Note: Where available, the PDF/Word icon below is provided to view the complete and fully formatted document
Treasury modeling claims incorrect

Download PDFDownload PDF


Peter Reith



The fundamental flaw of the Government's attack on Fightback! is now being accepted by leading commentators, for example:

‘ Treasury's bottom line is based on a nonsense, and is therefore clearly wrong.'

(Tim Colebatch, "The Age", Friday 6 March 1992)

And, now that the technical experts are reading the fine print of the Treasury's methodology, further technical errors are revealed.

The Treasury papers released under Freedom of Information papers last Tuesday contain a critique of the STATAX economic model developed by Dr Neil Warren. It was this model that was used to analyse the distributional impact on households of the taxation and expenditure changes proposed in Fightback!

Dr Warren is widely acknowledged as Australia's leading expert in his field - which is the reason he was commissioned by the Coalition.

Treasury even sought his advice - albeit in a limited way - when it first started to put together its GST Scare Campaign Model.

Unlike Dr Warren's work, this latter model has not been subject to public and academic scrutiny.

Dr Warren has provided comments on the Treasury critique of his model and the points are set out below:


This Appendix would have benefited considerably from a discussion with me about STA TAX and how it models the indirect tax system. It would appear that the Treasury have assumed that I have followed their 1985 methodology to the letter - this is not the case since STA TAX has been considerably extended over recent years to address a broad range of issues other than those for which the Treasury model was originally


Below, I shall address the three criticisms o f STATAX made by Treasury in Appendix 2 o f their Summary document.


Criticism 1: STATAX applies a Price equation which assumes markups are ad valorem in nature, modelling the following equation:

STATAX models prices using the following basic methodology:

P = (1+E). (1 + WM). (1 + WT). (1+RSM). (1+RST) (1)

where 1 = Product Price (or ex-factory price) E = Excise Duty per unit o f output WT = Wholesales Sales Tax rate

WM = Wholesales Markup rate RST = Retail Sales Tax rate RSM = Retail Sales Markup rate

Instead, the Treasury suggest that their model makes a ‘significant advancer by assuming margins are fixed in nominal terms, resulting in the following price equation:

P = ((1+E+NWM)(1 + WT))+NRSM. (1+RST) (2)

WT = Wholesales Sales Tax rate NWM = Nominal Wholesales Margin RST = Retail Sales Tax rate NRSM = Nominal Retail Sales Margin

The question to be answered here is on what basis did the Treasury assume that markups are fixed in absolute terms rather than in ad valorem terms. If markups are fixed in absolute terms, tax pyramiding would not be such a major issue - but in the past, the Treasury itself has gone to great lengths to make an issue out o f tax pyramiding. Moreover, if markups are designed as compensation for the holding cost

o f stock, then an ad valorem markup approach is more correct. They argue that assuming an ad valorem markup or ‘fixed percentage margins implies that a 10 per cent increase in the cost o f a good will lead to a 10 p e r cent increase in the cost o f transporting and selling i f and that this is unrealistic.

Quite simply, the Treasury have no firm basis on which to assert that fixed percentage margins are how shops set prices. A quick questioning o f a shopkeeper will indicate that they work on a ‘cost plus" basis. No shopkeeper would want to fix their margins in absolute terms on every product - after all, how would they administer such a system in a shop with say 7000 item lines. Moreover, while the freight cost o f a good may be fixed, the holding cost o f this good is not. A t most we have to admit that some elements o f margins are fixed and some are ad valorem.

The Treasury cannot assert that they have made a "significant advancer.

Criticism 2: STATAX does not feed back the reduction in the cost o f capital goods to reduced costs o f producers

This criticism is accepted to the extent that STATAX does not currently highlight how reductions in capital costs influence producer cost and therefore estimates o f the price index. However, STA TAX does allocate the benefits o f these reduced costs of capital to consumers and factor owners along the lines o f the assumptions set out in

Warren (1987: Appendix C).

Criticism 3: STA TAX operates at a level o f detail substantially below that o f PRISMOD

This is probably the weakest o f the criticisms by Treasury o f STATAX. While the Treasury m ight work with a transaction table which has some 1400 commodities with 116 users, when they come to undertake their model their "calculations are performed in respect o f 109 industries and 116 users." It is then said that "It is understood that the level o f detail in STA TAX fails well short o f this.' Treasury only had to ask for this issue to be clarified but d id not. In STATAX, "calculations are performed in respect o f 109 industries and 116 users? as in PRISMOD. The only difference between STATAX and PRISMOD is that when STATAX works with the Transaction table, only

109 industries and 116 users are identified, while PRISMOD considers 1400industries and 116 users. This difference is minor however, given that in PRISMOD itself, the 1400 industries are aggregated to 109 before the actual m odel is run.


Property responding to the myriad o f side points made in this Attachment can only rightly be done b y a comparison o f the facts rather than mere assertion. The Treasurys comments are based primarily on their understanding o f how STATAX operates at its most detailed and the data with which it operates.

For example, the Treasury questions how the Fightback! aggregate CPI differs from the average derived across all HES households. The reasons fo r this can be explained quite simply despite a Treasury assertion that an inconsistency exists. The simplest explanation is that the Official CPI is the price index which applies to "wage and salary earners in major metropolitan areas excluding the top 10 percent (by income) o f this group1 . Clearly, the average price index across the 140 groups will not result in this average being repeated. This is because including non-working households in the average would increase it because the lower income groups have a higher price index than middle and higher income groups who typically contain

workers. In fact, the Treasury estimates a lower HES based average (Attachment 4, p4, para 11).’

* * * * *

6 March 1992 Somerville

Contact: (06) 277 4045 D35/92