|
Article Excerpt
Abstract
This paper first empirically investigates the cost structure of the Greek banking sector. Secondly, it provides measures of economies (diseconomies) of scale and quantifies technical change and its sources. Finally, this paper measures total factor productivity growth and identifies its sources. Bank production is presented with two different approaches (the intermediation and the production approach) which are used to specify a translog cost function. The two different translog cost models are estimated through the full information maximum likelihood method of estimation on pooled time series and cross sectional data. The results obtained are not significantly affected by model specification. Both models indicate significant economies of scale and negative annual rates of growth in technical change and in total factor productivity. (JEL G24, G21)
**********
Introduction
In recent years, several studies attempted to investigate, empirically and theoretically, the structure and performance of financial institutions. Most recently, these attempts have been intensified due to significant changes that are taking place in financial industries around the world. For instance, in the United States, deregulation aims toward nationwide banking, particularly through the elimination of geographic restrictions. In 1993, the European Committee of the European Union issued a Second Banking Directive. It was designed for the liberalization of financial markets so that banks could operate freely throughout the European Union. In order to investigate the impact of deregulation on the banking industry, it is necessary to know the cost structure of the industry.
Over the past 12 years, the Greek banking industry has undergone substantial changes due to the fact that Greece became a member state of the European Union. This paper investigates the cost structure, technical change, and productivity growth of the Greek banking industry. Two different approaches, the intermediation and the production approach, are used to specify a translog cost function and, thus, two different models are investigated. The intermediation approach treats deposits and other liabilities as inputs, while the production approach treats them as outputs. Both approaches are investigated because they have been extensively used in the literature, [Benson et al., 1982; Hunter and Timme, 1986; Ashton, 1998; Lang and Welzel, 1998].
Based on the translog cost function, this paper first empirically investigates the cost structure of the Greek banking industry. Second, it obtains measures of economies (diseconomies) of scale. Third, it measures the rate of technical change and its sources and, finally, it measures the rate of growth in total factor productivity and its sources.
Empirical studies investigating the cost structure of the U.S. banking system include the works of Hunter and Timme [1986], Humphrey [1993], Bauer et al. [1993], and Mahajan et al. [1996], among others. Studies examining the cost structure of banks outside the U.S. include the works of Kim and Ben-Zion [1898] for Israeli banking, Parsons et al. [1993] for Canadian banking, Dietsch [1993] and Muldur and Sassenou [1993] for French banking, Ashton [1998] for U.K. banking, and Lang and Welzel [1998] for German banking.
This paper is organized as follows. The first section presents the model, while the second section gives the model specification. The next two sections present the data and discuss the statistical results. The final section provides the conclusions and the policy implications.
Model Definition
In the banking literature there is some debate about what constitutes inputs and outputs for banks. Generally speaking, researchers follow one of the two main approaches of the input and output specification, such as the intermediation approach and the production approach. In this paper, both approaches are used to specify cost functions of the Greek banking industry. Furthermore, banks are assumed to minimize costs for both model specifications.
The intermediation approach [Sealey and Lindley, 1977; Lang and Welzel, 1998; Ashton, 1998] considers banks as financial intermediaries that convert deposits and purchased funds into loans and financial investments. This approach treats loans and other financial assets as outputs, while deposits and other liabilities are treated as inputs. In this paper, a cost function related to the intermediation approach may be presented as:
C = g([y.sub.1], [y.sub.2], [w.sub.1], [w.sub.2], [w.sub.3]) , (1)
where [y.sub.1] and [y.sub.2] represent loans and investment assets and [w.sub.1], [w.sub.2], and [w.sub.3] correspond to the price of labor, capital, and borrowed funds. Thus, according to this approach, the banking industry is viewed as transforming labor, capital, and borrowed funds into loans and investment assets.
The production approach [Benson et al., 1982; Hunter and Timme, 1986; Ashton, 1998] considers banks as transforming labor and capital inputs into two output groups of assets and liabilities. In this paper, a cost function corresponding to the production approach may be presented as:
C = g([y.sub.1], [y.sub.2], [y.sub.3], [w.sub.1], [w.sub.2]) , (2)
where, [y.sub.1], [y.sub.2], and [y.sub.3] represent loans, investment assets, and deposits and [w.sub.1] and [w.sub.2] correspond to the price of labor and capital.
Model Specification
The translog joint cost function for m outputs and n inputs can be written as follows:
ln C = [a.sub.0] + [m.summation over (i=1)][a.sub.i] ln [y.sub.i] + [n.summation over (i=1)][[beta].sub.j] ln [w.sub.j] + [1/2][m.summation over (i=1)][m.summation over (j=1)][[delta].sub.ij] ln [y.sub.i] ln [y.sub.i] + [1/2][n.summation over (i=1)][n.summation over (j=1)][[gamma].sub.ij] ln [w.sub.i] ln [w.sub.j] + [m.summation over (i=1)][n.summation over (j=1)][[rho].sub.ij] ln [y.sub.i] ln [w.sub.j] + [a.sub.t]T + [1/2][a.sub.tt][T.sup.2] + [m.summation over (i=1)][[delta].sub.it] ln [y.sub.i]T + [n.summation over (j=1)][[gamma].sub.jt] ln [w.sub.j]T + [summation over (k)][f.sub.k]B[N.sub.k] (3)
where C is the total cost, [y.sub.i] is the quantity of output i, [w.sub.j] is the price of input j, T is the time trend, and B[N.sub.k] is a bank specific dummy. By Shephard's lemma, the translog cost function yields the following cost share equations and is used with the translog cost function (3) to form the system to be estimated:
[S.sub.j] = [[[partial derivative] ln C]/[[partial derivative] ln [w.sub.j]]] = [[beta].sub.j] + [n.summation over (i=1)][[gamma].sub.ij] ln [w.sub.i] + [m.summation over (i=1)][[rho].sub.ij] ln [y.sub.i] + [[gamma].sub.jt]T (j = 1, 2, ...n) . (4)
The cost function must satisfy the following regular properties: twice continuous differentiability, linear homogeneity in input prices, and monotonicity and concavity in input prices. Twice continuous differentiability implies that the second order derivatives of the cost function should be invariant with respect to the order of differentiation. This holds when the following symmetric equalities are satisfied:
[[delta].sub.ij] = [[delta].sub.ji] and [[gamma].sub.ij] = [[gamma].sub.ji], i [not equal to] j . (5)
Linear homogeneity in input prices implies that the share equations are homogeneous of degree zero in prices (so that only relative...
|
