Spatial dependence in the technical efficiency of local banks

Cooperative banks primarily compete with one another because they target niche markets that large typically ignore. The current study shows in this competitive environment, the connection between financial intermediaries affects operational efficiency of small banks. findings indicate capitalization, diversification strategies, funding costs, liquidity, credit quality, and risk bank neighbors have spillover effects on technical efficiency. Thus, networks trigger a cascading effect demands attention stakeholders. Los bancos cooperativos compiten principalmente entre sí porque se dirigen nichos de mercado que los grandes suelen ignorar. El presente estudio muestra que, en este entorno competitivo, la conexión intermediarios financieros afecta eficiencia operativa pequeños. resultados indican capitalización, las estrategias diversificación, costos financiación, liquidez, calidad del crédito y el riesgo vecinos tienen efectos indirectos sobre técnica. Así, redes bancarias desencadenan un efecto cascada exige atención inversores bancos. ??????(cooperative bank)??????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????????? extensive use relationship lending procedure loan issuance offers market edge over other banks, particularly when dealing micro enterprises household customers (e.g., Hauswald & Marquez, 2003; Schenone, 2010; Yosano Nakaoka, 2019). Furthermore, benefit from their customer relationships to collect set soft information (Agarwal et al., 2018; Baas Schrooten, 2006; D'Aurizio 2015; Ergungor Moulton, 2014; Garcia-Appendini, 2011; Petersen, 2004) process private inherently qualitative (see, among others, Cassar Liberti, 2018). In comparison, larger having stronger hierarchy greater distance collectors decision-makers, tend implement rationing policies toward opaque borrowers (Liberti Indeed, regarding kind can be only extrapolated personal interactions and, even it is collected, challenging transmit such highest level management 2012; Berger Udell, 1995; Stein, 2003). Moreover, adopt practices must wear additional costs establishing adequate internal controls (Cerqueiro 2009). According stream literature, banks' proximity consumers may an source advantage, burden borne by creditors institutions rises as bank's office increases (Almazan, 2002; Chiappori Gehrig, 1998; Sussman Zeira, 1995). Once has been established, maintain advantages acquired soft, non-transferable “opaque firms” (Elsas, 2005; Hasan 2021; Ongena Smith, 2001; Petersen Rajan, 1994). From perspective, local are advantaged hierarchical for households (Berger 2005, 2017; Degl'Innocenti Kysucky Norden, Liberti Mian, 2008; 2002). summary, enjoy location-based cost thanks technique, achieve power respect (Coccorese Santucci, Against background, dominate segment monopolistic competition condition impair Technical Efficiency (TE). Conversely, cluster close discipline turn improve Despite awareness possible rents roots, almost no research conducted determine whether these affected spatial interdependence. Accordingly, article draws externalities induced characteristics neighboring For example, reasonable infer cooperative face stiff capital market, regulation restricts channels through which raise capital. This implies make effort attract shareholders mutual As result, if direct competitor's risk-adjusted capitalization higher, similar equity structure. Alternatively, offset better capitalized advantage providing services at lower price, reduces both production Similarly, geographic occur result income or strategies. On hand, unconventional sources gain suffer losses result. However, scheme hold exploit economies scale capability endowments pursue strategies increase productivity (Wu 2020; Iosifidi 2021). Equally, significant revenue generated traditional activity nearby rival eroding, contend same market. homogeneous where provide comparable services, rate induces strategy reorganize avoid losing share. would suffer. regard, focus core banking activities incurring inefficient alternative interest-bearing assets (Bernini Brighi, McKee Kagan, 2016; Van Cuong 2020). Finally, negative externality TE across network area. Specifically, harmed increasing exposure, results higher screening monitoring DeYoung, 1997; Ding Sickles, 2019), depleting confidence. Similar inefficiencies caused liquidity (Altunbas 2007; Fernandes potential inefficiency, very few studies examined industry. novel strand literature examines dependence measures profiles (Ding well (Glass Kenjegalieva, 2019; Glass extent, Tabak al. (2013) evidence location key factor determining extent. Burgstaller (2020), positioned farther away closest competitors less efficient due lack competition. although Zhao (2020) show presence geographical effects, contiguous levels regional environment performance Chinese urban commercial authors do not find lag independent variables. general, idea co-movements scores should strengthened light statement discussed above. detail, present aims further Italian (Credit Banks, CCBs). 1 we apply robust two-step methodology first estimates score via nonparametric frontier technique then regresses group explanatory covariates terms. second step involves application Simar Wilson's (2007) method estimation variables; unlike methodologies, allows control assumption separability inclusion determinants Skevas Grashuis, addition standard environmental variables influence score, our empirical model incorporates lags account effects: specifically, how corporate CCBs i t h efficiency, vice versa. Our main experience pressure create struggle compete. Therefore, more effective competitors. needs explored many affect score. existence all estimated models. lagged factors shown drivers proceeds follows. Section 2 overviews literature. 3 describes methodology. 4 introduces data used specifications. test truncated bootstrap regression reported 5. 6 reports robustness check using estimate TE. 7 provides some concluding remarks. top five themes analysis (Ahmad vast based either parametric techniques (for reviews, see Ahmad Aiello Bonanno, 2018b; Bhatia 2018), but investigated role spillovers within framework. A body stochastic (SFA) More geographically weighted panel framework unobserved US approach precise estimates, compares competing facing economic shocks. Their weight matrix, elements defined Euclidean connections units specified bandwidth. focuses impact coefficients, overlooks treatment determinants. Interestingly, shortcomings nonspatial studies, applied intermediary operating exclusively Bonanno (2018a) follow focused multilevel models rather than approaches investigate clusters provincial-level characteristics, concentration, Sickles (2019) introducing error term resulting product residual vector weighting matrix (SWM). specification captures heterogeneity. adopted SWM binary contiguity take value two network—namely, j ten least banks—and zero otherwise. positive multiplier terms Tier ratio, NPL, cost-efficiency performance. shock nonperforming loans, 1, cause aggregate shift network's impaired risk-weighted respectively. Kenjegalieva develop decomposing total context carry out bank-level data. assume will diminish increases, thus inverse headquarters indirect allocative captured Durbin specification, integrates regressors' lags. return medium function analysis, dependent related elasticity output. considering translog includes represented areas indicates sharing common boundaries. distinctive feature does weights level; is, there SWM. Instead, ( n × n) row each region, taking region borders 0 With discovered loans provided had beneficial institutions. differs examining determinant different industry developed country development heterogeneity, characterized big players controlling share microcredit intermediaries, clients. mainly interaction, therefore probably also effect, stronger. Another interesting considers envelope (DEA) nonfinancial firms. Some single-step relate proximity, without investigating Fusco DEA takes estimating decision-making unit (DMU) array determinants, including capturing (Skevas Oude Lansink, yet analysis. represents attempt co-movement technique. section methodological study, follows Algorithm 1. stages. stage, obtain scores; detailed Appendix (A.1.1). regressed covariates. case, Grashuis Lansink include (i.e., regressor multiplied SWM), X (SLX) specification. Tobler's (1970) law geography states socioeconomic phenomena exhibit single one. necessitates econometric approaches, bias estimations co-movement. deterministic, nonnegative w : , = … expresses various types connectivity, values principal diagonal equal ? discard self-influence (Cliff Ord, 1968; Kelejian Prucha, 2010). Of kinds SWMs, 5 transformation Anselin's (2003) k-nearest (k-NN) metrics, nonzero entities orthodromic coordinates units. k-NN overcome biases attitudes people living rural mountain willingness travel compared citizens cities. addresses issue mountainous isolated areas, linear meaningless. symmetric, x's neighbor, y's neighbor necessarily x He, 2004). To address disparity, symmetric nearest (k-sNN) used, replacing W k ) new called k-sNN s ), computed + ? . Each becomes transformation. Consequently, observations neighbors. We choose six-nearest-neighbor (k=6) correspond median number queen constructed Thiessen polygon tessellation point locations. latter several artificial linkages densities pattern. polygons notion area (under highly simplifying assumptions), proxy “spatial” any given likely interact with. criterion consider actual distance, corrected unrealistically long intersecting band 73 km, shortest neighborless isolates). intersect matrices d them critical distance. combination fully located beyond 73-km boundary. row-standardized so rows add up fashion, normalize outside unit, making meaningful comparisons basis matrices. i, ? belongs N ? ^ distribution left censored ? Z ? WZ ?; adjusted ? ˜ ? ; parameters ? define maximum likelihood procedure. complete process, original combined confidence intervals errors 10 sample 230 2011 2020 2300 observations. 11 subsample following criteria: (1) availability years interest, 12 (2) exclusion six 13 Orbis Bank Focus (Bureau van Dijk). 14 geospatial geo-referencing CCB coordinates. Figure connecting criteria design three SWMs. figure confirms clustered supports A1 depicts indicators association (LISA) statistics Anselin (1995), backs influenced level. intermediation (Sealey Jr Lindley, 1977), multiple inputs deposits, labor, capital, etc.). accordance previous (inter alia, Asimakopoulos Barth 2013; Diallo, Drake Kulasekaran Shaffer, 2002), particular outputs earning assets) 15 four funds, staff expenses, fixed assets, loss provisions). last input (loan provisions) accounts risk/loan quality (Aggelopoulos Georgopoulos, Fukuyama Matousek, 2017). 16 expressed monetary deflated CPI. year dummies time D Although orientations estimated, profit maximization purpose pursuing objectives earlier Coccorese Ferri, Harimaya Ozaki, 2021) focusing input-orientation findings. Table descriptive calculate ); ). summarizes reveal contains 45 other. included refer funding, soundness, risk, quality. 17 gross (such favorable (Barth Bitar 2020), measures, adequacy report associations Minviel Ben Bouheni, document ratios (either 2) strongly degrade strong ratio (CAP) prudential management, turn, Wu examine (Income div) financing structure (Funding overlooked Income div net interest signifies reliance (Foos Fraser retail-oriented variable Because endowments, urged broaden (Iosifidi fortiori diseconomies concentrating conventional fields operation diversifying into nontraditional operations (McKee contingent size: noninterest whereas dramatically consequence, indefinable priori. exposed asset-liability mismatch deposits suitable measure capture effect. If bank, CCB, collects resources short-term source, precludes above prior evidence, Ferri deposit impairs CCBs. Curi 2020) opposite Funding offered collected namely expense divided funds. resource constraints makes price lever mitigate rising expect (Pérez-Cárceles Roman ?argu, 2013). LIQ liquid sum cash balances money instruments, marketable securities) scaled assets. Vazquez Federico (2015) posit soundness smaller extent directly low spur most practices. negative. magnify maturity become liquid, profit-efficient (Bitar Sakouvogui, contrary, Altunbas positively associated efficiency; indeed, (2018) lowers verify Islamic showing Earlier developing countries argue holding bear opportunity certain investment options could generate high profits, jeopardizing (Brissimis Kirkpatrick Sufian, expected unknown. NPL stands asset Mester, Podpiera Weill, 2008). retrospective indicator (Gaganis Pasiouras, Le Luo Mamatzakis, 2015). portfolio degradation reflects deteriorating general outlook flight part prime clients, operations. coefficient sign. Score stability, (ROA) plus deviation ROA. adjust temporal yield dynamic trend, rolling window years, recommended e.g., Barra Zotti, An elevated consistent institution (Shim, Howev