Housing prices have risen faster than incomes across many advanced economies, while homeownership rates have fallen, renewing attention to policies that influence housing demand. The international literature has extensively examined the tax treatment of mortgage interest in the context of owner-occupied housing, often justifying the deduction as a means of promoting homeownership. However, far less empirical attention has been paid to analogous deductions for rental property. In Australia, landlords may deduct mortgage interest and other rental expenses against unrelated income such as wages. In a recent working paper, I provide new evidence that the ability to deduct rental losses substantially raises demand for rental property by individual investors.
From a theoretical perspective, allowing full deductibility of rental losses—including against labour income—can be justified on the basis of tax neutrality. Under a comprehensive income tax system, all income (including capital gains) should be taxed on the same basis, with deductions permitted for expenses incurred in generating taxable income, regardless of an investor’s income mix. Under this framework, investing in rental property is conceptually similar to investing in a business or financial asset, where losses in one activity can offset income from another.
In Australia, capital gains are taxed at the taxpayer’s marginal tax rate, but for assets held longer than 12 months only half the gain is taxed. In contrast, rental losses are fully deductible against labour income (negative gearing). Investors can therefore deduct rental losses—including interest expense—at high marginal income tax rates and realise lower-taxed capital gains. This mismatch between the timing and rate of deductions versus gains breaks neutrality and may encourage investment for tax reasons rather than expected economic return. In practice, rental losses are overwhelmingly driven by use of debt.
While the ability to deduct rental losses against ordinary income in Australia is longstanding, tax reforms have changed tax benefits of rental property ownership for different groups of taxpayers over time. I estimate individual taxpayer demand for rental property ownership in Australia using changes in marginal income tax rates caused by reforms in the mid-2000s that lowered marginal income tax rates for upper-income taxpayers. Data are from the Australian Taxation Office Longitudinal Information Files (A-Life).
Figure 1 provides graphical evidence, focusing on upper-income taxpayers who were differentially affected by an increase in the top tax bracket threshold from $95,000 to $180,000 between the 2005/06 and 2008/09 financial years. I follow two income ranges: incomes greater than $180,000, which remained subject to the top marginal tax rate, and incomes in the range $95,000-$150,000, which moved from the top to the second-highest bracket. The marginal tax rate on incomes in the range $95,000-$150,000 declined by 5 percentage points relative to that on incomes above $180,000 in the 2006/07 financial year. The share of taxpayers reporting a negatively geared rental property declined in the $95,000-$150,000 range relative to the $180,000 and above range. In contrast, the fraction of taxpayers with a positively geared rental property moved similarly for both income groups. This provides evidence that high marginal tax rates increase the incentive to hold negatively geared rental property.
This graphical evidence is confirmed with formal econometric methods using the full set of changes in marginal tax rates over the period 2001/02 to 2010/11. I estimate that a 10 percentage point increase in the net-of-tax rate (i.e., one minus the marginal tax rate) reduces the probability that a taxpayer reports a rental property by 2.7 percentage points. This almost entirely reflects fewer negatively geared rental properties being reported. Accordingly, there is a drop in interest deductions claimed. My results are robust to a battery of sensitivity tests.
I show in the paper that my empirical findings are consistent with a relatively standard model. In the model, an investor chooses between potentially leveraged investment in real estate and financial assets, which cannot be leveraged. Modeling the portfolio choice problem of an investor is important because the changes in marginal tax rates I use for identification affect returns to property and non-property investment. The model predicts that a decrease in the marginal tax rate lowers demand for leveraged rental investment but has a negligible effect on unleveraged rental property investment, as estimated empirically. A change in the marginal tax rate has little impact on unleveraged property investment because it affects returns on other assets, like equities, in much the same way. In contrast, when investors use debt, the tax system effectively subsidises their property investment.
Proponents of tax concessions for rental property argue that their elimination would lead to an undersupply of private rental property. This echoes the argument that positive externalities to homeownership justify the mortgage interest deduction for owner-occupiers. Opponents argue that greater investor housing demand mainly pushes prices up because housing supply responds weakly to changes in price, which in turn makes it harder for owner-occupiers to purchase. Adjudicating between these views is beyond the scope of my paper. However, my findings provide a key new piece of evidence to the debate. The high sensitivity of demand for rental property to individual marginal income tax rates is a necessary condition for investor tax concessions to either address undersupply in the rental market or, conversely, create inefficient over-investment.
Recent Comments