What “No Harm” Really Means for Bank Mergers and Mortgages
The strongest answer to the question “does bank consolidation automatically mean worse mortgage terms?” is no—but the reason is not that mergers are harmless by definition. This series has examined completed U.S. bank mergers through three borrower outcomes: mortgage rates, application approvals, and subsequent delinquency. The evidence supports a conditional judgment: in the selected completed mergers studied by the Federal Reserve, national consolidation did not produce systematic worsening of those measured outcomes because local competition, portfolio composition, merger selection, and regulatory screening mediate the effect.
That answer is more useful than either reflexive alarm or blanket reassurance. It tells a reader what to check before accepting a claim about concentration. A smaller number of banks is a meaningful structural fact. It is not yet a finding that a borrower’s local mortgage options have narrowed, that the price of a comparable loan has risen, or that newly originated loans will perform worse. Those conclusions require the market, the comparison, and the outcome to line up.
A structural fact is not a borrower outcome
The concern about consolidation is well grounded. From 1985 to 2020, the number of active U.S. banks and thrifts fell from 14,417 to 4,379, while the number of institutions controlling half of banking assets fell from 91 to 10. Mortgages account for more than two-thirds of U.S. household liabilities. Fewer institutions and greater asset concentration can therefore justify scrutiny of whether banks have gained bargaining power over households.
But the relevant market for a mortgage is not the national list of banking institutions. It is the set of credible lenders offering the relevant product in the place and period where a borrower seeks it. The Federal Reserve study counts mortgage originators at the parent-company level and includes both banks and nonbank lenders. Its bank–county–quarter observations average 164 active lenders, with a median of 132; at the unweighted county-quarter level, the mean is 75 and the median is 50. These are different weighting perspectives, not contradictory facts. Both show why a national count of banks can overstate the direct competitive effect of a single mortgage merger.
The distribution of shares reinforces the point. The median lender captures 0.39 percent of a county market, the mean is 1.53 percent, and the 90th-percentile lender holds 3.6 percent. A few large firms matter, but the typical lender is a small participant, and nonbank mortgage companies add competitive alternatives. Most banks also remain geographically localized in mortgage activity: the median originates loans in five counties per year-quarter, while the mean of 29 is pulled upward by a small number of national institutions.
These structural indicators do not prove that every borrower has abundant choice. Rural markets are less dense, digital and branch-based lending may create different choice sets, and lender counts do not measure service quality, switching costs, or the ease of entry. They do something more limited and important: they change the prior expectation about how easily a merged institution can exercise power. A lender facing dozens of alternatives may lose applications if it raises rates or tightens standards. A lender with a large local share and few substitutes may face less discipline. The market must be measured where the outcome occurs.
The borrower outcomes then provide the test. For mortgage rates, the study adjusts for borrower and loan characteristics and compares the treated institution with other lenders in the same county and quarter. For approvals and delinquency, it uses the corresponding event-study comparison without pretending that denied applications or pre-origination loans can be treated like originated contracts.