RDP 2025-01: Are Investment Tax Breaks Effective? Australian Evidence 3. Literature

Hall and Jorgenson (1967) first analysed the impact of tax incentives on business investment and the intervening (nearly) sixty years have seen many economies implement large-scale business tax reforms with an aim of stimulating investment, both in normal times and as a countercyclical tool. Questions around the efficacy and cost effectiveness of such measures remain a live part of the policy debate.

Investment tax incentives are generally thought to affect investment through two channels. The ‘traditional’ user cost channel, and the ‘non-traditional’ financing frictions channel.

The user cost channel postulates that firms will invest as long as the marginal benefit of that investment is higher than the cost. Increasing tax deductions for investment expenditure reduces the costs, often referred to as the user cost of capital (UC), by increasing the present value of depreciation allowances that firms can use to lower their tax. This can be seen using a measure of the the UC derived by Hall and Jorgenson (1967):

where P is the real price of investment goods, M is the weighted average cost of funds (debt and equity), and $\tau$ is the business tax rate. Z is the ratio of the present value of future depreciation allowances, relative to the initial purchase price of the asset. These allowances permit businesses to deduct the cost of their capital investment from their taxable income over several years, with the period approximately reflecting the economic life of the asset. By increasing or bringing forward the timing of these allowances, investment incentives can increase Z and so lower the UC.

A permanent policy change that lowers the UC leads to a burst of investment as firms adjust to their new higher desired capital stock.^[3] Investment may remain higher going forward as firms invest to maintain the higher capital stock. Temporary policies may have a larger and more immediate effect as firms bring forward investment to the current period when it is cheaper (Abel 1982). However, empirically this bring-forward effect may be hard to detect, particularly in the case of long-lived assets, as the future ‘hole’ in investment could be spread over many years (House and Shapiro 2008).

Imputation systems such as exist in Australia change these costs and benefits. Under Australia's imputation system, company tax operates as a withholding tax. When dividends are issued to individuals, those individuals receive a credit for the company tax paid. Individuals are taxed annually on all of their income and can claim the credits for tax already paid. In particular, as they provide the end owner of the company with a tax offset for the tax paid by the company, they can ‘wash out’ the benefits of the depreciation allowances. Dividend imputation has been removed in many countries, including the United Kingdom and New Zealand, as corporate tax rates have been lowered. They remain an important feature of the Australian system.

Under the old view of corporate finance where businesses fund new investment through raising equity, the UC can be augmented for imputation as in Officer (1994):

Imputation is captured by the term $\gamma \in \left[0,1\right]$, which measures the value of a dollar of tax paid at the company level to shareholders. This value reflects the fact that, under dividend imputation, tax paid at the company level is a credit for personal income tax for resident shareholders. If all shareholders are domestic (and in the presence of full refundability of imputation credits), as may be the case for small firms, the value of $\gamma$ is 1 – tax paid at the company level lowers shareholders' taxes dollar-for-dollar. In this case increases in Z do not affect the cost of capital nor do they affect investment. If some shareholders are non-residents, as may be the case for larger firms, $\gamma$ may be below 1. In contrast, the new view posits that the marginal source of funds for firms is retained earnings. In this case the imputation system has little influence on the UC.

These two views of corporate finance lead to two very different conclusions regarding whether firms that are subject to imputation will respond differently to the ITB than those which are not. Under the old view, firms not subject to imputation (i.e. unincorporated businesses) should respond more strongly. Under the new view, both firm types should respond similarly.^[4] As such, ITB can be used to assess which view of corporate finance is most consistent with the empirical evidence.

Numerous studies have examined similar ITB overseas. Empirical studies on the United States (House and Shapiro 2008; Zwick and Mahon 2017) and the United Kingdom (Maffini, Xing and Devereux 2019) have found strong investment responses to tax incentives (and associated decreases in the user cost of capital).

House and Shapiro (2008) test if businesses increased their expenditure on longer-lived assets, and whether there was a change in aggregate investment behaviour in response to a change in policy. To look at asset lives, they use data from the Bureau of Economic Analysis to construct a quarterly panel of investment quantities and prices by asset type. They match 36 asset types to the Internal Revenue Service (IRS) depreciation schedule to track eligible and ineligible investment as well as short/long asset lives between 1959 and 2005. A structural model is then estimated to provide a baseline for investment activity, GDP and employment without the bonus investment. House and Shapiro then compare forecast errors across asset types. A strong response to the tax break was found in terms of the investment composition, with an investment supply elasticity of between 10 and 20 per cent. However, only modest increases in aggregate output and employment were found, as the policy narrowly benefited a small subset of investment goods.

Zwick and Mahon (2017) study the same bonus depreciation policy as House and Shapiro (2008) but use detailed IRS corporate tax return data that include information on investment in equipment and structures. In 2008, the sample represented about 1.8 per cent of the total population of 6.4 million C and S corporation returns. In each non-bonus year, the authors compute the share of eligible investment a firm reports in each asset life and property category. They then compute the present discounted value of one dollar of deductions for eligible investment, and combine this with the eligible investment shares to construct an industry-average deduction value. As the policy raises the net present value of longer-lived investment by more, they use this variation in asset lives in a DD regression to identify the policy effect. They find evidence that the policy affected investment. They also find evidence that the policy had a larger effect on the investment of firms that appeared to be financially constrained, indicating that the policy worked at least partly by loosening constraints.

Maffini et al (2019) study UK corporate tax returns to analyse the impact of a 2004 change in policy. It allowed more firms to access small business depreciation concessions, which took the form of a 40 per cent depreciation rate compared to the 25 per cent available for larger companies. Similar to what we do, they exploit this exogenous change in the qualifying thresholds and compare the response for companies newly qualified for the greater depreciation deductions to those that remain ineligible. Consistent with the US studies, Maffini et al find a substantial positive effect of the more generous depreciation allowances on firms' investment. Relative to the control group, the depreciation allowances raised the investment rate in eligible assets of newly-qualified companies by between 2.1 and 2.5 percentage points within three years of the change.

Zhang, Chen and He (2018) and Liu and Mao (2019) analyse the effect of a permanent 17 per cent tax credit for fixed investment for six industries in northwest China between 2004 and 2009. Using geographical restrictions to identify the policy effect, Liu and Mao (2019) find that the tax credit raised fixed investment of eligible firms by 28 per cent on average during 2004–2007, relative to 2001–2003, corresponding to a user cost elasticity of 1.84. The strong response by firms found by Liu and Mao (2019) is consistent with Zhang et al (2018), who found that, on average, the reform raised investment of the treated firms relative to the control firms by 38.4 per cent.

Australian evidence is sparse, partly due to data availability issues in the past. Rodgers and Hambur (2018) analyse the effect of tax breaks for machinery and equipment during the GFC (this is the first of the seven reforms that we consider in this paper). They use RDD and DD methods around the turnover threshold to compare the investment of small and large businesses using business tax microdata and unpublished survey data from the ABS.

Consistent with the difference in corporate taxation regimes (i.e. classical taxation in the United States and dividend imputation in Australia) and the old view of corporate finance, Rodgers and Hambur (2018) find a lower responsiveness of investment to the ITB compared to Zwick and Mahon (2017), as measured by the elasticity of investment. However, they find a much larger elasticity, more in line with the US studies, when focusing on unincorporated businesses who face a classical taxation regime similar to US corporations. They find some evidence that incorporated companies also responded. Given that the policy was targeted at smaller companies, who have very few non-resident shareholders, they argue that under Australia's dividend imputation system, $\gamma$ equals one for these companies. They conclude that their findings suggest that the policies worked partly by relaxing financing constraints.^[5] Our results presented below are consistent with this previous research on Australia which only examined one of the seven policies that we consider.

Footnotes

Given tax incentives work through this user cost channel, these policies have often been used to assess the effect of changes in the UC on investment more generally. They are seen as providing more plausibly exogenous variation than changes in financing costs (Cummins, Hassett and Hubbard 1994) and are less prone to measurement error. [3]

For a more detailed discussion see Sobeck, Breunig and Evans (2022). [4]

The broader literature on the value of $\gamma$ is mixed. Many large Australian companies raise equity from international investors, who derive no benefit from dividend imputation. As such, $\gamma$ is unlikely to be one for these businesses. Applied evidence based on samples of listed companies finds a range of values for $\gamma$, see Swan (2019) and Sobeck et al (2022). At the same time, studies overseas have found evidence in support of the new view (i.e. $\gamma$ is 0), with Yagan (2015) finding evidence that cuts to dividend taxes in the United States in 2003 had little effect on investment. [5]