There is a long‐standing debate in economics about whether governments should engage in Keynesian‐style countercyclical fiscal policy. During the recent global financial crisis, this debate has gained new momentum as many countries implemented fiscal stimulus packages. A prime reason for this was the fact that conventional monetary policy as an instrument for stabilization was no longer sufficient or feasible in an environment where interest rates had hit historically low levels in many countries. The objective of this paper is to re‐visit the effectiveness of such stabilization policies, in particular of a reduction in consumption taxes, using a difference‐indifference approach in combination with firm‐level data that exploits a temporary consumption tax cut in Turkey during the recent crisis.

There is a large body of empirical macroeconomic literature that addresses the question of whether fiscal shocks, in particular a debt‐financed increase of public spending or debt‐financed tax cuts, can have a positive impact on output over the short run. Contrary to the existing literature and in the absence of detailed and higher frequency household data, we use the change in firm sales using firm‐level data as an endogenous variable which, in aggregate, is likely to be closely related to the change in aggregate private demand. While our approach does not allow precisely estimating aggregate fiscal multipliers, it avoids the type of simultaneity problem that arises when using macro‐level data. The simultaneity problem arises because fiscal aggregates and GDP are interdependent with causation running in both directions; by contrast, the behavior and the performance of individual firms do not affect macroeconomic policies so that simultaneity does not arise.

Turkey has recently implemented a temporary consumption tax cut, namely in the value added tax (VAT) and the special consumption tax (SCT), at the peak of the financial crisis in 2009 as part of its fiscal package in response to the global economic crisis. In combination with the data we use, this policy change is particularly well‐suited for the purpose of our empirical research. On the one hand, the tax cuts were temporary and affected mainly durable and luxury goods (rather than necessity and non‐durable goods) so that it can be expected that consumers shift consumption forward in time. On the other hand, the consumption tax cuts were not universal and covered some but not all durable goods. Given that our firm‐level data covers the period during and after the tax change, we are able to implement a difference‐in‐difference approach where those firms primarily relying on goods covered by the tax cuts represent the treatment group and firms which primarily sell goods not covered by the tax cuts represent the control group. Our results indicate that the consumption tax cuts in Turkey indeed increased firm sales and private demand. While the data we use has limitations, our results nevertheless appear to be robust when we address a number of potential concerns about their reliability.