External indicators and the powers of a statistical office: Port statistics in Europe and Viet Nam

National statistical offices publish figures they did not produce, and what adopting an external indicator costs an office is rarely quantified. This paper makes it observable along three powers an office may lose: it defines the phenomenon, audits the value against an inspectable source, and governs revision; at issue is data governance, not procurement. Three public indicators of port activity (calls and inferred tonnage from PortWatch; vessel-density hours from EMODnet) cover 56 European ports over 36 months, tested against Eurostat criteria at 29, then transferred to Viet Nam, which publishes no port-level criterion. Within a port, cross-platform series correlate at 0.151 and same-platform series at 0.578 and 0.843; no construct is measured on both platforms, so platform and construct cannot be separated, but the three are not interchangeable. An identity decomposition places 93.8 per cent of vessel-hours variance in hours per call, not call counts, and substitution reverses three-year growth at 25 of 56 ports. In Viet Nam, tonnage tracks official total-cargo growth within 1.1 percentage points in two annual comparisons, supporting a provisional annual early indicator and nothing finer. The evidence supports free indicators as early estimates of an aggregate still officially compiled, not as replacements for discontinued collection.