

Covid-19 adjustment

To estimate the impact of the pandemic on GST receipts last year, itemised receipt data which provide a split of receipts according to 70 sectoral categories for the years 2018-2020 have been considered.

Whilst the sectoral totals for these receipts are incomplete due to late submissions, it has been assumed that all sectors are equally incomplete and thus the sectoral proportions are useful to inform our understanding of the pandemic's impact. They show substantial changes in three categories for 2020, namely "Accommodation" which saw a 75% fall (£3.3m), "Food and beverage service activities" fell by 27% (£1.5m) and "Retail trade, except of motor vehicles and motorcycles", which rose by 7% (£1.8m). Changes to the first two categories are self-explanatory, an indication of the industry's reliance on visitors and the consequences for hospitality venues of their enforced closure. The retail group is the largest category by a factor of three and will be influenced by a number of factors including purchases made to adjust to longer periods at home for many people, displaced demand due to restrictions on certain activities (e.g. fewer visits to restaurants will lead to higher purchases of unprepared food, a lack of holiday options may lead to higher local leisure spending by residents), the availability of bricks and mortar retail options versus the convenience of online purchasing as well as potential difficulties in importing goods due to Brexit-related complications.

Change in % share of GSTx receipts '20 vs '19

-5% -4% -3% -2% -1% 0% 1% 2% 3% 4% 5%

The pandemic-driven changes influencing the level of GST receipts are a combination of a) a fall in spending by overseas visitors and b) the impact of physical-distancing restrictions and business closures on the spending of residents. Working from the GST numbers that we have for 2020, it is possible to derive an estimate of how these restrictions translate into changes in receipts and apply similar adjustments to the 2021 forecast, according to the expected experience of restrictions and visitor numbers.

Firstly, we need to assign a quantity to account for the degree of disruption to local consumption patterns.  There  are  potentially  a  lot  of  ways  to  apply  a  consistent  measure  of  the  severity  of restrictions affecting the economic activity of residents, but one simple, binary measure is to take the length of time spent under 2-meter physical distancing guidance as a proxy for the intensity of wider restrictions. With four months of 2-meter distancing last year and another three at the beginning of this year, a range of outcomes has been considered, from: no further restrictions this year to a third wave of the virus that results in a similar duration of restrictions as the second wave - four months. This translates into either 75% of the local restrictions in comparison to last year (i.e. the ratio of three months this year compared to four months last year) or 175% (ie. three months until early April, plus four more, divided by last year's four).

To account for the disruptions to the visitor economy, 2020 saw a reduction in passenger numbers from both the airport and harbour to be around 80% (both arrivals and departures) when compared to the average over 2018-19 and thus we can use this figure as a measure of the disruption to the visitor economy. The likely degree of disruption to foreign travel for this year is difficult to predict. The latest data for off-island travel are from May 2021, which showed passenger arrivals and departures were 87% below the average over 2017-2019. However, the peak summer period is key and travel for the remainder of the year remains highly uncertain. Despite the recently announced delays to the island's reconnection plan, restrictions placed on tourists from the Common Travel Area by competitor destinations may benefit the local visitor economy. If this was sufficient to offset the decline in visitors due to uncertainty, route availability and restrictions on incoming passengers, this could mean visitor numbers are similar to normal years (a "high" scenario). A "low" scenario of a disrupted summer and the same effect on visitor numbers as last year is also possible. Between these two (a low of 80% reduction and a high of no reduction on tourist numbers), any number of scenarios in between are possible and thus a 50% level of reduction has been assumed as a central scenario.

By applying these factors proportionally to the 2021 estimate across all categories, according to the change they saw in 2020, we derive the following factors by which to adjust this year's final forecast. It has been assumed that 90% of demand for accommodation comes from visitors whereas visitors make up 20% of spending for restaurants and bars.

The FPP provided a high and low scenario in their latest letter on economic assumptions, which can provide a range of forecasts of GSTx receipts before pandemic adjustments are included.

In the high scenario, the FPP state that "it is assumed that restrictions are eased in line with the reconnection strategy and inbound visitor travel recovers strongly – resulting in a stronger recovery particularly in the hospitality sector" and compensation of employees is forecast to grow by 4.7%. Combining this with the high' assumptions above around the level of restrictions and inbound travel results in a forecast of £88.9m.

In the low scenario, the FPP state "the path of the pandemic is more severe, with a weaker global recovery and a greater degree of restrictions including restrictions on inbound travel. The result is a weaker recovery, particularly in the financial services and hospitality sectors" and compensation of employees is forecast to contract by 1%. Combining this with the low' assumptions above around the level of restrictions and inbound travel results in a forecast of £78.9m.

The second column of the table below shows the impact of the three scenarios for CoE; whereas the third column also includes the three scenarios for restrictions and inbound visitors.

scenario GSTx receipts  GSTx receipts 2021

2021 (£m) as  (£m) post-pandemic per CoE adjustment

low 84.2 78.9 central 87.1 84.3

high 90.0 88.9

In the previous (September 2020) forecast, the 2019 outturn for GST suggested that there was a potential  shift  to  the  relationship  of  CoE  to  GST-liable  expenditure  and  therefore  a  downward adjustment of 1.9% was made to the GST forecast from 2021 to 2025. The outturn for GST receipts in 2019 has now been revised upward £4m, so this adjustment is no longer necessary and has not been included in this forecast.

Further, the September 2020 forecast included an additional 2% downward adjustment for 2022 to reflect the potential for a more cautious consumer in the post-recession period. However, given that the impact of the pandemic on 2020 GST receipts was smaller than anticipated, we have not assumed any ongoing impact for 2022 and beyond, other than a lower level of CoE. Conditioned on the FPP's central assumptions on CoE to 2025, the forecast is as follows:

In constant values, annual GSTx receipts have fallen slightly since 2012. Growth in nominal receipts has been short of inflation in most years since then, with the notable exception of 2015 when Jersey hosted the Island Games. The large increase in 2011 is due to the increase in the GST rate from 3% to 5%. A gradual real-terms fall in GST receipts is assumed to continue over the forecast period with a higher forecast for inflation than GST receipts for each year.

Annual real growth in GST receipts

50% 40% 30% 20% 10% 0% -10% -20%

2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025

outturn forecast

ISE receipts are forecast to remain stable at £12.6m until the end of the forecast period. Additionally, the reduction in the de minimis level is assumed to lift GST receipts by £0.8m in each year from 2021 onward.

In summary, the forecasting process can be described as follows:

•re-estimation of model coefficients with new outturn data Re-estimation of  •re-assessment of potential change to CoE-GST receipts relationship

CoE to GST

relationship

•analysis of changes to receipts by sector for 2020 •isolation of pandemic effects to visitor and local economy

•estimate of likely 2021 pandemic impacts, based on expectation for the paAccndemic ountingefffecort relative level of disruption compared to 2020

•re-assessment of likely ISE receipts and change through reduction in de

minimis level

Changes to

adjustments

Thus the changes to the September forecast are as follows.

• There has been a rise of £13.1m from the previous forecast for 2021, of which £3.7m is an upwards revision to ISE receipt forecasts across the forecast period with the higher annual fees agreed in the 2021-24 Government Plan.

• An increase of £6.4m from the smaller adjustment now made to account for the disruption from the pandemic with the new information showing the impact of restrictions was less than anticipated (the adjustment has been reduced from £9.3m down to £2.9m for 2021). For 2022 following, the tapering downwards adjustments of £5.2m, £3.5m and £1.8m have been removed given the impact of restrictions on GST-liable expenditure was less than anticipated in 2020.

• An increase of £0.6m from the upwards revision to the FPP assumption for CoE in 2021 and the same from the impact of the higher than expected2020 outturn for GSTx receipts.

• An increase of £1.6m due to the revised 2019 outturn for GSTx. Last September's forecast was adjusted downwards to account for an unexpected underestimation from the model which was considered a potentially new relationship of GST to CoE receipts. This has been removed after GSTx receipts for 2019 were revised upwards by £4m.

• The remaining difference is from re-estimating the equation that is used to explain the relationship between compensation of employees and GSTx.

The following chart shows spending liable to GSTx relative to CoE from 2009 until the end of the forecast period. The outturn data show taxable expenditure falling as a proportion of earnings, which could be caused by a number of different factors. With the propensity of those on lower incomes to consume more of their income, it could reflect a shift in the income distribution towards higher paid workers. Similarly, with private sector rents growing faster than overall inflation in recent times, this may reflect a fall in income after housing costs. Perhaps most convincing, however, would be the idea that this fall in GST-liable expenditure is due to a shift to smaller online purchases that are not liable for GST. This third cause will be progressively mitigated as the GST de minimis level falls.

Forecasts for GSTx (spending) relative to compensation of employees (income)

75% 70% 65% 60% 55% 50% 45% 40%

2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025

Impôts forecast

Appendix D – Impôts forecast

Customs and Immigration Service Impôts estimates 2021-2025

1 Introduction

This paper summarises the Impôts yield for Customs and Excise duty for 2020 and the estimates of yield for the years 2021 to 2025.

In calculating these estimates, the RPI figures provided by the Fiscal Policy Panel (FPP) in March 2021 have been used together with measures introduced in the Government Plan 2021-2024.

2  Assumptions and uncertainties

The impact of the coronavirus outbreak continues to bring uncertainty to the forecasting of Impôts duty albeit that the impacts of the pandemic upon the economy, consumption and travel are much better understood. Actual yields for 2020 should now allow a much more accurate forecast for 2021.

The FPP economic assumptions published in April 2021 highlight uncertainty about the future path of the pandemic, both globally and locally, and a range of Impots estimates are therefore provided to take account of the following scenarios:

• The high scenario. For 2021, it is assumed that restrictions are eased in line with the reconnection strategy and inbound visitor travel recovers strongly – resulting in a stronger recovery particularly in the hospitality sector [];

• The low scenario. For 2021, the path of the pandemic is more severe, with a weaker  global  recovery  and  a  greater  degree  of  restrictions  including restrictions on inbound travel. The result is a weaker recovery, particularly in the financial services and hospitality sectors

3  Alcohol

1. The pandemic during 2020 resulted in significant changes to behaviours resulting in the following changes to the quantities of various alcohol commodities as follows:

 Table 1

2. The 2021 forecast is therefore highly contingent on the nature and extent of the Island's reconnection strategy (inbound visitors, Jersey residents travelling abroad, business travel) as well as the level of local restrictions affecting recovery in the hospitality sector.
3. Based upon known effects of the pandemic during 2020 and attributing these to low, central and high scenarios for 2021, does not result in particularly significant overall changes to the forecast as one commodity offsets another

Table 2 – Alcohol Forecast 2021

4  Fuel

1. Excise receipts for fuel fell approximately 12% during 2020 as a result of local pandemic  restrictions.  The  trade  are  predicting  an  approximate  5%  decrease  in consumption during 2021 based upon pandemic restrictions being eased and travel recommencing from May.
2. Table 3 below shows the assumptions that have been made taking into account the length and nature of the coronavirus restrictions and an estimate of the effect on fuel duty revenue.

Impôts forecast

Table 3 – Fuel Forecast 2021

5  Tobacco

1. The 2020 tobacco forecast was repeatedly revised throughout the year in response to changes in consumption and travel as a result of Covid. The August 2020 forecast for IFG provided the following narrative:

Actual excise figures show that dutiable tobacco quantities (in particular cigarettes and hand rolling tobacco) have increased by between 15% and 20%.

It is assessed that the vast majority of this increase is due to the effect of lack of duty- free imports and this is borne out by feedback from the trade who have seen an increase of between 30% and 40% in sales volumes. Supply chain issues, such as long lead-in times for orders, have resulted in a lack of available stock and sales would have otherwise have been even more buoyant.

As borders start to only cautiously open, and with the real possibility that travel disruption and restrictions will continue for the foreseeable future, the excise forecast for tobacco has been revised upward to show a 25% increase in quantities charged to excise duty for 2020. This may prove be a slightly over-cautious estimate but it is believed to be a realistic central scenario for the purpose of this forecast.

2. The forecast was based up a central scenario of a 5 month lockdown plus a 2 month sharp return to normality in September / October (this approach was agreed by IFG) but  of  course  the  reality  was  somewhat  different  with  severe  travel  restrictions remaining  in  place  for  the  remainder  of  the  year.  This,  together  with  a  huge corresponding reduction in business and leisure travel, resulted in a very sharp drop in the number of duty free goods being imported for local consumption which is no doubt the primary reason why tobacco excise receipts for 2020 are so inflated – increased domestic  sales.  Anecdotal  evidence  exists  that  tobacco  consumption  may  have increased during 2020 however this is not thought to be a significant factor in the large increase in excise receipts observed.
3. The original 2020 forecast of £16.9m eventually rose approximately 42% to achieve an actual outturn of £23.9m.

51

4. In 2019 Statistics Jersey, using data arising out of the Jersey Opinions and Lifestyle Survey, tried to assess the proportion of duty-free cigarettes being consumed in Jersey. Their conclusion was that between 20% and 46% of cigarettes consumed in Jersey were duty free. We now have strong evidence for the first time to indicate that the upper end of this estimate appears to be the most accurate.
5. Table 4 below shows the assumptions that have been made taking into account the length and nature of the coronavirus restrictions and an estimate of the effect on tobacco duty revenue.

Table 4 – Tobacco Forecast 2021

6  Other

1. For the purpose of this forecast the lower Vehicle Emissions Duty (VED) scenario for 2011 has been reduced by 21% in line with the actual outturn for 2020.
2. The forecast for Customs duty (CCT) has been increased from £200,000 to £800,000 as a result of increased numbers of goods being subject to Customs tariffs post-Brexit.

7  Impôts Forecast 2021 to 2025

In addition to the measures detailed above as a result of the coronavirus pandemic the below figures include the GP measures of:

• Alcohol  RPI increase for 2022-2025

• Cigarettes  RPI + 5% increase for 2022 - 2023, RPI increase for 2024-2025

• HRT  RPI + 8% increase for 2022 - 2023, RPI increase for 2024-2025

• Fuel  RPI + 2ppl increase for 2022, RPI increase for 2023-2025

Impôts forecast

Table 5 – Impots forecast 2021-2025

53

Appendix E – Stamp duty forecast

1. Summary

1. The property market in 2020 proved resistant to the Covid-19 pandemic, experiencing a shorter than anticipated reduction in housing market activity which resulted in the outturn exceeding the Autumn 2020 forecast by c.£7.9m.
2. The strong outturn for 2020 improves the base for future years of the forecast and is the main reason for the increase against Autumn 2020.
3. The revised FPP economic assumptions (April 2021) (Figure 1) provide a slight (c.£150k) decrease for 2021, however a number of large transactions forecast for the year results in an overall c.40% increase to the forecast compared to Autumn 2020.

2. Outturn for 2020
   1. The outturn of total stamp duty in 2020 was £36.9m which is c.£7.9m higher than the Autumn 2020 forecast of £29.1m.
   2. This increase over forecast was experienced by all components except for the stamp duty relating to the registration of Wills.

4. Transactions Q1 2020

1. The early timing of the forecast restricts the availability of in-year data, however a number of transactions of high value properties has resulted in a strong start to 2021.
2. Intelligence from Locate Jersey suggests several large transactions are likely to generate material amounts of Stamp Duty in 2021, and therefore a further adjustment (+£3m) to the forecast has been made for these

4. Revised FPP Economic Assumptions (April 2021)[1]

1. The housing market in 2020 exceeded expectations, resulting in the outturn figures being higher than previous assumptions, by 4.5% for house prices and 16.2% for housing transactions.
2. Despite the latest economic assumptions having a slight reduction for house prices in the later years of the forecast, compared to August 2020 there is a stronger increase in the assumptions for housing transactions which provides a slight increase to the forecast overall.

Figure 1: FPP Economic Assumptions April 2021

Figure 2: Variation to FPP Economic Assumptions from August 2020

3. The recently released Statistics Jersey House price report for Q4 20209 suggests that whilst the turnover of properties was 10% lower compared with 2019, the overall house price index was 5% higher.
4. The February 2021 UK House Price Index released by the Office for National Statistics (ONS)10 suggests similar increases have occurred in the UK housing market.

5. April 2021 proposed forecast for Stamp Duty

1. The stamp duty forecast has been revised to incorporate the outturn of transactions from 2020 and the updated FPP economic assumptions.
2. Compared to the autumn 2020 forecast, there is a material increase in all years of the forecast, from c.40% in 2021 decreasing to c.32% in 2024.

Transactions under £2m

3. A slight decrease in transactions during the first period of lockdown was balanced by an increase in numbers throughout the remainder of the year.
4. This has resulted in an increase to the base, which provides a £6m increase to this component for 2021
5. The seasonal variation has been reviewed, however due to the restrictions in housing market activity in 2020, it would not be appropriate to include that year within the three- year average. The seasonal variation is therefore based upon the average monthly transactions over the preceding three years.

Transactions over £2m

6. The tapering of Stamp Duty means that property transactions over £2 million are increasingly difficult to forecast, with recent transfers of property producing significant amounts of duty from single transactions.
7. Despite the restrictions in place during 2020, there continued to be a number of higher value property transactions, with the IFG Autumn 2020 forecast being exceeded by c.£1.2m.
8. Intelligence from Locate Jersey suggests that during 2021 there are several significant transactions anticipated, for which a £3m increase has been made to the forecast for 2021.

Land Transaction Tax (LTT

9. The Statistics Jersey House Price Index shows an annual decrease in the transactions of flats, which form the majority of transactions which are subject to LTT.
10. However there was an overall increase in the value of LTT transactions in 2020, with the outturn being £0.5m above the autumn 2020 forecast.
11. As with the transactions of properties under £2m, there was a noticeable decrease from trend during Q2 of 2020. The seasonal variation from the preceding three years has therefore been retained.

Wills and Probate

12. The forecasts for the Stamp Duty on Wills and Probate are both based upon a five-

year average.

13. There was a particularly high outturn in the stamp duty of Wills (c.£1.7m) during 2015 which has not been repeated since. The removal of that year from the five-year average has resulted in a £140k decrease to this component.

Figure 4: Spring 2021 stamp duty central forecast 2021 – 2025

q

6. Upside/Downside scenarios for Stamp Duty

1. The outturn for 2020 has demonstrated the difficulty with forecasting receipts from stamp duty.
2. To present an upside and downside scenario, an adjustment to house prices has been made to reflect the assumptions for earnings in the FPP's high and low scenarios in 2021.
3. No changes have been made to the assumptions for housing transactions or for the transactions of high value properties.

Figure 5: Upside/Downside scenarios for total stamp duty

Appendix F – Other income forecast