Deleveraging, Traditional versus Capital Markets Banking and the Urgent Need to Separate and Recapitalise G-SIFI Banks

Abstract

Since the crisis, even with massive support from governments and central banks, widespread regulatory changes and promises from bank executives to improve the governance of risk, the world continues to see failures of Globally Systemically Important Financial Institutions (G-SIFIs, like Dexia), and huge losses (most recently from JP Morgan). Banks refuse to lend to each other, the central banks have become the interbank market and ‘bad deleveraging’ bears down on the economy forcing job losses in small- and medium-sized companies. ‘Good deleveraging’ occurs via building capital, and in this respect the US approach to dealing with the crisis provides something of a lesson that policy makers in Europe should take note of. With respect to regulations, the paper shows that capital and liquidity rules create a bias against lending to the enterprise sector (that drives jobs and economic growth). With respect to G-SIFIs, the paper shows how movements in their balance sheets are dominated by derivatives, the exposure to which varies with the cycle in risk. Netting of derivatives provides no protection against market risk, and the collateral and margin calls associated with these swings is both pro-cyclical and dangerous. The paper argues the OECD case that the best way to deal with all of these issues – both materially reducing the risk that arises from too-big-to fail while encouraging well-capitalised retail banks get on with the job of lending to create jobs – is to separate retail banking from securities business and ensure the former is (particularly in Europe) well capitalised. In this respect the paper argues that the non-operating holding company approach with ring-fenced subsidiaries (close to the Vickers proposal in the UK) is perhaps a better model than the US Volcker rule.

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 — 31 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.