Stochastic semi-nonparametric frontier approach for tax administration efficiency measure: Evidence from a cross-country study

Under globalisation, countries all over the world have been impacted differently by both their macroeconomic conditions and other random factors particular to them. As such, governments and policymakers require a comparison of tax systems across different countries and, consequently, a study on the performance of Tax administration (TA) at a cross-country level would be a necessary reference for governments when designing tax policy. This paper seeks to measure the performance of TA across 44 countries, while considering the presence of contextual variables, using the recently developed and advanced frontier estimators, such as the semi-nonparametric StoNED (Stochastic Nonparametric Envelopment of Data) approach by Johnson and Kuosmanen (2011, 2012) and the conditional order-m (Daraio and Simar, 2005, 2007) approach, for two periods between 2008–2011 and 2012–2015. The results show that Tax agencies in these countries could have increased tax revenue, on average, by about 58.7% and 34.2% for the two periods, respectively. Equivalently, $7,737 and $4,667 PPP (purchasing power parity) per capita of tax revenue could have been increased for the two periods, respectively. It is also suggested, in general, the latter period (2012–2015) shows a higher level of efficiency than the former period (2008–2011), justified by both estimators.