Discussion

Abstract

The authors opened the discussion by addressing three concerns raised by the discussants. First, they pointed out their model assumes that asset holdings and liabilities grow at an exogenous rate. As a consequence, the balance sheet of banks is inelastic with respect to government guarantees, and the value of deposit insurance is not competed away, as noted by Lawrence Summers during his discussion. The authors agreed that changes in regulation or government guarantees could affect the size of the banking sector and that this issue is important for policy making. They argued that this response is slow due to adjustment costs in the banking sector, and the market-to-book ratio is expected to increase during the transition. Second, the authors acknowledged that both the reaction time before bailout and the presence of jump risk are key determinants of the value of government guarantees, as emphasized by Summers. Their calibration assumes a market-to-book ratio of 2, corresponding to that observed during the period of interest, and a risk-neutral probability of a crisis of 5%. With a reaction time of a year, their model predicts a government bailout worth half of book equity. They noted that with a precrisis book equity value of banks around $1.1 trillion, this amounts to a subsidy of $550 billion. In the authors’ view, this figure is comparable to the actual government support during crisis,which speaks in favor of a 1-year delay before bailout. Third, the authors confirmed that theirmodel abstracts from interest rate risk, as highlighted by Juliane Begenau as part of her discussion. They argued that such risk does not generate large enough losses over the course of a year to rationalize the value of government guarantees. Credit risk is more attractive in this respect, theymentioned, due to larger jumps and tail risk. In the authors’ opinion, the role of interest rate risk during earlier episodes, including the savings and loans crisis, was associated with longer reaction times by regulators.