RDP 2018-07: The GFC Investment Tax Break 1. Introduction

The effect of business taxation on investment remains a relevant question, more than 50 years after Hall and Jorgensen (1967) kick started its modern study. As was the case then, policymakers still frequently use investment tax incentives as a countercyclical tool.[1] The potential for higher investment is a key argument in current debates around lowering company tax rates.[2] The relationship between tax and investment has implications for the potency of monetary policy, as monetary policy is often considered to act through a similar intertemporal substitution channel.[3]

Most recent work indicates that tax incentives raise investment. Zwick and Mahon (2017) find that accelerated depreciation available in the United States during two periods in the 2000s raised investment in eligible capital considerably. House and Shapiro (2008) found a similar result for the earlier of the two periods. The more interesting question now is probably: what mechanisms underlie the response of investment? Zwick and Mahon found that financially constrained companies react more strongly than other companies, and that companies only respond when incentives reduce tax payable in the current year. These findings suggest that tax incentives operate at least partly through mechanisms other than lowering the cost of capital.

We study the effect of an investment tax break that was in effect in Australia during the global financial crisis (GFC). An (almost) unique feature of Australia's tax system – corporate dividend imputation – means that our work provides direct evidence on the mechanisms underlying the investment response to tax incentives.[4] Under dividend imputation, tax paid by companies becomes a credit for personal income tax for resident shareholders. This means that the (small) companies that we study faced no fall in their cost of capital from the tax break, so any measured response is being driven by other, ‘non-standard’, mechanisms. Some features of the policy itself also make it worth studying. It was around five times more generous than the US policies that were in effect during the 2000s, was available for a much shorter period, and provided the same benefit no matter the effective life of the asset. This means it provides a test of intertemporal substitution: the primary mechanism through which tax incentives operate in a standard model.

The tax break we study was part of the Australian Government's stimulus response to the global financial crisis.[5] No previous empirical work has been done assessing the effectiveness of the investment tax break, so our work is also of direct interest to Australian policymakers. Differences in taxation systems (other than dividend imputation) and economic structure mean that the response of investment to tax incentives is likely to vary across countries.

What did the tax break involve? During the first half of 2009, all businesses received an extra tax deduction of at least 30 per cent on investment in equipment, plant and machinery. For business profits taxed at a rate of 30 per cent, this is equivalent to a fall of around 12 per cent in the after-tax cost of investment, when evaluated in a standard investment model. During the second half of 2009, smaller businesses – those with revenue below $2 million – received extra deductions of 50 per cent, generating even larger falls in the after-tax cost of investment, whereas larger businesses received a benefit of 10 per cent. Australian Taxation Office data indicate that businesses claimed at least $15 billion of extra deductions under the investment tax break (ATO 2016).

We use large business-level datasets and modern policy evaluation techniques to study the effect of the tax break. The design of the tax break means that the best way to study its effect is to look at differences between small and large businesses, so a dataset separating the investment of small and large businesses is needed. Data that distinguish between investment in equipment, plant and machinery (‘equipment’ investment) and investment in building and structures (‘building’ investment) are also useful, as the tax break applied only to the former. Our datasets, tax data on the population of businesses from the Business Longitudinal Analysis Data Environment (BLADE)[6] and business-level data from the Survey of New Capital Expenditure (both provided by the Australian Bureau of Statistics (ABS)), have these features.

We apply both difference-in-differences (DD) and regression discontinuity (RD) methods to the datasets. Our DD strategy compares the investment of small and large businesses within a given industry, and, after accounting for a number of other controls, takes this difference as the effect of the tax break. This strategy controls for differing industry conditions, a key improvement on past studies. Our RD strategy refines this by using the large discrete difference in the generosity of the tax break around the $2 million revenue threshold to identify its effect.

All of our statistical work indicates that businesses responded strongly to the tax break. Moreover, we find that – despite the existence of dividend imputation in Australia – companies increased their investment significantly in response to the tax break. This indicates that the effect of tax incentives on investment operates at least partly through non-standard mechanisms. One potential such mechanism is a relaxation of financial constraints, as the tax break frees up additional cash flows by allowing the company to delay the taxation of income. To the extent that this is the case, it may suggest that such polices are more effective during downturns when financial constraints are more binding. We also find no indications of intertemporal substitution, which provides more evidence for the importance of non-standard mechanisms. Specifically, we find that businesses eligible for the highest rates of the tax break did not invest less than other businesses in the years after its expiry.

Our estimate of the aggregate tax elasticity of investment is below that provided for the United States by Zwick and Mahon (2017), consistent with the different tax treatment of companies in Australia. When we instead focus on unincorporated businesses, which are subject to a similar taxation regime to US companies, the estimates are similar to those for the United States.

On the policy front, we provide partial and general equilibrium estimates of the macroeconomic importance of the tax break. These estimates suggest that both GDP growth and the cash rate would have been significantly lower in 2009 in the absence of the tax break.

While our work suggests that tax rates and breaks can affect real decisions for Australian companies, despite the existence of the dividend imputation system, it provides only limited guidance on the potential effects of policies other than the one we study. Other policies will differ in terms of their timing, permanence and targeting, all of which could influence their effectiveness.


The United States, the Netherlands, and South Korea used investment tax incentives during the global financial crisis (GFC). More recently, the 2017 US tax reform incorporates a significant investment tax incentive (full expensing). A similar policy has recently been suggested in Australia (Tingle and Coorey 2018). [1]

See The Economist (2017), and for an Australian perspective, Coorey (2018). [2]

See Atkin and La Cava (2017) for a discussion of the ‘saving and investment’, or intertemporal substitution, channel. [3]

Currently, only New Zealand, Chile and Mexico have dividend imputation systems comparable to that in Australia. Other OECD countries have had dividend imputation in the past. [4]

See Leigh (2012) and Li and Spencer (2016) for studies of the other components of the stimulus. The investment tax break is mentioned in a Senate inquiry into the effectiveness of the stimulus held in late 2009 (Senate 2009), but received almost no attention in a recent review of the stimulus (Makin 2016). [5]

See the Copyright and Disclaimer Notice. [6]