Budgetary Decision Making in the Twentieth Century: Theories and Evidence
Reddick, Christopher G., Journal of Public Budgeting, Accounting & Financial Management
ABSTRACT. Three rival internal change theories of budgeting have characterized the twentieth century. Internal change theories consider the importance of change made inside governments; exogenous factors do not have a significant bearing on outcomes in these approaches. The century started out with the rational/scientific, then moved to incrementalism by mid to late-century. The century ended with the garbage can model. This paper tests the descriptive power of each of these models on budget outcome data. The methods used in this study were time series modeling of monthly U.S. national government budget outcomes from 1968 to 1999. The results provide support for both incrementalism and rational/scientific budgeting. There was no support found in this study for the anarchical tendencies associated with the garbage can theory of budgeting. These results suggest that the garbage can model may not deserve its reputation as the dominant budgeting paradigm at the close of the twentieth century.
Three rival internal change theories of budgeting have characterized the twentieth century. Internal change theories consider the importance of change made inside governments, exogenous factors do not have a significant bearing on outcomes in these approaches. The century started out with the rational/scientific, then moved to incrementalism by mid to late-century. The century ended with the garbage can model. This paper tests the descriptive power of each of these models on budget outcome data. The methods used in this study were time series modeling of monthly U.S. national government budget outcomes from 1968 to 1999 The results provide support for both incrementalism and rational/ scientific budgeting. There was no support found in this study for the anarchical tendencies associated with the garbage can theory of budgeting. These results suggest that the garbage can model may not deserve its reputation as the dominant budgeting paradigm at the close of the twentieth century.
The U.S. government in the late 1990s achieved budgetary surpluses only for the first time in almost 30 years. This new phenomena was expressed in the public's view toward the budget surplus. Recent public opinion polls have asked questions about what the federal government should do with the additional revenue.
- A Washington Post/Kaiser/Harvard 2000 Election Economy Survey (IPOLL, 2001) conducted in October 2000 asked "Which of these do you think should be the top priority for any surplus money in the federal budget: Cut federal income taxes, put it toward reducing the national debt, strengthen Social Security/Medicare, increase spending on other domestic programs, or national defense/military?" The highest response recorded at 44% was to strengthen Social Security/Medicare. The second highest response at 24% was to put the surplus toward reducing the national debt. Cutting federal income taxes received only 20% of respondents in this poll.
- In November 2000 a CBS News Poll (IPOLL, 2001) asked "Most people think the country is doing well economically, and there is currently a surplus in the government's budget. Do you think the surplus money should be given to taxpayers whose tax money contributed to it, or do you think the surplus money should be used to solve big problems the country hasn't been able to afford to solve before?" Half of the respondents believed the surplus money should solve big problems the country has not been able to afford. Only 38% believed that the surplus money should be given back to taxpayers.
- A Fox News/Opinion Dynamics Poll (IPOLL, 2001) conducted in October 2000 asked: "Which of the following statements do you agree with more? I'd rather pay higher taxes to support a larger government that provides more services. I'd rather pay lower taxes and have a smaller government that provides fewer services." The majority of respondents, 52% believed that they would rather pay
- Lower taxes and have smaller government while 32% wanted to pay higher taxes to support larger government.
These public opinion polls in combination clearly demonstrate inconsistent preferences by the public. On the one hand, the public wants more spending through Social security, Medicare, defense, and the federal government solving problems it was not able to address in tougher economic times. On the other hand, the public wants lower taxes and a smaller government. However, what theories have been used to describe the budgetary process in the United States in the twentieth century that can account for this dramatic change from surplus to deficit to surplus?
Historically, following WWII there was the incorporation of the rational/scientific method of budgeting. In this model, decision makers are assumed to have access to all the necessary information and ability to make informed policy choices when determining the budget. Planning-Programming-Budgeting System (PPBS) first introduced in the Defense Department in 1961 by Robert MacNamera and prescribed for all federal agencies by President Lyndon Johnson was an excellent example of this method of budgeting. Then came along incrementalism, the challenger and reaction to the rational approach. It denied the assumptions that decision makers have the requisite time and resources to make rational decisions in a political setting; marginal changes would rule the day. Finally, at the close of the century we witnessed garbage can decision making, which rejects both the rational and incremental approaches and argues that decision making has chaotic qualities; decisions are very much dependent on time and chance. This paper demonstrates the historical and institutional relationship between these budgetary decision making theories and outcomes and tests the validity of these three dominant approaches on budget outcomes.
This paper first examines the pattern of budgetary inflows and outflows over the past 30 years, demonstrating the movement from surplus to deficit to surplus as an example of a period of dramatic change in budgeting. The second section briefly reviews the literature of the dominant roles of rational/scientific, incrementalism and garbage can theory in the twentieth century. The third and fourth sections demonstrate the data set used and the time-series method for testing budget outcomes. Next, we examine the findings from the tests of these rival theories. The concluding section examines the significance of the results found and implications for budgeting in the 21^sup th^ century.
PATTERNS OF BUDGETARY INFLOWS AND OUTFLOWS
A close examination of Figure 1 reveals that there was a substantial rise in the federal debt from about 1982 to the mid-1990s, peaking at almost 70% of GDP in the mid-1990s, due to the large string of budgetary deficits.1 Not until mid-1990s did the budget head back toward surplus due to the longest expansion of the economy in the postwar. Figure 2 shows receipts and outlays as a percentage of real GDP. It illustrates that receipts decreased from the early 1980s and peaked again in the mid-1990s with the favorable economy. Outlays increased steadily after the mid-1970s, and reached a peak during the mid-1980, declined markedly in the early 1990s. Interestingly, Figure 2 demonstrates the massive fiscal dividend the Clinton administration benefited from in the late 1990s. There were several reasons for the rise and fall of budgetary deficits. The historical battle against the budget deficit can be dated back to the late 1960s because of the Vietnam War.
The U.S. experience with sustained budgetary deficits started because of the Vietnam War. This War was unpopular; Congress was reluctant to pay for it with tax increases. Rather than considering the deficit as an appropriate means of financing military operations as it had been thought of in previous wars, many Americans regarded it as evidence of an abandonment of fiscal responsibility. Furthermore, when the war ended, federal spending did not recede and the budget did not return to balance. There is also a political explanation for the budget deficit.
One political explanation for sustained budgetary deficits was Congress' request for its own budget process culminated in the Congressional Budget Act (CBA) of 1974, which President Nixon signed into law less than a month before he was driven from office by the Watergate scandal. In 1973, President Nixon claimed the authority to impound congressional appropriations; this impoundment had considerable effects on the budget process. The ensuing budget fight culminated in the CBA. The act limited the President's ability to impound congressional appropriations and imposed a measure of discipline on the internal congressional budgeting process. Some political scientists have argued that the act resulted not in more fiscal discipline but less (Kamlet and Mowery, 1992). Regardless, deficits soared after the 1974 CBA. The involvement of Congress explicitly with the creation of the Congressional Budget Office (CBO) may have something to do with this. The lack of tax revenue and slow to modest economic growth was another factor for sustained deficits.
Other primary reasons for unprecedented deficits were President Reagan's 1981 tax cuts, poor to modest economic growth and the rise in entitlement spending. Recent budgetary history might have been much different if the Reagan administration had not seriously miscalculated the impact of its spending and tax policies in 1981 (Schick, 1990). Reagan signed both a $750 billion tax reduction (over five years) and a $130 billion spending reduction (over three years) into law on August 31, 1981. Furthermore, the deficit was not sufficiently vigorous to keep pace with the rise in outlays. The inadequacy of growth is evident when the performance of the 1980s is compared with that of earlier decades. Since most of the federal budget is programmed for growth, it takes more than a 2 to 3% annual expansion to avoid substantial deficits (Schick, 1990). At the same time as the lack of revenue growth, the built-in expenditures for numerous entitlement programs were rising rapidly (e.g., Social security, Medicare and, Medicaid). The deficits themselves caused a cumulative increase in interest costs to be paid on the national debt.
In summary, the overriding causes of sustained budgetary deficits can be attributed to entitlement spending, the CBA of 1974 and poor to modest economic growth. Most notably, the national government was not able to increase taxes to a level that would achieve budgetary balance due to taxpayer fatigue (e.g., Proposition 13 in California). The lack of revenues and tax expenditures, which benefit some groups to the exclusion of others, are major reasons for the sustained budgetary deficits this government faced.
There were various institutional reform efforts as the budget crisis deepened in the 1980s. Notably Congress established annual targets to reduce the deficit. The Balanced Budget and Emergency Deficit Control Act of 1985, more commonly referred to as Gramm-Rudman-Hollings (GRH), called for the progressive reduction in the deficit in each fiscal year from 1986 through 1990 and for a balanced budget in 1991. This approach promised to deliver automatic cuts to spending if Congress' budget work was not completed on time. More importantly, it set the tone that would last throughout the 1990s of the necessity of solving America's deficit dilemma.
However, 'eyeballing' the data and providing a historical-institutional account may not provide enough evidence as to the impact of such theories as garbage can, budgetary incremental ism or the rational/scientific approaches as description of outcomes. Visual inspection perhaps provides no substitute for formal testing to find statistical evidence of these processes at work. Therefore, in this study we formally examine and test three alternative conceptions to locate the most robust theory or theories that explain the process in the twentieth century. In this study we have not tested all the potential budget theories, but the three we have chosen are representative of very dominant theories of the twentieth century. The following section provides a framework that can be used to discern why we have chosen these three approaches.
INTERNAL CHANGE BUDGETARY THEORIES
This paper examines policy outcomes in public budgeting. We specifically examine what are labelled "internal change" approaches, which examines change in the budgetary process within government. This approach does not consider exogenous factors outside the immediate control of government as having any substantial influence on budgetary outcomes. It is termed a "change" model because it examines the amount of change in budgetary choices from one period to the next and/or the stability of the size of this change. Within this class of models, there are three specific approaches that evolved in the twentieth century starting with the rational/scientific, moving to incrementalism by mid century, and ending the century with the garbage can theory. There are other approaches to budgetary decision-making, but we will reveal why these three were the dominant approaches for the twentieth century.
Rational/scientific budgeting was the dominant paradigm in the first half of the twentieth century. This approach assumed the desire of the budget makers to consider all possible outcomes when devising his/her budget. Performance budgeting was introduced in 1913 with the Taft Commission and was implemented in the Department of Agriculture in 1934 with mandate of linking inputs (dollars) with outputs (results). We also witnessed PPBS in the 1960s as another rational method. However, the idea of rational budgeting that is explored in this study is the notion that decision makers will act rationally and think about the future, rather than just use past decisions to determine the next period's budget. A version of the rational approach can be applied to public budgeting as demonstrated by a balanced budget constraint for the public sector. Hamilton and Flavin (1986) first tested this model in the public finance literature to see if governments are subject to a constraint. The question posed "When a government runs a deficit, is it making an implicit promise to creditors that it will run offsetting surpluses in the future?" If governments are subject to this constraint, the policy of running a permanent deficit is unfeasible. This balanced budget constraint has a startling implication for budgetary decision-making: in the short run budgetary deficits do not make a difference, since in the long run the budget will be balanced. In the long run, these budgetary reforms will have some accumulated impact on the deficit of a country's budget. In the short-run, deficits do not immediately respond to changes in budgetary policy. Budgetary reforms do not immediately alter the course of the budget since many components of the budget such as mandatory spending are uncontrollable in the short-run. Furthermore, a poorly performing economy may not bring in enough tax revenue to balance the budget. Therefore, the budget is "sticky" in the short-run and "variable" in the long run. This long-run approach differs from traditional notions of budgeting, such as incrementalism, which primarily considers previous spending decisions as the main determinant of next year's budget. By contrast, rational budgetary decision makers are prospective and look into the future.
Recently, Wood (2000) tested what he labels a "balanced budget force" in the United States modeled by annual data between 1906 and 1996. He indeed finds evidence of an equilibrium tendency in federal budgeting measured by an error correction model. This force demonstrates the long-run relationship, where the budget may be out of balance in the short-run, but it tends toward balance over the long run. Wood's theory is very similar to the extensive literature on the intertemporal budget constraint by Hamilton and Flavin (1986) and the work done by Baghestani and McNown (1994) on the theory that revenues and expenditures are cointegrated and react to each other. In the rational/scientific approach, decision makers are assumed to have the time and resources to make informed decisions; this is in contrast to budget incrementalism that uses rules of thumb when making decision.
The second paradigm in public budgeting was a reaction against the assumptions of the rational model. Since the publication of Wildavsky's (1964) The Politics of the Budgetary Process, more than three decades ago, incrementalism has been the pre-eminent theory of budgeting, and a leader in explaining how the budget process works. At the national level in the United States, this well known study emphasized the role of agencies in the budget process, assumed their desire for growth, and discussed their strategies in dealing with the congressional review process, especially the appropriations committees. Two years after Wildavsky's (1964) book, Fenno's (1966) The Power of the Purse: Appropriations Politics in Congress, was published. This book is still treated as a classic, and together they framed the incrementalist assumptions about budgeting at the national level in the US. They emphasized the centrality of a legislatively dominated budget, the importance of agencies in the process, the decentralization of the process, and the lack of comparison between alternatives for spending.
In the incrementalist approach, the budget policy maker does not have to try to understand all spending decisions from the ground up. Instead, his or her job is to comprehend how the various budget allocations differ from the status quo. Therefore, in this approach budget makers concentrate their evaluations on the margin, or on the increments by which the output differs from one policy to another (Braybrooke and Lindblom, 1963). This approach simplifies the complex decision-making process by removing the comprehensive evaluation of all alternative budget proposals. In most cases, policies that are politically feasible are only incrementally different from the existing spending levels. Incrementalism limits executive discretion because budget decision-making outcomes are determined at the margin, not according to the powers of the executive to initiate policy change.
A variant of budgetary incrementalism is punctuated incrementalism, which was first demonstrated by Davis, Dempster and Wildavsky (1974). This implies a constant style of budgetary decision making occasionally disrupted by exogenous shocks. It suggests that rather than making moderate adaptive adjustments to an ever-changing environment, budgetary decision-making is characterized sometimes by incremental decisions, when existing decisions are routinely employed, and sometimes by punctuations, when a slowly growing condition suddenly appears on the agendas of budgetary policy makers. It takes increased pressure such as a crisis situation to dislodge established ways of thinking about policies. The result is periods of stability inter-dispersed with occasional, unpredictable and dramatic change (Baumgartner, Jone & True, 1998).
The last internalized change theory has the least amount of rationality of the three models mentioned thus far and was the dominant paradigm of policy making at the close of the twentieth century.
At the close of the twentieth century a new paradigm arose in the policy-making literature, which was a rejection of both the rational and incremental approaches. The garbage can model was first developed by Cohen, March, and Olsen (1972) to describe decision-making in colleges. This theory portrays that the modern college faces decision situations involving unclear goals, unclear technology, and fluid participants. In this model, active decision-makers and problems track one another through a series of choices without great progress in solving problems. However, the anarchical features of this model perhaps can be extended by saying that public institutions, in general, may have many of the characteristics associated with the garbage can theory (see Kingdon, 1984).
Extensions of the theory came from the work of March and Olsen's Ambiguity and Choice (1976), March and Weissinger-Baylon's Ambiguity and Command (1986), and March and Olsen's Rediscovering Institutions (1989). The latter work after the original 1972 article dramatically extends the scope and complexity of garbage can like thinking and ultimately develops a new theory, the March-Olsen variant of the new institutionalism, which makes frequent use of the garbage can theory ideas. Few studies have examined this theory by other authors. A notable exception was Kingdon's Agendas, Alternatives, and Public Policies (1984), one of the most highly regarded treatments of policy making in the last two decades. However, the original authors, particularly March and Olsen have done most of the basic theoretical work in this tradition (Bendor, Moe & Shorts, 2001).
The most important attribute of the garbage can theory is the four independent streams of decision-making: problems, solutions, participants, and choice opportunities. This kind of organization is a collection of choices looking for problems. Decision-makers have already devised strategies to solve problems but are waiting for the appropriate time and place to implement them. Issues in the organization are available to be solved and the decision-maker must be quick to note any window of opportunity. This model is not a rational-comprehensive decision-making approach. In this approach, people do not set out to solve problems. By contrast, solutions usually search for problems. People work on problems only when a particular combination of problems, solutions, and participants in a choice situation makes it possible. Choices often just happen, with no clear connection to what participants want. They arise from dynamic organizational processes that are complex, highly contextual, and unpredictable, driven more by accident and timing than individual intention. This gives much of organizational life an almost chaotic appearance. In this model choices happen for no apparent reason. Outcomes are divorced from problems. People wander aimlessly in and out of decision arenas. Some decisions do get made in this model, but some problems do get solved. But it is largely due to the time-dependent circumstance of events, not to the rational effects of plans or goals. The original theory by Cohen, March, and Olsen (1972) is essentially a macro theory of organizations. Since we are using budget outcomes across time we can easily measure the macro level phenomenon of the garbage can theory.
The garbage can theory is founded on a radically different approach to choice than the rational/scientific model. Its premises are that decisions in an organized anarchy cannot be understood using the intentions of organizational participants, and imposing a rational explanation on organizational behavior can only distort what is really going on (Bendor et al., 2001). The authors have clearly abandoned the conventional rational approach to decision making.
The National Performance Review (NPR) provides an excellent example of the garbage can decision-making model. The NPR and its emphasis on managerialism and performance budgeting is also largely similar to other studies such as Grace, Rayner, and Nielsen Commissions, in the US, UK and Canada, respectively. Much of what the NPR recommends on performance budgeting can be found in reports ten years old, if not older, such as the Hoover, Glassco and Fulton commissions of the 1950s and 1960s (Savoie, 1994). This illustration of the garbage can model demonstrates its fundamental premise that oftentimes ideas are polished up and claimed as being new. Budgetary decisions in this approach are very much dependent on the current problems the organization faces, solutions that are feasible to implement, participants in the process and choice opportunities available. This theory clearly does not provide for rational-ordered outcomes.
The garbage can model can be illustrated by the following simple budget example. Consider an emergency meeting between, for example the secretary of Education and six departmental managers over the upcoming budget due for submission the following morning. In the meeting, these six department managers are requesting money to fund a proposed project they favor. However, the secretary has only enough money to fund one of the proposals. Perhaps the secretary has many proposals to consider and very little time to consider them; therefore, he or she will have to rely solely upon the information that the departmental managers provide. Furthermore, each of the proposals is similar and will solve the department's problem. As a result of this situation, the secretary finds himself or herself in, he or she uses a random guess to find the most suitable proposal. Therefore, with six departmental managers requesting money from one secretary for one project, the probability of a manager's proposal being chosen is approximately one-sixth. Budget makers as a result would use random guesses when making decisions under these circumstances.
If we apply this scenario to all government budgetary decision-making proposals, a startling implication occurs. Governments, which have to rely on policy entrepreneurs for advice, as we demonstrated above, will have budgetary policy outcomes which resemble a random process where the current outcome has no relationship to the previous outcome. In this process, budgetary decision-making is obviously not rational since outcomes are not considered by weighing each proposal's merits against one another. Furthermore, it is not incremental since there is no reliance on the previous spending decisions. The following section discusses the methods and data used to test the three internal change theories.
As previously demonstrated, there are three rival internal change models that illustrate the link between budget outcomes and decisionmaking. By definition, these models do not take into account systemwide exogenous factors such as elections, the role of Congress, Presidents, demographic shifts and macroeconomics impacts. At its simplest level the fundamental differences between the three internal change models are that garbage can budgeting is essentially a random process (no relationship between problems, solutions, participants, and choice opportunities), while budgetary incrementalism is a backward looking process (last period(s) spending decision), and rational choice budgeting is a forward looking process (systematically planning for the future using all available information). We first need to outline the methods that we are going to use to measure the evidence for these theories on budget outcomes.
Rational budgeting as previously outlined advocates a budget constraint based on the hypothesis that a government that runs a deficit is implicitly promising to generate sufficient surpluses in the future in order to repay the accumulated debt and interest. The rational budget balance theory requires that the budget be mean stationary. That is, it should exhibit a tendency to revert to a constant mean, not fly off in different directions. The long-run mean is expected to be zero since the budget should be in balance in equilibrium. If we look again at the deficit in Figure 1, we notice that it remained heavily out of balance during the 1980s, but reverted into balance in the late 1990s. Visually it appears that the process as previously illustrated in Figure 1 is stationary, but we test for this formally in the findings section of this paper.
As previously discussed, the theory of budgetary incrementalism is another rival model of decision-making in the internal change framework. This theory states that spending decisions at the present are entirely driven by the spending decisions made in the past (last month(s)' spending track). This is simply tested by measuring the spending series to see if it follows an Auto-Regressive (AR) process. For instance, an AR(I) process is where a disturbance or shock at the present period in a series is echoed in the error term of the previous observation, which gradually diminishes over successive observations. We have also added a revenue constraint (total receipts) to the incrementalism model. This accounts for "bounded rationality" in spending decisions. If revenues or receipts are lesser or greater than normal during a particular period, this will have an impact on the allocation. When they are less than available receipts, we expect that this will put a damper on the budget request.
The theory behind the garbage can model was formally dealt with as it relates to budgetary decision-making. To reiterate, the theory of garbage can budgeting states that decisions are made with no relationship to what was decided in the past or what may be decided in the future. It is a random process. A random walk model or a meandering process is a test of the garbage can theory. In terms of a budgetary example, garbage can spending decisions are made independently of each other. In essence, then, there is an equal probability of choosing either budgetary policy A or B. Thus, the best way to test for this is to see if the data generating process of public spending is stationary or not. Stationary simply implies that the data that is being measured does not drift chaotically. It reverts to a constant value. A non-stationary process, for example, would have the appearance of the Dow Jones stock index. If we follow the Dow Jones over several years we would notice that it may drift up for a while, then drift down, and so on. It does not usually head in any particular direction for an extended period. If stocks or stock markets only headed in one direction, everyone would know the true direction, and there would be no purpose for this investment. The easiest way to test for the garbage can model is to use a Dickey-Fuller and/or Phillips Perron tests for unit roots. A unit root in the data-generating process provides evidence of this non-stationary wandering process. If there is evidence of non-stationary budget outcomes, then outcomes are random, which is essentially what the garbage can budget predicts.
The statistical tests used in this study are those methods used m to measure econometric time series. Some of the methods used are the Augmented Dickey Fuller tests and Phillips Perron tests for unit roots. These tests determine if a time series, in this case budget outcomes are stationary. Non-stationary times series implies that it has a wandering quality. An analogy can be applied to flipping a fair coin 50 times. If we plot the outcomes of the coin toss, we might get a run of heads for a while or tails for a stretch. It is essentially a random outcome. The Phillips-Perron and Augmented Dickey Fuller tests are used to detect this random process. If we find that the calculated value of these test statistics is less than the critical value at the 95% confidence level, we can conclude that the budget outcomes wander and have a random quality. The opposite can be said if the calculated value is greater than the critical value. Therefore, the null hypotheses for the tests of the three theories are as follows:
H^sub 0^: If the Dickey-Fuller and Phillips-Perron calculated values are less than the critical value (at the 95% level), we can conclude that there is a garbage can process in budget outcomes.
H^sub 0^: If the AR test statistic is greater than the critical value (at the 95% level), we can say that there is budgetary incrementalism in budget outcomes.
H^sub 0^ If the Phillips-Perron and Dickey-Fuller test statistics are greater than the critical value (at the 95% level) we can say that there is a rational/scientific process in budget outcomes.
All three of these hypotheses demonstrate the extent to which we can measure with time series data the descriptive power of these three dominant theories of twentieth century budgeting.
Budget allocations to governmental programs and agencies provide a unique view of the political processes. Not only is the U.S. budget the primary instrument for the allocation of resources but such allocations serve as natural dependent variables: they are in convenient quantitative form; they are generated by a process that repeats itself in a regular cycle; and they are subject to public disclosure. Budget allocations have lent themselves to analysis using regression and other associated techniques (Su, Kamlet & Mowery, 1993). There is mounting evidence to suggest that different spending categories in the budget are influenced by varying political and macroeconomic factors (see Su et al., 1993; Baumgartner et al., 1998). Therefore, using only aggregated total outlays may not provide us with the necessary evidence of micro processes.
The database was taken from the Standard and Poor's DRI databank and runs monthly between 1968 and 1999 (DRI, 2000). The data was collected by DRI from the U.S. Treasury Department's monthly statement. This database, which has been used by political scientists to measure government growth, is unusual in the public budgeting literature since we are using monthly data whereas most of the existing work has modeled annual data.2
The traditional view is that budget outcomes can only be determined annually. We reveal several reasons why this may be inappropriate. LeLoup (1978) is of the opinion that incrementalism theorists especially have traditionally focused on annual appropriation decisions. This is an apparently obvious choice since yearly submission and review of appropriation requests are perhaps the most common feature of the budgetary process. However, annual appropriations are not the only important budgetary decisions and may not even be the most important. Along with annual decisions are a host of multi-year spending decisions that affect the budget. These include a number of substantive legislative and administrative decisions on the uncontrollable or statutory spending items that are not reviewed every year. Most decisions are non-annual, having a duration of more or less than one year. Operational decisions in this regard are important to consider. Included are decisions on the timing of spending, the amount of spending, carrying over balances, future funding, reprogramming, and transferring funds between accounts. Operations decisions are continual; they occur daily, weekly, monthly, quarterly and are the most specific type of budgetary actions.
The analytical choice of single years as the time frame precludes theoretical explanation of most of the budget for the United States. The impact of long-term spending decisions has become so great that one cannot hope to understand budgeting without considering them. As a result, the consideration of non-annual perspective is imperative; it is important to recognize here that selection of annual appropriations represents an analytical choice that virtually excludes these other important budgetary decisions. Incrementalism in the future should be modeled by considering time as important. By relying on annual estimates of spending, decisions have perhaps ignored changes that occur throughout the fiscal year.
In summary, there are two reasons why we have chosen to use monthly data instead of the standard annual data. First, with monthly data we are better able to be more precise about changes in budget allocations throughout the fiscal year. Annual data distorts these shortterm changes. Second, we have 384 observations for each time series. This large n factor gives much more statistical confidence in the findings, compared to studies of budgeting that use annual data having sometimes less than fifty observations, if they include the entire post-war period.
In this study, all budget outcomes were taken as a percentage of real GDP. The influence of the macroeconomy on spending is controlled by translating these disaggregated spending categories into a percentage of real GDP. Public spending is driven more by the macroeconomy; it will have a counter cyclical response to changes in GDP. Budget outcomes will increase slightly when the economy sags and decline when the economy improves. In addition, all the monetary variables in this study have been transformed into real terms by the GDP deflator. The effects of inflation have been controlled to ensure the series closely correspond to actual budgetary decision-making. An assumption is made in all the models tested in this study that decision makers are rational enough to discount or compensate for changes in the price level ((see Fischer and Kamlet (1984) for a similar conclusion)). It is also assumed that the inflation rate is clearly known by all participants and is without cost for them to process. We have not logged our dependent budget outcome variables since we are going to use an Auto-Regressive Conditional Heteroskedastic (ARCH) model if there is a detection of heteroskedasticity in the data generating process. The following section of this paper provides the findings for the three internal change decision-making models.
As mentioned the appropriate tests to measure stationarity for rational budgeting are the Phillips-Perron and Dickey Fuller tests for unit roots.3 A unit root found in the data generating process indicates a non-stationary process and consequently a rejection of rational choice budgeting. A non-stationary process cannot be balanced, being dependent on time. However, all the test results in Table 1 provide strong support that there is not a unit root found in the deficit variable. The Phillips Perron test statistic for the deficit was -17.60. The critical value for the deficit at the 0.01 significance level was -3.45. This provided an acceptance of rational budgeting and a rejection of a unit root in the data generating process. Similar tests were performed on the national debt and the results reject a process that meanders. Governments do indeed worry about paying off their debt and accumulated deficits. They act rationally since they do not want to default on an unsustainable debt and are attempting to balance the budget over the long-term.
In summary, no evidence was found for a unit root in the deficit and debt categories. The significance of these findings demonstrates a balancing process at play here where there were deficits during certain economic times, such as the 1980s. During these times, politicians were constrained from the lack of tax revenue the government collected. However, during more prosperous periods such as the late 1990s, there was increased tax revenue, which made it much easier for the government to balance its budget (Figure 2). The same analysis can be applied to the debt. As a percentage change the debt does not wander. It has increased by approximately the same percent month over month. There is also evidence in support of budgetary incrementalism.
Reported in Table 2 are the test results for budgetary incrementalism and punctuated budgetary incrementalism. In the dissagregated spending categories tested there were 13 out of 18 categories with budgetary incrementalism and/or punctuated incrementalism. We further notice that commerce, education, health, and justice have a revenue constraint on incrementalism - cutting back on spending when it is greater than normally allowable. These results demonstrate that most of the spending categories follow a process where present spending is an extrapolation of previous spending decision(s). These results confirm the existing evidence found in the literature that supports budgetary incrementalism in outcomes. There is also a small amount of evidence of punctuated incrementalism with three of the 13 categories displaying an ARCH process. An ARCH process is a technique that measures volatile data. This is especially relevant when, on occasion, crises (e.g., oil shock) to the budget system interrupt stability (incrementalism). After the crisis subsides, spending returns to its original incrementalist path. ARCH models are able to capture punctuated incrementalism by measuring the sudden shock (punctuation) to the budget system. The following model tested has the least support of the three descriptions of budgetary decisions tested in this study.
Finally, Table 3 presents the results of garbage can budgeting. Budget outlays were converted into monthly percentage changes and as a percent of GDP. We want to take into account changes month over month and the impact of the economy on outlays. We notice that when we use either the Augmented Dickey Fuller or Phillips Perron tests for unit roots there is an overwhelming rejection of garbage can budgeting. Specifically Table 3 demonstrates the rejection of a random walk model. The spending category, labeled veterans, has a t-statistic of -5.16, and the critical value of -3.45 at the 0.01 significance level, a definite rejection of a unit root and consequently garbage can budgeting. Spending does not appear to wander all over the block as the garbage can model dictates. There is some order to the process since all the test statistics presented in Table 3 exceed the critical value by such a large magnitude. The significance of this finding indeed demonstrates some order and direction in the process. The evidence found in this study does not support an organized anarchy perspective.
After reviewing the results for the internalized change models, we have found the most support for budgetary incrementalism and rational budgeting. There was no support found in this study for garbage can budgeting. As a result, the process appeared to be much more rationally oriented than garbage can theorists assume. The last section of this paper analyzes the results from the internal change models.
SUMMARY AND CONCLUSION
In this study, we have demonstrated three dominant internal change paradigms of budgetary decision making for the twentieth century. These theories consider the primacy of factors that occur within the control of government as the most important determinate of outcomes, exogenous factors are not considered as important in these approaches. The first approach illustrated was the rational/scientific. The second approach budgetary incrementalism was the challenger to this model. Finally, the century rounded off with the garbage can theory. Each of these approaches was tested in this study with a new dataset (monthly rather than the traditional annual data) not previously modeled in the literature. The empirical evidence indicates that budget outcomes follow incrementalism at the micro level (disaggregated spending) and rational/scientific at the macro level (debt, surplus or deficit). There was no evidence that the system is somewhat anarchical as advocated by the garbage can theory.
The evidence found in this study indicated that rational process and incrementalism are the dominant paradigms for the twentieth century. The significance of these findings demonstrates that the dominance of the garbage can theory, at least as it related to the time series model done here, may not deserve its reputation at the close of the twentieth century. The evidence seems to suggest that there is both a long run and short run view of budget process outcomes. The short run view holds that budget decision makers in the United States were constrained due to the poor performance of the macroeconomy and other political factors that may not have created an environment suitable for balancing the budget. Therefore, the incremental theory seems plausible when decision makers are operating under such constraints. However, decision makers also have a long run view of the budget, they know they will eventually have to pay off the accumulated debt, so they will act rationally and plan to balance the budget over the long run. If politicians only listened to the preferences of the public as expressed in the public opinion polls (as demonstrated at the beginning of this paper), the U.S. may never have achieved balanced budgets in the late 1990s.
However, there are some shortcomings of this study that should be addressed. The first is that we are using time series data. Time series data has a long history of being used to test budgetary incrementalism. This type of data may not be as suitable for testing the garbage can theory since most of the existing work in the area uses cross sectional data. Perhaps a future study might look at this process across both space and time using pooled times series analysis. The second shortcoming is that we are using budget outcome data after it has been authorized to agencies and departments. A more suitable type of data would be appropriations sent to Congress, which may provide us with a closer picture of the planning process. Third, in the rational model we have only considered aggregated categories such as the deficit and debt. For garbage can model and budgetary incrementalism we have used disaggregated outcomes. We may be providing too harsh a restriction on the latter two models. The version of rational budgeting we test here considers the budget process similar to a market clearing to reach equilibrium of budget balance. However, in spite of these criticisms this paper does provide a start in an exploratory analysis of testing alternative rival conceptions of budgetary decision-making.
The findings of this paper also demonstrate the problems in the current literature of using only one paradigm to explain outcomes. Maybe future research should look at a combination of approaches possibly across different countries. The broader significance of these findings demonstrates that budgeting at the national level in the United States possesses a great deal of rationality. Evidence found here supports rational budgeting and provides an alternative perspective to traditional notions such as incrementalism. Rational budgeting may provide a reason why the federal government in the late 1990s faced unprecedented budget surpluses after almost 30 years of persistent deficits.
J. OF PUBLIC BUDGETING, ACCOUNTING & FINANCIAL MANAGENENT, 15(2), 251-274 SUMMER 2003
Copyright (C) 2003 by PrAcademics Press
1. The data represented in Figures 1 and 2 have been seasonally adjusted by the U.S. Census Bureau X-11 method.
2. The data represented in Figures 1 and 2 have been seasonally adjusted by the U.S. Census Bureau X-11 method.
3. Similarly to garbage can budgeting the deficit and debt categories were converted into monthly percentage changes to measure overall change.
4. These disaggregated spending categories represent all the dependent variables tested for garbage can and incrementalism models.
Baghestani, H., & McNown, R. (1994). Do revenues or expenditures respond to budgetary disequilibria? Soulhern Economic Journal, 61, 311-322.
Baumgartner, F.R., Jones, B.D., & True, J.L. (1998). Policy punctuations: U.S. budget authority, 1947-1995. The Journal of Politics, 60, 1-33.
Bendor, J., Moe, T.M., & Shotts, K.W. (2001). Recycling the garbage can: An assessment of the research program. American Political Science Review, 95, 169-190.
Berry, W.D., & Lowery, D. (1987). Understanding United States government growth: An empirical analysis of the postwar era. New York: Praeger Publishers.
Braybrooke, D., & Lindblom, C.E. (1963). A Strategy of decision: Policy Evaluation as a social process. London: Collier-Macmillan Limited.
Cohen, M., March, J., & Olsen, J. (1972). A garbage can model of organizational choice. Administrative Science Quarterly, 17, 1-25.
Davis, O.A., Dempster, M.A., & Wildavsky, A. (1974). Towards a predictive theory of government expenditure: US domestic appropriations. British Journal of Political Science, 4, 419-452.
DRI Basic Economics. (2000). DRI basic economics quarterly database. Toronto: CHASS Data Center, University of Toronto.
Fenno, R.F. (1966). The power of the purse: Appropriations politics in congress. Boston: Little, Brown and Company.
Fischer, G.W., & Kamlet, M.S. (1984). Explaining presidential priorities: The competing aspiration levels model of macrobudgetary decisionmaking. American Political Science Review, 78, 356-370.
Hamilton, J.D., & Flavin, M.A. (1986). On the limitations of government borrowing: A framework for empirical testing. American Economic Review, 76, 808-819.
IPOLL (2001) IPOLL online database: The Roper Center for Public Opinion Research. Storrs: The University of Connecticut.
Kamlet, M.S., & Mowery, D.C. (1992). The first decade of the congressional budget act: legislative imitation and adaption in budgeting. In A.C. Hyde (Ed.). Government Budgeting: Theory, Process, and Politics (pp. 119-134). Pacific Grooves, CA: Brooks/Cole.
Kingdon, J. (1984). Agendas, alternatives, and public policies. Boston: Little Brown.
LeLoup, L.T. (1978). The Myth of Incrementalism: Analytical Choices in Budgetary Theory. Polity, 10, 488-509.
March, J., & Olsen, J. (1976). Ambiguity and choice in organizations. Bergen. Norway: Universitetsforlaget.
March, J., & Olsen, J. (1989). Rediscovering institutions: The organizational basis of politics. New York: Free Press.
March, J., & Weissinger-Baylon, R. (1986). Ambiguity and command: Organizational perspectives on military decision making. Marshfield, MA: Pitman.
Savoie, D.J. (1994). Thatcher Reagan Mulroney: In search of a new bureaucracy. Toronto, Canada: University of Toronto Press.
Schick, A. (1990). The capacity to budget. Washington, DC: The Urban Institute Press.
Schultz, K.A. (1995). The politics of the political business cycle. British Journal of Political Science, 25, 79-99.
Su, T.-T., Kamlet, M.S., & Mowery, D.C. (1993). Modeling U.S. budgetary and fiscal policy outcomes: A disaggregated systemwide perspective. American Journal of Political Science, 37, 213-245.
Wildavsky, A.B. (1964). The politics of the budgetary process. Boston: Little, Brown.
Wood, B. D. (2000). The federal balanced budget force: Modeling variations from 1904-1996, Journal of Politics, 62, 817-845.
Christopher G. Reddick*
* Christopher G. Reddick, Ph.D., is an assistant professor of public administration at the University of Texas at San Antonio. His teaching and research interests are in public budgeting and finance.…
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: Budgetary Decision Making in the Twentieth Century: Theories and Evidence. Contributors: Reddick, Christopher G. - Author. Journal title: Journal of Public Budgeting, Accounting & Financial Management. Volume: 15. Issue: 2 Publication date: Summer 2003. Page number: 251. © PrAcademics Press, Florida Atlantic University Spring 2007. Provided by ProQuest LLC. All Rights Reserved.
This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.