Under the view that the market is a weighted and directed network (Barabási, 2003), this document is a first attempt to model the Colombian money market within a spatial econometrics framework. By estimating two standard spatial econometric models, we study the cost of collateralized borrowing (i.e. sell/buy backs) among Colombian financial institutions, and its relationship with the effects induced by traditional variables (leverage, size and borrowing levels), and by spatial variables resulting from observed linkages among financial institutions. The model that best fits the data is the Spatial Durbin Model, whose main findings indicate that (i) traditional variables are of low explanatory power by themselves; (ii) there exists a significant spatial dependence with regard to the cost of collateralized borrowing; (iii) the inclusion of spatial lags of the same traditional factors results in a model able to explain the existence of borrowing spreads that vary across financial institutions despite the collateralized nature of sell/buy backs; (iv) direct and spill-over effects from the spatially lagged value of financial leverage are the most significant for determining the cost of collateralized borrowing. Results are valuable since making connectedness an explanatory variable break with the traditional (reductionist) understanding of financial markets, which concurs with the current interest in the macro-prudential perspective of financial stability.