Bank Monitoring with On-Site Inspections

Banks are supposed to have a unique edge – not just in lending, but in understanding borrowers, monitoring risk, and intervening when things go wrong. For decades, theory has argued that this monitoring advantage is what makes banks special relative to nonbanks. Yet direct evidence has been limited. This paper opens that black box. Using detailed, loan-level data and actual inspection reports, it shows how banks monitor borrowers in real time, how they act on that information, and how monitoring causally reduces default risk.

Bank Monitoring with On-Site Inspections

• Amanda Rae Heitz, Christopher Martin, Alexander Ufier
• The Journal of Finance, 2026
• A version of this paper can be found here
• Want to read our summaries of academic finance papers? Check out our Academic Research Insight category

Key Academic Insights

Monitoring targets riskier borrowers and projects
Banks monitor more when risk is higher. Loans with lower credit scores, higher loan-to-value ratios, speculative projects, and inexperienced borrowers receive more frequent inspections. Monitoring is not random. It is actively allocated where the marginal benefit is highest.

Monitoring substitutes for stricter loan terms
Loans with more intensive monitoring tend to have lower spreads and fees, and larger loan sizes. This suggests a trade-off. Instead of pricing risk entirely through higher interest rates, banks rely on monitoring to control risk dynamically over time.

Banks actively use monitoring information in real time
Inspection reports matter. Negative language in inspection reports is strongly associated with draw denials, while positive language increases approval likelihood. This is direct evidence that banks process and act on soft information, not just hard metrics.

Monitoring responds to changing economic conditions
When local housing markets improve, banks monitor less and deny fewer draw requests. When conditions worsen, especially with higher foreclosure rates, banks become stricter and deny more funding. Monitoring is adaptive, not static.

Monitoring reduces defaults, causally
Using an instrumental variable approach, the paper shows that increased monitoring significantly reduces loan defaults. The causal effect is large. Roughly, a modest increase in inspection frequency can meaningfully lower default probability, and standard regression methods actually underestimate this effect.

Practical Applications for Investment Advisors

Monitoring is a core source of alpha in credit

This paper reinforces that credit investing is not just about selection. It is about ongoing oversight. Active monitoring can materially improve outcomes, especially in opaque or complex investments.

Dynamic risk management beats static underwriting
Risk cannot be fully priced at origination. The ability to update views and act on new information over time is critical. Monitoring provides that flexibility.

Soft information matters
Quantitative models are not enough. Text, judgment, and qualitative signals can drive real decisions. The fact that inspection language predicts funding decisions highlights the importance of qualitative analysis.

Incentives shape outcomes
The most powerful effect of monitoring is often indirect. When borrowers know they are being watched, behavior improves. Structuring investments to create these incentives can be as important as selecting the right opportunitie

How to Explain This to Clients

“Banks do not just lend money and wait to be repaid. They actively monitor borrowers throughout the life of a loan. This research shows that when banks closely track how projects evolve and act on new information, they can significantly reduce losses. Even more importantly, the mere presence of monitoring encourages borrowers to behave more responsibly, improving outcomes before problems even arise.”

The Most Important Chart from the Paper

Figure 1 shows the relative frequencies for some loan-level variables for the full sample of loans. Panels A and F report, respectively, the combined loan-to-value (CLTV) of projects (reported as a percentage where 100 is 100% loan-to-value), original project loan amount in thousands, loan term to maturity in months, total lifetime inspections, total lifetime draws, and time to first inspection in months.

The results are hypothetical results and are NOT an indicator of future results and do NOT represent returns that any investor actually attained. Indexes are unmanaged and do not reflect management or trading fees, and one cannot invest directly in an index.

Abstract

This paper provides direct empirical evidence on bank monitoring using proprietary data on nearly 30,000 construction loans and associated on-site inspection reports. The authors examine what drives monitoring decisions, how banks use information gathered through monitoring, and whether monitoring affects loan outcomes. They find that monitoring is more intense for riskier borrowers and projects and is actively used in decision-making, with negative inspection reports leading to funding denials. Monitoring intensity responds to changing economic conditions and increases as banks approach distress. Using an instrumental variable framework, the authors show that monitoring causally reduces loan defaults, with much of the benefit arising from improved borrower behavior due to the threat of monitoring. Overall, the findings support theoretical models that emphasize monitoring as a key mechanism through which banks manage credit risk and improve lending outcomes.