Chapter 3
ECONOMIC MODELLING
Economic modelling of macroeconomic effects
3.1 An economic model is a representation of an economic system. In applied economics a
model is likely to be expressed in a computer program in which data (the `input') are
processed and manipulated to produce results (the `output').
3.2 Model building usually consists of two main stages. The first is to develop the
structure of the model, setting out what factors affect which variables. The second, using
econometrics, [1] is to estimate the actual strength of the
relationship postulated.
3.3 In estimating the economic effects of a change in policy or some other external
factor, the model is 'shocked' by varying one or more of the parameters.
3.4 Two Australian models are the MONASH Model [2] and the
Murphy Model. [3] These two models were pitted against each
other during the debate surrounding the introduction of a goods and services tax (GST)
which will commence on 1 July 2000. The models produced different results.
3.5 The hallmark of a good model is the accuracy of its predictions. No one is yet in a
position to assess either model on this criterion, however, because the actual impact of
the GST is as yet unknown.
3.6 Both the MONASH and Murphy models are computable general equilibrium models. This
means that they follow the impact of external changes all the way through the economy. It
is the branch of economics that is the most technical and least accessible to the
non-professional - even a definition of general economic equilibrium is difficult to
formulate. The main difference between the MONASH and Murphy models is the different
elasticities they assume - the differences in the estimated strength of relationships
between economic variables.
3.7 The concept of economic equilibrium suggests that market forces are in balance,
that there is equality between the demand and supply for each commodity.
3.8 General equilibrium analysis is distinguished from partial equilibrium analysis by
the requirement that the price of every good is free to vary and that all markets must
clear. In partial equilibrium theory a specified part of an economy is analysed while the
influences impinging on this sector from the rest of the economy are held constant.
3.9 General equilibrium models are designed to provide projections rather than
predictions. That is, over the projection period factors outside of the model (natural
disasters, new inventions, mineral discoveries and so on) will have their own economic
effects. General equilibrium models estimate the economic impacts of policy changes,
rather than trying to predict the state of the economy at some time in the future.
The revenue neutrality requirement for the Review of Business Taxation
3.10 The terms of reference for the Ralph Review of Business Taxation required its
recommendations to be revenue neutral in respect of the outcome from reforms to taxation
of income from investment and from changes in the capital gains tax.
3.11 The Ralph Report concluded that the total package of business measures (those
proposed in ANTS and those recommended in the Ralph Report) is significantly revenue
positive against the revenue generated by the current legislation and practices.
3.12 It did, however, caution that although the revenue estimates are based on the best
available data and methodology it is not realistic to attach a high degree of precision to
individual revenue estimates because the data upon which the estimates are based are often
seriously inadequate. In addition it is necessary to make judgments regarding behavioural
changes by taxpayers in response to a measure and these are often not easily judged.
3.13 History has shown that estimates relating to new tax measures, such as capital
gains tax, can sometimes be significantly in error and understate the revenue actually
raised. At the 22 October public hearing, the committee learnt that the estimate of what
the capital gains tax would raise was between $500 million and $600 million, but what was
realised was much more than that, and was in the billions very quickly. [4] It is also true that the revenue losses from avoiding tax by
converting income into capital gains may be underestimated. In terms of revenue outcomes
for this proposal it is clear that estimates of the position and the revenue impact will
vary with the application of different economic models and assumptions. The extent to
which differences between estimated and actual revenue outcomes for one measure will be
offset by those for other measures will influence the margin of error surrounding the
overall revenue position reported for the package of reform measures.
3.14 On balance, the Ralph Report concluded, revenue estimates are likely to understate
the overall positive impact on revenue of the recommended package, possibly to a
significant extent. [5]
Modelling underpinning the Ralph Report and the Government's response
3.15 The aim of the modelling done for the Review of Business Taxation was to estimate
how the proposed reforms might affect the distribution of economic activity between
different industries. It was not intended to estimate the impact of the reforms on the
overall level of economic activity and consequently does not include the anticipated
growth dividend in the modelling results.
3.16 Mr Chris Murphy of Econtech conducted the modelling with assistance from the
Department of Industry, Science and Resources and the RBT secretariat.
3.17 The modelling was conducted in two stages. The first was to estimate the change in
taxes paid by industries as a result of the business tax reforms and the second involved
entering these estimates into Econtech's MM303 model to estimate the extent to which they
might alter the pattern of industry production. MM303 models the production of over 300
commodities by 107 industries.
3.18 The results of the modelling were included in the Ralph Report as a guide to how
the impact of the reforms might differ across particular industries. It does not indicate
the overall impact of reforms on the level of activity for individual industries.
3.19 The major conclusion of the Ralph Report modelling is summarised as follows:
Given the degree of uncertainty that must be attached to estimated effects on
individual industries, the major conclusion that can be drawn from the model results is
that the business tax reform measures will neither advantage or disadvantage, in relative
terms, any industry sector to a significant degree. [6]
3.20 The modelling of the estimated changes in long-term production for different
industries indicated that the effects of reform on any particular industry are not likely
to be markedly different across all industries.
For example, the difference between the communication services industry, which receives
the greatest relative benefit (0.3 per cent), and the industries with the largest relative
detriment (-0.2 per cent) [mining, electricity, gas and water and transport] is only about
half a percentage point, which is not a large figure compared to the changes in production
generated by indirect tax reform and the influence of other factors affecting these
industries. [7]
3.21 The changes in production are relative to what would have happened in the absence
of the reforms and, as noted above, do not include the general increase in production from
the growth dividend expected by Mr Murphy and the Ralph Committee to flow from the
recommendations in the Ralph Report.
3.22 Mr Murphy presented evidence to the committee at its public hearing on
12 November 1999 based on modelling industry effects of the Ralph Report
recommendations and the Government's response of 21 September and 21 October 1999.
3.23 His results were that the combination of the ANTS reforms and business tax reforms
put the four sectors of agriculture, mining, manufacturing and finance and insurance 'well
ahead' (2.6 per cent, 7.2 per cent, 3.3 per cent and 1.9 per cent respectively). In
addition, he concluded that the recommendations of the Ralph Report 'really make very
little difference to the industry impacts from ANTS. The reason for that is that the Ralph
package is only one-eighth the size of ANTS.' [8]
3.24 In looking only at the impacts of the recommendations in the Ralph Report,
Mr Murphy modelled the move to a 30 per cent company tax rate and the elimination of
accelerated depreciation. He applied assumptions concerning elasticities of substitution
for labour and capital (0.75) and between seven types of capital with a variation in tax
lives (0.5). He modelled a representative project that used all of the different types of
capital in the proportions they are used in the economy.
With employment assumed to be held fixed in aggregate, there is a long-term modelling
assumption that this flows through to a gain in GDP of 1.5 per cent, which in turn flows
through to a gain in consumption of 0.4 per cent or $1.7 billion per year. [9]
3.25 He also looked at the sensitivity of the estimates of elasticities he used in
respect of the substitution between labour and capital and found that the GDP gain may be
as low as $1.15 billion and as high as $2.17 billion per annum. Similarly the range of
elasticities of substitution between different types of capital provides estimates of
between $1.58 billion to $1.75 billion per annum. His main conclusion was that the capital
stock in the long term (10 years) increases by 4.6 per cent.
The Dixon modelling
3.26 The original intention of the committee was to obtain an analysis of the broad
macro-economic consequences of the recommendations in the Ralph Report. Professor Peter
Dixon, Director of the Centre for Policy Studies at Monash University, was commissioned by
the committee to undertake such analysis. In his report to the committee he advised the
committee that to reach firm conclusions about the likely effects of the Ralph Report
proposals would require considerably more time than the ten days allowed for the
preparation of the report.
3.27 Professor Dixon's report considered three recommendations of the Ralph Report:
- the reduction of the company tax rate from 36 per cent to 30 per cent;
- the elimination of accelerated depreciation; and
- the halving of the rate of capital gains tax while eliminating indexing and averaging.
He thought it reasonable to take this approach because their implementation does not
depend on the implementation or non-implementation of the other recommendations.
3.28 Professor Dixon constructed a hypothetical project and sought to ascertain if such
a project would be financed, or would be attractive to different classes of investors,
including Australian residents and non-residents. He considered that the post-tax rate of
return would be the key to determining if a project would be financed, and not the pre-tax
position. The post-tax rate of return depended on the pre-tax characteristics of the
project together with the taxation rules and the characteristics of the financier. For
these purposes Professor Dixon made certain tax, dividend, retention and interest
assumptions.
3.29 Professor Dixon reached tentative conclusions on the basis of work able to be
completed as follows:
- the proposed cuts in company tax rates will have a very minor positive impact on the
attractiveness to Australian residents of investment projects in Australian companies but
will have a greater positive effect on the attractiveness of these investments to
foreigners;
- the proposed elimination of accelerated depreciation allowances will have significant
negative effects on the attractiveness of investments in Australia to all classes of
financiers;
- the proposed changes to capital gains tax will increase the attractiveness of
investments in Australia to Australian residents but not to foreigners;
- in total, the impact effects of the Ralph proposals will be to reduce the attractiveness
of investment projects in Australia to all classes of financiers;
- the Ralph proposals will generate windfall gains to owners of recently established
projects and windfall losses to owners of long-established projects; and
- the proposals amount to an increase in the rate of taxation of capital income accruing
to all classes of financiers.
3.30 In evidence Professor Dixon confirmed that in his assessment capital stock would
decrease over the long term. 'In the long run you wind up in a position where you actually
have less capital in the economy that you otherwise would have had.'
3.31 Professor Dixon also indicated that his findings would not be the same if he had
modelled the impact on small business because accelerated depreciation is not being
abolished for such taxpayers. He considered that he would need to do some calculations in
this area to be more definite in his conclusions, particularly in respect of the matter of
the effect it may have on the long-term negativity his findings to date have produced.
3.32 The committee is mindful of Professor Dixon's statements that his conclusions are
tentative and are derived from modelling three of the Ralph Committee's recommendations,
based on one hypothetical project. It notes Professor Dixon's opinion that it would be
technically feasible to make a detailed and convincing quantitative analysis of the Ralph
proposals, given more time, and his frustration that so little time was allowed for his
work on a project of such significance. There are obvious difficulties in drawing
conclusions from such a limited study.
3.33 Professor Dixon estimates an increase of aggregate capital of 0.25 per cent in the
short term and a decrease in aggregate capital between 0.25 and 0.4 per cent in the long
run with a concomitant reduction in the incentive to invest. The Dixon model produced an increase
in aggregate tax on investment, which conflicted with the views of Treasury.
3.34 The results achieved are due in good part to the removal of accelerated
depreciation and the detrimental impact that such a measure had on the particular project
that Professor Dixon modelled, which assumed an asset life of 20 years.
3.35 The Ralph Report and Treasury claim that the net effect of the abolition of
accelerated depreciation and the decrease in the company tax rate is a reduction,
not an increase, in the total tax wedge. [10]
Summary comments on modelling
3.36 A key differentiating factor between the results of the MONASH and Murphy models
appears to be the underlying project modelled. There appeared to be no disagreement
between all parties as to the validity of the underlying economics involved, as Treasury
made very plain. [11]
3.37 The tax effect appears to be the reason for the difference in results. Professor
Dixon's model produced a tax increase and the data used in the Ralph Report estimated a
tax decrease. Treasury attributed the difference in results to the comparably long assumed
effective life of the asset of twenty years in the project modelled by Professor Dixon and
speculated that using a three to seven year asset life as Treasury had done would produce
results on the other side of zero. The result of this would have been more compatible with
those in the Ralph Report.
3.38 The dilemma for the committee is in identifying the correct economic variables
and, as the expert witnesses have put to the committee, there is no clear right or wrong
in this area. The challenge is to find the right balance - drawing on the best available
but seriously inadequate data, choosing from a variety of acceptable methodologies and,
using valid underlying economics, to make a judgment regarding behavioural changes in
response to each measure.
Growth dividend
3.39 The Government's growth dividend is an estimate of the increase in GDP
attributable to the Government's proposed business tax reform, reflecting an increase in
Commonwealth tax revenue.
Informed judgment or modelling? The Government's response
3.40 The Ralph Report indicated that at the most fundamental level the growth dividend
is simply one aspect of the behavioural responses typically taken into account when
developing revenue estimates for particular tax measures. However, the effects of
behavioural responses are extremely difficult to estimate.
3.41 The Ralph Committee did not commission a study of the likely impact of the
proposed business tax reforms on Australia's economic growth because it concluded that
such studies typically involve models requiring a large number of assumptions that are
difficult to validate. In addition overseas experience has shown that alternative models
can give markedly different results.
3.42 The Ralph Report concluded that a growth dividend of 3/4 per cent of GDP by
2009-10 is likely to be conservative and this would deliver additional Commonwealth
revenue of $650 million by 2004-05. However, in order to ensure that the overall outcome
of the recommendations was clearly conservative, only $500 million of the expected revenue
gain was included in the revenue estimates. Similarly, contributions to revenue in the
earlier years were also scaled back. The expectation is that the recommendations will
significantly reduce compliance costs, however, a compliance dividend was not included in
the revenue estimates but was identified as further supporting the case for a significant
growth dividend.
Conflicting estimates
3.43 Professor Peter B Dixon and Maureen T Rimmer of the Centre of Policy Studies,
Monash University, hold a view contrary to that expressed in the Ralph Report. They have
concluded that it is technically feasible to successfully model the likely impact of the
business tax reforms on the growth of the economy by running a MONASH [12] simulation of the effects of an increase in capital taxation
combined with the effects of a compensating reduction in other taxes. Such analysis would
however, take some three months and obviously the short reporting deadline associated with
the committee's inquiry has prevented the completion of such analysis.
To complete our work for the Senate Inquiry we hope to run a MONASH simulation of the
effects of an increase in capital taxation combined with the effects of a compensating
reduction in other taxes. We think that such a simulation will indicate the broad
macro-economic consequences of Ralph. However, we should emphasise two points: (a) on the
basis of our analysis so far we think that it would be technically feasible to make a
detailed and convincing quantitative analysis of the Ralph proposals and (b) at least 3
months would be needed to complete such an analysis. [13]
3.44 Mr Murphy of Econtech also differed from the conclusions in the Ralph Report
concerning the feasibility of modelling economic growth. He provided the committee with a
report based on the RBT model which he had constructed specifically for this purpose. He
estimated that the general economic effects of the proposal to cut the company tax rate
and abolish accelerated depreciation would be a gain in GDP of 1.5 per cent.
Should the growth dividend be included in the estimates?
3.45 The Ralph Report was of the view that because of the constraint of revenue
neutrality imposed upon the Review, it was reasonable to include in the estimates the
amount of projected increase in Commonwealth revenues as a result of the business tax
reform measures outlined in the Report.
3.46 The economic environment against which the revenue estimates were prepared is the
same as that used for the purposes of preparing Budget revenue estimates.
3.47 The issue for consideration is therefore whether a growth dividend of this nature
would usually be considered a genuine offset to be included in revenue estimates for
budgetary purposes. The committee notes that it is not the accepted budgetary practice to
include such a figure.
3.48 However, inclusion of the growth dividend, if one is actually realised, does give
a more accurate picture of whether the proposed reforms meet the criteria and constraint
of revenue neutrality. As the figures have been included by the Government, it is even
more important for the other revenue estimates to be robust.
Size of growth dividend
3.49 The Ralph Report is of the view that the growth dividend figures included in the
revenue estimates are conservative. However no evidence was provided to support this
assertion. Although Professor Dixon indicated that it was feasible to model a growth
dividend he was not in a position to proffer an informed opinion at this stage on the
outcome of what any such modelling may produce.
3.50 The committee is, therefore, in the unenviable position of not being able to make
a judgment in respect of the existence or size of any growth dividend from these proposals
without access to any economic modelling on the subject to provide guidance.
Footnotes
[1] The setting up of mathematical models describing economic
relationships.
[2] Since 1993, the Centre of Policy Studies has developed
MONASH, a dynamic computable general equilibrium model of the Australian economy designed
for forecasting and for policy analysis. Like its predecessor, ORANI, MONASH has a high
level of microeconomic detail. Unlike ORANI, it has a strong forecasting ability. The
model generates forecasts for 113 industries and 115 commodities, which can be transformed
into forecasts for 860 sub-commodities, 341 labour occupations, 56 regions and many types
of households. The initial conditions used by MONASH are obtained from an updated version
of the 1986-87 input-output tables published by the Australian Bureau of Statistics.
[3] MM2 is a quarterly macroeconometric model of 700 equations
used for macro industry policy analysis and forecasting. It has short run Keynesian
properties and recognises that financial markets are deregulated and forward looking and
has a neoclassical long run which gives an integrated and consistent treatment of
investment, employment and pricing decisions in the business sector. MM2 has 18 industry
sectors. MM303 is a highly detailed model of the Australian economy, designed to model the
long-run effects of micro-economic reforms. It distinguishes 305 separate products by 107
industries.
[4] Evidence, 22 October 1999, p. 51.
[5] Review of Business Taxation, A Tax System Redesigned,
1999, p. 693.
[6] ibid., Section 25, p. 749.
[7] Review of Business Taxation, A Tax System Redesigned,
1999, p. 747.
[8] Senate Finance and Public Administration References
Committee, Hansard Transcript of Evidence (hereafter Evidence) , 12
October 1999, p. 191.
[9] ibid., p. 193.
[10] See Mr Smith in Evidence, 12 November 1999, p.
233, Professor Dixon in Evidence, 12 November 1999, p. 238.
[11] Mr Smith, in Evidence, 12 November 1999, p. 233.
[12] A dynamic computable general equilibrium model of the
Australian economy designed for forecasting and for policy analysis.
[13] Peter B Dixon and Maureen T Rimmer, Centre of Policy
Studies, Monash University, Company taxes, Depreciation allowances and Capital gains
taxes: some effects of Ralph, Paper prepared for the Senate Finance and Public
Administration References Committee: Inquiry into Business Taxation Reform, 10 November
1999, p. 10.
Top
|