120 likes | 298 Views
Location Decisions of Foreign Banks and Competitive Advantage. Stijn Claessens and Neeltje Van Horen Discussion by Jan Svejnar 13 th Dubrovnik Conference June 2007. Overall Assessment. Nice study Important issues Innovative way to collect and use data Careful analysis
E N D
Location Decisions of Foreign Banks and Competitive Advantage Stijn Claessens and Neeltje Van Horen Discussion by Jan Svejnar 13th Dubrovnik Conference June 2007
Overall Assessment • Nice study • Important issues • Innovative way to collect and use data • Careful analysis • Plausible results/recommendations • Also • Data strengths and limitations • Questions stemming from the analysis
Main Findings and Conclusions • Competitive advantage (similarity in institutional environments between host and source countries) increases flow of banks from source to host countries • This effect is stronger than that of (absolute) differences in institutional environments • There may be benefits from increased cross-border banking among developing countries; but there are also risks
Key Ideas (Hypotheses) • Foreign banks have been entering other markets because of deregulation, technological advances and globalization • Do banks enter countries where institutional familiarity provides them with a “competitive advantage” over other banks? • Existing literature implicitly assumes that a bank’s location decision is made independently of • the location decisions of other banks • a bank considering all countries for investment • => This study: bank takes its competitive position relative to all other banks into account
Data • Newly collected data on ownership of banks and bank FDI entry and exit over 1995-2005 • 137 countries covered • 50% or more foreign ownership = bank foreign owned • Source country = country with highest share ownership in the foreign owned bank • Direct rather than ultimate ownership used
Basic Data Patterns (Fig. 2) • Banks from high income countries tend to invest in high or lower income countries • Banks from middle and low income countries tend to invest in similar income level countries (banks from lower middle income countries tend to invest in upper middle income countries) • Institutional quality in the host and source country displays slightly positive, statistically significant correlation
Empirical Methodology • Key hypothesis: • Banks from countries with relatively weak (strong) institutions compared to their competitors will enter countries with relatively weak (strong) institutions • Need a variable that captures the quality of institutions of the host and source country, but also that of competitors • Construct country-pairs for all possible host-source country combinations => 9,957 pairs for each year
Empirical Methodology • Dependent variable = 1996-2005 change in the number of foreign banks from source country j present in in host country i • Claim that exit is not taken into account, but the above measure includes it (= entry minus exit)
Empirical Methodology • Measure of competitive advantage of a bank from j wrt i at t • CompAdvijt = (Instsourcejt – InstcompSt) (Insthostit - InstcompHt) InstcompSt = GDP-weighted average of institutional quality of all possible source countries InstcompHt = GDP-weighted average of institutional quality of all possible host countries Except for large countries (US), InstcompSt and InstcompHt do not vary much across countries and are similar, so that • CompAdvijt ~ (InstsourceAdjjt) (InsthostAdjit) • = product of relative institutional quality of host and source country pair (could be a sum, difference, …) • Not a concept of strategic interaction
Empirical Methodology • Regress • dForpresenceij = α0 + α1dCompadvij + α2dInsthosti + … controls + αiDdiffInst • Dependent variable is 1996-2005 change; explanatory variables are 1995-2004 change – too much overlap (use subperiods and lag across them) • Test of α1 versus αi , with α1 winning (product v. difference of institutional quality => question of what exactly is measured) • With CompAdvijt ~ (InstsourceAdjjt) (InsthostAdjit), the regression constrains coefficient on both components to be identical - α1 - (could relax and test what drives the result: InstsourceAdj or InsthostAdj ?)
Other ideas about Methodology • Work more on IV-Tobit (Table 3) – not clear that can use 1995-99 dCompadv as an IV for 2000-05 dCompadv -- 1995-99 dCompadv may be used by banks in deciding whether to enter in 2000-05; relevant explanatory variable for second stage? • GDP-weighted estimation because more banks may potentially engage in cross-border investment in large countries (could use a more direct measure, e.g., number of banks in source country minus number of banks already in host country • Alternatively, estimate P( bank j enters country i | not yet there) • Develop and estimate a model that takes into account more explicitly strategic interaction among competing banks (Instcomp is a “passive” measure of average institutional situation in other countries) • HA: relevant competition matters => try to capture how unsaturated the market is (opening up; relaxation of capital controls – e.g., ECA accounts for more than 50% of increase in number of foreign banks [T 1]) => refine & interact Entryres
Overall Assessment • Important topic • New, plausible interpretations of observed outcomes • Additional work can be done to increase credibility of results • Product versus difference of source and host country institutional environment