Global Bank Complexity and Balance Sheet Management Linda S. Goldberg

Transcription

1 Global Bank Complexity and Balance Sheet Management Linda S. Goldberg ACPR Banque de France Conference: Monitoring Large and Complex Institutions, December 2017 The views expressed in this presentation are those of the author and do not necessarily represent those of the Federal Reserve Bank of New York or the Federal Reserve System

2 Traditional portrayal of a bank Liquid assets Loans Other assets Deposits Wholesale funding Equity Balance sheet defines business of banking Understanding business of banking means understanding how banks manage their balance sheet Balance sheet also defines the boundary of the banking firm 2

3 Banking systems have evolved to support a mix of conglomerate types Complex conglomerate Simple conglomerate Liquid assets Deposits Liquid assets Deposits Loans Other assets Wholesale funding Equity Loans Other assets Wholesale funding Equity Possibly generates different risks and balance sheet management Bank in simple conglomerate is run according to canonical business model of banking. All that matters is own balance sheet. Bank in complex conglomerate. Resolution issues. Plus, own balance sheet management and structure possibly depend also on BHC business model and organizational complexity. 3

4 This presentation in two parts Facts about BHC complexity US Bank Complexity: What has Changed? with April Meehl One direction of why we care: relationship between BHC complexity and bank balance sheet management Organizational Complexity and Balance Sheet Management in Global Banks with Nicola Cetorelli (NBER 22169) 4

5 Measuring complexity Uses organizational structure entities, so Counts -based [Herring and Carmassi 2010; Cetorelli and Goldberg 2014] US BHCs based on US National Information Center (NIC) database Foreign Banking Organizations (FBOs) in US, Bankscope Ownership module (March 2013) Observe full ownership tree of BHC Count the bank and non-bank affiliates Include only those 50% owned (or 25 % owned) Use industry information, geography information Some balance sheet info mainly for banks, BHC 5

6 US Bank Complexity: What has Changed? with April Meehl for internal use only

7 Around 800 Large US BHCs: Asset Size & Numbers

8 10 largest US BHCs account for most of assets

9 Complexity build up pre-crisis, as reflected in Subsidiary counts. Across U.S. BHCs, growth in mid 2000s driven by increase in non-bank subsidiaries. Vast expansion in BHCs organizational footprint Additions in non-bank financials, non-financials Source: Cetorelli and Stern,

10 Complexity is mainly within the largest of the largest BHCs, with most of 800+ BHCs relatively simple. Size and complexity are correlated but distinct. 2006Q2 2016Q2 Percentiles calculated by BHC assets 0-25p 26-50P 51-75P P 0-25p 26-50P 51-75P P Number of Subsidiaries % Commercial Bank Number of 4 Digit NAICS Number of Countries

11 Within the top 50 largest US BHCs, post crisis declines in complexity. 2006Q2 2016Q2 Percentiles calculated by BHC assets Number of Subsidiaries 0-25p 26-50P 51-75P P p 26-50P 51-75P P % Commercial Bank Number of 4 Digit NAICS Number of Countries

12 Top 50 Industry Breakdown: large post-crisis increase in subs in funds, portfolio management; declines in broker dealers, insurance and other intermediaries With total subsidiary counts by type and total nonbank subsidiaries

13 Organizational Complexity and Balance Sheet Management in Global Banks with Nicola Cetorelli (NBER 22169) for internal use only

14 Does the complexity of the parent conglomerate matter for business conduct of its bank affiliates? May be it doesn t: Separate entities run as if independent of one another, In aggregate the conglomerates achieve diversification benefits. [open question on evidence for this] Or may be it does: Conglomerates form when synergies and complementarities across affiliated entities exist. Diversification benefits, efficiencies. Optimization of synergies result in subordinating business model of individual entities to interest of broader organization [Rajan and Zingales (2000, 2001a, 2001b), Stein 2002] Negatives around agency problems could have adverse effects. 14

15 Conjecture: Lending (and balance sheet) of banks in complex families is less sensitive to funding shocks, more to the BHC structure. Test: Use an ideal funding shock, November FDIC proposed rule changed assessment fee for deposit insurance U.S. banks with access to deposit insurance paid a fee proportional to total amount of insured deposit liabilities New rule: fee assessed over total bank liabilities taxes banks reliance on less stable wholesale funding U.S. banks shifted away from wholesale funding sources. Deposit to asset ratio rose by 7 pp (66 to 73%) in 7 qtrs. Rates on overnight wholesale funding dropped 5-10 bp (Kreicher, McCauley and McGuire, 2013) 15

16 Testing. Measure impact of FDIC-driven funding shock on U.S. branches of FBOs U.S. branches of FBOs do NOT have access to deposit insurance. Exclusive reliance on wholesale funding On the receiving end of a positive funding shock that is Unrelated to the branches themselves Unrelated to the structure of foreign financial conglomerates From a mandate from Dodd Frank Act, thus not driven by market conditions at implementation, conditions abroad Branches of FBOs are organizational simpletons. Terminal ends with no complexity of their own Economically significant banking units (15-20% US banking assets) 16

17 Implementation Compare growth in lending before and after FDIC ruling, between branches of simple family and branches of complex families (140) Y = α + β size + β counts + β X + β X counts + ε i i, HH i, HH i i i, HH i 0 pre 1 2 pre 3 pre ( post pre)/ pre Conjecture: ββ 11 negative and significant Explore various hypotheses about the composition of subsidiaries of the BHC to inform alternative hypotheses Alternative loan metrics, and other balance sheet variables Control for branch or BHC size; Define pre period around the FDIC announcement 4 quarters averages (skip the quarter during implementation), Pre: 2010q1-2010q4 Post: 2011q2-2012q1 17

18 18

19 Results show statistical and economic significance of complexity for lending response to funding shock The lending sensitivity of an FBO branch decreases by about 30% for each quintile increase in family s subsidiary count, after conditioning on size Branches of complex conglomerates are structurally different. Larger, greater access to external funding, and investment priority more toward support to the family internal funding flows prioritized more liquid branch balance sheet maintained Robustness to alternative Count constructions using industry and geography information 19

20 Implications Especially for large BHCs, diverse business models as reflected in the structure of the organizations Complexity matters, and is distinct from organizational size Organizational decisions of financial conglomerates impact banks as a vehicle of funding shock transmission Effects magnitude of transmission Effects direction too (from bank balance sheet to balance sheet of related parties, bank and nonbank entities) Novel channel of regulatory transmission (curbing complexity of financial conglomerates has impact on commercial banking activities) 20

21 Thank you. for internal use only

22 Table 2: Foreign Banking Organizations with US Branches, by Highholder region (2010:Q4) Highholder Region Number of Highholders Highholder Data Highholder Total Assets ($Billions) Total Affiliate Count Median Affiliate Count Number of U.S. Branches US Branch Data Branch Total Assets ($Billions) Branch Median Assets ($Billions) Euro Area 27 21, United Kingdom 4 6, Japan 8 5, China 5 6, Switzerland 2 2, Canada 7 2, Other Americas 18 1, Other Asia 37 3, Other 24 5, All ,973 28, ,

23 High Diversity in the 132 FBOs in the United States, with branches,

24 Results. C&I Lending Growth Table 3 C&I Lending Growth (1) (2) (3) (4) (5) Log Count *** *** ** *** *** (0.024) (0.029) (0.033) (0.023) (0.033) Log HH Assets (0.033) (0.037) Log Branch Assets (0.029) (0.033) Liquid Assets Share (0.200) (0.199) Constant 0.425*** 0.320** *** (0.113) (0.152) (0.553) (0.127) (0.579) Observations Adjusted R * p<0.10 ** p<0.05 *** p<

25 Results. C&I lending Growth Dropping branches with C&I share<5% Table 4 C&I Lending Growth (Branches where C&I lending >= 5%) (1) (2) (3) (4) (5) Log Count *** *** ** *** ** (0.022) (0.031) (0.033) (0.022) (0.035) Log HH Assets (0.034) (0.036) Log Branch Assets (0.029) (0.032) Liquid Assets Share (0.207) (0.209) Constant 0.507*** 0.481*** *** (0.104) (0.141) (0.546) (0.120) (0.568) Observations Adjusted R * p<0.10 ** p<0.05 *** p<

26 Results. Total lending Growth Table 5 Total Lending Growth (1) (2) (3) (4) (5) Log Count ** *** ** ** (0.023) (0.028) (0.034) (0.022) (0.033) Log HH Assets * (0.029) (0.032) Log Branch Assets (0.027) (0.031) Liquid Assets Share (0.186) (0.183) Constant 0.358*** 0.251* *** (0.103) (0.133) (0.491) (0.108) (0.521) Observations Adjusted R * p<0.10 ** p<0.05 *** p<

27 What explains this result? Ex Ante Balance Sheet Composition of US Branches of Foreign Conglomerates Shares Ln Branch Assets Funding Liquid Assets Total Loans Net Due Diff Ln HH Assets.96***.71***.04*** *** -.05** ***.08*** Ln Count.34***.07*** ***.07** - Constant 16.43*** 16.29***.20***.17**.59***.59***.52***.55*** -.86***.89*** Adj R * p<0.10 ** p<0.05 *** p<0.01 Branches of complex conglomerates structurally different. Larger, greater access to external funding and investment priority more toward support to the family 27

28 Table 7 Robustness to Geography C+I Loans C+I Loans Total Loans Total Loans Log Count -0.13*** -0.13*** -0.09*** -0.10*** Log HH Assets * Log Branch Assets Liquid Assets Share US+Peg Share -1.20** -0.76* USPegXLnCount 0.34*** 0.24** Tax+Sec Share TaxSecXLnCount 0.56* 0.38 Constant Observations Adjusted R * p<0.10 ** p<0.05 *** p<

29 Table 8 Time Series Panels of Balance Sheet Growth, Complexity, and Geography (1) (2) (3) (4) (5) Log Count -0.01** ** -0.01** Log HH Assets (lag 1Q) 0.02*** 0.03** 0.02*** 0.02*** 0.02*** Log Branch Assets (lag 1Q) -0.01* -0.02** ** Liquid Asset Share (lag 1Q) *** 0.06* GDP Growth (lag 1Q) Interest Rate (lag 1Q) 0.00** Constant *** Observations Adjusted R Cumulative Growth Effect after FDIC Shock Q1 to Q * -.107** -.083* -.084* -.077* Q1 to Q ** -.163** -.118* -.121** -.111* Control Variables USPeg and Tax/Sec x x x x x Time Dummies x x x x x Family Dummies x Region Dummies x GIIPS Dummies x * p<0.10 ** p<0.05 *** p<

30 Lending sensitivity may be lower because: Family complexity yields overall better access to wholesale funding. At the limit, if funding access unconstrained, lending should have no sensitivity at all to funding shocks Organizational choice to always prioritize internal funding flows Organizational choice toward maintaining a more liquid balance sheet 30

31 Evidence of an overall funding advantage Funding C&I Loans Total Loans Internal lending Ln Count *** *** *** *** (0.034) (0.029) (0.028) (0.031) Ln HH Assets *** ** (0.032) (0.037) (0.029) (0.033) (0.025) (0.029) (0.026) (0.038) Constant 0.268* 0.339** 0.285* 0.320** * (0.159) (0.148) (0.161) (0.152) (0.134) (0.133) (0.133) (0.130) Observations Adjusted R^ * p<0.10 ** p<0.05 *** p<0.01 Note: All dependent variables are winsorized at +/-100% growth. Branches of more complex family already have less constrained access to wholesale funding. An improvement in access has less of an impact 31

32 Robustness Robustness to alternative Count constructions using industry and geography information Time-series specifications for 2009Q1 to 2013Q4 Document this is not just a trend differential. Actual break after FDIC ruling Includes time varying controls for FBO parent country interest rates and demand conditions (not driven by stuff going on in countries of origin) 32

33 Conclusions Conglomerate complexity matters for the business model of a bank affiliate Banks in complex families exhibit lower investment sensitivity to funding shocks The balance sheet of banks in complex families structured to prioritize internal lending Better access to funding sources 33