The Programmatic Expansion of the U.S. Government

• Author: Kenneth N. Bickers
• DOI: 10.1177/106591299104400406
• Date: 01 December 1991
• Published date: 01 December 1991
• Subject Matter: Articles

THE
PROGRAMMATIC EXPANSION OF
THE
U.S. GOVERNMENT
KENNETH N. BICKERS
Rice University
esearch on the correlates and causes of government growth
~~c has constituted something of a cottage industry in recent years
(cf. reviews by Lybeck 1988; Larkey, Stolp, and Winer 1981;
Borcherding 1977; and Tarschys 1975). Research has focused on the
empirical utility of factors that have been proposed as potential sources
of government growth (e.g., Berry and Lowery 1987; Cameron 1978).
Research also has focused on methodological debates about how to
measure and compare the scope of the public sector over time and
cross-nationally (e.g., Klein 1985; Lewis-Beck and Rice 1985; Beck
1976). This paper, however, is not primarily concerned with the question
of what factor (or factors) best explains government growth; nor is it pri-
marily concerned with methodological issues. Instead, it seeks to show
that the selection of explanatory variables as well as questions of meth-
odology must turn on substantive decisions about what is to be explained.
My argument is that focusing on the size of government in the
aggregate (however measured) tends to mask the issue of what aspects
of the government’s responsibilities are increasing or decreasing in size
and importance. This is consistent with Rose’s injunction that &dquo;we
must consider what government does as well as why it grows.&dquo; (1984:
1). On his view, size is only one issue, and not necessarily the most
important one: &dquo;To measure the totality of government by one undif-
ferentiated observation reduces everything to a denominator so com-
mon that it tells us nothing in particular&dquo; (1984: 5). Rose proposes a
programmatic approach as a way of analyzing the mix of qualitatively
different activities in which governments can be engaged. Such a focus
permits a wider and richer range of comparisons than a focus only on
aggregate measure of government size.
The activities on which governments expend resources can vary
significantly. Two different governments may differ in size because
Received: June 23, 1989
Revision Received: November 19, 1990
Accepted for Publication: November 20, 1990
NOTE: This research was supported by the National Science Foundation (Grant No.
SES-8921109).

---

892
they invest in essentially the same mix of programs at two different
levels or because they have a significantly different mix of programs.
Likewise an individual government may increase in size because all of
its programs grow at about the same rate or because some of its pro-
grams grow rapidly while others lag behind.
Failure to recognize the different mix of things that governments
undertake may be partly to blame for the relatively poor empirical
record of government growth theories. Berry and Lowery (1987), in
an extensive comparison of government growth theories, conclude that
none of the hypotheses they examine adequately explains real growth
in the aggregate size of the U.S. public sector. They suggest, however,
that the attempt to explain government size at the aggregate level may
be partially responsible for the lack of empirical support. When they
decompose government size into four components - the public sector
inflation rate relative to that for the economy as a whole, transfer pay-
ments, domestic purchases, and defense purchases - they find that some
types of explanations are well supported when applied to particular
components of government size, even though none works for all of its
components. Government size in the aggregate may simply be too
gross a measure to expect that any particular variable should be capa-
ble of explaining its variation.
A reconceptualization of the government growth issue is needed.
The main purpose of this paper is to show that a programmatic approach
offers a potentially fruitful avenue for rethinking the dynamics that
contribute to government growth. The first section of the paper argues
that government growth can be viewed as a combination of change in
expenditures on ongoing programs and change in expenditures that
stem from program creation and termination. The second section intro-
duces a program-level data base covering U.S. federal domestic pro-
grams for the period 1971 through 1990. The third section presents
empirical data on the pattern of programmatic expansion over the
1971-1990 period. This section estimates the proportion of the overall
change in domestic spending that is due to incremental changes in
ongoing programs and to net program creation and termination. The
paper concludes with an assessment of the utility of this approach for
the study of government growth.
A
PROGRAMMATIC APPROACH
How to define and identify a program is not as simple as one
might wish. The functions, responsibilities, and activities of govern-

---

893
ment can be disaggregated and grouped together in an almost infinite
number of ways. Although no definition is likely to be fully satisfac-
tory, the approach taken here is to define a program in terms of the set
of official activities designed to accomplish a specific public purpose.
Linking programs to purposes places the analytic focus on the array of
things that governments do. Such a definition has an intuitive appeal.
In ordinary usage, we tend to think of programs in terms of the goals
for which tax revenues are collected and spent, the assignments on
which government employees devote their energies, and the particu-
larized benefits enjoyed by the recipients of government aid. Opera-
tionally, of course, differentiating between individual programs may
sometimes be problematical. First, the purposes to which government
resources are mobilized may be partially overlapping. Second, some of
the purposes to which government resources are mobilized may be so
broad or ill-defined as to defy clear delineation. Third, some resources
(e.g., staff, government office buildings) may be devoted to multiple
purposes more or less simultaneously. These problems mean that some
degree of ambiguity will always exist when trying empirically to estab-
lish boundaries between programs. These and other empirical prob-
lems will be discussed below. For now, however, I want to discuss the
utilities of such an approach.
One is that it becomes possible to conceptualize government size
as a product of the set of a government’s programs and its expendi-
tures on each program, respectively. In its simplest form, this relation-
ship can be stated in the following identity:
where GOVT is the size of government in the aggregate, EXP is an
expenditure level on a given program, i identifies a program from the
population of n programs that the government pursues at a given
point in time.
The main utility of this identity is to draw attention to the possi-
bility that changes in the size of government may be due to expendi-
ture increases in the existing array of programs or to an expansion of
the government’s programmatic commitments, or some combination
of both. Most theories of government growth have failed to differen-
tiate between these two possibilities, but the distinction may be criti-
cal. For example, Baumol’s (1967) argument about the consequences
of the differential between public sector and private sector productivitiy
growth rates seems to assume that government growth is driven by

---

894
expenditure increases in existing programs, rather than from the cre-
ation of new programs. Niskanen’s (1971) theory of self-aggrandizing
bureaucrats likewise seems to assume that most government growth is
due to increasing expenditures on existing programs. By contrast, one
interpretation of Wagner’s law (Lybeck 1988) holds that government
growth is mainly a product of new government programs which are
created in response to the need for public provision of health care, old
age assistance, education, unemployment insurance, etc., that accom-
panies the shift out of traditional society into industrial society. Simi-
larly, Buchanan and Tullock (1962) and their successors (e.g., Becker
1983) view most government growth as the result of interest group
mobilization for new programs that will provide them particularized
benefits. Other examples could easily be adduced. My point is that
theories of government growth need to make clear what type of growth
they are trying to explain. Failure to do so is likely to lead to inappro-
priately designed tests and thus false conclusions about the causes of
government growth.
This point becomes clearer when Equation 1 is restated in terms of
change in the size of government. To restate Equation 1 in this form,
it is necessary to decompose the right hand side of the identity into
two parts, corresponding to (a) changes in outlays on the &dquo;ongoing&dquo;
stock of programs (i.e. those that are neither added nor terminated in a
given year), (b) a component that represents the amount of change in
expenditures that stems from movement into new programmatic com-
mitments and out of others. This second component is itself composed
of two parts: the set of newly added programs and the set of programs
terminated in a given year. This expanded identity can be stated as
follows:
where GOVT and EXP continue to have the same interpretation as
before; A is a delta sign denoting change from the previous time
period to the current time period; a identifies a program from the set
of x programs that survives from the previous time period into the
next time period; b identifies a program from the set of y programs
that have been added since the previous time period; and c identifies a
program from the set...