In recent times of fiscal stress during and after the Global Financial Crisis many countries announced tax amnesties. Spain, Greece, Portugal and Italy were among the European states struggling with revenue shortfalls and rising borrowing cost that announced tax amnesties. The obvious driver was the expectation to raise a non-trivial amount of revenue when it was most needed. Another potential motivation for calling a tax amnesty is a large stock of estimated undeclared or hidden income. In 2012, the Internal Revenue Service (IRS) of the United States initiated a tax amnesty that aimed at income from assets that were hidden offshore. In 2014, Australia followed suit and initiated its DO IT programme, which also aimed at income from offshore assets. Indonesia, a country that has traditionally faced problems with a large fraction of their citizens’ assets being hidden in Singapore, has – after announcements in 1964, 1984 and 2008 – this year announced its most enticing amnesty yet Assets that are brought back from offshore only attract a concessional tax rate of 5 percent or less.
In some cases it is quite clear why some tax evaders want to make use of a tax amnesty. If, for example, new information (such as a CD-ROM including names of offenders) has surfaced, offenders have an incentive to indicate past wrongdoing if an amnesty offers to do so free of fines. Sometimes amnesties provide additional incentives to be taken up, such as reduced tax rates or threats like increased fines and detection effort once the amnesty program has run out. Often tax amnesties only open the door for tax delinquents to undo their former decisions without changes in tax, fine and enforcement regimes. In this case, the traditional economic theory of tax evasion predicts that taxpayers stay put and do not take up an amnesty. Tax cheaters should have no reason to change their mind when an amnesty is announced because they should have carefully weighed the risks and benefits at the time when the cheating decision was originally made. In reality we observe that some people take up amnesties that they should not have taken up according to standard theory – the amnesty-take-up puzzle.
Bayer, Oberhofer and Winner (2015) present a unified equilibrium model that models the strategic interaction between a government that is conscious of its budget, the likely extent of tax evasion, and a population of taxpayers, who form their beliefs about the likelihood of a tax amnesty being introduced in the future. Our taxpayers are taking up amnesties to a certain extent even if the conditions do not change because they are short-sighted in the sense that they discount future fines when they decide to evade. Then, once they are getting closer to the point in time at which detection and punishment could take place, they get scared and prefer to repay some taxes to reduce their punishment if caught.
In addition to finding a way to resolve the amnesty-take-up puzzle, our model makes some interesting predictions:
- If fiscal requirements increase, then tax amnesties become more likely and taxpayers anticipate this and react by evading more aggressively.
- Governments that have less to lose from calling an amnesty (such as no re-election concerns) are more likely to call an amnesty, which again leads to more evasion as taxpayers anticipate this. This implies that governments that are perceived weaker cause higher levels of evasion, even if they do not call an amnesty in the end.
- Increased tax rates do have an ambiguous effect on the probability of amnesties taking place because while higher taxes raise more revenue and reduce budgetary pressure, they also lead to higher tax evasion, which increases the recoverable revenue from an amnesty.
In general, the model predicts a self-fulfilling prophecy to some degree. If anything exogenous changes, such that the taxpayers’ believed probability of a future tax amnesty increases, then they evade more aggressively, which in turn nudges the government into the direction of a tax amnesty.
The role of self-fulfilling beliefs is so strong that under certain circumstances there might be multiple stable situations, which economists call equilibria. The equilibria differ considerably in the level of evasion and likelihood of tax amnesties from each other. For example, the same country could have an equilibrium with hardly any evasion and a negligible likelihood of a tax amnesty being introduced, an equilibrium with considerable evasion and a moderate likelihood of an amnesty but also an equilibrium with widespread evasion and frequently reoccurring tax amnesties. This is interesting for two reasons. First, it can explain why very similar countries have vastly different evasion and amnesty histories without reverting to hard to grasp concepts such a culture. Second, it shows that governments should be careful to avoid actions that could lead to a jump in amnesty beliefs, as that could catapult a country from a desirable low-evasion to an undesirable high-evasion equilibrium, which requires frequent amnesties.
In order to test the plausibility of our model we use a dataset that records all state tax amnesties in the United States (from 1982 when Arizona was the first state to initiate an amnesty). This dataset is ideal for our purpose, since the legislative framework across states is comparable, while there is considerable variation in the frequency of amnesties. We add macroeconomic, fiscal, labour market and political economy data and run a variety of panel-data estimations. The analysis confirms the main predictions made by our model. Particularly interesting is that amnesties that take place in the year before in other states have a strong positive impact on the likelihood of a tax amnesty. We interpret this as confirmation of the importance of beliefs because recent amnesties in other states seem to increase the believed likelihood of an amnesty announcement in a taxpayer’s home state.
Our central message to policy makers is that of caution. Due to the self-fulfilling and contagious properties of beliefs about the likelihood of an amnesty it is very important to manage beliefs. A snap announcement of an amnesty for a quick grab of revenue might have to be paid for with subsequent increases of evasion and the resulting increased pressures to initiate amnesties in the future.
This post is based on the paper ‘Bayer, R.-C., Oberhofer, H., Winner, H. (2015): The Occurrence of Tax Amnesties: Theory and Evidence, Journal of Public Economics 125, 70-82.’