Please download to get full document.

View again

of 28
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.


Publish on:

Views: 19 | Pages: 28

Extension: PDF | Download: 0

SWEDISH SCHOOL OF ECONOMICS ND BUSINESS DMINISTTION THE YJÖ JHNSSON WOKING PPE SEIES IN INDUSTIL ECONOMICS 5 () Thoma Gehrg & une Stenbacka INFOMTION SHING IN BNKING: N NTI-COMPETITIVE DEVICE? Key word: Informaton harng, Poachng, Swtchng cot, Credt market, Bankng JEL Clafcaton: D8, G, L5 Swedh School of Economc and Bune dmntraton Thoma Gehrg & une Stenbacka Dtrbutor: Lbrary Swedh School of Economc and Bune dmntraton P.O.Box 47 Helnk Fnland Phone: , Fax: E-mal: SHS ntreebyrå IB (Oy Caa Securty b), Helngfor ISBN X Informaton Sharng n Bankng: n nt-compettve Devce? Thoma Gehrg and une Stenbacka July, btract: We analye the nttuton of nformaton harng n a model of repeated bankng competton. In the preence of wtchng cot we fnd that nformaton harng render poachng more proftable n future round of competton, nce the poachng actvte can be targeted to credtworthy borrower. Thu borrower poachng may occur even when t would not be proftable wthout nformaton harng. t the ame tme nformaton harng reduce relatonhp beneft, and competton for ntal market hare weakened. Overall we fnd that nformaton harng enhance equlbrum proft weakly n general and trctly n the preence of wtchng cot. Keyword: nformaton harng, poachng, wtchng cot, credt market, bankng JEL-clafcaton: D8, G, L5 We would lke to thank Sudpto Bhattacharya, John Bryant, Gabrella Chea, Mark Flannery, Han Haller, Phlpp Hartmann, Matthew Jackon, Jay tter a well a conference partcpant of the VI. World Congre of the Econometrc Socety n Seattle, the EIE annual conference n Lauanne, the Southeatern Theory and Internatonal Economc Meetng n Houton, the European Scence Foundaton Exploratory Workhop at LSE, and emnar partcpant at Caltech, Unverty of Florda Ganevlle, Unverty of Frankfurt, the IMF, the VPI Blackburg and the jont lunch emnar of the European Central Bank and the Deutche Bundebank for helpful comment. We gratefully acknowledge the hoptalty of ce Unverty (Gehrg) and Stockholm School of Economc (Stenbacka) a well a fnancal upport from The cademy of Fnland, the Hanken Foundaton, the DD, the Deutche Forchunggemenchaft and The Yrjö Jahnon Foundaton. Inttut zur Erforchung der Wrtchaftlchen Entwcklung, Unvertät Freburg, D-785 Freburg, Germany. E- mal: Swedh School of Economc, P.O. Box 47, Helnk, Fnland. E-mal: . Introducton Credt bureau and credt regter play an mportant role n communcatng credt htore of borrower to lender, and, thu, a wdely aerted, contrbute to more effcency n credt market. ecent contrbuton n the lterature have ubtantated th vew by preentng model wth ocally benefcal mplcaton of nformaton harng (Japell and Pagano (), Pagano and Japell (3), a well a Padlla and Pagano (7, ). Whle th lterature largely concerned wth the cot of lendng and the probablty of default of funded project, t rae lttle concern about potental ant-compettve mplcaton of nformaton exchange among lender. To the extent that nformaton exchange coordnate nformatonal aymmetre n the future th lterature tend to eparate trategc concern about future lendng from current lendng decon. We how that th property may gnfcantly reduce the ntenty of competton n the preent lendng market. Informaton harng, n fact, a commtment to reduce nformatonal aymmetre between bank n future lendng and thu wll relax prce competton among bank n the frt tage when cutomer relatonhp are formed. In the abence of nformaton harng the propect of extractng future nformatonal rent from etablhed cutomer relatonhp enhance current competton among lender. We preent a two-perod model of a bankng duopoly wth wtchng cot. In perod exante dentcal bank compete for borrower. t th tage the competton ymmetrc nce bank have acce to the ame general pool nformaton. In perod bank can effectvely beneft from a lendng relatonhp etablhed n perod, becaue they learn the type of ther cutomer durng perod. In perod competton aymmetrc becaue bank enjoy an nformatonal advantage wth repect to ther own borrower and an nformatonal dadvantage wth repect to the clent of the rval bank. Conequently, bank can prce dcrmnate accordng to the borrower htore. Moreover, due to wtchng cot bank enjoy ome degree of market power wth repect to ther captve clentele. ccordngly, they can prce dcrmnate n perod. Bank want to keep and protect ther credtworthy borrower on one hand and poach the credtworthy borrower of ther rval on the other hand. In our model, we fnd that bank can ecure the proft they would earn f they entered n perod only,.e. they can ecure the hypothetcal perod- poachng proft f they attempted to poach the whole market n perod. ny addtonal revenue are elmnated by ex-ante competton n perod. In fact, competton n perod elmnate any revenue that are related to market hare n perod. Hence, bank dcount the expected relatonhp beneft by prcng even below cot. The nttuton of nformaton harng enhance the proftablty of poachng, becaue wth nformaton harng bank can focu ther poachng offer n perod excluvely to credtworthy borrower. poachng proft re relatonhp beneft declne. However, the reducton of relatonhp beneft doe not affect nter-temporal equlbrum proft, nce thee ncumbency advantage are elmnated by ex-ante competton n perod. Conequently, wth nformaton harng perod -dcount by bank are reduced and perod- competton for market hare weakened. Wthout nformaton harng the pool acqured through poachng wll be adverely elected, whch reduce the return from poachng. Conequently, wthout nformaton harng the return from acqurng cutomer n perod are larger a the rval return from poachng decreae. Thu, wthout nformaton harng bank have ncentve to offer larger perod- dcount n equlbrum. In th repect the nttuton of nformaton harng can be regarded a an ant-compettve devce. The ant-compettve effect of nformaton harng reult from the emergence of relatonhp nformaton after the frt perod of lendng and the extence of wtchng cot that enable bank to earn potve poachng revenue. Informaton harng enhance poachng revenue, and, thu, the proftablty of the whole bankng ndutry. There ample evdence that relatonhp nformaton relevant and emprcally gnfcant n bankng market. For the U.S. Peteren and ajan (4) and Berger and Udell (5) fnd that lendng term are mproved n lendng relatonhp. Smlarly Hoh, Kahyap and Scharften () document beneft of relatonhp for Japan and Ela and Krahnen (8) for Germany. Swtchng cot n bankng have been etmated by Shy ( (Chapter 8), ) for Fnland and documented by Farnha and Santo () for Portugal. In fact, Shy conclude from real-lfe data orgnatng from the Fnnh depot market n 7 that wtchng cot account for between % and % of the average balance a depotor mantan wth a For a ytematc characterzaton of thee nttuton n an nternatonal context we refer to Jappell and Pagano (). Equlbrum ntal dcount n general duopoly model wth wtchng cot and prce dcrmnaton have alo been oberved by Chen (7). 3 bank. Km, Klger and Vale () nvetgate the Norwegan loan market durng the perod They etmate average wtchng cot to be 4, %, whch approxmatvely equal one-thrd of the market average nteret on loan n Norway durng that tme nterval. 3 Smlar to our model Peteren and ajan (5) alo fnd that bank ntally charge low rate n early lendng perod. However, n contrat to that tudy, n our model bank charge low rate n order to buld up market hare rather than to contan entrepreneural moral hazard. In contrat to the early paper on lendng relatonhp (e.g. Haubrch, 8, Sharpe, ) our model doe allow for effectve ex-pot competton. Due to dfferentated wtchng cot our model generate equlbrum wtchng depte the extence of relatonhp nformaton. Moreover, equlbrum wtchng doe occur n pure tratege rather than mxed tratege a n von Thadden (). Vrtually all the recent lterature on nformaton exchange n credt market tree the value of communcaton among lender n reducng default probablte of borrower n tuaton of lmted trategc nteracton among lender. We how that the uppoedly benefcal conequence of nformaton harng are a conequence of the lack of ex ante competton for the etablhment of relatonhp generatng nformatonal rent n thoe model. Typcally, n th lterature, nformatonal advantage are arbtrarly agned at an ntal tage of the lendng market nteracton between rval bank. By enrchng the market tructure to allow for potental nteracton of rval lender n each perod, we how that nformaton harng hould rather be vewed a an ant-compettve devce nce t reduce the compettvene of current lendng market dratcally. We provde a model n whch the future gan aocated wth nformaton harng are more than offet by the loe of current compettvene. Padlla and Pagano (7) demontrate advantage of nformaton harng n the context of a two-perod model, n whch bank are nformatonal monopolt n both perod. Bank oberve the true rk clae of ther clentele but not thoe of ther compettor. Hence, n the abence of nformaton harng bank can extract all the project return accrung to borrower, thu reducng the ncentve for entrepreneur to nvet n project-pecfc and ablty-enhancng technologe that ncreae repayment probablte. Bndng ex ante agreement to hare nformaton at the end of perod commt the bank to compete n the credt market n tage under condton of ymmetrc nformaton. Th mple a commtment to more effectve competton and, thu, to hare perod- urplu wth the entrepreneur. ccordngly, nformaton 3 See alo Tarkka (5) and Sharpe (7) for further tude of wtchng cot n the bankng ndutry. 4 harng wll ncreae the ncentve nto development of entrepreneural ablty and thereby the repayment probablte. In partcular, Padlla and Pagano prove that credt market may operate under a regme of nformaton harng, whle t would collape wthout communcaton. We how that th argument break down when lender are ymmetrcally nformed ntally. Lke Padlla and Pagano we conder contellaton n whch bank wll become nformatonal monopolt at tage. However, we generalze ther framework by allowng bank to compete n a genune ene for clent n perod. In th tuaton the propect of future rent ntenfe competton n perod. Ung a mple bankng model wth borrower wtchng cot we fnd that lender overall proft are hghet when they can commt to hare nformaton. t the ame tme entrepreneural ncentve to ncreae repayment probablte are lowet under nformaton harng. Hence, we have good reaon to nterpret nformaton harng n credt market a a potentally ant-compettve devce. Thu, polcy concluon regardng nformaton exchange between lender hould be drawn qute carefully a the welfare mplcaton are entve to the prece nature of the trategc mechanm whereby bank can create nformatonal advantage relatve to ther rval. Fnally, our analy related to a large body of lterature on nformaton harng. 4 Promnent example of th lterature nclude Shapro (86) and Gal-Or (85, 86), whch focu on the ncentve for olgopolt to exchange prvate nformaton concernng common market condton or frm-pecfc effcency. Thee model are two-tage game of the followng character. Pror to the actual obervaton of the prvate nformaton the frm have to make bndng commtment whether to reveal ther prvate nformaton or to keep t prvate. t the econd tage market competton (baed on Cournot and Bertrand competton) take place. Th lterature generally fnd that the drecton of the ex ante ncentve for nformaton exchange depend on the nature of market competton (Bertrand or Cournot) and on the type of uncertanty (uncertanty concernng common demand condton or frm-pecfc cot). Th lterature tend to agree, however, that nformaton harng ncreae ocal urplu n many cae. However, concern about ant-compettve conduct by nformaton harng agreement are not explctly addreed n th reearch approach. The model of a two-perod bankng duopoly ntroduced n ecton. Secton 3 preent the market equlbrum of the bac credt market. The role of nformaton harng explored n 4 See ath (6) for a comprehenve urvey. 5 Secton 4. Secton 5 compare full and partal nformaton harng and ecton 6 conclude. Some techncal detal and mathematcal proof are delegated to the ppendx.. Two-Perod Model of Bank Lendng Conder a mple two-perod model of competton between two lender, whch we wll, for convenence, henceforth call bank, and B. The bank compete for borrower wth a horzon of two perod, t= and t=, by offerng hort term credt contract. The lendng rate may potentally dffer acro perod. 5 For mplcty, we aume the bank opportunty cot of fundng to be contant over tme. In lne wth Padlla and Pagano (7, ) we dentfy borrower wth entrepreneur, who requre external fundng of one (monetary) unt a they have no fund of ther own. There are two type of entrepreneur, talented and untalented one. Only talented entrepreneur can generate potve cah flow. Thee entrepreneur have acce to a project that yeld a return of v wth probablty π and wth probablty π. Untalented entrepreneur never generate any potve cah flow but they derve potve utlty from controllng a project. 6 The proporton of talented entrepreneur commonly known to be λ . Entrepreneur are rk neutral. Bank ntally have no pecfc nformaton about borrower type. Ther knowledge retrcted to the general charactertc of the pool of applcant. In perod bank compete for borrower by announcng lendng rate, =, B. 7 t the end of the perod they oberve the type of ther cutomer. The bank alo make project-pecfc obervaton of whether thee cutomer have been ucceful or not. We can make the nterpretaton whereby th nformaton ha been generated wthn the framework of the cutomer relatonhp n perod (ee e.g. Sharpe, ). It prvate and relatonhp-pecfc nformaton of the bank, but t may be communcated between bank under a regme of nformaton harng. Wthout nformaton harng, however, competton at tage take place under condton of aymmetrc nformaton acro bank. ccordngly, n perod a bank wll fnd t optmal to charge dfferent prce to clent wth dfferent htore and announce lendng rate (, Q ), =, B, for extng good- 5 The pecfc contractual form could be motvated n a tandard way, for example, by reortng to ome form of cotly tate verfcaton. 6 Hence, under lmted lablty, they wll undertake project even when they expect nolveny wth certanty. 6 type borrower belongng to t perod- cutomer and new borrower, repectvely. 8 We can make the nterpretaton of belongng t rval perod- clentele. Q a a poachng prce, whereby bank tre to attract entrepreneur We aume that t cotly for at leat ome borrower to wtch lender n perod. More precely, we aume that an entrepreneur ha to ncur an doyncratc wtchng cot to break t perod- cutomer relatonhp and wtch to a competng lender n perod. Thee wtchng cot may capture for example the cot aocated wth another applcaton procedure at a le known bank, or mply the fnancal cot of tranferrng fnancal account from one bank to another. In the cae of bundled fnancal ervce offered by bank the wtchng cot could alo capture cot to entrepreneur aocated wth dtrbutng account change to etablhed longterm cutomer n the product market. The wtchng cot may be relatvely low for ome borrower and ubtantally hgher for other. In fact, the evdence reported n Shy ( (Chapter 8), ) a well a Km, Klger and Vale () content wth cutomer beng dfferentated accordng to ther wtchng cot. Thee tude report a an emprcal regularty that wtchng cot tend to re wth the ze of the bank n depot market, wherea wtchng cot tend to declne a a functon of bank ze n loan market. The borrower-pecfc wtchng cot are prvate nformaton of thee borrower. In the preent analy we aume that the heterogenety n wtchng cot acro borrower can be ummarzed by a dtrbuton functon F() on a upport [, ]. In the central part of our analy we wll concentrate on the cae of wtchng cot that are unformly dtrbuted on an nterval [, ]. ddtonally, we aume that borrower are not aware of ther doyncratc wtchng cot realzaton untl perod. Thu, the ntal choce between bank partcularly mple a t cannot be condtoned on borrower-pecfc wtchng cot. Followng Sharpe () and Padlla and Pagano () we aume that ucceful entrepreneur wll conume ther perod- revenue at the end of perod. Hence all 7 To be prece the lendng rate pecfe the repayment ncludng nteret and prncpal. 8 In fact, the development of the nformaton technology ha dramatcally mproved the ablty of frm to condton prce on purchae htore (ee, cqut and Varan ()). ecent tude by Vlla-Boa (, ) a well a Fudenberg and Trole () have explored a number of nteretng feature of prcng dynamc n model wth cutomer recognton. In the cae of depot account Shy (, ) ha recently etmated ubtantal average wtchng cot n the order of -% of the average balance a depotor mantan wth a bank, wherea Km, Klger and Vale () fnd the average wtchng cot to conttute one-thrd of the market average nteret rate n Norwegan loan market. The unform dtrbuton wll allow u to mplfy the algebrac analy. In prncple, we could allow for more general dtrbuton ncludng dtrbuton wth a potve ma on zero wtchng cot. 7 entrepreneur are n equal fundng need at the begnnng of perod,.e. they all demand loan of the ame ze. In ummary, the credt market open twce. t tage bank compete for unattached borrower by announcng lendng rate. t tage bank announce lendng rate (, Q) to perod- cutomer of t own and of t rval, repectvely, and the attached borrower decde whether they want to wtch and ncur the aocated wtchng cot. t the end of the econd perod cah flow are realzed and the market wnd down. Entrepreneur are protected by lmted lablty. Moreover, we aume perod-end nformaton avalable to bank to be verfable, and, therefore, that t can be ued to enforce the contractual arrangement mmedately. For the ake of comparablty, our etup reemble the model of Padlla and Pagano (7) wth the excepton that we tart wth a ymmetrc dtrbuton of nformaton at the begnnng of tage. Thu, nformatonal market power n perod emerge endogenouly a a conequence of trategc prcng n perod. In contrat, Padlla and Pagano (7) aume that bank, for reaon whch are not explaned, are endowed wth an nformatonal monopoly already n perod. n ther model, bank wll, however, end up wth aymmetrc nformaton at the end of perod. The bank enjoy uperor nformaton about the pat performance of ther clent, whch trengthen the compettve poton n perod. Th nformatonal aymmetry ha farreachng conequence for the bankng equlbrum a the next ecton develop n detal. 3. Credt Market Equlbrum In the mult-tage credt market delneated above equlbrum determned n the tandard way by backward nducton. Thu, ntally we olve for the prce game n perod for gven borrower htore and for a gven confguraton of market hare nherted from perod. The ndvdual htore are relevant becaue n perod bank want to prce dcrmnate between known and unknown cutomer. The bank try to keep known talented borrower and attract the rval unknown cutomer a long a poachng proftable. Hence, typcally, perod- prce wll Th aumpton not crtcal. In an earler veron of the paper we mantaned the aumpton that ucceful entrepreneur could actually fund econd perod project from ther perod revenue. In that cae the overall project qualty reduced n perod. Neverthele, the bac reult are not qualtatvely affected. 8 depend on market hare nherted from perod. Thee market hare reult from the perod- prce, the determnaton of whch we analyze ubequently. Equlbrum n perod Denote the market hare acqured by bank (=,B) n perod a µ. Thee market hare repreent the outcome of perod- competton, and, are, hence, fxed for the determnaton of equlbrum n perod. Each bank ha an nformatonal advantage on t captve clentele, nce t oberve the true type, wherea n the abence of nformaton harng rval only have acce to tattcal pool nformaton for unknown borrower. Hence the return from lendng to captve clent are π, =, B. The return from poachng aocated wth talented clent are π Q, =, B, wherea thoe aocated wth untalented one are. Snce the poachng bank attract all the untalented entrepreneur from t rval pool t face an advere electon problem, and, thu, a cot dadvantage relatve to the ncumbent bank. In addton to the nformatonal advantage nformatonal nder enjoy a compettve advantage due to wtchng cot. Borrower wll only accept a poachng offer, f poachng prce undercut the ncumbent quote by a uffcent margn o a to overcome the wtchng cot barrer. More precely, a talented entrepreneur wth wtc
Related Search
Similar documents
View more...
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks