Liquidity, Bank Runs, and Bailouts: Spillover Effects During the Northern Rock Episode

Lindgren et al. (1996) show that, during the period 1980–96, of the 181 IMF member countries, 133 experienced significant banking problems. Such problems have affected developed as well as developing and transitional countries. Also see Dell’Ariccia et al. (2008) for an analysis of the real effects of banking crises.

See MacKinlay (1997) and Chapter 4 in Campbell et al. (1997) for a detailed discussion of event studies in economics and finance.

See De Bandt and Hartmann (2002) for an excellent survey of theoretical and empirical models of systemic risk and the references therein for further discussion.

The timeline and the discussion here rely on the description of the Northern Rock episode in the October 2007 Financial Stability Report of the Bank of England (2007); the articles “Down the Drain” and “Lessons of the Fall” in the September 6, 2007 and October 18, 2007 issues of The Economist, respectively; and the BBC website: http://​news.​bbc.​co.​uk/​2/​hi/​business/​7007076.​stm.

Northern Rock had only 76 branches in 2007 and retail deposits accounted for only 27 percent of its liabilities, whereas wholesale funding accounted for 68 percent of its liabilities, according to its interim results in June 2007. In January and May 2007, Northern Rock raised £10.7 billion through securitization, which made it the top securitizer among British banks.

This can arise due to the costs associated with changing banks that unsophisticated depositors incur, such as switching and searching costs. For example, data collected on current account switching behavior from the Financial Research Survey of the National Opinion Poll for the UK imply that a representative current account holder would only change banks every 91 years (Gondat-Larralde and Nier 2006).

According to the tripartite structure, the Bank of England, the Financial Services Authority, and the Treasury are mandated to deal with financial crises.

Acharya and Yorulmazer (2008) analyze various policy options to resolve bank failures such as liquidation, bailouts, and sale of a failed bank to a healthy bank, with and without government assistance. They show that, from a social welfare point of view, the sale of a failed bank to a healthy bank with government assistance can be preferable to bailing out the failed bank.

An announcement on September 21 confirmed and clarified the guarantees announced on September 20.

An important issue is whether there were other events, not directly related to the run on Northern Rock that might have confounded the effects. For example, Wall and Petersen (1990) extend Swary’s (1986) analysis of Continental Illinois and show that events related to U.S. banks’ holdings of claims on Latin America were confounding factors that needed to be included in the analysis. One such issue arises in our analysis for the effects of the bailout announcement. On September 19, at around noon, the Bank of England announced an injection of liquidity into the money markets and the extension of collateral to include mortgage debt, in an attempt to ease the liquidity crunch in wholesale markets. Hence, the returns on September 19 can be jointly affected by the bailout announcement on September 17 and the Bank of England’s announcement on September 19. To focus solely on the effect of the bailout announcement, we prefer to concentrate on returns for September 18.

The total assets of the ten banks in the study are approximately £4441 billion, which is around 317 percent of the U.K. GDP (£1401 billion) in 2007. The total assets of Alliance & Leicester, Bradford & Bingley and HBOS—the three banks on which we found significant spillover effects—are approximately £710 billion, which is around 51 percent of the U.K. GDP in 2007. Furthermore, spillover effects during September 14–17 wiped out 59, 32 and 14 percent of Tier 1 capital of Alliance & Leicester, Bradford & Bingley, and HBOS, respectively.

O’Hara and Shaw (1990) study the effect of the announcement of the too-big-to-fail guarantee by the U.S. Office of the Comptroller of the Currency, and find a positive value effect for the largest banks.

On September 19, at around noon, the Bank of England announced an injection of liquidity into the money markets and the extension of collateral to include mortgage debt. As a result, we chose to focus only on September 18 to analyze the effect of the bailout announcement. However, on September 19, Bradford & Bingley experienced a significant positive abnormal return (7.5 percent, significant at the 1 percent level). Hence, the cumulative abnormal return for Bradford & Bingley is 11.8 percent for September 18–19, which would only make the results of the paper stronger. While the failure of Northern Rock is clearly bad news for other banks that pursue the same business model, the announcement on September 18 contains some uncertainty about the government’s future response to difficulties in other banks. Even though the announcement on September 18 may imply that similar guarantees might be provided for other banks in the future, this potential guarantee is not explicit. Hence, the uncertainty about future government interventions still remains, which mitigates the effect of the government intervention. However, one can interpret the announcement on September 19 as indicative of the authorities’ willingness to intervene, which would have reduced the uncertainty about government interventions. Hence, the effects we find for the government intervention could be expected to serve as a lower bound for the effect of the bailout announcement.

Inclusion of Northern Rock makes the results stronger.

See Fig. 2 for an illustration of the relation between abnormal returns and wholesale funding. Note that Northern Rock is excluded from Fig. 2.

The other three explanatory variables, deposit, size, and capital, mostly have the expected signs, but they are highly insignificant.

Wholesale is significant at the 5 percent level for September 14–17, significant at the 10 percent level for September 14 and September 17, whereas the p-value is 0.12 for September 18.

This finding is consistent with Shin (2009), who argues that Northern Rock’s reliance on institutional investors for short-term funding made it extremely vulnerable to the drying up of liquidity in credit markets.

Results are available from the authors upon request.

A further confirmation for the robustness of our results is the subsequent failures (or near failures) of the institutions we show to have been affected. In particular, in the aftermath of Lehman’s failure, the takeover of Alliance & Leicester by the Spanish bank Grupo Santander was ratified by shareholders on September 16, 2008; the U.K. government announced that Bradford & Bingley had been partly nationalized and that Grupo Santander had purchased the savings business on September 29, 2008; Lloyds TSB’s acquisition of HBOS was announced on September 19, 2008, effective on January 19, 2009; and HBOS’ pre-tax loss of £10.8 billion in 2008 hit Lloyds TSB, which had to be recapitalized by the U.K. government. As the sequence of events confirms, the effects we found were neither spurious nor idiosyncratic; rather, they can be systematically explained by the business model of the affected banks.