Behavioural responses to estate taxation: evidence from Taiwan
Following the increase in government debt due to COVID-19, many policymakers proposed a hike in estate taxes to support the public finances (IMF, 2021; OECD, 2021). The proposals have drawn attention to the possible behavioural responses to a rise in estate taxation. Specifically, would an increase in the tax produce higher tax revenues for the government or will this incentivise individuals to change their behaviour by reducing savings or increasing tax avoidance and evasion? The answers to these questions are crucial for establishing an optimal tax design; however, empirical evidence is limited. In my doctoral research, along with Tzu-Ting Yang, I analyse Taiwanese data to provide evidence of behavioural responses to estate taxation using an earlier tax-cutting reform. This will potentially have implications beyond Taiwan for various forms of wealth taxation.
Using the rich administrative data on individual wealth and estates at death in Taiwan, we aim to quantify the impact on individual behaviour of the reform in estate taxation introduced in January 2009. This reform increased the exemption threshold by 150% and lowered the marginal tax rate by 80%. Past literature has encountered an empirical challenge in estimating the causal effect of a tax reform. A standard technique in economics is to compare behaviour before and after a reform. However, because people only die once, the estate at death cannot be tracked for the same person before and after a reform. It is, therefore, unclear whose behaviour is affected when a reform occurs. This is addressed by applying machine learning tools to identify who is exposed to the reform. We train an algorithm to predict individuals’ estate at death using their pre-reform wealth information. By doing so, we can uncover one’s level of exposure to the policy change and establish a ‘difference-in-difference’ design.
We have found that responses in net estate, which is defined as reported estate minus deduction, are substantial and immediate. Shown in the figure below, right after the tax rate is lowered, net estate jumps by 8% in the first 180-day period. The estimated elasticity of net estate with respect to net-of-tax-rate is 0.97: a 1% increase in the net-of-tax rate raises net estate by 0.97%. Additionally, there is differential behaviour between those who face a repeal and those who face a decrease in taxes. The increase of the exemption threshold from the reform creates two regions, a repeal region where taxes are totally removed and a decrease region where tax rates are decreased but not removed. It is estimated that the elasticity of the repeal region is four-times higher than that of the decrease region. The mechanism behind this response is analysed by breaking down the estate item-by-item. The results suggest individuals misreport unmonitored and liquid assets and adjustable deductions.
This figure presents the evolution of net estate before and after the reform. It plots the difference-in-difference estimates and their 95% confidence intervals. The event is on 23rd January 2009. A period is 180 days. The baseline period is 0, which is 180 days before the reform.
Indications are emerging from this research that behavioural responses to the reform are driven by avoidance and there is a heterogeneity in sheltering costs across estate items. An impact on saving decisions is, however, less apparent. The dominant role of avoidance suggests the need for better enforcement, for example, extending the coverage of third party-reported and offshore items. Without a stronger enforcement design, it is not necessarily the case that an increase in estate tax rate would raise tax revenues as expected. The next stage of this research will use the empirical findings to better understand the heterogeneous responses and welfare effects of estate taxation to deliver an optimal exemption threshold and tax rate for estate taxation.
