Models

Review of economic modelling at the Treasury

Independent Computable General Equilibrium (CGE) model: original model for tax policy analysis

CGETAX model: improved model for tax policy analysis

Independent Macro-econometric model: for economic forecasts and macro policy analysis

Independent Short-term Forecasting model: for timely, short-term economic forecasts

Review of economic modelling at the Treasury (March 2017)

In 2016, the Australian Department of the Treasury commissioned me to undertake a review and recommend areas for further developing The Treasury’s economic modelling capability. In brief, the five higher priority recommendations of the review are as follows.

  • The new economy-wide forecasting model being developed by The Treasury should be a macro-econometric model. This allows an appropriate balance between strong short-term empirics and well-defined long-run properties.
  • Judgemental adjustments should be applied to the forecasts from the macro-econometric model, especially in the first year or two of the forecast period. This is to allow for information not already taken into account in the model.
  • For its CGE model analysis of the economy-wide impacts of tax policy, The Treasury should upgrade from using the IE CGE model to using the CGETAX model. This will increase the depth of the analysis for the five categories of tax that are covered by IE CGE, as well as add a further seven categories of taxes.
  • The Treasury should consider upgrading CAPITA, its model of the equity impacts of changes in taxes and benefits. The existing CAPITA reports impacts on current incomes, but this could be extended to include both current incomes and lifetime incomes to give a more complete picture of equity impacts.
  • The Australian IGR should be further developed to world best-practice. This means that it should be issued in the name of the Treasury rather than of the government, and long-term revenue projections should be developed from the bottom up rather than on a top-line basis.

Download the review of Treasury modelling

Download my report to The Treasury (March 2017). My report is also available from The Treasury web-site.

Independent CGE model (2012-2015)

The Independent CGE model was used for analysing tax policy, but has now been superseded by the CGETAX model, which incorporates several important improvements. This section describes the Independent CGE model, while the following section highlights the improvements incorporated in CGETAX.

As a Computable General Equilibrium (CGE) model, the Independent CGE model analysed the economic impacts of policy changes on economic activity, employment, trade and investment at the level of individual industries, impacts on households and impacts on the Australian economy as a whole. Compared to other CGE models, this model had a special focus on tax policy. This is reflected in the following model features.

  • sophisticated modelling of production in each industry, distinguishing nine types of produced capital, three fixed factors to capture economic rents, and eight occupations for labour. This allows more precise modelling of the impact and incidence of taxes on different inputs into production such as company income tax.
  • builds in highly-developed modelling of the business tax system and its economic impacts. It takes into account factors such as: the different tax treatments of debt and equity financing; the complex system of depreciation allowances and tax concessions; franking credits; and the potential for international profit shifting.
  • provides a rigorous measure of changes in consumer welfare or living standards, so that policy changes can be correctly evaluated in terms of the public interest.

History

The Independent CGE model was developed early in 2012 as a completely new model. Later in 2012 the modelling of business tax was further developed, with input from Treasury, to meet the modelling needs of the Australian Government’s Business Tax Working Group (BTWG). This model has been used by Treasury under licence on four occasions.

  • In 2012 modelling of a cut in the rate of company tax was included in the Final Report of the BTWG.
  • In 2014 Treasury officers published a more detailed analysis of a cut in company tax in the Treasury Economic Roundup.
  • In 2015 Treasury officers published an analysis of the efficiency of major taxes in a Treasury Discussion Paper prepared to support a review of the tax system.
  • In 2016 the model was used in another Treasury Discussion Paper, this time modelling the proposed cut in the company tax rate from 30 to 25 per cent.

CGETAX model (updated December 2017)

From 2015 to 2017, Chris Murphy developed CGETAX, an improved version of the original Independent CGE model. Compared to the original Independent CGE model, the major enhancements in CGETAX are as follows.

  • The number of industries distinguished was extended from 120 to 288. This adds to the robustness of modelling specific taxes such as those on different forms of alcohol, fuel, gambling and insurance. It makes it possible to model the distortions from applying different rates of tax to the different forms.
  • The model now distinguishes 24 different taxes, up from six taxes in the original model.
  • The model's database was updated, including by using the latest input-output tables from the ABS.
  • A 2-tier approach to modelling consumer demand was introduced. This takes into account that households often have more scope to substitute within 19 broad categories of spending than between those broad categories. This is important in robustly modelling the effects of changes to the scope of taxes on consumer spending such as the GST.
  • The model now allows for progressivity in the tax system. This progressivity adds to the excess burden of personal income tax.
  • The model now takes into account superannuation tax concessions, covering both contributions and earnings.
  • The model now allows for the distorting effect of personal income and superannuation taxes on household saving decisions.
  • A regional module was added to extend the modelling of economic impacts from the national level to 50 regions of Australia.

Independent Macro-econometric model (updated August 2015)

As a macro-econometric model, the Independent macro model is designed for economic forecasting and for analysis of fiscal, monetary and labour market poicies. It aims to balance economic principles and evidence from historical data in capturing the broad workings of the Australian economy. This balanced approach is in contrast to: (i) calibrated macro models that place most weight on economic theory and are mainly designed for policy analysis; and (ii) vector autoregressive (VAR) models that place most weight on the historical data and are mainly designed for short-term forecasting.

History

The Independent macro model is the latest in a series of macro-econometric models developed by Chris Murphy since 1988. It shares common design features with earlier models in the series as follows.

  • It uses quarterly data, and the parameters are estimated econometrically.
  • Stickiness in wages and prices means it is demand-driven or Keynesian over short time horizons.
  • Market clearing means it is supply-driven or neoclassical over long time horizons.
  • A representative business in each industry maximises profits in the long run.
  • Financial markets are forward-looking, with model-consistent expectations, while other markets are generally backward looking.

The original model was based on one industry (Murphy, 1988) but this was extended to 12 and then 18 industries in Murphy Model 2 (Powell and Murphy, 1997). The current macro model has six industries: Agriculture, Mining, Manufacturing, Government Services, Other Services and Housing Services. This broad industry detail is designed to provide a stronger base for macro policy analysis and forecasting.

The first edition of the Independent macro model was developed by us as a completely new model from 2012 to 2014. In 2015 the second edition of the macro model was developed by making economic growth semi-endogenous. The first edition of the model incorporated the following new design features.

  • There is a short-term interest rate rule based on the Reserve Bank's inflation targeting approach.
  • Land and mining resources are introduced as fixed factors of production.
  • The GFC is taken into account in modelling consumer and investment behaviour.
  • There is a new approach to modelling household consumption that uses a target for asset holdings based on labour income.
  • There is a new approach to modelling household consumption that uses a target for asset holdings based on labour income.

The second edition of the model retains all of the new design fetures of first edition, and also makes economic growth semi-endogenous. This follows similar work with the Quest III model at the European Commission (Varga and Veld, 2011). Drivers of economic growth are modelled as follows.

  • Demography: the population by age and gender is projected using the cohort-component method from assumptions about fertility, mortality and migration.
  • Education: the link from government eduction funding to education attainment (school, VET, university) is modelled.
  • Skill: the link from education attainment to occupational skill level (high, mid, low) is also modelled. Different skill levels have different productivity, labour fore participation and sustainable unemployment rates.
  • Migration: the model captures the different labour force experiences (in terms of labour force status and occupation) of five permanent and four temporary visa categories.
  • Infrastructure: the model allows for economies of scale in the provision of government infrastructure as a productive input for the business sector.
  • R and D: high-skilled labour is used to produce patents that raise productivity.

Download the Information Paper on the macro model

Download the information paper on "The Independent macro model" (updated August 2015).

Independent Short-term Forecasting model

The Independent Short-term Forecasting model is a structural Vector Autoregression (SVAR) model of Australia designed for a post-GFC world. Hence, it focusses on the linkages from the world financial markets to the world economy, and from the world to Australian financial markets and the Australian economy.

For Australia, this model provides forecasts of economic growth, unemployment and inflation, and their financial market drivers of the stock market, interest rates and the exchange rate.

For more information on our models, email my office.

Home Page