Four exclusion categories, each with a specific theoretical justification: (1) Resource-dependent economies — grow via commodity windfalls unrelated to government size, creating noise without signal. (2) Externally-funded states — have aid-distorted budgets that make the spending/GDP ratio misleading. (3) Conflict/fragile states — have suppressed growth from instability that is independent of spending level. (4) GDP-distorted economies — have artificially inflated or deflated GDP figures. With these filters, the sample drops from ~151 to ~113 countries; R² rises from ~0.24 to ~0.42, confirming the excluded groups add noise rather than signal.