23 Performance and Budgeting Under the Separation of Powers

Marc Robinson
Published Date:
October 2007
  • ShareShare
Show Summary Details
Philip G. Joyce1

Many nations are attempting to change their budget processes to focus more on the connection between resources and results, rather than focusing budget decisions solely on inputs or activities of government. In this context, the role of the legislature in budget reform can be important, particularly in those countries where the legislature plays a significant independent role in resource allocation. This chapter focuses on three key obstacles to performance-oriented budget reform in these countries; namely, that:

  • the fragmentation of power that is characteristic of independent legislatures often leads to ambiguous or conflicting policy objectives

  • legislatures often base decisions on information that has little to do with overall performance, relying instead on either input data or on anecdotal information

  • legislatures frequently do not conduct detailed programmatic oversight.

Against this background, the chapter considers changes that may promote more attention by legislatures to performance.

Government programs and institutions exist to further societal ends. Many government budget processes, however, have focused on inputs—resources required—or outputs—the work performed—as if they were the end products of government. Increasingly, in countries throughout the world, this trend has been countered through the establishment of systems that attempt to connect the resources used with the effects that those resources bring about. This type of a system is part of an overall reform movement in which decisions within government—by elected officials and government employees—are informed by considering the performance implications of those decisions. Governments with budgeting systems in which performance information is used apply many analytical tools—including cost allocation, performance measurement, and evaluation—in the process of allocating and managing resources.

Multiple positive effects can occur as a result of making explicit linkages between results information and the allocation of resources. First, such a linkage fosters a more efficient use of government resources, since questions about the success of government activities are tied to the allocation of resources. Second, a performance-focused budget system can help demonstrate to key stakeholders—who fund government activities and/or receive public services—that they are receiving “value for money” Third, when either additions or deletions from the government budget need to be made, these can be targeted in a way that government performance is optimized. Government resources are scarce; therefore the way in which these resources are allocated is crucial to maintaining some degree of domestic consensus and reaching societal objectives.

The challenge of ensuring effective use of scarce resources has led to a number of specific past and current reforms, designed to more closely link performance data to the allocation and management of resources. In the US, these reform efforts date back more than 50 years, and include such reforms as program budgeting (and its federal manifestation, the Planning, Programming, and Budgeting System—PPBS), zero-base budgeting, and management by objectives (see Schick, 1966; Harkin, 1982). More recently, all three levels of government (federal, state, and local) have been engaged in renewed efforts to establish closer linkages between performance evidence and the budget. At the federal level, the main impetus was the passage of the 1993 Government Performance and Results Act (GPRA), requiring strategic plans and performance measures for all federal agencies. The Bush administration has carried that reform a step further, moving beyond the production of data, to emphasizing the use of performance information in the budget process (see Joyce, 2004). Some US state governments—most notably Washington, Virginia, Utah, Michigan, Louisiana, and Texas—have progressed even further than the federal government in the use of performance information for budgeting and management.2 Local government experiences are harder to summarize (there are approximately 88,000 local governments in the US), but organizations such as the International City/ County Management Association have been encouraging their member governments to focus more explicitly on focusing upon and assessing performance.

Similar reforms are being pursued in both developed and developing countries all over the world. Many OECD countries have attempted to reform their budget processes to incorporate performance measurements. Four countries in particular are frequently mentioned as being at the forefront of public budget reform efforts: Australia, Canada, the UK, and New Zealand (Premchand, 1999). In addition to developed countries, however, developing countries have also been pursuing these reform efforts, supported by international institutions such as the World Bank and the International Monetary Fund. In the interest of space, the experience of these countries will not be reviewed in detail here, but all of them have in common an emphasis on: planning; measuring results; and using performance information to allocate resources.

What is the role of the legislature in budgeting?

The type of governmental budget system will affect substantially the manner in which the legislature deals with the budget. Depending on the country, the legislature may play a crucial role in determining the level of funding for public agencies and programs. In some nations, however, the legislature is, in effect, an extension of “the government” or plays a role of providing a check on what the government or governing coalition does. Still others occupy some middle ground between these two extremes.

Pelizzo and Stapenhurst, in a recent World Bank Institute paper, characterized governmental systems of 48 different countries as presidential, semi-presidential, or parliamentary (see Table 23.1). Presidential systems, which include countries such as the United States, Indonesia, Brazil, and Chad, generally have an independently elected chief executive and legislature. In the parliamentary systems, which include the Westminster-style countries (such as the UK, Canada, Australia, and New Zealand) in addition to countries such as Germany, Turkey, and the Czech Republic, the party (or parties) in control of the parliament are actually in charge of the executive. These countries normally have a Prime Minister, who is the head of the executive and a member of parliament. The semi-presidential systems, such as France, Romania, and Niger, represent some hybrid of these two forms. In these instances, there is frequently both a Prime Minister and a President who has some independent constitutional authority (Pelizzo and Stapenhurst, 2004).

While governments in any of these types of systems may attempt to adopt a more performance-oriented budget process, legislative support is not equally important in each country. The countries of primary interest in this chapter, therefore, are those in which the legislature exercises independent control over the budget. While it is tempting to assume that this sort of legislative independence operates on a continuum where presidential systems are at one end (with substantial legislative control) and parliamentary systems are on the other (with virtually none), research into legislative budgetary power suggests otherwise. Another recent study, by Lienert, in attempting to characterize legislatures in terms of the level of control that they exercise over the budget, divided governments into five separate forms—presidential, semi-presidential, parliamentary republics, non-Westminster parliamentary systems, and Westminster parliamentary systems. It then characterized the separation of powers among these countries. At the two extremes, the results were not surprising. The presidential systems had the most substantial separation of powers between the branches, while the Westminster systems had none. There was virtually no difference, however, between the other three cases—each had similar degrees of legislative/executive separation (Lienert, 2005).

The Lienert study goes a step further, however, in attempting to characterize these countries in terms of the separation of budgetary powers. The study looks at several factors in determining the extent of legislative separation, the most important of which is the extent of formal power to make changes in the budget. Significantly, this analysis finds no consistent relationship between the extent of political separation of the executive and extent of budgetary power exercised. The presidential systems and the Westminster systems still prove the extreme cases, but the study finds that (for example) the average non-Westminster monarchy has a greater degree of independent legislative budget power than a typical semi-presidential system. It concludes, not surprisingly, that the form of government doesn’t necessarily matter, but rather a series of country-specific characteristics. The study summarizes factors influencing legislative budget independence as follows:

  • Political factors. If there are fewer strong parties in the legislature, or if the government results from a parliamentary minority, there tends to be more legislative power exercised. This is also true of presidential systems with many political parties, or countries with bicameral legislatures.

  • Constitutional and legal limitations on executive budgetary power. In many countries, legislatures have formally limited executive power, usually in response to a perceived excessive exercise of power by the executive. In parliamentary systems, laws tend to explicitly enhance the power of the executive. Alternatively, constitutional changes may specifically curb or enhance the legislature’s budget power. Further, laws may create greater staffing capacity in the legislature, or may affect the level of debate on the budget.

  • Non-legal constraints. In some countries, there is a practical requirement to gain agreement among many different political parties prior to the enactment of the budget. This happens, for example, in countries ruled by coalition governments. Country-specific cultural (for example, a history of a strong monarchy) or political (for example, the existence of legislative term limits) realities may also impose idiosyncratic constraints.

Table 23.1Form of government and number of parliamentary oversight tools
No. parliamentary oversight tools
Form of government4567
PresidentialCôte d’voireNicaraguaBeninCosta Rica
Guinea BissauCroatia
JamaicaCzech Republic
United KingdomGreece
Source: Pelizzo and Stapenhurst (2004, Table 5, p. 6).
Source: Pelizzo and Stapenhurst (2004, Table 5, p. 6).

Despite these elaborations on the typology, the study confirms the presumed strong distinction between the presidential and Westminster-style systems. On a scale of 0 to 10 (with 0 representing the weakest independent legislative budget power and 10 representing the strongest) New Zealand, Ireland, and Greece (0) have the weakest legislatures, with Sweden (9) and the United States (10) occupying the other end of the continuum among the 28 countries studied. The other Westminster countries—Australia, Canada, and the UK—all prove weak in terms of legislative independence. Most countries occupy a middle ground, with “budgetary independence” scores between 4 and 7—these include Argentina, Bolivia, Hungary, Italy, and Japan.3 (See Table 23.2 for a full listing of these countries according to legislative budgetary powers.)

These findings are significant because, while the level of independent budgetary powers varies quite a bit, in the majority of non-Westminster countries it is reasonable to think in terms of some “separation of powers” between the branches. Where such a separation exists, it is appropriate to consider what effect that separation has on various aspects of budgeting and management. For the purposes of this review, this means considering what effect the separation of powers would have on the movement of countries toward a performance-informed budgetary process.

The observations that follow in this chapter, therefore, apply most specifically to those governments where the legislative branch exercises substantial independent control over the budget. In these countries, the legislature has a particularly important role to play in establishing the environment for government performance, making decisions concerning how resources are allocated, and providing important follow-up that attempts to ensure that agreed-upon performance levels have been achieved. For this reason, it is important to understand precisely the manner in which these legislatures can contribute to (or detract from) performance, and how interested parties can think about the necessary incentives for these legislatures to better enable the effective use of scarce resources.

It is also worth noting that the choice of a political system in which the legislature has more independent budgetary power is just that—a choice, made for specific reasons and with political ramifications. While it may be much easier to set clear direction and unified policy for a political system without such a separation, that does not necessarily mean that the likely greater efficiency and clarity in policy is worth the price that might be paid in a given country. Many nations (for reasons of history, culture, or political theory) consciously choose messier power-sharing arrangements over arrangements in which direction for the government is set (at least at a given point in time) by one coalition or party. While it would be, in one sense, easier to operate in a system that encouraged greater goal clarity, that does not mean that the benefit of making such a change would be worth the cost. The purpose of this chapter, therefore, is not to argue that political systems with concentrated power are “better” than those with separated powers, but rather to identify the ways in which independent legislatures often impede results-oriented budgeting. The assumption is that understanding these limitations, and the reasons for them, can assist in devising solutions, as appropriate, for the challenges created by separated executive and legislative power.

Table 23.2Indices for the legislature’s budget authority in 28 countries
Type of government1.

Medium-term framework

Amendment powers

Time for scrutiny of budget

Technical support legislature

Restrictions during execution
Total index
BelgiumParlia. monarchy031004
DenmarkParlia. monarchy031015
GermanyParlia. republic031004
IcelandParlia. republic031026
ItalyParlia. republic231017
JapanParlia. monarchy031127
NetherlandsParlia. monarchy031116
New ZealandWestminster000000
NorwayParlia. monarchy031026
SpainParlia. monarchy011013
SwedenParlia. monarchy131129
TurkeyParlia. monarchy011013
United Kingdom Westminster000011
United StatesPresidential1322210
Source: Lienert (2005, Table 2, p. 23).
Source: Lienert (2005, Table 2, p. 23).

One other contextual observation seems in order. Most public budget processes in countries with independent legislatures have clear (if not always smoothly functioning) stages of the budget process, any of which can be affected by the separation of powers (see Lee et al., 2004, ch. 3):

  • budget preparation, as agencies or ministries develop internal budget allocations and requests that are eventually (after some give and mostly take) integrated into a proposed (or recommended) budget

  • budget approval, at which point the budget proposal is acted on and officially approved by the relevant governing body (a parliament, or a state legislative body)

  • budget execution, as agencies/ministries implement the budget, within the constraints established in the budget law(s)

  • audit and evaluation, as agencies/ministries and auditors/evaluators decide (after the fact) what the effects (financial and performance) of budgetary activities have been.

If a given budget process is to become more informed by performance data, two factors must be considered simultaneously at each stage of the budgeting cycle. The first is the availability of appropriate information—on strategic direction, results, and costs—in order to make budgeting more results-focused. The second is the actual use of that information to make decisions (Hilton and Joyce, 2003; Joyce, 2004).

It is important that these questions be asked throughout the budget process, since there are multiple decisions at many stages of that process that can be illuminated by performance information. Often the focus is only on decisions that are made by the Ministry of Finance or the Budget Office, and on approval of the budget by the legislature, as if these are the only places where “budgeting” occurs. This encourages an overly narrow view of the budget process. It, for example, ignores the substantial discretion that ministries may have in allocating and managing resources in the budget execution stage, or the important role played by performance auditors in the evaluation stage. Decisions made at these stages are also about budgeting, in the sense that they directly or indirectly affect the allocation of resources. Thus, they are important to focus on in our understanding of how performance is affected by the budget.

A full review of all of the ways that performance information can be produced and used at each of these stages is beyond the scope of this chapter.4 For our current purposes, the important thing is to understand that the legislature—in different ways and to varying degrees—can affect the level of effective integration of performance information at all stages of the budget, not just at the point of approval. That is, the legislature’s orientation towards the budget affects the preparation of the budget in the executive branch, the implementation of programs, and the evaluation of government activity after the fact. The next section of this chapter will outline the ways in which legislatures can particularly impede development of a performance-oriented budget process.

Greatest obstacles legislatures create for performance-informed budgeting

There are many ways in which legislatures can promote or impede the ability of their respective governments to emphasize performance. The most important of these fall into three categories. First, laws passed by legislatures are frequently quite unclear in setting overall goals and direction for programs, leaving ministries and program managers in a position of having to establish objectives, and strategies for achieving them, without clear direction on goals and priorities. Second, legislatures often base budget decisions on information that has little if anything to do with overall performance. They may instead focus on anecdotal information, or information with limited relevance to results (for example, the distribution of inputs). Third, legislatures often fail to engage in oversight designed to focus on performance in a comprehensive way, choosing instead to either ignore the oversight function entirely or to focus attention on reacting only to allegations of wrongdoing.

In addition to these three obstacles, of course, one way that legislative bodies can impede performance is through the excessive use of budget controls. Substituting accountability for the use of inputs for accountability for results can create counterproductive incentives. For example, the US Congress routinely prohibits federal agencies from closing particular facilities, or requires them to open new ones, without regard to the performance implications of these actions. It also frequently imposes personnel floors or ceilings on agencies that prevent these agencies from exercising flexibility in managing their human resources. These broader performance constraints are somewhat beyond the scope of this chapter, but should be kept in mind by governments thinking about the relationship between legislatures and the creation of an environment that is supportive of performance.

Establishing and promoting unclear policy objectives

The development of appropriate performance information, whether it is to be used for budgeting or any other purpose, implies that public entities need to know what they are supposed to accomplish. Malcolm Holmes, who was an architect of the Australian budget reforms, noted that a key condition for results-oriented management in government was “clarity of task and purpose” (Holmes, 1996). The desire for goal clarity as a precondition for effective performance is the reason that most governments that have embraced a greater performance orientation have started with some sort of strategic planning effort. Strategic planning enables decisions to be made that establish clear direction for government programs and ministries. This is often quite difficult to carry out in practice, particularly in countries with a horizontal and vertical diffusion of authority, responsibility, and political decision-making. It is relatively easier in parliamentary systems (especially, as noted, of the Westminster variety), where the majority party or coalition actually runs cabinet ministries.

Despite its difficulty, however, strategic planning is an important focus of budget reformers. It explicitly establishes the context in which performance and dollars can be linked. In order for any organization to evaluate either its performance, or its use of resources in pursuit of that performance, it must first know what it intends to do. Legislatures can impede the development of effective strategic planning, and the coincident establishment of policy direction, because of a tendency to send unclear and ambiguous—or even contradictory—signals concerning what agencies or programs are to accomplish.

Legislatures can produce vague and sometimes conflicting legislation for several reasons. Put simply, the process of establishing legislative coalitions may demand it. Further, legislatures may either: lack the expertise required to specify precise actions in advance and/or conclude that changing circumstances require that they delegate some power to executive agencies. On the first of these points, legislation is normally produced as a result of carefully crafted compromise. Making laws more explicit often has the effect of revealing clear winners and losers. While this may be desirable in one respect (it promotes transparency, for example), it may tend to make the law in question more difficult to pass. This is not a recent phenomenon. In 1949, Paul Appleby argued in Policy and Administration that public officials were necessarily involved in policy-making (rather than simply carrying out out policy in a more ministerial fashion) because legislatures tended to pass vague legislation. This leaves executive branch agencies to “fill in the details/’ meaning that the ends of government themselves are partially under the control of the bureaucracy.

This is realistic, largely because of the technical limitations of legislative bodies. Even where legislatures operate through relatively specialized committees, these committees and their staffs do not possess the level of expertise that resides in administrative agencies. Particularly in departments where laws require scientific or other technical judgment, the power to specify implementation details may be delegated to appointed officials. For environmental policy, for example, it may be essential to delegate the specifics of regulation to administrative agencies, since the expertise required to determine how to translate “clean air” into a specific permissible level of a given pollutant may only reside within the bureaucracy. This is not just a matter of expertise. Given the large scope of legislative activity, and the limited time and the relative small size of legislative staffs to assist in lawmaking, it is not surprising that laws are relatively vague. Finally, legislation must frequently permit flexibility, in order to be able to adapt efficiently to unforeseen circumstances.

As long as there is consistency between the intent of a law and administrative interpretation of that law, the delegation of power doesn’t necessarily create constraints to establishment of objectives, and (subsequently) performance measures.5 Often, however, vague laws and subsequent administrative interpretation yield legislative reactions to those interpretations. These reactions might be unified—that is, reflective of the legislature as a whole. More often, they are fragmented (generated from individuals or committees within the legislature). Legislative response can in turn create substantial ambiguity for the ministry or agency involved, because it is not clear whether the agency is to follow: (1) the law as written; (2) the law as interpreted by the administrative agency; or (3) the law as interpreted by the administrative agency and then reinterpreted by the legislature or some sub-set of the legislature. If it is truly not possible to develop performance measures independent of clear objectives, certainly the objective-setting process under the separation of powers can impede measurement of performance.

In an evaluation of the management effectiveness of US federal agencies conducted between 1997 and 2002, lack of goal clarity was identified as an issue in numerous agencies. These included the US Customs Service, which is buffeted between twin goals of promoting the efficient movement of goods into the country and preventing harmful substances from entry. Also included are agencies such as the Forest Service (conservation versus harvesting forests), the Federal Aviation Administration (promoting timeliness versus security), and the Immigration and Naturalization Service (enforcement of immigration laws versus providing services to immigrants). In each of these cases, agencies are left in the position of having to deal with multiple (and often conflicting) goals, with little clear direction in terms of how to resolve the inevitable ambiguity created in the legislative process. Lack of goal clarity, and lack of clearly established priorities between goals, often makes it difficult to identify performance measures that can be used to evaluate progress (Ingraham et al., 2002, pp. 94–102). Notably, in reviewing lessons learned from reviewing management effectiveness at all levels of government in the US, Ingraham, Joyce, and Donahue identified four conditions under which management effectiveness could make the most difference in delivering results; one of these conditions was “clear purpose and mission” (Ingraham et al., 2002, pp. 123–4).6 This is another way of saying that without a clear articulation of purpose, it is difficult to manage effectively.

Making ill-informed budget decisions

Even if agencies have clear objectives, this does not by itself guarantee that appropriate performance measures relevant to those objectives will follow. In addition to setting appropriate objectives, then, a related condition for performance-informed budgeting is the collection of appropriate performance and cost information, and use of that information in resource allocation and resource management decisions. This creates problems beyond the frequently-cited conceptual challenges of defining relevant indicators. Most public sector organizations reasonably resist being held accountable for outcomes, since they are influenced by so many factors that are beyond their control. Further, there are conceptual problems with measuring costs—many public organizations cannot track how much it costs to deliver an output, largely because of problems with allocating indirect costs (Anthony and Young, 1984). It is that much harder to go the next step to measuring outcomes.

Even if these problems related to producing performance and cost data could be overcome, however, legislatures frequently do not have incentives to demand appropriate performance and cost information when making decisions on funding for agencies or programs. There are several specific problems that inhibit linkages between performance and budgeting in this regard. Perhaps the most significant is that legislatures frequently see no observable connection between having—and acting on—information on results and the electoral benefits that may be paramount in their minds. This point is particularly salient in cases where legislators need to stand for election frequently. Political scientist Morris Fiorina, writing about the US Congress, argues that the members of that legislative body (and particularly the House of Representatives, who must stand for election every two years) have little incentive to spend their time in legislating (by this, he means engaging in lawmaking designed to promote national interests), and instead focus on more localized pursuits with more a more explicit short-term electoral connection. Fiorina highlights two such localized activities: casework, which represents efforts to intervene on behalf of constituents with administrative agencies; and “pork barreling,” which represents efforts to bring resources home to individual legislative districts (Fiorina, 1989).

The latter activity—pork barreling—is antithetical to a broad performance orientation on the budget. The definition of policy success under pork barrel practice is decidedly input-focused, rather than focused on results. For example, the typical legislative press release targeting the delivery of local benefits will not discuss the broad policy outcomes to be achieved from the expenditure, but will normally present the expenditure as more or less an end in and of itself. In fact, in their book, Perpetuating the Pork Barrel, political scientists Robert Stein and Kenneth Bickers (1995) argue that “[a]t the heart of most policy subsystems is a set of government programs” and that these programs are “bundled together” in ways that address the preferences of the actors in the sub-system. These actors include interest groups and executive branch agencies (two-thirds of the traditional “iron triangle”), but also include legislators. For legislators, the key preference is re-election, and a key path to re-election is the delivery of benefits—usually defined as jobs and money—to the home district.

But, while apparently inefficient, pork barrel spending is not without its justification. A recent book on the practice of pork barrel spending in the US argued that, as much as pork barrel spending may be reviled (and it is certainly easy to criticize on a case-by-case basis), it has the positive effect of fostering coalitions to pass legislation that otherwise might not achieve a majority. To the extent that this legislation may deliver general, rather than particularistic, benefits, these benefits may be worth the cost of a little pork (Evans, 2004). Further, it may not actually represent that much money—informal estimates are that “pork” represents less than 1 percent of the US federal budget.

In terms of impediments to performance, there are less visible, but probably more important, problems than the direct attempts to focus attention on inputs. Among these are misconceptions concerning the appropriate ways to measure performance, or to understand its relationship to budgeted amounts. Unfortunately, many legislators may be inclined to either not use performance information at all, or to use it in ways that do not really relate to performance. This last problem has at least two manifestations. First, there is the simple, but incorrect, approach (allegedly embraced by some members of the US Congress) that argues that the way to use performance data for budgeting is to “simply” take money from those who fail to meet performance targets, and give more money to those who meet targets.7 While this may sound good, it relies on heroic assumptions, one of them about the causal link between money and results. For many programs, we do not know whether the failure to achieve performance targets is an argument for more money (the program could not achieve its objectives because it was insufficiently funded) or less (it is an ill-conceived or badly managed program). In fact, for any program, sorting out the contribution of funding versus other factors requires a full understanding of the logical relationship between inputs, outputs, and outcomes, taking into account both internal and external factors that influence performance. Further, budget decisions are appropriately influenced by other (non-performance) concerns, such as relative public priorities, unmet needs, and equity concerns, to name three.

Second, there is the tendency to assume that anecdotal information is the same as performance information. It may be true that an extremely large number of anecdotes might equate to some overall sense of agency or program performance. Often, however, legislators rely on sporadic anecdotal information, or data limited in geographic scope. A legislative leader in a large US state once noted that he was sure that the state Department of Transportation was doing a bad job “because there are lots of potholes in my district.” Without doubting the veracity of that statement, this piece of data by itself neither yields information about how the Department is performing overall, nor does it in any way indicate what level of resources would need to be provided in order to improve performance. In the US federal government, Kenneth Kizer (Undersecretary of the Department of Veterans Affairs in the first Clinton administration), decried what an article later called the “tyranny of anecdotes.” In this case, individual stories about the health of military veterans prevented members of Congress from focusing on the big picture of veterans’ health—rather than the more episodic and localized issues related to particular states and Congressional districts. Kizer argued that “[i]f you govern by anecdote, you’re chasing your tail because you’re continually chasing that one person. You can’t judge the process if you’re looking at a series of one” (Laurent, 2002).

Successful performance-informed budgeting occurs only when those involved in the budget process move beyond mere production of information to the use of information to make decisions about resource allocation and management. This can only occur if all budgetary actors have effective incentives (and resources) to collect and use information. In fact, the incentive question is probably the most important one on which to focus on in analyzing the extent to which performance information will actually be used as an input in the various stages of budget decision-making. Aaron Wildavsky, writing in 1969 about an earlier US reform, argued convincingly that legislators would not use performance data because it was supplied; rather it was crucial to focus on the demand for analysis. Giving legislators data that they do not want and they did not ask for is not an effective means toward getting those data used to make legislative decisions (Wildavsky, 1969).

Inadequately focusing oversight

As legislatures recognize the extent to which power to make specific policy out of vague legislative direction is delegated to administrative agencies, they often develop a greater focus on oversight. Development of an oversight orientation allows legislatures to couple the grant of substantive policy-making with the capacity to discover what the executive branch has done with that power. In the US, for example, the recognition that administrative agencies were involved in policymaking led to specific efforts on the part of the Congress to provide checks on federal agencies. This effort was characterized by specific legislative reactions to bureaucratic power (such as the Administrative Procedures Act and the Legislative Reorganization Act, both passed in 1946) as well as a more general orientation toward greater oversight (see Rosenbloom, 2000). In fact, the 1946 Legislative Reorganization Act directed Congressional committees to exercise “continuous watchfulness” of executive branch agencies (Aberbach, 1990).

In nations with strong independent legislatures, oversight represents the key tool that legislatures can use to maintain control over bureaucratic agencies. Problems discovered during the oversight process can be rectified through subsequent legislative, or other, action designed to influence bureaucratic behavior. Legislatures have responded to this need for increased oversight by establishing organizations designed to provide information to them on the operations of government programs. One of the reasons for the establishment and subsequent expansion of the mandate of the General Accounting Office (GAO, now the Government Accountability Office) in the US has been the desire on the part of the Congress for increased information on the operations of federal programs and agencies. Prior to around 1970, the GAO had mainly been responsible for conducting financial audits. It has moved substantially into more “programmatic” audits over the past 30 years. This also has been a trend at the state level. A recent review of “performance auditing” in the states cited 12 examples of particularly strong performance audit capabilities in US state governments.8

Outside the US, virtually all countries use some oversight tools in reviewing actions by the executive branch. Pelizzo and Stapenhurst found, in their survey of 84 countries, that all of them employed some oversight tools, ranging from hearings, to “question time,” to employing an ombudsman. In fact, in their systematic review of the use of 7 different oversight tools, they found that roughly 80 percent of these countries (67 out of 84) used at least 5 of these 7 oversight tools (Pelizzo and Stapenhurst, 2004, pp. 3–5; esp. Table 3, p. 5) (see Table 23.1 for a listing of some of these countries and the extent of the use of oversight tools). Many countries have either independent auditors (elected or appointed) or legislative staff ministries designed to assist in the performance of the oversight function. An example of this is Russia’s Audit Chamber, which employs more than 500 staff and conducted more than 3,000 investigations between 1995 and 2000. (Although it should be noted that the nature of the work of the Audit Chamber seems to be more “financial” and less “performance” related (Remington, 2004).) The Polish Chief Board of Supervision is another example of an audit agency that is under legislative control (Olson, 2004).

Capacity to perform oversight and even oversight quantity are important, but quality must also be considered. Having the capacity for oversight is not the same thing as having the will to perform oversight in a systematic fashion. Systematic oversight—the kind that attempts to discern whether programs or agencies are performing well on a comprehensive basis—occurs with substantially less frequency than more episodic forms of legislative review. This is chiefly because systematic oversight is time consuming, and does not carry with it the kind of rewards that are associated with other forms of legislative activity. Therefore much of this oversight tends to be episodic, and focused on specific instances of wrongdoing or perceived failure that may or may not be indicative of any general trend within the agency or program. In fact, the criticism of Congressional oversight historically is that this oversight has not been focused on the extent to which programs have achieved their objectives. Rather, oversight has been used to focus attention on politically sensitive or high-profile issues.

Political scientists Matthew McCubbins and Richard Schwartz, in an attempt to describe this phenomenon, differentiated between two types of oversight as practiced by the US Congress. The first, which they called “police patrol” oversight, occurs when the Congress (or specifically a Congressional committee) attempts to discern how an agency is performing through routine reviews initiated by the committee. That is, the committee (or the committee staff) initiates a review designed to bring any necessary issues to light. The second case, “fire alarm” oversight, occurs when committees respond to issues that are brought to their attention by interested parties, and then hold hearings designed to “get to the bottom” of problems that have been raised (McCubbins and Schwartz, 1984).

The McCubbins and Schwartz thesis is that fire alarm oversight is more cost-effective than police patrol because it enables committees to spend their attention on those issues worthy of their attention, without the cost of discovery. Thus, they can focus on those issues that carry with them the most political benefit, but rely on constituents, the media, and others to bring those issues to their attention. In a later comprehensive analysis of Congressional oversight, Joel Aberbach found that the police patrol approach actually dominated, in spite of McCubbins and Schwartz’s arguments to the contrary. Aberbach (1990, pp. 98–100) argues, interestingly, that the method may have changed over time, in response to resource scarcity; that is, that when resources are scarce, Congress is more likely to be proactive in engaging in oversight, as a means to justify reductions in expenditures for programs that are found lacking.

This would seem to offer some ray of hope for performance-focused oversight, particularly when budgets are tighter. It seems appropriate to observe, however, that in neither of these approaches does oversight represent an effort to look at programs or agencies “from the ground up.” Thus, while “police patrol” oversight may involve more comprehensive investigations than “fire alarm,” it is still constrained by the questions that the committee—or legislature—wants to ask. If the committee does not want to ask “How is this agency performing overall?” then the oversight agenda will not reflect this kind of broad attention to performance. In fact, in surveys that Aberbach conducted of members of Congress and their staffs, both identified “evidence of malfeasance in administration: scandal” as the top factor explaining a decision to conduct oversight hearings (Aberbach, 1990, p. 110).

In the end, then, once again the problem is one of incentives. In order for detailed oversight to occur, someone must believe that the benefits (in this case, more detailed knowledge about performance) are worth the cost (in this case, precious legislator/staff time and resources). This matters because if legislative oversight attention to what is “right” and “wrong” is to contribute to possible legislative and policy changes, it is necessary that these reviews be as “performance-focused” as possible. If, as discussed above, legislation is more explicit about specifying expected performance, it will be far more likely that oversight will also focus on these performance issues. A benefit of more detailed oversight, in fact, is that it would communicate to agencies that performance is important to the legislature. This is rarely the case at present. The cold fact is that there are currently limited incentives for members of oversight committees—in most countries—to focus in detail on the performance of programs or ministries. Such oversight has few electoral benefits for legislators who must stand for election on relatively short cycles. More emphasis on oversight is unlikely to occur without substantial attention to the question of why the legislature would want to conduct that type of oversight.

Another issue affecting the quality of oversight is the availability of information for the legislature. Following the trend set by Australia, New Zealand, and other (primarily OECD) countries, the last 15 years have seen substantial legislative impetus for performance measurement in the US and, therefore, for a greater performance focus in audit and evaluation. The Chief Financial Officers Act, the Government Performance and Results Act, the Clinger-Cohen Act and other laws had in common the notion that policy-makers and the electorate alike should better understand the relationship between resource use and results. The present Bush administration’s initiatives share this focus, perhaps particularly manifested in the Program Assessment Rating Tool. This system requires after-the-fact knowledge of performance and inputs in order to succeed. If performance reports are required of agencies, to the extent that gaps between expected and actual performance are highlighted, the reports themselves will be useful tools for future oversight planning.

How to support performance in systems with independent legislatures

The reader might be forgiven at this point for believing that constraints presented by a separation of legislative and executive powers are so great as to make the development of a more performance-oriented budget system very difficult, if not impossible, in these nations. This concluding section takes a different view, arguing that there are actions that might be taken to contribute to a greater performance orientation (or, at a minimum, not detract from such an orientation). In most cases, government is essentially incremental, therefore incremental changes can have non-trivial consequences. There are four separate kinds of actions that might be taken to contribute to better public sector performance in governments with independent legislatures:

  • establish a clear schedule for authorizing or reviewing programs

  • invest in production of performance data, and legislative capacity to package and review these data

  • increase transparency of legislative decision-making

  • recognize the performance implications of legislative constraints placed on executive branch agencies.

Establish a clear schedule for reauthorizing programs

Public programs, once created, have a tendency to live in perpetuity. Establishing some set schedule for review by the legislature may, at a minimum, encourage attention to issues of program design, implementation and therefore performance. Systematic conduct of reviews has the added advantage of permitting comprehensive rather than piecemeal analysis of a given program. Among the important questions that can be asked are:

  • What are the program’s current objectives?

  • Have the purposes of the program been achieved? If not, through what means, and when, is achievement likely?

  • What is the historical relationship between funds provided for the program or agency and the achievement of objectives?

Establishment of clear objectives, and systematic reauthorizations, for a program can go a long way in focusing the funding and oversight agenda around comprehensive performance questions, rather than more narrow parochial concerns. Another means of ensuring such review may be to identify a specific sunset for programs, such that on some predetermined schedule (for example, every five years) legislation needs to be passed in order to keep the program operating. It is worth noting, however, that sunset reviews have a rather checkered history. The main problem with sunset proposals is that they need to be taken seriously in order to work. That is, if legislative committees comply with the letter of the sunset review, but not the spirit, the end result may be to create a lot of work for the executive branch without the desired benefits associated with more detailed program review.9

Invest in production of performance data and legislative capacity to review programs

Related to the periodic review of programs, if legislatures express more interest in specific measures used to gauge performance, this will send signals to ministries that will in turn focus their efforts. In such cases, involving the legislature as a full partner in establishing performance measures in the first instance will go a long way toward a unified, consistent approach to performance. Involving legislative stakeholders and affected interests up front—developing agreed-upon measures—can assist in avoiding the temptation to treat reform as an entirely executive-focused activity. It is also important to pay attention to the manner in which performance is reported to the legislature. Performance information is most likely to engage legislatures if it meets two tests. First, it should provide the legislature (or individual legislators) with information that they consider useful. Second, a limited number of high-level measures should be reported to that level. Nothing discourages attention to performance measures like reporting a large number of very detailed measures to people who neither need nor can digest that volume of detail.

There are also legislative capacity issues to consider. Legislatures that develop an internal capacity to both review executive branch performance data provided, and to initiate performance reviews of their own, are more likely to engage in discussions concerning legislative affects on performance. Legislative bodies should consider development of performance audit capacity independent of the executive branch as one means of providing the legislature with the ability to initiate and check performance data. Another reason that attention to institutional capacity is important is that legislatures may tend, particularly in systems with a separation of budgetary powers, to distrust information coming from the executive branch. In this context, if a legislative branch agency is empowered to review and summarize data for the legislative branch, those data are more likely to be viewed as accurate, and are therefore more likely to be used.10 Bourdeaux (2005, p. 16) argues that legislative use of performance information is in part related to “legislative staff capacity, legislator ability to assimilate information, and developing a culture of trust/’ and concludes that the development of an independent or legislative staff capacity to audit or summarize performance data may be crucial to promoting legislative use of those data. Here there is hopeful news. Schick (2002) notes that the general trend in many countries is toward more staffing capacity for legislatures, and more independent audit offices that legislatures can rely on for information on fiscal and programmatic performance.

Some have argued that movements to legislate performance contribute to the development of an agreed-upon set of performance measures. Many countries that have adopted performance budget reforms have done so in response to the passage of laws requiring strategic planning and performance measurement. Certainly, this is true in Australia, New Zealand, and the UK. In the US, the Government Performance and Results Act at the national level, and the myriad state laws mandating performance measurement, have provided an impetus for the development of better measures and have created a number of forums for discussion of the appropriate measures that should be used.11

Increase transparency of legislative deliberations

In the end, perhaps nothing can encourage attention to performance more than transparency. If legislatures are making budget decisions based exclusively on input or output concerns, or based on incremental changes from past practice, the opportunity cost associated with those decisions must be illuminated. In any resource allocation process, the opportunity to ask the question “compared to what?” must be preserved. Here the role of citizens, interest groups, and the media is important in publicizing legislative budgeting practices, and making transparent the cost of paths not taken. Practically, incentives to change these practices must be connected to things that legislators value: for example, how they are viewed by constituents, and likely success in re-election efforts. Sunshine tends to be the enemy of inefficient resource allocation, not because inefficiency is “wrong” in some normative sense, but because it is embarrassing in a political sense. This potential embarrassment can be the factor that leads to the kind of demand for performance data that Wildavsky suggested is so crucial to the use of more performance information by the legislative branch.

Understand importance of budget execution and constraints created by legislatures

As important as it is for legislatures to specify performance expectations, to support the development of appropriate performance measures, and to make decision processes transparent, it is also important that they understand the ways in which legislative practices may impede agencies in executing budgets in an effective manner. Legislated constraints have the intention of, or at least result in, compromised executive branch flexibility in implementation. Overly-specified line-items can make it difficult for an agency to effectively manage spending. For example, fixed lines for hiring personnel versus contracting can prevent agencies from determining which represents the more cost-effective method to achieve a given outcome. Legislative prohibitions against reducing staff at given locations or closing particular facilities—and their mirror image, mandates for a given staff level or a particular amount of expenditure in a given location—inhibit the ability of the agency to manage resources well.

There are perfectly legitimate reasons for legislatures to try and constrain executive activity. Holding bureaucracies accountable is an important function of legislatures, particularly given the substantial power that can be delegated to bureaucracy. It is also important, however, to recognize the negative consequences of control.


Decisions by countries to empower legislatures to make decisions about the budget can make it more difficult to implement budgetary reforms that rely on common understanding of what is to be accomplished, common standards for review of competing claims for resources, and commitment to understanding the effects of the implementation of programs. The independence of legislatures, however, should neither be used as an excuse for inaction nor as a justification for taking power away from the legislatures. Rather, it is crucial to understand the incentives present for these independent legislatures and promote reforms based on the particular needs of these elected bodies. It is tempting, and also potentially incorrect, to assume that the answer must be to weaken the legislature at the expense of the executive.

On this last point, a recent analysis of the 50 US states conducted by Bourdeaux finds support for an earlier argument in the literature suggesting that more powerful legislative bodies actually exercise more budgetary responsibility that weaker ones. The argument, in brief, is that having both a strong legislature and a strong Governor may promote responsibility, and that strong gubernatorial powers (such as the line-item veto) may promote irresponsible behavior on the part of elected legislative representatives, which seek to shift the burden of responsibility for budget outcomes to the Governor.12 Bourdeaux’s own research concludes that “given authority over the budget, legislators may be more interested in reforms that promote administrative efficiency and that this engagement may positively affect implementation across multiple dimensions” (Bourdeaux, 2006, p. 136). If this is true, then the problem of squaring performance-informed budgeting and the separation of powers is not insoluble, but demands creativity in devising tools and incentives that make it more likely that legislative action will promote, rather than impede, government performance.


The author wishes to thank Rita Hilton and Marc Robinson for helpful comments, and Nathaniel Taylor for research assistance.

The Government Performance Project is a joint academic/journalistic evaluation of management of the 50 state governments, funded by the Pew Charitable Trusts. For current results of this project, see <>.

There are numerous other presidential countries that were not included in this study. These include, for example, Brazil, Guinea, Chad, Tunisia, and the Côte d’voire.

See Joyce (2004) for such a comprehensive review of performance and the budget at each stage of the budget process.

There might be problems of political legitimacy through providing the unelected bureaucracy with more power than some would be comfortable with, but that is a different story.

The other three were: flexibility to pursue agency or government purpose; where predictable action is valued for linking managerial action to performance; and where new leadership requires institutional strength and support for effective change.

Paul Posner, then of the US General Accounting Office, described this relationship as motivating some members of Congress, in a presentation before a conference on performance-based budgeting, Queenstown, Maryland, March 31, 2003.

The Government Performance Project, at <http:/>.

Aberbach (1990) discusses the rather limited prospects for sunset to make much difference in Keeping a Watchful Eye (particularly pp. 205–7).

Grizzle and Pettijohn, in discussing the state of Florida’s experience with budget reform, note that “[t]he Florida legislature’s organizational culture includes beliefs that agency staff cannot be trusted to do what is right.” See Grizzle and Pettijohn (2002, p. 56).

On US state-level requirements, see Melkers and Willoughby (1998).

The earlier articles cited by Bourdeaux include Barrilleaux and Berkman (2003), and Abney and Lauth (1998).


    Other Resources Citing This Publication