Understanding Compliance with Laws and Regulations: A Mechanism-Based Approach

Abstract

This introductory chapter aims to provide an in-depth theoretical account of the ways in which legal and regulatory compliance can occur. Following the recommendations of analytical sociologists, a mechanism-based approach is adopted, in order to explain how and why compliance occurs in various situations. Four principal mechanisms of legal compliance are identified: they are instrumental (or prudential) compliance; normative compliance; situational compliance; and compliance based on habit or routine. Within the first three of these mechanisms, important sub-types are also identified: for example, within normative compliance are the sub-types of compliance through legitimacy, normative belief, normative attachment and normative cues. Interactive effects between the principal mechanisms are also discussed. A brief comparison is made between this explanatory approach and Nielsen and Parker’s Holistic Compliance Model of regulatory compliance.

1 Introduction

The purpose of this chapter is to lay some conceptual groundwork for later contributions in the book, by exploring and categorizing the reasons why people comply with laws and regulations. The focus of the discussion will, therefore, be upon what the sociologist Frank Parkin (1982: 79) once memorably described as compliance ‘as viewed from below’: that is to say, a ground-level look at how compliance occurs. By contrast, later chapters are—entirely properly—mostly concerned with how those who have a responsibility to enforce laws and regulations can achieve their ends with optimum efficiency and effectiveness—an approach that can reasonably be described as compliance ‘as viewed from above’. Despite their differences, however, these two types of analysis are in principle interconnected, because an enhanced understanding of ‘compliance from below’ (how and why compliance occurs) should enable enforcement agents to develop more sophisticated and effective compliance strategies.

In presenting this analysis in the context of a book whose main substantive focus is regulation in the field of business, I am very conscious that my own research career has concentrated on (in the clichéd phrase) ‘crime in the streets’ rather than ‘crime in the suites’. It follows that my examples are principally derived from my main field of study. However, some business examples are included, and some attention will also be paid to a leading strand of theorization about regulatory compliance, the ‘Nielsen–Parker holistic compliance model’ (Nielsen and Parker 2012; see also Parker and Nielsen 2011, 2017).

2 A Mechanism-Based Approach

The conceptual approach adopted in this chapter is based on the view that, when explaining social phenomena, we need to pay special attention to social mechanisms. In contemporary social science, an early advocate of this view was the social philosopher Jon Elster. He argued that ‘we will never have any general theory of collective action’, because the variety of potentially interacting motivations is ‘simply too large’ to be encompassed in such a theory (Elster 1989b, p. 205). That did not mean, however, that social scientific analysis is impossible. Rather, Elster concluded that social analysts should instead focus on ‘small and medium-sized mechanisms that apply across a wide range of social situations’ (p. 205), or, alternatively stated, ‘plausible, frequently observed ways in which things happen’ (p. viii). In focusing on mechanisms, he argued, we are also focusing on explanation: ‘To explain an event is to give an account of why it happened. Usually ….this takes the form of citing an earlier event as the cause of the event we want to explain…..[But this] is not enough: the causal mechanism must also be provided, or at least suggested’ (Elster 1989a, pp. 3–4).

Advocacy of the importance of social mechanisms in the explanation of social phenomena has subsequently been taken forward by a number of writers, and it has now burgeoned into a sub-field known as ‘analytical sociology’, with its own internal debates about key topics such as theories of action and causality (see, e.g., the collections of essays in Hedstrӧm and Bearman 2009a; Demeulenaere 2011a). For present purposes, it is not necessary to delve into these complexities, but it is important to outline some basic points about mechanisms as explanations. In doing so, I shall rely in particular on some writings by the Swedish sociologist Peter Hedstrӧm (2005; see also Hedstrӧm and Bearman 2009b), who is widely seen as having been particularly ‘responsible for the [systematic] theorization of [the analytical sociology] approach’ (Demeulenaere 2011b, p. 24).

At the beginning of his monograph on analytical sociology, Hedstrӧm (2005, p. 1) emphasizes the desirability of developing ‘precise, abstract, realistic and action-based explanations for various social phenomena’. This goal is of obvious relevance to this chapter, which tries to answer the explanatory question ‘why do people obey laws and regulations?’ In pursuit of his stated objective, Hedstrӧm (2005, Chapter 1) advocates a number of key features of sociological explanations, some of which I shall paraphrase here.

Firstly, then, explanations must be truly explanatory and not simply descriptive—that is, they must address the question why things happen. It is a commonplace of social science education that ‘a correlation is not a cause’, yet it is not always recognized that even advanced statistics are often simply correlational. Analytical sociology emphasizes that descriptions and correlations, although certainly valuable, are not enough; instead, in any given social situation it is crucial to ask the ‘why’ questions. Analytical sociology further argues that the best way of answering such questions is ‘by detailing mechanisms through which social facts are brought about’, also that these ‘mechanisms invariably refer to individuals’ actions and the relations that link actors to one another’ (Hedstrӧm and Bearman 2009b, p. 4, emphasis added).¹ In the present context, the ‘why’ questions of interest are of course centred upon why, in a given social context, the level of compliance with a law or regulation is as it is (whether this be high or low, expected or unexpected).

Secondly, given the focus in analytical sociology on the actions of individuals (see the italicized phrase above), it is important also to emphasize that ‘sociology, as a discipline, is not concerned with explaining the actions of single individuals. [Hence] the focus on [individuals’] actions, is merely an intermediate step in an explanatory strategy that seeks to understand change at a social level’ (Hedstrӧm 2005, p. 5). Put another way, the overall strategy is to ‘explain why, acting as they do, [individuals] bring about [specified] social outcomes’ Accordingly, in our context, faced with a given level of compliance with a law, we need to explain how the actions of individuals are, in aggregate, producing that level of compliance.²

Thirdly, since analytical sociology is not ultimately concerned with the actions of single individuals, it must to an extent rely on generalisations, and its analyses must therefore contain a degree of abstraction. However, analytical sociologists insist that in developing explanatory theories, researchers ‘must refer to the actual mechanisms at work’ and resist the temptation to build models of mechanisms ‘that could have been at work in a fictional world invented by the theorist’ (Hedstrӧm 2005, p. 3).³ This realism is an important characteristic of analytical sociology. It is certainly also a helpful characteristic when one is seeking (as this chapter does) to analyse ‘compliance as viewed from below’ in a way that might be useful to people—such as compliance officers or police officers—who are facing real-life challenges in delivering effective and just compliance.

Finally, it is noteworthy that Hedstrӧm’s (2005) book is called Dissecting the Social—a title which he chose in order to emphasize that analytical sociology aims to ‘gain understanding by dissecting the social phenomena to be explained’ (p. 2). More specifically, in Hedstrӧm’s theorization, the term ‘to dissect’ means: ‘to decompose a complex totality into its constituent elements and activities, and then to bring into focus what is believed to be its most essential elements’ (p. 2). Taking this point together with the earlier ones, when analysing a given social situation a researcher should not only consider in detail the mechanisms in play among the people involved, she/he should also address the social relations in operation in that specific social situation (including the interactions of actors with differing mechanisms), in order to build an overall explanation of the social outcome.

In the present context, the implication of this approach is that we need to construct a typology of the principal mechanisms that, in the real world, sometimes cause legal compliance—or in other words, what Elster (1989b, p. viii) called the ‘plausible, frequently observed ways in which things happen’. In any specific situation where compliance is an issue, we will further need to consider how these mechanisms operate, given the social relations in play in that specific context.

One further issue must be addressed before we move on. The late Martin Hollis (2002), in his wonderfully clear textbook on the philosophy of social science, paid special attention to the existence of, and the potential tensions between, two major traditions in social scientific theorization, which he called, respectively, ‘explanation’ and ‘understanding’. The ‘understanding’ tradition is, unlike the explanatory tradition, primarily interpretative, and in its pure form, it proposes ‘that the social world must be understood from within, rather than explained from without…Instead of seeking the causes of behaviour, we are to seek the meaning of action’ (Hollis 2002, pp. 16–17). Hedstrӧm (2005), whose focus is firmly on explanation, shows only limited interest in the interpretative tradition,⁴ but in my view, the best social science takes full account of both traditions and seeks to develop them in creative synthesis (Bottoms 2008). From this perspective, it is encouraging that within the field of regulatory compliance, there is a consensus that research in both these traditions has made important contributions to the field (Parker and Nielsen 2011, pp. 3–8).⁵ While the present chapter focuses especially on mechanisms as a crucial tool with which to develop explanations of compliance, it endeavours to take full account of interpretative research that might help us to understand what ‘compliance’ means in specific situations—and to recognize that it might mean different things in different situations.

To illustrate the potential of a mechanism-based approach to explanation that also pays attention to interpretative issues, I shall first discuss an example of compliance processes in action, derived from a research project in which I was involved. After that, I shall turn to a full discussion of the main mechanisms of compliance.

2.1 Explaining Lower-Than-Expected Compliance: A Study in English Prisons

In the late 1990s, the minister responsible for English and Welsh criminal justice policy in the then UK government⁶ decided to introduce a new policy for prisons in this jurisdiction known as the ‘Incentives and Earned Privileges’ policy (or ‘IEP’). The intention of the new policy, shortly stated, was to improve prisoners’ behaviour by linking prison ‘privileges’ (i.e. certain non-standard benefits available to prisoners) more closely to prisoners’ good behaviour than had previously been the case. In other words, privileges would not be granted easily, but would have to be earned, and, as the name of the policy implies, the hope was that the ‘carrot’ of valued privileges (such as extra family visits) would act as incentives to improved behaviour.

The theoretical model underpinning this policy is, of course, one of rational choice. Such a policy strategy, to be effective, requires that the incentives really are incentives—that is, that the benefits available through the policy are truly valued by the people to whom the policy is applied. In the present case, this condition was met: the privileges within the IEP scheme were indeed seen as valuable by the great majority of prisoners. The government and its advisers therefore confidently expected that prisoners’ behaviour would improve when the IEP scheme was implemented.

The government then commissioned a research project, led by my Cambridge colleague Alison Liebling, to evaluate the effects of IEP during the first year of its operation. Unexpectedly, the principal finding of the research study was that, taking together the five prisons studied, there was no overall change in prisoners’ behaviour during that first year (Liebling et al. 1999; Bottoms 2003).⁷ Perhaps understandably, this result was met with some resistance by our research sponsors, who regarded it as seriously counter-intuitive.⁸

We were, however, able to demonstrate that the result was fully explicable. During the research project, we had been able to develop (from questionnaire responses at the beginning and the end of the year) a series of scales measuring prisoners’ perceptions of their particular prison; the main such variables were ‘relations with staff’, ‘perception of staff fairness’, ‘perception of regime fairness’ and ‘possibility to make progress’.⁹ Aggregating the results from the five prisons, the scores for each of these four measures were found to have declined significantly during the first year of operation of IEP; moreover, for three of the variables (staff fairness, regime fairness and progress) there was evidence of decline in all five prisons. Other data showed that most prisoners regarded the principles of the IEP scheme as fair, but they thought that the way in which the policy was being implemented was unfair. A main reason for this perception was that (for reasons too complex to explain here) the new policy gave basic-grade staff more power, and this extra power, prisoners thought, was often being applied arbitrarily.

The concept of ‘fairness’ has loomed large in the preceding paragraph, and fairness is of course a normative principle, closely linked to that of ‘justice’.¹⁰ Also, there is a great deal of evidence that people will, generally speaking, behave more co-operatively with authorities when they perceive their treatment by authorities to be fair rather than unfair (see Chapter 2 by Justice Tankebe in this volume). So the explanation of the initially puzzling results of the IEP research was that they embodied a clash of compliance mechanisms. The creators of the IEP policy, adopting an instrumental, rational choice theory of compliance, had assumed that valued privileges would lead to improved behaviour, and it would seem that this could have occurred, because the recipients of the policy (the prisoners) were not hostile to the principles of IEP, and they regarded the privileges on offer within the scheme as valuable incentives.¹¹ However, this result was not achieved, because the way in which the policy had been implemented was seen to be contrary in significant ways to normative principles of fairness. This led to a greater reluctance to comply with the incentives-based scheme; in essence, many prisoners took the view that, given the way the policy was being implemented, they were not going to ‘jump through the hoops’ that the scheme required. Thus, two kinds of compliance mechanism, incentives and normativity, were in conflict.

2.2 Purposes of the Analysis of Compliance

The IEP study is, of course, a very specific example, but it can readily be seen to have wider implications, not least because we have already identified two broad types of compliance mechanisms (instrumental and normative). From this starting point, it is necessary to take the analysis further in two main ways. Firstly, we need to identify other principal mechanisms of compliance, beyond the instrumental and the normative, as well as the sub-types of each principal mechanism. Secondly, we will need to identify, at least in outline, possible ways in which these mechanisms might interact (as the instrumental and normative mechanisms did in the prisons example).

I have tackled these issues in previous work (Bottoms 2001, 2002), but I now believe that some modifications are necessary to the analyses presented in those earlier papers.¹² I have not, however, altered my view that the best conceptual framework for understanding ‘compliance from below’ is to work with four principal mechanisms. These are now described as (i) instrumental/prudential compliance; (ii) normative compliance; (iii) situational compliance; and (iv) compliance based on habit or routine. These main mechanisms, with their sub-types, are set out in Fig. 1, which it is hoped will act as a useful road map for the discussion in the remainder of this chapter.

Fig. 1

(Source Author)

A mechanism-based typology of legal compliance

Since the principal purpose of the chapter is, in Hedstrӧm’s (2005, p. 2) helpful terminology, to ‘dissect’ the phenomenon of legal compliance, the main focus of the discussion of the various mechanisms will be upon their characteristics as causal mechanisms. Issues of effectiveness will occasionally be mentioned, but within the framework of a single chapter, it is not possible to provide a thorough discussion of the effectiveness literature in addition to a ‘dissective’ analysis.

A final introductory comment concerns the meaning of the term ‘compliance’—a matter which, as Decoene and Beyens (2013, pp. 216–221) have pointed out, has been the subject of debate among researchers in both medicine and social psychology. As regards the main topic of this chapter—compliance with laws and regulations—the approach taken is straightforward, defining such compliance in a behaviour-based manner as ‘acting in a way that is consistent with (or at least, not inconsistent with) the requirements of a given law, regulation or official programme (such as the IEP)’.¹³ In some parts of the discussion, however, the examples that best illustrate the conceptual issues being discussed concern broader forms of compliance, such as compliance with social norms that are not part of any regulation or programme. Where this is the case, I have not hesitated to include the example.

3 A Mechanism-Based Typology of Legal Compliance

3.1 Instrumental (or Prudential) Compliance

The first main mechanism to be considered is that of instrumental, or prudential, compliance. This is, normally, compliance based on self-interest: a person decides that it is in his or her own best interests, and therefore it is prudent, to comply with the regulation.¹⁴ The principal sub-types of this mechanism are, obviously, compliance as a response to incentives and compliance as a response to disincentives, and I shall deal with these before turning to the third and rather different sub-type, that of ‘creative compliance’.

3.1.1 Incentives and Disincentives as Mechanisms of Compliance

Incentives and disincentives as modes of compliance can be considered together, although, of course, in the field of legal regulation disincentives are more commonly deployed, and this has generated a large research literature on the subject of deterrence (for a recent summary, see Nagin 2013). Four issues concerning the characteristics of incentives and disincentives merit special attention.

Firstly, incentives and disincentives must always be considered from the perspective of the persons whom it is hoped to influence. It is all too easy for members of an enforcement agency to believe that they know what will influence those at whom the strategy is aimed, yet to get that prediction wrong. A well-documented case of this kind arose in youth crime policy in England and Wales in the early 1980s, when the then Conservative government announced with much publicity that it was introducing a more rigorous, or ‘tougher’, regime into detention centres for male young offenders, in order to act as a deterrent. (This was colloquially referred to as the ‘short, sharp shock regime’, and it was influenced by the concept of the military ‘glasshouse’.) At the Conservative Party conference, a senior minister said that these centres ‘will be no holiday camps … we will introduce on a regular basis drill, parades and inspections’ (Thornton et al. 1984, p. 1). However, evaluative research by prison psychologists showed that the new centres had similar reconviction rates to existing centres. Crucially, also, what would deter had been misperceived; for example, in one centre it was found that ‘the “new” activities (drill, extra physical education, etc.) reduced the amount of time spent on work …. [but since] work tended to be less popular than these new activities, [this] … involved the displacement of a relatively unpopular activity with a relatively popular one’ (Thornton et al. 1984, p. 203, emphasis added). This experience shows how important it is for those who introduce incentives or disincentives to ensure that they understand the situation from the point of view of potential regulatees.

Secondly, it is an inherent feature of enforcement strategies based on instrumental or prudential principles that, to be effective, the law or regulation must be constantly enforced. Unlike some other modes of compliance (see further below), an instrumental policy does not, in itself, aim to induce internalized compliance; instead, those subject to the policy are expected always to calculate the costs and benefits of compliance in particular situations. Accordingly, if a legal or regulatory agency significantly decreases a relevant element of its enforcement activity, and the recipients of the policy notice this, then one can in principle expect lower compliance rates after the diminution in enforcement activity. Empirically, this kind of decay after initial deterrent effects has been frequently (although not invariably) noted after police ‘crackdowns’ in small locations (Sherman 1990). It is therefore not surprising that a recent overview of research findings on random breath testing (RBT) for drink-driving in Australia concluded that RBT best practice must include, on a jurisdiction-wide basis, credible and consistent enforcement practices that receive appropriate publicity.¹⁵ The review adds that the application of these best practice principles is ‘critical’ for ‘RBT to be an effective drink-drive deterrent’ (Ferris et al. 2015, p. 80).

Thirdly, a number of matters need to be considered about how regulatees might respond to an instrumental strategy. An obvious point—but one that is easy to overlook—is that incentives and disincentives are only relevant to people who might contemplate engaging in the specified activity. Hence, increasing a lottery prize will not increase ticket sales among people who disapprove of lotteries on principle; similarly, an enhanced probability of detection for drug supply, or stiffer sentences for that activity, will make no difference to the behaviour of people who deliberately eschew all aspects of illegal drugs. Another reliable research result is that more immediate disincentives are more likely to have an effect on compliance than a disincentive that will come into operation only at some time in the future, and this is a principal reason for the now very well-established finding that a higher probability of detection is significantly more likely to have a deterrent effect than is the probability of receiving a more severe sanction if one is apprehended (Nagin 2013). Finally, it must be noted that there can be different reasons for a compliant response even among those who comply with a law or regulation for instrumental reasons. A good illustration of this is found in Valerie Braithwaite’s (2009, 2013) research into compliance with (or defiance of) tax laws, in which she described the two main ‘motivational postures’ of compliers as ‘commitment’ and ‘capitulation’. A similar distinction—described as a difference between ‘substantive’ and ‘formal’ compliance—has been drawn by Robinson and McNeill (2008) in the field of community penalties. An example of this difference, they suggest, would be that of two offenders on an unpaid work order, one of whom genuinely engages with the spirit of the penalty (he works hard at the assigned task, seeing this as reparation to the community for his offence), while the other unenthusiastically does the minimum necessary to count as fulfilling the terms of the order.

Fourthly and finally, it is a pervasive feature of the research on instrumental compliance that the degree of such compliance may be affected by the normative context in which the incentive or disincentive is offered. A result of this kind has already been seen in the IEP study, but a further example will illustrate a slightly different point, more directly related to incentives.

Klepper and Nagin (1989) administered a scenario-based survey on tax evasion to adult part-time masters-level students of management in the USA, nearly all of whom had experience of working as middle-level managers in either the private or public sectors. The results showed that many respondents were willing to consider under-reporting income to the tax authority when the sanctions would be financial penalties imposed confidentially by the authority. (As predicted by rational choice theory, such ‘gambles’ were much influenced by the degree of probability of detection.) However, the judgements of respondents changed significantly when the scenario included the possibility of criminal prosecution, with its attendant publicity: in this situation, ‘any non-zero chance of criminal prosecution [was] an absolute deterrent’ (p. 239). This result presumably arose because of respondents’ perceptions of the likely social consequences to them of a criminal conviction, including social stigma and shame. These wider normative consequences therefore constituted a powerful disincentive to cheating on one’s tax liability.

3.1.2 ‘Creative Compliance’

We now turn to the third sub-type of instrumental compliance, ‘creative compliance’¹⁶ (see Fig. 1). This sub-type is perhaps of special interest in the context of this volume, because creative compliance has been identified as occurring especially in relation to regulation in the field of business, notably in financial services. Indeed, it has been described by Doreen McBarnet (2013, p. 71) as ‘the dominant culture of compliance in business, and …among those “high net worth” individuals with the resources and power to manage law to suit their own purposes’ (emphasis added). McBarnet’s telling analysis of the character of creative compliance deserves extensive quotation:

Rather than simply accepting and meeting the substantive obligations imposed by law, those opting for creative compliance focus on ways of arguably complying with the technical drafting of the law, while simultaneously frustrating its purpose.¹⁷ Practices may well be adjusted, but they will be adjusted technically rather than substantively, so that the same practice continues substantively…repackaged into another legal form. (pp. 71–72)

She describes this kind of activity as ‘legal engineering’ and adds:

Legal engineers know that they are not following the intentions or spirit of the law. Bankers and banking lawyers talk in interview about their legal practices as ‘bullish’ or ‘sailing close to the wind’ ….. In the mindset that underlies and fosters legal engineering, all the responsibility for control is placed on the regulators. If they can’t make regulations ‘legal engineering-proof’, or spot the failings in the schemes, it is fair game to exploit that situation. Ideas such as responsibility, the public good, morality, ethics or integrity do not enter into the equation. (p. 86)

In short, then, creative compliance can be described as a response that is technically compliant, but with the intention of being as non-compliant as the relevant law allows. As such, it clearly merits identification as a separate sub-type under the heading of instrumental compliance. We should also note the severe challenge to regulators that is posed by this kind of activity. Those engaging in creative compliance frequently have available significant financial resources, with which they can employ the best lawyers, whose job is then to devise schemes, sometimes of formidable technical complexity, in order to defeat the avowed purpose of the law. If the regulators nevertheless manage to achieve an imposition of liability, there will be another attempt next year to devise a ‘better’ scheme and so on.¹⁸

Arguably, the practice of creative compliance can in some circumstances have wider social consequences. Doreen McBarnet (2013, pp. 72–74) has suggested that, contrary to received wisdom, the global banking crisis of 2008 was not caused only by ‘innovative financial products that were too readily believed to disperse risk’ (‘financial engineering’). A second cause (or mechanism) creating the crisis was, she claims, ‘legal engineering’—as described above—which saw lawyers ‘creatively removing the “obstacles” of prudential regulation, accounting requirements [etc.]…intended to control or disclose risk….Circumventing capital adequacy regulation was a crucial driver behind much structured finance’. This is, of course, an empirical claim about the events of 2008 which can be assessed only by those with the relevant knowledge (which does not include this author). The claim is, however, important in the present context because it is illustrative of an analytical sociology approach to explanation—that is, it aims to refine the explanation of the banking crisis by drawing attention to the actions of individuals (lawyers seeking ‘creative compliance’), which, it is claimed, was one of the mechanisms leading to the crisis.

3.2 Normative Compliance

The word ‘norm’ has several meanings in English, so what is its connotation in the present context? A helpful basic source is a (slightly truncated) dictionary definition, according to which a norm is ‘a principle or standard …. that reflects people’s expectations of behaviour … and serves to regulate action and judgment’ (Longman 1984). A key phrase in this definition is that of the ‘expectations of behaviour’ among a given social group, and we need to note that ‘expectations’ is a value-laden term.¹⁹

Norms as expectations of behaviour have a dual character: they enable and they constrain (Giddens 1984). If we are familiar with the norms of a given group or context, this allows us to navigate daily life more easily in accordance with the local customs and expectations: thus, norms can smooth, enable and encourage social action. (Conversely, of course, unfamiliarity with the norms of a different culture can create problems, as—e.g.—first-time Western visitors to Japan quickly discover.) But in addition, because norms regulate action and judgement, they act as a constraint, nudging people towards behaving in the way that the norm expects. Individuals will not necessarily follow these expectations, but even if they act in clear defiance of the norm, they will know what the social expectations are, and we can be certain that very often people will be influenced to act congruently with the conduct that the norm prescribes.

The cultural and moral norms of a society will often (but not necessarily) underpin its laws and regulations, a point sometimes described as the ‘mirror thesis’ (‘the law mirrors the norms’).²⁰ Where this is the case, we can expect to find that normative factors are of great importance in shaping legal compliance, especially since many norms will have been fully internalized by members of the society. I shall return to these more personal normative judgements shortly, but first I shall consider legitimacy—a sub-type of normative compliance that is always of great importance when one is considering compliance with formal laws and regulations.

3.2.1 Legitimacy and Compliance

Legitimacy, in its social-scientific sense, has been defined by a leading authority as ‘power which is acknowledged as rightful by relevant agents, who include power holders and their staff, those subject to the power and third parties whose support or recognition may help confirm it’ (Beetham 2013a, p. 19, emphasis in original).²¹ Such acknowledgement is, however, not necessarily either total or unchanging. This means that relations between power holders and those subject to power can be likened to a continuous dialogue (Bottoms and Tankebe 2012), in which:

What we [as subjects] acknowledge as legitimate, here and now, is what, here and now, makes sense as a legitimation of power as authority; and discussions about whether it does make sense will be engaged, first-order discussions using our moral, social, interpretive and other concepts. (Williams 2005, p. 11)

We have already seen this process in action in the IEP study. A new policy for prisons was introduced by a properly elected government minister. Prisoners did not contest the fact that he had the right to introduce the policy, nor did they object to the principle that ‘privileges must be earned’. But they made it very clear that they considered many of the ways in which the policy was being introduced in their particular prison were unfair, so that—‘here and now’, as Williams puts it—staff were behaving in a less legitimate way than they had done before the policy was introduced. This example also illustrates the point that judgements of legitimacy are not made only in relation to major issues of principle (such as ‘Is the law against the possession of cannabis justified?’); they can concern much more mundane matters concerning the detailed enforcement of laws and regulations by small groups of staff or even in some circumstances by an individual officer.

In the IEP study, the delegitimating factor was a widespread sense of unfairness, and there is strong support in empirical studies that the fairness or unfairness of state agents is a key dimension of legitimacy—both in the sense of procedural fairness and that of distributive fairness (see Bottoms and Tankebe 2017, also Tankebe’s chapter in this volume). Other delegitimating factors that have been identified are unlawful actions by the power holder and evidence that power holders are ineffective in performing their primary tasks (Bottoms and Tankebe 2017).

Judgements of legitimacy are clearly normative judgements,²² and this emphasizes the point that, as Paul Tucker (2018, p. 157) put it in a recent text on independent, non-elected, government bodies, ‘values are part of the fabric of a political community’. Tucker goes on to make an interesting point about the relationship between instrumental and value-based principles in independent agencies within governmental structures. We do not, he suggests, typically respond to obviously poor performance by agencies by suggesting that the incentives structures were incorrect; indeed, if we think about agencies solely in terms of instrumental principles, we will quickly go wrong.²³ To maximize legitimacy, we therefore need to achieve what he calls ‘incentives - values compatibility’; that is, we need to align the incentives structures within the agency to the achievement of its true purposes, delivered in a legitimate way.

3.2.2 Acceptance of, or Belief in, Social Norms

This sub-type of compliance requires little elaboration. If a person sincerely holds a particular normative belief, it is reasonable to predict that she/he is more likely to act in accordance with that belief than someone who does not hold that belief. (To take some questions from Braithwaite’s [2013, pp. 113–114] survey, someone who affirms that ‘Paying tax is the right thing to do’ will obviously be less likely to under-report income than someone who says ‘I don’t care if I am not doing the right thing by the tax office’.) Of course, we do not always act in accordance with sincerely held beliefs (we can be tempted to do otherwise), but this does not detract from the importance, in many circumstances, of normative beliefs as a mechanism of compliance.

Since people derive many of their normative beliefs from their childhood experiences, drawing a distinction between compliance based on normative beliefs and habitual compliance based on early socialization can, in practice, be difficult (Decoene and Beyens 2013, p. 210). However, in principle these two mechanisms are separable, so it is important to include both in a dissective typology of mechanisms. An empirical example of the separateness of the two mechanisms would arise where a teenager who has been brought up to believe that theft is wrong, and has habitually behaved accordingly, is suddenly invited by school friends to join a shoplifting expedition. If, after considering the matter, he refuses, then his habitual legal compliance has, in this situation, become compliance based on his normative beliefs. In the business context, a similar situation might arise where a new employee is invited by longer-serving colleagues to ‘bend’ some rules relating to transactions that he is handling.

3.2.3 Normative Attachment and Compliance

Although normative beliefs can and do on occasion act as causal mechanisms of legal compliance, research in social psychology shows that most of the time people do not ‘function on this explicit moral reasoning level, or habitually make conscious norm-based decisions’ (Decoene and Beyens 2013, p. 210). There is, by contrast, extensive evidence that people may be influenced by the moral context of the situations, or the social groups, in which they find themselves, or which they choose to join—as has been shown by, for example, classic experimental studies such as those of Solomon Asch and Stanley Milgram. (For a summary and assessment of these studies, see Smith and Haslam [2012, Chapters 5 and 7].) In this area of analysis, therefore, Decoene and Beyens (2013, p. 222) are right to say that criminology ‘could benefit from a renewed friendship with social psychology’.²⁴

In my 2002 paper on compliance, I discussed issues of this kind more criminologically, and specifically with reference to Travis Hirschi’s (1969) concept of ‘attachment’. Within Hirschi’s broader theory of social control, the concept of attachment is linked to the notion of individuals’ ‘bonds to society’—which are clearly normative. In that earlier paper, I gave examples of such attachments both in the field of desistance from crime (normative links to romantic partners) and as regards the normative climate within schools, which has been shown to influence juvenile delinquency rates over and above the individual propensities to crime of those attending the school. I further pointed out that attachments to non-legally compliant individuals or groups, such as criminal peers, can have the reverse effect (Bottoms 2002, pp. 34–36).

All this remains valid and pertinent to both explanation and policy in the field of legal compliance. A related topic that was not discussed in my 2002 paper, but which is clearly relevant to compliance in the field of business regulation, is that of corporate reputational issues. There is significant evidence that businesses can be influenced to comply with regulations if there is a normative climate in relation to that particular industry such that non-compliance will be reputationally damaging to the firm (see, e.g., Kagan et al. 2003 on the compliance of pulp and paper manufacturing mills with pollution regulations).²⁵ Such a normative climate then acts as an incentive, analogous to the incentive provided by public disclosure in the Klepper and Nagin (1989) study—see above.

A contrasting example might be that of an employee who, in terms of her personal morality, was originally a ‘committed complier’ (in Valerie Braithwaite’s terms), but whose prolonged cultural exposure to the culture of a firm operating an aggressive version of ‘creative compliance’ alters her motivational posture. The policy implication, of course, is that if regulators can work with industry representatives and the general public to create a culture normatively favourable to compliance, this is likely to be beneficial to the overall compliance rate.

It is necessary, finally, to return briefly to the relationship between this sub-type of compliance and that of normative beliefs. As has been seen, there is an analytical difference between these two sub-types, yet they remain related in complex ways. For example, when a leading researcher on legitimacy asks questions such as how morally wrong it would be to break various specific laws (on a five-point scale from ‘very wrong’ to ‘not wrong at all’), he can very reasonably describe the resultant data as a measure of ‘personal morality’, in contrast to measures of legitimacy (see Tyler 1990, p. 44). In the language used in this chapter, however, this measure of ‘personal morality’ almost certainly contains both an element of ‘normative belief’ and an element of ‘compliance arising from normative attachment’. Yet even within such a measure, one might be able to make some further, and relevant, differentiation. Thus, for example, Wikstrӧm et al. (2012), in their UK study of juvenile delinquency, constructed a ‘generalized [personal] morality scale’ (pp. 132f.) which, as a key part of their ‘crime propensity index’, was a good predictor of criminality.²⁶ However, the researchers also found that, even after controlling for individual crime propensity, ‘young people’s rate of crime …. was highest … in [social] settings with a weak law-relevant social context’, such as the city centre (p. 363). In other words, both the personal morality of the individual and the moral context of particular social settings were, in this study, relevant to the explanation of whether a young person chose to commit crime or to comply with the law.

3.2.4 Compliance Through Normative Cues

A recent strand of research in social psychology has shown, intriguingly, that people can be influenced to comply not only by the normative climate of a particular social setting, but also by normative cues that are signalled in one-off encounters. Major theoretical texts that underpin this approach are Cialdini’s (2014) Influence and Thaler and Sunstein’s (2009) Nudge, but here, I will simply illustrate the possibilities of the approach with two striking examples of norm compliance.

Keizer (2010, Chapter 3, Study 5) carried out an experiment in which a confederate of the researcher wheeled a bicycle along a street and then ‘accidentally’ dropped some oranges while she/he was trying to put them into a bag (all the time continuing to hold the bicycle). Would passers-by help the confederate to pick up the oranges? The answer: in the ‘baseline’ condition, as described above, 40% of passers-by helped the cyclist, but in a second condition, when a further confederate could be seen sweeping up litter nearby, the helping rate for the cyclist doubled (83%: P < 0.001). As the author concludes, this result (together with others from related experiments) supports ‘the expectation that subtle cues of respect for a norm’—in this case, that it is good to keep streets tidy—can influence compliance with a similar or related social norm—in this case, that it is good to help strangers in difficulty. Indeed, ‘one person armed with a broom was able to boost helping others in need by … 100 per cent’ (p. 51).

The second example shows that normative cues towards compliance can be triggered even by words in a letter. Researchers in the UK Cabinet Office became interested in the science of normative cues and carried out an experiment in collaboration with the tax authorities. In the UK, personal income tax that is not deducted at source is payable on two set dates each year. There are penalties for late payments, but these still occur, and delays in payments cost the government very large sums of money. In a randomised controlled trial, the tax authorities sent out five differently worded reminder letters to those who had not paid their tax by the due date. Details are shown in Table 1, which also shows that the wording that was most effective in triggering payment combined references to (i) what ‘most people do’ (with the implication: ‘you don’t want to be different, do you?’); (ii) ‘your local area’ (so: ‘we’re not talking about people who live in other places’); and (iii) ‘a debt like yours’ (subtext: ‘you’re a bit exceptional in having a debt, and you’ll want to clear it, won’t you?’).

Table 1 Tax payment rates by late-paying individuals, one month after receiving differently-worded reminder letters

This kind of letter improved early payment by 5.4 percentage points over the control condition, which is, as Halpern (2015, p. 114), rightly claims, ‘not bad for a single extra line of text in a letter that was going out anyway’. Clearly, this kind of result has implications for compliance officers, who may wish to consider in what ways normative cues could be effectively used in their particular regulatory environment.

3.3 Situational Compliance

In my earlier writing on types of compliance (Bottoms 2001, 2002), this third type of compliance was described as ‘constraint-based compliance’, and in the more developed 2002 paper, it was divided into two sub-types, respectively titled ‘physical restrictions’ and ‘social-structural constraints’.

Richard Jones (2007, p. 174) rightly criticized this categorization on two grounds. Firstly, he pointed out that the category ‘physical restrictions’ effectively excluded analogous forms of restraint that are virtual rather than physical, such as blockages placed in software programs. Jones suggests, following Lessig (1999), that a better term than ‘physical’ would be ‘architectural’, a word which Lessig uses in a broad sense ‘to refer to such wide-ranging phenomena as … built architecture, … the laws of physics insofar as they affect human movement and communication, … computer and network hardware, … and the “virtual” architecture of software’ (Jones 2006, p. 177). Secondly and more importantly, in relation to ‘architecture’ Jones (2007, p. 174) argues that rather than emphasizing only ‘constraints’ as a way of achieving compliance, ‘it seems more theoretically desirable to speak of the ways in which architectural compliance-seeking approaches always simultaneously both constrain and enable’. While the simultaneity of constraints and enablements in relation to compliance can be questioned,²⁷ Jones is certainly correct in emphasizing that ‘architectural’ features can sometimes enhance compliance by the way in which they enable certain forms of action. It is equally the case that the other sub-type in this category identified in my earlier papers (‘social-structural constraints’) is better theorized as including social-structural enablements as well as constraints.

As shown in Fig. 1, we can therefore appropriately identify four sub-types within this mechanism of compliance, namely (i) ‘architectural constraint-based’; (ii) ‘architectural enablement’; (iii) ‘social-structural constraint-based’; and (iv) ‘social-structural enablement’.

A useful generic name embracing all these sub-types is situational compliance. This term has the merit of emphasizing that this type of compliance is truly distinct from both instrumental and normative compliance: that is to say, the compliance—whether architectural or social-structural—is caused by specific features of the situation and not by instrumental or normative mechanisms. As all criminologists will instantly recognize, the term ‘situational’ is borrowed, with gratitude, from the genre of ‘situational crime prevention’, which has been influential in crime prevention policy since the 1970s (see Clarke 2018).

3.3.1 Architectural Constraint-Based Compliance

Architectural constraint-based compliance itself has three subdivisions, depending on whether the constraint is applied to the potential target, the potential rule-breaker or the possible means used to break the rule.

Applying constraints to the potential target, thus making the target harder to access, is an extremely common way of attempting to promote legal compliance. At a basic level, it happens every evening in millions of homes and businesses as people lock doors and close windows. More advanced physical constraints are used in some shops (bars on jewellers’ windows) and in banks (very thick doors to vaults), and these may be supplemented by technological devices such as controlling entry to certain spaces with numerical codes. As noted above, within the cyber world, virtual constraints may also be utilized to prevent access to, for example, sites containing images of sexual abuse.

Applying constraints to potential rule-breakers is also a very familiar practice, seen most obviously in the use of imprisonment as a penalty. When someone is in prison, by definition he/she cannot commit crimes in the community, and this has led to the development of a scholarly literature on ‘incapacitation’, which has attempted to estimate the crime-reductive effects of various versions of an increased (and usually selective) use of imprisonment.²⁸ It is important to recognize, however, that imprisonment is not the only way of applying architectural constraints to potential law-breakers in an attempt to improve compliance. For example, in many countries the range of available non-custodial penalties now includes the possibility of ordering the defendant to stay at home at specified times, electronically monitored by wearing a ‘tag’. Less formally, teachers in schools often deliberately place potentially troublesome pupils at some distance from one another within the classroom, so that they are not tempted to start fighting, and at professional soccer matches in England, the police routinely adopt a similar tactic in relation to the fans of the two teams, channelling them to the stadium by separate routes, under police supervision.

These examples all relate to personal offenders, but of course constraints can also be applied to corporate offenders, forbidding them for a time from engaging in certain activities, or, in extremis, removing their licence to trade, or closing down the company altogether. In this connection, it is worth remembering that the final two sanctions in the first iteration of Ayres and Braithwaite’s (1992, p. 35) well-known ‘responsive regulation enforcement pyramid’ are those of ‘licence suspension’ and ‘licence revocation’.

‘Architectural’ compliance can be enhanced not only by placing restrictions on potential targets and on potential offenders, but also by restricting access to the means of committing an illegal or disapproved-of act. This is why, for example, virtually every state has strict controls on the sales of poisons, and also why most states have extended the same logic to the availability of firearms. However, rather than enter here into the contentious US-dominated literature on gun control, I will illustrate the relevance of this form of situational control through a well-documented example in which greater compliance with the social norm against committing suicide was an unintended consequence of an altered situation. This occurred in Great Britain in the 1970s, when extensive supplies of natural gas were discovered under the North Sea (off the east coast of England) and the government decided for economic reasons to ‘mine’ this supply, and simultaneously to discontinue production of the previously used gas (‘coal gas’, derived from the burning of coal stocks). Many families in Great Britain use gas cookers in their kitchens, and significant numbers of suicides had previously taken place using the toxic coal gas readily available in so many domestic ovens. But North Sea gas is non-toxic, so after the change, this method of suicide was no longer available. The national suicide rate dropped, and careful scholarly analysis has shown that the only plausible reason for the reduction was the alteration in the nature of the gas supply (Clarke and Mayhew 1988). This research has therefore shown that even a deeply personal decision such as trying to end one’s own life can be significantly affected by a situational change in the available means to achieve one’s purpose.

3.3.2 Architectural Enablement

Good examples of compliance through architectural enablement may be found in the sphere of road traffic. In their book Nudge, Thaler and Sunstein (2009, pp. 41–42) provide an illustration from their home city of Chicago concerning a stretch of lakeside road which has ‘stunning views’ but also ‘a series of S curves’. Approaching the S curves, there is a mandated speed reduction (to 25 mph), but in the past, many drivers failed to obey this requirement, and the accident rate was high. In an attempt to improve the situation, the city authorities painted a series of stripes on to the road as a visual signal to drivers:

When the stripes first appear, they are evenly spaced, but as drivers reach the most dangerous portion of the curve, the stripes get closer together, giving the sensation that driving speed is increasing…..One’s natural instinct is to slow down. When we drive on this familiar stretch of road, we find that those lines are speaking to us, gently urging us to touch the brake before the apex of the curve. (p. 42)

Thus, ‘nudging’ can work in a situational manner, as well as by offering cues to normative principles that are held by an observer (see previous section).²⁹ But architectural enablement can also go beyond nudging, and a good illustration here is the fact that traffic accidents are significantly lower (per thousand vehicles) on designated motorways than on other types of road.³⁰ The reasons for this are straightforwardly situational and include the banning of pedestrians and very slow vehicles on motorways, and the fact one can only enter or leave such roads at a limited number of designated points.

In other contexts, such as pollution control from industrial waste, it will no doubt be possible for those with appropriate technical knowledge to identify analogous examples of strategies of architectural enablement, both of a ‘nudging’ and of a wider situational character.

3.3.3 Social-Structural Constraint-Based Compliance

This sub-type is, as we shall see, not a mechanism of compliance that can be ethically commended as something that law enforcement personnel should seek to cultivate. It arises, as I explained in my first essay on compliance, in situations where people are ‘cowed into submission by the coercion [of] a power-based relationship’ (Bottoms 2001, p. 93). In order for such compliance to be truly situational, the person complying will not be motivated by instrumental concerns, nor will he or she find any normative reasons to comply. The best term to characterize the nature of this kind of compliance is therefore—as David Lockwood (1992, p. 43) has suggested—that of ‘fatalism’. As he puts it, in conditions of entrenched social inequality a sufficient condition of obedience ‘is simply that the structure of power, wealth and status is believed to be inevitable’, and so it is fatalistically accepted by those without power.

Colleagues and I encountered such fatalistic compliance when conducting research in a so-called Vulnerable Prisoner Unit in a high-security prison (Sparks et al. 1996, Chapter 6). This unit (consisting of two prison wings) was reserved for those prisoners—such as sex offenders and those with debts to other prisoners—who would face hostility, and perhaps injury, from other prisoners if they were located in the mainstream prison system. Given this situation, they had little option than to accept the regime that the staff offered. Unfortunately, some staff realized this and behaved in an unprofessional manner: as one staff member candidly explained, ‘the power can go to your head easy’ (p. 214). But given the prisoners’ powerless condition, even staff malpractice was usually accepted by prisoners without formal complaint.

It is instructive to compare this mode of compliance with that of ‘creative compliance’. Creative compliance is possible only when those expected to comply have considerable economic and social resources; social-structural constraint-based compliance occurs only when those expected to comply feel powerless. In different ways, both modes of compliance show that, in real-life social situations, social power can significantly affect the dynamics of compliance.

3.3.4 Social-Structural Enablement

A good example of social-structural enablement is to be found in policies relating to the public availability of criminal records. Concern has been expressed, particularly by American criminologists, that in recent years, given informational and internet developments, ‘criminal records have become cheaper to access and more widely accessible’ with the result that ‘millions have moved from the category of “potentially discreditable” to the category of “formally discredited”’ (Uggen and Blahnik 2016, p. 229, emphasis in original). This development has consequences in a number of spheres, but particularly in relation to employment, making it significantly more difficult for those with criminal records to obtain jobs, especially satisfying jobs. This has led to some increased attention to the potentialities of policies (already adopted in some countries) of ‘sealing’ criminal records in order to enhance the possibility of desistance from crime. Such policies can take a variety of forms,³¹ but on almost any analysis, they seem likely to have either a positive or neutral impact on offending levels (Kurlychek et al. 2016), so they seem well worth pursuing as a policy to enhance compliance through social-structural enablement.

3.4 Compliance Based on Habit or Routine³²

The fourth principal mechanism of compliance is concerned predominantly—and to a greater extent than any of the other three mechanisms—with unthinking compliance. Within this mechanism will be found compliance based on both routine and habit, which themselves need to be distinguished.

Turning first to routine, consider the type of law that, in many countries, requires the parents of children between certain ages to send them to a school (whether public or private) that is registered with the state. Compliance with such laws is typically high. But compliance requires that active steps be taken, on every weekday in school term, to send children to school. In the average household, there is no debate about whether the law should be complied with; instead, on mornings in school terms there is a semi-organized bustle involving breakfast, packing a school bag with sandwiches and homework and so on. In other words, unthinking routine prevails. This is not a unique example—the law requiring the wearing of seat belts in cars similarly results in routine belt-attachment as people get into their vehicles, the driver checks that everyone has belts fastened, and so on.

Both of these examples concern laws that require specific and positive actions to be taken if one is to comply with them, and ‘legal compliance by routine’ occurs particularly (although not exclusively³³) in relation to that kind of law. However, many laws simply require one to refrain from doing a particular act (say, burglary or street robbery), and in these circumstances, compliance by habit comes more into the picture. Obviously, most adult members of the general population would never even consider committing a burglary or a street robbery—we can say that they have a settled mental disposition (or habitual way of thinking and acting) that rules out this kind of behaviour. An interesting attempt to describe mental dispositions was made by the philosopher Richard Wollheim (1984, p. 34):

Mental dispositions …. are persistent phenomena, which manifest themselves intermittently. They do not occur, nor are they events. They are mutable. Dispositions have histories, which are made up of events, and these histories are varied … Dispositions vary in their beginnings, for some are innate, some arise in the mind, and some are acquired. They persist in different ways, for some remain constant and some change, and they may mature or decline or fluctuate.

This characterization is, in my view, helpful because it draws attention to two contrasting points. Firstly, habits or mental dispositions are described as relatively ‘persistent phenomena’. This is important because if such phenomena exist—and the criminological evidence on ‘crime propensity’ (Wikstrӧm et al. 2012) suggests that they do—then they refute strong versions of psychological situationalism, which argue that people’s behaviour is so situationally influenced that concepts such as ‘character trait’ have to be abandoned. (On this debate in a philosophical context, see Sreenivasan 2013.) Secondly, however, habits or mental dispositions are not immutable, and they can and do change as people encounter fresh sets of circumstances. This point is also confirmed in the criminological literature, where research on desistance from crime has shown that offenders—even persistent offenders—can and frequently do, over time, reduce their crime propensity and acquire a fresh set of routines and habits of thought (for a summary, see Shapland and Bottoms 2017). These observations of course refer to individuals, but similar features are seemingly also found in corporations, since the concept of a ‘corporate culture’ within a given firm or organization is frequently used, and of course such cultures can change over time.

4 Interactions Between Different Mechanisms of Compliance

The analysis in this chapter has proposed four principal mechanisms of compliance: instrumental, normative, situational and habitual. Following the recommendations of analytical sociologists, uncovering these mechanisms, and their sub-types, has involved a process of ‘dissection’, in which the aim has been ‘to decompose a complex totality into its constituent elements and activities, and then to bring into focus what is believed to be its most essential elements’ (Hedstrӧm 2005, p. 2).

However, in any real-life analysis of compliance—whether undertaken as a social scientist or a law enforcement officer—initial ‘dissection’ is only a preliminary step. One also needs first, to consider whether—and, if so, how—different mechanisms of compliance might be interacting with one another, and second, how the mechanisms are operating within the social structures and relationships in that particular situation. The latter point is considered briefly in the next section, but first we will discuss potential interaction effects between the three most ‘active’ mechanisms of compliance, namely the instrumental, the normative and the situational.

4.1 The Instrumental and the Normative

Interactions between the instrumental and the normative have been fairly frequently noted in the research literature, and we have already seen examples of such interactions in the IEP prisons study and in the Klepper-Nagin questionnaire study on taxpaying.

Interactions between the instrumental and the normative can in principle work both ways (i.e. normative → instrumental or instrumental → normative). In the Klepper-Nagin research, the addition of a significant normative dimension (exposure to public disapproval as a ‘tax cheater’) acted as a major incentive to compliance. The data in that paper were derived from a ‘scenario-based’ questionnaire, but its core finding has been confirmed in real-life research, perhaps most notably in the path-breaking series of experimental studies on misdemeanour-level domestic violence in the USA, reported by Lawrence Sherman (1992).³⁴ From these various sources, a reasonable conclusion from the empirical evidence is that there is an integral connection between normative and instrumental compliance, such that incentives and disincentives work best for persons or corporate entities that have strong ties to a social group (e.g. a family, a local community or fellow members of a business community), in a context where reputation matters to the social standing of the individual or corporate entity in that group, and members of the social group have clear normative expectations that are related to the behaviour at which the law or regulation is aimed.

The previous paragraph describes situations in which normative factors act to enhance instrumental compliance. However, as the IEP study shows, other kinds of normative factor (in that research, perceptions of unfairness by staff, i.e. lack of legitimacy) can act to inhibit instrumental compliance.

In the reverse direction of causality, can instrumentally conceived enactments enhance normative compliance? In principle, they can, although the empirical evidence here is more tentative. Perhaps the main examples of this kind of effect have occurred when a law created for its hoped-for instrumental effect subsequently generates a normative change. An interesting illustration of this kind of process has been observed as regards drink-driving in Great Britain. A new law on this topic was passed in 1967, for the first time authorizing breath tests on drivers, and (save in exceptional circumstances) mandating compulsory disqualification if one was found guilty of driving when over the legal limit. At the time of its enactment, this law was very much seen as an attempt to secure compliance by instrumental (deterrent) means, since public normative attitudes on drink-driving were, at the time, certainly not consistent with the ‘mirror thesis’ (see Note 20 above). A degree of normative contestation continued for a decade or two,³⁵ but perceptions gradually changed, and now—half a century after the law was passed—there is a very widespread public consensus that this is a normatively just law.

4.2 Interaction Effects from Situational Measures

Situational prevention measures can and do, in some circumstances, have secondary effects both of an instrumental and of a normative kind. By contrast to instrumental–normative interactions, however, in this instance the causation does not seem to operate in both directions.

That situational prevention can create longer-term instrumental effects should come as no surprise to anyone. Thus, for example, an agency that is managing a gold bullion store will obviously wish to protect it with some serious hardware, to prevent thieves from accessing the bullion (= reduction of opportunity, a key feature of situational crime prevention). The serious hardware will, however, inevitably have the secondary effect of acting as a significant disincentive to all but the most determined and professional thieves. Secondary instrumental effects of this kind can even occur as an unintended consequence of a situational measure. Thus, in the 1970s in Great Britain, a law was passed requiring motorcyclists to wear a crash helmet, in order (through a situational measure) to reduce the severity of injuries in accidents. In the year after this new law came into force, in Greater London the rate of thefts and unauthorized takings of motorcycles decreased by a quarter, although the equivalent rate in respect of other motor vehicles rose significantly. Why did the reduction occur? Home Office researchers explained the change by noting that many unlawful takings of motorcycles are ‘opportunist’ (i.e. committed without planning), so ‘it is reasonable to think that some potential users (aware of what was a well-published requirement [to wear a helmet]) have been deterred from illegally taking two-wheelers because of their increased visibility if not wearing a crash helmet’ (Mayhew et al. 1976, p. 18). A situational change focused on safety had therefore promoted instrumental compliance with the theft laws, based on an increased probability of detection.

Situational prevention can also result in an enhancement of normative compliance. The leading example of this kind of effect is that described by Welsh and Farrington (2008) in their Campbell Collaboration review of research on the effects of improved street lighting on crime. The authors conclude that the five most recent such evaluations (all of which were UK studies) ‘showed that improved lighting led to decreases in crime’ (p. 19). However, a puzzling feature of the results was that, in these areas, night-time crimes did not decrease to a greater extent than daytime crimes: in other words, crime decreased as much when the lights were off as when they were on. This was surprising because improved street lighting had been promoted as a form of situational crime prevention, reducing opportunities for crime by lighting dark places, as well as making offenders more visible (and so easier to detect). Welsh and Farrington suggested that the most plausible explanation for the unexpected ‘daytime’ finding was that improved street lighting resulted in ‘increasing community pride and informal social control’. To support this suggestion, the authors cited evidence from one area where the improvement in lighting was very obvious and where data showed that ‘tenants thought that their quality of life had been improved’ (p. 19). In other words, improved lighting had given rise to a normative effect of improved well-being among residents, which in turn seemed to have led to improved informal social control even in daylight hours, and hence less crime.³⁶

5 The Nielsen–Parker Holistic Compliance Model

In the field of business (or regulatory) compliance, most writers focus on what I have described as ‘compliance from above’ rather than ‘compliance from below’. It is therefore of special interest, for the purposes of this chapter, that a leading pair of authors on regulatory compliance (Christine Parker and Vibeke Lehmann Nielsen) have written in both of these ‘voices’. Their leading paper on ‘compliance from below’ (Nielsen and Parker 2012) sets out an explanatory model, which is now described as the ‘Nielsen–Parker Holistic Compliance Model’ (see Parker and Nielsen 2017, p. 220). This is reproduced here as Fig. 2,³⁷ and—as may be seen—it combines three distinct sets of explanatory variables. The core set concerns the regulatee’s interests and motives (box on left-hand side of the diagram); to these are then added (i) three ‘psychological or organizational factors inside the regulatee’ (Nielsen and Parker 2012, p. 450), which I shall describe as ‘internal influences’ (boxes above the horizontal line), and (ii) the ‘external influences’ shown in the box below the horizontal line.

Fig. 2

(Source Nielsen and Parker [2012, p. 448], as modified in Parker and Nielsen [2017, p. 220])

The Nielsen–Parker holistic compliance model

It will be useful to consider each of these sets of variables separately and to compare each with aspects of the analysis in this chapter—which I shall refer to in shorthand as the ‘Typology of Compliance Mechanisms’ or TCM. As will be seen, a significant point arises from each of these comparisons.

5.1 Business Motives

The regulatee’s interests and motives are central to Nielsen and Parker’s (2012) explanatory model. In their 2012 paper, they hypothesize that firms will have three principal ‘business motives for compliance’, namely economic or material motives, social motives and normative motives. Economic motives are said to relate ‘broadly to the motives of the stereotypical amoral calculator’ (p. 431); they therefore, in the language of the TCM, constitute a mechanism for compliance based on incentives and disincentives. Social motives are said to relate to ‘earning the approval and respect of significant others with whom an actor interacts’, a group which could include ‘other businesses … employees, customers, local communities, the wider public, family and friends’ (p. 432, emphasis deleted). Again, this corresponds closely to a sub-type in the TCM, namely that of compliance through normative attachment. Nielsen and Parker note that, in the field of business compliance, some theorists have proposed merging the categories of economic and social motives, given that ‘social stigma might lead to … economic and other losses’, but they resist this suggestion because, they argue, acts of non-compliance arising from economic and from social motives sometimes require different policy responses by regulators (p. 432). As has been seen, the TCM analysis, while recognizing that ‘social motives’ can sometimes have instrumental effects, also supports keeping these categories separate, because differing mechanisms are in play. Nielsen and Parker’s third category of motives, ‘normative motives’, refers to ‘compliance as a result of internalized moral judgements by people about the substance and procedures of the law’ (p. 432). This category clearly maps on to the TCM sub-type of ‘normative beliefs’, although close examination of Nielsen and Parker’s text suggests that they also intend ‘normative compliance’ to include what in the TCM is the further sub-type of legitimacy.

In an empirical study of Australian firms (see Note 20 above for details), Nielsen and Parker showed through a principal components analysis that firms’ motives for compliance (or otherwise) could indeed be appropriately classified into the economic, the social and the normative.³⁸ The same empirical study also showed, however, that the great majority of firms ‘rate[d] all the motives fairly highly’ (p. 443): that is, they held a plurality of motives, although the researchers found that there were ‘small gradations of difference as to how they do so’ (p. 445).³⁹

Comparing Nielsen and Parker’s compliance model with that of the TCM, there is clearly considerable congruence between the three principal motives that these authors have identified and the main sub-types within the instrumental and normative mechanisms of the TCM. Perhaps surprisingly, however, there is nothing in the Holistic Compliance Model about situational compliance, but on reflection, this difference probably reflects the differing theoretical purposes of the two typologies. The TCM is explicitly focused on identifying the main mechanisms of compliance ‘as seen from below’, whatever the source of the mechanism. By contrast, the Holistic Compliance Model is more concerned with the relationship, within business regulation, between regulators and regulatees, with a strong policy focus on improving compliance, and perhaps situational compliance mechanisms are rarely discussed within such relationships.⁴⁰ There is, of course, room for both approaches, and it is hoped that the analysis in this chapter will complement that of Nielsen and Parker. Moreover, on one point there is a firm concurrence between the two analyses: both agree that much compliance is multifaceted and that the interconnections between different (and often plural) reasons for compliance need to be carefully considered in any serious look at compliance or non-compliance in a particular social situation.

5.2 Internal Influences

The ‘internal influences’ identified by Nielsen and Parker (Fig. 2, above horizontal line) refer to matters such as the characteristics of decision-making within the firm (are decisions emotionally or rationally driven?; is compliance seen as important or as an ‘afterthought?’); the capacity of the firm to act strategically in implementing decisions; and the resources available to the firm, since differences in resource levels are known ‘in large part to explain differences in compliance behavior’ between firms (Nielsen and Parker 2012, p. 449). In relation to compliance with the mainstream criminal law, the analogue of these ‘internal’ matters would of course be psychological differences between individuals.

It is at this point that a note of caution must be raised. For certain purposes, including those important to regulatory compliance officers (and therefore to Nielsen and Parker), these ‘internal’ variables can be very important—as would also be attested, for example, by probation officers working to promote desistance from crime with individual offenders. But that is less the case in other contexts, and these other contexts include the work of analytical sociologists and overall analyses of particular social policies. As regards analytical sociology, we need to recall that, as Hedstrӧm (2005, p. 5) put it, ‘the focus on [individuals’] actions is merely an intermediate step in an explanatory strategy that seeks to understand change at a social level’—which means that such analyses, although hopefully realistic, must necessarily contain an element of generalization, rather than individuation (see the introduction to this chapter). As regards social policy, it is useful to look back to the IEP example and to recognize that no detailed individual-psychological research was needed to identify the main mechanisms in play in that situation, nor to understand why the policy did not have the desired effect. Accordingly, an important difference between Nielsen and Parker’s Holistic Compliance Model and the analysis of this chapter is that ‘internal’ influences relating to regulatees are more important for the former than the latter.

5.3 External Influences

This final element in the Nielsen–Parker model is extremely wide-ranging in scope. As can be seen in Fig. 2 (box below horizontal line), it covers ‘external agents, environments and events’—which means that it embraces matters as diverse as the actions of the regulator; decisions of professional bodies such as those for accountants or lawyers; social pressures in relation to a particular industry (see, e.g., Note 25 above in relation to pulp and paper mills); and specific events such as fresh legislation. In the TCM, this wide range of potential influences would be handled more specifically—for example, relations with the regulator would be considered under both ‘legitimacy’ and ‘instrumental compliance’, and social pressures would be seen as potentially influencing compliance by normative attachment.

Despite this analytical difference, both approaches to understanding compliance emphasize the potentially very great importance of these ‘external’ influences. As previously indicated, in the TCM the analytic strategy is first to identify the mechanisms in place, in a ‘dissective’ analysis; then to consider possible interactions between mechanisms; and finally to relate this mechanism-based analysis to the social structures and relations in play in the particular context. The authors of the Holistic Compliance Model similarly see compliance as a very complex matter, and they note that this has implications for compliance officers. Citing various authors such as Baldwin and Black (2008) on ‘really responsive regulation’, they conclude that such officers need to develop:

understanding and insight into the multifarious actors and mechanisms that interact with one another to influence and create compliance. It is foolhardy to assume that just because one or two factors have been addressed….compliance will automatically increase. …. [I]t is up to regulators to have the skill and ultimately the courage … to craft solutions and alliances that are responsive to the complex social, economic and political contexts in which they work. (Parker and Nielsen 2017, pp. 230–231)

6 Conclusion

The quotation from Parker and Nielsen (above) rightly emphasizes the multifaceted and interactive complexity of compliance processes. The analysis in this chapter has tried to provide a set of tools with which one can begin to unravel this complexity.

It has been argued, following the recommendations of analytical sociologists, that a mechanism-based approach offers the best framework for the explanation of legal and regulatory compliance, while at the same time granting the ‘interpretative’ tradition of social science greater weight than most analytical sociologists would allow. Particular attention has been paid to the ‘dissective’ dimension of mechanism-based analyses, as recommended by Peter Hedstrӧm; and in a ‘dissective’ analysis, four principal mechanisms of legal compliance have been identified, namely instrumental (or prudential) compliance, normative compliance, situational compliance and compliance based on habit or routine. The first three of these mechanisms are also shown to have some distinct sub-types within the general mechanism, and the specific features of these sub-types require careful attention (see Fig. 1 for an overview). To complete the analysis, interactions between the mechanisms, and their functioning in real-life situations, have been considered.

As has been indicated, this chapter builds on some earlier work of this author on legal compliance (Bottoms 2001, 2002), some aspects of which have seen constructively criticized by other scholars (see Note 12 above). This chapter tries to provide an improved analysis, while also to an extent widening the substantive scope of the discussion in an attempt to address some features of compliance in the field of business regulation. How far it succeeds in these dual aims is, of course, for readers to judge.⁴¹

Notes

• 1. See further Hedstrӧm (2005, Chapter 2), where he defends the view that mechanism-based explanations are more satisfactory as causal explanations than are two alternative possibilities, namely ‘covering law explanations’ and ‘statistical explanations’.

• 2. This approach is commonly, and fairly, described as a version of ‘methodological individualism’(MI). However, as various writers—including Hedstrӧm (2005, p. 5, n. 4)—have pointed out, acceptance of this proposition does not entail acceptance of more extreme MI statements, such as Elster’s (1989b, p. 248) ‘there are no societies, only individuals who interact with one another’. For a helpful discussion of MI and analytical sociology, see Demeulenaere (2011b, pp. 3–9).

• 3. On this point, Hedstrӧm (2005, p. 3) follows the ‘analytic realist’ approach of Talcott Parsons (1937, p. 730), in which ‘the concepts of science are not fictional…[yet they]correspond not to concrete phenomena, but to elements in them which are analytically separable from other elements’.

• 4. He agrees with Max Weber that the ‘individual is the sole carrier of meaningful conduct’ (p. 153), but is much more interested in quantitative than in qualitative research, and—for example—deeply sceptical of Pierre Bourdieu’s concept of habitus, which he regards as ‘mystifying’ (p. 4).

• 5. For a good example of interpretative research in the field of business compliance, see the work of Lauren Edelman in the area of employment discrimination in the USA, summarized in Edelman and Talesh (2011). Edelman showed that organizations initially responded to new laws on discrimination by ‘developing written rules, procedures and policies in an attempt to achieve legal legitimacy, while simultaneously limiting law’s impact on managerial power and unfettered discretion over employment decisions’ (p. 107). This could be fairly described not as committed compliance, but rather as ‘creating structures designed to symbolize attention to law’ (p. 110, emphasis added). Edelman then outlines a sequential process whereby ‘organizational actors tend to construct laws in ways that are consistent with traditional managerial logics and goals. As these constructions of law become institutionalized over time, they subtly and gradually affect how courts …. understand the meaning of law and what constitutes rational compliance with law, [since] judges take … cues from norms and practices that become institutionalized within organizations …. [In sum], …. organizational constructions of law gain not only organizational, but also legal legitimacy’ (p. 110). Thus, what is meant by ‘legal compliance’ gradually and subtly shifts over time.

• 6. Within the UK, there are three separate legal jurisdictions, and therefore, three separate prisons systems—England and Wales; Scotland; and Northern Ireland. The IEP policy was applied only to England and Wales.

• 7. This was assessed by a simple measure of change from the beginning to the end of the year in which the IEP policy was implemented. Methodologically, the research would have been stronger if it had included control group prisons as well as IEP prisons; however, the research team’s suggestion to this effect was vetoed by the minister on the grounds that he wanted the policy to be implemented in all prisons.

• 8. One of the five prisons showed an improvement in behaviour, but there were compelling reasons to conclude that this improvement was attributable to factors other than IEP: see Bottoms (2003, p. 157).

• 9. These quantitative measures were based on extensive prior qualitative research by Alison Liebling in other prisons. They are therefore an example of a methodological approach that she has subsequently come to describe as ‘quantified ethnography’.

• 10. For example, ‘justice as fairness’ is a central feature of John Rawls’ (1972) seminal text on justice.

• 11. As the research team made clear in its report, the implication of this point is that with different implementation strategies, prisoner behaviour would probably improve. Since the time of the research, there have been a number of changes to the implementation of the IEP scheme, but no formal further research evaluation has been undertaken.

• 12. While some of these modifications have arisen from my own further reading and reflection, I wish also to acknowledge with gratitude the helpful critiques of these earlier papers by Richard Jones (2006, 2007); by Gwen Robinson and Fergus McNeill (2008); and by Stef Decoene and Kristel Beyens (2013).

• 13. Decoene and Beyens (2013, pp. 219–220) particularly advocate the potential importance for criminology of Herbert Kelman’s social-psychological work on compliance, identification and internalization, which he summarized in an important semi-autobiographical paper (Kelman 2006). Decoene and Beyens quote at length Kelman’s original definitions of these three concepts (p. 219), without making it clear that these were developed in the experimental context of his early work, i.e. ‘a one-time and one-way persuasive communication [in the laboratory], intended to influence a specific attitude or behavior of individual members of the audience’ (Kelman 2006, p. 7). Subsequently, Decoene and Beyens refer to Kelman’s later work on social influence in the broader context of social systems, but they do not specifically quote his rather different definition of compliance in this context, i.e. ‘adherence to the rules or norms of a system (including its laws and customs) – i.e. the behavioral requirements it sets for its members’ (Kelman 2006, p. 11)—a definition that is similar to that adopted in this chapter. Kelman (pp. 17–18) also makes clear that in this broader social context the concept of legitimate authority significantly complicates the model presupposed by the original experiments, in which ‘people [were] presumed to react [simply] on the basis of their personal preferences’.

• 14. Although instrumental compliance is usually undertaken to further the interests of the regulatee, this is not necessarily the case, since as Unger (1976, p. 25) has pointed out, an instrumental choice can be made because of an ‘altruistic concern for welfare of others, as long as the ultimate basis of this choice is one’s own will’.

• 15. In more detail, the stated best practice principles (always jurisdiction-wide) are: (i) random but strategically deployed RBT stops; (ii) full enforcement of the programme; (iii) publicity; (iv) ‘no one gets off’; and (v) targeted strategies for recidivist drink-drivers.

• 16. This is McBarnet’s (2013) term; Braithwaite (2013) calls the same phenomenon ‘game-playing’. While there is certainly an element of game-playing in this kind of activity, I have preferred ‘creative compliance’ because of the seriousness of the intent, and of the consequences, that can be involved.

• 17. The term ‘arguably’ is used deliberately in this sentence. As McBarnet (2013, p. 72) goes on to explain: ‘The argument may fail …. But the existence of an arguable case usually provides immunity from any suggestion of non-compliance. It is simply a failed case rather than a fraud. Creative compliance thus creates “fraud insurance”, enabling business, as simple non-compliance does not, to simultaneously avoid both the criminal label and legal control’.

• 18. In this connection, a further issue raised by some is concern about potential conflicts of interest arising from the so-called revolving door, i.e. the movement of individuals between public office and the private sector: see Transparency International UK (2012).

• 19. A more technical and formal definition of a social norm, which helpfully elaborates the element of expectation within such norms, has been provided by Bicchieri (2017, p. 35): ‘A social norm is a rule of behaviour such that individuals prefer to conform to it on condition that they believe that (a) most people in their reference network conform to it (empirical expectation) and (b) most people in their reference network believe they ought to conform to it (normative expectation)’.

• 20. The caveat ‘but not necessarily’ is important. As Brian Tamanaha (2001, p. xi) has pointed out, the mirror thesis very often does not apply in colonial societies; for example, in Micronesia—where he worked—the law had been ‘transplanted in its entirety from the United States’, with the result that the norms of the legal system were very different from the culture and values of the indigenous society. In the field of business regulation, there can also on occasions be a significant difference between the norms enshrined in regulations and the norms of at least some regulatees. For example, Nielsen and Parker (2012) carried out a survey of nearly 1000 businesses who were subject to the Australian federal law on competition and consumer protection (the Trade Practices Act 1974, or ‘TPA’). The authors classified a minority of respondent businesses (80 out of 943) as ‘dissenters’ because, although these firms recognized as important ‘a variety of social and normative motives that would [in general terms] tend to support compliance as important’, nevertheless they reported ‘a principled disagreement with the very goals of the TPA’ (p. 444).

• 21. In this source, Beetham distinguishes between the social scientist’s understanding of legitimacy and that of the political philosopher, for whom legitimacy is ‘power which is rightful according to rationally defensible standards or principles’ (emphasis in original). For a more extensive discussion of this distinction, see Bottoms and Tankebe (2017).

• 22. This might be questioned when a judgement of illegitimacy is made on the grounds of lack of effectiveness—for is not effectiveness simply an instrumental concept? For a persuasive rejection of this view, see Beetham (2013b, pp. 136f.).

• 23. On this point, see also Elster (1989b), who treats ‘rational, selfish, outcome-oriented behaviour’ (p. 37) as a starting point in his analysis of social order. However, he eventually concludes (‘with some reluctance’: p. 250), firstly, that social norms independently motivate individuals’ order-related behaviour, and secondly, that such norms cannot be regarded as merely rationalizations of self-interest (Chapter 3).

• 24. However, this will require careful attention to definitional issues (see the comments in Note 13 above). It is also very important to remember that the present chapter is an exercise in analytical sociology and that, ultimately, ‘sociology, as a discipline, is not concerned with explaining the actions of single individuals’ (Hedstrӧm 2005, p. 5). See further discussion in the concluding section of this chapter.

• 25. In this industry, the authors described the near-ubiquity of ‘social pressures’ on mill managers from local communities and environmental activists, such that ‘many … managers spoke to us of having to meet the terms not only of their regulatory license but of their “social license”’ (p. 69). A qualitative analysis within the research project also suggested that ‘variations in social pressures have a significant effect on firms’ relative environmental performance’ (p. 69).

• 26. In this study, the construct ‘crime propensity’ is comprised of two variables: ‘generalized morality’ and ‘ability to exercise self-control’. However, the overall crime propensity score is, over time, more closely related to ‘morality’ than to ‘ability to exercise self-control’—see Wikstrӧm et al. (2012, graph on p. 139).

• 27. It is of course correct that any ‘architectural’ feature simultaneously constrains and enables. However, use of some types of feature (such as a prison cell) will generate significantly more constraint than enablement, while other features will have the reverse effect.

• 28. For a short summary of differing methods of attempting to estimate incapacitation effects empirically, see Bottoms and von Hirsch (2010, pp. 113–120). Given that any form of incarceration will probably produce at least a small reduction in crime, a key issue is the cost effectiveness of various sentencing strategies with an incapacitative element, and students of incapacitation quickly learn that cost effectiveness is very difficult to achieve, given the expensiveness of imprisonment.

• 29. Indeed, apparently similar physical interventions can trigger different mechanisms of compliance. For example, painting stripes on the road in Chicago produced situational compliance because of the illusion created by the decreasing distance between stripes; by contrast, painting footprints leading to litter bins on sidewalks in Copenhagen led to less litter being left on the sidewalk, but this was because of the normative cue (‘good citizens put litter in bins’) that the footprints had created (see Halpern 2015, p. 94, reporting results obtained by Pelle Hanson in Copenhagen).

• 30. In Great Britain, motorways carry 21% of vehicular traffic, but only 6% of road fatalities occur on them, despite high average speeds: Department for Transport (2018, p. 16).

• 31. For example, one type of policy prevents the public disclosure of convictions received as a juvenile, provided that there are no adult convictions. Another kind of policy ‘seals’ a criminal record after a certain period without convictions, thus allowing an ex-offender to lawfully reply ‘no’ to a question on an employment recruitment form asking whether the applicant has at any time been convicted of a criminal offence.

• 32. This section draws on aspects of the discussion in Bottoms (2001, pp. 93–94).

• 33. An important form of compliance that is frequently used by offenders themselves, when they are trying to move away from crime, involves structuring their daily routines so as to avoid locations where temptations to re-offend might arise. (Temptations might occur, e.g., through chance meetings with friends who are still offending, or by going to a bar where fights often break out). This might be regarded as a cross between compliance by the creation of routines and a version of architectural compliance. For a full discussion, see Bottoms (2013).

• 34. Sherman reports on results from several experimental studies in which police officers responded to less serious domestic violence incidents either by arresting the suspect or by some less severe action, the choice of officer response being randomly mandated in advance. One key set of results was summarized by the author as follows: ‘Evidence that the effects of arrest vary by suspect comes from …. [three] cities, …. [where the] data suggest that unemployed suspects become more violent if arrested, but that employed suspects do not. This consistent pattern supports a hypothesis that the effects of criminal punishment depends upon the suspect’s “stakes in conformity”, or how much he has to lose from the social consequences of arrest’ (Sherman 1992, p. 17): i.e. those who had greater ‘stakes in conformity’ were more likely to comply in future because of the social shame of arrest.

• 35. For a detailed examination of the enactment and early history of the Road Safety Act 1967, see Light (1994).

• 36. Welsh and Farrington (2008, p. 19) recognize that one might hypothesize a reverse causal effect, i.e. the normative change ‘comes first, causing improved street lighting on the one hand and reduced crime on the other’. The available data did not permit scrutiny of this hypothesis in all areas. However, in the two areas where time-related data were available, the authors were clear that this alternative hypothesis could be rejected.

• 37. The text included in some of the boxes in the model differs somewhat in the two sources cited at the foot of Fig. 2. Where there is such a difference, the figure always uses the text of the 2017 source, except that the text in two boxes (those that, in the 2017 source, contain only the words ‘Characteristics’ and ‘Capacities’) has been amplified by reference to the 2012 source.

• 38. However, in this empirical analysis four, not three, factors were identified, because ‘normative motives’ were found to split into two distinct factors. One of these related to ‘a general attitude that one should obey the law because it is the law’; the other related to specific attitudes towards the particular legislation that regulators were, in that empirical study, enforcing (Nielsen and Parker 2012, p. 441).

• 39. As an exception to this general picture, a minority of firms (about 8%) expressed normative disagreement ‘with the very substance of the law’ (p. 445) that was being enforced by the regulators—see further at Note 20 above.

• 40. One can imagine reasons why this might be so; although from the perspective of regulatees, situational measures are likely sometimes to be potentially sensible ways of improving compliance.

• 41. I am grateful to my Cambridge colleague Justice Tankebe for his helpful comments on an earlier draft of this chapter.