How do banking analysts behave around unanticipated news? Evidence from operational risk event announcements

Abstract

We study earnings per share (EPS) forecast revision and accuracy of banking analysts around operational risk event announcements in U.S. banks. We find that first announcements of operational risk events are more informative than their settlement announcements. Optimistic banking analysts revise their forecasts downward more aggressively around operational risk disclosures, thereby improving forecast accuracy. Career concerns of banking analysts cause an upward bias in forecast revision and deterioration in forecast accuracy only if the potential employer is a systemically important bank (SIB). We find consistent evidence linking competition among banking analysts with optimistic and inaccurate forecasts, which is consistent with analysts seeking to use inflated forecasts to curry favour and attract businesses to their brokerage house around the time of operational risk disclosures. Global settlement has no favourable impact on analyst forecast accuracy around operational risk event announcements. We find evidence supporting a materiality threshold of $10 million for the informativeness of operational risk event announcements in SIBs. Overall, our results shed light on optimism bias in banking analyst behaviour upon the arrival of unanticipated news.

KEYWORDS:

• Operational risk
• unanticipated news
• banking analyst
• forecast revision and accuracy
• career concerns
• analyst competition
• optimism bias

JEL Classifications:

• G14
• G21
• G41

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 An exception is Item 2.02.

2 U.S. banks are required to disclose particular data items on their operational losses in FR Y-14Q filings. However, these regulatory filings are not publicly available.

3 We understand that our sample is not big according to the common standards in the finance literature. However, due to strict sample selection criteria (see Table 1), relatively small samples are common in the operational risk literature. For example, Perry and de Fontnouvelle (2005) have used a global sample of only 115 operational risk events in the period 1974 – 2004. Gillet et al. (2010) analysed only 103 operational risk events in U.S. financial firms in the period 1990 – 2004. Other studies (e.g. Fiordelisi et al. 2013, 2014) have used quite large samples (215 and 430 events, respectively) due to a low materiality threshold ($1 million) and wider coverage (U.S. and European banks).

4 Settlement announcements usually arrive sometime after the first announcement and are hence expected. This is because there is usually a wealth of information publicly disclosed at the market or privately discovered by analysts between the first announcement and settlement announcement of the operational risk event.

5 Optimistic forecasts (also known as optimistically biased forecasts or upward biased forecasts) are EPS forecasts whose error is positive (i.e. EPS forecast is greater than the actual EPS) as computed from the I/B/E/S Detail History file.

6 A systemically important bank (SIB) is a bank whose total assets exceed $250 billion.

7 We focus our analysis on U.S. banks to mitigate concerns about the regulatory, institutional, and cultural environments of different countries driving analyst forecast revisions.

8 We have considered using a materiality threshold that is relative to bank size. However, most of the operational loss to market value ratios in our sample are lower than 1% and show little meaningful variation, thus making it practically difficult to examine different relative materiality thresholds. The same difficulties hold if we use the operational loss to total assets ratio.

9 Analyst reaction ratio is computed as the number of analysts who revise their EPS forecast (either upward or downward) during the event window (-5,+5) divided by the total number of analysts following the bank on day -6.

10 Apart from Appendixes I and II, all other tables 1 – 10 have the same order for the main results and online appendix. For example, the descriptive statistics are reported in Table 2 (the full sample), Table A.2 (the severe sample), and Table B.2 (the minor sample), and so forth.

11 Since our sample also includes operational risk events with no loss amount disclosed at first announcements, we expect a slowness on the part of analysts to respond. Extant literature on operational risk uses longer event windows including (-10, +10) and (-20, +20). However, we use only (-5, +5) to avoid losing too many observations due to the overlap of operational risk event announcements with other announcements such as 10-Qs, 10-Ks, and 8-Ks.

12 As Rubin et al. (2017) explain, because the analyst’s information set consists of emerging forecasts of other analysts as time evolves, along with more private and public information released, the drop in an analyst forecast error is expected to be greater if a longer time has elapsed from the analyst’s previous forecast.

13 We do not include bank fixed-effects and analyst fixed effects in all regressions to avoid multicollinearity because: a) some bank-level and analyst-level variables are either time-invariant, such as Potential Employer, or show little if no variation over time, such as Optimistic Analyst, and b) 40% of banks and 30% of analysts appear only once in our sample. We address the issue of within correlation at the event-level by using robust standard errors clustered at the event-level in all regressions.

14 ‘Bad News’ include only negative CARs and ‘Good News’ include only positive CARs. There are no zero CARs in our final sample. It is noteworthy that equity markets may react favourably to some operational risk event announcements for several reasons such as: 1) the bank has recognised the operational risk event quickly and promised to take a prompt corrective action or 2) the disclosed operational loss amount in the settlement announcement is lower than previously expected.

15 Our sample does not include any zero CARs.

16 The Economic Growth, Regulatory Relief and Consumer Protection Act became effective on May 24, 2018. It has increased the threshold of SIFIs from $50billion(previouslyimposedbytheDodd-FrankActof2010)to$250 billion in total assets.