Bankruptcy or Bailouts?

Abstract

The usual reaction if one mentions bankruptcy as a mechanism for addressing a financial institution's default is incredulity. Those who favor the rescue of troubled financial institutions, and even those who prefer that their assets be promptly sold to a healthier institution, treat bankruptcy as anathema. Everyone seems to agree that nothing good can come from bankruptcy. Indeed, the Chapter 11 filing by Lehman Brothers has been singled out by many as the primary cause of the severe economic and financial contraction that followed, and proof that bankruptcy is disorderly and ineffective. As a result, ad hoc rescue lending to avoid bankruptcy has been the preferred solution. In this Article, we seek to provide the first careful assessment of the belief that governmental rescues are preferable to bankruptcy. While the interaction of financial firms, systemic risk, and Chapter 11 is complex, our analysis suggests that the widespread belief that bankruptcy should not be used to resolve the distress of financial firms is misguided, and that it has had serious costs in the recent crisis. Although bankruptcy is not always the optimal response to financial distress, it is more effective than is generally realized. In Parts I and II of the Article, we describe the principal problems created by financial distress - debt overhang and creditor runs - and the mechanisms bankruptcy provides for addressing these problems. We then provide historical context in Part III, looking to Drexel Burnham's bankruptcy in 1990 for further lessons about the efficacy of bankruptcy. In Part IV, we turn to firm-specific bailouts, describing this strategy's benefits and the distortions it causes. We then shift our focus back to bankruptcy, considering the (legitimate) concern that it may not adequately counteract systemic risk in Part V, and exploring its treatment of derivatives, one of the chief new habitats of systemic risk, in Part VI. Part VII is a brief conclusion.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 86 citations exist upstream.