Abstract : Taxation is one of the main components of a country’s fiscal space. Its internal origin and the accountability it creates between rulers and populations make it a key element in financing public expenditure. Tax capacity differs between countries and depends on structural factors. A number of empirical studies attempted to determine countries’ overall tax potential and tax effort (Lotz and Morss, 1967; Stotsky and Wolde Mariam, 1997; Fenochietto and Pessino, 2013). However, the methodologies used tend to underestimate or overestimate countries’ tax potential and thereby their tax effort. The purpose of this study is to better assess countries’ non-resource tax potential and VAT’s tax potential independently using a more appropriate method. It is in line with the study of Brun et al. (2014) and rests on a large sample of developing countries over the period 1980/2014. We first employ the previous models and discuss about their shortcomings, after we use the stochastic frontier model of Kumbhakar, Lien and Hardaker (2014). This model allows to disentangle the overall tax effort into a persistent tax effort due to policy economy decisions and a time-varying tax effort relating to tax administration efficiency. The results are more realistic. Low income countries have higher tax effort along the period even if their tax effort decline at the end of period on the opposite of resource depending countries. In fact, the latter characterized by lower tax effort compared to non-resource countries improved consequently the efficiency of their system since 2010. The results also suggest that inefficiency in taxation depends more on policy decisions than on tax administration performance.