Measuring the Cost of State: A Controlled Benchmark of Seven React State Management Libraries in Table-Based Dashboards

We present a controlled, reproducible benchmark comparing seven React state management approaches — Redux Toolkit, Zustand, Jotai, React Context, Valtio, MobX, and an idiomatic Redux variant — across thirteen interaction scenarios on React 19.1 production builds. The primary finding: selector-based libraries (Redux, Zustand, Jotai, Valtio, MobX), combined with React.memo, produce zero surplus re-renders when unrelated state changes; React Context produces one re-render per update regardless of which slice changed (50 surplus renders; ~1,632 ms scenario-level overhead). Three multi-panel scenarios confirm that isolation generalizes beyond a single component tree. Bundle overhead: Redux +10 KB, Jotai +5 KB, Zustand +2 KB; Context adds zero library overhead. A practitioner-critical finding: the React Compiler silently over-memoizes proxy-based libraries (Valtio, MobX), suppressing required re-renders — an incompatibility invisible to production monitoring. Cross-hardware replication on four platforms (Apple M5 Max, M3 Max, Intel Core i9, Apple M1 CI) and three JS engines (V8, SpiderMonkey, JavaScriptCore) confirms hardware- and engine-independence of all render-count results. The benchmark artifact is fully reproducible with a single command (pnpm benchmark) in under two minutes. Source code: https://github.com/kvashaihor/state-management-benchmark