Abstract
The relationship between corruption and tourism has been sporadically examined over the years. According to the existing theory, there is an inverted U relationship which implies that tourism demand initially increases as corruption increases (greasing the wheels) and after a certain threshold level of corruption, tourism demand decreases (sanding the wheels). Empirical studies so far concentrated on capturing the nonlinear relationship, by applying a simple linear model and by including corruption as a quadratic term. In the current paper, the authors revisit the "greasing and sanding the wheels" hypothesis by applying an advanced econometric technique, the threshold regression model, which deals with a key element of model uncertainty, namely parameter heterogeneity. In particular, using a sample of 83 countries from 2001 to 2018, the authors firstly examine if there is a nonlinear relationship between corruption and tourism, and then, they estimate the threshold value of corruption. According to the results, the null hypothesis of a linear model against the alternative of a threshold model with two regimes is strongly rejected. Furthermore, while the effect of corruption on tourism is positive in the low corruption regime and negative in the high corruption regime, a heterogeneous relationship is also found between other politico-socio-economic variables and tourism demand in the low and high corruption regimes. SUPPLEMENTARY INFORMATION: The online version contains supplementary material available at 10.1007/s00181-021-02193-2.