Concentration in Local Tax Portfolios: A Sign Reversal and What It Reveals About Unit Selection

Abstract

Whether larger governments hold more diversified revenue is an empirical question with an established expected answer. This article computes concentration for every Indonesian local government from the national tax realisation file for fiscal year 2024, and finds that the answer depends entirely on which units are included. Collapsing 97 reported tax objects into eight groups and computing a Herfindahl-Hirschman index across them, the correlation between concentration and the logarithm of realised tax revenue is positive at 0.244 across all 536 reporting governments and negative at minus 0.326 across the 498 districts alone. The reversal is not a statistical curiosity. Provincial governments hold a tax base consisting almost entirely of vehicle and fuel taxation, which falls into a single group, so 35 of 38 provinces record perfect concentration by construction while also recording the largest revenues in the file. Including them manufactures a positive slope that inverts the relationship holding among comparable units. The article reports the district distribution, in which mean concentration is 0.330 against a theoretical floor of 0.125 and no government approaches even eight-way balance, identifies five districts whose recorded concentration reflects reporting failure rather than revenue structure, and argues that concentration measures computed across constitutionally unlike tiers are uninterpretable.