The impact of GST reform on Australia’s state and territory economies

ABSTRACT

We use a multi-regional general equilibrium model and a multi-regional microsimulation model to explore the economic consequences of a Goods and Services Tax (GST) rate rise and broadening of the GST base for Australia’s eight states and territories. We identify the states and territories adversely affected by GST change, and explain the regional structural factors responsible for differences in regional impacts. These findings have policy relevance in the context of Australia’s GST framework, which requires unanimous state and territory support for GST change.

KEYWORDS:

• Goods and services tax
• value added tax
• dynamic regional cge
• microsimulation

JEL CLASSIFICATION:

• C68
• H25
• H73
• R13
• D57

Disclosure of potential conflicts of interest

No potential conflict of interest was reported by the author(s).

Notes

¹ To put this in context, the next three largest revenue lines for Australian states and territories are payroll tax (7.1% of total revenue), property duties (6.8%) and motor vehicle taxes (3.1%), which collectively generate less annual revenue than the GST for state and territory governments. See ABS Cat. Nos. 5506.0 and 5512.0.

² See Freebairn (2011), Daley et al. (2015). The latter note that the low GST rate and exemptions result in an under-reliance on GST relative to global peers. For example, the share of revenue raised by GST in Australia (almost 13%) is well below the OECD average (20%). Similarly, Australia’s GST base is narrow by international standards, applying to 47% of consumption (compared with an OECD average of 55% and 96% for New Zealand).

³ These papers span the full period 1987–2021. References available from authors on request.

⁴ See Adams, Dixon, and Horridge (2015) for details of the VURM model without GST detail. The GST detail is described in Giesecke and Tran (2018) and expanded upon herein.

⁵ For many commodities, EEX is not quite zero, because businesses that are unregistered for GST purposes cannot reclaim VAT paid on their inputs.

⁶ Low but pervasive levels of non-registration for GST purposes mean that, for most industries, REF values are slightly below 1. Our model recognizes two causes of non-registration: (1) ‘legal’ non-registration (NRL), reflecting entities that choose not to register for GST purposes because their scale is below the mandatory registration threshold; (2) ‘informal’ non-registration, reflecting businesses who choose not to register for GST when they are legally required to do so. In our model, NRL and NRI rates differ across industries. As industry compositions differ between regions, our model thus captures the effects of differences in non-registration via industry composition effects. However, values for NRL and NRI are low, and differences between regions via industry composition effects are small. Thus, NRL and NRI exert no material effect on the differential state impacts discussed in Section 3.

⁷ This is the same as the value used in other recent studies including Bento and Jacobsen (2007) and Khanna and Nelson (2008) (who use +0.15 for the U.S.), and Cao et al. (2015) (on the basis of Dandie and Mercante 2007) and Fraser and Waschik (2013) (who use +0.15 for Australia). It is in line with, but a little higher than, the indicative value (+0.10) suggested by Boeters and Savard (2013) citing Sørensen (1999).

⁸ Boeters and Savard (2013) argue that the labour/leisure ratio should be calibrated to produce an empirically plausible income elasticity of labour supply. They suggest −0.1 as a plausible value. A similar value for Australia can be justified by inspection of the income elasticities reported in Dandie and Mercante (2007). They report income elasticities estimated by Breunig, Cobb‐Clark, and Gong (2005) of −0.01 to 0.00 for married men, and −0.01 for married women. Breunig et al. do not provide estimates for other labour market groups. Again, reported by Dandie and Mercante, income elasticities for lone parents of −0.52 (Bingley, Symons, and Walker 1992) and −0.24 (Jenkins 1992) for the UK. A range of Australian estimates by Murray (1997) for lone parents are also reported by Dandie and Mercante (see their Table A2). These range between −1.38 and −0.14. On the basis of this range of estimates, a plausible income elasticity of labour supply by lone parents might be −0.5. On the basis of Breunig, Cobb‐Clark, and Gong (2005) we might reasonably assume that an estimate for couple families is −0.01. ABS (2019) reports a distribution of family types comprising couple families (83%) and one parent families and other families (17%). This points to an average income elasticity of labour supply of approximately −0.093 (=0.83*-0.01 + 0.17*-0.5), very close to the −0.1 value that Boeters and Savard (2013) suggest is plausible.

⁹ See Table 1, row 2, which reports government expenditure shares in gross state product. These are highest for the ACT and the NT.