MultiStrategic Behaviour of Croatian Banks by Sanja Jakovljevi

Multi-Strategic Behaviour of Croatian Banks by Sanja Jakovljević Comments by Ralph De Haas (EBRD) YES – 17 th Dubrovnik Economic Conference – June 2011 1

Overview of the paper l l Spatial competition model with banks that produce two outputs – Loans – Deposits Local market power allows these banks to use: – Interest-rate strategies – Strategies in terms of the number and location of branches

Overview of the paper l l Data on 32 Croatian banks (2004 -2010) to estimate demand elasticities for loans and deposits: – Own and competitor interest rates – Own and competitor branch networks Main findings: 1. Interest-rate strategies more important than branching strategies 2. Before crisis: Deposit demand more sensitive than loan demand 3. During/after crisis: Loan demand more sensitive than deposit demand

Overall assessment 1. Well-written, clear, and to the point 2. Careful analysis 3. More to explore. . .

Some remarks (1) Literature review l Clear summary l What about information asymmetries between banks and their clients and the relationship with distance, spatial credit rationing (e. g. Degryse and Ongena)? What is the link? l At the end: clarify how this paper contributes to the literature

Some remarks (2) Missing l • Add a short section early on in the paper on the structure and development of the Croatian banking system (incl. current Figure 2) Show some descriptive statistics: • Number of banks, variation in branch networks across the regions (include a map? ) • Deposit and loan interest rates across banks • Other variables too (area, urban, highway, etc).

Some remarks (3) l General criticism of this literature: treats all bank activities as one ‘market’ l Competition structure for corporate and household clients may be very different. Same holds for different types of corporate clients l E. g. the market for blue-chip companies may be very competitive whereas SME market leaves more room for (local) market power l Impact of distance may be different in particular l Acknowledge this somehow. . .

Some remarks (4) • Not only different clients but also different banks • Bank. Scope: 28 Croatian banks of which 14 foreign owned • • Do foreign and domestic banks differ systematically? Or large banks versus small banks? Foreign banks’ interest-rate strategies may differ • Different client structure? Different opportunity costs of lending? • Different funding structure, less interested in local deposits? • Branching strategies may differ too

Some remarks (5) l l “When potential clients are choosing between branches, they do so on the basis of several determinants: – Distance – Interest rate – Transportation costs τ related to distance” What is the difference between (1) and (3)?

Some remarks (6) • • Clarify link between reality, theoretical model, and empirical model Reality: banks compete in different markets (i. e. Croatian regions). In each market they set: • an interest rate (let’s assume customers are homogeneous) • a deposit rate (perhaps the same everywhere) • and operate one or more branches • How exactly does this translate to the models?

Some remarks (7) l Imposed restriction: deposit and loan interest rates uniform across regions l “Unfortunately, bank data on regional interest rates is not currently available” l How much of a problem is this if your analysis is at the regional level and you are analysing interest rate strategies. . . ?

Some remarks (8) l For 20 regions: – Quarterly deposit volume – Quarterly loan volume – Number of bank branches (interpolated semi-annual data) l Not all banks span all 20 regions, so instead use three NUTS II regions l Really necessary? You throw away a lot of nice info. . .

Some remarks (9) 1. Before crisis: Deposit demand more sensitive than loan demand 2. After crisis: Loan demand more sensitive than deposit demand • Current explanation quite ad hoc and not very convincing • The fact that the deposit insurance scheme became 4 times as generous made depositors switch from large to smaller banks • But doesn’t this imply an increase in the sensitivity to deposit rates? • Room to dig deeper

Some remarks (10) • “So what? ” What is the main (policy) lesson? • Two options to expand this paper a bit more: 1. Use Croatia as an interesting case study to derive broader lessons (e. g. on the impact of foreign bank entry on strategic behaviour of banks) 2. Go deeper into the Croatian case and derive lessons for Croatian policy makers: l l l E. g. how does competition look like compared to some benchmark (link back to the findings of the extant empirical literature) More investments in physical infrastructure? Probably not the main recommendation you want to make on the basis of this paper “Branch networks are relevant for market power” (Are they? ) “So contemplate regulation of branching networks across markets”. Think twice. . . (see US evidence)