The value of being systemically important: event study on regulatory announcements for banks

Abstract

It is assumed that the awarding of a ‘systemic importance’ seal by the regulator has a positive effect on the equity value of its holder. By employing an event study analysis on a new set of regulatory announcements, we find that financial market participants react to these announcements which are, in effect, judgements that a certain credit institution is systemically important. However, the stock returns found for the respective banks are not exclusively positive; a phenomenon for which we provide explanations. Furthermore, our results show that market reactions on the most present event are weakest, indicating that the announcements’ informational value to market participants diminished.

Keywords:

• too big to fail
• systemic risk
• moral hazard
• financial crisis
• bailout
• banking regulation

JEL Classification:

• G01
• G14
• G21
• G28

Notes

¹ At that time, Continental Illinois was the seventh biggest bank in the US in terms of the volume of deposits.

² See US Congress (1984, p. 300): ‘Chairman St Germain: “That is one of the prime reasons for these hearings. We have quite a few, but one of our principal reasons is we have to make a decision. Do we allow, ever, a large bank to fail?” … Mr. Conover: “I think it is important that we find a way to do that.”’

³ The debate culminated in the following pointed remark by deputy McKinney: ‘We have a new kind of bank. It is called too big to fail. TBTF, and it is a wonderful bank’ (US Congress, 1984, p. 300). On the following day, the Wall Street Journal (WSJ) published an article with the title ‘U.S. Won’t Let 11 Biggest Banks in Nation Fail’ along with a list of the eleven biggest US banks ordered according to their balance sheet total (annual financial statement 1983) (Carrington, 1984). The list overlapped to a large degree with the internal list of the regulator (O’Hara and Shaw, 1990, p. 1598). Positively on discrepancies causing welcome ambiguity, see Morgan and Stiroh (2005, pp. 8–9).

⁴ For a differentiation of systemic importance and size, see, e.g. Morrison (2011, pp. 500–1).

⁵ A bank’s living will is a ‘detailed, binding statement of how its assets will be allocated in case of impending default’ (Solow, 2010, p. 28). For an introduction to the implementation and the problematic aspects of living wills, see Packin (2013). On 28 December 2010, the Japanese newspaper Mainichi published an (until that point in time) internal list of 60 systemically important financial institutions by the Financial Services Agency of Japan (Uranaka, 2010). Since this event is an announcement of a national regulator and, thus, should have a considerably lower impact than that of an international one, we use only the (comparable) events (1), (2) and (3) for our analysis.

⁶ Shortly thereafter, the rating agency Moody’s tried to assess 28 banks with a comparable methodology and published their own list in a Weekly Credit Outlook (25 July 2011) (Basel Committee on Banking Supervision, 2011). However, the FSB publication of the final official list in conjunction with concrete regulatory measures in November 2011 should have had information and, thus, a shock effect from the view of the capital market participants.

⁷ See for event (1): Jenkins and Davies (2009), (2): Financial Stability Board (2011, annex p. 4) and (3): Basel Committee on Banking Supervision (2011, p. 15) in connection with Financial Stability Board (2012, pp. 2–4).

⁸ Later, the term ‘abnormal’ prevailed.

⁹ Critically, on a number of questionable, but popular modifications of the original Fama et al. (1969) event study, see Brown and Warner (1980, p. 249).

¹⁰ Of his influential contributions, see von Hayek (1976, p. 117): ‘The sum of information reflected or precipitated in the prices is wholly the product of competition, or at least of the openness of the market … Competition operates as a discovery procedure not only by giving anyone who has the opportunity to exploit special circumstances the possibility to do so profitably, but also by conveying to the other parties the information that there is some such opportunity. It is by this conveying of information in coded form that the competitive efforts of the market game secure the utilization of widely dispersed knowledge’; comparably ground-breaking, see von Mises (1949, p. 253): ‘As soon as one abandons … rigidity of data, one finds that action must needs be (sic!) affected by every change in the data.’ Further, p. 325: ‘In an economic system in which every actor is in a position to recognize correctly the market situation with the same degree of insight, the adjustment of prices to every change in the data would be achieved at one stroke. It is impossible to imagine such uniformity in the correct cognition and appraisal of changes in data except by the intercession of superhuman agencies’.

¹¹ Fama (1970) calls a market semi-strong efficient, if publicly available information (along with historical data) is utilized for allocation decisions of investment liquidity.

¹² See also retrospectively Fama (1991, p. 1601): ‘The typical result in event studies on daily data is that, on average, stock prices seem to adjust within a day to event announcements. The result is so common that this work now devotes little space to market efficiency’.

¹³ The interval would mean in that case that all returns from the trading day t=−2 till trading day t=2 are added up.

¹⁴ See fundamentally Markowitz (1952) and building on this Sharpe (1964), Lintner (1965) and Mossin (1966).

¹⁵ For a comparison of the OLS regression with other estimation procedures, see Brown and Weinstein (1985).

¹⁶ For further references, see Section II.

¹⁷ For an overview of the historical events in Greece, see, e.g. Sauernheimer (2011).

¹⁸ Unlike the majority of studies that focus on the US, Pop and Pop (2009) analyse a Japanese sample of banks, while Schäfer et al. (2013) analyse, apart from a US-based sample, Swiss, British and German bank samples. The samples investigated normally comprise the ‘x’ largest banks, classified according to total assets. This is also the case for our study because banks with a large amount of total assets have a higher probability of receiving the G-SIB seal.

¹⁹ Aegon, Allianz, Aviva, Axa, SwissRe and Zurich Ins.

²⁰ The TBTF banks from our sample were excluded first. For more information regarding the index, see www.stoxx.com/indices/index_information.html?symbol=SXG83P

²¹ For more information regarding the index, see http://us.spindices.com/indices/equity/sp-global-1200

²² Tables A2 and A3 in Appendix 3 provide the underlying (cumulative) abnormal returns of the control sample.

²³ Not least the innumerable bank bailouts in the course of the sub-prime and, later, the financial crises might have changed the TBTF perceptions of market participants.

²⁴ For empirical results on this topic, see, e.g. Hirshleifer et al. (2009).

²⁵ However, those two alternatives are questionable, since fundamental input values may significantly change due to the event. For this reason, the estimation window should not reach into the period after the event (t > 0). Otherwise, the returns around the event would influence the estimation parameters of the expected return, which would reverse the logic of the event study. That procedure would only be supportable in the case of missing market data prior to the event (e.g., for event studies of stock emissions).

²⁶ For a short explanation of the neutralization of confounding events, see with further references McWilliams and Siegel (1997, p. 637).

²⁷ An elimination of all possible biasing events in the estimation window could easily result in a high number of identified confounding events, and lead the estimation procedure into absurdity.

²⁸ For the handling of confounding events in the event window, see in detail Foster (1980, pp. 52–7).

²⁹ See event studies listed in Table 2.

³⁰ See here and in the following Corrado (1989, pp. 387–8).

³¹ Tests for homogeneity of variances using the Levene-statistic showed significant differences in variance (see Appendix 4).