Credit Cycles and Financial Stability (DFG)

Systemic banking crises are a recurring phenomenon in modern economic history. Understanding their causes and consequences is an urgent priority for the discipline of economics. What makes our modern financial systems so fragile, and crises so costly for the real economy? Recent work in macro-finance has made some important inroads. Empirical evidence points to a general pattern where buoyant conditions in credit markets, measured either via increases in the quantity of credit (Schularick and Taylor, 2012; Jordà, Schularick, and Taylor, 2013; Mian, Sufi, and Verner, 2017) or via low expected returns on credit assets (Greenwood and Hanson, 2013; Krishnamurthy and Muir, 2017), presage banking crises and severe economic downturns. Moreover, these relationships appear robust across space and time.


While the link between buoyant conditions in credit markets and risk of financial instability is well understood, important questions remain open: what makes economic agents borrow and lend so much if this predictably leads to a financially fragile economy? Put differently, what are the driving forces of periodic episodes of excessive risk taking? Is it linked to governance problems in the financial sector, possibly caused by insufficient equity capital? Does competition enhance discipline and make banking sectors safer, or do competitive pressures induce riskier behavior, f.i., by decreasing the charter value of banks? Ten years after the global financial crisis, these remain first-order questions for macroeconomists and policy-makers.


The point of departure of this research proposal is that crossing the next level in understanding the deep sources of financial fragility in modern economies will depend on a new type of data. So far, the macroeconomic literature on credit cycles and financial instability has typically worked with aggregate data over longer time spans, while empirical research in banking and finance has often relied on short-run micro data at the bank-level (or at the loan-level). Both long-run macro and short-run micro data have distinct advantages with respect to achieving external and internal validity. But they have not yet been combined, and the proposed research project aims to make the first step in this direction.


Collaborators: Kaspar Zimmermann and Matthew Baron