Financial covenants have gone in and out of style over the last 30 years. They serve to transfer control rights to lenders when a borrower’s financial metrics breach pre-set contractual thresholds and so provide an interesting laboratory to study debtholder-shareholder conflicts and how they are potentially resolved. First, covenants can enhance efficiency by making contracts between lenders and their borrowers more complete. Second, through the contingent transfer of control during the term of the loan, they provide scope for renegotiation and wealth transfers that are primarily at the discretion of the lender.

In practice, covenant violations are associated with a series of economically significant consequences for borrowers. These consequences include changes in investment, governance, and executive compensation, debt issuance and capital structure, innovation, employment, and financial reporting. Essentially, lenders use control to constrain and lower the risk of their borrowers, making the borrowers more likely to repay. However, this constraint on potentially profitable ventures represents a cost to borrowers. In a new paper, we investigate the price supporting the market transfer of covenant-contingent control rights: What is the cost of control in the corporate loan market?

To answer this question, we develop a simple theoretical framework. All else equal, lenders prefer both high interest rates and acquiring control but will trade them off. Likewise, borrowers prefer both low interest rates and retaining control but will trade them off. If homogeneous lenders compete to make a loan, the winner must offer terms that weakly exceed the borrower benefits of the next best offer (i.e., this follows the intuition of second price auctions). By this logic, if a borrower obtains multiple loans, these loans should generate the same benefits to the borrower. Therefore, when a borrower obtains multiple loans, the observed loan terms trace out the borrower’s willingness to  trade off loan spreads and control rights. We establish theoretical conditions under which the slope of the borrower’s indifference curve can be interpreted as the cost of control. Empirically, our objective is to estimate this slope.

Our preferred estimate of the cost of control is 52.7 basis points. This estimate accounts for unobserved borrower quality by relying only on variation in loan terms for the same borrower in the same year. For the average loan in our sample with covenants, this implies a 14.8 percent discount on the cost of borrowing compared with a loan with no covenants. We find that this control discount varies significantly over time, and with the exception of a spike during the 2007-2008 financial crisis, has been decreasing since a peak in 2000. Using loan spreads alone as a measure of credit conditions likely understates credit tightness, particularly during credit contractions. The gap between the observed spread and the control-adjusted spread has also systematically narrowed over time as the market has shifted toward covenant light or “cov-lite” loans. Our method demonstrates how market participants and regulators might adjust observed spreads to account for the price-control tradeoff.

An important part of our theoretical framework is that we identify the conditions necessary for the logic underlying our approach to hold. The first condition requires that we observe multiple loans for the borrower in the same time period. We find that the estimated cost of control is monotonically decreasing as we lengthen the period in which the multiple loans must be observed from one quarter to two years. This pattern is consistent with the theoretical intuition that with longer periods, unobserved borrower quality is more likely to change between the two loans. A similar concern is that loan terms could be related across multiple loans because giving up control in one loan makes it less costly for the borrower to give up control in the other. If this effect is significant, it would imply that the presence in our sample of borrowers receiving more than two loans in the same period should attenuate our estimates. However, we find no statistical difference in our estimates when we look at borrowers receiving more than two loans in the same year. This suggests that complementarities across loans are not a major determinant of the cost of control and makes it likely that our estimates would generalize to borrowers that only receive a single loan in a given year.

The second condition is that lending to the borrower is sufficiently competitive to require that winning lenders must each offer the same surplus. To assess this condition, we construct several measures of potential lending competition in the borrower’s location, industry, and size segment. For each measure, we find no evidence that more competition leads to significant differences in the estimated cost of control, consistent with sufficient competition for the theoretical condition to hold in our sample.

Holding non-price and control terms constant is the third required condition. A potential concern is that non-price and control terms are related to the price-control tradeoff we estimate. We explicitly address this possibility by flexibly controlling for other terms (amount, collateralization, and maturity) and find relatively stable estimates. Of equal or greater interest is whether these other terms interact with the cost-control tradeoff, consistent with the presence of complementarities across contracting terms. We find insignificant interactions for all terms other than maturity. Longer-maturity loans imply a moderately lower cost of control, consistent with the trust underlying longer relationships substituting for explicit contracting.

Our findings have implications for our understanding of the pricing of debt claims and the value of contingent control rights. Specifically, our findings highlight the control discount implied by the equilibrium tradeoff of effective interest rates and control rights made by borrowers and lenders.  Because control transfers between borrowers and lenders are cyclical, our results imply a systematic relationship between the magnitude of the control discount and credit cycles. This relationship underscores the importance of adjusting for the control discount when drawing inferences about borrower risk from loan interest rates.

This post comes to us from Professor Andrew Bird at Chapman University’s George L. Argyros School of Business & Economics. It is based on a recent paper written by him, Stephen A. Karolyi, and Thomas G. Ruchti, “Estimating the Cost of Control Rights in the Corporate Loan Market,” available here.