An Empirical Theory of Public Management

In recent years Larry O'Toole and I (O'Toole and Meier 1999; 2003; Meier and O'Toole 2003; 2006b) have been working with a parsimonious theory of public management that represents what we think is the essence of the public management literature. In an effort to be clear about our contentions and facilitate the empirical testing of the model, we have opted for a precise, formal theory:

where

A brief discussion of the four basic concepts in the theory will serve as a starting point for illustrating how the study of public management might inform political science.

Legislatures

The study of legislatures and legislative behavior is a central, perhaps defining, activity of political science. An interesting comparative literature is focused on one legislative output—survival, or how long the governing coalition will last (Laver and Shepsle 1998; Martin and Vanberg 2004). Implicitly, and more recently explicitly, the notion of government performance is relevant; the success and policy performance of the coalition contributes to its longevity and ability to survive. This literature considers a variety of variables similar to those included in our model of management. Structures such as electoral laws influence the longevity of the coalition. Environmental conditions such as the number of political parties or the state of the economy are also thought to play a role.

This empirical literature, I believe, would be greatly aided by some focus on management. Coalitions are managed, and the degree to which coalition leaders are successful at management should affect how long the coalition survives. These management functions include M₁, internal management, the ability to control the members of one's own party and to keep the back-benchers happy. It includes M₂, or managerial networking, keeping coalition partners happy (that is, other parties in the coalition) or at least not so unhappy that they abandon the coalition. Thies's (2001) recent work on the allocation of cabinet posts and the use of junior cabinet members to monitor coalition partners is one illustration of managing the coalition/network. The M₂ function can also include the cultivation of other network nodes that might be useful; as an illustration, Druckman and Thies (2002) find that support in the upper house of a bicameral legislature lengthens the coalition's life even though upper houses often do not have significant independent powers.⁴

Could one measure these aspects of legislative management? In terms of both the various actions that constitute these functions of management and the quality of that management, I think the answer is yes. McCurley and Mondak (1995) successfully created a measure of individual legislative effectiveness as well as a measure of honesty for members of the U.S. Congress using available public documents and media. Coalition leaders tend to establish long political histories that could be used to construct similar management measures.⁵

At the risk of being an area studies person, I think public management has a great deal to say about the U.S. Congress also. A recent article by Clinton and Lapinski (2006) created a long time series for congressional outputs. This provides one measure of performance; earlier work on outputs relative to the president, in terms of passing or failing to pass significant legislation, provides an additional output measure (see Edwards 1985; Edwards, Barrett, and Peake 1997). Clearly the point about coalitions, that they are managed, also applies to the U.S. Congress. Current House Speaker Dennis Hastert (R-IL) can be compared to former House Speaker Newt Gingrich (R-GA) systematically, and such measures should explain some additional variation in the actions of the Congress.⁶

One obvious point in applying public management to Congress is that Congress is a network, not a hierarchy, and network leaders must gain cooperation with only modest ability to coerce. Fortunately, we know a fair amount about managing in networks. The work of O'Toole (1997) and Provan and Milward (1995; see also Meier and O'Toole 2003) finds that networks work best when they operate in practice more like hierarchies, that is, they create more formal ties, stabilize linkages, and engage in long-term rather than short-term relationships. Relative to the U.S. House of Representatives, this might be termed “the Tom Delay rule”: The use of coercion via both threats and inducements can be used to gain cooperation in a network. Beyond this recent example, we know that an alternative way to generate organizational stability is to impose a set of uniform values on members of the organization (Kaufman 1960) either by selective recruitment (Carpenter 2001) or via socialization. In this regard, partisanship might be considered a value, and the rise of partisanship in Congress in recent years can be interpreted as an effort to gain control over the network that is Congress (Bond and Fleisher 2000).

The structural aspects of the management model are also relevant to both Congress and other legislatures. The House with 435 members requires greater structure to operate than the Senate with only 100 members. Similarly, as the literature demonstrates, rules matter. The debates over the filibuster in recent years and whether to limit its use are debates over the distribution of power in the Senate. One of the most significant structural changes in the House of Representatives in the last few decades was the decentralization of the committee structure after 1974, which more broadly distributed power to individual members (Baumgartner and Jones 1993, 199).

Structural variables might well be the answer to an interesting substantive question in the legislative literature. Students of representation have generally failed to find an independent effect of racial representation in Congress (Lublin 1999; but see Espino 2004) and in state legislatures (see Owens 2005; Preuhs 2006), but it is always found in school boards (Meier and Stewart 1991; Meier and O'Toole 2006a). The size of the legislative body and the obvious structural implications might be the most promising explanation for this difference. Large legislatures need to be organized; and thus, they create structures and rely heavily on partisanship (both of which limit inter-member interactions). Whereas in school boards with five, seven, or nine members, small group dynamics is likely to replace structural and partisan methods of coordination. A minority representative, thus, is more likely to be pivotal on some issues on a school board and more likely to influence colleagues even when the representative is not the pivotal voter (see Meier 2005).

Finally, environmental turbulence (the X variables) also influences legislative actions. All organizations need to adapt to environment changes or alternatively figure out a way to stabilize the environment. One interesting illustration is how Congress has decided to deal with the environmental instability caused by elections: by consistently stressing buffering the environment (M₄) over exploiting environmental opportunities (M₃) by gerrymandering out the competitiveness of individual elections.⁷

Political Executives

Political executives are the most obvious application for the theory of public management. Indeed, several works have incorporated the ideas of management into the study of the U.S. presidency. Karen Hult and Charles Walcott (2004; Walcott and Hult 1995) have done yeoman work on the creation of White House structures and how they matter. Peri Arnold (1986) has written on the managerial presidency, and Andrew Rudalevige's (2002) Managing the President's Program is a creative effort to deal with management issues and the presidency.⁸

I would like to take as my starting point on the presidency, however, a recent, yet-to-be-published paper by James Pfiffner (2006) entitled “The First MBA President.” Pfiffner characterizes President George W. Bush as a classic MBA-style manager who stresses secrecy, speed, and top-down control and shows how this pattern results in a predictable pattern of decisions.

The MBA approach, within the context of our theory, is a pure M₁, internal management style, with no need for networking or M₂ at all. As President Bush stated about the War on Terror, “I have no outside advice. Anybody who says they're an outside advisor of this Administration on this particular matter is not telling the truth” (Lemann 2004, 158). The obvious point is that such a management style becomes highly dependent on the quality of decisions made by “The Decider.” The implications for decision making and performance should be obvious.

A second key aspect of public management is also relevant to the presidency. We are currently in the midst of a performance appraisal revolution (Holzer and Callahan 1998; Ferlie and Shortell 2001). Performance standards are set, and agencies and programs evaluated in terms of how well they meet these standards. Although schools are at the forefront of this performance revolution, it is occurring virtually everywhere—the PART system for the U.S. federal government, the elaborate audit system for UK local governments, and in many other governments throughout the world (Gilmour and Lewis 2006; Pollitt and Bouckaert 2000; Pollitt 2006). A basic law of organizations is that if you set a goal or a performance system, the organization will try to maximize the performance system. Goal displacement, especially when the performance system does not exactly measure what the organization is to accomplish (a virtual guarantee), is a clear and present danger (Blau 1956).

What are the implications of what we know about the management of performance systems that are relevant to the presidency? I think the president's performance appraisal system is public opinion. That presidents play to public opinion polls just as schools teach to the test is fairly obvious. In fact, with the extensive use of polling by the White House and its allies, we see a unique form of teaching to the test. Presidents now look to see what public opinion says and then select the policy options that fit public opinion (Eisinger 2003).

Interpreting public opinion as a performance system for the president raises the question of cheating, an issue John Bohte and I (Bohte and Meier 2000) have examined in the case of schools. If enough administration spokespersons contend that weapons of mass destruction were found in Iraq, will the public believe it? Recent public opinion polls show movement in this direction (Harris Poll 2006). Metaphorically, this process is equivalent to a school district that did not like its test results and so it went back and changed the answers to the test to produce scores more to the district's liking.

Courts

Courts are organizations with very little if any M₁ term. While it may be genetic, lawyers for some reason appear to have no management skills whatsoever. In a law firm the managing partner is one who is not considered valuable enough to deal with the firm's prime clientele. Some law firms have even gone so far as to admit their lack of management skills and to hire non-lawyers to run their operations.

By eschewing internal management, legal organizations such as courts rely exclusively on M₂, external management, and S, structuring, to deal with the environment and its threats. A directly relevant and impressive research agenda using neoinstitutional approaches to judicial politics has revealed the relevance of many management-related variables (see Brace et al. 2006; Spriggs and Wahlbeck 1997; Wahlbeck et al. 1998). The theme of this literature is that court decisions reflect not just the preferences of judges but that these preferences are shaped and limited by institutional factors.

The importance of structural elements is illustrated by a recent time-series analysis by Kevin McGuire (2004). He demonstrates how creating structures, or institutionalization as he terms it, has affected the volume of work done by the U.S. Supreme Court. Such structural changes as the elimination of circuit riding, the creation of the clerk's office, the establishment of a library, and the relocation into a separate building increased the capacity of the Court to generate greater outputs and thus expand the role of the judiciary in American politics. Equally impressive is the way the U.S. Supreme Court has structured and managed its workload over the past 50 years, the result being that the volume of its workload has dropped dramatically (McGuire 2004).

The downside of relying exclusively on M₂, managing the environment, is that courts seem to take the norm of judicial self-restraint seriously. Others need to have real cases or controversies to get on the courts' agenda; cases have to be ripe and meet a variety of rules of standing. What this means is that of the M₂ general options, buffering versus exploiting the environment, courts generally cannot exploit the environment because they do not have any ability to interact with outside actors on their own initiative (see Goerdel 2006 on initiation). Despite occasional legislative wish lists, courts as a result have limited influence on the actions of other political institutions in regard to the courts. Judicial concerns about the detrimental impact of mandatory minimum sentencing, for example, have generally been ignored by legislative bodies (“The Verdict Is In” 1993; Berman 2005).

Public Policy

I will not deal with public policy in depth because public administration has made and continues to make significant contributions to the study of public policy. Some of us, in fact, are considered students of public policy and even teach courses in the area despite never having taken a course on the subject.

My illustration in the area of public policy draws on the work of Baumgartner and Jones (1993), who have made a series of contributions to the public policy literature with their theory of punctuated equilibrium. More recently, Bryan Jones and colleagues (Jones, Sulkin, and Larsen 2003) have addressed the concept of punctuated equilibrium in the policy process by using analogies from tectonics, the study of the movement of the earth's plates. Similar to the buildup of geological pressures that at some point are released by a major earthquake, Jones et al. suggest that political institutions create stickiness that results in the buildup of pressures for major change. Eventually, the pattern of small incremental changes is disrupted (punctuated) by a major change. Institutions, according to this theory, generate outputs that are characterized by high-peaked distributions (many small changes) with narrow shoulders (fewer moderate changes) and heavy tails (a few very large changes) rather than being normally distributed.

The heavy tails are regions of dramatic change, some with opportunities, some with serious dangers. Dramatic change might include hyperinflation, the collapse of a political regime, or other major polity-shaping phenomena. Jones et al. (2003) have illustrated this with different types of institutions and have shown that greater institutionalization (what we would term structure) generates greater stickiness and thus more leptokurtic distributions.

In recent work with Scott Robinson, Floun'say Caver, and Laurence O'Toole (Robinson et al. 2007), we ask what happens when management is added to this process. Including management is quite logical because political institutions are human systems, not natural systems. Human systems are managed systems; and, as a result, successful managers will try to smooth out and normalize these processes. That is, the dangers of the extreme cases in leptokurtic distributions should be apparent to managers, and they should take actions that release the buildup of pressures for change before they can reach dangerous levels. Indeed, our empirical analysis shows that organizations with a larger administrative component generate outcomes that look more like normal distributions than leptokurtic distributions.⁹

Spatial Modeling

Another flourishing approach to political science that might benefit from some exposure to public administration is the literature on spatial modeling. Some of that literature explicitly includes bureaucracy in its models, particularly the political control models or their offshoots, the delegation models (McCubbins, Noll, and Weingast 1989; Epstein and O'Halloran 1999). While this scholarship has produced a flurry of theoretical work and a modest empirical literature, much of the literature is hampered by an assumption of unitary actors. As one might expect, when that literature allows for non-unitary actors, it does so for legislatures (see Baron and Ferejohn 1989).

An equally interesting approach would be to relax the assumption that bureaucracies are unitary actors. Clearly, career bureaucrats have different policy ideal points from agency heads, and everything we know empirically about bureaucracy indicates that hierarchies cannot eliminate the discretion exercised by organization members (Barnard 1938; Simon 1997; Downs 1969; Brehm and Gates 1997).¹⁰

Spatial models become even more interesting when one relaxes the assumption that all the key actors (legislature, elected executive, bureaucracy) are unitary actors. Such an assumption means that any given institution must first manage itself (that is, pay attention to the M₁ function) and only then seek to exercise the M₂ function and enter into the various games with other political institutions. The general practice of examining games one at a time might also be adjusted in valuable ways. Although setting up models for a single game makes solutions more tractable, institutional players face several potential games simultaneously (that is, agencies often interact with multiple congressional committees and administer several programs that have vastly different game players). One management function, therefore, is what Fritz Scharpf (1997) calls deciding which games to play—or, in our terms, deciding whether to buffer the environment or exploit it.

Policy spaces are almost always multidimensional; and as the number of dimensions increase, more than one equilibrium is possible, and agenda setting becomes crucial. The study of public management has long recognized that, rather than ideal points, bureaucratic actors have zones of acceptance or zones of indifference (Barnard 1938; Simon 1997); such zones provide the potential to manage the game by trading off space within one zone of acceptance to protect more highly prized space in another dimension. Another strategy might be to combine games, to link (see Tsebelis 1990) a new game such as homeland security to an existing game such as transportation planning, or to drop out of one game (the FBI's stance from 1928 to 1980 in drug law enforcement) to maximize its impact in other games (high profile crime areas such as bank robbery). In short, bureaucracies build and manage coalitions over a series of games (Rourke 1969; Carpenter 2001). Given the multiple dimensions plus the incorporation of management, there is always some spatial area where more than one equilibrium might be possible. All of these suggestions make the job of our colleagues in game theory and rational choice more difficult, but I also think more interesting. Managing the various games that bureaucracies play is an essential part of the M₂ process.

Governance Structures

Polities are governed not by a single institution but by a collection of institutions, many of which are specialized. Public management has an extensive focus on governance systems (Lynn, Heinrich, and Hill 2001), a body of knowledge that would be useful to scholars of institutions in multi-actor processes. I would like to illustrate this potential by reference to recent work (Meier and O'Toole 2004) that asks, what happens in a multi-institutional governance systems when one of the institutions fails?

We approach this question using the perspective of the famed politics-administration dichotomy. Frank Goodnow (1900), in Politics and Administration, a book everyone cites but no one actually reads, never contended that politics could be separated from administration.¹¹

In other words, politics can never be separated from administration; the functions always overlap. Politics and administration are essentially symbiotic functions: both need to be performed to have a successful governance system.

How might one test Goodnow's contention that politics and administration are symbiotic functions or, in more general terms, what happens to other institutions in a governance system when one of the institutions fails? Our study (Meier and O'Toole 2004) examines this question in the context of the Texas education system by using data from local school districts. Specifically, we ask what happens when a school board fails to perform one of its key functions: representing the range of interests in the community. When that occurs, what are the implications for public management (the administration function)?

A school board might fail in its representation function in one of two ways. First, it might suppress conflict. When a school board suppresses conflict, the conflict does not go away; it merely moves into the administrative branch (see Truman 1951). The manager must now deal with both the conflict that the political branch failed to resolve and the implementation job. Second, the school board might exacerbate conflict, thus generating more dissension than is present in the broader community. This action makes the manager's M₂ function much harder. The manager first needs to contain the conflict and then also to attend to the M₁ function. This logic implies that the bureaucracy, as a result, becomes more important on issues that should be the domain of the political branch, that is, politically salient ones. This does not mean that the bureaucracy succeeds, only that it now accounts for more variance in the outcomes.

An analysis of data from 1,000 school districts over a period of eight years supports the hypotheses generated using the logic of the politics-administration dichotomy. On salient educational issues, the failure of the school board to represent the population adequately results in the actions of the manager increasing in size and significance. In short, management actions become more important in determining whether the school district succeeds or fails.

We believe that this illustration is part of a more general process. Numerous examples from comparative politics also illustrate how one political institution becomes more important when another chronically underperforms. Legislatures frequently cannot exercise the discipline needed for consistent fiscal policy; in such circumstances the role of an autonomous central bank becomes more crucial (Franzese 1999). When central governments cannot manage ethnic conflict, the role of decentralized regional governments is more influential in maintaining regime stability (Lustick, Miodownik, and Eidelson 2004; Lijphart 1977; Hale 2000). Similarly, until recently the failure of political institutions generally in Latin America led to an increased governance role for the military (O'Donnell 1973). All these illustrations imply that the phenomenon studied here, how one governing institution responds when another underperforms or fails, is a fairly common problem of modern governance.¹²

But So What?

The theme of this lecture is that our colleagues in political science could learn a great deal about questions they care about by paying attention to the literature in public administration. Institutions are managed—often badly managed—but managed nonetheless. The study of public management has built up an extensive body of knowledge about how organizations are managed and what difference management makes in terms of an organization's success or failure. I have deliberately focused on the public management literature, that stream of public administration least likely to be read and incorporated by political scientists. Taking the extreme position, I hope that I have raised the possibility that other subfields of public administration might also inform the study of political science.

I know that when delivering the Gaus Lecture I am preaching to the choir. I also know that you find your own research more exciting and interesting than what political scientists do. But let me suggest that missionary work is a valuable contribution. I urge you to visit political scientists in their native habitats and bring them the word. There was once a vital and flourishing joint tradition of public administration and political science. That tradition can only be recreated if those in public administration carry the torch.

With the publication in PS, there remains the possibility that some political scientists not in public administration might read the lecture. Those scholars I would encourage to check out the public administration literature; I have presented only one small subset, public management, and that only partially.