Note: Where available, the PDF/Word icon below is provided to view the complete and fully formatted document
Dawkins $3 billion hole becomes over $4,2 billion hole with withholding tax

Download PDFDownload PDF


, Γ

Peter Reith



The statements by Treasurer Dawkins and his parliamentary secretary, Senator McMullan, about simplifying the income tax system by extending the withholding tax are a monumental furphy attempting to distract attention from the fact that the Government needs to raise $3 billion or more from secret tax measures.

The fact is that the introduction of a flat 20% withholding tax could cost more than $1.2 billion in lost revenue - not save money from tax evasion as has been suggested to the media.

Mr Dawkins is simply worsening the fiscal problem he outlined in the budget, not improving it.

Detailed economic modelling done by leading expert Dr Neil Warren of the University of New South Wales for the Tax Institute of Australia (copy attached) shows that a 20% withholding tax with no limit on the amount of interest income earned by a taxpayer eligible for the tax concession would cost the Government more than $1.2 billion in 1990-91 dollar terms (see table 4).

Further it should be noted that the figure of $3.7 billion in tax evasion from interest income that has been bandied about in the media in recent days is based on dated information. Dr Warren's paper notes that this figure is calculated from Tax Office and Bureau of Statistics figures from 1988-89.

That is, a full 2 to 3 years before the introduction of the Tax File Number system on bank accounts in July 1991. The Tax File Number system was specifically designed to capture any tax evasion on interest income and according to recent government reports has done so. Is the Government correct or incorrect on the TFN?

Therefore the question still remains: what is the Government's secret tax agenda for raising $3 billion in revenue to reach a budget deficit of 1% of GDP in 1995-96, or for raising over $8 billion to balance the budget as Mr Dawkins has said he will?

Indeed, over the next few years the Government will have created over a $50 billion fiscal hole. How will this be paid for?

Obviously a lot of money will need to be raised from the fringe benefits tax or from the prescribed payments system. Is he in fact going to raise tax on interest earnings of self-funded retirees, for example. Or is the Government really considering a GST?

25 August 1992 Canberra

Contact: David Turnbull (06) 277 4277 D133/92



WAKRJETJ VEXJTTJRJSS I> t:y L ,t:d 3 Rosewall St Willoughby 2068 19 March 1991

PH/FAX 02 958-6011

Geoff Petersson -

Technical Director The Taxation Institute of Australia 64 Castlereagh Street Sydney 2000

Dear Geoff Γ': ;;

RE: Project on "Resident Withholding Tax"

Please find attached the report commissioned from Warren Ventures P/L by the Taxation Institute on the costing of your Resident Withholding Tax proposal.

As you will see, we have not made any comment on what the rate and threshold for a Resident Withholding Tax should/could be. I feel that this is ultimately a decision for the Taxation Institute, using the information in this report.

If you should have any questions, please do not hesitate to contact me.

If possible, I would appreciate a copy of your final report on the Resident Withholding Tax when it is released.

Yours sincerely,

Dr Neil Warren Director

- 2 -


Costing the Ή Α "Resident W ithholding Tax" Proposal


The objective of this report is to estimate the rate (t) and threshold (IT) of a Resident W ithholding Tax which when introduced, would be revenue neutral in its impact on the Personal Income Tax.

This report will assume that the revenue expected by the Federal governm ent from the introduction of the second phase * of the Tax File Number (TFN) system - assigning TFN to financial institution accounts and shareholdings - would also accrue from the introduction of a Resident W ithholding Tax system.

In the first section below, we shall examine the estimates made by the Federal Treasury of the revenue expected from the second phase of the TFN system. Attention will then be focussed on the cost of various Resident W ithholding Tax systems. "


The primary constraint on the design of any Resident Withholding Tax is the funds to be allocated to financing the proposal. This report will assume that the funds raised from the second phase of the TFN would also arise from the introduction of a Resident W ithholding Tax. The difference in the Tax Institute's proposal is that the surplus funds raised would be used to offer a tax concession to recipients of interest income (but not to dividend income recipients).

The Federal Treasury estimates of the revenue from the second phase of TFN are shown in Table 1.

TABLE 1 Revenue from Income Matching using TFN

1989-90 1990-91 1991-92 1992-93 1994-95

$10m $15m $115m $225m $310m

Source: "1988 Business Tax Reform" Statement, 25 May 1988, p44

1 The first phase involved attaching the TFN to all earned income sources and more recently, to all government transfers.

- 3 -

There are several factors which might suggest that these estimates are extremely conservative.

1) The forecasts in Table 1 were m ade in early 1988 for a proposal which was to come into full force some three years hence. Therefore, these forecasts should be revised by the Treasury and placed in 1990-91 values, using the current level of savings, interest rates and personal income tax rates.

2) The information in Table 2 indicates that the Australian Taxation Office is assumed to take some years before it is able to fully utilize the data which will be made available to it under the new system; or that individuals will take some time to comply with the system's requirements - in fact up to three years. This seems a

considerable am ount of time, especially since the deeming rules under the Social Security system, which operated from 1 March 1991, are likely to force greater accountability amongst interest and dividend income recipients earlier than previously expected^.

3) In 1988-89, the Australian Taxation Office reported some $8696m in "Gross Interest" from individuals (excluding trustees or manually assessed and issued assessments^). In contrast, the Australian Bureau of Statistics (ABS) reported "Household Income" from "other interest,etc received" defined as income from "receipts from companies on notes and debentures, from private and government banks on deposits and from public authorities on public authority securities"'* at

$21159nA The discrepancy between the figure reported by the ABS and by the Australian Taxation Office is considerable - $12463m. If the ABS estimate of interest is an accurate measure of interest income from individuals, then only approximately

41% of this income source is reported to the Australian Taxation Office. Therefore, if we were to assume that the average marginal tax rate confronting taxpayers with investment income was say 30% in 1988-89, then the personal income tax collections should be higher by some $3740m in 1988-89. Even if there are numerous reasons why there are difference between the ABS and Australian Taxation Office measures, they are unlikely to be able to explain all the difference away. This leads us to

2. It is also likely that the 10% deeming rule which has applied to all savings by Government transfer recipients since March 1 1991, could increase the interest income of individuals markedly and this may mean an increase in personal income tax collections and

not just a reduction in Government transfer payments. While any additional tax revenue resulting from deeming cannot rightly be said to arise from a Resident Withholding Tax or from the second phase of the TEN system, its introduction has the potential to influence a Resident Withholding Tax. This would arise from more current pensioners loosing their pension and becoming subject to the Resident Withholding Tax. It is our expectation though,

that this factor may not be that significant in the short term. 3. Taxation Statistics 1988-89, Australian Taxation Office, AGPS, Table 1.5, p42 4. Australian National Accounts, National Income and Expenditure, 1975-76, Australian Bureau of Statistics, Catalogue No. 5204.0, p 91.

5. The figures for 1988-89 are shown in Table 14, National income and Expenditure I9M-90, Budget Related Paper No. 2, 1990-91 Budget, Australian Bureau of Statistics Catalogue No. 5213.0, Table 14, p l2.

- 4 -

conclude that the Treasury estimate of the potential revenue from investm ent income through income matching using the TFN could be considerable - much more than the figure in Table 1.

The fact is that the Treasury estimate of increased income tax revenue from attaching TFN to investment income sources is very much a m inimum estimate. In the examination of the Resident W ithholding Tax System below, we shall assume that some $300m to $400m would be raised from the second phase of the TFN and likewise, from the introduction of a Resident Withholding Tax. This is in our view, a very modest assumption.


The Resident W ithholding Tax is applied to each individual on the basis set out below. However, before proceeding, let us firstly define the key variables in our analysis:

Y2 = Y1 - 1 Y3 = Y2 + IT WT = t . I

where: Y1 = total taxable income I = interest income IT = resident withholding tax threshold t = withholding tax rate WT = resident withholding tax on I


T1 = Personal Income Tax on ΥΊ T2 = Personal Income Tax on Y2 T3 = Personal Income Tax on Y3

Rule Ί. No tax payer shall pay more tax ui.der the new system than they did under the original system. If it happens that more tax was paid under the new system, then the difference would be refunded. This will ensure that the new effective tax

rates on interest income of individuals is less than or equal to that under the current tax system.

- 5 -


4 1

4 4 8 1





« 1







Rule 2. If the tax payer is receiving any governm ent benefit or pension, it is assumed that they are entitled to apply to their bank for exemption from Resident W ithholding Tax. This ensures that no pensioner or government benefit recipient is any worse off or would have to start filing tax returns when they previously did not fill out such returns.

All pensioners and government benefit recipients are therefore excluded from this study and its costings.

Rule 3. If the individual’s interest income (I) is less than or equal to the threshold (IT), then their final tax liability (T) is:

T = T2 + WT

ie the tax liability is the withholding tax plus the personal income tax on their other sources of income.

Rule 4. If the individual’s interest income (I) is greater than the threshold (IT), then their final tax liability (T) is:

T = T2 + (T1 - T3) + t.IT

The person's tax liability is therefore the sum of: i. the personal income tax on income other than interest income (T2), ii. the withholding tax on interest income up to the threshold (t.IT), iii. and personal income tax on interest income above the threshold when it is

taxed at the marginal tax rates that apply to total income above (Yl+IT). This is equal to (T1-T3).

This method of administration ensures that the concession on the threshold amount of IT accrues at the lowest marginal tax rate that an individual confronts on interest income (I), not the highest. This is shown in Chart 1 for several different individuals

using the personal income tax schedule effective from 1 January 1991 which is shown in Table 2 and illustrated in Chart 2.


Costing the Resident W ithholding Tax proposal requires detailed data on the income of individuals in the population.

The data source used in the estimates below was the 1988-89 Household Expenditure Survey unit record file. This data file contains detailed information on some 7225 households surveyed for their income and expenditure patterns in 1988­ 89. This data source was adjusted to take the values up to the figures forecast for

- 6 -

1990-91 using a methodology detailed in W arren(l987,1990) and forecasts supplied by ACCESS ECONOMICS and published in their Budget Monitor, February 1991, Chapter 2.

The costing of the proposal was done against the personal income tax schedule effective from 1 January 1991 which is shown in Table 2 and Chart 1. This is different from the composite schedule for fiscal 1990-91 which is shown in Table 3. The reason for using the post 1 January 1991 schedule is that this gives an appropriate indicator of the cost of the proposal if it was introduced with the current personal income tax schedule. However, as will be show n below, the estimated cost of the Resident W ithholding Tax is only marginally different using either the schedule in Table 2 or 3.

Table 2 1 January 1991 Personal Income Tax Schedule

Income Range($pa) Marginal Tax Rate

0 - 5400 0%

5401 - 20700 20%

20701 - 36000 38%

36001 - 50000 46%

50001 and above 47%

Table 3 1990-1991 Personal Income Tax Schedule

Income Range($pa) Marginal Tax Rate

0 - 5250 0.0%

5251 -17650 20.5%

17651 - 20600 24.5%

20601 - 20700 29.5%

20701 - 35000 38.5%

35001 - 36000 42.5%

36001 - 50000 46.5%

50001 and above 47.0%

The cost in 1990-91 of a Resident W ithholding Tax when using the personal income tax schedule effective from 1 January 1991 is shown for a range of withholding tax rates (t) and thresholds (IT) in Tables 4,5 and 6.^

\6 . It is worth noting that the receipts from the Medicare Levy are not affected by the Tax concession on interest income.

- 7 -

The sensitivity analysis in Table 4 indicates that the cost of the Resident W ithholding Tax proposal is relatively insensitive to variations in the threshold. This appears to be due to the major cost of the proposal coming from the many small savers whose interest income is small in size but adds up to a considerable am ount in total.

TABLE 4 Costings of the Resident W ithholding Tax on Individuals at a 20% Rate using different Thresholds


30000 x300000 1230

20000 200000 1115

10000 100000 905

5000 50000 675

4000 40000 605

3000 30000 525

2000 20000 425

1000 10000 285

* assum ing 10% average return on investments

Table 5 indicates that the cost of the proposal can be reduced if a higher Resident W ithholding Tax rate is applied. The trend evident in Table 4 is again clear, but now with a lower cost.

TABLE 5 Costings of the Resident W ithholding Tax on Individuals at a 25% Rate using different Thresholds


30000 300000 930

20000 200000 850

10000 100000 690

5000 50000 515

4000 40000 465

3000 30000 405

2000 20000 325

1000 10000 220

* assum ing 10% average return on investments

- 8 -

Table 6 indicates just how sensitive the proposal is to changes in the rate of Resident W ithholding Tax. It should be noted that by setting the Resident W ithholding Tax rate at or above 20%, individuals with incomes over $20700pa are the only ones to benefit from the proposal (that is, their m arginal tax rates are greater than 20%).

The costs differs only marginally if we are to use the composite personal income tax schedule for fiscal 1990-91 as shown in Table 3. For example, with a Resident W ithholding Tax rate of 20% and a threshold of $2000pa, the cost using the personal income tax schedule in Tables 2 and 3 w ould be $430m and $450m respectively and

with a Resident Withholding Tax rate of 20% and a threshold of $5000pa, $680m and $715m respectively.

TABLE 6 Costings of the Resident W ithholding Tax on Individuals: Rate Sensitivity Testing


5000 20% 675

5000 25% 515

5000 30% 360

5000 35% 215

2000 20% 425

2000 25% 325

2000 30% 230

2000 35% 130


The above results indicate that a threshold on the Resident W ithholding Tax of $2000 (which represents savings of approximately $20000) and a Resident W ithholding Tax rate of 20% would cost just over $400m while a rate of 25%?, a cost of over $300m. There are however, advantages in opting for a 20% rate of Resident W ithholding

Tax. Firstly, it lines up with the lowest marginal tax rate, secondly, it minimizes the advantage to families from shifting investment income between the primary and secondary earner to reap any tax advantage (income splitting). Also, with a rate of 20%, this should minimize the incentive for couples to change their current income splitting behaviour^. Moreover, it should also offer similar advantages to couples

7. Income splitting occurs when there are marginal tax rate differentials between the adults in the family. This provides an incentive to shift any income which can be shifted between the partners (primarily investment income), to the tax payer with the lowest income, thus reducing the family's total tax liability. If the Resident Withholding Tax is set at 20%, this minimizes the need for splitting investment income by couples since they would both

confront very similar marginal tax rates on investment income, as long as their investment

- 9 - >

who do not currently income split, as currently accrues to those couples who do income split.

APPENDIX Data: Qualifications and Reservations

The estimates presented in this study are derived by applying the STATAX Income Tax model to the 1988-89 Household Expenditure Survey Unit Record data. The limitations of household surveys and their implications for personal income tax forecasting deserve outlining.

It is likely that the estimates of the cost of the Resident W ithholding Tax in the report above are if anything, underestimates of the true cost of such a proposal. This is because experience has shown that high income groups are often less likely to be respondents to such surveys.

In relation to the study in this report, it means that the cost of the proposal may be underestim ated to the extent that these higher income groups are undersam pled in the survey or they under report their income. In relation to the latter (ie underreporting of investment income) this may be less of a problem since we are costing a concession to a limited am ount of investment income (ie up to a threshold).


W arren, N.A., (1987), The Distributional Impact of a Change in the Tax Mix in Australia. (Australian Tax Research Foundation, Sydney,1987), Research Study No 6, pp 123.

W arren, N.A., (1990), A Goods and Services Tax for Australia: A Evaluation of the National Farmers' Federation Outlay Tax Reform Proposal. (Australian Tax Research Foundation, Sydney), Research Study No 11, pp 84.

income is below the Resident Withholding Tax threshold). It would also mean that couple currently splitting investment income, need not change how their investment income is split.

Chart 1 Effective MTR Schedule for different individuals when a Resident Withholding Tax is levied at a rate of 20% and has a

threshold of $2000 A. Y1-$2S000. US 1000


40 -

30 ■



01 *----1 --------1 ---1 ----1 -----1 — —u0 5 10 15 20 25 30 35 40T e e ti · lneeme<$000) ' · ' · 1 Jen PIT S * e d 4 e Eflecu* MTR

C. Y1-S35000. US 1000

T ut**· Income (5000)

— Effect*· MTR

B. Y1-S25000. US5000

Peeent 5 0 1---------





O' ■---------- -------------1------------1------------ 1------------ -------------*------------u

0 5 10 IS 20 25 X 35 40

Taxable Income (1000)

1 Jen PIT 9dte0ute — E ffect* MTR

D. Y1-S35000.1-S5000





36 20 30 40 5 10 Taxable Ineem# (S000)

1 Jen PIT Schedule I J w PIT Schedule Effect*· MTR







Chart 2

Personal Income Tax Schedule: 1 Jan 1991






) % $5400

I I I l

5 10 15 20 25 30 35 40 45 50 55 60 65 70

Taxable Income ($000)