Organizational structure, rule enactment, and relational embeddedness in small business finance
From its origins, a central question of organizational theory is how firms balance the efficiency and reliability that result from centralization and standardization of decisions with the adaptability, flexibility, and innovation that result from decentralization, local customization, and experimentation. This tension is especially salient for organizations that pursue multiple missions (e.g. hybrid organizations), or those that must rely on relational ties (e.g. financial firms).
For example, young and small firms depend disproportionately on external finance for their survival and growth. They also lack substantial collateral, formal reporting mechanisms, or track records, which makes them difficult to find, evaluate, and monitor. This creates pressures for banks and microcredit organizations that lend to small firms to find economies of scale, standardize processes, and automate lending decisions to reduce costs. But because small firms lack structure and face more uncertainty, the “soft information” that is best conveyed through relational ties is uniquely valuable in lending decisions. Employees with personal ties to clients use their local knowledge and private information to interpret, enact, and improve rules and procedures. Lending firms, therefore, are constantly pulled in the opposite directions of standardization and flexibility and a long research tradition in finance has studied the benefits and costs of each.
In my work exploring these questions, I straddle levels of analysis and combine rich ethnographic data with longitudinal loan-level datasets from lending organizations. I also exploit different sources of exogenous variation, such as random loan officer rotations, exogenous determinants of branch locations, or natural variation in loan officer enforcement styles.
In a first paper with Ramana Nanda (2012), we analyze these tensions at the organizational level. We contrast the performance of centralized vs. decentralized lending structures. In decentralized banks, branch managers have greater autonomy over lending decisions, so they have an incentive to use “soft information” when setting loan terms. Client firms that are smaller or have less verifiable information are more likely to be approved for and receive larger loans from decentralized banks, as theorized by prior finance research. We show, however, that this only holds true when there is local competition between banks. In concentrated markets, decentralized banks are more likely to cherry pick the best clients, give them smaller loans, and charge them higher rates. Branch managers in decentralized structures, therefore, use soft information profitably. But for small borrowers, the benefits of decentralized lending structures are dependent on the context.
In a next set of papers I straddle the team and individual levels of analysis. In two related papers (2011, 2012), I show that loan officers exercise discretion productively, to (i) better serve clients, (ii) improve bank rules, and (iii) defend officer status within the organization. This, however, runs counter to organizational pressures to centralize decisions to reduce prohibitive costs. While rule bending provides substantial benefits to the organization, it also carries significant costs. Officers who bend the rules are more difficult and costly to manage. Furthermore, it is not automatic that actors will use their discretion to benefit the firm. They could become too close to clients or withhold private information for their personal benefit.
To manage these contradictions, firms can follow several approaches. In my 2014 paper, I compare branches that contain higher concentrations of agents who follow rules strictly, adhere to a flexible relational model, or blend both strategies. Branches that have a preponderance of agents who blend both styles experience a decrease in organizational performance, but so do branches with high concentrations of agents who practice either pure style. In contrast, branches that contain “discretionary diversity”—a balanced distribution of agents with opposing styles—perform best. This is because loan officers process lending decisions in branch-level credit committees, where they must justify their actions to peers. Discretionary diversity within teams creates a productive tension that pushes loan officers to justify their decisions according to broader organizational goals and promotes a balance between routines and discretion.
A complementary paper with Minkyung Kim, K. Sudhir, and Kosuke Uetake (2019a) focuses on the individual level. It shows that loan officers do engage in moral hazard and withholding of private information. This is reduced, however, when they are made accountable for their decisions by tying their pay to loan performance and repayment. To reduce loan officers’ grip on private client information, they can be randomly rotated. While this creates some information loss and severed relationships, in the aggregate the firm benefits from improved alignment of loan officer effort. Next, with Jason Greenberg (2015), we straddle the individual and organizational levels, to study whether individual employees or the firms that employ them retain the relational capital created between a firm and its clients. We show that loan officer rotations increase the probability that a client will become delinquent, underscoring that relational contracts can only exist between individuals (the loan officer and the client). This disruption, however, is mitigated when the loan officer is replaced by another who follows a similar relational style. Thus, relational contracts between organizations and individuals can continue in the face of turnover if agents are replaced with others who exhibit consistent relational styles, and organizations can create a sustainable competitive advantage through routines that train their agents (and clients) to develop consistent interaction styles.
My work in this area contributes to organization theory by showing that even though bureaucratic (standardized, centralized) and relational (customized, decentralized) models represent contradictory philosophies, they can coexist and enhance one another, but only if explicit structures are established to ensure that they remain in productive balance.