The Policy Studies Journal, Vol. 31, No.

4, 2003

Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations: Using Evolutionary Theory to Explain Policy Change?
Peter John
This article reviews the current state of public policy theory to nd out if researchers are ready to readdress the research agenda set by the classic works of Baumgartner and Jones (1993), Kingdon (1984) and Sabatier and Jenkins-Smith (1993). After reviewing the inuences of institutional, rational choice, network, socio-economic and ideational approaches, the article pays tribute to the policy streams, punctuated equilibrium and policy advocacy coalition frameworks whilst also suggesting that future theory and research could identify more precisely the causal mechanisms driving policy change. The article argues that evolutionary theory may usefully uncover the micro-level processes at work, particularly as some the three frameworks refer to dymamic models and methods. After reviewing some evolutionary game theory and the study of memes, the article suggests that the benets of evolutionary theory in extending policy theories need to be balanced by its limitations.

Objectivity cannot be equated with mental blankness; rather, objectivity resides in recognizing your preferences and then subjecting them to especially harsh scrutinyand also in a willingness to revise or abandon your theories when the tests fail (as they usually do).Stephen Jay Gould, The lying stones of Marrakech: Penultimate reections in natural history (pp. 104105). Ten years has elapsed since the last major advance in public policy theory. For it was in 1993 that two key books were published: Baumgartner and Jones Agendas and Instability in American Politics and Sabatier and Jenkins-Smiths Policy Change and Learning. Although these volumes had their antecedents, such as in the work of Kingdon (1984) and in the earlier writings of both sets of authors, they marked a punctuation in thinking about public policy. Both attracted a great deal of attention; they were complementary, and they set off research programs in the forms of detailed empirical research, edited collections of studies (e.g.,
481
0162-895X 2003 The Policy Studies Journal Published by Blackwell Publishing. Inc., 350 Main Street, Malden, MA 02148, USA, and 9600 Garsington Road, Oxford, OX4 2DQ, UK.

482

Policy Studies Journal, 31:4

Baumgartner & Jones, 2002), and extensive commentary in the rest of political science. Important work has emerged since, such as Joness Politics and the Architecture of Choice (2001) and the collection of essays in Sabatier (1999), but nothing has changed the direction of thinking in the same way that the cluster of books and articles at the beginning of the 1990s did. Appearing to conrm this claim, the APSA Public Policy Section voted Baumgartner and Jones a decisive rst in their list of most important works in public policy in the last 10 years Shoup (2001). That social science should take the route outlined above is no great surprise, given the prevalence of research programs and how they generate knowledge. What may be more useful to speculate is whether this particular kind of thinking about public policy has come to the end of its line of development, and whether there is potential for a synthesis and further innovation, in particular whether evolutionary theory has the potential to push ahead the study of public policy. Theorizing About Public Policy When considering about how to theorize about public policy, there are two things to bear in mind. One is the nature of theory in the social sciences; the other is the character of public policy. For empirical researchers, theory is a body or system of propositions about the causal relations that link together elements of the social, economic, and political worlds. These relations are regularized, having applicability over a range of cases, both in space and time. Theory in social science is usually based on claims about the nature of human action and power relationships, and seeks to provide a coherent and consistent account of reality. Sometimes the more wide ranging theories are called frameworks that organise diagnostic and prescriptive inquiry (Ostrom, 1999, p. 40). What distinguishes one social science theory or framework from another is the primacy attached to certain core social processes, which often act as a set of determinants of others and to social and economic outcomes. The result is a system of human action, in which the observer can nd, even with the subtleties built in, the links that derive from an original cause. Theories differ, of course, in their applicability; but they are linked by the aim to generalize, and in themselves they do not yield hypotheses. What theories do is to generate models, which are more restricted assumptions about social and political relationships from which hypotheses can then be derived and tested. Researchers in the eld of public policy want to understand why public decisions and their outcomes change, stay stable, vary from sector to sector, and differ in their consequences for the publics that consume and appraise them. It is a distinctive and problematic area of study, far more inclusive than others. In contrast, the empirical subtopics of most parts of political science tend to concentrate on one institution, such as Congress, or explore a eld of activity, such as voting behavior, which limits the range of what is to be explained even if most aspects of social and political processes need to be referred to if the explanation is to have

John: Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations

483

depth. Public policy, on the other hand, tends to include in its baseline all political activity and institutionsfrom voting, political cultures, parties, legislatures, bureaucracies, international agencies, local governments, and back again, to the citizens who implement and evaluate public policies. In addition, decision making varies vastly from sector to sector, a claim that is the core contribution of public policy studies to political science knowledge, but which complicates the task at hand. The problem is compounded by the absence of a clear chain of causation from public opinion to parties and bureaucracies and back again. As many writers on public policy have lamented (cf. Sabatier, 1999), there can be no stages model of the political process to provide a simple map because of the multiple sources of causation, feedback, and the sheer complexity of what is going on. The heterogeneity of the institutions under study and the complex networks among them preclude sequential sorts of explanation and rule out the use of theories that seek to understand how democratic the policy is, such as pluralism and its successors. Coming up with theory that creates some simplicity or parsimony and that takes account of complexity is quite a challenge. The move to simplicity may simply impose a tautology or overextend a set of plausible and partial models of political action to the whole of the policy process. But if complexity is the theme, then the models become redundant and degenerate to descriptionnot poor description, such as lists of committees, laws, and public decisions, but good description, which provides an account of human action based on a contextual understanding of the links and transactions among decision makers, but in which social science hypothesis testing is not particularly relevant. Importing Theory from Mainstream Political Science One answer to the search for theory is to take ready-made ones already in use in political science, as they often have a policy dimension. The problem is that such theory may not be well adapted to the many faceted character of the policy process; moreover, many of these theories have difculties of their own. Institutionalism: Old and New The best candidate for such an approach is institutionalism. This is the idea that formal structures and embedded norms have an effect on human action, which has a long pedigree in political science and appears in the classic writings, from Plato to Montesquieu, deTocqueville, and Woodrow Wilson. In the 1980s and 1990s, it was conventional to distinguish between the new and old variants, but given the limited life-span of any ascription of new, it is better to talk of them in one sense, with different instantiations. In part, institutions are formal arrangements, such as electoral systems, the division of powers, and the salience of the higher courts; but there are also the practices embedded in formal organizational

484

Policy Studies Journal, 31:4

arrangements, which are sometimes called standard operating procedures. The former sense is better for empirical testing, and comes out with the hardly startling nding that institutions matter for policy outputs and outcomes (e.g., Steinmo, Thelan, & Longstreth, 1992; Lane & Ersson, 1999). Unless institutions are entirely circumvented by networks and power relations, they generally affect how policy is made as they inuence the speed at which political systems attend to public problems, the efciency with which they aggregate public preferences, and the way in which policies attract rent seekers and principals seek to control their agents (Strom, Muller, & Bergman, 2003). Given that institutions constrain public action and affect the costs and benets of political participation, such an effect is to be expected. But does institutionalism explain policy change? In part, it does, but institutionalists nd it harder to explain bursts of change, such as improvements in policy performance or the imminence of policy disasters, which are some of the crucial issues. Institutions can account for change when they adapt, especially in relation to one set of interests and policy concerns. When new pressure groups rub up against established sets of institutions, the resultant change may be greater than would be expected. Institutions often are the grit in political systems, promoting reform, for example, by blocking changes initially. The other alternative is when new groups gain institutional power by conquering one branch of the state and can advance their interests. For example, one way to explain the advance of the civil rights movement in the 1950s and 1960s is through the institutional variation within U.S. polity. Of course, social change must play its role, but an institution, the Supreme Court, which embodiedperhaps by chancethe values of the reformers and gave them leverage, was useful in propelling progressive social policy ideas into the mainstream. Institutional reform can also promote change, say, between levels of government. Moreover, it is possible that institutions themselves adapt. They may evolve according to their own rules, and so affect the choices of policymakers. In spite of these nuances, it is not certain that institutional approaches offer an all-encompassing theory of policy change, mainly because institutions are better at explaining the dampening rather than the amplifying of political processes. They are generally stable, which means they set out routines and constrain human action. Socioeconomic Change Socioeconomic changes must play their role in explaining policy change in the form of shocks and inuences on the political system. A lot of academic energy was spent on the socioeconomic causes of policy change before the 1980s, but then doubts about macro schemes of politics set in and systems theorists of all sorts fell out of fashion. It is possibly the case that the intellectual reaction against systems theory and functionalism has gone too far, and social science should start examining complex systems again, perhaps through the idea of coevolving social processes. But social scientists do not now accept the basic assumption that there

John: Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations

485

is a transmission belt from society and the economy to the political system and its institutions, as the latter inuences the former, and both are highly variegated. Socioeconomic changes may act as a constraint on political action, but the action needs particular conditions to indicate the strength and nature of that inuence. In short, socioeconomic factors do not offer a theory of policy change except to show that policymakers react in different ways to the wider environment. Rational Choice Theory Rational choice theory examines policy change, variation, and stability by examining the strategies of actors located within political institutions and in society at large. Outcomes are the effects of these choices, which may not reect the aggregation of preferences of decision makers but rise from strategic interaction imposed by the structure of the choices, such as from the size of the payoffs. Collective action may fail in many contexts, but it is possible to overcome this dilemma through the evolution of cooperation, the operation of smart institutions, the presence of selective incentives, and the actions of entrepreneurs. Thus environmental policy or urban development policy may result from the degree of collective action possible, which may lead to policy change or stability. Evolutionary game theory is the most appropriate application of rational choice theory to understand the movement from policy stability to change and vice versa. Here the tness of the players comes from the success they have in playing the game. Rather than being stuck in a one-shot game, the players can alter their strategies. With the prisoners dilemma, it is possible to model supergames or tit-for-tat simulationsto understand the ways in which cooperation emerges (Axelrod, 1984; Taylor, 1987). If Ostrom is right and ideas about cooperation apply to common resource solutions, rational choice helps researchers understand how societies may move from noncooperative to cooperative equilibria (Ostrom, 1998). But rational choice does not offer solutions for all cases and contexts. It is better at explaining outcomes when preferences are settled; it nds it harder to explain where those preferences come from and why they should change. Thus it becomes part of the analysis of public policy without being all encompassing: an essential part of the toolkit of political scientists, ready to apply to certain contexts, such as coalition building, for example, which provide testable hypotheses. Adapting Theory from Mainstream Political Science When taken as a set of hypotheses and claims, few researchers would want to work without a theory of choice, an account of socioeconomic change and an understanding of the inuence of institutions, but these approaches do not work as well as all-encompassing theories of public policy. So do models developed by public policy researchers fare better?

486

Policy Studies Journal, 31:4

The best candidate is group or network theory, which has emerged through the subgovernment literature and from organizational sociology. This theory claims that the structure of the coalitions across the complex policy sectors determines policy outputs. The long-term relationships between interest groups and executive agencies evolved into more complex networks between public and private organizations as the number of institutions and participants in the policy process grew after the 1960s, when public problems became more intractable and policy sectors less distinct from each other. Thus agricultural policy became harder to organize without thinking of international rules and economic competition, and it now links closely to environmental issues. The claim is that the values, power relations, and routines in these networks explain the differences across policy sectors and account for long periods of stability, which observers of the literature on U.S. subgovernments had long believed, and this idea was adopted and extended by the policy network literature in the 1990s (cf. Marsh & Rhodes, 1992; Browne, 1995). In part, such explanations could be seen as a subtle form of institutionalismthe inuence of regularized constraints on public action, which operate outside and across the formal institutions. But these accounts suffer from some of the same problems as institutionalism because networks are static in character. They may not even be much of a constraint, being epiphenomenal to the social and political systems they occupy. Thus social or organizational change usually affects networks, which occurs at the same time. When power relations are at stake, the existence of a network among organizations does not seem to be a particularly strong inuence or constraint on human action. In the brutal world of politics, loyalty is a luxury that few can afford. In the United Kingdom, Dowding (1995) has led a cogent attack on policy network approaches, with an inuential article, and the debate continues (Dowding, 2001; Marsh & Smith, 2001). A less polarized position would acknowledge that networks are more than contacts and power relations; they are sites for the exchange of ideas and perpetuation of social practices, which have a long-term impact on the content of public policies. But this claim takes the analysis to a new topicthe impact of ideas in public policy. Ideas Ideas were the hot public policy topic of the 1990s. Public policy scholars became interested in the effects of knowledge on public policy. To what extent do thoughts or ideations about an issue such as poverty or deprivation act independently of the interests that advocate them? Knowledge in the policymaking world consists of claims about the origins and solutions to public problems, which link to normative claims and may either be accepted or debated by the participants in the policy-making process. Even though most knowledge is social in its construction, the argument is that the process of collecting evidence, the creativity in generating the ideas, and the skills at deploying them give ideas independence.

John: Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations

487

One problem is that it is not clear exactly what ideas are. Depending on the author, they are sometimes policy proposals, new techniques or solutions, systems of ideas, or discourse and language. In the strong sense, ideas-based approaches involve the claim that they constitute reality, making history a contest of ideas rather than interests. But most public policy analysts would want to veer from such a strong articulation of ideas and would like to see them playing a role within the power play of institutions and choices, guiding preferences, comprehensible when seen against a more conventional political backcloth. Thus it makes sense to answer the question to what extent did ideas about the causes of poverty inuence social policy in the United States in the 1960s by examining the growth of research on the issue; the campaigning role of groups; and the interests of different political actors, such as the positions of political parties and runners for elected ofce, which seize on ideas as ways to articulate their interests. In this way ideas are closely connected to political interests, neither determined by them nor determining of them. As with the other approaches in public policy, the application of ideas to public policy is not a theory but the identication of a set of causal processes that link to others. These are contingent on each other, and one is neither logically nor empirically prior. It might seem bland to say that institutions, socioeconomic processes, networks, choices, and ideas interact with each other, but it is a truer statement than saying that one of these processes drives the others. The question to ask is whether chance and history do not have a clear pattern because of variability of these relationships or whether they interact in a systematic fashion, and then allow for the tests of models of political action derived from theory. If contingency and context take over, then researchers have an endless variety of processes to observe, making description much better than the application of theory. Is there any way out of this trap? Synthetic Accounts of the Policy Process In the 1990s, more complex accounts of policy streams, advocacy coalitions, and punctuations emerged. What these theories or frameworks offer is a conceptualization of the relationship between the ve core causal processes. These frameworks may be called synthetic, largely because they bring together much of the research on institutions, networks, socioeconomic process, choices, and ideas (John, 1998). Policy Streams and Windows The rst is Kingdons (1984, 1995) account. He found that policymakers often do not always know where policies come from, particularly when they are asked to explain how it was that a proposal emerged rather than another. Kingdon argues that there is an element of chance or a stochastic element, which explains the uidity and rapid change of the policymaking process. He does not say that

488

Policy Studies Journal, 31:4

randomness dominates or provides the sole explanation, as there is an interaction between randomness and the more recognizable processes of problems, policies, and politics. The garbage can of conjunctions among the policy streams may be crucial in setting off deviations from existing policies, but existing institutions, parties, and public opinion still inuence how and when they are introduced. Kingdons account is close to an evolutionary model of public policy. He writes that policymaking is a complex adaptive system (1995, p. 224) in which agents react to changing environments and there is continual Darwinian selection. Kingdons conception of public policy as a primeval soup also implies evolution. New organisms are only partially formed until they nd the right environment, and then they emerge as policies, which Kingdon sees as akin to biological natural selection (p. 226). He refers to Eldridge and Gould when discussing the concept of discontinuous change, the punctuated equilibrium model, and notes that politics moves in ts and starts just like the natural world. Kingdon uses evolutionary ideas to highlight the dynamic and contingent aspects of his account. It is a useful component of his account of policy change, without being an evolutionary model. There are, however, some useful clues as to how one could emerge. Kingdon argues that possibilities and limits of combinations create unique outcomes because [e]verything cannot interact with everything else (1995, p. 207). In other words, there are certain combinations of ideas and proposals that have the potential to evolve, but not others. Here is a possible causal mechanism. It is not enough to call the policy process evolutionary because in some general way it resembles how biologists explain the natural world. There is a causal claim that aims to show how policy proposals are adopted and where some aspect of the policy process resembles genetic selection as identied by the biologists. The Punctuated Equilibrium Model The link to the study of evolution seems clearer in the punctuated equilibrium model of agenda change. The term draws from debates in evolutionary biology in the 1970s. Indeed, Baumgartner and Jones write, Policy diffusion, with its S-shaped curve, is remarkably like the punctuated equilibrium model in which the system shifts rapidly from one stable point to another (1993, p. 17). Baumgartner and Jones cite Eldredge and Goulds work, though they are cautious in their comparison. Overall, these researchers tend to use punctuations as an analogy to the natural world, discussing a few elements of evolutionary theory rather than adopting it wholesale. There are several aspects to the explanation, which involves a realistic account of the limits of individual decision-making capacity; an examination of overlapping jurisdictions in legislatures, which is one of the ways in which agendas expand; and a critique of the idea that policies and agendas simply continue over time and only gradually change. Baumgartner and Jones use all these

John: Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations

489

elements to examine the core of the model, which is the existence of punctuations themselves, and which are explored using longitudinal data on agenda setting and decision making (Baumgartner & Jones, 1991, 1993, 2002; Jones, 1994, 1998, 2001; Jones, Baumgartner, & Talbert, 1993; Jones, True, & Baumgartner, 1997; Jones, Baumgartner, & True, 1998, 2002a, 2002b; Jones, Sulkin, & Larsen, 2003; Jones & Sulkin, 2001; True, 2000; True, Jones, & Baumgartner, 1999, John & Margetts, 2003). It is an important research program, which has the ability to question established models of the policy process, such as incrementalism. It also casts doubt on the claims of stable policy networks and the gridlock of the U.S. political system, by showing how policy monopolies can dissolve. Why do punctuations occur? Baumgartner and Jones argue that once an idea gets attention, it will expand rapidly, and once that happens, as Kingdon notes, it becomes unstoppable. But what causes the idea to get just that little bit ahead of the others in the rst place? As Kingdon notes, there are many ideas competing for attention, but then something happens to make one more applicable than the others at a particular point in time. In one account, the process comes about from external events that disrupt the political system, particularly the ones that are big enough to puncture the equilibrium. Either political systems respond to crisis, such as the way in which the political system was hit by the earthquake of oil price rises of the 1970s, and then they respond to it with a different set of policy instruments, or there is partisan change when a new party enters power, reecting a shift in electoral preferences, which has its expression in new policies. Indeed, if the rst task of the empirical program is to show that punctuations occur in the form of leptokurtotic distributions of policy or agenda changes around a median point, the second shows that these punctuations occur because of political changes (Jones et al., 1998). Time series analysis, with dummy variables placed in the theoretically correct years, can test the partisan change hypothesis. With these two types of tests, the bulk of the explanation rests with what is happening outside the policy process, making the model explaining the trajectory of change. But there may be some important questions unanswered: at what point do these external processes impact on decision making; what is the exact shape of the Sshaped relationship between external change and policy outputs and outcomes; at what point in the bend does the punctuation start to emerge; what is the relationship between the nature of the policy input and the character of the policy-output; and does the punctuated equilibrium tell us what kind of policies emerge or is it just about the quantity of them? The last question is the most crucial one of them all. Jones traces the process of policy change in two single-authored books. In Reconceiving Decision-Making in Democratic Politics (1994, p. 18, fn. 4), he argues that there is a great deal of sloppy thinking about punctuated equilibrium; it is not descriptive but an explanatory model: [I]n biology and paleontology the ideas of gradual and punctuated change both have causes; they are not solely descriptions. There is more biology in Politics and the Architecture of Choice

490

Policy Studies Journal, 31:4

(2001). Jones starts from biological arguments about human cognitions and adaptation, arguing that the characteristic of cooperation has been an advantage and that it is learned. Conformity is inherited genetically because social learning provide an evolutionary advantage to the individual (2001, p. 117). There is a link to policy here because solutions may be transmitted over time, which may help overall efciency, though it may lock policymakers into suboptimal solutions. In this way, Jones is close to accepting the existence of memes. He argues that organizations evolve using existing structures to meet new environmental demands (p. 159). He refers to the random processes of trial and error learning, which is at the heart of social adaptation. Like modern evolutionary theory, Jones rejects the idea that such adaptive mechanisms lead not to a general equilibrium but to a series of partial equilibria (p. 186). Error correction can help avoid policy failure, but some genetic dispositions, such as emotions, can give rise to mistakes by policymakers. Though Jones considers political institutions to be shaped by human choice, which makes politics fundamentally different to natural evolution, he argues these, too, are rarely set in stone and change through interaction and adapt on the y, rather like the evolution of committee jurisdictions (Jones, Baumgartner, & MacLeod, 2000). Although the book closes by an examination of the implications of adaptation, Jones does not develop a full-scale evolutionary model; such elements are hinted at in the text, as indicated above, but the book is more about decision-making theory. The Policy Advocacy Coalition Framework Sabatier and Jenkins-Smiths (1993) policy advocacy coalition framework is different than the two discussed above. The key idea of the framework is that there exist sets of core ideas about causation and value in public policy. These coalitions form because certain interests link to them. There are several such ideas, which create about two to four coalitions, and it is possible to map these networks of actors within a policy sector. Change comes from the ability of these ideas to adaptin their noncore aspectranging around a whole series of operational questions and what works in any one time or place. Partly in response to wider social and economic changes or from political events and also from policy learning, the balance of power in these networks changes and the structure and memberships of the coalitions alter. The cogency of this approach is threefold. First, it leaves behind the idea that policy sectors are composed of integrated networks; instead, they are a political terrain through which different collations ght it out, which is far closer to contemporary reality. Second, the ideational approach to policymaking is fully integrated into the way in which the coalitions operate, so it provides a grounded way of understanding the importance of discourse in the political process. Third, the advocates have developed an effective research program to map the development of advocacy coalitions, through the coding of representation to legislative committees.

John: Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations

491

The approach has a neat account of policy change, which occurs through the interaction between wide external changes or shocks to the political system and the success of the ideas in the coalitions, which may cause actors in the advocacy coalitions to shift coalitions, even just for tactical reasons. It is just possible that there is an implied evolutionary process, as some adaptation happens when partners maneuver for strategic advantage within and across the coalitions. No coalition mindlessly advocates arguments that are unlikely to win; new situations may deserve new arguments, and perhaps switches across the coalition divide. So some success may be explained by tness for purpose, others by adaptation. Of course, there are limits on this adaptation: the strength of the core ideas. But that is entirely consistent with evolution, which is not predicated on plasticity, merely the possibility of adjustment within limits set by past adaptation. An Evolutionary Theory of Public Policy Because some writers on public policy use evolutionary ideas in their arguments, it may be possible to argue that the next logical step is to develop the theory more fully, which would be similar to other areas of the social sciences that have developed evolutionary thinking, such as economics, game theory, economic geography, social theory, and evolutionary psychology. There are a number of ways of applying evolutionary ideas in the social science, but some are not particularly relevant for public policy. Part of the problem is that the debate about evolution is not clear, with many terms being used vaguely or rhetorically. Evolution in biology refers to a change in the gene pool of a population over time. The core mechanism is largely as Darwin argued: Genes randomly mutate, individual organisms are selected by their environment, and populations evolve. There may be small variations; also, large variations may cause the emergence of new species. But the core is some process of selection. It is possible to use evolution loosely in everyday speech to mean slow change, but if it is to make sense in social sciences, similar processes must be at work: a gene and a form of mutation and selection, which means that the population changes in composition and functioning. There is causal processone from gene mutation, the other from selection. If these elements are not present, then the evolutionary label is a heuristic device or an analogy. Thus the work on complexity and chaos theory, which is interested in the prevalence of nonlinear, S-type relationships in the natural world, when applied to society and to politics (e.g., de Greene, 1994), is part of, but not the whole of, an evolutionary account. Thus the identication of chaotic process goes some way to indicate that an evolutionary or evolutionary-like mechanism is at work, but it is not the evolutionary process as such. The other set of writings about policy and social theory uses the evolutionary label to examine patterns of rapid policy change or system change and to study where cultures and policy ideas seem to evolve in stages. The evolutionary take on social history offered by Runciman (1989) resembles this pattern, whereby cultures evolve. Closer to public policy are some recent accounts of British politics,

492

Policy Studies Journal, 31:4

such as the study of the U.K. Conservative government 19791997 (Kerr & Marsh, 1998), and recent historical sociology (e.g., Hay, 1998, 2002). Hay regards evolution as incremental, path-dependent and directional change which is circumscribed in some way (1998, p. 31, fn. 69), which is broader than the biological denition. Path-dependent change need not be evolutionary, though. Even though many of these authors are looking for selection mechanisms, they do not get very far. Just saying that selection mechanisms exist does no more than state the obvious, as in some way selection happens all the time with new policies, policy instruments, and personnel; what matters is whether that selection is itself evolutionary. As Ward (1998, p. 4) writes, [A]ttention is best restricted to areas where there is selection of variants which are produced blindly. The most powerful application of evolution to the social science is evolutionary psychology, but it does not make sense to use this in general public policy theory, though it is useful as a guide to decision-making. If modern skulls house stone-age minds, which evolved in a path-dependent way, decision making will be suboptimal and deviate from the rational model. But again, such processes and outcomes do not add up to an evolutionary model of public policy; they are about the impact of biological evolution on public policy. Rational Choice and Memetic Models Rational choice theory and memetic models, or a combination between the two, can offer an account of the causal process at work. The advantage of the rational choice approach, particularly evolutionary game theory, is that it can provide a rigorous account of change over time. The cost is the need to adhere to the assumptions behind the game; but if they hold, even partially, there is a form of evolutionary selection, especially as it is possible to relax some of the more problematic assumptions and assume some bounded rationality. An example of what is possible is Wards game theoretic accounts of state change (1998, 2003). He sets out a coordination game to explain why the state may adopt certain policies to regulate business. There is a population of rms and a population of governments. He examines the phenomenon of coevolution, whereby new styles of economic and political governance evolve together. The evolutionary take on this is how mutants or alternative strategies affect the game. In the equilibrium situation, the switching of a strategy is punished, given all the other players strategies, which can exist in the new or the old arrangements. This outcome comes from a Nash equilibrium in which changing strategies would get a lower payoff, because the rms in this game get certain rewards, which are partly based on the regulatory regime in place. There is an incentive against change. The interest of evolutionary game theory is to model what happens when the equilibrium shifts. Ward suggests that the payoffs from the game change if rms get monopoly rents from adopting a new technology. If they are likely to trade with each other, they can create a subset of rms that can stay in business even under

John: Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations

493

the old regulatory regime. This creates the evolutionary path where replication and genetic advantage have been fostered. In turn, the governments also change their strategies and move to a new equilibrium point. From the policy point of view, there would need to be a number of policymakers, say branches of the state or state and local governments, and a range of other actors, and what matters is whether the subset can get selective benets from innovation and can support each other. In this way, Wards model is similar to allopatric theory, popularized by Mayr, in which new species arise in very small populations, though the difference is that this theory is not about the rents of innovations but species that become isolated from their parental group, which often appear in marginal areas. It is possible that Wards model could be extended to accounts of diffusion and innovation, where the state governments are seen as a strategic actor. And Wards model is not the only way of doing it. Farkas (1996), for example, uses a formal model, with chance and tness built in, to generate a simulation to policy solutions over time. Although it is not as efcient as the biological example, it moves in that direction. Another variant is diffusion in social networks as an example of replication and the simulation of the results of idea propagation. Proponents of such accounts of evolution need to think carefully about what is being selected. To answer this question, social scientists have debated the existence of memes. As invented by Dawkins (1976), a meme is an information pattern, held in an individuals memory, which is capable of being copied to another. Memes transmit over time, inuence behavior, and become successful. They are different to genes in that they are more mutable, but for the evolution idea to work, there must be some stickiness or permanence to them and a sense that the ideas exist apart from each other. So although an environmentalist and an economic development advocate may share some ideas, it is clear what distinguishes them. Thus it makes sense to examine the success and failure of environmentalist ideas, for example. The unit of selection is the idea or practice, which needs expression by a human agent, who has an interest in the ideas survival. The human agent is dened by the interest but needs the idea to provide direction and identity. Ideas need to have concrete expression in human agencythey need a carrier. Ideas are made of different elements, which may correspond to the genes in the mimetic account. Mutation occurs through random processes of trail and error of these elements, either through errors or through what biologists call recombination when elements recombine to make new genes. Stronger forms of evolution occur when new species are createdspeciation. For each individual, such selection may appear as a conscious activity. But when seen as a set of individuals each trying out new strategies, it appears as a stochastic process, with some individuals being lucky in hitting on the combination that has the chance of transmission and replication. In terms of policy, the key people are Kingdons policy entrepreneurs, who are activists with a particular interest in the success of the policy, though in a less acute sense everyone is

494

Policy Studies Journal, 31:4

an entrepreneur who has a stake in the policy outputs and outcomes: the citizens who vote for policies, politicians who seek to maximize votes and capitalize on policy opportunities, and bureaucrats who have a stake in the implementation of particular policy choices. Evolution suggests a key moment when trial and error or recombination allows a successful replication to take place and can then expand and take hold and challenge an existing policy monopoly. What happens is that the idea not only takes hold in peoples heads, but it become institutionalized in practices and routines, such as by governments as in Wards game, which then continue over time. Similar to the punctuated equilibrium model, the evolutionary equilibrium is hard to dislodge. Empirical Investigations Evolution does not disrupt the claims of the punctuated equilibrium and policy streams accounts; what it claims to offer is a better account of the causal processes at work in policy change. In that sense, evolution is not a new paradigm; it is just an extension of the public policy theory of the 1990s. To understand replication through time, empirical studies can map the elements to the model and nd out how they t together: the nonlinear processes, the separation of the ideas into nonmalleable units of selection, the random recombinations of ideas from the gene pool, the competition between units of selection, the way in which the intentions of central policymakers are driven by these processes, and examination of the payoffs in the development of evolutionary pathways. Unfortunately, detailed studies are rare, possibly because selection is a messy business and it is hard to identify memetic processes. Examples include tax policy (Steinmo, 2003). Another is the U.K. poll tax (John, 1999), which examined the fate of different proposals for local taxes that the U.K. government faced, including the controversial decision in 1990 to introduce a single person tax, the poll tax, and was subject to a high degree of attention and opposition. Essential was the role of a small group of policy entrepreneurs who drove the policy forward. The single person tax beat its competitors; the policy process become unstoppable, but further selection during the implementation process killed the idea. This account has attracted criticism (Dowding, 2000), largely because the speed of events in the poll tax episode makes it unlikely candidate for a nonintentional process, though it is not easy to separate out unconscious from conscious policy evolution because every policy usually has an element of conscious choice, just as every game player does when acting for strategic advantage (John, 2000). In Dowdings view, it all happened too suddenly and at too high a level of conscious choice, whereas a more defensible account of policy would examine the long-term evolution of memes through unconscious reection and transmission. Rather than ideas, memes are practices that have become institutionalizedquite the opposite of the poll tax episode. Another familiar criticism is that it is possi-

John: Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations

495

ble to say that much of what happens in politics is not the result of imitation but of power. Even Runciman (1999) recognizes this: And what about cultural behaviour which is the result not of imitation or learning but of enforced obedience to instructions from lawgivers and other people in positions not merely of inuence but of power? (page 2112) There are also problems with memes, as they lack the clarity of genes and seem to be an amalgam of thought processes and practices. Midgley (2000) cautions that the dream of standardizing the world into comparable units is an old chimera. Even if memes exist, they are anything but standard. Researchers on evolution need to address the problems of transferring models from the natural to the social world, mainly because the causes are different. As it is not possible to identify mechanisms in the same way, it may be the case that evolutionary theory becomes just another way of representing social and political facts, not adding very much to conventional narratives and models. The other objection comes from the nature of politics itselfdoes the fact that politics is capable of imagining itself more than any other aspect of human existence automatically rule out evolution? The conscious choices that political actors make are made not only to advance their interests but are about what kind of world they prefer. Evolutionary approaches based on the automatic adjustment of tness for purpose will never work smoothly and probably serve as the background for the emergence of new political ideas rather than their sole determinant. Conclusion This journey through the policy theory literature does not reveal an imminent change to the central debates in public policy theory. In the main, writers on public policy are working within the lines of argument set out by Frank Baumgartner, Bryan Jones, John Kingdon, Paul Sabatier, and Hank Jenkins-Smith. This review seeks to show that it is a further possibility for advance, largely within the same mode but with more attention to the core causal processes than before. With more ne-grained tools of analysis, writers in public policy can proceed with less fear that they are applying social science labels that describe rather than explain the policy process. Evolutionary theory is one possible line of advance, which might be able to uncover processes not normally observed by political scientists. Although much is uncovered by institutional processes, long-term social and economic ideas, networks, strategic interaction, and the conscious adoption of ideas, the claim is that random processes, competition, and selection exert a background inuence, which can drive the policymaking and implementation processes. But if evolution is to avoid being another tag or heuristic device, it needs a great deal of attention to methods and denitions. Careful tests of decision making is the strength of the researchers testing the policy advocacy coalition framework, policy windows, and the punctuated equilibrium model, and no doubt future tests of policy theory will show the same attributes.

496

Policy Studies Journal, 31:4

Peter John is Professor of Politics in the School of Politics and Sociology, Birkbeck College, University of London, where he teaches the masters degree in Public Policy and Management. He is author of Analysing Public Policy (1990) and Local Governance in Western Europe (2001). Note
This article was rst given as a paper to the panel, Theories of Policy Change in Democracies: a 30 Year Perspective, which took place at the Midwest Political Science Association Conference, Chicago, April 36 2003. I am grateful to the panel conveners and the Policy Studies Organization for inviting me to speak and to the discussants and other participants for their stimulating comnments.

References
Axelrod, R. (1984). the evolution of cooperation. New York: Basic Books. Baumgartner, F., & Jones, B. (1991). Agenda dynamics and policy subsystems. Journal of Politics, 53, 10441074. Baumgartner, F., & Jones, B. (1993). Agendas and instability in American politics. Chicago: University of Chicago Press. Baumgartner, F., & Jones, B. (Eds.). (2002). Policy dynamics. Chicago: University of Chicago Press. Browne, W. (1995). Cultivating Congress: Constituents, issues, and interests in agricultural policy making. Kansas University Press, Kansas. de Greene, K. (1994). The challenge to policy-making of large-scale systems. Journal of Theoretical Politics, 62, 161188. Dawkins, S. (1976). The Selsh Gene. Oxford: Oxford University Press. Dowding, K. (1995). Model or metaphor: A critical review of the policy network approach. Political Studies, 431, 136158. Dowding, K. (2000). How not to use evolutionary theory in politics: A critique of Peter John. British Journal of Politics and International Relations, 2, 7280. Dowding, K. (2001). There must be an end to confusion: Policy networks, intellectual fatigue, and the need for political science methods courses in British universities. Political Studies, 491, 89105. Farkas, A. (1996). Evolutionary models in foreign policy analysis. International Studies Quarterly, 40, 343361. Hay, C. (1998, March). Political time and the temporality of crisis On institutional change as punctuated evolution. Paper presented at ECPR Joint Sessions, University of Warwick. Hay, C. (2002). Political analysis. Basingstoke, United Kingdom: Palgrave. John, P. (1998). Analysing public policy. London: Cassell. John, P. (1999). Ideas and interests; agendas and implementation: An evolutionary explanation of policy change in British local government nance. British Journal of Politics and International Relations, 11, 3962. John, P. (2000). The uses and abuse of evolutionary theory in political science: A reply to Alan McConnell and Keith Dowding. British Journal of Politics and International Relations, 11, 8994. John, P., & Margetts, H. (2003). Policy punctuations in the UK: Fluctuations and equilibria in central government expenditure since 1951. Public Administration, 81, 411432. Jones, B. (1994). Reconceiving decision-making in democratic politics: Attention, choice, and public policy. Chicago: University of Chicago Press.

John: Is There Life After Policy Streams, Advocacy Coalitions, and Punctuations

497

Jones, B. (1998). Bounded rationality. In N. Polsby (Ed.), Annual review of political science. Jones, B. (2001). Politics and the architecture of choice. Chicago: University of Chicago Press. Jones, B., Baumgartner, F., & MacLeod, M. (2000). Jurisdictional evolution in the legislative process. Journal of Politics, 62, 321349. Jones, B., Baumgartner, F., & Talbert, J. (1993). The destruction of issue monopolies in Congress. American Political Science Review, 87, 657671. Jones, B., Baumgartner, F., & True, J. (1998). Policy punctuations: U.S. budget authority, 19471995. Journal of Politics, 60, 133. Jones, B., Baumgartner, F., & True, J. (2002a, March). Patterns and punctuations in the U.S. budget. Discussion paper for conference, Budgetary Policy Change: Measures and Models, Oxford. Jones, B., Baumgartner, F., & True, J. (2002b, March). Policy Macropunctuations: How the U.S. budget evolved. Discussion paper for conference, Budgetary Policy Change: Measures and Models, Oxford. Jones, B., & Sulkin, T. (2001). Is politics normal? Information processing and institutionally-imposed costs in policy change. Unpublished paper, Center for American Politics and Public Policy, Department of Political Science, University of Washington, Seattle. Jones, B., Sulkin, T., & Larsen, H. (2003). Policy punctuations in American political institutions. American Political Science Review, 971, 151169. Jones, B., True, J., & Baumgartner, F. (1997). Does incrementalism stem from political consensus or from institutional gridlock? American Journal of Political Science, 41, 13191339. Kerr, P., & Marsh, D. (1998, March). Explaining Thatcherism: Towards an evolutionary and multi-dimensional approach. Paper presented at the ECPR Joint Sessions, University of Warwick. Kingdon, J. (1984). Agendas, alternatives, and public policies. Boston: Little, Brown. Kingdon, J. (1995). Agendas, alternatives, and public policies (2d ed.). Boston: Little, Brown. Lane, J., & Svante, E. (1999). The new institutional politics. London: Taylor and Francis. Marsh, D., & Rhodes, R. (1992). Policy networks in British government. Oxford: Oxford University Press. Marsh, D., & Smith, M. (2001). There is more than one way to do political science: On different ways to study policy networks. Political Studies, 49, 528541. Midgley, M. (2000). Why memes? In H. Rose & S. Rose (Eds.), Alas, poor Darwin (pp. 6784). London: Jonathan Cape. Ostrom, E. (1998). A behavioral approach to the rational choice theory of collective action: Presidential address, American Political Science Association, 1997. American Political Science Review, 92, 122. Ostrom, E. (1999). Institutional rational choice. In P. Sabatier (Ed.), Theories of the policy process (pp. 3571). Boulder, CO: Westview. Runciman, W. (1989). A Treatise on Social Theory (vol. 2). Cambridge, United Kingdom: Cambridge University Press. Runciman, W. (1999). Darwinian soup. London Review of Books, 21, p. 2112. Sabatier, P. (1999). The need for better theories. In P. Sabatier (Ed.), Theories of the policy process (pp. 117). Boulder, CO: Westview. Sabatier, P., & Jenkins-Smith, H. (1993). Policy change and learning: An advocacy coalition approach. Boulder, CO: Westivew. Sabatier, P., & Jenkins-Smith, H. (1999). The advocacy coalition framework: An assessment. In P. Sabatier (Ed.), Theories of the policy process (pp. 117166). Boulder, CO: Westview. Shoup, B. (2001), Policy section members vote for the most important works in public policy, Policy Currents, 11, 1415. Steinmo, S. (2003). The evolution of policy ideas: Tax policy in the 20th century. British Journal of Politics and International Relations, 52, 206236.

498

Policy Studies Journal, 31:4

Steinmo, S., Thelan, K., & Longstreth, F. (1992). Structuring politics: Historical institutionalism in comparative analysis. Cambridge, United Kingdom: Cambridge University Press. Strom, K., Muller, W., & Bergman, T. (2003). Delegation and accountability in parliamentary democracies. Oxford, United Kingdom: Oxford University Press. Taylor, M. (1987). The possibility of cooperation. Cambridge, United Kingdom: Cambridge University Press. True, J. (2000). Avalanches and incrementalismmaking policy and budgets in the United States. American Review of Public Administration, 30, 318. True, J., Jones, B., & Baumgartner, F. (1999). Punctuated-equilibrium theory: Explaining stability and change in American policymaking. In P. Sabatier (Ed.), Theories of the policy process (pp. 97115). Boulder, CO: Westview. Ward, H. (1998, March). Bandwagons rollor not: The co-evolution of modes of regulation and patterns of rule. Paper presented at ECPR Joint Sessions, University of Warwick. Ward, H. (2003). The co-evolution of regimes of accumulation and patterns of rule: State autonomy and the possibility of functional response to crisis. New Political Economy, 8, 179202.