DISTRIBUTIONAL IMPACT OF POSSIBLE TAX REFORM PACKAGES

Transcription

1 National Centre for Social and Economic Modelling University of Canberra DISTRIBUTIONAL IMPACT OF POSSIBLE TAX REFORM PACKAGES Neil Warren, Ann Harding, Martin Robinson, Simon Lambert and Gillian Beer Report to Senate Select Committee on a New Tax System 1 April 1999

2 National Centre for Social and Economic Modelling University of Canberra The National Centre for Social and Economic Modelling was established on 1 January 1993, and currently receives core funding from the University of Canberra and the federal departments of Family and Community Services; Health and Aged Care; Education, Training and Youth Affairs; and Employment, Workplace Relations and Small Business. NATSEM aims to be a key contributor to social and economic policy debate and analysis by developing models of the highest quality, undertaking independent and impartial research, and supplying valued consultancy services. Policy changes often have to be made without sufficient information about either the current environment or the consequences of change. NATSEM specialises in analysing data and producing models so that decision-makers have the best possible quantitative information upon which to base their decisions. NATSEM has an international reputation as a centre of excellence for the analysis of microdata and the construction of microsimulation models. Such data and models commence with the records of real (but unidentifiable) Australians. Analysis typically begins by looking at either the characteristics of or the impact of a policy change on an individual household, building up to the bigger picture by looking at many individual cases through the use of large datasets. It must be emphasised that NATSEM does not have views on policy: all opinions are the authors own and are not necessarily shared by NATSEM or its core funders. Director: Ann Harding National Centre for Social and Economic Modelling University of Canberra ACT 2601 Australia 170 Haydon Drive Bruce ACT 2617 Phone Fax natsem@natsem.canberra.edu.au Website

3 NATSEM - Distributional Impact of Possible Tax Reform Packages Abstract In 1998 the Howard Government presented the electorate with a comprehensive plan for tax reform. The Senate Committee on a New Tax System commissioned us to independently estimate the likely distributional impact of this tax reform package, using a range of modelling and distributional assumptions specified by the Committee. Author note Ann Harding is Professor of Applied Economics and Social Policy and Director of the National Centre for Social and Economic Modelling (NATSEM) at the University of Canberra. Neil Warren is Associate Professor of Economics at the Australian Taxation Studies Program (ATAX) at the University of New South Wales. Martin Robinson is a Research Fellow and Simon Lambert and Gillian Beer are Senior Research Fellows at NATSEM. Acknowledgments Various officers from the Treasury provided technical advice on the details of the ANTS package and other officers within the Department of Family and Community Services provided technical advice about the details of their programs. This advice is very gratefully acknowledged. We would also like to thank Penny Gallagher for her assistance with finalising the printed and electronic versions of this report.

4 Contents 1 INTRODUCTION DESCRIPTION OF THE OPTIONS Option 1: The Government s Package Option 2: New Homes and Tobacco in CPI Option 3: Household Specific Price Effects and Dissaving Option 4: 70 Per Cent Pass Through of Price Falls Option 5: Food Out and Income Taxes Increased Option 6: Food Out and GST Rate Increased Option 7: Food In with Increased Compensation Summary of the Seven Options OPTIONS 1 AND 2 : ANTS AND TOBACCO AND NEW HOMES PRICE CHANGES Option 1: The Government s Package Option 2: New Homes and Tobacco in CPI OPTION 4: ANTS WITH 70 PER CENT PASS THROUGH OF PRICE FALLS OPTIONS 5 AND 6: FOOD OUT OF GST BASE OPTION 7: FOOD IN BUT COMPENSATION INCREASED CONCLUSIONS...40 APPENDIX A TERMS OF REFERENCE FOR CONSULTANCY: DISTRIBUTIONAL IMPACT OF TAX PACKAGE...42 APPENDIX B PHASING OF ANTS REFORMS...44 APPENDIX C IMPLICATIONS OF ADOPTING THE 13 TH SERIES CPI...46 APPENDIX D MODELLING METHODOLOGY AND ASSUMPTIONS...48 REFERENCES...59

5 NATSEM - Distributional Impact of Possible Tax Reform Packages Executive Summary The Senate Committee on a New Tax System commissioned us to model the Government s ANTS tax package (Treasurer, 1998); to assess the sensitivity of some of the key assumptions made by the Treasury in its modelling of the distributional impact of the package; and to examine the impact of specified possible major changes such as excluding food from the tax base of the proposed new Goods and Services Tax (GST). For most of the options examined, the brief from the Senate Committee asked us to abstract from the timing issues associated with the introduction of the various indirect tax reforms, and to model the distributional impact of the ANTS package reforms in as if they applied in full to households in July In effect, this means that this report examines the expected longer-run distributional impact of the package. In many cases, it was not clear what was the most appropriate set of assumptions to make about the likely tax/transfer system in the face of some of the possible changes requested by the Senate Committee. For example, the Senate Committee asked us to model the likely distributional impact if businesses failed to immediately pass on to consumers (via lower prices) the proposed cuts in the existing Wholesale Sales Tax. Such an outcome would result in a higher increase in the Consumer Price Index (CPI). In such a case, would it still be fair to assume that the Government would increase pensions by only 3.4 per cent, as intended when the expected CPI increase was substantially lower? In the event, we have modelled ten scenarios, the key parameters of which are summarised in the table overleaf. The first scenario is our attempt to replicate exactly the methodology followed by the Treasury in assessing the distributional impact of the tax reform package. Our estimated CPI increase under this set of assumptions is 2.0 per cent, which is very close to the Treasury s 1.9 per cent estimate. Our estimates of the distributional impact of the package are extremely close to the Treasury s in almost all cases (see Part 2 of this report for the detailed cameo household results). In Option 2, we again followed the Treasury s methodology, with the single exception that the expected impact of the package on tobacco and new home prices was included within the CPI measure. It should be noted that we do not endorse using this measure as the appropriate indicator of the likely change in prices facing households. There are strong arguments for not compensating smokers for the likely increase in prices facing them under the ANTS tax reform package. In addition, we have not been able to model the distributional impact of the proposed First Home Owners Scheme, so that the adverse distributional effect of the GST on house prices is included within this measure but not the positive effect of the new scheme specifically designed to overcome this adverse effect. Two of the Treasury s assumptions in modelling the distributional impact of the ANTS package have been particularly controversial. The first is that the CPI will provide an adequate measure of the likely increase in prices facing different types of households. The second is that any possible dissaving by households should not be taken into account when calculating the likely gains and losses from the ANTS tax package. Option 3 examines the outcomes when these two 1

6 assumptions are changed (in both this and all subsequent Options). First, the estimated average increase in prices facing each of the 29 cameo household types is used to guage the likely impact of the indirect tax reforms upon them. Second, households with zero private income are assumed to be spending 10 per cent more than their disposable income, with the extent of possible dissaving assumed to decline linearly with income, reaching zero at $20,000 of private income. When these assumptions are made, a number of cameo households are estimated to make small losses from the impact of the ANTS Option 3 package in particular, age pensioners and disability support pensioners with no private income. It should be noted that these results are contingent upon the estimated impact of the change in tobacco and new house prices being included in the measure of the change in prices facing households. We do not support this as the most appropriate measure of the impact of the tax reform package on prices, preferring instead a measure of price change which excludes the likely impact of tobacco and new house price changes. Option 3B shows the outcomes when this assumption is used and the compensation rate for pensions and allowances is 3.5 per cent (equal to the estimated CPI of 2 per cent plus the Government s buffer of 1.5 per cent). Under Option 3B, age and disability support pensioners are not net losers under the ANTS tax reform package, but their gains are very marginal at low income levels. Option 4 provides estimates of the likely distributional impact of a 70 per cent passing on by businesses of the reductions in their Wholesale Sales Tax and Stamp Duty burdens in the first year of operation of the GST (rather than 100 per cent, as assumed in the preceding options). We consider this 70 per cent figure to be well at the upper end of likely medium term outcomes. Under this set of assumptions, the estimated price effect of the ANTS tax reform package is almost 1 per cent higher. If it is assumed that the Government still increases pensions by only 3.4 per cent then, not surprisingly, some social security recipients make losses from the ANTS tax package. But if the possibly more realistic option of the Government increasing pensions by their stated 1.5 per cent buffer plus the CPI is followed, and if the price effect excludes tobacco and new houses, then essentially only age pensioner couples and self-funded retirees with modest or no private income are net losers from the ANTS package (Option 4B). Options 5 and 6 consider the impact of removing food from the GST base in a revenue-neutral manner. In Option 5, the lost $5 billion of revenue is financed by reducing the generosity of the income tax cuts. In Option 6, the lost revenue is recouped by increasing the GST rate to 12 per cent on the remaining tax base. The removal of food from the GST base particularly assists lower income families and especially the aged. Under Option 5, lower income families are better off and higher income families are worse off than under the Government s original ANTS package (Option 3). Taking food out of the GST base under Option 5 causes a 1.1 percentage point decrease in the CPI. Under such circumstances, it is at least arguable that the Government would not still wish to increase pensions by the planned 3.4 per cent. Accordingly, Option 5B (as with all the other B options) looks at the likely outcomes if pensions are increased by the expected CPI change plus 1.5 per cent. Option 6 does not emerge as a particularly attractive method of addressing concerns about the impact of the ANTS package on lower income groups and families with children. 2

7 Option 7 examines the impact of leaving food in the GST base but increasing the compensation package for social security recipients. This additional compensation is financed via reductions in the generosity of the income tax cuts. The distributional outcomes for Option 7 should be measured against those for Options 3B (the Government s package) and 5B (the food out/tax up package). In all of these cases, the measured impact of the tax reforms excludes any impact on tobacco and new house prices. Broadly speaking, Option 7 delivers higher gains to social security recipients and higher gains to families on incomes above about $60,000 than does Option 5B. It is hard to make generalisations, as the results vary so much by family type, but, on the whole, middle income families make roughly comparable gains under Option 7 and Option 5B. In general, what this study shows is that the distributional outcomes from making food GST-free (Options 5 and 5B) can be reasonably replicated by an alternative approach where food remains in the GST base but compensation is more generous and better targeted (Option 7). 3

8 Package Estim- Measure New Increase Dissaving Marginal Marginal ated of Price Houses in Social Assumed Tax Tax CPI Effect on and Security Rate Rate Impact Hholds Tobacco From From $20k- $50k- $50k $75k 1. Government s ANTS Package 2. Same as Option 1 but tobacco and housing in price effect 3. Same as Option 2 but own price effects and dissaving 4. Same as Option 3 but only 70% pass through WST and Stamp Duties 5. Same as Option 3 but with food out and income taxes up 6. Same as Option 3 but with food out and GST rate at 12% 2.0 CPI Out 3.4% No 30% 40% 2.7 CPI In 3.4% No 30% 40% 2.7 Own In 3.4% Yes 30% 40% 3.6 Own In 3.4% Yes 30% 40% 1.6 Own In 3.4% Yes 33.5% 43% 2.9 Own In 3.4% Yes 30% 40% 7. Same as Option 1 but own price effects, dissaving, pensions up 6%, income tax up and tobacco/housing out 3B. Same as Option 3 but excluding tobacco/housing + pensions up 3.5% 4B. Same as Option 4 but excluding tobacco/housing + pensions up 4.2% 5B. Same as Option 5 but excluding tobacco/housing + pensions up only 2.3% and higher tax cuts 2.0 Own Out 6.0% Yes 30.5% 43% 2.0 Own Out 3.5% Yes 30% 40% 2.7 Own Out 4.2% Yes 30% 40% 0.8 Own Out 2.3% Yes 33% 43% 4

9 1 Introduction In August 1998 the Howard Government outlined a blueprint for tax reform and won the subsequent election. The proposed tax changes and their impact were comprehensively outlined in a report published by the Treasurer (1998) colloquially referred to as the ANTS document (A New Tax System). The impact of the proposed tax reforms is now being assessed by the Senate Committee on a New Tax System. This Committee commissioned this study in an effort to assess the distributional impact of the proposed reforms and to undertake analysis of the sensitivity of key assumptions made by the Treasury in its modelling of the tax package in ANTS. The terms of reference for this consultancy are contained at Appendix A. The STINMOD-STATAX microsimulation model, constructed jointly by NATSEM at the University of Canberra and Professor Neil Warren of the Australian Taxation Studies Program (ATAX) at the University of NSW, was used to undertake this analysis. This model is highly comparable to the model used by the Treasury, PRISMOD. It should be appreciated that both the STINMOD-STATAX and the PRISMOD models are exceptionally complex models that attempt to provide an indication of the likely impact of substantial tax reform packages, rather than absolutely precise measures of outcomes (Warren and Lambert, 1999; Harding and Warren, 1998). This report is divided into two Parts, of which this is the first. Section 2 of this Part describes the tax reform options that the Senate Committee asked to be modelled and explains the reasons why such sensitivity analysis was requested. Sections 3 to 8 describe the outcomes of the modelling. Details of the exact policy parameters used and of the methodology underlying the modelling is contained in Appendix D. Part 2 contains the hundreds of tables which underlie the analysis presented in Part 1. 2 Description of the Options In commissioning this report, the Senate Committee on a New Tax System specified a number of different options for close examination. These are detailed briefly below and elaborated upon in greater detail in the following sub-sections. Option 1. Option 1 is the ANTS cameos replicated, using the same methodological assumptions as adopted by the Treasury Option 2. The impact in July 2000 of a the Government s base package with 100% pass through of both indirect tax cuts and increased costs in the first year assuming (a) a single CPI price effect but including housing and tobacco and (b) disposable income equals consumption 5

10 Option 3. Option 2 but with (a) cameo specific price effects inclusive of housing and tobacco price effects and (b) hypothetical dissavings ratios for different income groups which vary within the cameos Option 4. Option 3 but with 70% pass through of indirect tax cuts in the first year but with 100% of increased costs passed through in one year Option 5. Option 3 with Food GST-Free but with the cost of this change funded through a progressive adjustment to the personal income tax rate schedule Option 6. Option 3 with Food GST-Free and with the cost of this change funded through increasing the GST rate to 12% Option 7. Option 3 but with the cost of increased compensation funded through adjustments to the personal income tax rates and cameo specific price effects exclusive of housing and tobacco price effects The Committee specified that the assumptions about dissavings ratios not be derived from the Household Expenditure Survey (HES), but instead be specified as hypothetical possibilities to illustrate the sensitivity of the results to different savings assumptions. Also, the definition of food used in the study was specified as excluding expenditure on restaurants and takeaway food. The remainder of this section examines the specific detail of these various options. 2.1 Option 1: The Government s Package The Government s tax reform package outlined in the ANTS report included: substantial income tax cuts; the introduction of a broadly based 10 per cent goods and services tax (GST); the abolition of a range of existing indirect taxes, including wholesale sales tax; major reforms in cash transfer payments to families with children; and increases in pensions and allowances to compensate for the expected increase in the prices facing social security recipients. The overall package involved a tax-mix shift, away from income taxes and towards indirect taxes. The distributional impact of this reform package was shown in a series of 25 cameos, which presented the estimated impact of the reforms in July 2000 for hypothetical types of families at various income levels (Treasurer, 1998, pp ). The Government estimated that the ANTS package would increase the Consumer Price Index (CPI) by 1.9 per cent. For reasons explored in more detail below, this excluded the estimated impact of the ANTS package on tobacco and housing prices. It also emerged that this was an estimate of the impact of the tax reform package on prices in (In the longer term the CPI impact would be even further reduced by the impact on prices of the move to 100 per cent GST credits for motor vehicles). 6

11 As a result of transitional factors, the Treasury told the Senate Committee that they estimated that the initial CPI increase in the first year following the reforms would be 2.5 per cent, because FID and business stamp duties are not repealed until January and July 2001 respectively. In the ANTS cameos, the 1.9 per cent CPI impact was modelled. Combined with a promise to compensate pensioners above the CPI effect by a margin of 1.5 per cent, this resulted in an increase in social security payments of 3.4 per cent for households in the long term and 4 per cent on 1 July The impact of the GST upon cameo households was calculated as 1.9 per cent of the disposable income of households. The first option modelled in this report is thus the Government s ANTS tax reform package in but in terms of their impact on July In modelling ANTS we have therefore abstracted from timing issues and assumed that the second year CPI impact applies instantaneously in July This is the approach sought by the Senate Committee on A New Tax System to ensure the results from this report are comparable with those in ANTS (Treasurer, 1998, p ). 2.2 Option 2: New Homes and Tobacco in CPI In deriving the estimated 1.9 per cent effect of the ANTS package on the CPI, the Treasury excluded the impact of changes in the prices of tobacco products and housing. Option 2 proposed by the Senate Committee seeks to examine the impact of the reform package using exactly the same assumptions as the Treasury, but with the single exception that tobacco and new housing price changes be included in the estimate of the effect on the CPI. What are the arguments for excluding tobacco and housing from the estimate of the CPI? Tobacco Tobacco has around a 4.0 per cent weight in the CPI and tobacco is expected to increase in price by 13.3 per cent in ANTS (1998, p 170). Consequently, the CPI effect of this reform alone is estimated by us to be around 0.5 per cent. Lower income earners are more likely to smoke (Harding and Percival, 1997), so there has been some concern expressed at the regressive nature of these tobacco excise changes. Yet many would argue that there are sound reasons for these effects to be excluded from estimates of the price increases caused by the tax reform package. Most important amongst these explanations is the fact that these taxes are designed to distort consumer behaviour and so any notion of compensating those adversely affected by such taxes is an anathema to the basic goal of this tax. The tobacco excise is therefore a Pigouvian tax one designed to correct for the failure of the tobacco market to correctly factor into its pricing the negative externalities associated with tobacco consumption (such as its health effects). Where the distributional impact of such taxes might be an issue is when the level of taxation imposed is far higher than that necessary to correct for market failures. In this context, the distributional impact of such a tax might become an issue. While it could be considered controversial that the distributional impact results in ANTS (1998) excluded the effect of tobacco 7

12 price increases, it is equally controversial to include such effects. The preferred position is probably to exclude the effects of tobacco excise from any distributional impact analysis. Housing The ANTS CPI measure also excluded the effect of ANTS on new house prices (1998, p56). This approach is defensible on two grounds. The first is that current home owners will benefit from new house prices increasing because, since their homes are almost perfect substitutes for new homes, they will benefit from their used homes increasing in value by a comparable amount to new homes. However, if they wish to upgrade their house, the magnitude of the difference will increase by the increase in new home prices. The second defence of the treatment of new home prices in ANTS comes from the proposal in ANTS to introduce a First Home Owners Scheme. The group most affected by excluding new homes from the CPI are first home buyers. Such buyers will not benefit from the increase in existing home prices 1. In response to this problem, ANTS proposes a $7000 lump sum compensation for these effects on first home owners. As was stated in ANTS (1998, p97): If nothing else were done, a GST would raise the prices of new homes by about 4.7 per cent. Although existing home owners would benefit from a rise in the value of their homes, those who do not currently own a home would face higher purchase prices and possibly higher rents. To maintain home affordability at its present high levels, the Government will require the States to assist home buyers through a First Home Owners' Scheme as a condition of receiving GST revenue. People who qualify for assistance will receive a lump sum payment of $7,000 from July Payment will not be means tested. Because it is a flat amount, people purchasing more expensive first homes will receive proportionally less benefit than those buying less expensive homes. People who buy an older house will find that this payment assists them to fund renovations. A first home owner will be more than compensated on the purchase of a home with a construction value (excluding land) of up to $150,000. Good reasons therefore exist to exclude new home price increases from the distributional impact analysis, as was the case in the ANTS report. If the effects on new homes are to be included, then the effects of the First Home Owners' Scheme should also be included. However, this has not been done here, as there is no clear methodology available for distributing the benefits of the proposed First Home Owners Scheme among the cameo households. This report includes distributional impact results where the effect of the GST on new homes is both included (Options 2 to 6) and excluded (Options 1, 3B, 5B and 7) from the CPI. It is important therefore to qualify some of the results in this report. Where the results include the effects of GST on new home prices, the effect of the GST on households is being overstated. This is because the First Home Owners Scheme is not considered in Options 2 to 6 and nor is the positive impact on current owners, 1 An exception to this might be if they are the beneficiaries of bequests whose primary source is residential homes. This could be as a result of bequests from the parents of persons who do not currently own a home. 8

13 whose homes should increase in price in line with new home prices. However, it should be noted that the extent of this overstatement of the CPI is not very great of the order of about 0.2 percentage points. This is because the 12 th Series CPI used to estimate the CPI effects of the tax reform package does not include the impact of the package on new house prices only the impact on renovations is picked up. The extent of possible overstatement will increase greatly with the 13 th Series CPI, which specifically includes new houses within its basket of goods and services (see Appendix C). As a result, caution should be attached to the results which include the effects of new home prices but overlook the benefits to current owners through increases in the price of their current home and the compensation to non-owners through the First Home Owners' Scheme. The group who do appear most likely to lose are those who are renters and who will never be owneroccupiers. 2.3 Option 3: Household Specific Price Effects and Dissaving Two critical assumptions underlie the Government s estimates for cameo households of the net gains from the tax reform package. The first is that it is fair to assume that the prices facing each of the hypothetical households will rise by 1.9 per cent, which is the estimated change in the CPI. It is thus assumed that the price increases facing different types of households will not vary greatly, despite differences in consumption patterns. This is a particularly important assumption in ANTS, as social security pensions are increased by the general CPI effect plus 1.5 per cent. If the general CPI effect does not provide an adequate indicator of the likely changes in prices for pensioners, then there is a possibility that social security recipients are not being adequately compensated for the shift in the tax mix towards consumption taxes. The second critical assumption is effectively that dissaving does not occur or, if it does, it is not a problem which requires addressing. This assumption is implicit in the calculation that the impact of the new indirect taxes will be limited to 1.9 per cent of current disposable income, rather than 1.9 per cent of current expenditure in cases where expenditure exceeds income. Both of these assumptions have been the subject of some controversy, with the result that the Senate Committee requested that this study model a third option, in which the CPI was not assumed to provide an adequate indicator of the possible price changes and the impact of possible dissaving was taken into account. Household-specific price effects The change in the all-households CPI is currently used as the basis for indexing many social security payments, and the Treasury modelling in ANTS assumes that it will provide an adequate measure of the price changes facing social security recipients after the implementation of the ANTS package (1998, p162-3). One potential problem, however, is that the 12 th series CPI does not relate to welfare recipients, as it is constructed from information on wage and salary earner households in the 8 capital cities and excludes the richest 10 per cent of such households. 9

14 Even the new 13 th series does not fully cover pensioners, since it only covers all private households in the 8 capital cities. However, the ABS has indicated that it intends to: Develop analytical indexes specifically designed to measure changes to the cost of living of subgroups in the population. These indexes will be constructed on an outlay basis and will be published at approximately annual intervals. The first such indexes are expected to be published by This should mean that actual household-specific measures of the impact of ANTS on key household groups should be available before the GST is introduced. The assertion in ANTS (1998, p163), which cites an ABS publication as showing that there was almost no difference in the overall price increase each household group faced over a period of almost eleven years, despite prices rising over 120 per cent over that period, will then be able to be tested. It is, of course, not certain that such conclusions will still hold in the context of a tax-mix shift, in which there will be significant relative price shifts. To estimate whether different types of households are likely to face different total price increases, data about the consumption patterns of households are required. The only possible source of such comprehensive information is the Household Expenditure Survey data, issued by the Australian Bureau of Statistics. As many researchers have observed, there are some problems with using the HES data to estimate the price effects for small sub-groups within the Australian population (Carnahan, 1998; Harding and Warren, 1998; Johnson et al, 1998). However, we have estimated price effects for 29 groups within the population, using the HES records. The price effects are an average for all single income unit households that fall within that particular category. In some cases, relatively few households fall within one of the categories requested by the Senate Committee, and the resultant estimated price effects should be treated with extreme caution. (The sample size is shown in Table 1.) In all cases, the relevant average price changes should only be regarded as very broadly indicative. The outcomes for real Australian households within each of the categories may vary greatly, if there is enormous variation in their expenditure patterns. 2 See ABS Information Paper, Introduction of the 13 th Series Australian Consumer Prices Index, 1998, ABS Cat No , p4. 10

15 Table 1 Estimated Number of Households Within Each Cameo Type in the Household Expenditure Survey * Cameo Number Household Type Estimated Population Weight Number in HES Sample 1 single person sole parent 1dep child< sole parent 2dep 1< single income no deps single income 1dep< single income 1dep single income 2dep 1< single income 2dep single income 3dep 1< double income(50:50 income split) no deps double income(50:50) 1dep< double income(50:50) 1dep double income(50:50) 2dep 1< double income(50:50) 2dep double income(50:50) 3dep 1< double income(67:33) no deps double income(67:33) 1dep< double income(67:33) 1dep double income(67:33) 2dep 1< double income(67:33) 2dep double income(67:33) 3dep 1< single age pensioner age pensioner couple single self-funded retiree self-funded retiree couple couple 1dep5-12 with rent assistance single youth allowee away from home disability support couple no deps self employed couple 2 deps 1< * Households within the HES which contain more than one income unit or did not meet the classification requirements for allocation to one of the 29 cameos were not used to calculate the cameo-specific price effects but were included when calculating the all households price effect. 11

16 Dissaving The second area where there has been some controversy over the methodology adopted by ANTS is the implicit assumption that dissaving is not important. A simple example will illustrate what the potential problem is. Suppose an unemployed person has a disposable income of $200 a week but is spending $250 a week (financed by running down savings). Suppose further that the CPI increases by two per cent as a result of a tax reform. The government s methodology assumes that this person requires an extra $4 a week of income to be fully compensated for their new indirect tax burden due to the reform (i.e. 2 per cent of $200). But this ignores the extra $1 that the unemployed person will have to pay to buy the same goods and services as they are currently purchasing out of their $50 of savings. Thus, the net impact of the ANTS package has not been correctly estimated for those who are dissaving. But how do we know who is dissaving and by how much? Unfortunately this is one area where the HES is particularly weak, and the ABS have explicitly cautioned against using the HES to provide an estimate of savings (1995, p. 27). The paper by Carnahan from the Treasury focuses on the limitations of the Household Expenditure Survey data as a source of information on savings by households (1998). This is not a new criticism. In the 1985 Draft White Paper, the Treasury imposed its own savings rates on the various household groups it chose to study. In the Treasury attack on Fightback! though, the Treasury used the HES data, although with some adjustments. In ANTS, past methodologies have been cast aside as apparently flawed and replaced with a better measure (ANTS, 1998, p163) which assumes expenditure equals household disposable income which implies that savings are zero for all households. The problem though is that there is not a clear-cut case for the assertion that this approach is better. While it might provide some indication of the impact in the long term on households who enter with nothing and leave with nothing, any one of a number of other approaches could have been adopted. The fact is households do save (especially those on high incomes), and others do dissave (such as those who are retired or who are experiencing a transitory period of reduced income because, for example, of unemployment). A partial offset for those who are retired is the proposed Aged Persons Savings Bonus, which is a form of compensation for the erosion of the real purchasing power of their savings. However, this compensation is only available to the retired who are aged 60 or more and, even then, it is means-tested. While most of us would not deny that there are problems in using the HES to undertake distributional impact studies, there is no universally accepted or preferred set of assumptions and the Treasury has shown this by the differing approaches it has itself adopted over time. Accordingly, the Senate Committee also asked us to take into account the possible impact of dissaving, so that they could gain some idea of the sensitivity of the distributional results to the assumption made by the Treasury of zero savings. In doing this, we have accepted the Government s position that those who are currently saving should be compensated for the decline in the purchasing power of those savings when they are ultimately spent. Thus, if a single person is earning $1000 a week but spending only $800 a week, we have assumed in the modelling that if prices increase by two per cent then they require a $20 12

17 increase in their income to maintain their position, not an $16 increase in their income. 3 In effect, we have only looked at changing the assumptions for dissaving households, rather than for saving households. This approach is also logical given that the primary goal of the ANTS proposal for an Aged Persons Savings Bonus and a Self-Funded Retirees Supplementary Bonus is to effectively compensate for dissavings by the aged. Given that STINMOD models entitlements to these Bonuses, it is logical to model those factors (dissaving) which have made consideration of such Bonuses necessary. Further, it is important to recognize that groups other than the aged also dissave and part of the governments planned compensation of 1.5 per cent above the CPI is no doubt, in part, designed to account for this fact. The dilemma then is what dissavings rates to assume for lower income households. Given the very tight time frame in which this report has had to be completed, it has not been possible to undertake a comprehensive analysis of the comparative dissavings rates revealed in the HES. As a result, we have simply imposed completely arbitrary dissavings rate assumptions upon different types of households, based upon the savings rates used by Treasury in the 1985 Draft White Paper on Tax Reform (1985, p 259), and then have undertaken some limited sensitivity analysis to demonstrate the impact of changing the dissavings rate assumptions. Tobacco/Housing In or Out? Debate since the ANTS package was released indicates that many groups strongly oppose the Treasury s assumptions of no dissaving and of the CPI providing an adequate measure of the price changes facing different types of households. But there seems to be less consensus about whether the impact of tobacco and housing price increases should be included within the CPI measure used to gauge the impact of the package. Accordingly, we have also modelled Option 3B. This is exactly the same as Option 3 in that it uses household-specific price indexes and allows dissaving but it excludes from the estimate of the price changes the impact on tobacco and housing prices. 2.4 Option 4: 70 Per Cent Pass Through of Price Falls How taxes come to be borne by households is a complex and difficult economic issue. Obviously, the assumptions about how different taxes are shifted through to the ultimate bearers of all taxes households will directly affect the apparent distributional impact of those taxes. In ANTS, it was assumed that the effect of all tax changes flowed immediately through to households. This is the basis of the 1.9 per cent CPI estimate for It is also the basis for the expected CPI increase of 2.5 per cent in July 2000 (after excluding the effect of GST on new home prices and of tobacco excise changes). This 2.5 per cent is different from the CPI estimate because FID is not repealed until January 2001, Business Stamp Duties until July 2001, and the car GST credits are phased in. 3 That is, two per cent of $1000 is $20, while two per cent of $800 is only $16. 13

18 Obviously, the assumption of full shifting forwards to consumers of all indirect tax changes has important implications for the distributional impact of the ANTS package relative to a situation where only partial shifting forward of selected taxes occurs. There has in fact been some concern expressed about the possibility that some businesses might attempt to make windfall gains, by not fully passing to consumers (via lower prices) some or all of the reduction in their existing tax burdens resulting from ANTS. As a result, the Senate Committee requested that this study model a scenario in which only 70 per cent of the expected price reductions for selected indirect taxes were passed onto consumers in the first year. In an effort to gain some insight into the sensitivity of different tax shifting assumptions on the distributional results in ANTS, this study examined the distributional impact of assuming, in the first year, that 100 per cent of all indirect tax changes were shifted forward except in the following cases, where only 70 per cent was assumed shifted forward: (a) repeal of WST; (b) business stamp duties It is important to recognize that this 70 per cent scenario should not be viewed as the standard case. A strong defense can be mounted for the ANTS assumption of 100 per cent shifting forward. In particular, one could cite the powers being given to the Australian Competition and Consumer Commission (ACCC) to pursue businesses who do not fully pass on their tax related price changes. To quote ANTS (1998, p85): The removal or reduction in a number of taxes will moderate the price impact of the GST. For example, the average impact of the wholesale sales tax on consumer prices (as measured by the CPI) is over 4 per cent. This burden is generally hidden from the public by the nature of the wholesale sales tax. Other taxes being abolished or reduced should also act to reduce prices. In general, it can be expected that the competitive pressures that already exist within the economy will ensure that a fall in the tax rate on a product will flow through to consumers in the form of lower retail prices. However, the Government recognises that with the transition to a new tax system, there will be legitimate community concern regarding the possibility of consumer exploitation and excessive profit taking. In measures designed to counter excessive profiteering, the Government will legislate to provide the Australian Competition and Consumer Commission (ACCC) with special transitional powers to formally monitor retail prices. The transitional price oversight regime will begin 12 months prior to the implementation of the GST and will continue for a further 2 years after the date of introduction. The ACCC will be required to monitor retail prices in order to identify instances where consumers have not fully benefited from reductions in the tax rate, or have been exposed to greater than necessary prices rises. The Commonwealth will consult with the States with a view to ensuring the ACCC has statutory authority to take action under its Act against unfair business practices that adjust prices in a manner that is not consistent with changes in tax rates. Breaches of the Act carry penalties of up to $10 million. The ACCC will report to the Government on a regular basis. Those reports will be made publicly available. This aggressive strategy is likely to ensure 100 per cent shifting of the indirect tax changes does act to offset the typical economic argument that the market will take some time for the tax-related price increase to flow through. Therefore, in the case of the ANTS reforms, 100 per cent shifting is probably 14

19 the most reasonable assumption in the medium term. As a result, the 70 per cent WST and Stamp Duties shifting assumption should be viewed as at the extreme of the range of possibilities. In Option 4B, we have assumed, as with the other B packages, that the Government would increase pensions and allowances by the estimated CPI effect excluding tobacco and housing (which is 2.7 per cent see Table 6) plus 1.5 per cent. This would result in a 4.2 per cent increase in pensions. Although this would cost marginally more than the original 3.4 per cent pension increase, we have not changed the income tax scales to fund this higher increase. Thus, unlike the other options examined, this is not a revenue-neutral option. 2.5 Option 5: Food Out and Income Taxes Increased Principally due to concerns about the possible impact of the tax reform package on low income families, a number of organisations have argued that food should be excluded from the tax base of the GST. As Michael Raper from the peak welfare body, the Australian Council of Social Service, argued: One of the key ways to restore some equity to the package is to remove food from the proposed GST (because including food burdens low income people much more than high income people) and to make up the revenue shortfall by reducing some of the proposed income tax cuts (ACOSS, 1999, p. 3). The Senate Committee thus asked us to look at the distributional impact of a package where food was excluded from the GST base and the consequent revenue shortfall was made up by reducing the generosity of the income tax cuts. Expenditure on restaurants and take-away food was not to be defined as food that is, the GST base was still to include restaurants and take-away food. We estimated that this narrowing of the GST base to exclude food would cause a loss of revenue of a little over $5 billion in It is not entirely clear, however, what the most appropriate option to model is when looking at the distributional impact of food-out. One possible scenario, for example, is that pensions would still be increased by 3.4 per cent (as originally planned by the Government) and that the lost $5 billion in revenue would be fully funded via reductions in the generosity of the proposed income tax cuts. To provide an illustration of the possible distributional impact of this option, which we have called Option 5, we used a slightly amended version of the new income tax scales proposed by ACOSS 4. We estimated that the new tax scales would generate additional revenue of about $5 billion, relative to the income tax cuts proposed in the ANTS document. The income tax scales modelled were as shown in Table 2. 4 See Option 1 in Table 1(b) of Making the Tax Package Fairer, ACOSS, Sydney, p14, February

20 Table 2 Income Tax Scales Modelled in Options 5 and 5B Taxable Income Government s Option 5 Option 5B original proposal Marginal Rates Marginal Rates $0-$6, $6,001-$20,000 17% 17% 17% $20,001-$50,000 30% 33.5% 33% $50,001-$75,000 40% 43% 43% $75,001 47% 47% 47% However, with food out the estimated CPI effect of the package, with tobacco and housing included, falls from our original estimate of 2.7 per cent to only 1.6 per cent (ie. by 1.1 per cent). As noted earlier, a strong case can be made for leaving the estimated impact of changes in tobacco and housing prices out of the CPI estimate. In this case, the estimated impact of the ANTS package on the CPI falls from 2.0 per cent with food in to 0.8 per cent with food out (ie. by 1.2 per cent). If a package with food out were actually to be considered, it seems likely that the Government might consider a 3.4 per cent increase in pensions too generous, when the estimated general increase in prices is only 0.8 per cent. Thus, we also modelled an Option 5B. In this Option, pensions were increased by only 2.3 per cent. This maintained the conceptual integrity of the Government s original plan, which was to increase pensions by the general CPI effect (minus housing and tobacco), plus 1.5 per cent. Because the simulated pension increases were lower in Option 5B than in Option 5 (2.3 per cent rather than the original 3.4 per cent) the income tax cuts could be slightly more generous than under Option 5. As shown in Table 2 above, in Option 5B the marginal rate applying between $20,000 and $50,000 was reduced to 33 cents (down from 33.5 cents in Option 5). Overall, therefore, Option 5 simulates a 3.4 per cent pension and allowance increase and the marginal tax rates shown in the middle column in Table 2, while Option 5B simulates a 2.3 per cent increase in pensions and allowances and the marginal tax rates shown in the right hand column in Table It should be noted that, in this and other Options, a range of other possible scenarios could have been validly modelled. For example, another variant in the case of Option 5 would have been to have taken food out (resulting in a 1.6 per cent CPI increase if tobacco and housing were included), and then increased pensions by 3.1 per cent. This would have maintained the Government s intended 1.5 per cent buffer for pensioners (ie. 3.1 per cent equals 1.6 per cent plus 1.5 per cent). But it would have applied that buffer on top of a CPI measure which included tobacco and housing within the CPI measure. If the package was to be kept revenue-neutral, this would also have meant changing the marginal income tax rates. As noted above, we took another possible variant and applied the 1.5 per cent buffer on top of a CPI measure which excluded the impact of tobacco and housing (and also changed the income tax scales accordingly). Overall, therefore, in this and other Option cases, there is no single right answer. We could have modelled a wider range of scenarios, but then the number of tables in Part 2 would have increased to about 500, raising the distinct possibility that the key findings would have been drowned in the static. So we have presented only a subset of the possible modelling scenarios in this report, while accepting that this may leave us open to criticism that every possible counter-factual has not been modelled. 16

21 2.6 Option 6: Food Out and GST Rate Increased The Senate Committee also requested that this study examine a sixth option, where food was excluded from the GST base and the lost revenue was made up via a higher GST rate being imposed upon the remaining tax base. We estimated that a 12 per cent rate would have to be imposed upon the remaining goods and services if food was removed from the base. As in Option 5, restaurants and take-away food were not defined as food. 2.7 Option 7: Food In with Increased Compensation The Senate Committee has relatively little time before it is required to report to the Government. Our analysis of the preceding options suggested that there were likely to be calls to increase the generosity of the compensation package. In an attempt to assist the debate we therefore modelled a seventh option that was not initially requested by the Committee. This tax reform package contained the following features: the impact of tobacco and housing price changes were excluded from the CPI measure; assumed 100 per cent pass through of indirect tax cuts and indirect tax increases in the first year; food was retained in the GST base; pensions and allowances were increased by six per cent and income test thresholds by 2.5 per cent (resulting in an estimated increase in outlays of about $1.1 billion); the proposed changes to family assistance remained as outlined by the Government (as in all the other Options modelled); because self-funded retirees were not benefiting from the pension increases, the Self-Funded Retirees Supplementary Bonus was increased from $2,000 to $4,500 (see Appendix D for details), at a cost of about $210 million; and the generosity of the income tax cuts was reduced by $1.3 billion to fund the higher pension/allowance increases and the higher Savings Bonus. The income tax scales modelled were as shown in Table 3. When comparing the cameo results for Option 7 to other options, it is important to note that we have assumed that the appropriate measure of the price effect of the tax package is one exclusive of tobacco and housing price changes. Thus, the Option 7 results compare the distributional outcomes with no compensation for increased tobacco prices. Conceptually, therefore, the distributional outcomes for Option 7 should be compared with those of Options 3B and 5B. In both of these cases, the estimated distributional impact of the tax package does not include any effect on tobacco and housing prices. This package was designed to be revenue-neutral, relative to the Government s current ANTS package. However, it should be noted that we do not have access to the taxpayer files held by the Australian Taxation Office, and so our estimates of the revenue cost of the income tax changes is not exact. But the proposed changes are well within the ballpark. 17