KE Tax Report__230310 MM900 KPMG Econtech Efficiency Of Taxes Report

User Manual: MM900

Open the PDF directly: View PDF .
Page Count: 178 [warning: Documents this large are best viewed by clicking the View PDF Link!]

KPMG Econtech

CGE Analysis of

the Current

Australian Tax

System

Final Report

26 March 2010

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Inherent Limitations

This report has been prepared as outlined in the Engagement Letter from KPMG Econtech to

Commonwealth Treasury dated 6 February 2009. The services provided in connection with this

engagement comprise an advisory engagement which is not subject to Australian Auditing Standards or

Australian Standards on Review or Assurance Engagements, and consequently no opinions or

conclusions intended to convey assurance have been expressed.

No warranty of completeness, accuracy or reliability is given in relation to the statements and

representations made by, and the information and documentation provided by Commonwealth Treasury

as part of the process.

KPMG Econtech have indicated within this report the sources of the information provided. We have not

sought to independently verify those sources unless otherwise noted within the report.

KPMG Econtech is under no obligation in any circumstance to update this report, in either oral or

written form, for events occurring after the report has been issued in final form.

The findings in this report are subject to unavoidable statistical variation. While all care has been taken

to ensure that the statistical variation is kept to a minimum, care should be taken whenever using this

information. This report only takes into account information available to KPMG Econtech up to the date

of this report and so its findings may be affected by new information. Should you require clarification of

any material, please contact us.

The findings in this report have been formed on the above basis.

Third Party Reliance

This report is solely for the purpose set out in contract and is for Commonwealth Treasury’s information.

This report has been prepared at the request of Commonwealth Treasury in accordance with the terms of

KPMG Econtech’s Engagement letter dated 6 February 2009. Other than our responsibility to

Commonwealth Treasury, neither KPMG Econtech nor any member or employee of KPMG Econtech

undertakes responsibility arising in any way from reliance placed by a third party on this report. Any

reliance placed is that party’s sole responsibility.

Forecasting Disclaimer

In the course of our work, projections have been prepared on the basis of assumptions and methodology

which have been described in our report. It is possible that some of the assumptions underlying our

projections may not materialise. Nevertheless, we have applied our professional judgement in making

these assumptions, such that they constitute an understandable basis for estimates and projections.

Beyond this, to the extent that certain assumptions do not materialise, then you will appreciate that our

estimates and projections of achievable results will vary.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

CONTENTS

Executive Summary 1

1 Introduction 11

1.1 Overview of the Australian Tax System 12

1.2 The Structure of this Report 13

2 Key Literature 14

3 The Cost and Incidence of Taxes 17

3.1 The Excess Burden of Taxation 18

3.2 Using MM900 to Estimate the Excess Burden of Australian Taxes 20

3.3 The Incidence of Taxes 22

3.4 Using MM900 to Estimate the Incidence of Australian Taxes 24

4 The MM900 Model 28

4.1 Over-arching Assumptions 29

4.2 Households 31

4.2.1 Labour Supply versus Leisure 31

4.2.2 Consumption versus Saving 32

4.2.3 Pattern of Consumption 32

4.3 Producers 33

4.3.1 Low and High Skilled Labour 33

4.3.2 Structures and Other Capital 34

4.3.3 Land 35

4.3.4 Natural Resources and Other Generators of Economic Rents 35

4.4 Government Sector 37

4.5 Foreign Sector 38

5 Results 40

5.1 Summary of Excess Burden Results 44

5.2 Summary of Incidence Results 47

5.3 Detailed results 51

5.3.1 Petroleum Resource Rent Tax 51

5.3.2 Land tax and Municipal rates 52

5.3.3 Company Income Tax 54

5.3.4 Resource Royalties and Crude Oil Excise 58

5.3.5 Labour income tax 59

5.3.6 Payroll tax 62

5.3.7 GST 65

5.3.8 Tobacco Excise 68

5.3.9 Alcohol taxes 69

5.3.10 Import duties 71

5.3.11 Luxury Car Tax 72

5.3.12 Fuel excise 73

5.3.13 Motor Vehicle taxes 74

5.3.14 Conveyancing duties 77

5.3.15 Stamp duties other than real property 79

iii

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

5.3.16 Insurance taxes 80

5.3.17 Gambling taxes 81

6 References 83

Appendix A – Detailed Methodology by Individual Tax 85

A.1 Petroleum Resource Rent Tax 85

A.2 Land Taxes and Municipal Rates 87

A.3 Company Income Tax 90

A.4 Crude Oil Excise and Resource Royalties 95

A.5 Labour income tax 97

A.6 Payroll tax 105

A.7 GST 107

A.8 Tobacco Excise 109

A.9 Alcohol Taxes 111

A.10 Import Duties 114

A.11 Luxury Car Tax 115

A.12 Fuel Excise 119

A.13 Motor Vehicle Taxes 121

A.14 Conveyancing Stamp Duties 124

A.15 Stamp Duties Other than on Real Property 127

A.16 Insurance Taxes 129

A.17 Gambling Taxes 131

Appendix B – Modelling Taxes with Partial Coverage 133

Appendix C – Industry Impacts 134

Appendix D – MM900 Parameters 137

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Executive Summary

Background

In May 2008, the Treasurer announced an extensive review into Australia’s tax and transfer

system (the Henry Review). Governments aim to raise tax revenue in a way that meets their

funding needs, while paying regard to the three principles of good tax design:

• simplicity (keeping administration and compliance costs low);

• equity (or fairness, which is partly a subjective judgement); and

• economic efficiency.

KPMG Econtech was commissioned to model the existing tax system against the third aim of

economic efficiency, although the modelling also provides some insights into equity. Most

taxes result in losses of economic efficiency by distorting economic behaviour. For example,

most taxes reduce incentives to work or invest, or distort consumption patterns. This leads to

losses in consumer welfare that can be compared to the amount of revenue that is being raised.

An efficient tax system relies on taxes that result in relatively low losses in consumer welfare

per dollar of revenue raised (excess burden).

With this in mind, the Treasury commissioned KPMG Econtech to undertake a rigorous

economic analysis of the economic costs of the Australian tax system. For this, KPMG

Econtech has estimated the economic inefficiencies, or the excess burdens, that arise from

Australia’s major federal, state and local taxes. This report also examines the economic

incidence of each tax, or how the final burden is shared between various sources of real income

(labour, capital, land rents and other economic rents).

The results for each tax have been estimated using KPMG Econtech’s computable general

equilibrium (CGE) model, MM900. MM900 has been constructed by further developing our

existing MM600 model specifically for this study. MM900 goes well beyond previous

Australian modelling in capturing the economic effects of the tax system on the Australian

economy. It distinguishes 19 different major taxes at the Federal, State and Local levels. For

each tax, it identifies the true tax base as closely as possible, and aims to capture the main

behavioural responses to the tax’s imposition. The modelling also allows for certain negative

externalities in consumption that may justify certain specific taxes.

MM900 contains a fine level of detail. For example, in MM900 the economy produces 889

different products, which represents eight times as much product detail as other comparable

models. This allows the model to more accurately capture the application of certain product-

based taxes. For example, MM900 treats beer, wine and spirits as separate substitutable

products within one broad group. Less disaggregated models aggregate all alcohol products

together, and therefore miss the excess burdens that arise from taxing closely substitutable

alcoholic beverages at different rates.

Another example of the fine level of detail in MM900 is that each of the 109 industries uses up

to six different primary factors (or types of labour, capital and fixed factors). The different

fixed factors include land and natural resources, allowing for much more robust modelling of

the effects of taxes based on the value of land or the value of natural resource use.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

In addition, in comparison to earlier studies, this analysis incorporates a more comprehensive

analysis of the behavioural responses to each tax. In MM900, taxes can cause households to

change their supply of labour and their levels and patterns of spending. Taxes can cause

businesses to change their choices between the six primary factors of production, affecting

employment, investment and valuations of land and natural resources. Finally, taxes can affect

the propensity to import and export each of the model’s 889 products.

These behavioural responses to taxes, by impairing the functioning of the economy, can reduce

consumer welfare. In MM900, these welfare losses are captured appropriately by using a utility

function in which households derive welfare or utility from leisure, saving and consumption of

products, and then deriving all household behaviour from that same utility function. Other

comparable models include more ad hoc elements in modelling household behaviour.

This modelling approach leads to a more robust analysis of the economic inefficiency of

particular taxes as well as their economic incidence. .

Key Results on Efficiency

This analysis shows that some of Australia’s taxes are much more inefficient than others. This

variation is explained mainly by two economic principles. The mobility principle recognises

that the excess burden of a tax is higher, the higher the mobility of its tax base. When a tax is

applied to a highly mobile tax base, that tax base is likely to shrink, distorting economic activity.

The narrowness principle recognises that the excess burden of a tax is likely to be higher, the

narrower the tax base. A narrow tax base may make it possible to respond to a tax by shifting to

untaxed close substitutes. Such shifts add to economic inefficiency and reduce the revenue

yield. Taxes on goods and services with consumption externalities are exceptions to this. For

example, although the base of the tobacco excise is narrow, shifting consumption away from

tobacco entails a social benefit, so tobacco excise is expected to have a low excess burden.

Chart A illustrates the variations in inefficiency between taxes using a selection of six out of the

19 taxes that are modelled. The inefficiency of a tax is measured by its marginal excess burden.

This refers to the effects of a small increase in a tax from its existing level, and is calculated as

the ratio of the loss in consumer welfare relative to the net gain in government revenue. Put

more simply, it is the economic harm expressed in cents per dollar of additional revenue.

The mobility principle is clearly reflected in the modelling results. Petroleum Resource Rent

Tax (PRRT) is modelled to have a marginal excess burden of zero, because it is applied to a tax

base that is assumed to be completely immobile – oil resources. Similarly, municipal rates have

a marginal excess burden of only two cents of welfare loss per additional dollar of revenue

raised because they are applied to land, which is also completely immobile. Municipal rates

only generate an excess burden to the extent that they are applied non-uniformly to different

land users, distorting the pattern of land use.

At the other extreme, company income tax is modelled to have a high marginal excess burden of

40 cents in the dollar of additional revenue, because it is applied to capital, which is highly

mobile. This high mobility is generated by the international competition for funds.

The economic incidence of the remaining three taxes shown in Chart A falls mainly on labour.

Labour has an intermediate level of mobility, and so under the mobility principle these taxes

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

would be expected to have medium excess burdens. However, the narrowness principle leads

their excess burdens to vary.

Payroll tax has a relatively narrow base because, under the small business exemption provision,

it applies to only around one-half of labour income. This provision provides an exemption from

payroll tax on the first tranche of labour income for small and large business alike, undermining

revenue raising, while doing little to reduce disincentive effects. This narrowness of the payroll

tax base leads to a high marginal excess burden of 41 cents in the dollar.

Chart A Marginal Excess Burden of Major Australian Taxes

(cents of consumer welfare loss per dollar of additional revenue)

40 41

PRRT Municipal rates GST Labour income

tax

Corporate

income tax

Payroll tax

Source: KPMG Econtech estimates from MM900

Personal income tax (as applied to labour income) also narrows the tax base, although to a lesser

degree. The progressive rate structure provides an exemption from tax on income earned up to

the tax-free threshold, and beyond that lower marginal tax rates are applied at lower incomes

than at higher incomes. This narrowing of the tax base (compared with a flat rate tax) leads to a

medium marginal excess burden of 24 cents in the dollar of revenue. Of course there are

compelling equity reasons for the progressive nature of the personal income tax scale, but its

efficiency implications should still be understood.

The GST can also be analysed alongside taxes on labour income. Taxes on labour income and

GST both act as a disincentive to supply labour, by reducing the purchasing power of the

additional pay earned from a given amount of additional labour. Labour income tax does this by

removing tax from the additional pay, while GST does this by reducing the purchasing power of

that pay through raising consumer prices. The GST has a relatively broad base, applying to

close to 70 per cent of consumer spending. This breadth helps keep its marginal excess burden

low, at 8 cents per dollar of additional revenue raised. Also contributing to this favourable

result is the gain in the terms-of-trade associated with applying GST to some expenditures of

international tourists.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

These results are closely in line with those found by Johansson et al. in a 2008 study for the

OECD. They used both theory and empirical evidence to conclude that recurrent taxes on

immovable property (particularly residential property) are the “least distortive tax instrument in

terms of reducing long-run GDP per capita” (Johansson et al., 2008, p7). Broad based

consumption taxes, such as the GST, are found to be the second least distortive, followed by

personal income taxes and then corporate income taxes. This is the same as our ranking implicit

in Chart A.

Table A below presents more detail on the relative efficiency of each of Australia’s taxes. It

presents the marginal excess burdens (MEB) and average excess burdens (AEB) for each of the

taxes modelled for this study.

As noted above, the marginal excess burden is the economic harm from a small increase in the

tax, expressed in cents per dollar of additional revenue. The average excess burden is a similar

measure, being the economic harm from introducing the whole tax, expressed in cents per dollar

of additional revenue. The marginal excess burden is useful for considering the impact of small

changes in the tax, while the average excess burden is useful for considering the impact of

abolishing the tax.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Table A: Marginal and average excess burdens of Australian taxes

(cents of consumer welfare per dollar of revenue)

Rating Tax MEB AEB

Low Tobacco excise

a -8 -23

Low Import duties

b -3 -7

Low Petroleum resource rent tax c 0 0

Low Municipal rates 2 1

Low GST 8 6

Low Land taxes

d 8 6

Low Alcohol excise and WET a 9 7

Medium Fuel taxes 15 10

Medium Stamp duties other than real property e 18 18

Medium Luxury car tax

f 20 9

Medium Labour income tax 24 16

High Conveyancing stamp duties g 34 31

High Motor vehicle registration h 37 32

High Motor vehicle stamp duties h, i 38 38

High Corporate income tax 40 23

High Payroll tax 41 22

Very High Insurance taxes 67 47

Very High Royalties and crude oil excise 70 50

Very High Gambling taxes j 92 54

Source: KPMG Econtech’s MM900 model estimates

Notes to Table 5.1:

a. The excess burden for tobacco excise and alcohol tax will be influenced by the assumptions for externalities

associated with tobacco and alcohol consumption.

b. The low (negative) excess burden estimate for import tariffs is an under-estimate in some contexts. It refers to the

efficiency of import tariffs post the tariff reductions scheduled for 1 January 2010. Those reductions mean that

all import tariffs will be five per cent or less, apart from a tariff of 10 per cent on clothing. These rates are below

a theoretical ‘optimal’ tariff of around 11 per cent (based on average export demand elasticities in MM900 of

around -10), so such low rates of tariff are not distorting. However, Australia has had much higher import tariffs

in the past, which were highly distorting. Further, the notion of an optimal tariff ignores the risk that other

countries will impose tariffs in retaliation, leading to welfare losses for Australia.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

c. The zero excess burden for the PRRT rests on the assumption that it is designed so that it only taxes the excess

profits of petroleum extractors, which they derive from access to a natural resource which is in limited supply.

d. The estimate of the excess burden for land tax does not take into account that the rate structure is progressive

because of the exemptions for land holdings under a certain threshold value and an increasing tax rate for higher

valued land holdings. Also, the data for land tax collections by industry was highly aggregated and does not

reflect the full extent of the variability of land tax paid by industry. The excess burden may therefore be an under-

estimate.

e. The excess burden estimates for stamp duties other than on real property are likely to be under-estimates because

there are a number of distortions from these stamp duties that are not readily amenable to CGE modelling. For

example, these stamp duties may reduce the frequency of transactions.

f. The excess burden estimates for luxury car tax are likely to be under-estimates, because the modelling does not

capture the point that the luxury car tax distorts the choice between luxury and non-luxury cars.

g. The excess burden estimates for conveyancing duties have a downward bias because there are a number of

distortions from conveyancing duties that are not readily amenable to CGE modelling, such as the distortion

between renting and buying housing.

h. The excess burdens for motor vehicle registration and motor vehicle stamp duties are only for the proportion of

those taxes paid by businesses. This is because motor vehicle taxes paid by households do not appear in input-

output tables used to construct the model. The impact that this has on the size of the excess burden is ambiguous.

i. The excess burden estimates of motor vehicle stamp duties are likely to be under-estimates because there are a

number of distortions from motor vehicle stamp duties that are not readily amenable to CGE modelling. For

example, motor vehicle stamp duties may lead to less frequent motor vehicle transactions.

j. The excess burden estimates for gambling taxes are likely to be over-estimates if there are negative externalities of

gambling, which have not been taken into account.

k. The excess burden of state taxes (excluding land tax and resource royalties) have a conservative bias, because any

economic cost arising from differences in tax regimes between the states has not been taken account of in the

modelling.

Key Results on Incidence

While excess burdens give information about the size of the overall cost to the economy of

taxation, the question of who actually bears the burden of taxes is also important. In

considering the burden of taxes, it has long been appreciated that there is an important

distinction between who has the liability to government for a tax, and therefore carries its

impact, and who bears its final burden after all economic adjustments, and therefore bears its

incidence. It is the incidence of a tax, not its impact, which matters for economic analysis, and

often they are different.

The concept of mobility of a tax base is important for determining the incidence of a tax, just as

it is important for determining the efficiency of a tax. Highly mobile primary factors of

production are unlikely to bear much of the final incidence of a tax. Rather, they are likely to

partly move to lower taxed alternative uses. This partial withdrawal of supply from the original

use generates a shortage that is likely to push up its price for that use, partly or wholly offsetting

the tax impost on the mobile factor. The incidence of the tax will then be passed on to other

factors of production. In contrast, completely immobile factors of production have no scope for

passing on taxes, so when they carry the impact of a tax, they will also bear its final incidence.

The importance of the mobility principle can be seen in Table B below, which shows the final

incidence for each of the 19 taxes under reference. The results in the table identify where most

of the incidence of each tax falls, based on model simulations. The simulations involved

abolishing a tax, and observing the impacts on consumer prices and different sources of private

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

income. The full numerical results of the incidence analysis are contained in the main body of

the report.

Each of the columns in Table B corresponds to the different ways that the incidence of a tax can

be borne. For example, a tax may reduce the nominal income from any of the primary factors of

production: labour; land; other fixed factors; or capital. Or it may also be transmitted into

higher prices, reducing real incomes.

Table B reveals that none of the taxes have their main incidence falling on capital. This result

comes about because of the highly internationally mobile nature of capital. The mobility of

capital means that rather than accepting any burden of the tax, capital will move to alternative

uses in which the prevailing global after-tax rate of return can be obtained.

Table B shows that the final incidence of a large number of taxes falls mainly on consumer

prices. These taxes can be divided into two categories. The first is the group of taxes which are

levied on consumption directly. This group includes the GST, excise duties and taxes on fuel,

insurance and gambling. In this case, the party carrying the impact of the tax, consumers, is the

same as the party bearing its final incidence. The second category is taxes that are not levied on

consumption directly, yet have their final incidence on consumer prices. Thus, the party

carrying the impact of the tax is different to the party bearing its final incidence. These taxes

include payroll tax and business motor vehicle taxes.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Table B Final Incidence or Burden of Each Major Tax (Long run)

Tax labour land other rents capital

Tobacco excise X

Import duties X

Petroleum resource rent

tax X

Municipal rates X

GST X

Land taxes X

Alcohol excise and WET X

Fuel taxes X

Stamp duties other than

real property X

Luxury car tax X

Labour income tax X

Conveyancing stamp

duties X

Motor vehicle

registration X

Motor vehicle stamp

duties X

Corporate income tax X X X

Payroll tax X

Insurance taxes X

Royalties and crude oil

excise X X

Gambling taxes X

Source: KPMG Econtech estimates from MM900

Note: The table shows the party that bears the main portion of the final incidence for each tax.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Such taxes that increase consumer prices can be thought of as having their final incidence on

labour. Higher consumer prices decrease the purchasing power of the wage, reducing the return

to work. Similarly, labour income tax (i.e. personal income tax applied to labour income)

reduces the incentive to work, but through the different channel of reducing the after-tax wage.

Therefore, the results from Table B show that labour bears the main incidence for the majority

of Australian taxes – either directly through labour income or indirectly through consumer

prices.

Labour bears much of the final incidence of Australian taxes because, in contrast to capital,

labour is relatively immobile. For example, payroll tax has its impact on employers, but its final

incidence is on labour. Payroll tax increases the cost of employing labour, and since prices are

equal to the marginal cost of production, firms must pass this cost on. This will either be

through decreased nominal wages or increased consumer prices. In MM900, firms pass the cost

of payroll tax on, in the form of higher consumer prices1. This, in turn, reduces the real wage

that labour receives, and in response, labour supply falls. This fall creates a shortage of labour

that will push up the wage that employers must pay. However, households’ supply of labour

has a low responsiveness to the after-tax real wage, and so the fall in labour supply is relatively

small. Thus, the increase in the wage offsets only a small part of the initial increase in

consumer prices, and labour is left to bear the main burden of payroll tax in the form of higher

prices and lower real wages.

Land and other fixed factors also bear the burden of a number of taxes. These factors of

production are on the opposite end of the spectrum to capital, with their supply completely

fixed. When a tax is applied to any of these fixed factors, such as petroleum resource rent tax to

oil resources, municipal rates or land tax to land, the supply does not fall in response to the

lower rate of return. With no change in supply, the pre-tax price of fixed factors will not

change. Instead, the after-tax return that owners of the fixed factors are able to receive falls by

the full amount of the tax.

Company income tax has a more complex incidence. Company income tax is applied to profits,

or return to capital, land and other fixed factors, spreading its impact three ways. The fixed

supply of land and other fixed factors means that they will bear the full incidence of the

company tax that is applied to them, as described above. However, capital will not bear the

incidence of company tax that is applied to it. Instead, the supply of capital will fall until the

increase in its pre-tax return fully offsets the increase in company income tax. This process may

take several years so that returns to capital fall initially, but ultimately the higher cost of capital

will be passed on to consumers in the form of lower real wages. In MM900, this is transmitted

through higher prices2. This means that part of the final incidence of company income tax is

borne by labour. It is not borne by capital because it is highly mobile, at least in the long run.

Resource royalties and crude oil excise are applied to the output of mining industries rather than

to the profits derived from the natural resources used in mining industries. However, these

natural resources, as fixed factors, still bear some of the incidence of these taxes. In the same

way as company tax, the mobility of capital means that it will not bear the incidence of resource

1 The incidence of payroll tax is transmitted through higher prices rather than a lower nominal wage because the

nominal wage is the numeraire in MM900. However, the choice of numeraire does not affect the real outcome of the

model, which is that the real wage falls in response to a payroll tax and so labour bears the final incidence.

2 The incidence of company income tax is transmitted through higher prices rather than a lower nominal wage

because the nominal wage is the numeraire in MM900. However, the choice of numeraire does not affect the real

outcome of the model, which is that the real wage falls in response to a company income tax and so labour bears part

of the final incidence.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

royalties and crude oil excise. When some of the capital in mining industries is withdrawn, this

reduces the productivity of natural resources, leading to lower rents for those natural resources.

The higher cost of capital also flows through into consumer prices, lowering the real wage and

implying that labour also bears some of the incidence.

In summary, while the impacts of our taxes are widespread, most of their final incidence falls on

labour through a reduction in the real wage either directly in the case of labour income tax, or

indirectly through higher consumer prices. The remaining incidence falls on the fixed factors of

land and natural resources. However, these fixed factors are ultimately owned by households

either directly, or through the corporate veil as shareholders. So in the end consumers bear most

of the burden of the tax system.

Since consumers bear the final burden of virtually all taxes, an informed approach to tax design

should not focus superficially on who carries the initial impact. Rather it should focus on the

three principles of good tax design – efficiency (choosing taxes that keep excess burdens low –

see Table A); equity (but with judgements based on who bears the final burden of a tax – see

Table B); and simplicity.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

1 Introduction

In May 2008, the Treasurer announced an extensive review into Australia’s tax and transfer

system (the Henry Review). Governments aim to raise tax revenue in a way that meets their

funding needs, while paying regard to the three principles of good tax design:

• simplicity (keeping administration and compliance costs low);

• equity (or fairness, which is partly a subjective judgement); and

• economic efficiency.

KPMG Econtech was commissioned to model the existing tax system against the third aim of

economic efficiency, although the modelling also provides some insights into equity. Most

taxes result in losses of economic efficiency by distorting economic behaviour. For example,

certain taxes reduce incentives to work or invest, or distort consumption patterns. This leads to

losses in consumer welfare that can be compared to the amount of revenue that is being raised.

An efficient tax system relies on taxes that result in relatively low losses in consumer welfare

per dollar of revenue raised (excess burden).

With this in mind, the Treasury commissioned KPMG Econtech to undertake a rigorous

economic analysis of the economic costs of the Australian tax system. For this, KPMG

Econtech has estimated the economic inefficiencies, or the excess burdens, that arise from

Australia’s major federal, state and local taxes. This report also examines the economic

incidence of each tax, or where the final burden falls – which may be on consumer prices, or

various types of income (labour, capital, land rents and other economic rents).

The results for each tax have been estimated using KPMG Econtech’s computable general

equilibrium (CGE) model, MM900. MM900 has been constructed by further developing our

existing MM600 model specifically for this study. MM900 goes well beyond previous

Australian modelling in capturing the economic effects of the tax system on the Australian

economy. It does this by distinguishing 19 different major taxes at the Federal, State and Local

levels. Further, for each tax, the model identifies the true tax base as closely as possible, and

aims to capture the main behavioural responses to the tax’s imposition.

Some of the important features of MM900 are listed below.

• MM900 contains 109 industries producing 889 products. This is around eight times as

many products as other models, so that economic impacts of selective taxes on individual

products – such as fuel, alcohol, tobacco and gambling taxes – can be captured more fully.

• MM900 has a more detailed treatment of primary factors. The model recognises not only

capital and labour, but also land and natural resources as primary factors of production in

each industry. This allows for more appropriate modelling of the economic impacts of taxes

on land (land taxes and municipal rates) and taxes on natural resources (the existing

petroleum resource rent tax and other potential resource rent taxes).

• In MM900, particular attention has been paid to the structure of taxes. For example, the

modelling of the application of payroll tax to labour inputs takes into account the economic

impacts of the small business exemption. The modelling also recognises that the

progressive nature of personal income tax adds to work disincentives (while accepting the

compelling equity argument for such progressivity).

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

• The modelling allows for negative externalities that may justify certain narrowly-based

taxes. These include negative externalities from consumption of alcohol and tobacco.

• The GST is modelled in fine detailed in MM900. That is, MM900 identifies the individual

GST treatment for each of the 889 products - taxable, GST-free or exempt.

1.1 Overview of the Australian Tax System

There are currently around 125 taxes in Australia levied at the national, state and local

government level (Treasury, 2008). These taxes have two broad aims.

• First, taxes are levied to raise government revenues to meet government spending

requirements, which is often referred to as the revenue-raising requirement.

• Second, some taxes are levied to alter the behaviour of firms and households in a welfare-

improving way. Alcohol and tobacco excises are examples of this.

The structure of the Australian tax system affects the operation of businesses, households and

governments. Therefore, Australia’s tax-transfer system forms an integral part of the economic

structure – through its influence on decisions of: saving; consuming; investing; and working.

The combined revenues of federal, state and local governments in Australia was around $350

billion in 2007/08, or more than 30 per cent of the national Gross Domestic Product (GDP).

Chart 1.1 shows that Australia’s tax mix is heavily reliant on income tax revenue from both

businesses and individuals, with these taxes making up 55 per cent of total revenue. The next

most important revenue source is GST, accounting for a further 13 per cent. The efficiency of

these taxes will therefore be important for assessing the overall efficiency of our tax system.

Chart 1.1 Government Revenue Composition

36%

19%

13%

12% Personal income tax

Company income tax

GS T

Excises and levies

Payroll tax

Property taxes

Taxes on financial and

capital transactions

Other taxes

Source: ABS Taxation Revenue, Australia, 2007/08, cat. no. 5506.0, Canberra.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

1.2 The Structure of this Report

This report is structured as follows.

Section 2 sets the context of this report by comparing its scope and methodology to other

studies into the efficiency and incidence of taxes.

Section 3 describes costs of Australian taxes and who pays them. It explains the concepts of the

excess burden and incidence of taxation, and how they are measured in this report.

Section 4 describes the key features of MM900 that makes it the most appropriate tool to

conduct this comprehensive evaluation.

Section 5 presents the results and a discussion of the excess burdens and incidence of each tax.

Section 6 includes the references used in the report.

Appendix A provides details on the modelling methodology for each tax individually. For each

tax, this section includes:

a description of the tax;

a literature review of previous studies into the impact of this type of tax;

theoretical and modelling issues pertinent to estimating the excess burdens and incidence of

the tax; and

considering these issues, it then describes KPMG Econtech’s modelling approach.

Appendix B includes technical details on modelling taxes with partial coverage.

Appendix C provides tables identifying the industries most impacted by each tax.

Appendix D provides additional detail in relation to the MM900 model.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

2 Key Literature

There have been a number of theoretical and applied studies surrounding the excess burden and

incidence of different taxes. This literature analyses a broad range of taxes in a number of

countries, and helps to develop an idea of the likely excess burden of the major Australian taxes.

This study goes further than any previous study done in the Australian context. It analyses 19 of

the major Australian taxes in a consistent framework, allowing policy makers to easily compare

their economic costs. It is also a more detailed modelling than has been undertaken in the past,

with the model identifying the base of each tax to a high level of precision, and a large range of

behavioural responses are taken into account.

In a 2008 study for the OECD, Johansson et al. examine the efficiency of a number of taxes

commonly used among OECD countries. They determined a ranking of taxes that is similar to

the results of this study, although their conclusions are necessarily less specific because of the

cross-country nature of the study. They conclude from theory and empirical evidence that

recurrent taxes on immovable property (particularly residential property) are the “least distortive

tax instrument in terms of reducing long-run GDP per capita” (Johansson et al., 2008, p7).

Broad based consumption taxes, such as the GST, are found to be the second least distortive,

followed by personal income taxes and then corporate income taxes. These findings are

summarised in Table 2.1 below.

Table 2.1 Tax Ratings from OECD study(a)

Rating Tax

Low Immovable property tax

Low Consumption taxes

Medium Personal income taxes

High Corporate income taxes

(a) ratings by KPMG Econtech based on rankings from the OECD study

These rankings give the same qualitative results of the relative efficiency for each tax as the

estimates made with MM900 in this report.

A range of both state and local government level taxes are investigated in a 1998 report by the

Productivity Commission (PC). The PC report includes a partial equilibrium assessment of the

efficiency costs (marginal excess burdens) of a number of state taxes. The partial equilibrium

approach focuses only on the impact of the tax in one market – the market in which the tax is

levied. This means that the effects of a tax in one market on the consumption and tax

collections in other markets will not be taken into account. Therefore, such estimates of

efficiency costs of taxes will be incomplete. For example, as a tax on labour, payroll tax may

reduce employment which in turn will reduce income derived from labour. As consumption

falls in line with incomes, tax collections from consumption-type taxes will also fall. This

second-round loss of tax revenue is an additional cost of payroll tax, which a partial equilibrium

estimation will not capture. Conversely, the computable general equilibrium (CGE) model used

in this study means that all second round effects are taken into account.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The Centre for International Economics (CIE) conducted a qualitative analysis of state taxes,

focusing on those levied on businesses (CIE, 2009). The report identifies state taxes as having

high deadweight losses, partly because of the narrow bases of some state taxes, such as payroll

tax. Rather than assessing each tax individually, as is done in this study, the CIE study

estimates the economic impact of introducing a number of tax reforms concurrently. Using the

MMRF model, they model the impacts of reducing a number of state taxes and replacing the lost

revenue through an increase in the GST. The CIE modelling finds that this tax policy switch

leads to an increase in GDP and consumption. This indicates that the package of state taxes has

a higher economic cost than the GST. However, the results of the CIE study do not allow an

assessment of the relative inefficiencies of each individual tax.

The CIE findings are broadly in line with the results of the current analysis, which finds that

GST has a lower economic cost per unit of revenue raised than many state taxes. However,

features of the MMRF model used by CIE will result in some differences between the results of

the CIE study and this study. For example, the modelling of payroll tax and land tax is more

sophisticated in the MM900 than the MMRF. MM900 treats land as a fixed factor, as explained

in section 4.5.3., including distinguishing between residential, urban and rural land, and has a

detailed modelling of the payroll tax, as explained in Appendix A, section A.2.

Access Economics has also conducted a CGE analysis of states taxes, for the Financial Industry

Council of Australia (2008). Their rankings for the efficiency of these taxes are included in the

table below.

Table 2.2 Tax Ratings from Access Economics study(a)

Rating Tax

Low Municipal rates

Low Land Tax

Low Gambling tax

Medium Payroll tax

Medium Conveyancing Duty

Medium Insurance tax

Medium Motor Vehicle tax

(a) ratings by KPMG Econtech based on rankings from the Access Economics study

Similar to the results in this study, Access Economics find a low excess burden of taxes on land.

The remainder of their excess burden rankings are low compared to this study.

• While the current report does not include the consideration of any potential externalities

from gambling, the Access Economics report does. This means that the Access Economics

ranking for gambling taxes is much lower.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

• The more sophisticated treatment of payroll tax in the MM900 model includes the impact of

the small business exemption, which contributes to the excess burden by reducing the

revenue raising ability of the tax, as discussed in section 5.3.

• The lower excess burden ranking for conveyancing duty, insurance tax and motor vehicle

tax reflects the smaller amount of choices that are available in the model used by Access

Economics, the Monash Multi-Regional Forecasting Model. MM900’s more comprehensive

coverage of economic decisions improves the estimates of excess burdens, as discussed in

Section 4.

Baylor and Beauséjour (2004) use a CGE model of the Canadian economy to estimate the

marginal excess burden of a number of different taxes. Their results are broadly in line with

both the literature and this study. They estimate that corporate income tax has a high economic

cost, at 37 cents per dollar of tax revenue. This is followed by personal income tax, at 32 cents

per dollar of tax revenue. Consumption taxes have the smallest economic costs, at 13 cents per

dollar of tax revenue. Baylor and Beauséjour’s estimation of the cost of payroll tax (15 cents

per dollar of tax revenue) is in line with our estimate of the cost of a payroll tax if the small

business exemption were removed, which again is discussed in section 5.3. A further

comparison of Baylor and Beauséjour’s results with the KPMG Econtech estimates is contained

in chapter 5 of this report.

Studies focused on estimating the economic costs of particular taxes are also widely available.

The results of some of these studies are compared with the results in this study in Section 5.

However, estimating the efficiency of all taxes in a consistent framework, rather than using a

series of separate studies, will be the most appropriate way to compare the relative economic

costs of different Australian taxes.

This study therefore builds on the work of the authors mentioned above, and forms the most

extensive study of the efficiency of Australian taxes to date. The contributions of this report to

the literature are summarised briefly below.

• This study uses a computable general equilibrium model, giving quantitative results which

are consistent with qualitative research that has been undertaken by the OECD.

• The general equilibrium nature of the model used takes into account both how all markets

within the economy interact as well as the constraints on the macro-economy.

• The separate analysis of each tax allows policy makers to compare all taxes on a consistent

basis.

• It is the most thorough analysis done to date, in capturing where taxes impact on the

economy, and how it responds.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

3 The Cost and Incidence of Taxes

As background to the results on the efficiency and incidence of each tax presented in section 5,

this section explains the importance and meaning of the concepts of tax efficiency and

incidence, the main drivers, and how they have been modelled.

For any tax, it is usually straightforward to assess the amount of revenue that is collected and

from whom. But that only gives a superficial view of the tax system. Most taxes distort

economic incentives giving rise to excess burdens – economic costs over and above the amount

of revenue that is raised. Further, taxes may lead to economic responses such as price increases

that mean that the party who bears the economic incidence or final burden of the tax, is

different from the party who pays the tax. In assessing the merit of any tax, understanding its

economic incidence and excess burden is fundamental. This section explains these concepts and

how they have been modelled.

This study estimates the excess burdens arising from Australian federal, state and local taxes.

Most taxes carry some excess burden. For example, certain taxes reduce incentives to work or

invest, or distort consumption patterns, adversely impacting on economic performance and

ultimately on consumer welfare. This excess burden can be compared to the amount of revenue

that is being raised to assess the degree of economic inefficiency that each tax entails.

The study also identifies the incidence of (or “who really pays”) each tax. The party paying a

tax, and so carrying its impact, is sometimes different from the party who actually bears the

final burden or incidence of a tax, after economic responses. For example, while company tax is

collected from businesses, a business may pass the burden of the tax on to households through

higher prices or lower wages. In that case, while the business would carry the impact of the tax,

households would bear its economic incidence.

Understanding the incidence of a tax is an important step in assessing its implications for equity.

However, further analysis is also required, linking the incidence of a tax to the situations of

different types of households. This distribution analysis is outside of the scope of this study.

Further, good tax design is not only about economic efficiency and equity of taxes, but also their

simplicity. Taxes are considered to be simple when they involve low administration costs for

tax collectors and low compliance costs for taxpayers. An assessment of the simplicity of each

tax is also outside of the scope of this study.

MM900, KPMG Econtech’s CGE model of the Australian economy, was used to estimate the

excess burden and incidence of key Federal, State and Local taxes. The excess burden results

show, for every dollar in tax revenue raised, the additional cost to consumers in excess of that

dollar. The incidence results show how the economic incidence of each tax is spread between

consumer prices and various types of income (labour, capital, land rents and other economic

rents).

The remainder of this section discusses in further detail the excess burden and economic

incidence of taxes and how these have been modelled. Section 3.1 looks more closely at the

excess burden of taxation, section 3.2 discusses how excess burdens have been modelled,

section 3.3 discusses the economic incidence of taxes, and section 3.4 explains how incidence

has been modelled.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

3.1 The Excess Burden of Taxation

Most taxes change the behaviour of households, firms and/or the foreign sector. These changes

in behaviour create distortions in economic activity. Taxation may influence behaviours in the

following ways.

• Households may change their level of consumption, their consumption patterns or the

amount that they work.

• Firms may change what they produce or how they produce it.

• The foreign sector may demand less of Australia’s exports or supply Australia with fewer

imports.

The “excess burden of taxation” is a measure of the economic costs associated with these

distortions. More specifically, the excess burden is defined as the reduction in welfare3 from

imposing or raising a tax, after the additional tax revenue has been returned to individuals as a

lump-sum payment. The higher the tax rate, the higher will be the reduction in welfare.

Moreover, if the tax rate increases, the excess burden increases at a greater rate. In fact, the

excess burden of a tax is generally roughly proportional to the square of the tax rate4. Figure 3.1

below illustrates the excess burden of a tax in a market that has variable producer and consumer

prices.

Figure 3.1 Excess Burden of a Distorting Tax

Figure 3.1(a) shows a partial equilibrium model with no taxes. In this case, an equilibrium will

be reached at the point where the demand and supply schedules intersect (at Q*). Here, the

marginal value of the last unit consumed is equal to the marginal cost of the last unit supplied.

3 Technically speaking, welfare is defined as the collective level of utility of Australian households, where the utility

of each household is determined by the commodities and leisure they consume.

4 This is because if, for example, the tax rate is doubled, then it is likely that not only will the loss in economic

activity approximately double, but also the gap between the marginal value and marginal cost of that activity will also

approximately double.

Quantity

Price

Quantity

Q* QT Q*

Demand

Supply

(

)

No Tax

(

)

Distortin

Tax

Supply

Demand

q = p + t

q* = p*

DWL

Price

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

When a tax is imposed, as in Figure 3.1(b), it raises the price paid by consumers and lowers the

price received by producers. This drives a wedge between the marginal value and the marginal

cost and, consequently, reduces market activity. The size of this wedge, the Deadweight loss

(DWL) shown in Figure 3.1(b), is the loss to consumer and producer surplus that is not

recouped as tax revenue. This can be thought of as the welfare loss associated with the tax.

The implication is that, in practice, raising a dollar of tax revenue usually creates a welfare loss

(a positive excess burden). This report expresses the excess burden (EB) per unit of additional

tax revenue, as shown in Box 3.1.

Box 3.1: Definition of Excess Burden

revenuetax

welfare

−≡EB

Where:

(a) the change in welfare is defined as the amount of (lump sum) compensation required after

the tax change to restore the consumer’s utility back to its original level.

(b) the change in tax revenue from the tax increase is defined as the amount of the lump sum

transfer back to consumers financed by the tax increase. This is equal to the total change in tax

revenue, including any changes in revenue stemming from changes in other markets.

As indicated above, the change in welfare is the compensation required to restore the

consumer’s utility back to its original level. For this purpose, it is assumed that the revenue

raised by the tax is returned to the consumer as a non-distorting or lump sum transfer, so that the

consumer does not directly lose any income from the change in the tax. This means that the

change in welfare only arises from the distortions to economic behaviour generated by the tax.

This measure of the change in welfare is known as the compensating variation (CV). In the

MM900 model, the CV can be interpreted as the real change in consumption5, saving and leisure

enjoyed by the consumer.

The change in tax revenue from the tax increase is defined as the amount of the lump sum

transfer to consumers that would be financed by the tax increase, while keeping the budget in

balance, and given fixed government expenditures. This will reflect both the additional revenue

from the tax whose rate is increased, as well as any second round effects on tax revenues (due to

impacts on tax bases).

There are two types of excess burden that can be measured.

• The marginal excess burden (MEB) is defined as the additional welfare loss imposed by

increasing a particular tax by a small amount, divided by the change in government revenue.

The MEB is useful for considering small reforms of the current system or ways of raising

additional revenue within the existing tax framework.

5 The model uses a generalised linear expenditure system utility function, in which households derive utility from

consumption of products, saving and leisure. It specifies an essential level of consumption for each product. The

consumer then maximises their utility by choosing how to allocate the remainder of their full income.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

• The average excess burden (AEB) is defined as the total welfare loss from imposing a

particular tax, expressed as a proportion of total revenue raised by that tax. These measures

are useful for considering the abolition of a certain tax.

A concept related to the excess burden is the marginal cost of funds, which is one plus the

marginal excess burden. This gives the cost to the economy of raising an additional dollar in tax

revenue, if the revenue were not returned to consumers. The marginal cost of funds exceeds one

dollar because it takes into account that raising a dollar of revenue also entails an excess burden.

To be worthwhile, the value to the community of a public spending program must exceed the

marginal cost of funds.

3.2 Using MM900 to Estimate the Excess Burden of Australian

Taxes

Estimating the Marginal Excess Burden and Average Excess Burden for each Tax

The MM900 model has been constructed so that both the change in welfare and the change in

tax revenue are standard outputs of the model. Thus, MM900 estimates the numerator and

denominator of the excess burden formula. This allows the excess burden of each tax to be

calculated and compared (see section 5 of this report). As noted above, MM900 uses a measure

of the change in welfare known as the compensating variation (CV), while the change in tax

revenue is measured by the transfer payment to consumers that can be financed from the tax

change, while keeping the budget in balance.

For each tax, two versions of the excess burden have been simulated in MM900. Specifically,

the Marginal Excess Burden (MEB) is estimated by simulating in MM900 a small

(five per cent) increase in the tax being considered and performing the excess burden

calculations. The Average Excess Burden (AEB), on the other hand, is estimated by simulating

in MM900 the abolition of the tax being considered, and performing the excess burden

calculation.

When a tax is imposed, it impacts upon behaviour not only in the taxed market, but also in other

related markets. For example, in the case of two goods that are gross substitutes, the imposition

of a tax on one good will cause consumers to reallocate their consumption spending from the

taxed good towards the untaxed good. CGE models like MM900 are designed so as to pick up

these ‘related market effects’. CGE models also identify other second round effects, such as

changes in aggregate consumption due to changes in consumers’ income. Related market

effects impact on the MEB and the AEB when the tax affects activity and tax collections in

other distorted markets. The features of MM900 relevant to modelling the excess burden of

taxes are described in Section 4.

Excess Burden in MM900

Some of Australia’s taxes are thought to be more inefficient than others. This is because of two

economic principles. The mobility principle recognises that the excess burden of a tax is higher,

the higher the mobility of its tax base. When a tax is applied to a highly mobile tax base, that

tax base is likely to shrink, distorting economic activity. The narrowness principle recognises

that the excess burden if a tax is likely to be higher, the narrower is the tax base. A narrow tax

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

base may make it possible to respond to a tax by shifting to untaxed close substitutes. Such

shifts add to economic inefficiency and reduce the revenue yield.

Under the mobility principle, resource rent taxes and land taxes would be expected to have low

excess burdens. This is because they are assumed to be applied to immobile tax bases – natural

resources and land. MM900 treats both land and natural resources as completely immobile. At

the other extreme, company income tax is expected to have a high marginal excess burden,

because it is applied to capital, which is highly mobile. This high mobility is generated by the

international competition for funds. In fact, in MM900, capital is treated as perfectly mobile.

Labour has an intermediate level of mobility, and so under the mobility principle taxes can fall

on it, such as payroll tax or personal income tax as applied to labour income, would be expected

to have medium excess burdens.

However, the narrowness principle adds another dimension to the analysis. Payroll tax has a

relatively narrow base because, under the small business exemption provision that is modelled

in MM900, it applies to only around one-half of labour income. On the other hand, the GST has

a broader base, applying to close to 70 per cent of consumer spending in MM900. Narrowing a

tax base tends to make a tax less efficient.

Estimating the Excess Burden from Non-Uniformity

A tax may have a high excess burden for one of two reasons. First, it may have a high excess

burden because of its inherent nature, which might indicate that the tax should be abolished. On

the other hand, it may have a high excess burden because it is implemented in a non-uniform

way that narrows the tax base, which might indicate that the base of the tax should be

broadened. It is important to be able to distinguish these two cases of high excess burden,

because the appropriate policy responses are different.

Non-uniformity in a tax may arise in a number of different ways, which can be illustrated

through three examples. For payroll tax, there is a tax-free threshold or small business

exemption. For alcohol taxation, closely-substitutable products are taxed at different rates.

Finally, for the GST, some products are taxable, some are exempt, and some are GST-free.

In MM900, simulations were conducted to identify the contribution of non-uniformity in taxes

to their excess burdens. For each tax, this involved beginning with the baseline scenario in

which the tax is in place, and conducting two simulations. In the first simulation, the tax is

abolished. In the second simulation, the tax is ‘reformed’, by replacing it with a uniform tax

that is designed to raise the same amount of revenue. The difference in consumer welfare

between the tax abolition scenario and the baseline scenario shows the excess burden of the tax.

The difference in consumer welfare between the reform scenario and the baseline scenario

shows the excess burden due to the non-uniformity of the tax. The difference between these

excess burdens shows the excess burden due to the inherent nature of the tax. Thus, this

procedure allows us to dissect the excess burden of a tax into the component due to any non-

uniformity in its application and the component due to its inherent nature.

This procedure was applied to three different non-uniform taxes - payroll tax, GST and alcohol

taxes. The results allow us to understand the different efficiency gains that could be made by

either reforming or abolishing each tax.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Modelling State Taxes

In this study, state taxes are modelled at the national level, using a weighted average of the

regimes across states. This methodology provides a reasonable overall estimate for the

efficiency of the state tax in question. However, there is potentially some downward bias in the

excess burden estimates for some state taxes. The differences between states in their tax

regimes may distort a firm’s choice of location, adding to the excess burden, and the modelling

results will not take this into account.

3.3 The Incidence of Taxes

While excess burdens give information about the size of the overall cost of taxation, the

question of who actually bears the burden of taxes is also important for tax analysis. The party

who is liable to government for a tax carries its impact. However, some taxes lead to economic

responses, such as price increases, that mean that the party or parties who bear the economic

incidence or final burden of a tax may be different from the party carrying the initial impact.

That is, economic incidence is concerned with who ultimately bears the cost of the tax.

The cost of a tax is borne by a group when the tax changes the income that they receive or the

prices that they face. When a tax is imposed in a market it either: raises the price paid by users;

lowers the price received by sellers; or both. These price changes will flow through the

economy, and alter the return to the different factors of production. Incidence can therefore be

measured by considering the changes in prices of goods and services and changes in income

from each factor of production.

The economic incidence of each tax depends on the characteristics of each market. The panels

in Figure 3.2 show three different markets, and how the demand and supply elasticities affect

the incidence of taxes using partial equilibrium analysis. In general, the side of the market

which is relatively price insensitive (inelastic) will bear more of the tax.

In each case, without any tax, the market-clearing price would be P*. However, a tax drives a

wedge between the price that consumers pay, Pc, and the price that suppliers receive, PP. The

deviation of Pc and PP from P* indicates the effect on prices of introducing each tax.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Panel (a) shows a taxed market with elastic

demand and supply curves. When a tax is

applied in this market, both the price that

consumers pay and the price that producers

receive will change. Therefore, the

incidence of a tax in this market will fall on

both consumers and producers.

Panel (b) shows a taxed market with a

perfectly elastic supply curve. This is

representative of a market with perfect

competition, and which does not use factors

of production which are fixed in supply6.

Under perfect competition, suppliers of the

good are already producing at minimum

average cost when there is no tax.

Therefore, they will not reduce their prices

when a sales tax is introduced because this

will reduce their profits below the required

rate of return. Instead, the full incidence of

the tax is passed on to consumers in the

form of higher prices.

Panel (c) of Figure 2 shows a taxed market

with a perfectly inelastic supply curve. In

MM900, this is representative of the market

for a fixed factor of production, such as

residential land. With a fixed supply of

land, its value to users (or marginal

product) will be determined only by

demand (at P*). No user will be willing to

pay an amount higher than P*, even after a

tax is applied. Therefore, the owners of

residential land must bear the full incidence

of any tax in this market, in the form of a

lower return for ownership of their land.

6 Under perfect competition, all firms will produce at minimum average cost and receive zero profits. They exit and

enter an industry until demand is satisfied and price is equal to marginal cost. This gives the flat supply curve for

any commodity, at a price equal to marginal cost. A fixed factor of production would imply increasing marginal

costs to the industry, which would result in an upward sloping supply curve, even under perfect competition.

Quantity

Demand

Supply

Price

(a) relatively elastic demand and supply

Demand

Supply

Price

PP = P*

Demand

Supply

Price

Pc = P*

(b) perfectly elastic supply

Figure 3.2: Economic incidence

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The simple analysis above only reveals part of the story. In the analysis above, the incidence of

the tax is determined by considering only the market to which the tax is applied. While this

partial equilibrium analysis is instructive, a more in-depth investigation of the incidence of taxes

would use a general equilibrium approach and consider all of the flow-on impacts to other parts

of the economy. For example, a tax on labour income will reduce the return to labour, having

an impact on labour supply. This would reduce incomes in the economy, and impact on the

demand for goods and services. Thus, in turn, the demand for other factors of production may

also be reduced, reducing the income derived from them. At the same time, the increased labour

costs would lead to a switch away from labour intensive production technologies towards

production technologies more intensive in the other factors of production.

As a general equilibrium model, MM900 traces the impact of a tax in one market to all parties

across the whole economy. In this way, both the direct and indirect impacts of a tax can be fully

taken into account. The following section describes the methodology in more detail.

3.4 Using MM900 to Estimate the Incidence of Australian Taxes

To analyse the incidence of a tax, agents in the economy must first be divided into groups.

Since this study is concerned with the welfare of Australian households, the incidence analysis

divides domestic households into groups. This allows some understanding of the distributional

effects of taxes on different types of domestic households. Groups typically used in incidence

analysis include (Metcalf and Fullerton, 2002):

• consumers (as opposed to producers);

• owners of each of the factors of production;

• groups of individuals with different levels of income;

• groups of individuals living in different regions; and

• intergenerational groups.

The choice of groups will depend on the type of model used. This study divides households into

two sets of groups, as discussed below.

• Firstly, the incidence of each tax on owners of the different factors of production is

analysed.

• Second, the incidence of each tax on individuals with different levels of income is

considered.

A key observation is that Australian households are simultaneously consumers and producers.

Households derive welfare from their consumption of goods and services, which in turn,

depends on their income as owners of factors of production including labour, capital, land and

other fixed factors. Therefore, any change to either consumer prices or factor incomes will

affect overall welfare. It follows that distinguishing between the incidence on consumers and

producers would not be a meaningful analysis.

Therefore, this study divides households according to their ownership of the factors of

production – labour, capital, land and non-land fixed factors. Different taxes will have different

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

impacts on the income derived from each of these factors of production. Changes in factor

incomes after a tax is introduced will indicate where among the groups of households the

incidence falls. MM900 results can be used to identify these changes. The variables of interest

are listed in Table 3.1 below.

Table 3.1 Identifying Tax Incidence

Agent Variable indicating tax incidence

Labour Change in after-tax labour income

Owners of capital Change in after-tax private capital income

Owners of residential land Change in after-tax private residential

land income

Owners of rural land Change in after-tax private rural land

income

Owners of urban land Change in after-tax private urban land

income

Owners of non-land fixed factors of

production e.g. natural resources

Change in after-tax private income from

other economic rents

Examining the dollar changes in these types of income when a tax is introduced will show

which groups the incidence of the tax falls on. This may give some indication as to the impact

on different socio-economic groups.

The above analysis shows changes in nominal, private, after-tax incomes, before allowing for

any changes in consumer prices. In reality, some taxes result in changes in consumer prices,

which must be taken into account in determining the effect on real incomes. This requires some

estimate of the percentage change in the cost-of-living.

However, spending patterns vary between households, which means that percentage impacts on

the cost-of-living can vary from one household to the next. To give some indication of this

variation, this report uses two different measures of the cost of living. These are the price of

essential consumption and the price of non-essential consumption. MM900 identifies a basket

of goods and services that is considered essential consumption. Consumers will always

consume this same basket of goods and services, no matter the price of each good. After they

have purchased this basket, they then choose how to spend the remainder of their income,

depending on both the relative prices of each product and their overall income. This extra

consumption is termed non-essential consumption.7 The essential and non-essential baskets

have different mixes of the same goods and services. For example, food is more important in

the essential basket while recreation is more important in the non-essential basket.

7 Consumption decisions are explained in more detail in section 4.3.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The spending of poorer households will be dominated by essential consumption, while spending

of richer households will be dominated by non-essential consumption. Hence, other things

being equal, any tax which, in percentage terms, increases the price of essential consumption

more than it increases the price of non-essential consumption will have a greater proportional

impact on poorer households. Such a tax can be considered a regressive tax. Conversely, any

tax which, in percentage terms, increases the price of essential consumption less than it

increases the price of non-essential consumption will have a greater proportional impact on

richer households. Such a tax can be considered a progressive tax. Therefore, percentage price

changes for both essential and non-essential consumption are reported as part of the incidence

analysis. Here progressivity is implicitly defined relative to consumption levels, but it can also

be defined relative to income levels.

A full analysis of the equity of a tax would also involve an analysis of the changes in nominal

factor incomes summarised in Table 3.1 against information on the sources of factor incomes

for different types of households. For example, if poorer households rely relatively more on

labour income and richer households more on capital income, the relative impact of a tax on

labour and capital income would also influence its equity.

Importantly, the analysis from MM900 will be a long-run analysis. Long-run results may look

different to results from a short-run analysis, mainly because some prices take time to adjust in

response to tax changes. For example, an increase in the company tax rate may initially lower a

firm’s profits if they do not have the ability to change their prices straight away. However, in

the long run, as lower profits lead to a withdrawal of capital and tighter supply conditions, firms

will be able to adjust their prices upward and pass the tax burden on to consumers.

In any general equilibrium model, all prices are determined in relative terms, but expressed in

nominal terms. To do this conversion to nominal terms, one good is chosen to be the

“numeraire”. This good has a set fixed price, and all other prices can be expressed relative to

that fixed price. In MM900, the average nominal wage has been chosen as the numeraire.

Although the choice of numeraire does not affect the modelling results in real terms, it does

need to be taken into account in interpreting nominal price changes.

Tax Incidence in MM900

In the long run, the economic incidence of most Australian taxes can be expected to fall largely

on Australian labour, and this is also the case in MM900. To understand this conclusion, there

are three issues to consider in turn.

The first issue is the extent to which the incidence of Australian taxes might fall on foreigners

rather than Australians. As a relatively small part of the world economy, Australia is generally

considered to be close to being a ‘price taker’ on world markets. This implies that Australian

taxes applied in world markets can be expected to have little impact on prevailing world prices.

Rather, such taxes will largely be absorbed on the Australian side of the market, will little or

none of the incidence being borne by foreigners through changed world prices. In MM900,

Australia is a price taker for capital and imports, and is close to being a price taker for most

exports. This means that the incidence of import duty on imports, company tax on capital, and

other Australian taxes will fall largely on Australians, not foreigners.

Given that most of the incidence of our taxes can be expected to fall on Australians, the next

issue is the extent to which that incidence falls on the incomes of Australian factors of

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

production. MM900 makes the common assumption that production occurs under constant

returns to scale and perfect competition. This implies that the entire value of production is

distributed to the factors of production – labour, capital, land and other fixed factors – without

any leakage. It follows that the incidence of production-related taxes is borne fully by factors of

production.

Given that the incidence of our taxes falls largely on Australian factors of production, the final

issue is identifying how that incidence is shared between those factors of production – capital,

labour, land and other fixed factors. This depends on their relatively mobility. Immobile

factors of production cannot move to avoid a tax, while mobile factors can.

As already noted, capital is highly mobile internationally, as it seeks out higher rates of return.

In that environment, it is common to assume that Australia is a price taker in world capital

markets, and that approach is followed in MM900. This means that foreign investors don’t

ultimately bear a burden from Australian taxes; rather taxes that impact on capital limit

investment and are ultimately passed on into higher prices, reducing real incomes for other

factors of production, especially labour. Capital bears none of the burden of the tax system in

MM900.

In contrast, land and other fixed factors of production are completely immobile in MM900.

This means they bear the full burden of any taxes that are applied to them. It also means they

cannot avoid the burden of taxes that are passed on to them. However, fixed factors are less

important as factors of production than either labour or capital, so they bear a relatively small

share of the tax burden.

Labour is generally considered to have a low degree of mobility, and this is also the case in

MM900. This means that the incidence of taxes on labour falls largely on labour, although a

small part of it is shifted to fixed factors. It also means that labour, as the major factor of

production, also bears the largest share of the incidence of other taxes, with the exception of

taxes that are applied to fixed factors. Hence, labour bears most of the incidence of the tax

system in MM900, through reduced real wages. This reduction comes about through reduced

after tax nominal wages, in the case of personal income tax, and through increased consumer

prices, in the case of a wide range of other taxes. 8

In summary, the incidence of most Australian taxes can be expected to fall largely on labour.

This reflects the common modelling assumptions of perfect competition with constant returns to

scale, and that Australia is (almost) a price taker in world trade and capital markets. However,

land and natural resources bear some of the incidence of the tax system, and when taxes are

applied directly to them they bear the full incidence.

8 The incidence of most taxes borne by labour is transmitted through higher prices rather than a lower nominal wage

because the nominal wage is the numeraire in MM900. However, the choice of numeraire does not affect the real

outcome of the model, which is that the real wage falls in response to a wide range of taxes and so labour bears the

final incidence.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

4 The MM900 Model

While section three explained the nature and significance of the concepts of tax efficiency and

incidence, this section outlines the economic model that has been used for the tax simulations.

It completes the groundwork needed for section 5, where the modelling results are presented for

the efficiency and incidence of each tax.

KPMG Econtech’s MM900 model is the latest edition of our detailed computable general

equilibrium (CGE) model of the Australian economy focussing on tax analysis. The first

edition, MM303, was developed for the South Australian Department of Treasury and Finance

in the late 1990s to assist it in participating in a developing debate on indirect tax reform. That

debate culminated in the introduction of the New Tax System (NTS) in July 2000. In the lead

up to the introduction of the NTS, MM303 was further developed to MM600+ to assist the

Australian Competition and Consumer Commission (ACCC) in its price surveillance work. For

this study for the Federal Treasury, MM600+ has been re-developed as MM900, extending its

tax analysis capabilities from indirect taxes to also include direct taxes.

This development process has meant that MM900 goes well beyond other Australian modelling

in capturing the economic effects of the tax system on the Australian economy. It does this by

distinguishing 19 different major taxes at the Federal, State and Local levels. Further, for each

tax, the model identifies the true tax base as closely as possible, and aims to capture the main

behavioural responses to the tax’s imposition. The modelling also allows for certain negative

externalities in consumption that may justify certain specific taxes. A key feature of MM900 is

its fine level of detail.

MM900 contains even a finer level of product detail than MM600+. For example, in MM900

the economy produces 889 different products (i.e. about 900 products), which represents eight

times as much product detail as other comparable models. This allows the model to more

accurately capture the application of certain product-based taxes.

As noted in Han (1998), a greater level of detail will also lead to a smaller amount of

aggregation bias in CGE estimates of excess burdens. The greater the product detail, the more

distortions the model will pick up, and the better the estimates of the excess burdens. For

example, MM900 treats beer, wine and spirits as separate substitutable products within one

broad group. Less disaggregated models aggregate all alcohol products together, and therefore

miss the excess burdens that arise from taxing closely substitutable alcoholic beverages at

different rates.

MM900 includes a detailed treatment of consumer’s responses to taxes that cause relative price

changes. MM900 does this by specifying a two-tier consumer demand system covering the 889

products in the model.

• First, the consumer decides between 17 different broad groups of products using a linear

expenditure system.

• Second, and importantly, MM900 also allows for substitution within these 17 broad groups,

between individual products, with the degree of substitutability able to vary from one group

to the next, adding extra sophistication.

This detail in consumer decisions means that MM900 will produce high quality estimates of

excess burdens.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

MM900 also contains a fine level of detail in modelling production processes. Each of the 109

industries use up to six different primary factors (or types of labour, capital and fixed factors),

compared with two primary factors in MM600. The different fixed factors include land and

natural resources, allowing for much more robust modelling of the effects of taxes based on the

value of land or the value of natural resource use.

In addition, in comparison to earlier studies, MM900 incorporates a more comprehensive

analysis of the behavioural responses to each tax. In MM900, taxes can cause households to

change their supply of labour and their levels and patterns of spending. Taxes can cause

businesses to change their choices between the six primary factors of production, affecting

employment, investment and valuations of land and natural resources. Finally, taxes can affect

the propensity to import and export each of the model’s 889 products.

These behavioural responses to taxes, by impairing the functioning of the economy, can reduce

consumer welfare, that is, give rise to excess burdens on households. Capturing these welfare

losses accurately requires that the modelling of household behaviour is underpinned by a

consistent treatment of consumer welfare. In MM900, this occurs by using a utility function in

which a representative household derives welfare or utility from leisure, saving and

consumption of products, and then deriving all household behaviour from that same utility

function. Other comparable models include more ad hoc elements in modelling household

behaviour.

In summary, by better capturing where taxes impact on the Australian economy, how it

responds, and how these responses flow through to change consumer welfare, MM900 is

uniquely well suited to the analysis of the excess burdens of Australian taxes.

MM900 is also carefully designed to capture the economic incidence of each tax. This involves

recognising that, as a relatively small part of the world economy, Australia in close to being a

‘price taker’ in world trade markets. It also means taking into account that natural resources and

land are completely immobile, capital is highly mobile, and labour has a low degree of mobility.

Other comparable models have more limited or no treatments of natural resources and land.

The following sections systematically summarise the main features of MM900, emphasising

those that are most pertinent to this tax study. It begins by noting the main, over-arching

assumptions, and then moves on to summarise the behaviour of the ‘economic agents’ in the

model – households, producers, government and the foreign sector.

4.1 Over-arching Assumptions

In keeping with most long-run CGE models, in MM900 economic agents engage in optimising

behaviour, markets are in equilibrium and the government and private sectors live within their

means.

Long-run Horizon

MM900 refers to a long-run equilibrium, after the economy has fully responded to shocks. For

example, stocks of capital in each industry have fully adjusted. This long-run focus is important

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

for tax policy, because good tax policy is based on the lasting effects of tax policy changes, not

the transitional effects.

Optimising behaviour

Economic agents engage in optimising behaviour. In MM900, this means that a representative

business in each of the 109 industries chooses inputs and outputs to maximise profit under

perfect competition subject to a production technology with constant returns to scale. It also

means that a representative household maximises utility, which depends on leisure, saving and

consumption of products, subject to a budget constraint. This focus on consumer utility is

important for drawing conclusions about how individual taxes affect consumer welfare.

Equilibrium

In keeping with MM900’s long-run horizon, all markets are assumed to have achieved

equilibrium. This includes markets for the six factors of production – low-skilled labour, high-

skilled labour, structures, other capital, land, and other fixed factors – and markets for the 889

products (goods and services) that are produced.

Government Budget constraint

Governments must always pay their way in the long run. For simplicity, in MM900 the

government is assumed to always balance its budget. To achieve this, a budget policy

instrument must be selected that, instead of being an input to the model, automatically adjusts to

balance the budget. For this study, a hypothetical lump sum tax/transfer is chosen as the swing

instrument, because the efficiency of any tax is traditionally assessed against a lump sum tax,

which by definition is perfectly efficient. Hence, when a change in a tax rate is simulated in this

study, the potential impact on the budget balance is automatically neutralized through a change

in lump sum tax. Any change in consumer welfare can then be attributed to economic

distortions associated with the tax that has been changed. This approach to tax efficiency

analysis is standard in the literature.

Private Budget constraint

Private saving behaviour must also be sustainable in the long run. As explained further below,

the private propensity to save is constant in MM900. Based on that saving rate, together with

the return to savings and the growth rate of the economy, the model then deduces the level of

private assets. Remaining assets are owned by the foreign sector and are supplied perfectly

elastically at the world required rate of return on capital. In the long run, the stock of foreign

liabilities (just like the stock of private assets) must also grow at the same rate as GDP,

requiring a particular current account deficit. In MM900, the exchange rate adjusts to deliver

that current account deficit (external balance).

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

4.2 Households

In MM900, a representative household maximises utility, which depends on leisure, saving and

current consumption of products, subject to a budget constraint. This is an important

development from MM600, in which utility only depended on consumption of products.

While MM900 is a ‘static’ model, saving generates utility on the basis that it represents future

consumption of products. This approach leads to a Generalised Linear Expenditure System,

which includes relationships for labour supply, total consumption expenditure, and its spread

across 889 products. These three sets of relationships are now discussed in turn, beginning with

the labour supply choice.

4.2.1 Labour Supply versus Leisure

In MM900 households face a choice in dividing the time in which they could be working,

between work and discretionary leisure, whereas in MM600 the labour supply was fixed. Under

the MM900 utility maximising approach, the amount of time they devote to work depends on

the after-tax real wage that is available from working, and their real ‘full income’, which is the

potential income they could receive if they take no discretionary leisure. MM900 makes full

allowance for the taxes that influence this work-leisure choice.

For example, higher labour income tax directly reduces the real after-tax wage, reducing the

economic return to work. This leads to lower labour supply, but higher leisure. Lower labour

supply leads to lower incomes, reducing levels of consumption and saving.

Many other taxes reduce the economic return to work, as measured by the real after-tax wage,

by increasing consumer prices. Put another way, higher consumer prices reduce the consumer

purchasing power generated by a given work effort. GST and ‘sin taxes’ on alcohol, tobacco

and gambling have obvious effects on consumer prices. Other taxes, such as company tax and

payroll tax, have different tax bases, but ultimately are also likely to be largely passed on into

higher consumer prices or lower wages. All of these taxes act as a disincentive to work by

reducing real after-tax wages. They also have other disincentive effects that are captured in

MM900.

Importantly, explicitly including leisure in the analysis helps to make the excess burden

estimates more robust. For example, increasing labour income tax (or many other taxes) will

reduce the after-tax real wage and reduce labour supply. While reduced labour supply will

reduce utility as lower labour incomes lead to lower consumption and saving, this will be partly

offset by the utility derived from increased leisure. Without the inclusion of leisure in the utility

function, the excess burden of labour income tax would be overstated. This will also be true for

other taxes that reduce real after-tax wages, such as GST, ‘sin taxes’, payroll tax, company tax

etc.

Given the amount of labour that households choose to supply, they will receive a certain

income. The next choice is how to divide this income between consumption and saving.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

4.2.2 Consumption versus Saving

Modelling saving behaviour poses an issue for long run models such as MM900. In particular,

saving (i.e. sacrificing present consumption for future consumption) can appear artificially

attractive. This is because, if saving rates are increased, long-run model results will show the

gain in future consumption, but not the sacrifice of present consumption. To avoid this

problem, the private propensity to save is constant in MM900, as a consequence of the

deliberate design of the household’s utility function. In particular, saving generates utility on

the basis that it represents future consumption of the same products that are consumed in the

present. Allowing for saving and the utility it brings is an important development in moving

from MM600+ to MM900.

Incorporating saving in the analysis of household behaviour, just like incorporating leisure,

helps to make the estimates of excess burden more robust. In particular, foregoing leisure for

the income from additional work allows additional consumption and saving, and it is important

that both of these lead to higher utility if leisure is not to appear artificially attractive.

At the same time, assuming a constant propensity to save means that MM900 is not useful for

estimating excess burdens for taxes that mainly distort the propensity to save. These taxes

include personal income tax on income from assets and taxes on superannuation earnings and

benefits.

Having determined the split of income between consumption and saving, the next choice is how

to divide consumption between the various products.

4.2.3 Pattern of Consumption

Similar to MM600, MM900 allocates total consumption expenditure between products using a

two-tier consumer demand system, corresponding to a Generalised Linear Expenditure System.

In MM900 there are 889 products and the two tiers are as follows.

• In the first tier, the consumer decides between 17 different broad groups of commodities in a

Linear Expenditure System.

• In the second tier, MM900 allows for substitution between individual products within these

17 broad groups, with the degree of substitutability able to vary from one group to the next,

adding extra sophistication.

This detail in consumer decisions means that MM900 will produce high quality estimates of

excess burdens. For example, MM900 treats beer, wine, spirits as separate products, and they

are all substitutable in consumption. Less disaggregated models treat all alcohol products as a

single product group, and will therefore miss the excess burden that arises from taxing

substitutable alcoholic beverages at different rates.

In short, MM900 includes a detailed treatment of consumers’ responses to taxes which cause

relative price changes, and so generates more robust estimates of excess burdens.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

4.3 Producers

In MM900, production occurs in 109 industries that produce 889 products. Within each

industry, a representative business operating under perfect competition chooses inputs and

outputs to maximise profit subject to a production technology. Apart from the unusually large

number of products in MM900, this approach is typical of CGE models.

For its production, each industry uses products produced by other industries as well as primary

factors of production. In MM600, the two primary factors of production of labour and capital

were recognised and treated as substitutable – this is typical of CGE models. However, in

MM900 the number of primary factors is extended from two to six. As discussed further below,

this extension is important for robustly modelling the excess burden and economic incidence of

certain taxes, such as resource rent taxes, land tax, municipal rates, and motor vehicle

registration.

In each industry, the demand for the six primary factors is modelled in a two-tier approach

based on a nested CES production function. In the first tier, the representative producer chooses

a mix of the three broad primary factors – labour, capital and fixed factors – taking into account

their relative prices. In the second tier, for each broad primary factor, they choose a mix

between two types. For example, for labour, they choose a mix between low skilled labour and

high skilled labour, taking into account their relative prices. These mixes vary from one

industry to the next.

The six primary factors, which are now discussed in turn, are as follows:

• low-skilled labour;

• high-skilled labour;

• capital – structures;

• capital – other;

• land; and

• other fixed factors such as natural resources.

Demand for these six primary factors is driven by producer decisions. In the following

discussion of each primary factor, each explanation of how producer behaviour determines

demand is followed by an explanation of supply, and how an equilibrium is achieved in which

supply and demand are balanced.

4.3.1 Low and High Skilled Labour

Labour is divided into two types, low skilled and high skilled. This is equivalent to dividing

labour into a ‘heads’ aspect, which keeps track of the number of people who are employed and a

‘quality’ aspect, which determines the average productivity of labour in each industry. The

reason for dividing labour into skilled and unskilled labour is so that the model can account for

both labour input in terms of the number of people, or total employment, as well as effective

labour input, or the total productivity of labour.

Following profit maximizing behaviour, the representative producer in each industry generates a

demand for low skilled labour and high skilled labour. This demand depends on the prices of

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

low skilled and high skilled labour, the prices of other inputs, and industry production. These

demands are aggregated over industries to determine the total demand for low skilled labour and

the total demand for high skilled labour. This raises the question of how this demand for low

skilled and high skilled labour is balanced with supply.

The supply of labour is determined by household behaviour in the way described in section

4.2.1. The mix of that supply between low skilled and high skilled labour is held fixed as an

input to the model. The wage for low skilled labour relative to the wage for high skilled labour

then adjusts until the mix of demand matches the fixed mix of supply.

Total labour demand equals total labour supply in MM900 by virtue of Walras’ Law. That Law

states that, when considering any particular market, if all other markets in an economy are in

equilibrium, then that specific market must also be in equilibrium.

The average wage is not determined by the model but rather is fixed as the “numeraire”. The

idea here is that in any general equilibrium model, all prices are determined in relative terms,

but expressed in nominal terms. To do this conversion to nominal terms, one good is chosen to

be the “numeraire”. This good has a set fixed price, and all other prices can be expressed

relative to that fixed price. In MM900, the average nominal wage has been chosen as the

numeraire. Although the choice of numeraire does not affect the modelling results in real terms,

it does need to be taken into account in interpreting nominal price changes.

4.3.2 Structures and Other Capital

Physical capital is split into two different types. These are:

• structures - residential and other; and

• other capital, which is all other capital goods, such as motor vehicles, machinery and

computers.

Structures include residential structures, which are only used by the ownership of dwellings

industry (in producing housing services) and other structures such as commercial buildings and

engineering construction, which are used by all of the other 108 industries in the model.

Other capital is also used by all of the other 108 industries in the model. Other capital covers

all capital other than structures, and includes motor vehicles, computers, machinery etc, and has

a fixed composition.

Following profit maximizing behaviour, the representative producer in each industry generates a

demand for structures and other capital. This demand depends on the prices of the services of

structures and other capital, the prices of other inputs, and industry production. These demands

are aggregated over industries to determine the total demand for structures and the total demand

for other capital. This raises the question of how this demand for structures and other capital is

balanced with supply.

In MM900, an industry can access as much capital as it needs so long as it can achieve the after

tax rate of return required by international investors. This perfect elasticity of capital supply

means that the required after tax rate of return on capital is effectively fixed on world capital

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

markets in the long run. Stocks of structures and other capital adjust in each industry until this

rate of return is achieved.

As noted, it is the after tax rate of return on capital that is determined on world capital markets

and governs the terms on which capital is supplied to Australia. This means that any tax that is

put on capital in Australia will ultimately increase the required before tax rate of return to offset

the tax impost. Capital will flow out of the country until its greater scarcity achieves the

increase in its before tax rate of return. The resulting lower capital intensity of production will

reduce the marginal productivity, and therefore the incomes, of other factors of production,

including labour, land and other fixed factors of production.

4.3.3 Land

MM900 includes land as a fixed factor, which is an unusual feature in CGE models of the

Australian economy. This feature was developed to enhance the estimates of the excess burden

of land tax and municipal rates.

Specifically, MM900 has the following three types of land:

1. rural land, which is land used only by sectors within the agriculture, forestry, fishery and

mining industries;

2. residential land, which is land used only by the ownership of dwellings industry; and

3. urban land, which is used by all other industries.

Following profit maximizing behaviour, the representative producer in each industry generates a

demand for land services. This demand depends on the rental price of land, the prices of other

inputs, and industry production. These demands are then aggregated to determine the total

demand for rural land, for residential land, and for urban land. This raises the question of how

this demand for each of the three types of land is balanced with supply.

The total supply of each of the three types of land is fixed, which reflects the overall availability

of land in the economy and somewhat mimics zoning legislations. This implies that the price of

each of the three types of land will adjust until total demand is reconciled with the fixed supply.

Any tax placed on a type of land will not change the supply of that land. With the fixed supply

of land to the market, the land rents that users will be prepared to pay will be unaffected by the

tax. This means that the burden of the tax will be fully borne by the land owners.

4.3.4 Natural Resources and Other Generators of Economic Rents

The zero profit assumption is a common assumption in economic modelling. However, in

reality, excess returns do exist in some industries. These excess returns can arise as a result of

factors such as access to a natural resource (such as petroleum or minerals deposits), monopoly

power, branding, patents or barriers to entry.

In MM900, these excess returns are modelled as a return on a fixed factor, that is as a “rent”.

To identify the excess returns in an industry, the rate of return on capital in each industry was

examined. When this return has consistently been greater than the “normal” rate of return on

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

capital, then the industry may be making excess returns9. Further, the idea that any particular

industry would systematically make excess returns has been subject to a “reality check”. This

was done by examining whether factors that generally lead to excess returns, such as access to a

natural resource or barriers to entry, appear to be present for that industry.

The following industries are identified as having “excess returns” that, for modelling purposes,

are attributed to a fixed factor:

• Coal;

• Oil and gas;

• Iron ore;

• Non-ferrous metal ores;

• Beer;

• Banking; and

• Non-bank finance.

Each of these seven industries is assumed to use a different fixed factor. The remaining 102

industries use no fixed factor.

Following profit maximizing behaviour, the representative producer in each of the seven

industries with a fixed factor generates a demand for fixed factor services. This demand

depends on the rental price of the fixed factor, the prices of other inputs, and industry

production. This raises the question of how demand for each of the seven fixed factors is

balanced with supply.

The total supply of each fixed factor is fixed, which reflects the overall availability of natural

resources and other fixed factors in the economy. This implies that the price of each of the

seven types of fixed factors will adjust until total demand is reconciled with the fixed supply.

Any tax placed on a type of fixed factor will not change the supply of that fixed factor. With the

fixed supply of the fixed factor to the market, the rent that users will be prepared to pay for use

of the fixed factor will be unaffected by the tax, and will continue to reflect the fixed factor’s

marginal productivity. This means that the burden of the tax will be fully borne by the fixed

factor owners.

This inclusion of fixed factors is particularly useful for measuring the excess burden of the

Petroleum Resource Rent Tax, which is intended to be a tax on the above normal profits of the

oil extraction industry.

9 Industries making “excess returns” have been identified in the following way. First, a “normal” rate of return of

return for each of the 109 industries in MM900 has been estimated using employment data and an estimate of output

per employee in each industry. Next, the actual GOS according to the national accounts was compared to the

estimate of the “normal” rate of return. Where the actual GOS was significantly larger than the estimate, then it is

considered that the industry may be making “excess returns”.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

4.4 Government Sector

Government sector spending accounts for part of final demand for various products, and is

financed by taxes. Table 4.1 shows the 19 key taxes in MM900 that are analysed in this study.

These include a range of Federal, State and Local taxes.

The table divides the key taxes into three groups, according to how they appear in MM900.

These are general taxes (that are applied at a single rate), industry taxes (that can be varied from

one industry to the next), and product taxes (that can be varied from one product to the next and

even between different categories of end users of a product). Some specific comments on

particular taxes can be made.

Table 4.1 Key Taxes

General By industry By product

Labour income tax Payroll tax GST

Corporate income tax Land taxes Alcohol excise and WET

Municipal rates Tobacco excise

Resource rent tax Luxury car tax

Motor vehicle registration Import duties

Fuel taxes

Royalties and crude oil excise

Gambling taxes

Conveyancing stamp duties

Motor vehicle stamp duties

Stamp duties (other)

Insurance taxes

“Labour income tax” refers to personal income tax as it applies to labour income. Other

personal income tax, which is collected from non-labour income such as interest income, is

much smaller in magnitude, and is not explicitly modelled in MM900. As explained in section

4.2.2, this is because the main excess burden from personal income tax on non-labour income

(such as income from assets) is likely to be through its influence on private saving, but the

private saving rate is fixed in MM900. Hence, MM900 is not designed to capture the excess

burden from personal income tax as it applies to non-labour income.

Company income tax is modelled in MM900. In addition, the reduction in personal income tax

collections from the claiming of franking credits is also taken into account.

In modelling payroll tax in each industry, the small business exemption is taken into account.

The model allows for resource rent tax to be applied at any rate in most sectors of the mining

sector. In constructing the model, resource rents were identified for: the coal industry; the oil

and gas industry; the iron ore industry; and the non-ferrous metal ores industry. In the baseline

scenario of the existing tax system, the existing petroleum resource rent tax is applied in the oil

and gas industry, while in the other three industries resource rents are not taxed. The model also

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

identifies economic rents in some other industries – the banking industry, the non-bank finance

industry, and the beer industry.

The model assumes that GST is applied at a single, specified rate, which is 10 per cent in the

baseline scenario of the existing tax system. However, GST status (taxable, GST-free or

exempt) is set product-by-product, separately for both residents and foreign tourists visiting

Australia.

In modelling fuel taxes, separate account is also taken of fuel subsidies that partly offset fuel

taxes. In simulating the model for this study, when fuel taxes are altered, associated fuel

subsidies are altered in the same proportion.

4.5 Foreign Sector

Australia’s interactions with the global economy are important for the domestic economy. As a

small country, Australia is generally considered to be close to being a ‘price taker’ on world

markets. In MM900, Australia is a price taker for capital and imports, and is close to being a

price taker for most exports.

In a world of highly mobile capital, Australia is assumed to be a price taker in world capital

markets. This means that the world supplies capital to Australia at a fixed real after-tax rate of

return. On the demand side, as explained in section 4.3.2, industries generate demand for

structures and other capital following profit-maximizing behaviour. They do not differentiate

between local and foreign owned capital. The supply of local-owned capital is determined by

saving behaviour, while remaining capital demands are met by foreign-owned capital.

Similarly, the rest of the world supplies Australia, as a small open economy, with as much

imports as demanded at the world price i.e. supply is perfectly elastic. On the demand side,

consumers and producers perceive imported and locally produced goods to be different from

one another, and choose their mix of imported and locally produced goods and services

depending on their relative prices.

For exports, Australia’s status as a small open economy is again recognised, but this time by

assuming export demand is highly elastic but not perfectly elastic. That is, Australia is close to

being a price taker, but has a small degree of pricing power. This pricing power may arise

through product differentiation or by supplying a large share of the world market. For most

goods, export demand elasticities in MM900 are set to a very responsive -12. For goods where

Australia is considered to have some market power, export demand elasticities are lower. The

smallest elasticity is for wool, where the value is -4, in recognition of our large share of the

world market. The same elasticity is used for tourism, which takes into account the product

differentiation between the tourism services that Australia offers compared with those offered

by other countries. All trade elasticities used in MM900 are reported in Appendix C.

On the supply side of export markets, in CGE models with perfect competition and constant

returns to scale, commodity supply is also elastic. The combination of highly elastic export

demand and supply commonly leads to problems because small shifts on either side of the

market can lead to large changes in export volumes.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Some models attempt to overcome this by making export demand less responsive to price

changes. However, this is not reflective of Australia’s position as a small open country. For a

more realistic approach, MM900 incorporates two features.

First, MM900 includes fixed factors of land and natural resources. This reduces the flexibility

in export supply of agricultural and mining products in a realistic way.

Second, MM900 introduces some friction between supplying domestic and export markets.

Intuitively, this friction may arise because some exported commodities are tailor made for

export, or are more narrowly defined than the corresponding home commodity. For example,

Australian consumers may eat all types of apples, but Australia may only export Fuji apples to

Japan. Thus, there will be a cost involved in switching from supplying the domestic market to

supplying the export market. This reduces flexibility in supplying the export market.

Australian taxes applied in world markets can be expected to have little impact on prevailing

world prices. Rather, such taxes will largely be absorbed on the Australian side of the market.

Taxes on foreign capital will increase the before tax rate of return on capital in Australia, and

accordingly reduce the demand for foreign capital, as discussed in section 4.3.2. Taxes on

imports will have no influence on the world price, and instead raise the price that Australian

consumers pay for imports and therefore reduce their demand for them. Likewise taxes on

exports will have little effect on the price that foreign consumers will pay because export

demand is highly elastic. Instead, they will reduce the price that Australian producers receive

and therefore reduce export supply.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

5 Results

This section presents the results of this study. For each tax, the results are reported in terms of

the economic cost of the tax and the economic incidence or who ultimately pays the tax. These

concepts were explained in section 3, while the model used to generate the results was

summarised in section 4.

For ease-of-reference, Box 5.1 on the following pages summarises the information from

section 3 on the concepts and measurement of the excess burden and economic incidence of a

tax. It also explains how to interpret the detailed, tax-by-tax results of economic incidence that

are reported throughout this section.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Box 5.1 Measures of the economic cost and incidence of taxes

The excess burden of a tax measures the cost of each tax over and above the revenue raised.

These costs arise because taxes may distort incentives to work and invest, as well as production

methods and consumption patterns. The marginal excess burden (MEB) is the cost arising

from a small increase in the tax rate. The average excess burden (AEB) is the full cost arising

at the existing tax rate, starting from a situation without the tax. The MEB and AEB are

measured as cents of economic cost per dollar of revenue raised. See sections 3.1 and 3.2 for

more a detailed explanation of excess burdens and how they are measured.

The incidence of a tax refers to who ultimately bears the burden of each tax. The party who is

liable to government for a tax carries its impact. However, some taxes lead to changes in the

economy that mean that the party who is liable for the tax is different from the party who

actually bears the final burden or incidence of the tax. For example, while company tax is a

liability of businesses, so they carry its impact, a business may pass the burden of the tax on to

households through higher prices or lower wages, so that households bear its incidence. See

sections 3.3 and 3.4 for more a detailed explanation of incidence and how it is measured.

Interpreting the incidence tables

The incidence results for each tax are reported throughout this section in the form of tables. The

interpretation of these tables can be understood by using the table for labour income tax, which

appears below, as an example.

Table 5.10 Incidence of Introducing Labour Income Tax (nominal changes)

change in $m

Tax revenue: a 109,422

Change in private income by source:

After tax labour income -149,505

After tax private capital income -6,349

After tax private land income - residential -583

After tax private land income - rural -835

After tax private land income - urban -1,585

After tax private other rent income -1,240

Transfers -3,084

Change in total private income: -163,179

% change

Change in total private income: -19.0%

Change in prices:

Price essential consumption -1.4%

Price non-essential consumption -1.4%

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

Notes to table 5.10:

a. Tax Revenue is the net revenue from introducing the tax. It includes any offsetting revenue losses from

changes to other tax bases. It is measured by the fall in lump sum tax revenue that is required to keep the

budget in balance given the income tax collections.

b. The changes in the table are in nominal terms. In MM900, all prices are expressed relative to the nominal

wage (the numeraire). However, the incidence of taxes depends on the real impacts, so care should be

taken when interpreting the nominal results. Thus, with a fixed nominal wage, an increase in consumer

prices can be interpreted as a reduction in the real wage.

c. Changes in capital and transfers are included only for completeness, and do not have policy implications.

The amount of revenue raised from labour income tax is shown in the top row of the table. This

amount refers to the year 2009/10, but under a ‘normalised economy’. This normalised

economy abstracts from the strength of Australia’s mining exports and terms of trade growth

since 2004/05. It also removes the effects of the global financial crisis. As such, revenue

generated from this process will not align with projected outcomes for the 2009/10 financial

year and the actual tax collected will differ from the amount shown in the table.

A key observation for each tax is whether the incidence will be felt through lower nominal

income or higher prices or some combination. This can be ascertained by comparing the

percent fall in nominal total private income with the percent rise in consumer prices. For labour

income tax, the main incidence is felt through nominal private income, which is 19.0 per cent

lower as a result of labour income tax. This is slightly offset through consumer prices, which in

fact are lower, not higher, as a result of labour income tax.

For taxes where the incidence is felt mainly through lower incomes, the table can then be used

to show what types of income mainly bear this incidence. Specifically, where the incidence is

felt through lower incomes, the table shows how the total fall in nominal private income,

expressed in $ millions, is distributed across different types of income. This information can

then be matched to the types of incomes of different households, to ascertain how different

households are affected, which would assist in making judgements about the equity of the tax.

In this case, the table shows that most of the reduction in nominal private income is through a

large fall in labour income, implying that the incidence of the tax is mainly borne by households

who supply labour.

Changes in incomes from labour, the three types of land and ‘other rent’ are of most interest for

incidence analysis. ‘Other rent’ refers to income from sources of ‘economic rent’ other than

land, such as natural resources, barriers to entry or brand names.

Changes in capital income and transfers are included only for completeness. They reflect

specific model assumptions, and do not carry policy implications. Any changes in capital

income only reflect the assumption of a fixed propensity to save for Australian households. As

total income (which includes the value of net government transfers) changes, savings and

domestically owned capital stock will change in the same proportion. Since the rate of return to

capital is fixed on international capital markets, any change in private capital income only

reflects changes in the capital stock owned by Australian residents. Likewise, changes in

transfers from overseas only reflect the assumption that these transfers are fixed as a share of

nominal GDP. Therefore, the movements in incomes from capital ownership and transfers will

move with incomes generally. They are not closely linked to the nature of the tax under

consideration, and do not have policy implications.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

While the incidence of labour income tax is felt through lower nominal incomes, for many other

taxes the incidence is felt mainly through higher consumer prices. The table reports two

different measures of the change in consumer prices, which refer to baskets of essential

consumption and non-essential consumption. This is so the relative burden on households with

different levels of income can then be examined. In the table above, labour income tax leads to

prices of essential and non-essential consumption falling by the same amount of 1.4 per cent,

implying a neutral result. However, for some taxes, the percentage rise in the price of essential

consumption may be larger than for non-essential consumption. This would affect low-income

households more than high-income households, because essential consumption will make up a

higher proportion of the budgets of lower income households. Hence, a comparison of the

percentage impacts of these two different measures of consumer prices assists with judgements

about the equity of a tax.

In any general equilibrium model, all prices are determined in relative terms, but expressed in

nominal terms. To do this conversion to nominal terms, one good is chosen to be the

“numeraire”. This good has a set fixed price, and all other prices can be expressed relative to

that fixed price. In MM900, the average nominal wage has been chosen as the numeraire.

Although the choice of numeraire does not affect the modelling results in real terms, it does

need to be taken into account in interpreting nominal price changes. For example, a fall in the

real wage could come about through either a reduction in the nominal wage or an increase in

consumer prices. In MM900, it will always be the latter, because the nominal wage is fixed.

Therefore, any price increases reported in the incidence tables can simply be interpreted as a fall

in the real wage.

Department of the Treasury

The Excess Burden of Australian Taxes

March 2010

ABCD

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

The KPMG logo and name are trademarks of KPMG.

Liability limited by a scheme approved under Professional Standards Legislation.

5.1 Summary of Excess Burden Results

The efficiency cost of each tax is reported in Table 5.1 below, which contains KPMG

Econtech’s estimates of the marginal and average excess burden of each of the 19 Australian

taxes in this study. It shows that some of Australia’s taxes are much more inefficient than

others. Excess burden estimates range from negative values (indicating a welfare improving

tax) to almost 100 cents (indicating that for every dollar of revenue raised, there is an additional

dollar of welfare loss).

Table 5.1: Marginal and average excess burdens of Australian taxes

(cents of consumer welfare per dollar of revenue)

Rating Tax MEB AEB

Low Tobacco excise a -8 -23

Low Import duties b -3 -7

Low Petroleum resource rent tax c 0 0

Low Municipal rates 2 1

Low GST 8 6

Low Land taxes d 8 6

Low Alcohol excise and WET a 9 7

Medium Fuel taxes 15 10

Medium Stamp duties other than real property e 18 18

Medium Luxury car tax f 20 9