private Indian banks showed almost no change, and
foreign-owned banks showed a remarkable increase
in efficiency during the sample period. It is noteworthy that foreign-owned banks exhibited below-average performance through 1990, and improved dramatically to above-average performance in the last
year of the sample period, when they were nearly as
efficient as the public sector banks.
A second feature of the DEA results concerns
their variability. Domestically-owned banks exhibit
less variability in performance than do foreign-owned
banks. This finding is not surprising, and presumably
reflects a greater familiarity with the regulatory system. The greater variability in the efficiency of
foreign banks is also due to their dependence on less
stable wholesale or corporate resources, interbank
market borrowings, and refinance of assets. Domestic banks, on the other hand, have a more extensive
branch network, assuring a more stable retail banking business. Overall performance variability tends
to increase through time. This pattern holds clearly
for publicly-owned and foreign-owned banks, and
somewhat less clearly for private Indian banks. Liberalization may be leading to improved overall performance, but it is also creating winners and losers.
We also conducted an analysis of returns to scale
among the Indian commercial banks. We characterize returns to scale with the sign of the intercept of
the supporting hyperplane in the DEA multiplier
problem dual to (2). The intercept of the supporting
hyperplane is not unique for frontier banks, however,
since they are located at the intersection of two or
more supporting hyperplanes. In general, banks exhibit increasing, constant or decreasing returns to
scale at their optimal radial projection according as
the maximum intercept of their supporting hyperplane is negative, the minimum and maximum intercepts bound zero, or the minimum intercept is positive, respectively; see Banker and Thrall (1992). Our
analysis shows that most banks displayed decreasing
returns to scale. Indeed every Indian bank in every
year in the sample period operated in the decreasing
returns to scale region of production technology. The
persistence of diseconomies of scale could possibly
result from the RBI's branching policy. Indian banks
are required to open branches under the branching
policy, but are not allowed to close unprofitable
branches. This policy prevents optimizing resources