Cross-border banking, parents bank performance and subsidiaries credit extensions: evidence from the CESEE region

Transcription

1 Cross-border banking, parents bank performance and subsidiaries credit extensions: evidence from the CESEE region L U C A G A T T I N I A N D A N G E L I K I Z A G O R I S I O U S T A R E B E I F I N A N C E D P R O J E C T U N I V E R S I T Y T U T O R : P R O F. N I K O L A O S G E O R G A N T Z I S

2 Structure Background Dataset Methodology & Results Concluding remarks

3 Background During the early years of transition foreign banks have speeded up economic convergence & raised standards of living in the CESEE region Result: exceptionally high market shares for foreign banks in the region Relatively large literature on cross border banking & credit growth Foreign banks beneficial as well as detrimental for credit growth & systemic efficiency contradicting views in the literature Pros enhance competition in domestic banking markets provide financial services with lower costs boost the efficiency of resource allocation mitigate credit crunches in host countries by bringing in additional liquidity from their headquarters in the home country (De Haas and Lelyveld, 2010) Cons lack information on creditworthiness of small businesses tend to have higher interest margins and profitability than domestic banks lead domestic banking markets to lower competition are more inclined to reversals of capital and credits during difficult times

4 Background - Some parent-subsidiary channels not fully explored in the literature: for example how the health of the parent affects the subsidiary bank, how the home business cycle affects the host s bank behavior - CESEE region not investigated in full in the literature - In addition the continuously changing nature of the ownership structure, parent banks strategic positioning and effects of ultimate ownership have not been accounted in full at times - It is imperative to revisit the determinants of credit growth in the region - The first part of our project revisits the mere feature of being a foreign owned bank and its consequences on the propensity to expand credit at an individual bank level - The second explores whether the nature of the home-host relationship is doomed to be unchangeable or - The third digs into the actual home-host linkages and exploits parents and subsidiaries balance sheet data including balance sheet health s indicators - Next, the analysis is further expanded to detect if/how parents risky behavior can affect subsidiaries credit extensions - Finally we research the effects of home/host business cycle on credit growth

5 Dataset Essential for our research: A database tracking the ownership of banks over time Connecting subsidiary and parent characteristics as well as macro environment features These type of databases does not exist in nature and the literature has very few attempts to build such a dataset All papers/datasets are identifying the direct owner of each bank - relatively easier shortcut to examine parent-subsidiary relationship; It lacks of precision and does not allow for the identification of the actual and final relationship We constructed a new and unique database: It identifies the Ultimate owner of each bank across time and ownership changes. This is the only way to capture the most relevant capital flows, interconnections and cross border economic and financial relationships among owners and owned entities

6 Dataset How the definition of the Owner (Direct vs Ultimate) impacts the database 1 st Example 2 nd Example Other databases Our database Other databases Our database German Financial Group H owns Arab individuals established a Croatian Bank Z owns Croatian Bank Z, which owns Fund headquartered in Bosnia owns fund headquartered in Bosnia, which owns Bosnian Bank U Bosnian Bank U Bosnian Bank B Bosnian Bank B therefore Bank U is Croatian owned Bank U is German owned Bank B is domestic Bank B is foreign

7 Dataset The different kinds of foreign ownership in our database Sample 1. The majority shareholder is a foreigner domestic banks with big dispersion of shares and the majority shareholder being a foreigner plus 2. and 3. 68% of Observations 2. A foreigner owns the absolute majority of shares over 50% of shares held by foreigners plus 3. 66% of Observations 3. A financial group owns at least 50% of shares 57 % of Observations

8 percent percent percent Dataset Foreign owned banks are 57% of total observations but on average represent much higher market shares in most markets Aggregate Market Shares of Foreign banks Bulgaria Romania Latvia foreign banks domestic banks foreign banks domestic banks foreign banks domestic banks Source: Own Calculations

9 % Dataset Our database and the contribution of foreign banks to the expansion of Banking Sectors in countries of the region Growth of Banking assets - foreign and domestic banks Source: Own Calculations Hungary Czech Republic Bulgaria Foreign Banks Domestic Banks Overall growth Banking theory predicts that each individual bank extents credit on the grounds of its own financial characteristics and macroeconomic conditions Is it possible that ownership plays a role too? How, ceteris paribus, foreign banks extend credit differently than local banks?

10 Dataset Descriptive statistics - We cover the banking sectors of 18 CESEE (host) countries which are connected with 27 home countries - Time frame: We include financial intermediaries with more than 1% Mk. Share in at least 1 year

11 1 st Model Does foreign ownership per se have an effect on individual bank credit growth? We estimate a dynamic credit growth model adding a control for ownership type. Since members of foreign financial groups constitute the vast majority of the subset of foreign ownership, results are driven by this subcategory and focus on it we also tested for it and results confirm the choice The preferred econometric method is a system GMM- Difference GMM and Fixed effects are also reported

12 1 st Model Estimation results Members of financial groups consistently have a propensity to extend credit more than other banks Why? Differences in levels? Independent sample t-tests yes Can those differences cause the determinants of credit growth to behave differently across the sub-samples? Chow test yes Solution? We cannot simply estimate this model separately for domestic banks and foreign ones Sub-sample of members of financial groups-> possible cross-sectional dependence due to sharing the same Parent bank We test this with Crossed Random Effects Model.

13 2 nd Model Controlling for the participation in a specific financial group and year effects Do member of financial groups resemble each other? If so in what respect? Answers to those questions will allow us to better specify our core model The structure of our data indicate the need for a crossed random effects model We estimate 2 such models & test if All banks are affected similarly by events associated with years. Members to the same financial group behave similarly In the 2 nd Model we also test if some events during year t might be more detrimental or beneficial to a certain financial group than others

14 2 nd Model Controlling for the participation in a specific financial group and year effects The selection criteria & a likelihood ratio test indicate the year- group interaction as significant. Which is the resemblance between A Banks operating in the same year but belong to different financial groups B Banks belonging to the same financial group but operate across different years C Banks operating in the same year and belonging to the same financial group 16% 19% 42% Therefore members of financial groups, given the year of operation, resemble very much each other in terms of credit growth. Why? They all face the same budget constrains imposed at consolidated level and set by the conglomerate s financial conditions.

15 3 rd Model Credit growth by Members of foreign financial groups and parent bank characteristics Given the results above, once we constrain our analysis to banks members of financial groups, we must control for time variant parent bank characteristics. We control for parent bank funding conditions, which we consider more relevant for subsidiary s credit growth, so we control for parent bank profitability, solvency, liquidity and asset quality. Finally we control for growth and demand conditions in the home country. Our Dynamic Model becomes:

16 3 rd Model Credit growth by Members of foreign financial groups and parent bank characteristics Banks manage their capital at consolidated level which serves to fund multinational operations. Once the parent bank becomes more risk averse, domestic credit growth declines (De Haas and Van Lelyveld, ). Asset quality its deterioration is detrimental to credit growth at consolidated and at subsidiary level. However at consolidated level its impact is multiplied. Parent asset quality of significance in previous studies. Our approximation of the variable more consistent results & of larger magnitude Liquidity and profitability at subsidiary level have an effect on domestic credit growth. On the other hand capital and asset quality are more meaningful at the parent level. 1 De Haas, R., Van Lelyveld, I., Internal capital market and lending by multinational bank subsidiaries. Journal of Financial Intermediation 19 (1), 1 25.

17 Parent Bank Risky behavior and Credit Growth Past Excessive loan growth as predictor for current loan growth Loan impairment charges at parent level consistently deteriorate credit growth. Loan impairments = bad project / financing decisions Aggressive credit expansion more bad projects financed Is it detrimental to future credit growth? We investigate this both at parent level and subsidiary level. Bank/ Parent bank engaged in excessive loan growth. FE diff GMM sys GMM 1 year before 3.139** (1.52) (1.61) (1.63) 2 years before (1.07) (1.20) (1.12) 3 years before (1.09) (1.25) (1.15) 4 years before ** ** (1.15) (1.36) (1.15) 1 year before (1.45) (2.04) (1.77) 2 years before * (1.57) (1.81) (1.52) 3 years before * (1.45) (1.46) (1.60) 4 years before * (1.34) (1.24) (1.29) Excessive loan growth has some significance at both levels BUT not consistently Past reports trace risky behavior in the region: Thinly capitalized banks inclined to excessive credit growth How this impacts their future lending capacity?

18 Risky behavior and Credit Growth Past Excessive loan growth & excessive drop in Solvency rate as predictor for current loan growth Credit expansion combined with deterioration of capital ratio always predicts a decrease in future credit growth Parent banks and subsidiaries differ only in the horizon the risks are materialized Consequences of risky behavior at Parent bank level materialize sooner This is the only study that explores risky behavior taking into account the parentsubsidiary nexus

19 The effect of business cycle on credit growth Models above & previous literature: Economic growth impacts (+) credit growth Our contribution explicitly test the effects of Business Cycle phases Host country Business Cycle Foreign banks accelerate economic growth. They are way more sensitive during positive phase rather than negative ones. Home country Business Cycle Only positive phases of the cycle have positive spillovers for the host countries.

20 Concluding remarks Novel dataset capturing ultimate ownership in a dynamic framework Proved ownership per se matters for subsidiaries credit extensions Foreign owned subsidiaries respond less to own balance sheet features while time-varying parents features are a fundamental driver in subsidiaries credit extensions Home and host macroeconomic factors affect host credit growth Profitability mostly at subsidiary level, liquidity (at subsidiary and parent level) and leverage at parent level influence credit growth A time varying measure of asset quality (at subsidiary and even more so at parent bank level) is a relevant driver of subs capabilities to extend credit Excessive risk taking behavior at both parent and subsidiary level constraints credit growth, more than past excessive credit growth Tentatively we detect an asymmetric influence of business cycles atypical upturns (at host and home level) propel credit growth