A €66 Billion Discrepancy and a Call to Rewire Global Economic Statistics for a Fragmenting World

The euro area and the United States cannot agree on the size, or even the direction, of their bilateral current account balance. According to European Central Bank data, euro area imports of goods, services and income from the United States exceeded exports by €57 billion in 2025. The U.S. Bureau of Economic Analysis, using its own methodology, reports the opposite: euro area exports to the U.S. were €9 billion higher than imports. The combined discrepancy reaches roughly €66 billion, a figure large enough to distort the policy debate on tariffs, competitiveness, and capital flows between the world’s two largest economic blocs.

This asymmetry was the centrepiece of a keynote address delivered by Bundesbank Executive Board member Fritzi Köhler-Geib at the External Statistics Conference in Kraków on 28 May. Speaking to an audience of central bank statisticians and international economists, Köhler-Geib used the transatlantic data gap to illustrate a broader problem: geoeconomic fragmentation is making accurate, comparable external statistics more important, and simultaneously more difficult to produce.

“If we want to understand all aspects of external trade properly, we need more consistent data,” she said. “Large and persistent bilateral asymmetries harm the analytic potential of our data for policymakers and researchers.”

The drivers of asymmetry are well known but hard to eliminate. Compilers in different jurisdictions use different source data and estimation methods. Multinational enterprise structures are complex, and confidentiality constraints prevent statistical agencies from comparing microdata. Divergent interpretations of international standards add a further layer of inconsistency.

Efforts to close the gaps are underway. European statistical coordination networks have established asymmetry resolution mechanisms, though Köhler-Geib noted these cannot easily be replicated elsewhere because they depend on the EU’s legal framework for official statistics. The OECD and the IMF now host publicly accessible databases that map bilateral asymmetries by country pair. A BOPCOM task team, mandated in autumn 2024, is working on recommendations.

But Köhler-Geib argued that these initiatives are insufficient. Her concrete proposal: include a dedicated recommendation on asymmetry reduction — with a monitoring system based on jurisdictions’ self-assessments — in the next iteration of the G20 Data Gaps Initiative. The G20 already operates a traffic-light monitoring framework for data commitments, and she suggested that fostering bilateral discussions between the statistical producers responsible for the largest discrepancies could make “a great deal of difference.”

Speed, standards, and StatGPT

The asymmetry problem is compounded by a speed problem. Official statistical standards move slowly: it took from Bitcoin’s 2008 launch until the 2025 publication of the BPM7 manual for a harmonised statistical treatment of crypto assets to appear, and full implementation is not expected until 2030. Köhler-Geib called for a more flexible approach — a stable core of concepts such as economic ownership, surrounded by a faster-moving layer of new breakdowns and supplementary items that can be updated on a rolling basis via the G20 Data Gaps Initiative, rather than waiting for once-a-decade manual revisions.

On the user side, she highlighted StatGPT, a platform under development at the IMF that aims to let users query official statistics in natural language, combining the accessibility of large language models with the reliability of SDMX-governed data. The SDMX sponsor organisations — the BIS, ECB, Eurostat, IMF, OECD, UN and World Bank — issued a joint statement in December 2025 pledging to make official statistics “discoverable, machine-readable, and globally usable.”

At the Bundesbank, internal reforms have shortened securities statistics production times by around three-quarters, from weeks to days. An integrated data and analytics platform now serves as an internal marketplace for data and tools, and the bank has centralised its financial market data management.

Statistics, trust, and democracy

The OECD’s 2024 Survey on Drivers of Trust in Public Institutions found that citizens who perceive official statistics as reliable are nearly four times as likely to trust their national government as those who do not (59 percent versus 15 percent). Those who believe government promises can be checked against official data are more than twice as likely to express trust.

Yet a majority of citizens in OECD countries — and in Germany — believe it is rarely or never possible to verify government claims using statistics. Köhler-Geib called this a “critical challenge” and argued that clear documentation, intuitive visualisation, honest communication of uncertainty, and a willingness to respond when statistics are misused are all part of the statistician’s responsibility to democratic governance.

She closed by returning to the Ambrogio Lorenzetti frescoes in Siena’s Palazzo Pubblico, which depict the consequences of good and bad government. “Independent statistics are a part of ‘good government’,” she said. “In a fragmenting world, we will only be able to provide good answers if we all work together.”