Buying Books, Buying Diversity? Public Procurement Theory and the Brazilian State as Book Buyer

Governments are among the largest single buyers of books, yet public procurement research rarely asks what state purchasing does to the diversity of the market it buys from. This article examines federal book procurement in Brazil using item-level electronic invoice (NF-e) data published by the Office of the Comptroller General (CGU). The data cover 60 months between October 2021 and September 2026 and R$21.9 billion in purchases, 96.7% of which were made by a single buyer, the National Fund for Education Development (FNDE). Drawing on Stirling's variety–balance–disparity framework, we build five indicators of bibliodiversity: supplier concentration (Herfindahl–Hirschman index), the share of the four and eight largest suppliers, supplier evenness, title evenness, and territorial inequality among suppliers. The results describe a market with high variety but low balance. A median of 280 suppliers invoice the federal government each month, yet the four largest take a median 68% of value, and two publishers, FTD and Moderna, account for 45% of the whole series. Concentration is episodic rather than trending: 13 of 60 months exceed the HHI threshold of 2,500, with no significant monotonic trend. Territorial inequality is extreme and stable (median Gini 0.93), with São Paulo-based emitters taking 93% of value. Title diversity falls when supplier concentration rises (Spearman ρ = −0.61) and has declined sharply since mid-2025. We read these patterns against transaction cost, principal–agent, competition, strategic-procurement and institutional theories. We argue that, in markets for unique cultural goods purchased without price competition, the locus of competition moves from the tender to the pre-selection stage, so programme design rather than tendering rules becomes the main lever for diversity. We derive five propositions and discuss the data-quality limits of invoice-based procurement analytics.