Abstract
The financial crisis of 2007-2009 revealed that the corporate complexity of most of the Global Systemically Important Banks (G-SIBs) presented a formidable obstacle to any plausible orderly resolution of these institutions. This paper documents the extent of this complexity making use of an historical time series, developed by the authors, that shows the evolution of the number of majority-owned subsidiaries of G-SIBs over time. After a very significant increase in complexity before the crisis and until 2011, this trend may be reversing, possibly in response to regulatory and market pressures on banks since then. Nonetheless the reduction in complexity has been uneven across institutions and may not persist. The econometric analysis of this new set of panel data produces two key results with relevant policy implications: first, the relationship found in previous studies between the number of subsidiaries and bank size loses significance when time effects are introduced; second, large mergers and acquisitions are a key driver of complexity and their effect remains significant even when time effects are considered.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
Undoubtedly, many banks also adopted the holding company structure because of the opportunities that it provided for double-leverage. By borrowing at the holding company level and down-streaming the proceeds as equity in affiliated banks, the holding company could satisfy capital adequacy guidelines at affiliated banks while simultaneously achieving its desired degree of leverage in the consolidated financial structure. Subsequent regulatory constraints on bank holding companies and increasing emphasis on consolidated regulation and supervision have reduced this flexibility.
The BCBS (2010, p. 29) cautions that the line between a branch and a subsidiary is often quite blurred in practice noting that “in some jurisdictions branches…may have to meet many of the requirements normally imposed on locally-incorporated subsidiaries, while in other jurisdictions subsidiaries may function much more like branches integrated into the parent”. Luciano and Wihlborg (2013) make a similar point.
The total number of controlled subsidiaries reported in FED/NIC database is often larger than the number we have reported in Table 1 based on the Bankscope database. One reason may be that the FED/NIC uses a lower ownership threshold (25 %) to define “control” than we do (50 %). In addition, the FED/NIC data include two other categories of subsidiaries: (1) entities that meet additional “reportability” criteria and (2) entities that are “of interest” to the Fed (but these do not necessarily meet the definition of control). For an analysis of corporate structures of US bank holding companies based on NIC/FED data see Avraham et al. (2012).
For G-SIBs that completed more than one significant acquisition, we have adjusted the ratio to reflect the additional large acquisitions. Each new transaction is represented by dividing the cumulative sum of the subsidiaries of all acquired firms in the year before the transaction by the number of subsidiaries of the acquirer in the year preceding the very first significant deal. In some cases where data on subsidiaries of acquired firms were not available in Bankscope, we have used FED/NIC or SEC data.
Of course, market pressures may also cause G-SIBs to simplify their corporate structures. Klein and Saidenberg (2005) found that bank holding companies face market pressures to simplify their structures. Laeven and Levine (2007) adopt a different approach, but also find a conglomerate discount in large complex financial institutions. They identify agency problems and diseconomies of scope that outweigh economies of scope as probable causes. These findings indicate that market pressures may reinforce regulatory pressures for banks to simplify their corporate structures, but it remains a puzzle why banks chose to form conglomerates just as non-financial firms were abandoning the model and emphasizing focus.
Since at any point Bankscope makes current data on bank corporate structures available, but not the historical data, we have also used Bankscope historical discs to get the information and data on majority-owned subsidiaries for past years, going back to 2002. The latter is the earliest year for which data on subsidiaries reported by Bankscope appear to have the same coverage as current and most recent data and for which numbers on subsidiaries can be consistently analyzed and enables us to use the 50.01 % filter for majority-owned subsidiaries for all G-SIBs and for all years from 2002 to 2013.
These broader criteria for inclusion of legal entities in the NIC/FED organizational hierarchy are likely to produce a significant impact. Specifically, we have noticed that the NIC/FED list of subsidiaries is particularly long for two of the largest firms, Goldman Sachs and Morgan Stanley, respectively with 14,814 and 8,998 entities as of 31 December 2013. The entities for which the relationship is “of interest to the Federal Reserve” are outside of the scope of Regulation Y definition of control and do not need to fall under such definition: Large Merchant Banking Investments are included in this category, and this might explain the very high number of entities for Goldman Sachs and Morgan Stanley.
References
Avraham D, Selvaggi P, Vickery JI (2012) A structural view of U.S. Bank holding companies. Econ Pol Rev 18(2):65–81
Basel Committee on Banking Supervision (BCBS) (2010) Report and Recommendations of the Cross-border Bank Resolution Group. Basel, March
Basel Committee on Banking Supervision (BCBS) (2013) Global systemically important banks: updated assessment methodology and the higher loss absorbency requirement. Basel, July
Basel Committee on Banking Supervision (BCBS) (2014) The G-SIB assessment methodology – score calculation. Basel, November
Calomiris C, Eisenbeis R, Litan R (2011) Financial crisis in the US and beyond. In Litan R (ed), The World in Crisis: Insights from Six Shadow Financial Regulatory Committees From Around the World, FIC Press, Wharton Financial Institutions Center, University of Pennsylvania, November
Carmassi J, Herring RJ (2015) Corporate Structures, Transparency and Resolvability of Global Systemically Important Banks. Systemic Risk Council, Washington DC, January 27
Cetorelli N, Goldberg LS (2014) Measures of Complexity of Global Banks. Economic Policy Review, Volume 20, Number 2, Special Issue on Large and Complex Banks, March 31
Dermine J (2006) European banking integration: Don’t Put the cart before the horse. Fin Mar Ins Ins 15:57–106
Federal Deposit Insurance Corporation (FDIC) (2011) The Orderly Liquidation of Lehman Brothers Holdings Inc. Under the Dodd-Frank Act. FDIC Quarterly, Volume 5, No. 2.
Federal Deposit Insurance Corporation (FDIC) and Federal Reserve Board (FRB) (2014) Agencies Provide Feedback on Second Round Resolution Plans of ‘First-Wave’ Filers. Joint Press Release August 5
Federal Reserve Board (FRB) (2015) Regulatory Capital Rules: Implementation of Risk-Based Capital Surcharges for Global Systemically Important Bank Holding Companies; Final Rule. Federal Register, Vol. 80, No. 157, 12 CFR Parts 208 and 217, August 14
Financial Stability Board (FSB) (2013) 2013 Update of groups of global systemically important banks (G-SIBs). November 11
Glasserman P, Loudis B (2015) A Comparison of U.S. and International Global Systemically Important Banks. OFR Brief Series, 15–07, August 4
Goodman L, Lucas D, Fabozzi F (2007) Financial innovations and the shaping of capital markets: the case of CDOs. J Altern Invest 10(1):62–70
Herring RJ (2003) International Financial Conglomerates: Implications for National Insolvency Regimes. In Kaufman G (ed.), Market Discipline and Banking: Theory and Evidence, Elsevier, pp. 99–130
Herring RJ (2014) The challenge of resolving cross-border financial institutions. Yale J Reg 31(3):853–881
Herring RJ, Carmassi J (2010) The corporate structure of international financial conglomerates: complexity and its implications for safety & soundness. In: Berger AN, Molyneux P, Wilson JOS (eds) Oxford handbook of banking, 1st edn. Oxford University Press, New York
Herring RJ, Carmassi J (2015) Complexity and systemic risk: what’s changed since the crisis? In: Berger AN, Molyneux P, Wilson JOS (eds) Oxford handbook of banking, 2nd edn. Oxford University Press, New York
Herring RJ, Santomero AM (1990) The corporate structure of financial conglomerates. Journal of Financial Services Research, Volume 4, Issue 4, December, pp. 471–497
Hume, N (2011) Protium’s acid reflux (updated). FT Alphaville, April 28, available at http://ftalphaville.ft.com/2011/04/28/555611/protiums-acid-reflux/
Kapur E (2015) The next Lehman bankruptcy. In: Scott KE, Jackson TH, Taylor J (eds) Making failure feasible – How bankruptcy reform Can End “Too Big to fail”. Hoover Institution Press, Stanford
Klein P, Saidenberg M (2005) Organizational Structure and the Diversification Discount: Evidence from Commercial Banking. CORI Working Paper 2005–06, University of Missouri, April
Laeven L, Levine R (2007) Is there a diversification discount in financial conglomerates? J Financ Econ 85:331–67
Laeven L, Ratnovski L, Tong H (2014) Bank Size and Systemic Risk. IMF Staff Discussion Note, SDN/14/04, May
Luciano E, Wihlborg C (2013) The Organization of Bank Affiliates: A Theoretical Perspective on Risk and Efficiency. International Centre for Economic Research, Working Paper No. 6/2013, May
Lumsdaine RL, Rockmore DN, Foti N, Leibon G, Farmer JD (2015) The Intrafirm Complexity of Systemically Important Financial Institutions. Available at SSRN: http://ssrn.com/abstract=2604166, May 8
Acknowledgments
We are grateful to the Systemic Risk Council for support of this work although our results do not necessarily reflect its views or those of its members.
The authors are grateful to Nicola Cetorelli, David Mayes, Haluk Unal, an anonymous referee and participants at the IBEFA 2016 ASSA Session on Banking Risk and Complexity for comments on an earlier draft.
Author information
Authors and Affiliations
Corresponding author
Appendix A
Appendix A
1.1 Data on subsidiaries: methodology issues and alternative sources
The aspect of the complexity of G-SIBs that should be easiest to measure is the number of subsidiaries. Unfortunately, it is not. Inconsistent and opaque definitions and gaps in disclosures present formidable hurdles. We lack an official data source with comprehensive and consistent data for all G-SIBs. Data provided by regulators often differ because they have differing objectives and statutory obligations. Sometimes the differences can be quite significant and raise troubling questions about consistency. Figures 2 and 3 illustrate this problem using data from two different US regulatory agencies, the Securities and Exchange Commission and the Federal Reserve Board, and the private vendor we use for this article. As shown by Fig. 2, in some cases two different sources may provide quite similar figures, with a third source giving very different results. In other cases, there may be significant differences among multiple sources (see Fig. 3). Although we have cross-checked the data with every available source, in the end we have relied on Bankscope because of the comprehensiveness of its international coverage, the consistency of its methodology, and the granularity of the detail.
Number of subsidiaries of Citigroup. This figure shows the number of subsidiaries of Citigroup (as of year-end 2013) reported in the FED/NIC database, SEC filings and Bankscope
Number of subsidiaries of JPMorgan Chase & Co. This figure shows the number of subsidiaries of JPMorgan Chase & Co. (as of year-end 2013) reported in the FED/NIC database, SEC filings and Bankscope
Inconsistencies across sources are likely to be produced by differences in regulatory objectives and criteria used to identify subsidiaries. Nonetheless, despite the inconsistencies across sources of data and uncertainties that may remain about the precise number of subsidiaries for each G-SIB, the totals are high – with hundreds or even thousands of legal entities for each institution. Of course, a simple count of subsidiaries is not a complete indicator of the complexity of G-SIBs, but the number of subsidiaries does indicate a key challenge to an orderly resolution.
Our analysis of bank corporate structures largely relies on the Bankscope database, which provides a clear and quite simple criterion to identify majority-owned subsidiaries. Bankscope reports the parent’s extent of ownership in each legal entity tracing the ownership share along the entire chain of control. We identify a subsidiary as controlled (“ultimately owned”, in Bankscope terminology), if the parent owns at least 50.01 % of the voting shares through each level of the chain of control. This is a very conservative standard that includes both directly and indirectly controlled subsidiaries. While other sources are available for institutions headquartered in the United States, most G-SIBs are headquartered elsewhere and Bankscope is the only source that provides consistent and detailed coverage of all banking groups worldwide. In addition to information on the number of subsidiaries, Bankscope includes data on location, assets, operating income and number of employees of subsidiaries and a classification of subsidiaries by industry codes. Moreover, Bankscope also offers comprehensive data for all G-SIBs over time. Our data cover the period 2002–2013.Footnote 6
Alternative sources, such as the Federal Reserve/National Information Center data and SEC filings, do not provide all these details. For example, FED/NIC data on organizational structures, based on confidential forms submitted by banks (e.g. form FR Y-10), are focused on US banks and on US operations of non-US banks. This omits a considerable amount of useful information on the non-US corporate structures of European and Asian G-SIBs. Moreover, publicly available FED/NIC data do not report the financials of subsidiaries, and the categorization of the business/industry is much less specific than the Bankscope classifications. Finally, the FED/NIC data are only available in pdf files, which are less amenable to statistical analysis with standard software.
The main criterion for the inclusion of legal entities in the FED/NIC organizational hierarchy is a definition of control under Regulation, which is essentially a 25 % control (Bank Holding Companies and Change in Bank Control, 12 CFR 225.2 (e)). However, additional entities that meet FR Y-10/10F “reportability criteria” are included, as well as entities for which the relationship is “of interest to the Federal Reserve”. The inclusion of the latter two categories makes it very difficult for an external analyst to compare these data with data available from other sources.Footnote 7 But the FED/NIC data also have strong points: first, they provide information on the level of ownership and hierarchy of control for each subsidiary (which, however, is available also in Bankscope); second, it appears to be the only database that provides corporate structure trees at a given time (Bankscope only reports current data in the version available to subscribers, and the historical discs refer to an unspecified day during each month).
The exhibits of SEC filings reporting the list of subsidiaries (in 10-K for US firms, in 20-F for non-US firms) only indicate the name and jurisdiction of subsidiaries, as of year-end. The minimum percentage of control for each may be indicated in an explanatory note preceding the list. Unfortunately, the SEC lists provide no information on the financial profile of the subsidiary or its principal line of business. Moreover, unlike Bankscope and FED/NIC data, the SEC filings do not provide an indication of the ownership/hierarchical structure (i.e. the chain of control from the top of the group to each of the subsidiaries). Most troubling, the SEC permits reporting firms to omit “non-significant” subsidiaries (Regulation S-K, 17 CFR 229.601 (b) (21) (ii) and Regulation S-X 17 CFR 210.1-02 (w)). This ill-defined category might include a large number of subsidiaries that could pose an obstacle to an orderly resolution. Finally, the SEC does not provide lists of subsidiaries in a format that may be easily analyzed with standard statistical software.
Annual reports of banks and other official documents published by banks on their websites often include a list of subsidiaries. However, the criteria used to build these lists may vary significantly across institutions and across countries, which will inevitably give rise to inconsistencies. Moreover, relevant information related to each of the subsidiaries such as the financial profile, sector of activity, and ownership level, are seldom included. Even the public sections of living wills submitted to the US regulators by US and non-US banking groups do not include a full list of subsidiaries, but focus only on material entities.
Other sources (e.g. the SNL database, or the new website www.opencorporates.com) do report bank corporate structures information, albeit with differing degrees of detail, but they generally provide less comprehensive information, and they do not seem as helpful as other sources for our statistical analysis.
For all of these reasons, we have chosen to rely on Bankscope as the primary source of information on the corporate structures of G-SIBs. Nonetheless, we used other sources as well to evaluate the consistency of information across sources. Some internal experts in various G-SIBs emphasize that Bankscope data may often be flawed; however, in the absence of stronger, more consistent disclosure regulations these are the best publicly available data for our purposes.
1.2 Appendix B
1.3 Appendix C
Rights and permissions
About this article
Cite this article
Carmassi, J., Herring, R. The Corporate Complexity of Global Systemically Important Banks. J Financ Serv Res 49, 175–201 (2016). https://doi.org/10.1007/s10693-016-0251-4
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10693-016-0251-4