A surprising way bank mergers can devastate poor neighborhoods

When banks merge, they often close branches in an effort to cut costs and eliminate redundancies. When regulators scrutinize these closures to look for the public interest, they typically focus on the idea of avoiding “bank deserts,” where residents would lack convenient access to a branch of any bank. Yet since the idea of branch closures is generally to reduce redundancy, they frequently pass regulatory muster.

But a new paper from MIT graduate student Hoai-Luu Q. Nguyen argues that regulators may be looking for harm in the wrong place. Branch closures in low-income neighborhoods turn out to have surprisingly large — and negative — impacts on the local economy even if other branches remain open.

The key finding of Nguyen’s paper is that when a branch closes in a low-income area, small-business lending disappears. The impact is very localized, which is likely why it hasn’t been uncovered in earlier research. Nguyen’s research looks at census tracts, a very small geographical unit that typically covers just a few blocks in a dense urban area, and finds that small-business lending declines by 13 percent within a small area. The impact dissipates within 8 miles and is concentrated in low-income and heavily minority areas.

Interestingly, branch closure does not seem to impact mortgage lending at all. This mitigates the adverse economic impact of a bank closure, but it also sheds light on a larger theoretical debate.

Traditionally, small banks and the trade associations that represent them have argued that small banks have a special approach to business that makes them uniquely well-suited to providing financing to small businesses. The idea is that while big Wall Street players chase huge deals and commodity loan products that can be homogenized and securitized, small banks can make deals based on local knowledge and local relationships.

This has long been conventional wisdom in the industry, but some recent research has challenged that idea.

Nguyen’s paper shows that relationships really do matter. Mortgage loans are an essentially standardized product at this point, so the presence or absence of any particular bank branch is basically irrelevant to the volume of lending. Everything is driven by national interest rates, computer models, and policies.

Small-business lending really looks to be different. It’s not just that a neighborhood needs a local bank branch for local businesses to be able to get credit — the presence or absence of a specific branch seems to have meaningful consequences.

The implication is that outside the web of algorithmic lending there are plenty of businesses that most bankers won’t lend to, but that particular bankers with specific knowledge of the community think are worth investing in.