6 Structures, Environn1ents, and

Con1plex Systen1s

Key themes of this chapter:

•
A focus on structures, policy conditions, environments, or complex systems is a
useful corrective to individualist and 'top-down'accounts of policymaking.
•
Policy and policymakingis not controlled by a small group of policymakers at the
'centre' of government. They do not control their policymakingcontext. Rather,
it influences their attention to problems and the feasibility of solutions.
Policy conditions include the demographic, social, technological, natural, and
•

economic changes that often seem to drive the policy agenda.

•
The wider policymakingenvironment includes the institutions, networks, and
ideas that new governments inherit before they make new choices.
•
Many policy theories use many different evolutionary metaphors to describe
policy actors adapting to their environments.
• In complex policymaking systems, policy often seems to 'emerge' locally in the
absence of central control.

INTRODUCTION: STRUCTURE AND AGENCY IN THE POLICY PROCESS

When we seek to understand policy change, we attribute the exercise of power to the
individuals that make policy but also recognize the context in which they operate and the
pressures that they face. This context prompts us to consider a different perspective to
'comprehensivelyrational' policymaking and the idea that the policy process begins with
the decision by a policymaker to identify a problem to solve (Chapters 2 and 4). Policy
analysis begins with the consideration of context, not a blank sheet of paper (Bardach and
Patashnik, 2015; Cairney and Weible, 2017). Context includes geographical and socio-
economic conditions,the governmentinfrastructure and policies that are already in place,
and the events that often seem to be out of the control of policymakers and prompt them
to act.
There are many ways to characterize this process, with different approaches striking
a different balance between the role of structure and agency (Chapter 1). Some accounts
seem to focus on 'structural power' and profound limits to agency (Chapter 3, pp.51-2),
but most are really trying to 'zoom out' from elite individual action to describe a wider per-
- spective on policymaking. The term 'structure' refers vaguely to a set of parts put together
in a particular way to form a whole. We tend to attribute two key properties to structures:

94
 Structures, Environments, and Complex Systems 95

they are relativelyfixed and difficult but not impossible to break down; and they influence
the decisions that actors ('agents') make. Examples include the structure of the domestic
-

and global economy and the rules and policies that new governments inherit when they
-

enter office.
Policy studies may not always use the language of structure and agency directly, but
they supplementa focus on policymaker action by highlighting context. We know that the
decisions made by policymakers can have a profound effect on society, the economy, and
government. However, we also know that the choices they make are often based on policy
problems over which they have limited control, such as an increasingly global economy;
demographic change; and events that propel issues to the top of their agenda (Chapter 9).
There are limits on their ability to anticipate the effects of policy and carry it out success-
fully (Chapter 2). Policymakers operate within the context of rules, norms and cultures
that influence their behaviour (Chapter 5). Further, such is the scope of the state that indi-
vidual policymakers could not hope to have a full understandingof it, and their decisions
while in office will likely represent a small part of overall governmentactivity. We miss a
lot if we only focus on their action, not the big picture.
This chapter identifies the main ways in which we can describe this policymaking
context. Policy conditions describe what policymakers take into account when identifying
problemsand deciding how to address them. They may be particularly aware of a politi-
cal system's size, demographic structure, economy and mass behaviour. These conditions
represent a source of pressure on how policymakers operate. For example, an ageing
population puts pressure on governments to address pensions policy and their provision
of social care, while an economic crisis generally rises to the top of a government'sagenda
(Box 6.1).
Policy environment is an evolutionarymetaphor. It brings together the key concepts of
this book: the spread of actors and responsibilities across many levels and types of govern-
ment (Chapter 8); the institutions associated with each venue (Chapter 5); the networks
between policymakers and influencers (Chapter 9); the ideas on which policy actors draw
(Chapter 11); and structural or socioeconomic factors (and events) (Box 6.1). Combined,
these factors produce the sense that the main influences on policymakingoften seem to be
overwhelmingor out of the control of individual policymakers. Further, each policy theory
brings together these factors to provide their own story about how policy actors engage
with their environment.
Finally, in this chapter, we focus on the ways in which complexitytheory tells that story.
The language of complexity seems to downplay the role of central governments and the
power of individual policymakers. Complexitytheory suggests that we shift our analy-
sis from individual parts of a political system to the system as a whole -
as a network of
elements and systemic behaviour that cannot be broken down easily into its constituent
parts. Complex system describes the interdependence between a huge number of actors,
albeit often in pockets of local activity from which outcomes 'emerge'. Policy results
from their interaction, the system seems to dampen or amplify the effects of action, and
policy outcomes often appear to emerge locally despite attempts at central control. This
approach serves as a corrective to accounts that focus too much on the ability of individual
policymakers to change policy single-handedly. It is useful as long as we do not go too far
towards the idea that structures determine behaviour and remove the role of agency, or
the ability of individuals to deliberate and make choices.
96 Understanding Public Policy

Box 6.1 Structural factors and policy conditions

Historic-geographic. Factors such as the climate and stock of natural resources

•

influenced how countries and regions developed. The population size of politi-
cal systems may inform how institutions are set up. Policies for dense, urban
areas may not be popular in, or appropriate for, rural areas (Hofferbert, 1974:
228-29).
Demographic. An ageing population puts pressure on the costs of social security
•

and personalcare. A 'baby boom' produces greater demand for schools.

• Economic. Global and domestic economic activity influences the size of govern-
ment tax income, its ability to attract foreign investment, and the effect of inter-
est rate changes on inflation and investment.
•
Social and behavioural. Social attitudes and behaviour influence the popularity
and effect of policy measures. Unhealthybehaviour in the population affects
demand for healthcare services.
•
Technological. The mass productionof cars produces new demands on transport
policy. Medical advances produce demands for new medicines and equipment
(Hagwood, 1992a: 191-208). Information technology changes the way we
engage in politics and design and use public services (Dunleavy et al., 2006).
•
Institutional. Formal institutions describe the structure of government.Informal
institutions and norms influence behaviour within it (Chapter 5).

DO STRUCTURAL FACTORS DETERMINE POLICY AND POLICYMAKING?

A focus 'structures' helps raise a fundamental question: how much do policymakers

on

shape, and how much of their behaviour is shaped by, their surroundings (Hay, 2002: 89)?
Box 6.1 outlines the main 'structural' factors that help shape policy, while John (1998: 92)
highlights how strongly we could associate them with policy outcomes:

The simple idea is that the policy process, far from being a rational weighing up of

alternatives, driven
is by powerful socioeconomic forces that set the agenda, structure
decision makers choices, constrain implementationand ensure that the interests of
the most powerful (or of the system as a whole) determine the outputs and outcomes of
the political system.

Historical approaches suggest that structural factors are

System 'integrated
-

something to
ignore at peril. For example, Hagwood
our
group of interacting
(1992a: 191-208) compares demography and changing par- elements designed
ties in government and concludes that the former often had a jointly to perform a
larger effect on trends in public expenditure, employment,and given function' (such as

regulation. a political system that

Further, many empirical studies inspired by systems functions to 'allocate
values' Dawson and
-

theory (Easton, 1953, 1965; Deutsch, 1970) test the sta-

Robinson, 1963: 267).
tistical association between the socioeconomic factors that
 Structures, Environments, and Complex Systems 97

produce demands feeding into the political system and policy outputs (Blomquist,2007:
261). Dawson and Robinson (1963), Dye (1966), and Sharkansky (1972: 21) argue that
these factors explain variations in, for example, the demand for and provision of public
services, albeit while mediated by political parties, legislatures, and executive adminis-
tration. Further, Hofferbert's (1974: 228-30; see also Wilder, 2017) 'funnel of causality'
describes how:

•
historic-geographic conditions contribute to the socioeconomic compositionof a region
(climate and natural resources influence populationdensity, employment,and levels of
prosperity);
• the concentration of social groups in regions may be traced back to historical events;
• the socioeconomic composition of a region contributes to 'mass political behaviour'
such as voter turnout, which influences the structure and fortunes of parties; and
• these factors combine with 'governmentinstitutions' to influence elite behaviour.

These descriptions are not deterministic. For example, Dye (1966: 299-301) rejects a
sole focus on individual and group action, then explores how socioeconomic factors can
be taken into account in the 'little black box labelled political system'. Hofferbert (1974:
231-33) argues that, while elites cannot ignore their policy environment, there is still
scope for 'leadership' and 'vigorous action against strong historical and economic forces',
and structural impact varies by policy issue over time (1974: 237-41). In other words,
they portray an iterative process: socioeconomic factors influence outcomes, but the
impact of policy feeds back to the policy environment. Still, there was a strong academic
reaction to the perception that the 'Dye-Sharkansky-Hofferbert'(DSH) approach down-
played the idea that 'politics matters' (Blomquist, 2007: 270-71). Studies sought to 'bring
politics back in' by comparing countries rather than US states and finding a much clearer
link between, for example, levels of social welfare spending and the role of political parties
(John, 1998: 105-6; Blomquist,2007: 264-65).
More generally, it is difficult to conclude that structural factors determine politics,
for two main reasons. First, the structure and agency-inspired policy literature presents
a more dynamic process in which structures influence agents and agents often make or
reconstitute the structural constraints within which they operate (Marsh and Smith,
2000: 5; Adler and Haas, 1992: 371; Jacobsen, 1995: 300). Different texts describe this
relationship differently (Hay, 2002: 118-21), but few present a completely structural
account (compare with Chapter 3).
Second, we cannot define structures
objectively and there are few conditions that, like
a meteor about to fall
on your head, cannot be ignored. Rather, they are subject to inter-

pretation; policymakers attach different meanings to what appear to be the same factors
(Blomquist, 2007: 274-75). They mediate the effects of structural factors and external
events by defining their importanceand acting on them in a particular way (Chapter 10).
They may pay attention to some and ignore others for long periods. Therefore, a focus on
structures resembles the study of agenda setting (Chapter 9) in which actors try with -

variable success to convince policymakers that certain factors are causing a crisis that
-

should not be ignored.

Yet we should not far in highlighting the role of discretion and attention.
go too
Indeed, the fact that policymakers in most countries could choose to focus on any issue
but actually focus on a small number of issues such as the economy
-
reinforces the
-
98 Understanding Public Policy

Box 6.2 Japan's nuclear policy

Japan's choice of nuclear power as a key of electricity demonstrates the

source

strong, but not determining, nature of external factors on decisions (see Park and
Sovacool, 2018). In this case, we are literally describing the policy environment
because energy policy depends on two potentially competing factors. First, Japan
has few indigenous sources of energy such as coal or gas, requiring it to import
natural sources or rely on nuclear power. Second, it is vulnerable to natural disasters
such as earthquakes that undermine reliance on nuclear power (even if considerable
measures are taken to make it secure). The resultant policy, including nuclear as a

key source, follows policymakers weighing up these sorts of constraints but with-
out policy being determined by them. This case demonstrates how difficult it is to
describe structural constraints. In the aftermath of the earthquakes in March 2011,
one scientific commentator (on the rolling BBC news coverage) still maintained
that Japanese governmentseffectively had to rely on nuclear power; that they had
no choice. This type of example may be what Ward (1987: 602) has in mind when
he describes situations which appear to make 'certain acts unthinkable or physi-
cally impossible' or 'so costly that actors are structurally constrained from carrying
them out'. This constraint is far less apparent elsewhere, and the Japanese crisis has
prompted many governments in other countries to reassess their energy portfolio
(Pidd and Goldenberg, 2011; Wittneben, 2012; Huang et al., 2018; Chapter 9).

importance of socioeconomic factors. A sensible way to address this problem analytically

is to think in terms of the structure-agency mix, with some issues providing more or less
constraint on choice than others. For example, the ageing population may give govern-
ments little choice but to plan for the consequences, but they can do so in a variety of
ways. Technology-drivenhealthcare is resistible, and public health policies are often more
cost-effective, but many governments still seem to favour medicines and acute hospitals
strongly. Adapting to our physical environment also involves a complex mix of structure
and perception (Box 6.2). For example, coastal conditions may appear to force us to build
protective barriers, but policymakers can ignore the issue for some time (Werritty, 2006).
Many policymakers also seem determined to deny or ignore climate change (Tangney,
2018).

THE ECONOMIC CONTEXT: MARXISM AND GLOBALIZATION

Chapter 3 discusses the continued relevance of Marxism to studies of power, while Kiely
(2017) describes its importance to studies of globalization.Marxist accounts (broadly
defined) help us construct a simple story of policymaking in which elites, and the structure
of government, seem devoted to protectingthe capitalistsystem of economic production,
giving economic interests a privilegedposition within the policy process. In this scenario,
our assumption is that the capitalist system benefits one class of people (those who own
or control capital) at the expense of another (the working class, who do not). Elites in

powerful positions in government share relationships with business elites. For example,
 Structures, Environments, and Complex Systems 99

the ruling and capitalist classes share a common, privileged, background that predisposes
them towards working with each other. Or, the world of business provides an incentive for
policymakers to cooperate, by providing campaign funding or a source of employmentin
the future (Cairney and Yamazaki, 2018: 259). Consequently,government elites forge close
networks with representatives of business, ensuring that the latter dominate access to pol-
icymaking at the expense of other interests (John, 1998: 94). It is in the interests of most
governments or policymakers to ensure that the capitalist system runs smoothly, since
this provides employmentfor its voters and government income through taxes. Overall,
we have a strong explanationfor the role of business within the policy process (compare
with Lindblom,1977).
It is possible to take this argument to its extreme, to treat capitalist protection as a
functional imperative, inevitable and immune from change, with agents merely func-
tioning according to particular roles, with no real ability to make choices (see Hay, 2002:
116). In this scenario, education and healthcare systems exist to socialize and maintain
a functioning workforce. However, the problemwith such arguments is that it is possible

to interpret all policy developmentsin this way without evidence or demonstratinghow

the process works (John, 1998: 96). When we 'unpack' the process, we find two problems.
First, many policies in many countries appear to reflect the beliefs of governmentsor the
power of unions, such as the welfare state, the living wage, or measures to protect union
rights. Second, we can identify several types of capitalistclass that may not share the same
interests. For example, the issue of interest rates may divide the manufacturer and the
banker.
The same can be said for 'globalization',an important but vague term to describe the
international spread of processes that used to be confined to a small number of countries.
Economic globalization describes global or deregulated financial markets, technology
which allows greater interactions -
such as trade -
across the globe, the role of 'global
corporations', migration, and/or the unequal relationshipbetween the Global North and
South (Hill, 2005: 45; John, 1998: 103; Gangopadhyay, 2017; Hopper et al., 2017). It
also describes the social and cultural integration that tends to follow greater interaction
between countries and their populations(Parsons, 1995: 242-43).
Globalization could describe what happens when nation-states experience a diminished
ability to control their own economic and monetarypolicies and the actions of large mul-
tinational corporations (MNCs). Governments appear to be forced to compete with each
other, react to shifts and crises in international financial conditions, and change to attract
business from MNCs. The focus is on economics determining politics, with moves from the
1980s towards 'deregulation, privatization and cutting back welfare' representinga world-
wide response to globalization(John, 1998: 103; Friedman, 2000; Quiggin, 2006: 536).
Such arguments are incompleteunless they identify a clear causal process and some evi-
dence (Hay, 2002: 114). For example, we may identify an inevitable 'race to the bottom' as
governments acting on the belief that MN Cs will not invest in countries with high taxes,
-

a public sector-dominated economy, and restrictive labour and environmental regulations -

reduce the size of the public sector and corporation taxes and deregulate to attract foreign
direct investment. Yet this race has not run in a uniform way and, for example, the overall
size of the public sector has not diminished (Hay, 2006b: 591). Rather, different policymak-
ers in different countries have reacted to their environment in different ways (Hoberg,
2001: 128-30; Hay, 2002: 253; Quiggin, 2006: 537). Policymakers are constrained by the
need to maintain a reasonable balance between revenue and spending (the budget) and
100 Understanding Public Policy

imports and exports (the balance of payments) (Quiggin, 2006: 530-35). However, there
are many ways to attempt to do so, including running a deficit in one term of office, on the
assumptionthat a future governmentwill be obliged to produce a surplus (Quiggin, 2006:
540). Public expenditureas a percentage of GDP also varies, quite markedly, over time
(Hogwood, 1992a: 43; Hindmoor, 2003: 208). More significantly, it fluctuates by policy
area, as a combination of inertia and political choices produces minimal change in most
areas but profound change in others (John and Margetts, 2003: 421; Chapter 9).

Therefore, empirical policy studies suggest that 'structural forces' are important but
they do not determine political behaviour; there is always some degree of choice. We may
agree that one choice seems inevitable because it seems much better than the rest. Or, the
appearance of constraint and inevitability may be convenient for policymakers attempting
to introduce unpopular policies or avoid responsibilityfor poor results (Hay, 2002: 259).
However, there is choice nonetheless and 'the constraints of globalizationare as much as
anything else, what political actors make of them' (Hay, 2006b: 587). Further, a key reason
for country-levelpolicies to stay distinctive is that they are influenced by other structural
factors, 'including distinctive national values, different political institutions, and the
legacy of past policies' (Hoberg, 2001: 127; Sinclair, 2004). Yet we should strike some note
of caution: most of these studies are of the 'West' or Global North. As Chapter 12 suggests,
policymakersin the South often face more political and economic pressures to reform to
become more like the international norm (see also Kiely, 2017: 123).

INHERITANCE BEFORE CHOICE, AND POLICY SUCCESSION

The 'legacy of past policies' is captured best by Rose (1990), who describes the cumulative
effect of decades of policies: newly elected policymakers inherit a huge government with
massive commitments. Most new decisions are based on legislation that already exists,
and the bulk of public expenditure is spent on government activities that continue by rou-
tine, carried out by public employees recruited in the past. Current policymakers choose
to 'uphold the laws of the land' before making new ones, and the effect of their new policy
choices is small compared to the sum of governmentactivity (1990: 263). Much current
government policy results from the choices made by former policymakers, and policy-
makers 'choose' to address the 'consequences of inherited programmes that would not
have been chosen by the current incumbents' (Rose, 1990: 264, 229). Note the difference
between inheritance and incrementalism (Chapter 4). Rose and Davies (1994: 31) suggest
that incrementalist analyses tend to be relativelyshort term; longer-term analyses show
that marginal steps made in the past have produced 'massive commitments' such as the
modern welfare state and tax systems (or the breakdown of such systems -
Richardson,
2018). These policies are maintained and reproduced by government organizations that
operate without requiringfurther authorization.
The logical implication of bounded rationality is that when policymakers pay attention
to one issue they must ignore 99 others. When they invest a significant amount of effort
and attention to policy change on one issue, they accept that they cannot do the same for
most others (Rose, 1986; compare with Chapter 9). Therefore, most of the day-to-day
delivery of policy takes place without significant policymaker involvement. Most public
sector organizations continue to implementpolicies that were legislated and budgeted for
in the past (Rose, 1990: 264). 'Change without choice' describes external effects, such as
 Structures, Environments, and Complex Systems 101

demographic changes prompting new demands for services, and almost automatic changes
in budgets (1990: 285). This argument supplements Rose's (1984) suggestion that a new
party in government may not make a major difference. They inherit legislation and a lot
of draft legislationthat tends to progress regardless of the party in government (Rose and
Davies, 1994: 133).
This theme of inertia is reinforced by 'policy succession', as 'the replacement of an exist-
ing policy, program or organization by another' (Hagwood and Peters, 1983: 1). While it may
look new, and is different from 'policy maintenance', it is 'directed at the same programme
and/or clientele' (1983: 18). Succession is likely for three main reasons. First, the size and
scope of the state is large. Second, existingpolicy is often 'its own cause' (Wildavsky, 1980),
since the delivery of policy often throws up problems that command the time of policymak-
ers (compare with policy feedback theory -
Mettler and SoRelle, 2018; Béland and Schlager,
2019). Third, the level of existingcommitments is high and there is little scope to increase
tax income (throughgrowth or higher taxes) to fund new programmes (Hagwood and Peters,
1983: 2-5). Succession is generally more likely than innovation because the conditions for
the introduction of policy are already in place: the issue already has legitimacy, the resources
are in place, and policy has an established clientele (Chapter 2).
These arguments are not about no choice. For example, Rose (1990: 288; Rose and
Davies, 1994: 5) argues that inherited commitments can be moulded to suit current prob-
lems, some programmes can be terminated or introduced, and the policy innovations made
during the administration may have a long-term,cumulative effect. Further, many of the
policy successions described by Hagwood and Peters (1983: 30), such as the introduction
of the National Health Service in the UK, were significant (in fact, they describe this as
a hybrid of succession and innovation). Similarly, the significance of a decision to reform
welfare policies by spending the same amount but changing the balance between recipi-
ents may not be captured well by the term 'consolidation' (1983: 66). Rather, it shows us
that any such reform may take a 'huge political effort' to produce 'marginalimprovements'
(1983: 129). It will not involve new money and will be carried out by the organizations
that policymakers inherited (compare with Box 5.2 on 'drift, layering, conversion, and dis-
placement', and Chapter 11 (pp.193-4) on first- and second-order change).

THE EVOLUTIONARY METAPHOR: CONTEXT AS A POLICYMAKING

ENVIRONMENT

Many theories draw on evolutionarymetaphors to describe the interaction between actors

and their environment. This approach can be useful in the abstract, to identify the key
dynamics that might constitute that environment, and prompt us to consider why anar-
row focus on elites will not tell us much about policymaking. For example, John (2003:

495) uses 'evolutionarytheory' to capture the dynamics of the 'five core causal processes',
described by Heikkila and Cairney (2018: 303-4) as follows:

l. Actors making choices. There are many policymakers and influencers spread across many
types of policymakingvenues (Chapter 8). If we only focus on a small group of elites, at
the 'centre' of government,we miss the spread of policymakingresponsibilityacross a
political system.
102 Understanding Public Policy

2. Institutions. Each venue contains its own formal and informal rules governing
behaviour (Chapter 5). Formal rules place limits on the roles of policymakers, and our
explanation of their behaviour is incompleteif we do not consider the often-unspoken
rules that they follow.
3. Networks. Each venue can produce its own networks of policymakers and influencers
(Chapter 9). Even in highly centralized systems, there is a strong logic to the
formation of policy 'communities'. Policymakers at the 'top' must ignore and delegate
responsibility for most of the issues for which they are responsible. Bureaucrats tend
to rely on specialists to help them process policy, and they often work closely together,
building up trust and shared ideas about policy problems. If so, the lines between
formal responsibility and informal influence are blurry (Chapter 8).
4. Ideas. Participantsin each venue draw on a dominant set of beliefs about the nature of
policy problems and the acceptable range of solutions (Chapter 11). Classic examples
range from a broad adherence to capitalistprinciples in many countries, to the sense
in some countries that some ideas seem off limits (e.g., when UK policymakers take for
granted,but US policymakers reject, a 'National Health Service').
5. Social and economic factors, and events, help set the policy agenda and limit
policymakers' abilities to address and solve policy problems. Events can be routine
(such as elections) or non-routine (such as social or natural crises).

Combined,these factors produce the broad sense that environmental influences on poli-
cymaking often seem to be overwhelmingor out of the control of individual policymakers.
Their environments constrain or facilitate individual action.
Since this process relates to an evolutionarymetaphor, it is worth being clear on how
evolutionary language translates into policymaking dynamics. For example, Lewis and
Steinmo (2008: 15-16) suggest that 'all living things including humans want to pass on
- -

their genes' and they 'inherit powerful instincts to follow social rules' to help them do so. In
other words, they combine their desires to pursue self-interest and cooperate with others to
protect themselves and their kith and kin (Cairney, 2013b: 289). In politics, the equivalent of
passing on genes is to ensure the survival of memes (Dawkins, 1976; John, 2003: 493; Kerr,
2002), to help communicate and protect our beliefs across gener-
Meme -
the ideational ations. The process involves three key evolutionarymechanisms:
equivalent of a gene:
'an information pat- 'Variation' refers to the different rules adopted by different
tern, held in an indi- social groups to foster the collective action required to sur-
vidual's memory, which
vive. 'Selection' describes the interaction between people and
is capable of being
their environments; particular environments may provide
copied to another'
(John, 2003: 493). It
an advantage to some groups over others and encourage
is used by Dawkins certain behaviours (or, at least, some groups may respond by
(1976) to explain why adapting their behaviour to their environment). 'Retention'
types of information-
describes the ways in which people pass on their genes
common sense, cultural (memes) to ensure the reproduction of their established
references, phrases, rules. (Cairney, 2013b: 289)
technology, traditions,
theories -
'survive' and In other words, we are trying to identify the selection mecha-
are reproduced across
nism that causes some ideas to be successful while most fail.
generations.
Evolutionarytheory requires the identification of competing
 Structures, Environments, and Complex Systems 103

memes; selection and deselection processes; changes to policy memes by political actors
seeking to adapt their aims to current circumstances; and the 'random mutation of ideas
and their fortuitous success in particular environments' (John, 1999: 47). In other words,
there aremany possible ways to understand the world and produce many different rules on
cooperation. However, policymaking environments advantage some groups and rules over
others, and the maintenance of rules over generations is imperfect. As Chapter 5 describes,
it is unclear how people produce, maintain, challenge, or pass down rules across generations.
The use of evolutionarylanguage in policy scholarship is also imperfect. Its meaning is
contested strongly (Dowding, 2000), and too many scholars use the language too loosely,
treating evolution as a vague synonym for change, or describing very different processes,
including:

l. Gradual change in policy or institutions, improvementover a long period.

or
2. 'Punctuated equilibrium', or a major disruption in policy or the institutions
used to
process policy.
3. The 'coming together of multiple factors to create the conditions for major policy

change', such as a 'perfect storm' or 'window of opportunity'.

4. The 'trial-and-error strategies adopted by actors, such as policy entrepreneurs, when
adaptingto their environment' (2013b: 279).

This lack of clarity will not go away, partly because it reflects wider debates on the meaning
and mechanisms of evolution in nature (Cairney, 2012a: 266-69). Therefore,the onus is on us
to think critically about the meaning of evolution in each case. For example, multiple streams
analysis (MSA) (Chapter 11) describes the slow and gradual improvement of policy solutions
towards political feasibility. In contrast, punctuated equilibrium theory (PET) (Chapter 9)
describes long periods of political stability and policy continuity followed (sometimes) by
instability and rapid and profound change. While MSA describes the need to 'soften' solutions
to make them acceptable to their audience in one venue, PET describes the potential for venue
shopping and shifting responsibility to a more sympathetic audience (Cairney, 2013b: 282).
On the crucial role of 'policy entrepreneurs', Kingdon (1984: 173) describes 'surfers waiting
for the big wave' rather than controllers of the sea. In other words, entrepreneurspossess the
skills to exploit decisions made by other actors, not to control them directly (Lustick, 2011:
204). Drawing on rational choice theory (Chapter 7), John (1998: 184) suggests that their
strategies change as actors learn to cooperate with each other and adapt to their environments.
They learn, and their preferences change, when they formulate and adapt ideas. This process
does not end when policies are selected at the formulation stage. Rather, policy changes fur-
ther as actors more powerfulat the point of delivery- influence the progress of policy. John
-

(1998: 186) describes the process as 'structured evolution' since the institutions, networks,
and socioeconomic constraints that affect behaviourare more static than strategies and ideas.

Box 6.3 The policy environment and 'evidence-based policymaking'

This focus on helps us understand why the prospect of 'evi-

policy environments
dence-based policymaking' (Box 2.2, p.19) is so unrealistic. Chapter 4 already ques-
tions its possibility by describing why individual policymakers must disregard most
104 Understanding Public Policy

evidence. Further, each of the five environmental factors provides a sense of the
immense scale of the task. Consider the advice that someone would give you if you
sought the greater use of your evidence in policy (Cairney and Oliver, 2019; Oliver
and Cairney, 2019). First, if there are so many potential authoritative venues, you
will need to devote considerable energy to finding where the 'action' is. Second,
even if you find the right venue, you will not know the unwritten rules unless you
study them intensely. Third, some networks are close-knit and difficult to access
because bureaucracies have operating procedures that favour some sources of evi-
dence. You could be a privileged insider in one and be excluded in others. Fourth,
if your evidence challenges an existing paradigm, you need a persuasion strategy
good enough to prompt a shift of attention to a policy problem and a willingness
to understand that problem in a new way. Finally, you can try to find the right time
to use evidence to exploit a crisis leading to major policy change, but the spark
for these opportunities is out of your control. Then consider how you could oper-
ate effectively in a complex system, in which the same action can be amplified or
ignored, and outcomes seem to emerge locally, often with no central direction. In
that context, policy studies recommend being entrepreneurialand investing your
time to build up alliances, and trust in the messenger, gain knowledgeof the sys-
tem, and spot 'windows of opportunity' (Stoker, 2010; Weible et al., 2012; Cairney,
2016a: 124; Cairney and Oliver, 2017). However, they offer no assurances that this
investment will pay off.

THE POLICY PROCESS AS A COMPLEX SYSTEM

Scholars use complexity theory to identify complex systems

Reductionist -

across the natural and social sciences. Broadly speaking, the

explaining complex
approach is based on a rejection of reductionist approaches phenomena in relation
to science, and there is a strong emphasis on the absence of to their separate
linear or orderly processes. It seeks to explain why complex or components.
system-wide behaviour emerges from the interaction between
'large collections of simpler components' (Blackman, 2001; Kernick, 2006; Mitchell, 2009:
x; Geyer and Rihani, 2010). Cairney (2012b: 348) identifies five common assumptions
regardingcomplexsystems:

l. The whole greater than the sum of its parts. A complex system cannot be explained
is

merely by breaking it down into its component parts. Instead, we must shift our
analysis to the system as a whole, as networks of elements that interact, share
information, adapt, and combine to produce systemic behaviour.
2. Non-linear dynamics. Complex systems provide 'feedback' to actions: dampening some
(negativefeedback) and amplifyingothers (positive feedback). As a result, small actions
have large effects and large actions can have small effects.
can
3. 'Strange attractors'. Although unstable, systems often display regularities of behaviour.
'Punctuated equilibrium' sums up the potential for long periods of stability disrupted
by short bursts of change.
 Structures, Environments, and Complex Systems 105

4. Sensitivity initial conditions. Initial events or decisions often produce a long-term

to
momentum. The 'butterfly effect' captures the idea of small factors producing large
consequences and a profound cumulative impact.
5. 'Emergence' in the absence of central control. Behaviour and outcomes result from the
interaction between elements at a local level rather than central direction.

For example, swarmingbehaviour in bees cannot be explained merely by the actions of

individual insects. Rather, we must study their interaction, the rules they follow, how they
communicate those rules, and the extent to which a minor change in rules causes a large
change in outcomes. The brain is also a complex system in which emergent processes, such
as thoughts and feelings, are difficult to break down into the performance of individual
cells and neurons. The brain also makes it difficult to treat political and social systems in
the same way as natural systems. The human 'capacity to reflect and to make deliberative
choices and decisions among alternative paths of action' makes the social world a different
object of study (Mitleton-Kelly, 2003: 25-26). For example, social science complexitystud-
ies are less deterministic than physical models tracing outcomes back to a specific cause
like the 'butterfly effect' (Box 6.4).
However, complexity theory's application to public policy is not obvious, and
'complexity' like 'evolution'
- -
is often used very loosely as a metaphor or to mean com-

plicated (2003: 26; Kernick, 2006: 389; Bovaird, 2008: 321). To make sense of 'complexity',
we need to describe the key dynamics of a complex policymaking system.
The whole is greater than its parts. Individual policy actors are one small part of a much
larger system over which they have limited control. Jervis (1998: 5-6) suggests that: 'We
are dealing with a system when (a) a set of units or elements is interconnected so that

changes in some elements or their relations produce changes in other parts of the system,
and (b) the entire system exhibits properties and behaviours that are different from those
of the parts.' The metaphor of a telescope is useful to describe 'zooming out' from indi-
viduals to consider the system as a whole (Cairney and Geyer, 2015: 2).
Non-linear dynamics and strange attractors. PET uses this language to explainpolicymak-
ing stability and instability (Chapter 9). The 'general punctuationhypothesis' describes the
importanceof 'feedback' when policymakers process information. Jones and Baumgartner
(2005: 7) define information processing as the 'collecting, assembling, interpreting and
prioritizing [of] signals from the environment'. Policymakers are surrounded by an almost
infinite number of 'signals' -
or information that could be relevant to their decisions
-

from their environment. Since policymakers are boundedly rational (Chapter 4), they do
not have the ability to process all signals. Instead, they must simplify their environment
by ignoring most information and issues (negative feedback) and promoting a few issues
to the top of their agenda (positivefeedback) (Baumgartner, 2017). Negative feedback may
produce prolongedperiods of equilibrium;existingpolicy relationshipsand responsibilities
are more likely to remain stable, and policy is less likely to change, when the issue receives
minimal attention from policymakers. Positive feedback may produce policy 'punctuations'
because, when policymakers pay a disproportionateamount of attention to an issue, it is
more likely that policy will change dramatically.
In a complex system with so many actors, non-linear dynamics may be found in many
places, such as when actors in different venues pay maximal or minimal attention to an issue,
or implementers pay unusually high attention to some directives and ignore others. Yet, sys-
tems also display regularities of behaviour over extended periods (Bovaird, 2008: 320). We
106 Understanding Public Policy

might call suchregularities 'strange attractors' and perhaps relate their cause to regularities
in ways of thinking summed up by the idea of organizational or political 'culture' or the
- -

relatively stable role of powerful interests (Givel, 2015: 71-74). However, this term seems
relativelyvague and less used in policy studies (Tenbensel, 2015: 372).
Sensitivity to initial conditions resembles the 'path dependence' described by historical
institutionalism (Chapter 5). A 'critical juncture' is the point at which certain events and
decisions led to the developmentof an institution. The timing of these decisions is cru-
cial, because it may be the order of events that sets institutional developmenton its path.
When a commitment to a policy has been established and resources devoted to it, over
time it produces 'increasing returns' and it becomes increasingly costly to choose a differ-
ent path (Pierson, 2000a). The link between complexitytheory and historical institution-
alism is strong: both identify the same sense of inertia and unpredictability,as relatively
small events or actions can have a huge and enduring effect on policy change, and both use
the black-and-red ball analogy and examples such as the QWERTY keyboard, outlined in
page 82 (Pierson, 2000a: 253; Room, 2011: 16; 2016). Complexitytheory may add a new
dimension to the study of institutions, exploringthe potential for unexpected instability
and dramatic policy change even when the rules seem to change in a minor way (like a sud-
den shift of direction of a swarm of bees, or an attention shift in the brain).
Emergence requires the most careful translation since there is no direct political equivalent
to the completeabsence of central control. For example, in biology, we may identify molecu-
lar interaction with no equivalent of a 'centre'. In politics, there is at least one centre -

central government and so we need to describe (1) the emergence of policy despite a centre,
-

or despite attempts by central governments to control outcomes, and (2) the role of many

possible 'centres' or authoritative sources of policymaking (Cairney and Geyer, 2015: 3). For
example, complexitytheory is one of three approaches exploringthe idea of 'multi-centric'
policymaking (Caimey et al., 2019a), alongside the Institutional Analysis and Development
(IAD) framework (Chapter 7) and multi-level governance (MLG; Chapter 8).
Further, emergence in the absence of central governmentcontrol is a key feature of
the implementation and governance literatures. Central governments face problems
when they do not recognize the extent to which policy will change as it is implemented
(Chapter2). The level of interdependencebetween governments and implementingorgani-
zations has prompted the identification of 'self-organizing networks' (Rhodes, 1997: 50)
and images of 'bottom-up' implementationthrough self-selecting clusters of organizations
in which a variety of public and private organizations cooperate.
Such arrangementshave often promptedgovernmentsto embrace 'new public manage-
ment', or the application of private business management methods to the public sector,
to try to impose order through hierarchy and targets (Geyer, 2012; Chapters 8 and 12).
However, implementation structures may not be amenable to such direct control, even
if the organizations,responsible for delivery, are completelyon board with central policy
(Teisman and Klijn, 2008: 294; Bovaird, 2008: 339). Constant reforms of service delivery
functions may be futile unless we recognize the nonlinear nature of policymaking. This
recommendation is one of many possible practical lessons for policymakers, including
(Teisman and Klijn, 2008: 288):

l. A policy that was successful in one context may not have the same effect in another. General
behaviour is difficult to identify because the policy process is 'guided by a variety of
forces'. X will only influence Y under conditions that are difficult to specify. We need to
 Structures, Environments, and Complex Systems 107

know why it was successful inthat instance, but there is so much interaction between
so many variables that we cannot account for them all (or their separate effects).
2. Complex systems are difficult to control. They have 'self-organizing capacities', and the
effect of the same central government intervention may be large or small. Policymakers
should not be surprised if their interventions do not have the desired effect.
3. Find ways to adapt to a changing landscape. The metaphor of the 'fitness landscape' -

or 'surroundingsin which living beings exist and behave' -

provides the context for
the choices of agents. It is unstable and often rapidly changing. Therefore, actors
must adapt quickly and not rely on a single policy strategy (2008: 289; Mitleton-Kelly,
2003: 35-36).

Complexity scholars tend to recommend that central government policymakers let go

of the idea of control, to experimentmore, and give more local actors the discretion and
flexibility to act. Sanderson (2009: 706) suggests that we do not know exactly how any
policy measure will make difference. Therefore, policymakers should be careful when
a

making interventions, makinggreater use of 'trial and error' and learning from pilot pro-
jects (2009: 707), or giving implementingorganizations the chance to learn from their
experience and adapt to their environment (Bardach, 2006: 353; Haynes, 2008: 326). Such
advice takes us back to the discussion in Chapter 4 on Lindblom (1959: 86), who rejects
the idea of top-downradical change because policymaking is 'a very rough process' and no
one knows 'enough about the social world to avoid repeated error in predicting the conse-
quences of policy moves'. Therefore, a 'wise policymaker' anticipates error and unintended
consequences and avoids too-radical moves. See also Dunlopand Radaelli (2015: 140), who
propose 'regulatoryhumility' to avoid accepting an 'illusion of control'.
Yet it is difficult to expect policymakers to accept such advice. For example, in
Westminster-stylesystems, national elections revolve around the idea that we can hold
elected politicians to account because we know they are in charge of policymaking.In that
context, they need to link their image of governing competence to the idea that they are in
control (Cairney, 2015a). Other forms of accountability- built on the idea of multi-centric
policymaking, 'co-produced' policy, and shared accountability may exist, but they remain
-

difficult to defend in the world of highly competitive and adversarial politics (Cairney
et al., 2019a).

Box 6.4 Are complex systems theories deterministic?

A common complaint regarding systems theories is that they describe systemic log-
ics that 'operate in some sense independentof and over the heads of the actors
- -

themselves' (Hay, 2002: 102). The danger is that, if the complex system is pre-
dominantly the causal factor, then we lose sight of the role that policymakers play;
there may be tendency to treat the system as a rule-bound structure, which leaves
a
minimal room for the role of agency. What we need is an understandingof how
agents perceive their decision-making environments; how they reproduce, accept, or
challenge the structural,institutional, and wider systemic constraints that they face
when making decisions. Policy studies seek to explain why different policymakers
make different decisions under the same circumstances.
108 Understanding Public Policy

If we being sympatheticto complexity theory, we can treat its approach to

are
structure and agency like Gidden's 'two sides of the same coin If we look at social
...

practices in one way, we can see actors and actions; if we look at them another way
we can see structures' (Craib, 1992: 3 on Giddens, 1984). On one side, we have 'self-

organizing landscapes' (Teisman and Klijn, 2008: 289) or complex systems that
adapt and change behaviour; behaviour is 'emergent' from the processes within
systems and is not readily broken down to the agents within it. Further, much of
the explanation for outcomes comes not from individuals but from the level of con-
nectivity between them (Mitleton-Kelly, 2003: 28). On the other, we have the 'self-
referential behaviour' of agents, reacting to 'external forces and changes' but also
'creating their own perception of what they want and how to behave in the landscape
they are in' (Teisman and Klijn, 2008: 289). Some actors are more powerful than
others because they are more able to adapt to the 'fitness landscape' (Room, 2016).
If we are less sympathetic, each side appears to contradict the other: a focus on sepa-
rable, independent actions by agents appears to contradict the idea that a complex
system cannot be broken down into its component parts (Cairney, 2012b). Perhaps
this is the meaning of 'two sides of the same coin': two argumentsthat represent
each other's opposite. This problem of conceptualizing structure and agency is not
unique to systems theory. Indeed, a common theme throughout the book is how to
attribute explanatorypower to entities that do not act but appear to influence the
actions of agents.

CONCLUSION

The strongest structural accounts suggest that powerful external forces constrain the abil-
ity of individuals or governments to make decisions, or that socioeconomic change deter-
mines policy change. Yet few contemporary theories take this approach. Rather, we can
identify a range of approaches that try to strike the right balance between structure and
agency. The DSH approach measures the association between socioeconomic factors and
policy outputs. Marxist accounts explore the strong imperative for governmentsto sup-
port the capitalist system and therefore the interests of the classes that benefit most from
that system. Globalization suggests that governmentscompete to protect their economy
and secure foreign direct investment. Inheritance before choice suggests that govern-
ments inherit massive policy commitments and tend to change policy only at the margins,
while the pervasiveness of policy succession reflects the constraints on policy innovation
and termination. Complexitytheory suggests that policymakers are part of a large com-
plex system that seems to defy control.
A focus on structural factors is really about the wider context in which policymaking
takes place. It allows us to 'zoom out' from a focus on individual policymakers and their
centrality to policymaking,towards thinking of the policymakingenvironment or system
in which they operate. Broadly speaking, the environment is a metaphorto bring together
key concepts actors, institutions, networks, ideas, socioeconomic factors and events
- -

and then consider how specific theories use them to provide an overall explanationof
policymaking.
 Structures, Environments, and Complex Systems 109

The advantage of such approaches is that they highlight the context within which policy
is made. We know that policymakers make choices but recognize that policymakingdoes
not begin with a blank slate or operate in a vacuum. Structural factors may influence what
they pay attention to and how they act. There may be events and policy conditions out-
side of their control. There are limits on their ability to anticipate the effects of policy and
the extent to which it is carried out successfully. They inherit a 'ship of state' that behaves
more like a super-tanker(notoriously difficult to turn around) than a rowboat. In short,
this discussion is a useful corrective to the idea that we could ever explain policymaking
well by focusing on a small group of policymakers at the centre.
The disadvantage is that such accounts often appear to favour structural explanations
at the expense of a focus on agency. Sorne Marxist accounts highlight inevitability rather
than choice. Sorne theories of globalizationdo not explainfully how policymakers inter-
pret and react to their environment. Sorne accounts overemphasize the vague notion of
evolution at the expense of careful analysis, or suggest that 'self-organizing' complex sys-
tems almost have a life of their own. It is difficult to present a convincing account of poli-
cymaking purely in terms of context and the constraints that policymakers face. We must
also explain how and why they act. In sorne cases, policymakers have many choices even
when the decision-making context appears to provide only one realistic option. We may
often feel that policymakers have no choice, but a more accurate description is that 'the
best course of action obvious' (Dowding, 1996: 44). While it may seem like a fine
seems

distinction, it helps separate social science contingency from the deterministic physical
and natural sciences.
Most modern theories try to conceptualize this dynamic process in which structures
influence agents and agents mediate or reconstitute the structural constraints within
which they operate. They identify how socioeconomic factors constrain behaviour and poli-
cymakers mediate these factors by interpreting or weightingtheir significance in different
ways. PET shows how policymakers process information in complex systems (Chapter 9).
MSA shows how policy entrepreneurs navigate policymakingenvironments (Chapter 11).
The advocacy coalition framework seeks to incorporate the DSH approach into a wider
analysis. Its flow diagram (Figure 10.1) includes a discussion of 'relatively stable param-
eters', such as the 'basic distribution of natural resources', and 'external events' such as
'changes in socioeconomic conditions', but it also seeks to explainthe 'black box' by identi-
fying the ability of advocacy coalitions to interpret external effects and compete to define
the policy problemwithin subsystems.