What Completing the Banking Union Can—and Cannot—Fix

The incomplete European Banking Union can remain fragmented even when the interest rates on realized cross-border loans look similar because price is observed at the end of the access process. A bank must first operate in a market, a firm and bank must form a relationship, and only then can their loan terms be compared. This final synthesis connects those stages with the regulatory evidence and the modelled macroeconomic response, so the reader can judge both the promise and the limits of completing Europe’s banking network.

The answer is hidden in the conditioning

The central distinction is now straightforward. Entry and relationship formation are extensive margins: they determine whether foreign banking capacity and a particular bank-firm link exist. Interest rates and loan quantities are intensive margins: they describe the terms of a link after it exists. The distinction is not technical decoration. It determines which population a statistic describes.

Under the paper’s operational-geography definition, cross-border lending averaged 5.6% of total euro-area lending from 2019 onward, while cross-border relationships accounted for less than 1% of relationships. A small number of unusually large links can produce that combination. In fact, cross-border relationships were on average about ten times larger in outstanding amount and more likely to be syndicated. The volume statistic therefore does not mean that a typical firm has a foreign lender. It reflects the weight of a selected set of large relationships.

The selection runs through both sides of the market. Roughly 40% of banks had at least one cross-border relationship, but very few firms borrowed across borders. The firms that did were older and much larger in turnover, balance-sheet size, and employment. The banks involved were also larger in assets, lending, and deposits. Cross-border finance is visible, but it is not distributed evenly across the potential borrower-bank network.

Once those relationships exist, the terms look much less divided. The preferred fixed-effect estimate places cross-border interest rates about 28 basis points below domestic rates among observed relationships, roughly 8.5% of the sample mean rate. Conditional cross-border loan shares are about 14% lower in the preferred quantity specification. Both results allow some pairs to be more or less favorable than others, but their average size is modest beside the relationship-formation gap. A foreign loan can therefore be slightly cheaper for the borrower who obtains it while foreign borrowing remains inaccessible to most firms that never reach the comparison.

The entry margin reveals the same issue from the supply side. The estimated cross-border entry gap is about 99 percentage points when a group’s foreign-market presence is compared with home operation under the paper’s stated specifications. A group that does not establish a branch or subsidiary cannot compete for local borrowers as an operating bank, regardless of how similar its potential loan rate might be to a domestic bank’s rate. Entry is not just another contract term; it changes the set of institutions with which firms can form relationships.

This is why Europe can be fragmented beneath price convergence. The observed rate is conditional on two earlier successes: the group has crossed the organizational border and the firm has crossed the matching border. If those successes are rare, the observed loans can look integrated while the market that generates them remains nationally concentrated. Price convergence among survivors does not imply convergence in access.

The same logic clarifies the difference between operational and nationality-based integration. A locally operated branch of a foreign group can be domestic under the residency definition used for lending relationships and foreign under the ownership definition used for group entry. Reassigning activity by ultimate parent raises the measured foreign share to 11% of loans and 14.6% of loan amounts, but the diagonal of the lending matrix remains dominant and cross-border branching remains low. Ownership integration can therefore coexist with local operating concentration.

The source’s answer is not that prices are irrelevant.