Measuring Indirect Tax Losses October 2007

Transcription

1 Measuring Indirect Tax Losses October 2007

2 Contents 1. Introduction 3 2. Estimating VAT Losses 4 Methodology 4 Results 5 MTIC 6 3. Estimating Excise Losses 7 Methodology 7 Alcohol 9 Tobacco 10 Hydrocarbon Oils 12 Annex A: Margins of Error 15 Annex B: Detail for VAT Estimates 17 Annex C: Detail for Alcohol Estimates 22 Annex D: Detail for Tobacco Estimates 28 Annex E: Detail for Hydrocarbon Oils 46 Glossary 55 Abbreviations 56 2

3 1 Introduction 1.1. This technical paper presents the latest annual update of HMRC s estimates of revenue losses in indirect taxes. Previous estimates have been presented in technical papers published alongside each Pre-Budget Report (PBR) since Estimating the scale of, and trends in, revenue losses is not only inherently difficult, but also a relatively untested area of work for governments in the EU and around the world. Nevertheless, HMRC have developed estimates for losses in indirect taxes and excises that they believe are the best possible, based on all the information presently available. However, uncertainty in the estimates means that HMRC continue to review the methodologies in the light of new information and data. For this reason, the estimates are subject to periodic change, as improved understanding and new methods become available The methodology for each estimate of revenue loss is outlined in the main body of the document, with more detail on data sources and methods in the Annexes. Of particular note is that the estimates presented are potentially subject to both random and systematic errors. The details of these are shown in Annex A HMRC welcomes feedback on this paper from businesses, academic experts and other interested parties, with a view to discussing and, where possible, improving these methodologies further. To comment, or to request further copies of these papers, please use the contact details below. Measuring Indirect Tax Losses KAI Floor 2 HM Revenue and Customs 100 Parliament Street London SW1A 2BQ

4 2. Estimating VAT Losses 2.1. This section discusses losses from the VAT system 1. Methodology Principle 2.2. The methodology for calculating the VAT gap was first published in November The total level of VAT losses can be estimated using a top-down approach by comparing the net theoretical tax liabilities with actual VAT receipts. The difference between these amounts is known as the VAT gap. VAT gap = Net Theoretical Tax Liabilities Actual VAT Receipts 2.4. The approach employs a gap analysis, which involves: Assessing the total amount of expenditure in the economy that is theoretically liable for VAT; Estimating the effective rate of tax due on individual types of expenditure based on commodity breakdowns of the expenditure data; Applying the effective rate of tax to expenditure in each category to estimate the value of tax on the VATable expenditure, to derive the gross VAT theoretical tax liability (VTTL); Subtracting any legitimate refunds (deductions), occurring through schemes and reliefs, to arrive at the net VTTL; Subtracting actual VAT receipts from the net VTTL; Assuming that the residual element, the gap, is the total VAT loss due to any cause. 1 This excludes losses through illicit activity in spirits and tobacco goods which are covered in the excise duty section of this document. 2 Measuring Indirect Tax Losses HM Customs and Excise, November

5 Detail Results 2.5. More detail of the methodology for estimating the VAT gap is given in Annex B. As with any statistical estimate, the estimated level of the VAT Gap is subject to a margin of error. This is explained further in Annex A Table 2.1 shows the VTTL, net VAT receipts and estimated revenue losses for the years to Due to substantial revisions to National Accounts data, which underlies the VTTL, these figures are different in level to those published at PBR Table 2.1: VTTL, VAT receipts and revenue losses Net VTTL ( bn) Net VAT Receipts ( bn 1 ) Revenue Loss ( bn) VAT Gap (per cent) 16.1% 12.6% 12.4% 15.5% 14.2% 1 Net VAT receipts are expressed net of payments and re-payments. 2 A proportion of the VTTL estimate for the final two years is based on a forecast The results show a fall in the percentage VAT gap between and , but then a rise in with a subsequent fall in Overall the level of the VAT gap is lower in than The movement in the VAT gap from one year to the next is determined by the growth in VTTL and the growth in receipts. Growth in VTTL is determined mainly by growth in the total value of transactions liable to VAT and, to a lesser extent, any changes that are made to the tax regime. Figure 2.1 shows that in receipts grew at a higher rate compared with net VTTL growth and so the VAT gap estimate shows a decrease. Figure 2.1: Percentage growth rates of net VTTL and net VAT receipts 10% 9% 8% 7% 6% 5% 4% 3% 2% 1% 0% Net VTTL Growth Net VAT Receipts Growth 5

6 MTIC Fraud 2.9. Missing Trader Intra-Community (MTIC) VAT fraud is a largescale organised criminal attack on the EU VAT system. The most serious form of the fraud known as carousel fraud involves a series of contrived transactions within and beyond the EU, with the aim of creating large unpaid VAT liabilities and fraudulent VAT repayment claims. Methodology Results HMRC currently produces a bottom-up estimate, based on operational evidence, that provides a range for both attempted MTIC fraud, and for the immediate effect this may have had on VAT receipts. It is not appropriate to reveal this methodology as to do so may have a detrimental effect on compliance activity HMRC s current assessment, based on operational data and first published in 2006, is shown in Table Table 2.2: Estimates of MTIC Fraud ( bn) Attempted Fraud Upper Bound 4¾ 3¼ Lower Bound 3½ 2¼ Impact on VAT Receipts Upper Bound 3 2 Lower Bound Both attempted fraud and the impact on VAT receipts fell in Moreover, other evidence, such as published Office for National Statistics (ONS) statistics for MTIC related activity 4, using overseas trade data supplied by HMRC, suggests that much of the activity took place in the first quarter of the financial year. 3 Estimates of MTIC fraud for to were published on a different basis in Measuring Indirect Tax Losses UK Trade First Release June 2007: ONS: 9 August

7 3. Estimating Excise Losses 3.1. This section discusses losses from the excise system 5. At present this includes losses for spirits, cigarettes, hand rolling tobacco (HRT) and hydrocarbon oils. Methodology Principle 3.2. The methodologies for estimating losses due to illicit activity in excise goods were first set out in Measuring Indirect Tax Fraud in November Since that time significant changes have been made to the methodology as HMRC has improved its knowledge and analytical techniques. Therefore this document provides a description of the full methodology for estimating the illicit market for each excise good Losses due to illicit activity correspond to the amount of duty and VAT not collected as a result of the purchases of illicit product. Although there are some differences in methodology between the various excise regimes, the basic methodology is a top down technique the basic principle of which is: Illicit Market = Total Consumption Consumption of Legitimate Goods 3.4. This is estimated for the volume of product where, Total UK consumption is estimated from consumer and other surveys. Consumption of Legitimate Goods is estimated using: Consumption of Legitimate goods = UK Duty Paid goods + Cross-border shopped goods + Duty Free goods Where: UK Duty Paid goods are measured using HMRC clearance data; 5 This includes both losses in duty and the losses in VAT due to smuggled product excluded from the VAT section. 6 Measuring Indirect Tax Fraud: HM Customs and Excise, November

8 Cross-border shopping estimates are derived from survey data collected from international passenger surveys; Duty free goods are estimated from trade data on sales of UK duty-free goods Illicit market share estimates, represented as a percentage, are then calculated as: Illicit Market Share = Illicit Market Total Consumption X Estimates of the illicit market are then combined with price data and duty and VAT rate information to produce an estimate of revenue losses attributable to the consumption of illicit goods. Using ranges to represent the illicit market share 3.7. Illicit market share estimates depend critically on the estimation of total consumption. However, the accuracy required to measure the relatively small changes in consumption in the illicit market is well beyond that ordinarily required of survey data. In addition, the particular products for which we are producing estimates are known to suffer from under-reporting of consumption. For these reasons it is not possible to provide an accurate single estimate of the illicit market, and results are presented as a range within which the illicit market is expected to lie The ranges for the illicit market shares consist of upper and lower bounds estimated from the survey data by varying the assumptions used to turn raw survey data into total consumption estimates. The actual assumptions that are changed are different for each of the excise goods 3.9. As the ranges are designed to deal with systematic rather than random errors they cannot be seen as confidence intervals. In this kind of estimate the true value is just as likely to fall at any point in the range as any other. The year-to-year changes in the illicit market share estimates are subject to significant statistical uncertainty, and so should not be interpreted as representing short-term changes in compliance. However the mid-point does give an indication of the longterm trend in the illicit market The estimates shown for oils consist of a central estimate and 95 per cent confidence intervals. 8

9 Detail Alcohol Results The actual calculations of the illicit market share estimates are significantly different for each of the excise goods, therefore these are discussed in detail, along with supplementary results, in technical Annexes to this document as follows: Alcohol products Annex C; Tobacco products Annex D; Hydrocarbon Oil products Annex E Estimates of revenue lost to the illicit market are currently available for spirits. Estimates are not included for other alcohol goods, though development is continuing with the aim of increasing the scope of published estimates Estimates of the illicit market share for spirits rely on data from the Expenditure and Food Survey (EFS), which only becomes available around 18 months after the survey period. For this reason, estimates for are not yet available. Table 3.1: Spirits: Illicit Market and Associated Revenue Losses 1, Illicit Market Shares: Upper Estimate 17% 14% 15% 7% 10% Lower Estimate Associated Revenue Losses ( million) Upper Estimate Lower Estimate Indicates estimates of negligible levels of fraud. 1 Includes duty and VAT although this will overstate losses to the extent that VAT is collected on sales of illicit alcohol through normal retail outlets. 2 Figures are independently rounded to the nearest 50m or 1%. 9

10 Figure 3.1: Spirits: Illicit Market Shares Estimated Range Table 3.1 and Figure 3.1 show the latest estimates of the illicit market share for the spirits market, this is presented as lying within the range defined by the upper and lower estimates. Table 3.2: Spirits: Illicit Market Mid-point of range Illicit Market Share Mid-Point 8% 7% 8% 4% 5% Tobacco The mid-point of the range is shown in Table 3.2. The trend in the mid-point shows a decline in the level of the illicit market between and from 8 per cent to 5 per cent Estimates of revenue lost to the illicit market are currently available for both cigarettes and Hand Rolling Tobacco (HRT). Results: Cigarettes Estimates of the illicit market share for cigarettes and HRT rely on data from the General Household Survey (GHS), which only becomes available around 12 months after the survey period. For this reason, estimates for are not yet available. 10

11 Table 3.3: Cigarettes: Illicit Market and Associated Revenue Losses 1, Illicit Market Shares: Upper Estimate 24% 18% 21% 19% 18% Lower Estimate 16% 12% 15% 10% 8% Associated Revenue Losses ( million) Upper Estimate 3,200 2,400 3,000 2,500 2,300 Lower Estimate 2,000 1,500 1,900 1,200 1,000 1 Includes duty and VAT. 2 Figures are independently rounded to the nearest 100m or 1%. Figure 3.2: Cigarettes: Illicit Market Shares Estimated Range Table 3.3 and Figure 3.2 show the latest estimates of the illicit market share for the cigarette market, this is presented as lying within the range defined by the upper and lower estimates. Table 3.4: Cigarettes: Illicit Market Mid-point of range Illicit Market Share Mid-Point 20% 15% 18% 14% 13% The mid-point of the range is shown in Table 3.4. The trend in the mid point shows a decline in the level of the illicit market between and from 20 per cent to 13 per cent. Results: Hand Rolling Tobacco Table 3.5 and Figure 3.3 show the latest estimates of the illicit market share for the HRT market, this is presented as lying within the range defined by the upper and lower estimates. 11

12 Table 3.5: HRT: Illicit Market and Associated Revenue Losses 1, Illicit Market Shares: Upper Estimate 63% 62% 60% 63% 62% Lower Estimate 47% 47% 44% 52% 50% Associated Revenue Losses ( million) Upper Estimate Lower Estimate Includes duty and VAT. 2 Figures are independently rounded to the nearest 10m or 1%. Figure 3.3: HRT: Illicit Market Shares - Estimated Range. Table 3.6: HRT: Illicit Market Mid-point of range Illicit Market Share Mid-Point 55% 54% 52% 58% 56% The mid-point of the range is shown in Table 3.6. The trend in the mid-point indicates a fairly stable HRT illicit market share between and at around 55 per cent. Hydrocarbon Oils HMRC currently provides estimates of the illicit market for both diesel and petrol in Great Britain (GB), and of the Non-UK Duty Paid (NUKDP) share in petrol and diesel for Northern Ireland (NI) within the goods subject to hydrocarbon oils duty. Results: Great Britain: Diesel Estimates Estimates of the illicit market in Great Britain are shown in Table 3.7 and Figure 3.4. Figures for are provisional. 12

13 Table 3.7: GB: Diesel: Illicit Market Share and Associated Revenue Losses 1, Illicit Market Shares: Upper Confidence Interval 11% 11% 11% 10% 8% Central Estimate 6% 5% 5% 4% 2% Lower Confidence Interval 1% 0% 0% 0% 0% Associated Revenue Losses ( million): Upper Confidence Interval 1,300 1,350 1,400 1,400 1,200 Central Estimate Lower Confidence Interval Estimates include duty and VAT. 2 Figures are independently rounded to the nearest 50m or 1%. 3 Figures for are provisional, as not all components of the total consumption estimate are available at this time. Figure 3.4: GB: Diesel: Illicit Market Shares Central Estimate and Confidence Intervals 1 1 The lower confidence intervals for 2003 onwards are truncated at zero. Results: Great Britain: Petrol Estimates Petrol is considered to be less vulnerable than diesel to illicit activity because of the low demand from commercial sectors and the flammable nature of the product. The small non-uk duty paid petrol market (less than 1%) and a wide margin of error mean that more detailed analyses are not possible. Results: Northern Ireland Estimates All estimates for Northern Ireland relate to total non-uk duty paid consumption, rather than the illicit market. This reflects the present difficulty in disaggregating total revenue losses between illicit activity and legitimate cross-border shopping. 13

14 Table 3.8: NI: Diesel Non-UK Duty Paid Market Share and Associated Revenue Not Collected 1, NUKDP Market Share: Upper Confidence Interval 62% 59% 56% 48% 48% Central Estimate 58% 53% 51% 42% 43% Lower Confidence Interval 51% 49% 47% 39% 40% Associated Revenue Losses ( million): Upper Confidence Interval Central Estimate Lower Confidence Interval Figures are rounded to the nearest 10m or 1%. 2 Estimates include duty and VAT. 3 Figures for are provisional, as not all components of the total consumption estimate are available at this time. Table 3.9: NI: Petrol: Non-UK Duty Paid Market Share and Associated Revenue Not Collected 1, NUKDP Market Share: Upper Confidence Interval 22% 22% 22% 20% 24% Central Estimate 18% 18% 18% 13% 14% Lower Confidence Interval 11% 11% 11% 9% 6% A ssociated Revenue Losses ( million): Upper Confidence Interval Central Estimate Lower Confidence Interval Figures are rounded to the nearest 10m or 1%. 2 Estimates include duty and VAT. 3 Figures for are provisional, as not all components of the total consumption estimate are available at this time. 14

15 Annex A: Margins of Error VAT A.1. The figures presented here are subject to statistical uncertainty caused by two factors: Random variation in sample data (the whole population is not questioned, therefore there is a possibility that an unrepresentative response is obtained); Systematic errors where the sample consistently under or over represents the true values for the population (e.g., people consciously under-report their true tobacco consumption). A.2. Where possible HMRC has estimated the effects of these errors. Where this is from the random variation in survey results, HMRC has calculated margins of error. These margins of error approximate to 95 per cent confidence intervals, which give the range within which the true value will lie 95 per cent of the time. However, it should be noted that these are not true confidence intervals as the correction for under reporting may introduce a systematic error, meaning that the error margins are not centred on an unbiased estimator. A.3. Systematic error is less straightforward to deal with, as it is not defined by statistical assessments that allow easy interpretation. In order to give an indication of the effect of these biases HMRC now present the illicit market share estimates as ranges. These are constructed as the range between an upper and lower bound, representing the degree of uncertainty associated with the systematic biases. A.4. The VTTL model and the top-down VAT Gap derived from it are broad measures, subject to a degree of uncertainty. They are based on an analysis of survey and other data, and include a number of assumptions and adjustments which add both random and systematic variation to the estimates. For the final two years estimates, there is also an element of forecasting in some of the spending data, which introduces further random variation. A.5. It is not possible to produce a precise confidence interval for the VAT revenue loss estimates. The VTTL estimate is constructed largely from ONS National Accounts data which are derived, in the main, from sample surveys and are thus subject to both sampling and nonsampling errors. The ONS does not publish error margins for the relevant input series and so it is not possible to construct a precise estimate of the impact of these errors on the VTTL. However, a broad 15

16 analysis of potential errors has shown that the margin of error could be up to ± 4 percentage points. A.6. It is important to note that this error margin relates to the point estimates of the VAT gap. Part of the error is likely to result from systematic errors in the source data, which do not contribute to error bounds for year on year changes in the VAT gap. Therefore it is likely that the error margins on the year on year changes observed in the VAT gap will be smaller than ± 4 percentage points. A.7. The VAT Gap is updated and revised as and when new data becomes available and new methodologies developed. Work is therefore continuing to identify and allow for sources of error and bias. Excise Duties Systematic Biases A.8. Systematic biases are explicitly considered for spirits and tobacco products, with results presented as a range. The definitions of the ranges are shown in Annex C for spirits and Annex D for tobacco products. A.9. For oils, HMRC continues to undertake analyses to define suitable ranges. No account is presently made for systematic biases in the oils estimates. Random Variation A.10. While the upper and lower estimates for spirits and tobacco will contain random variation, the resultant confidence intervals are not shown in this document. A.11. For oils an assessment of the effect of random variation is included using error margins. These are estimated by combining the random errors (where available) from all data sources used to calculate total consumption. The combined random errors are estimated by simulation techniques and the results presented as combined margins of error. These approximate to 95 per cent confidence intervals, standard across statistical analyses. 16

17 Annex B: Detail for VAT Estimates General Methodology B.1. The methodology for calculating the VAT gap was first published in November The total level of VAT losses can be estimated by comparing the net theoretical tax liabilities with actual VAT receipts. The difference between these amounts is known as the VAT gap. VAT gap = Net Theoretical Tax Liabilities Actual VAT Receipts B.2. The approach employs a gap analysis, which involves: Assessing the total amount of expenditure in the economy that is theoretically liable for VAT; Estimating the effective rate of tax due on individual types of expenditure based on commodity breakdowns of the expenditure data; Applying the effective rate of tax to expenditure in each category to estimate the value of tax on the VATable expenditure, to derive the gross VAT theoretical tax liability (VTTL); Subtracting any legitimate refunds (deductions), occurring through schemes and reliefs, to arrive at the net VTTL; Subtracting actual VAT receipts from the net VTTL; Assuming that the residual element, the gap, is the total VAT loss due to any cause. Calculation of Total Liability General Methodology B.3. VTTL is the theoretical amount of VAT that would be collected in the absence of any losses. It is calculated by multiplying appropriate categories of expenditure liable to VAT in the economy by their VAT rate and allowing for other relevant rules determining tax liability. B.4. The expenditure series used in the calculation are mainly constituents of National Accounts macroeconomic aggregates. All National Accounts data used to construct VTTL estimates are Measuring Indirect Tax Losses : HM Customs and Excise, November

18 consistent with the latest Office for National Statistics (ONS) Blue Book. B.5. A number of streams of expenditure contribute to the tax base, with most VAT deriving from consumers expenditure. The main expenditure categories that comprehensively cover VAT liabilities are: Input Tax Adjustments Deductions Household spending and non-profit institutions serving households final consumption expenditure; Central government current and capital expenditure; Exempt sector intermediate consumption and other input tax blocks; Housing expenditures - certain household and corporate capital expenditure which incurs non-refundable VAT. B.6. Net VAT liability is the difference between VAT due on taxable supplies made by registerable traders ( output tax ), and that recoverable by traders on supplies made to them ( input tax ). B.7. VAT liability for the relevant categories can be estimated directly from National Accounts data, with one exception - the exempt sector. Businesses making outputs that are exempt from VAT are generally not permitted to reclaim the VAT on inputs associated with their exempt outputs. In order to make an adjustment for this irrecoverable input tax, a separate HMRC survey is used to ascertain the proportion of purchases on which VAT cannot be reclaimed. B.8. A further adjustment is made for expenditure by businesses legitimately not registered for VAT and, as such, the VAT is not recoverable as input tax. This adjustment uses a combination of ONS data and HMRC information on the distribution of business turnover below the VAT threshold to estimate relevant expenditure. B.9. Finally, third party data sources are used in conjunction with National Accounts data to inform estimates of business expenditure on cars and entertainment, on which VAT is due. B.10. Because the calculation of non-recoverable input tax is complex, the level of uncertainty around input tax adjustments is larger than for the other elements. B.11. The sum of the VAT liability arising from each of the expenditure categories listed in paragraph B.5. gives an estimate of the gross VTTL 18

19 in each year. However there are a number of legitimate reasons why part of this theoretical VAT is not actually collected. These can be grouped into two broad categories: VAT refunds; Expenditure at traders legitimately not registered for VAT. B.12. VAT refunds are made primarily to government departments, NHS Trusts and regional health authorities for specified contracted out services acquired for non-business purposes. A number of other categories of expenditure cannot be separately identified in the overall VTTL calculation, for which VAT can be refunded. The value of these refunds is taken directly from audited HMRC accounts data. B.13. Traders who trade below the VAT threshold can legitimately exclude VAT on their sales. Expenditure on the output of these businesses will have been picked up in the theoretical liability. To adjust for this an estimate of relevant expenditure is made using a combination of ONS data and HMRC information on the distribution of business turnover below the VAT threshold. Methodology Changes B.14. The detailed calculations used to construct the estimated VTTL are continuously reviewed to identify improvements to the methodology. B.15. The proportion of household expenditure on ex-business cars in second-hand car sales is now taken directly from data used to estimate the amount of input tax blocked on expenditure on cars by businesses, which is derived from ONS National Accounts data. This replaces an adjustment previously taken from the Own Resources Account, prepared by HMRC for the European Commission. This change ensures consistency in the data being used throughout the VTTL calculations. B.16. The VTTL is now adjusted to reflect refunds of VAT made to the Isle of Man, under their agreement with the UK authorities to simplify tax collection procedures for businesses. Summary of VTTL B.17. Estimates of the contribution to the VTTL of each relevant expenditure component are given in Table B.1. Note that the household element excludes expenditure on purchases of illicit alcohol and tobacco. The revenue losses associated with such purchases are considered in the excise illicit market share estimates and to measure them here would constitute double counting. 19

20 Table B.1: Expenditure components of VTTL ( bn) Households Exempt Government Housing Gross VTTL Deductions Net VTTL Household component also includes estimates of expenditure by non-profit institutions serving households; estimates exclude expenditure on smuggled alcohol and tobacco. Measure of Tax Collected B.18. Figures for actual receipts of VAT are taken from HMRC s published Consolidated Fund figures. The receipts are adjusted, to reflect timing effects within each financial year, before being used in the model. Changes to data B.19. Since estimates of the VTTL and VAT gap were last published in December 2006 there have been data revisions affecting the estimates. The main changes are summarised in the following section. B.20. The National Accounts data used to construct the VTTL are subject to updates and revision by ONS throughout the year. This is part of the routine revisions to the ONS National Accounts data as final data become available. Supplementary Results Summary of changes to previously published estimates B.21. Table B.2 and Figure B.1 compare the current estimate of the percentage VAT gap series with the estimate published at PBR The main effect of revisions in the last year has been to increase estimates of the VAT gap for each year since Table B.2: Time Series of percent VAT gap: Current Estimate and Estimate Published at PBR VAT Gap - PBR % 12.4% 11.7% 14.5% - VAT Gap - Current Estimate 16.1% 12.6% 12.4% 15.5% 14.2% 20

21 Figure B.1: Current and previously published estimates of percent VAT gap time-series 18.0% 16.0% 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% PBR 2006 Current Estimate 21

22 Annex C: Detail for Alcohol Estimates General Methodology C.1 The estimate of the illicit market for spirits is produced using a top-down methodology. That is, the estimate is produced by first estimating total consumption of spirits purchased in the UK, and then subtracting legitimate UK consumption, the residual being the illicit market: Illicit Market = Total Consumption Consumption of Legitimate Goods C.2 For spirits, total consumption estimates only cover the consumption of products purchased in the UK. Details of the estimation of total consumption and of consumption of legitimate goods are provided in the following sections. C.3 This equation provides an estimate of the volume of goods supplied through the illicit market. This is then turned into an estimate of the proportion of the total market that is supplied through the illicit market the Illicit Market Share: Illicit Market Illicit Market Share = Total Consumption + Cross-border Shopping + Duty Free X 100 C.4 The total market is the total consumption of spirits bought in UK plus the cross-border shopping and duty-free elements of the legitimate consumption. This calculation is necessary as the Expenditure and Food Survey (EFS) only covers purchases within the UK. C.5 Revenue losses associated with the illicit market are then estimated by combining the illicit market share information with price data and duty and VAT rate information. C.6 As explained in paragraphs 3.7 to 3.10, uncertainties in the data sets used to create these estimates mean that it is only possible to provide estimates in the form of a range within which the level of the illicit market is likely to fall. For spirits, the determining factor for the size of the range is the estimate of total consumption. The specifics for the estimation of the upper and lower bound are discussed in the relevant section. Estimates of Total Consumption C.7 The estimate of total spirits consumption for each year is produced using the Family Expenditure Survey (FES) for the period prior to 2001 and the Expenditure and Food Survey (EFS) since its 22

23 inception in Estimates are based on data weighted by survey sample sizes in order to maintain representativity of the estimates. C.8 As the survey data for alcohol consumption are subject to under-reporting, an uplift factor is necessary to correct for the bias this causes. C.9 Total UK consumption in any given year is calculated using: Estimates of household spirits on-and off-licence expenditures from the EFS. HMRC converts these into consumption in volumes using data on average prices provided by a commercial market analysis company; Estimates of the adult population from the ONS; Estimates of the number of adults per household in the UK population from the ONS; A correction factor covering both under-reporting in each of the on- and off-licence sector. This leads to an estimate defined by: Total Consumption = Average Household Consumption Number of Adults per Household X Number of Adults X Correction Factor C.10 The estimation of average household consumption is complicated because a proportion of the other drinks category in the EFS and FES should be included as spirits consumption. This proportion is not accurately known, so the estimation exercise is undertaken twice to produce an upper and lower bound. C.11 The correction factor is calculated by comparing estimates of spirits consumption with HMRC clearance data. We use clearance data for a base year in which there is believed to be minimal illicit activity, and therefore gives a true indication of total consumption. In order to reduce error, we have created the correction factor using the average of three year s data: ; ; and C.12 Separate correction factors are calculated for the on- and offlicence markets. In order to do this, clearance data are split between on- and off-licence markets using market shares estimated from commercial sales data. C.13 The correction factor is therefore defined as: Correction factor = Average of Clearances for to Average Total Consumption to

24 The correction factors are held constant over time for both the on and off license markets. Bounds for Total Consumption C.14 The issue of the allocation of survey responses to the correct type of drink mean that estimates are variable. They also have the potential to contain systematic biases through the estimate of average household consumption. There is no reliable way to correct for this, therefore a range has been developed to represent the uncertainty in the estimates. Upper Bound Lower Bound C.15 The upper estimate is defined using the following assumptions: For and earlier years, a proportion of the other drinks category equal to the proportion of spirits in all known drinks has been included in the spirits figure; For to , the entire other drinks category is reallocated to spirits - this provides an upper bound for total spirits expenditure as the illicit market share cannot be higher than this; For the analysis of EFS data revealed further issues in the recording of the other drinks category. In order to provide more robust figures of consumption the other drinks category is re-allocated into spirits by: o Taking the other drinks consumption; o Adding to this a portion of the increase in other drinks in between and The adding portion being equal to the total increase in total drinks multiplied by the proportion of spirits in total known drinks for C.16 Redistributing the other drink category to produce a lower bound uniformly produces a figure that is less than zero. This leads to the conclusion that the lower level of fraud has to be zero as it cannot reasonably be lower. Therefore, in order to supplement the top-down approach, an analysis of direct measures of fraud has been undertaken. These have led to figures lower than 1 per cent and so the lower bound is reported as negligible levels of fraud. 24

25 Estimates of Consumption of Legitimate Purchases C.17 Estimates of legitimate consumption have two elements: UK duty paid consumption; Cross-border and duty free shopping. UK-duty paid consumption C.18 Estimates of UK-duty paid consumption are taken directly from returns to HMRC of the volumes of spirits on which duty have been paid, along with the actual amounts of money. Since these come from administrative sources (clearance data) they are assumed not to be subject to errors. The volumes of ready-to-drink products have been removed from the spirits clearances in order to obtain figures for spirits only. Cross-border and duty free shopping C.19 Estimates of consumption of goods purchased as cross-border shopping are based on figures produced from the International Passenger Survey (IPS). This provides estimates of the volume of spirits an average adult traveller brings into the country, separately for air and sea passengers. These estimates are then multiplied by the number of passengers arriving by each mode, provided by commercial traffic information, to give an estimate of total cross-border shopping. C.20 This estimate, however, does not cover sales made on-board ferries. Therefore, commercially provided data about deliveries of spirits to ferries are used to supplement the cross-border shopping estimate, and provide a complete figure. Conversion to monetary losses C.21 All calculations to this point have been made on volumes of spirits. However, it is also important to know the equivalent losses of revenue. The formula is: Losses = Illicit Volume x Spirits Duty Rate + Illicit Volume x Average Price x VAT Rate C.22 The average price is taken as the average retail price of spirits for the on- and off sector. This method assumes that VAT is also lost on all purchases. As, in some cases, the final product is sold in legitimate outlets this may not always be the case, and so this will be an overestimate of losses. 25

26 Summary of Methodology C.23 A summary of the calculation of the spirits upper bound is: Illicit Market = Total Consumption UK duty paid Consumption HMRC Clearance Data Household Consumption X UK Adult Population X Average Clearances ( to ) Adults per Household Average Consumption ( to ) C.24 Statistical analyses for the lower bound do not produce a meaningful result; therefore figures are estimated directly from HMRC operational results. These suggest a lower bound of less than 1 per cent. Supplementary Results C.1. Table C.1 shows estimates of market shares for the upper and lower estimate, along with the mid-point for spirits. Table C.1: Spirits: Breakdown of UK Market Shares 1, UK Tax Paid Consumption Upper Estimate 94% 95% 95% 95% 96% Lower Estimate 78% 82% 81% 88% 86% Mid-Point 86% 88% 88% 92% 91% Illicit Market Upper Estimate 17% 14% 15% 7% 10% Lower Estimate Mid-Point 8% 7% 8% 4% 5% Cross-Border Shopping 3 5% 5% 4% 4% 4% - Indicates estimates of negligible levels of fraud. 1 Includes duty and VAT although this will overstate losses to the extent that VAT is collected on sales of illicit alcohol through normal retail outlets. 2 Figures are independently rounded to the nearest 1%. As a result components may not appear to sum. 3 Cross-border shopping estimates include provisional data for and may be subject to revision when final data becomes available. 26

27 C.2. Table C.2 shows estimates of revenue for the upper and lower bounds, along with the mid-point for spirits. Table C.2: Spirits: Breakdown of Revenues ( million) 1, Tax Due: Total Consumption Upper Estimate 3,650 3,700 3,900 3,600 3,650 Lower Estimate 3,050 3,200 3,300 3,350 3,300 Mid-Point 3,350 3,450 3,600 3,500 3,450 UK Tax Paid Consumption Illicit Market 2,900 3,050 3,200 3,200 3,150 Upper Estimate Lower Estimate Mid-Point Cross-Border Shopping Indicates estimates of negligible levels of fraud. 1 Includes duty and VAT although this will overstate losses to the extent that VAT is collected on sales of illicit alcohol through normal retail outlets. 2 Figures are independently rounded to the nearest 50m. As a result components may not appear to sum. 3 Cross-border shopping estimates include provisional data for and may be subject to revision when final data becomes available. C.3. Table C.3 shows the upper estimate, lower estimate and mid point of volumes of spirits illicitly entering the UK market and the central estimate for cross-border shopping and UK duty paid consumption. Table C.3: Spirits: Breakdown of Volumes (million litres) Total Consumption Upper Estimate Lower Estimate Mid-Point UK Tax Paid Consumption Illicit Market Upper Estimate Lower Estimate Mid-Point Cross-Border Shopping Indicates estimates of negligible levels of fraud. 1 Figures are independently rounded to the nearest 5 million litres. As a result components may not appear to sum. 27

28 Annex D: Detail for Tobacco Estimates Cigarette Estimates Methodology General Methodology D.1. The estimate of the illicit market for cigarettes is produced using a top-down methodology. That is, the estimate is produced by first estimating total consumption, and then subtracting legitimate consumption, the residual being the illicit market: Illicit Market = Total Consumption Consumption of Legitimate Goods D.2. Details of the estimation of total consumption and of the consumption of legitimate goods are provided in the following sections. D.3. This equation provides an estimate of the volume of goods supplied through the illicit market. This is then turned into an estimate of the proportion of the total market that is supplied through the illicit market the Illicit Market Share: Illicit Market Share = Illicit Market Total Consumption X 100 D.4. Revenue losses associated with the illicit market are then estimated by combining the illicit market share information with price data and duty and VAT rate information. D.5. As explained in paragraphs 3.7 to 3.10, uncertainties in the data sets used to create these estimates mean that it is only possible to provide estimates in the form of a range within which the level of the illicit market is likely to fall. For cigarettes the determining factor for the size of the range is the estimate of total consumption, and the specifics for the estimation of the upper and lower bound are discussed in the relevant section. Estimates of Total Consumption D.6. The estimate of total UK consumption of cigarettes for each year is based on the estimates of cigarette smoking prevalence and consumption per smoker obtained from the General Household Survey (GHS). However as the survey data for tobacco consumption are 28

29 subject to under-reporting an uplift factor is necessary to correct for the bias that under-reporting causes. D.7. The estimation of this uplift factor is a difficult exercise, and uncertainties mean that it cannot be done accurately enough to produce a single point estimate of total consumption. Therefore, the exercise is undertaken twice once to produce an upper bound for total consumption, and once to produce a lower bound. This allows us to produce a range for total consumption that correctly presents the uncertainty in the underlying data. Upper Bound for Total Consumption D.8. The upper bound for total consumption in any given year is calculated using: Estimates of prevalence (proportion of the population that smokes cigarettes) from the GHS; Estimates of the adult population from the ONS; A correction factor covering both under-reporting and consumption per smoker. This leads to an estimate defined by: Total consumption = Prevalence x Adult Population x Correction Factor D.9. The correction factor is used to both correct the estimate for under-reporting and to turn prevalence estimates into total consumption estimates by forming a constant estimate of consumption per smoker. D.10. The correction factor is calculated by taking estimates of prevalence in a base year, and comparing these with independent estimates of total consumption. To do this we take a year in which there is believed to be little or no illicit market and use HMRC clearance data, duty free and cross border shopping estimates as a true indication of total consumption for cigarettes, is used. D.11. Theoretically, in order to reduce error, we would like to create the correction factor using the average of three year s data. However, as there is a strong trend in consumption per smoker figures over the base period using data for more than one year does not reduce random error, and has the potential of introducing a bias. For that reason a single year is used. D.12. The correction factor is therefore defined as: Correction factor = Legitimate Consumption for Prevalence Estimate for from GHS 29

30 D.13. This estimate of total consumption makes three assumptions: That there is no illicit market in ; That consumption per smoker has been constant over time; That under-reporting remains constant over time. An analysis of these assumptions suggests that they do not hold. In particular, there is strong evidence that consumption per smoker has fallen significantly since , which means that the estimate produced is too high for later years, and certainly since While evidence of a small illicit market in , and changes in under-reporting will reduce this effect, the reduction will be small and the combined effect means that the resultant estimate of total consumption is an upper bound. Lower Bound for Total Consumption D.14. The lower bound for total consumption in any given year is calculated using: Estimates of prevalence (proportion of the population that smokes cigarettes) from the GHS; Estimates of cigarette consumption per smoker from the GHS; Estimates of the adult population from the ONS; A correction factor covering under- reporting. This leads to an estimate defined by: Total consumption = Prevalence x Consumption per smoker x Adult Population x Correction Factor D.15. The correction factor is only used to correct the estimate for under-reporting. It is calculated by taking estimates of consumption in a base year, and comparing these with independent estimates of total consumption. In this case we take a year in which there is believed to be little or no illicit market and use HMRC clearance data, duty free and cross border shopping estimates as a true indication of total consumption - again is used. Theoretically, in order to reduce error, we would like to create the correction factor using the average of three years data. However GHS in the 1990s only collected data on smoking prevalence in alternate years. We do not have data for and , and we have to rely on data from the single year. This does not increase bias in the estimate, but will increase random variation, and hence the resultant confidence intervals. Figures for 30

31 legitimate consumption do exist for the three years, however, so an average for legitimate consumption is calculated for , and D.16. The correction factor is therefore defined as: Correction factor = Average Legitimate Consumption for to Prevalence Estimate for x Consumption per smoker for D.17. This estimate of total consumption makes two assumptions: That there is no illicit market in ; That under-reporting remains constant over time. An analysis of these assumptions suggests that they do not hold. Specifically, it is likely that there was a small illicit market in , and that there are likely to have been changes in the under-reporting of consumption per smoker that will mean that the estimate is too low. Of particular importance is the underreporting of consumption per smoker, which is thought to have increased over time since the base year, causing the estimate of total consumption to be too low. For this reason the estimate is believed to be a lower bound. Estimates of Consumption of Legitimate Purchases D.18. Estimates of legitimate consumption have two elements: UK-duty paid consumption UK duty paid consumption; Cross-border and duty free shopping. D.19. Estimates of UK-duty paid consumption are taken directly from returns to HMRC on the volumes of cigarettes on which duty have been paid, along with the actual amounts of money. In general these can be used directly; however there is a complication around the duty increases associated with the budget that usually happens at the end of each financial year. Not all cigarettes cleared in the two months before a budget will be consumed in that period, meaning that the figures do not match the consumption estimates. To correct for this the clearance data are corrected over three months, by taking an average over that period. Cross-border and duty free shopping D.20. Estimates of consumption of goods purchased as cross-border shopping are based on figures produced by the International Passenger Survey (IPS). This provides estimates of the number of 31

32 cigarettes an average adult traveller brings into the country, separately for air and sea passengers. These are then multiplied by the number of passengers arriving by each mode, provided by commercial traffic information from the Civil Aviation Authority (CAA), Eurostar & the Department for Transport (DfT) to give an estimate of total crossborder shopping. D.21. This estimate, however, does not cover sales made on-board ferries. Therefore, commercially provided data about deliveries of cigarettes to ferries are used to supplement the cross-border shopping estimate, and provide a complete figure. D.22. Estimates of duty free consumption are difficult to assess directly, therefore an indirect measure is used. An assumption is made that a similar number of cigarettes are purchased in duty free shops in the UK as are brought in to the country by UK passengers. Using this assumption an estimate of duty-paid consumption is made from the amount of cigarettes supplied by tobacco manufacturers for sale in UK duty free shops. Conversion to monetary losses D.23. All calculations to this point have been made on volumes of cigarettes. However, it is also important to know the equivalent losses of revenue. The formula for performing this calculation is: Losses = Specific Duty + AD velorum duty x average price + VAT rate x (average price VAT) D.24. The average price is taken as the retail price of the most popular specific brand of cigarettes. This method assumes that VAT is also lost on all purchases. As, in some cases, the final illicit product is sold in legitimate outlets this may not always be the case, and so this will be an over estimate of losses. Summary of Methodology D.25. A summary of the calculation of the cigarette upper bound is: Illicit Market = Total Consumption UK duty paid Consumption Cross-border Shopping Duty Free HMRC Clearance Data Average Carried per traveller x Number of Travellers Supply to UK Duty Free Shops Prevalence X UK Adult Population X Legitimate Consumption in Prevalence in