I develop a model of bank behavior in the over-the-counter overnight interbank loan market that can explain empirical patterns observed in a dataset from Mexico. With the model, I aim to understand the nontrivial network structure observed in the market. Understanding the network structure is important because a financial market typically has a small number of large banks occupying a large share of the market, and the way in which they are connected can determine the characteristics of the financial system.
As a first step, I present a simple framework that can describe the behavior of banks in the market. In the data, the size of a bank itself, measured by its total assets, has important effects on its behavior such as the interest rates on the loans that the bank lends and borrows. I show that how the assumption that a larger bank is less sensitive to the same amount of change in the level of liquidity can explain the observed effects of bank size.
Then, I consider the matching between a lender and a borrower in this market. One important feature in the data is that the market does not look as if it were being cleared through a centralized exchange or by a Walrasian market maker. A lender-borrower pair is distinguished from another lender-borrower pair, not only by the characteristics of individual lenders and borrowers involved, but also by the identity of the pair itself.
Therefore, I argue that there must be an unobservable factor that determines the matching. I represent this hidden factor as some type of ‘residuals’ or lender-borrower pair fixed effects that comes out of the constructed matching model. Then, I use the financial crisis in 2008 as an exogenous event affecting their distribution and discuss its interpretation.