Engström, Per Nordblom, Katarina Stefánsson, Arnaldur
Year of Publication:
Working Paper 2018:8
We study what determines taxpayers' deduction behavior when filing tax returns. Preliminary deficits might be viewed as losses assuming zero preliminary balance as reference point. Swedish taxpayers may escape these losses by claiming deductions after receiving information about the preliminary balance. Furthermore, the Swedish income tax system has a substantial kink (20 percentage points) where the central government tax applies. Taxpayers slightly above the governmental tax kink have substantially higher (standard economic) incentives to claim deductions than taxpayers slightly below the kink. Using a regression kink and discontinuity approach with individual fixed effects, we study a panel of 4.1 million Swedish taxpayers in 1999 to 2006. We find strong causal effects of preliminary deficits on the probability of claiming deductions. The initial empirical evidence for a kink in deduction probability at the central government threshold, anticipated by standard economic theory, is weaker but significant. However, a more detailed analysis reveals that the kink at the tax threshold is not likely due to the tax incentives per se. When controlling for the preliminary tax deficit, the kink at the tax threshold disappears. Taxpayers just above the tax kink are namely more likely to run a preliminary tax deficit than those just below it. Hence, the most plausible explanation also for the kink at the tax threshold is therefore loss aversion and not standard economic incentives. The Swedish taxpayers are thus "misbehaving", in a Thaler (2015) sense, on two separate margins: they are highly loss averse but surprisingly inattentive to standard monetary incentives.
tax compliance loss aversion prospect theory quasi-experiment regression kink regression discontinuity