Microfoundations and Macroeconomic Implications June 12-13, 2014, - - PowerPoint PPT Presentation

International Evidence on Bank Funding Profiles and Performance: Are Banks “Overbanked”?

by Jose A. Lopez & Mark M. Spiegel

Federal Reserve Bank of San Francisco

Federal Reserve Bank of San Francisco

IMF/DNB Conference on International Banking: Microfoundations and Macroeconomic Implications

June 12-13, 2014, Amsterdam, The Netherlands Views expressed are our own and not those of the Federal Reserve

Bank Funding a Policy Concern

April 15 speech by Federal Reserve Chair Janet Yellen:

• “…. Basel Committee's first task was to strengthen bank

capital requirements …. second task was to develop new liquidity standards for global banking firms…

• (These steps) …. do not fully address the financial stability

concerns associated with short-term wholesale funding. These standards tend to focus on … firms … in isolation, rather than on the financial system as a whole.

• Federal Reserve staff are actively considering additional

measures that could address these and other residual risks in the short-term wholesale funding markets.”

Some Countries are considered “Overbanked”

• Literature finds some nations are “overbanked” [e.g.

Eichengreen and Luengnaruemitchai (2006)]

• banks account for abnormal share of

intermediation

• benefits of diversification of lending markets
• Greenspan “spare tire” argument
• role in Asian financial crisis
• Overbanking said to hurt development of Argentina and
• thers at turn of 20th century [Davis and Gallman (1978)]

Research on funding liquidity

• Funding and lending strategies impact bank

vulnerability

• Brunnermeier et al. (2011)

– Metrics such as “CoVaR” to measure contributions

• f banks to financial system vulnerability
• Bai et al. (2013)

Optimal individual bank deposit funding share?

• Extend “overbanking” idea to individual bank liabilities
• Can banks “over-rely” on deposits?
• Alternative may be longer-term funding through securities
• Current regulatory stances favor deposits

Considered stable funding sources However, stable sources may be difficult to increase Raising funds through deposits requires interest rate increases to both old and new

• Banks may benefit from securities markets presence

Difficult for “unseasoned” to raise funds in securities markets

Also potential systemic concerns

• Hard to raise external funds during crisis

– Local conditions important

• Only so much liquidity in system

– Easier to raise funds in stable system – Prudent funding and lending decisions at bank level depend on rest of system

• We investigate spillovers in funding practices

at national level

Plainview: “I drink your milkshake”

We investigate impact of funding conditions on ROA

• Use Bankscope international bank-level data
• Cross-section for 2007, also (2004-2012)
• 16,000 banks in nearly 200 countries
• Examine several funding liquidity measures
• assets only
• liabilities only
• both as measured by bank net-stable-funding

ratios (NSFRs), official BIS metric .

• Also look at national funding environments
• Same measures for all other banks in system

Data (1)

• Bankscope data

– Publicly available data on financial institutions from nearly 200 countries in 2007

• Look at wide set of firm types to characterize

varied national financial systems

– standard depository institutions, specialty finance firms, investment banks, and govt credit inst. – Focus on unconsolidated subsidiaries to identify liquidity funding profiles at national level

Data (2)

• Truncated sample

– Disproportionate share of base sample in Bankscope are US banks

• Follow Claessens and Van Horen (2014) in

limiting sample to 100 banks from countries with >4 banks

– Reduces sample to <4,000 banks from 134 countries – 90% of assets in banking system

Results

• 1. Results suggest increased deposit funding lowers ROA
• Less profitable with more standard funding sources
• Profits also decrease with system reliance on deposits
• 2. In contrast, we obtain insignificant coefficient estimates on the

share of cash in a bank asset portfolio

• Also for share of cash in national banking system
• 3. Also obtain positive coefficient on NSFR of rest of banking

system

• Better off in a more stable system
• Own NSFR enters insignificantly

Robustness checks

• Truncated sample

– Results are qualitatively similar

• Domestic and foreign sub-samples

– Most domestic variables significant with expected signs – Foreign banks do better in systems with low NSFRs

• Samples from 2004-2012

– Negative deposit results robust over time – Others change over time

Liquidity funding measures

• One side of the balance sheet

– Liabilities only: dependence on retail deposits – Assets only: cash (& equivalents) on balance sheet – We examine these normalized by total assets.

• Recently, measures incorporating both sides
• f balance sheet have been proposed

– Liquidity mismatch index [Brunnermeier, et al (2011), Bai et al. (2013)]

Formal liquidity measures

• Codified into banking regulation by the Basel

Committee of Bank Supervisors (BCBS)

• Liquidity coverage ratio (LCR)

– Amount of one-month funding sources as well as high-quality liquid assets available

• Net stable funding ratio (NSFR)

– More useful for our analysis – One year measurement horizon – Covers larger set of assets and liabilities

Net Stable Funding Ratio (NSFR)

• Ratio of funds owed relative to assets

– Items weighted by expected availability

• Example: interbank funding more likely to be

lost during market disturbance than retail deposits

– Retail deposits receive higher weighting in numerator of NSFR than interbank funding

• Government securities more easily liquidated

than commercial loans

– Commercial loans therefore receive lower weight in denominator

Calculation of Net Stable Funding Ratio (NSFR)

• NSFR for bank i in country j at time t is denoted as

where P and Q represent liabilities and assets, and the weights are based on sensitivity to roll-over risk and market liquidity

Balance sheet data

• For BCBS regulatory purposes, weight formulas are rich in

detail, reflecting maturities and currency types

• Bankscope data is contains much less detail
• Use balance sheet and weights from Federico (2013)

National banking system liquidity

• In addition to individual NSFRs, interested in

liquidity profiles of national financial systems

– National NSFR measure would obscure bank size and would lose country fixed effects

• We therefore calculate NSFR for all other

banks:

Summary statistics: liquidity measures

• Retail deposits/TA (RETDEP) 52% avg.

– Japan and developing Asia between 75% and 79% – US average ratio 61%, due to smaller US banks. – Euro core banks have a lower average of 35%.

• Cash/TA (CASH) almost 2% avg.

– Banks from Eastern Europe, developing Asia, and

• ther non-OECD 5% - 7% avg.

– Banks in Japan and euro core below 1% (US 2.1%) – Variation highlights how differences in accounting and regulatory standards lead to challenges in measuring liquidity using narrow categories

NSFR summary statistics

• NSFR has average value of 0.88

– Recall NSFR < 1 suggests stable funding – More stable funding to illiquid assets

• Developed countries had lower NSFR

– US = 0.89; Euro-core = 0.76; Japan = 0.92

• Developing countries had higher values

– Latin American banks averaged 1.07 – Other non-OECD country banks averaged 1.24

Model specification Examine effect of bank funding profiles have on bank performances:

where Yijt is ROA for firm i in country j in year t; Lijt is a vector of liquidity metrics; L-ijt is the corresponding vector of aggregate liquidity metrics for all the other firms in country j; Xit is a vector of firm characteristics;

• log total assets and leverage ratio

Zjt is a vector of country-level characteristics

• either fixed effects or Rose-Spiegel

indicators

Full sample base specification results

RETDEP

• 1.307***
• 1.230**

(0.477) (0.497) RETDEP_cn

• 5.553*
• 6.224**

(3.165) (2.784) CASH 3.353 2.777 (2.281) (1.933) CASH_cn 11.548 8.725 (8.795) (6.315) NSFR 0.067 0.054 (0.160) (0.136) NSFR_cn 1.281* 1.722** (0.690) (0.700) LEV

• 0.030***
• 0.031***
• 0.030***
• 0.030***

(0.005) (0.005) (0.006) (0.006) LNASSETS 0.023* 0.017 0.058** 0.055*** (0.013) (0.014) (0.023) (0.013) r2 0.193 0.188 0.171 0.169 N 15673 15673 15673 15673

Truncated sample base specification results

RETDEP

• 0.849***
• 0.730***

(0.161) (0.173) RETDEP_cn

• 4.546**
• 4.902***

(1.996) (1.444) CASH 0.680 0.503 (0.501) (0.504) CASH_cn 11.223*** 4.551* (2.933) (2.398) NSFR 0.202*** 0.160*** (0.043) (0.040) NSFR_cn 1.824* 1.888*** (0.912) (0.449) LEV

• 0.036***
• 0.035***
• 0.033***
• 0.034***

(0.009) (0.010) (0.011) (0.011) LNASSETS 0.029 0.016 0.019 0.027 (0.031) (0.030) (0.020) (0.021) r2 0.232 0.226 0.219 0.222 N 3846 3846 3846 3846

Domestic and foreign sub-samples

• Would think that domestic bank would differ

in reliance of local financial system for funding

– Claessens and van Horen sub-sample allows for separation of domestic and foreign firms – End up with samples of 653 global banks and 1,665 domestic banks – Results differ dramatically

Domestic bank sub-sample

RETDEP

• 0.544***
• 0.342***

(0.121) (0.113) RETDEP_cn

• 5.822*
• 3.819**

(3.099) (1.436) CASH 1.326*** 1.332*** (0.239) (0.289) CASH_cn 23.912*** 15.991*** (1.341) (1.295) NSFR 0.343*** 0.315*** (0.092) (0.080) NSFR_cn 0.593 1.253 (2.024) (1.095) LEV

• 0.036***
• 0.035***
• 0.035***
• 0.036***

(0.011) (0.011) (0.011) (0.011) LNASSETS 0.013 0.005 0.018 0.031 (0.030) (0.026) (0.021) (0.022) r2 0.339 0.327 0.327 0.334 N 1665 1665 1665 1665

Foreign bank sub-sample

RETDEP

• 0.981
• 0.648

(0.584) (0.606) RETDEP_cn

• 5.840*
• 4.471

(3.025) (3.388) CASH 0.275 0.042 (1.737) (1.776) CASH_cn

• 5.660
• 8.975

(7.525) (7.540) NSFR 0.269*** 0.181** (0.086) (0.081) NSFR_cn

• 5.721***
• 5.654***

(0.824) (0.881) LEV

• 0.035***
• 0.036***
• 0.035***
• 0.034***

(0.007) (0.008) (0.008) (0.008) LNASSETS 0.279** 0.248** 0.243** 0.263** (0.116) (0.116) (0.106) (0.103) r2 0.349 0.337 0.333 0.339 N 653 653 653 653

Additional robustness tests

• 1. Alternative time periods
• Look at individual annual results between 2004 and 2012
• Consider 2004-2006 “pre-crisis” years, 2010-2012 recovery
• 2. Country characteristics substituted for fixed effects
• Use variables identifies as influencing performances during

crisis in Rose and Spiegel (2011)

• Also consider overbanking at system level
• 3. Other bank characteristics
• Loan as % of assets
• Loan growth 2005-2007
• Retail bank (depository inst.)?

Other years (full sample)

RETDEP RETDEP_cn CASH CASH_cn NSFR NSFR_cn ROA_99_2004: full sample (WLS)

• 1.097**
• 5.717**

2.485** 6.854 0.064 1.566* (0.455) (2.202) (1.008) (5.794) (0.094) (0.869) ROA_99_2005: full sample (WLS)

• 1.327***
• 4.607**

2.253

• 7.849

0.278** 3.108*** (0.435) (1.968) (2.404) (7.695) (0.122) (0.765) ROA_99_2006: full sample (WLS)

• 1.344***
• 6.751***

2.783 22.653*** 0.246** 2.982*** (0.403) (1.706) (2.459) (8.062) (0.120) (0.872) ROA_99_2007: full sample (WLS)

• 1.307***
• 5.553*

3.353 11.548 0.067 1.281* (0.477) (3.165) (2.281) (8.795) (0.160) (0.690) ROA_99_2008: full sample (WLS)

• 0.594
• 5.921***

4.006 8.135

• 0.033

0.978* (0.418) (2.068) (3.215) (10.063) (0.152) (0.556) ROA_99_2009: full sample (WLS)

• 0.518
• 1.036

2.385** 6.598** 0.055 1.660*** (0.400) (1.499) (1.068) (2.948) (0.072) (0.437) ROA_99_2010: full sample (WLS)

• 1.001*
• 2.742
• 1.294**

0.119 0.226

• 0.441

(0.504) (2.407) (0.521) (2.302) (0.135) (1.085) ROA_99_2011: full sample (WLS)

• 1.077**
• 2.287
• 1.494***
• 6.547***

0.261** 0.367 (0.477) (2.183) (0.174) (1.170) (0.105) (1.152) ROA_99_2012: full sample (WLS)

• 0.972
• 2.182
• 0.983***

6.756*** 0.259*** 0.252 (0.591) (2.832) (0.250) (0.986) (0.073) (1.106)

Other years (truncated sample)

RETDEP RETDEP_cn CASH CASH_cn NSFR NSFR_cn ROA_99_2004: claessens (WLS)

• 0.366**
• 4.534***

0.356

• 9.469**

0.042 1.557** (0.155) (1.209) (0.474) (3.670) (0.100) (0.750) ROA_99_2005: claessens (WLS)

• 0.685***
• 4.600***
• 0.758
• 12.844**

0.309*** 2.524*** (0.164) (1.015) (1.208) (5.194) (0.070) (0.768) ROA_99_2006: claessens (WLS)

• 0.852***
• 5.375***
• 0.232

12.764*** 0.254*** 2.242** (0.138) (0.818) (0.435) (3.491) (0.037) (0.831) ROA_99_2007: claessens (WLS)

• 0.849***
• 4.546**

0.680 11.223*** 0.202*** 1.824* (0.161) (1.996) (0.501) (2.933) (0.043) (0.912) ROA_99_2008: claessens (WLS)

• 0.178
• 9.675***

0.263

• 1.160

0.230***

• 1.319*

(0.154) (1.811) (0.565) (2.394) (0.061) (0.686) ROA_99_2009: claessens (WLS)

• 0.002
• 4.650***

0.886***

• 2.550**

0.096** 1.386** (0.137) (1.183) (0.211) (1.223) (0.043) (0.611) ROA_99_2010: claessens (WLS)

• 0.416***
• 3.281**
• 0.844***

0.901 0.135***

• 1.116

(0.141) (1.348) (0.187) (1.464) (0.033) (0.988) ROA_99_2011: claessens (WLS)

• 0.559***
• 9.914***
• 1.709***
• 16.866***

0.147**

• 0.149

(0.189) (2.105) (0.239) (3.380) (0.055) (0.736) ROA_99_2012: claessens (WLS)

• 0.116

2.785**

• 0.716***

8.230*** 0.215*** 1.846 (0.180) (1.215) (0.242) (1.990) (0.065) (1.513)

With Country characteristics (no FE)

RETDEP

• 1.491***
• 1.444***

(0.451) (0.477) RETDEP_cn 0.569 0.602* (0.384) (0.339) CASH 4.558** 4.246*** (1.784) (1.413) CASH_cn

• 3.316*
• 1.339

(1.652) (1.480) NSFR 0.021 0.025 (0.188) (0.156) NSFR_cn

• 0.297
• 0.099

(0.300) (0.274) LEV

• 0.032***
• 0.032***
• 0.031***
• 0.031***

(0.004) (0.004) (0.004) (0.005) LNASSETS 0.017** 0.009 0.054* 0.046** (0.007) (0.008) (0.028) (0.019) DCB 0.003* 0.002

• 0.001
• 0.001

(0.002) (0.002) (0.001) (0.001) DCB2

• 0.000**
• 0.000
• 0.000
• 0.000

(0.000) (0.000) (0.000) (0.000) r2 0.120 0.110 0.088 0.080 N 12831 12831 12831 12831

Overbanking at the country level

With other bank characteristics added

RETDEP

• 1.279**
• 1.156**

(0.525) (0.526) RETDEP_cn

• 6.656*
• 7.147**

(3.513) (2.860) CASH 2.430 2.095 (1.665) (1.579) CASH_cn 9.342 7.470 (7.375) (4.887) NSFR 0.232*** 0.158* (0.083) (0.078) NSFR_cn 0.368 1.360** (1.190) (0.618) LEV

• 0.034***
• 0.035***
• 0.035***
• 0.034***

(0.007) (0.007) (0.007) (0.007) LNASSETS 0.013 0.007 0.043* 0.043** (0.013) (0.014) (0.024) (0.021) r2 0.229 0.223 0.208 0.208 N 14800 14800 14800 14800

Coefficients on other bank characteristics

(Full specification) RETDEP RETDEP_cn CASH CASH_cn NSFR NSFR_cn LOANS

• 0.668***
• 0.968***
• 1.004***
• 0.898***

(0.127) (0.141) (0.199) (0.205) LOANS_cn

• 5.665***
• 5.668***
• 4.572***
• 6.015***

(1.036) (1.489) (1.562) (1.409) AssetGrth

• 0.004
• 0.003
• 0.003
• 0.003

(0.003) (0.003) (0.002) (0.003) AssetGrth_cn 0.013 0.019 0.003 0.006 (0.014) (0.014) (0.015) (0.013) Retail Bank

• 0.195
• 0.190
• 0.532***
• 0.525***

(0.189) (0.200) (0.147) (0.145) RetailAssetShare_cn

• 1.016
• 0.894
• 1.623*
• 1.960**

(0.908) (0.934) (0.906) (0.822)

Conclusion (1)

• Examine large cross-country micro data set to

investigate impact of bank funding

• Profitability decreasing in share of deposits

– Indicates could do better by relying less on deposits

• Also find evidence of spillovers across financial

system

– Profitability also decreasing in system deposit reliance for funding – Positive spillovers from greater system stability

Conclusion (2)

• Substantive differences between domestic

and foreign banks

– Domestic largely enter with expected signs – Foreign profits declining in local NSFR

• Profitability decreasing in share of deposits

– Could do better by relying less on deposits

• Macroeconomic implications

– Point estimates indicate 1 SD increase in deposits associated with 25 bp decline in GDP growth – 1 SD ↑ in NSFR with 7.5 bp ↑ in GDP growth

Caveat: Mixed results over time

• Pre-crisis results very different than post

–Coefficient on deposits becomes insignificant over time –CASH becomes significant

• Lose positive system NSFR spillovers
• Raises question of proper policy goal

– Policies designed to stabilize during normal times,

• r mitigate most severe downturns?

– Uncertain about frequency of steep downturns

RETDEP

RETDEP mean sd count EU Core .3497126 .2290198 2611 EU Periphery .4697421 .2378624 857 Eastern & Other Europe .5371743 .3028582 996 Latin America .5252667 .2419059 436 Japan .7900082 .1845817 587 Developing Asia .7478993 .2109334 500 United States .6105338 .3254791 8311 Total .5209982 .2910633 15673

CASH

CASH mean sd count EU Core .0077556 .0149325 2611 EU Periphery .0228859 .0571485 857 Eastern & Other Europe .0601656 .0623852 996 Latin America .0354 .05407 436 Japan .0103962 .0087194 587 Developing Asia .0454455 .0506985 500 United States .0216869 .0156935 8311 Total .0195007 .0347588 15673

NSFR

NSFR mean sd count EU Core .7610826 .5583812 2611 EU Periphery .8982984 .3998646 857 Eastern & Other Europe 1.050263 .408758 996 Latin America 1.067772 .5466903 436 Japan .9211721 .1413169 587 Developing Asia .9942192 .2572209 500 United States .8855665 .3097704 8311 Total .8794351 .4653529 15673

Approx of NSFR