Measuring Transaction Costs and Institutional Change in the U.S. Commercial Banking Industry

Measuring Transaction Costs and Institutional Change in the
U.S. Commercial Banking Industry

By

Margaret M. Polski
Indiana University



© 2001 by author

---

Abstract



There has been considerable theorizing in the new institutional economics about the
relationship between transaction costs and institutional change. However, there have been
few attempts to measure these changes in particular political economic environments.
This paper makes use of a long time series of financial data from a transaction costs
intensive industry, commercial banking, to begin to shed some light on these issues. A
preliminary analysis suggests that transaction costs can be measured over time in this
industry but not perfectly. Banking has two types of transaction costs, which move in
inverse relationship and are subject to different political and economic influences. Both
types of transaction costs are (almost) always changing even as they appear to be
converging to equilibrium. In 1998, total transaction costs were 77% of income, which is
8% higher than in 1934. During a recent period of intense economic and institutional
change in the industry, transaction costs moved out of equilibrium, increasing to 90% of
total income, and then returned to equilibrium as the intensity of change diminished. An
analysis of these events implies that the causal link between transaction costs and
institutional change works in both directions.

2

---

1
Introduction
The economic importance of transaction costs is widely recognized. Transaction
costs, which reflect the costs of economic organization both outside the firm and inside
the firm, are one means by which we can measure the efficiency of different institutional
designs in achieving economic outcomes in particular environments. While there has
been considerable theorizing about the relationship between transaction costs and
institutional change, there have been few attempts to measure these changes and the
relationship between them. Yet these are important issues in understanding the political
economy of institutional change. The goal of this paper is to shed some light on these
matters.
Scholarship in the new institutional economics provides the basis for several
different hypotheses about the relationship between transaction costs and institutional
change. One hypothesis is that changes in transaction costs induce institutional change.
However, changes in transaction costs can arise from technological change, other types of
relative price change, or policy changes. This suggests a second hypothesis in which
institutional change induces changes in transaction costs. A third possibility is that
transaction costs and institutional change interact. That is, institutional change could
induce changes in transaction costs, which could induce further institutional change to
adjust for the cost effects of the prior change.
Speculations about the relationship between transaction costs and institutional
change cannot be tested without specifying and measuring these concepts in particular
political and economic environments. Recent changes in the U.S. insured commercial
banking industry provide a good opportunity to do this kind of analysis. In the period

3

---

1960-1998, economic and policy changes reconfigured the competitive environment for
financial services in the United States. The availability of annual financial data for much
of the period 1934-1998, provides an opportunity to quantify changes in transaction costs,
as well as to study the sequence of changes in transaction costs and institutional change.
The first question this paper addresses is the problem of identifying and
measuring transaction costs over time in the U.S. commercial banking business. While
one might analyze transaction costs in the banking industry from either a micro political-
economic perspective or a macro perspective depending upon the issues in question, for
reasons more fully described in Section Two, this paper takes the macro view.
Banking activities generate two types of transaction costs, which are subject to
different political and economic influences. One type of cost, interest expense, reflects
the costs of funds for banking activities. A second type of transaction cost, noninterest
expense, reflects the costs of information and coordination. Over the period 1934-1998,
interest expense shows an increasing trend, whereas noninterest expense shows a
diminishing trend. Total transaction costs, the sum of interest and noninterest expense,
increase over this period from 69% of total income in 1934, to 77% in 1998.
When the behavior of both interest and noninterest expense is compared over the
entire time series, an inverse relationship emerges. This implies that when the costs of
funds increase, banking firms seek to reduce the costs of information and coordination
activities. After 1960, there is evidence that the two types of costs are converging to an
equilibrium in which each type is about 40% of total income. However, this equilibrium
is disturbed between 1972 and 1992, a period of profound institutional change in this
industry.

4

---

A second question this paper addresses is the relationship between transaction
costs and institutional change. An informal analysis of the evidence suggests that there is
interaction between transaction costs and institutional change: either could be the cause
of change in the other. However, further analysis is required to more precisely estimate
the relationship between these two processes.
The main conclusions about transaction costs and institutional change that can be
drawn from the analysis in this paper are first, that transaction costs can be measured in
the banking industry but not perfectly. The data used in this analysis include the costs of
organizational change. While the aggregated data generate some interesting questions, it
would have to be disaggregated to investigate the micro foundations of the implications
of the analysis in this paper. Given the nature of banking activities, this could prove to be
quite difficult.
Second, there are some regularities in the data that suggest that transaction costs
are nonlinear and subject, at least in part, to cyclical effects. There is also evidence that
suggests that transaction costs evolve over time. More work is required to understand
these dynamics.
Third, there is evidence that suggests that transaction costs are sensitive to
changes in the political economic environment. Regulatory stance and legislative activity
are important signals for those who are engaged in economic organization and transaction
cost management. Further work is needed to develop these linkages.
The analysis in this paper is related to and consistent with the work of a number
of other scholars. The continuous change in transaction costs and firm size that is
apparent in the data for the commercial banking industry over the period 1934-1998, is

5

---

consistent with Coase's (1937) prediction that equilibrium firm size is achieved by
business managers continually experimenting with alternative ways of organizing
economic activities. Moreover, the industry's investment in institutional change is a
logical implication of Williamson's (1985, 1997) analysis of the principles of transaction
cost economics.
An increase in transaction costs in the U.S. commercial banking industry in the
twentieth century is consistent with Wallis and North's (1986) finding that the transaction
cost sector in the U.S. is growing over the period 1870-1970. Moreover, their argument
that the growth in the transaction cost sector arises from both economic changes and
institutional changes has relevance for explaining the behavior of transaction costs in the
commercial banking industry.
The complex interaction between economic change, transaction costs, and
institutional change in the U.S. commercial banking industry over the period 1960-1998,
is consistent with North's (1990) analytical framework for explaining economic
performance and institutional change over time. Moreover, it is consistent with the
comparative empirical analysis of Berger, DeYoung, Genay, and Udell (Forthcoming
2000), who argue that cross-border consolidation in the banking industry has been
fostered by deregulation as well as economic change. The analysis in this paper is also
consistent with the micro level analysis of Klein and Saidenberg (1998), who argue that
banking firms that reorganize as Multi Bank Holding Companies (MBHCs) solve a
diversification problem in a transaction cost minimizing way.
The paper is organized as follows. Section Two analyzes the relationship
between transaction costs and institutional change. Section Three discusses measurement

6

---

issues. Section Four describes the data and analyzes changes. The last section
concludes.
2
Transaction costs and institutional change
Scholarship in the new institutional economics argues that transaction costs and
institutions are linked in important ways. However, there is a dearth of empirical and
theoretical analyses of the relationship between these two processes and how this
relationship changes over time. While it is generally accepted that transactions must be
governed and that some institutional arrangements are more efficient than others, the
nature and existence of an economizing selection mechanism is usually assumed rather
than explained (Shelanski and Klein, 1995). Recent changes in the U.S. commercial
banking industry provide a rich source of data for taking a closer look at these issues over
time.
Addressing the question of the economic justification for firms, Coase (1937)
argues that there are costs associated with transacting and that by internalizing
transactions in a firm, efficiencies can be achieved that cannot be achieved by transacting
in markets. The organization of credit allocation through banking firms and other forms
of financial intermediaries are a well-established demonstration of this fundamental
insight that is not inconsistent with more recent information-oriented theories of banking.i
While Coase predicts that business managers will be "constantly experimenting"
with alternative ways of organizing economic activities, his analysis is essentially static.
He anticipates that dynamic factors are important in explaining organizational change,
but he does not speculate about what these factors might be and how the economics of the

7

---

match between a particular set of transaction costs and particular organizational
structures might change over time.
Williamson (1985, 1997) explores the important issue of the relationship between
transaction costs and institutional design. He inextricably links transaction costs and
institutional change by defining the fundamental question of transaction cost economics
to be one of operationalizing the economizing problem in particular environments,
aligning transaction costs and governance structures in transaction cost economizing
ways, and comparing feasible institutional alternatives. Williamson anticipates the
potential for interaction between transaction costs and institutional change when he
argues that the alignment between transaction costs and governance is nested in an
additional set of institutional arrangements. If this set of arrangements change,
transaction costs change. Presumably, changes in transaction costs give rise to changes
in governance arrangements however, Williamson does not develop the transaction cost
implications of this interaction.
Menard (1997) applies transaction cost economics to analyze the internal
organization of firms. Reinterpreting Williamson, he develops a taxonomy of
governance structures that can be applied to a wide variety of internal contracting
dilemmas. While he anticipates the problem of change as well as the interaction between
firm governance structures and other governance structures, he does not tackle these
issues.
Addressing the question of transaction costs over time from a macro political
economic perspective, Wallis and North (1986) measure transaction costs in the U.S.
economy and find that the transaction sector grows from 25% of GNP in 1870, to over

8

---

50% of GNP by 1970. They attribute this growth intuitively to three main causes: (1)
increasing specialization and division of labor that creates more impersonal exchange; (2)
technological change accompanied by increasing firm size; (3) the diminishing cost of
using the political system to restructure property rights resulting from the development of
commissions. Wallis and North do not test the relationship between changes in
transaction costs and institutional change, however, their explanation of their empirical
findings are consistent with some form of interaction between transaction costs and
institutional change.
North (1990, 1997) further develops the empirical context of the relationship
between transaction costs and institutional change. He argues that transaction costs
change as a function of the interaction between changes in relative prices and existing
institutional structures. When changes in transaction costs reconfigure the bargaining
power of economic participants, transaction costs act as an incentive to make institutional
change. In this formulation, institutional change potentially enters on both sides of a
causal argument: changes in an existing institutional structure - institutional change -
could alter transaction costs, and changes in transaction costs could create incentives for
(additional) institutional change.

In sum, scholarship in the new institutional economic raises interesting questions
about the relationship between transaction costs and institutional change. The arguments
in the extant literature imply that there is a complex, interactive relationship between
transaction costs and institutional change. However, this dynamic has yet to be quantified
and investigated empirically over time in particular industries.
3
Measuring transaction costs and institutional change

9

---

Transaction Costs

Transaction costs have been defined in various ways. Coase, Arrow, and Wallis
and North have provided general definitions. Coase (1937) defines transaction costs as
the costs of using the price mechanism, which includes the costs of discovering relevant
prices, and negotiating and concluding contracts. Arrow (1969) defines transaction costs
as the costs of running the economic system.
Wallis and North (1986) draw a distinction between transformation activities and
transaction activities. Transaction costs are the economic value of inputs to a transaction
function as distinct from the economic value of inputs to a transformation function. In
their taxonomy, transaction costs are the costs of processing and conveying information,
coordinating, purchasing, marketing, advertising, selling, handling legal matters,
shipping, and managing and supervising.
Others have defined the concept in terms of particular aspects of transacting.
Williamson (1985), who focuses on contracting activities, defines transaction costs to
include the ex ante costs of drafting, negotiating & safeguarding an agreement, and the ex
post costs of haggling, costs of governance, bonding costs to secure commitments. Davis
(1986) defines transaction costs as those costs associated with "greasing markets,"
including the costs of obtaining information, monitoring behavior, compensating
intermediaries, and enforcing contracts. Barzel (1989) defines transaction costs as those
costs involved in transferring, capturing, and protecting rights.
In a survey of the transaction cost literature, Alchian and Woodward (1988)
transcend these differences by distinguishing between two types of transactions:
exchange transactions involving the transfer of property rights, and contracting

10

---