Working Paper Series no. 801: Aggregate Implications of Credit Relationship Flows: a Tale of Two Margins

What drives the fluctuations of credit over the business cycle and in the long run? How do banks adjust their credit supply in response to aggregate shocks or policy changes? These questions have been at the forefront of macro-finance and banking research at least since the seminal work of Bernanke (1983). Yet, our understanding of aggregate credit fluctuations and their implications for the real economy remains incomplete on several fronts. Bank credit is a significant source of financing for the majority of businesses. One particularly important aspect that has been extensively studied at the micro level, yet overlooked in macro, has to do with bank-firm credit relationships. Indeed, a vast theoretical and empirical literature has long highlighted the role of these relationships in terms of alleviating agency frictions and shaping credit supply at the lender-borrower level. It also emphasized the existence of cross-sectional heterogeneity in terms of match quality and inherent relationship characteristics such as duration, which can potentially hinder banks' ability to adjust their credit supply in a frictionless way. Conversely, the common view across most macro-finance models either simply assumes homogeneous borrowers and/or lenders, or abstracts from the long-term nature of financial contracts and any market frictions that may prevent banks from costlessly forming or severing these credit matches. These models thus downplay the value of relationships and their aggregate consequences and imply that banks can swiftly adjust the number of their borrowers in response to shocks. They also leave little room for analysing the process of credit reallocation across bank-firm matches and its dynamics throughout the cycle.

This paper proposes a novel macro perspective on the process of credit intermediation. It aims to provide further empirical evidence on the key and distinctive roles played by both the intensive and extensive margins in shaping aggregate credit fluctuations. Here, we attempt to look behind such fluctuations in order to address first-order questions such as: (i) When aggregate bank credit declines by five percent, is it because the average loan size (i.e., intensive margin) drops by five percent, or is it because five percent of bank-firm matches (i.e., extensive margin) are destroyed? (ii) Does the origin of aggregate credit fluctuations matter? (iii) Do monetary policy shocks impact these margins differently?

To answer those questions, we leverage a key source of information, the French Credit Register, which covers the commercial loan market in France, and is maintained by Banque de France. The data contains granular and nearly exhaustive records of bank-firm matches and corresponding credit exposures over the period 1998 - 2018. To study the properties of credit relationships flows, we develop an empirical methodology akin to the one pioneered by Davis and Haltiwanger (1992) for labor flows. Our methodology takes into consideration specific characteristics associated with credit market structure and available data. For example, we track data entries for each bank-firm match to determine the time of creation and inferred time of destruction in order to construct the associated gross credit relationship flows. We also account for cross-sectional heterogeneity and the nature of financial contracts through key attributes such as loan size, credit type and maturity, and relationship duration. Our empirical investigation establishes the following stylized facts about the extensive and intensive margins of credit: (i) extensive and intensive margins fluctuate continuously over time, (ii) although both margins are important at the business cycle frequency, the intensive margin plays a more prominent role, contributing about one half to three quarters of the variance in aggregate credit, (iii) in the long run, the extensive margin accounts for the bulk of aggregate credit variations (i.e., 90+%), and (vi) the intensive margin displays higher volatility relative to the extensive margin, while their persistence is roughly identical. It also highlights the following features pertaining to gross credit relationship flows: (i) the creation, destruction, and reallocation of bank-firm relationships coexist throughout the cycle, (ii) creation (inflows) and destruction (outflows) of relationships show greater volatility compared to net flows. Variations in net flows are driven mainly by inflows, and (iii) outflows exhibit greater volatility for small and short-term loans and credit relationships with duration of less than one year. Inflows exhibit greater volatility for relationships involving small loans and lines of credit.

Our empirical framework also provides us with tools to better understand the nature of the reallocation process occurring in credit markets and the channels through which bank shocks get transmitted to the real economy. In particular, we show that the excess reallocation rate of credit relationships is countercyclical, in line with the cleansing effect of recessions. In addition, yearly (excess) reallocation rates have been steadily declining over the past two decades. These results indicate the existence of factors hampering credit market fluidity and contain relevant theoretical and policy ramifications worthy of further investigation.