Tax Collection Efficiency in Selected OECD Countries: Evidence from Advanced DEA Extensions

Data Envelopment Analysis (DEA) modeling has been widely used in business analytics, but its application in public finance analysis is comparatively not as common. This study aims to analyze efficiency measurement in 37 OECD tax administrations from 2018 to 2022. Our methodology is based on step-by-step extensions of DEA. We will begin with a standard DEA model. The unique nature of public finance requires the use of undesirable output modeling, which will be the next step in expanding our model. To capture other dimensions of analysis, we will utilize “non-discretionary factors modeling”. Additionally, we will use the congestion modeling to ensure that excessive inputs are not being used. As tax administrations transition to electronic systems, we will incorporate “categorical variable(s) modeling” to address this aspect of revolution in our analysis. we will use stochastic DEA modeling to compare our results with those obtained under deterministic assumption. Finally, “super-efficiency modeling” will be added to determine just one efficient decision-making unit. Based on our findings from non-discretionary factors modeling, the results have improved the result for countries such as the USA, UK, Turkey, and Costa Rica. Additionally, other versions of extensions show that some countries like Finland, France, Denmark, Norway, Austria, Belgium, and Germany are less sensitive to changing DEA assumptions and remain efficient despite modeling changes.