Foreign Banks and the Vienna Initiative Turning Sinners into Saints? Ralph De Haas (EBRD) Yevgeniya Korniyenko (Bank of England) Elena Loukoianova (IMF) Alexander Pivovarsky (EBRD) Banks and Government in Globalized Markets OeNB, Vienna, September 12 th -13 th 2013
Introduction Emerging Europe: exceptional level of international banking integration Before 2008, foreign banks contributed to efficient and stable banking More competition, efficiency, know-how transfer, and access to credit (Fries and Taci 2005; Havrylchyk and Jurzyk 2011, Giannetti and Ongena 2012) Stable credit source during host country crises (De Haas and Van Lelyveld 2006) The 2008-09 shock originated in foreign banks home countries and led to a sharp deleveraging both at home and abroad (Cetorelli and Goldberg 2011; Giannetti and Laeven 2012; De Haas and Van Horen 2013)
Introduction Funding constraints and potentially biased government interventions raised concerns about foreign banks commitment to emerging Europe Lack of coordination could cause a run on the region Large exchange-rate fluctuations and BoP problems Sharp decline in lending with dire consequences for firms and households Vienna Initiative (VI): 1. Ad hoc coordination mechanism to overcome collective action problems 2. Guarantee macroeconomic stability in Emerging Europe by closing external financing gaps through bail-outs and bail-ins 3. Support banking groups to prevent a credit crunch
The Vienna Initiative in a nutshell September 2008: Collapse Lehman Brothers Autumn 2008: Initial talks between Austrian and Italian banks and Austrian supervisors. EC, host and home supervisors, IFIs (EBRD, EIB, WB), IMF, and 17 bank groups join to expand private-public coordination February 2009: Joint IFI Action Plan. 24.5 billion support to 17 banking groups March 2009: Emergency Summit EU leaders: National support packages should not lead to restrictions on lending by subsidiaries in Emerging Europe 2009: IMF-EU BoP support: Bosnia, Hungary, Latvia, Romania, Serbia ( bail out ) Bail in : foreign banks sign commitment letters to pledge support for subsidiaries in specific countries and to roll over debt (e.g. Swedish banks pledge to roll over 80 per cent of exposures to Latvia) Worries about negative spill-over effects to other Emerging European countries
Aim of this paper Contribute to the evaluation of the impact of the VI No uncoordinated withdrawal of banks from the region VI was successful stricto sensu at the macro level...... although we do not know the counterfactual No bank-level evidence yet on the impact of the Vienna Initiative on credit growth 5
Research questions 1. Were foreign banks relatively stable lenders during this crisis? 2. Were foreign banks that were part of the VI more stable lenders than those that remained outside the VI 3. Within the same country, were foreign banks that signed a commitment letter more stable lenders than banks that did not? 4. Did commitment letters lead to negative spillovers to other countries? 6
Contribution to the literature 1. Multinational banks as conduits of foreign shocks and absorbers of local shocks (Morgan, Rime and Strahan 2004; Peek and Rosengren 1997, 2000) 2. Can government intervention dampen or exacerbate intrabank shock transmission? Negative: financial protectionism (Rose and Wieladek 2012) Positive: Vienna Initiative? 3. Catalytic effect of IMF funding. Limited IMF funding can convince private creditors to roll over their commitments, thus filling external funding gaps Empirical evidence sobering (Corsetti and Roubini 2004) Korea (1997) and Brazil (1999) relatively successful: roll-overs were neither fully voluntary nor uncoordinated
Data Bank-level dataset with detailed information on about 500 banks across Emerging Europe Annual balance sheet and income statement data for 1999-2011 (BankScope) We add: Dynamic ownership info: state, foreign, and private domestic banks Information on banks participation in Vienna Initiative and on government support Macroeconomic covariates (IFS-IMF)
Data BOSNIA-HERZEGOVINA HUNGARY LATVIA ROMANIA SERBIA Support Support Bank name Bank name Bank name Bank name Bank name Support Support Support Parent participates in VI Parent does not participate in V Raiffeisen Bank Y UniCredit Bank Y DnB Nord Bank N Alpha Bank Romania Y Société Générale Bank Serbia Y Intesa SanPaolo Bank Y Raiffeisen Bank Y SEB Bank N Banc Post Y UniCredit Bank Y UniCredit Bank Y Erste Bank Hungary Y Swedbank N Banca Romaneasca Y Piraeus Bank Beograd Y HypoAlpe-Adria-Bank Y K&H Bank Y Piraeus Bank Romania Y Eurobank EFG Y ZepterKomercBank BanjaLuka Y CIB Bank Y Volksbank Romania Y Volksbank Serbia Y Volksbank BH Y Magyar Takarekszövetkezeti Bank Y Banca Comerciala Romana Y Alpha Bank Serbia Y NLB Bank N UniCredit Tiriac Bank Y Vojvodjanska Bank Y BRD - Groupe Societe Generale Y Banca Intesa Y Raiffeisen Bank Y HypoAple-Adria Bank Y Raiffeisen Bank Y Turkish Ziraat Bank Bosnia N KDB Bank N HVB Bank Latvia Y Egnatia Bank N NLB Bank N ProCredit Bank N Volksbank Y ProCredit Bank N ProCredit Bank N Bosna Bank International N Commerzbank Y ABN Amro Bank Y Erste Bank Y Banco Popolare Y OTP Bank N OTP Bank N Deutsche Bank N San Paolo IMI Bank Y Marfin Bank N Fundamenta-Lakaskassza N Banca de Creditsi Dezvoltare Romexterra N Moskovska Bank N Allianz Bank N Emporiki Bank N Credit Agricole Y Budapest Hitel-ésFejleszési Bank Y Findomestic Bank Y KBC Bank Y
Data Bank ownership Avg. No. Banks 1999-2011 2000-2007 2008 2009 2010 2011 (1) (2) (3) (4) (5) (6) Vienna countries Domestic state 12 0.40 0.13 0.15 0.00 0.03 Domestic private 32 0.42 0.19 0.31 0.03 0.00 Foreign Table 1 Credit growth across emerging Europe Vienna 24 0.57 0.15 0.01-0.05-0.04 Non-Vienna 30 0.43 0.26 0.01 0.00-0.01 Total 99 Non-Vienna countries Domestic state 14 0.26 0.07 0.21 0.09 0.07 Domestic private 57 0.44 0.22 0.20 0.11 0.10 Foreign 97 0.43 0.16 0.10-0.05 0.01 Total 167 Average annual credit growth
Empirical strategy Panel regressions: focus on annual gross credit growth rate of individual banks before and during the crisis Impact of: Ownership structure VI variables: Vienna country, Vienna parent, Parent signed here, Parent signed elsewhere Control for standard time-varying bank and country characteristics Include bank and year FE Lijt = 1 + γ 1Ownijt + γ 2Crisist + γ 3Ownijt Crisist + γ 4 α X + µ + η + ε ijt t ij ijt
Empirical strategy Cross-sectional regressions for foreign banks in VI countries only Zoom in on Vienna Initiative and role commitment letters Only countries that borrow from various multinational banks Use country dummies to control for credit demand L α γ L γ VI γ X + η + ε ij = + ij + ij + 1 1,2004 07 2 3 ij, 2007 j ij Diff-in-diff regressions
Empirical strategy Upfront: This is not a controlled trial with a randomized treatment... Mitigate concerns about selection bias through: 1. Quantification of omitted variables bias 2. Instrumental variables approach 3. Propensity score matching exercise
Panel results: bank ownership Overall common trends: credit growth started to slow in 2008, temporarily halted in 2009, but continued to decline in 2010-11 Foreign banks reduced credit growth more in 2009-10 when compared to domestic banks Table 2 Bank ownership and credit growth during the crisis All Private Domestic State Foreign (1) (2) (3) (4) Year 2008-11.195** -19.662** -6.250-14.641*** (0.024) (0.011) (0.132) (0.006) Year 2009-6.334-18.006-3.545-10.960 (0.526) (0.237) (0.705) (0.260) Year 2010-22.836*** -28.603** -25.521*** -28.746*** (0.009) (0.018) (0.001) (0.000) Year 2011-21.958** -25.806** -22.128*** -24.380*** (0.019) (0.032) (0.003) (0.003) State * Year 2008-0.910 (0.843) State * Year 2009 8.938* (0.066) State * Year 2010-0.166 (0.981) State * Year 2011 1.712 (0.841) Foreign * Year 2008-5.564 (0.107) Foreign * Year 2009-12.429*** (0.007) Foreign * Year 2010-11.444*** (0.001) Foreign * Year 2011-7.534 (0.153)
Panel results: the VI Regressions for foreign bank subsidiaries only (incl. year and bank FE as well as subsidiary, parent and macro covariates) Subsidiaries of VI parents outperform non-vi subsidiaries, even in non-vi countries: if anything, positive spill-overs from signing elsewhere In VI countries, subsidiaries of signing parents had a 10.1% higher credit growth in 2009 relative to subsidiaries of non-participating parents Role unobservable time-varying factors in VI participation decision? Quantification omitted variables bias cf. Altonji et al. 2005; Bellow and Miguel 2008 Table 3 The Vienna Initiative and credit growth All countries Non- Vienna countries Vienna countries (1) (2) (3a) (3b) Vienna parent*2008 6.336 3.122 7.873 (0.122) (0.547) (0.166) Vienna parent*2009 15.350*** 15.117** 12.307* (0.008) (0.014) (0.057) Vienna parent*2010 15.465*** 15.173** 12.506** (0.004) (0.025) (0.046) Vienna parent*2011 14.599** 18.616** -3.668 (0.018) (0.046) (0.428) Parent signed here*2008 1.919 (0.838) Parent signed here*2009 10.061** (0.035) Parent signed here*2010 9.38 (0.148) Parent signed here*2011-5.336 (0.246) Parent signed elsewhere*2008 37.187* (0.078) Parent signed elsewhere*2009 19.856 (0.276) Parent signed elsewhere*2010 24.885 (0.183) Parent signed elsewhere*2011 2.454 (0.912) No. observations 794 524 270 270 R-squared 0.655 0.671 0.713 0.723 Year FE Yes Yes Yes Yes Bank FE Yes Yes Yes Yes R=β F /(β R- β F ) 3.380 2.507 3.127 1.296
Table 4 Robustness tests Base 7+ years GMM PCSE Asset Net loans growth growth (1) (2) (3) (4) (5) (6) Parent signed here*2008 1.919 0.591 8.578 8.876 2.333-0.713 (0.838) (0.954) (0.361) 0.326 (0.816) (0.947) Parent signed here*2009 10.061** 7.909* 16.530* 17.978** 15.010** 9.025** (0.035) (0.059) (0.084) 0.018 (0.016) (0.015) Parent signed here*2010 9.380 6.367 14.820* 12.837* 4.359 6.608 (0.148) (0.411) (0.060) 0.09 (0.526) (0.171) Parent signed here*2011-5.336-4.637 11.388 8.047-1.609-6.001 (0.246) (0.433) (0.137) 0.315 (0.831) (0.180) Parent signed elsewhere*2008 37.187* 34.241* 23.548 29.199*** 15.796*** 35.730* (0.078) (0.083) (0.328) 0.008 (0.007) (0.078) Parent signed elsewhere*2009 19.856 19.661 8.423 2.524 1.683 17.256 (0.276) (0.288) (0.445) 0.792 (0.915) (0.404) Parent signed elsewhere*2010 24.885 25.156 13.847 14.581 0.332 26.698 (0.183) (0.239) (0.443) 0.133 (0.986) (0.196) Parent signed elsewhere*2011 2.454 7.474-2.781-8.974-7.121 3.285 (0.912) (0.726) (0.764) 0.398 (0.775) (0.891) No. observations 270 249 268 270 270 264 No. banks 54 49 54 54 54 54 R-squared 0.723 0.736-0.71 0.772 0.725 Bank controls Yes Yes Yes Yes Yes Yes Parent controls Yes Yes Yes Yes Yes Yes M acroeconomic controls Yes Yes Yes Yes Yes Yes Year FE Yes Yes Yes No Yes Yes Bank FE Yes Yes No No Yes Yes
Cross-sectional results Table 5 The Vienna Initiative, commitment letters and bank lending Credit growth (1) (2) (3) (4) (5) (6) (7) (8) Vienna parent -15.086* -23.769* -10.783** -0.152-12.249-19.331* -4.773 0.063 (0.079) (0.061) (0.014) (0.596) (0.344) (0.077) (0.439) (0.872) Parent signed here 23.746* 38.143** 15.272** 0.574* 21.015** 32.592** 12.120** 0.157 (0.056) (0.029) (0.018) (0.100) (0.016) (0.043) (0.037) (0.516) Pre-crisis average annual credit (asset) growth -0.356** -0.401*** -0.310*** -0.010** -0.231-0.232** -0.280** -0.010* (0.013) (0.000) (0.006) (0.020) (0.265) (0.017) (0.045) (0.080) Support 7.787 7.71 8.674 0.119 2.192 2.247 7.595-0.091 (0.335) (0.212) (0.314) (0.396) (0.682) (0.729) (0.174) (0.644) No. of observations 54 54 54 54 51 51 51 51 R-squared 0.603 0.574 0.646 0.511 0.701 0.678 0.732 0.522 M odel OLS OLS OLS LPM OLS OLS OLS LPM M odel period 2009 2009 2009-2011 2009 2009 2009 2009-2011 2009 F-test (p-values) (0.129) (0.090) (0.019) (0.189) (0.032) (0.112) (0.067) (0.787) Subsidiary Controls Yes Yes Yes Yes Yes Yes Yes Yes Parent Controls Yes Yes Yes Yes Yes Yes Yes Yes Country FE Yes Yes Yes Yes Yes Yes Yes Yes First S tage Results for Instrumental Variables Regressions M arket Share other Vienna Letter - f 0.091* 0.087* (0.054) (0.093) M arket Share other Vienna Letter - f ^ 2-0.001** -0.001** (0.026) (0.044) F-statistics for instrument relevance 4.680 4.280 F-test (p-value) (0.016) (0.023) Hansen test (p-value) (0.745) (0.578) Asset growth
Diff-in-diff results Difference-in-Difference Results All banks in emerging Europe Foreign banks in non-vi countries Foreign banks in VI countries [1] [2] [3] [4] Crisis -38.756*** -44.024*** -74.110*** -74.325*** (0.000) (0.000) (0.000) (0.000) Foreign bank 2.345 (0.531) Crisis*Foreign bank -8.676** (0.050) Vienna Parent 5.883-0.241 (0.336) (0.981) Crisis*Vienna Parent 1.421 20.051** (0.846) (0.040) Parent signed here 9.769 (0.509) Crisis*Parent signed here 18.773* (0.092) R2 0.538 0.615 0.761 0.767 N 506 160 92 82
Propensity-score matching Table 6 - Propensity Score Matching - Second Stage Results Vienna Parent CEB & S EE Vienna Parent non-vienna countries Method (1) (2) Nearest Neighbor M atching (n=1) 17.764*** 20.159** t-value (3.159) (2.552) N 146 91 Nearest Neighbor M atching (n=1) 17.764** 20.159* Bootstrap z-value (2.321) (1.793) N 146 91 Nearest Neighbor M atching (n=2) 14.317*** 18.262*** t-value (2.693) (2.708) N 146 91 Nearest Neighbor M atching (n=2) 14.317* 18.262* Bootstrap z-value (1.823) (1.711) N 146 91 Kernel M atching Epanechnikov 13.564* 19.274* Bootstrap z-statistic (1.836) (1.912) N 146 91 Kernel M atching Normal 11.087* 16.732* Bootstrap z-statistic (1.709) (1.916) N 146 91
Summary 1. Foreign banks transmitted crisis via their subsidiaries in 2009 Unlike in earlier, host-country crises, foreign banks were now a less stable source of credit Banking integration cuts both ways: a source of strength in case of internal shocks but also imports external shocks 2. VI foreign banks were relatively stable sources of credit, in particular in countries where they signed commitment letters 3. No evidence for negative spill-overs due to VI commitment letters
Policy implications Bail in-bail out approach with semi-voluntary and monitored rollover commitments and central coordination can help resolve collective action problems during a crisis Prevention: Host-home coordination between supervisors is key More centralised supervision of international banking groups? Nordic-Baltic supervisory agreement may set example Centralised supervisory authority for large banking groups?
Thank you