Local or State? Evidence on Bank Market Size Using Branch Prices

[...]The branch prices we study are less limited. Increasingly,banks are entering new markets by buying one or morebranches from other banks (Benz 1998). The price of a givenbranch should depend on the branch’s expected profits, andexpected profits, in turn, depend on competition. All elseequal, branches in less competitive (that is, more concentrated)markets will fetch higher prices because the absence ofcompetition enables branch owners to lower deposit ratesor service levels (or both).Using prices on 110 branch deals over 1992-99 in tennortheastern states, we run a type of “horse race” to determinewhether branch prices depend more on concentration at thelocal level (as the local-market paradigm implies) or at thelarger, state level. Our branch price data seem to work wellin the sense that branch prices are always correlated withconcentration at one level or another. Some of the specific findings are consistent with the state-market hypothesis; acrossall years in our sample, branch prices are more closelycorrelated with bank concentration at the state level than at thedesignated market level. State-level concentration also tends tomatter more for branch prices in dollars and cents, not just instatistical terms. However, the correlation between branchprices and state concentration depends partly on how we cutthe data, so we cannot conclude entirely in favor of the statemarkethypothesis. Branch price data certainly advance thelocal-versus-state debate, and with enough such data, thatquestion might be settled once and for all.The next section discusses conceptual definitions of marketsand summarizes actual Federal Reserve practices in designatingmarkets. Section 3 reviews some of the evidence on market size,most of which, it should be admitted, favors the local-markethypothesis. In Section 4, we present our findings, showing thatbranch prices also depend on concentration at the state level, notjust at the local level. Section 5 discusses robustness and caveats.