Operational risk is a first-order concern for firm management and stakeholders. Stark examples of losses due to corporate operations include BP’s $17.2 billion loss in June 2010 following the Deepwater Horizon incident (Wong and Yousuf, 2010) and Freeport-McMoRan’s $13.9 billion loss in 2008 due to plunging metal prices and difficulty with the acquisition of a rival company (James, 2009). The consequences of these and other operating losses are significant, and spill over to connected firms (Wu, 2016). Managers may be willing to make risky operating decisions if they are unaware of the risk or measure it poorly, or if their interests are aligned with equityholders who share in the upside of risky operating decisions. Lenders are both excluded from the upside of risky operating decisions and exposed to the downside. Accordingly, operating decisions should be at the heart of the conflict between debtholders and equityholders.

In our paper, we study the effect of lender control on corporate operating decisions in light of their conflicting incentive to ex-ante reduce exposure to operations that have high outcome variance. We focus on three key issues in our analysis. 

First, we shed light on the influence of lenders on corporate operating decisions, suggesting that the scope of lender influence extends beyond financing and investment decisions. Using a regression discontinuity design, we find that borrowers reduce operational risk-taking following covenant violations, corresponding to a marginal decrease in the probability of experiencing distress within one year by as much as 10%. This finding is robust to a variety of theoretically-motivated controls and alternative estimation methodologies.

Second, we examine the determinants of lender influence. Given the role of covenants in mitigating agency problems, we investigate cross-sectional variation in lender influence associated with ex-ante information asymmetry and agency problems with borrowers. We find that the magnitude of this covenant violation effect is positively correlated with several proxies for borrowers’ ex-ante information asymmetry and agency problems. We further find that lender influence is concentrated among borrowers without credit ratings, with high absolute abnormal accruals, and with dividend restrictions, suggesting that lenders take a more active role in the presence of information asymmetry and agency problems.

We also investigate cross-sectional variation in lender influence associated with supply-side characteristics. In particular, we investigate the coordination costs within lending syndicates, the predisposition of lenders toward active intervention, and expected losses given default. Individual lenders have the incentive to freeride when coordination costs are high (Bolton and Scharfstein, 1996), and this reduces the efficacy of control. Similarly, lenders with active management experience are more likely to have the requisite expertise to more efficiently influence corporate decisions. Lastly, lenders have stronger incentives to exert control when the expected loss given default is high. Our evidence suggests that coordination, active management, and liquidation incentives all contribute to the effectiveness of lender influence.

Third, we introduce a new measure of operational risk based on the financial risk measure of Campbell, Hilscher, and Szilagyi (2010) and industry-specific data on corporate operations. An important contribution of our study is to construct a novel, industry-specific measure of operational risk. We use industry-specific data on corporate operations from the Compustat Monthly Updates – Industry Specific Quarterly dataset to create industry-specific measures of operational risk for eleven industries: airlines, gaming, healthcare, HMO, homebuilding, lodging, mining, oil and gas, retail, semiconductors, and utilities.

We face two main empirical challenges in studying these effects. First, isolating lender influence from borrower behavior and performance is difficult. Lenders may influence corporate operations implicitly by withdrawing funding offers, refusing to rollover debt, or imposing harsh terms in renegotiation. Separating the lender’s decision to selectively use this implicit influence from the borrower’s condition that required refinancing, renegotiation, or additional funds is an identification challenge. We address this problem with a regression discontinuity design.  Second, ex-ante exposure to operating risk is difficult because this exposure is unobserved and subject to measurement error. Moreover, the assessment of any given project’s operating risk varies with the information and experience of the decision-maker, which varies across companies and time, and is likely different from the information set of the econometrician. This problem is addressed through our research design’s built-in attenuation bias due to manager risk-shifting incentives. 

Our research sheds more light on the classic conflict between equityholders and debtholders and provides a new methodology for estimating the exposure of firms to operational risk. Importantly, these results suggest that the scope of lender influence following covenant violations is not limited to financial risk. Though this study demonstrates one way that financing affects corporate operating decisions, it raises new questions. In particular, it calls for more work to understand the effect of operational risk-taking on future renegotiation and the extent to which lenders condition on borrowers' operating decisions at loan initiation.

Stephen A. Karolyi is an Assistant Professor of Finance and Accounting at Carnegie Mellon University, Tepper School of Business.

John Sedunov is The Michele and Christopher Iannaccone ’91 Assistant Professor at Villanova University, Department of Finance.