Policy instruments communicate messages about overarching policy objectives to target groups. Scholarly interest in the messages to citizens engrained in public policies is increasing (Shafir 2013; Thaler and Sunstein 2008). Policy tools have an informational component that affects how they work in practice (John 2016). Scholars of responsive regulation, behavioural public policy and administration point to the huge potential of behaviourally informed intervention as effective intervention strategies (Alemanno and Sibony 2015; Ayres and Braithwaite 1992; Grimmelikhuijsen et al. 2017; Oliver 2013). Surprisingly, however, this insight has not yet made its way into the comparative study of policy instruments. Although compliance with public policies is a key concern, we know surprisingly little about the precise link between policy instruments and outcomes (e.g. Gofen 2015; Knill et al. 2012; see Schaffrin et al. 2015 and Weaver 2014 for recent overviews). To improve our understanding of regulatory effectiveness, this paper adds the dimension of explicitness to existing taxonomies of properties of policy instruments. It illustrates different levels of explicitness and their link with policy outcomes using the case of organ donation regulation.

Policy instruments are a set of techniques used by public actors to ensure support and to either effect or prevent social change (Vedung 1998: 21). “Policy instruments” and “policy output” are closely intertwined concepts. I follow Schaffrin et al. (2015: 258) who “take a perspective where policy instruments are regarded as the core concept of policy output but also consider the respective content of the instruments focusing on its setting and calibrations”. I shall suggest that policy instruments can be calibrated differently in terms of their explicitness. Explicitness is a dimension or property of policy instruments which refers to their content.

Ideally, the state assesses the desirability of a given regulatory goal—for example, an increase in the availability of donor organs—and then chooses policy instruments that are suitable to achieve this goal (Jordan et al. 2005). However, public policy instruments are more than just means of addressing problems (Bagchus 1998). Public policy instruments also reveal a theorization of the relationship between the government and the governed, and set out the values and interests protected by the state (Lascoumes and Le Galès 2007). Specifically, the use of coercion (at the formal and substantive level) reflects the intensity with which governments attempt to influence target group behaviour—for instance, the consent to donate organs (Schaffrin et al. 2015). Moreover, regulatory density refers to the extent and amount of regulations. The double-edged nature of policy instruments—addressing problems and exerting power—is, however, only partly mirrored in the extant taxonomies.

Existing research emphasizes the density, formal intensity and substantive intensity of policy instruments (Adam et al. 2015; Engeli et al. 2013; Studlar and Burns 2015). Essentially, formal and substantive intensities describe the power relationship between the government and the target population (Bauer and Knill 2014; Hood 2007; Howlett and Cashore 2009; Sager 2009; Schaffrin et al. 2015; Vedung 1998). These existing taxonomies focus on regulatory content either as the governments’ resources at hand, or as the relationship between public authorities and private actors (Schaffrin et al. 2015). However, none of these categories captures the notion that each of these policy instruments can be neutral (objective) or involve a judgement of desirable behaviour (Vedung 1998: 33).

Let me illustrate this. While both Switzerland and Spain had fully non-coercive awareness-raising campaigns for organ donation in 2010, the sermons differed substantially in the message they communicated. Spain issued the slogan “Donate your organs, donate life”. This slogan promoted organ donation very explicitly, yet non-coercively as the desired behaviour. In contrast, the Swiss campaign issued slogans like “Organ donation yes or no”, “I know what I want”, “I decide myself” and “Does somebody else know what you want?” The Swiss government emphasized the procedural aspect of taking a decision regarding organ donation, but without involving any statement about the desired direction of this decision. Against this background, this paper argues that, while highly relevant, neither the notion of “instrument calibrations” nor the dimensions of density or intensity capture the fact that policy instruments differ in the explicitness with which they define a desired direction of behavioural change relative to a given technical policy goal.

The study proposes that existing taxonomies of the substantial intensity of policy instruments be extended to include the dimension of explicitness. It puts forward two main arguments: first, the explicitness dimension captures whether and how policy instruments define an underlying policy goal. On the one hand, policy instruments may or may not indicate a direction of behavioural change (Engeli and Varone 2011). On the other hand, this behavioural change may or may not be attributed positive or negative valence (Kahneman and Tversky 2000; Levin et al. 1998). Not only do information measures or “nudges” display differing degrees of explicitness, but so do more coercive measures (John 2016; Nelson et al. 1997). By decoupling the degree of explicitness from the kind of instrument, this study provides a tool to understand the effects of the intensity of policy instruments, transcending existing notions of “command and control” or “nudging”. Second, indicating a direction of desired behavioural change affects how target groups respond to policy instruments (Cialdini 2003; Johnson et al. 2005; Shafir 2013; Thaler and Sunstein 2008; Weaver 2014). A qualitative comparison of organ donation policies in the two cases of Switzerland and Spain illustrates these claims (Manatschal and Thomann 2011). The explicitness dimension provides additional insights that the degrees of density and restrictiveness of the instruments cannot offer.

The next section first reviews existing perspectives on policy design. I discuss how the new dimension of explicitness can complement these perspectives, how it can be measured, and how it might be linked with policy outcomes. The remainder of the article presents the research design and results, discusses the role of context, values and culture, and exemplifies explicitness in two other policy sectors.

Policy goals are generally recognized as an essential component of policy designs (e.g. Howlett and Cashore 2009). However, the literature features a variety of understandings and measurements of the concept of policy output (Schaffrin et al. 2015). The first, a holistic understanding, stresses that policy output includes both “ends” and “means” at different levels of abstraction (Howlett and Cashore 2009; Howlett 2009). Here, “policy output is composed of three levels of abstraction: the goals of policies, the instruments used to achieve these goals, and the settings in which the policies are applied” (Schaffrin et al. 2015: 258). Policy settings lay out goals through specific requirements. In accordance with these settings, means are calibrated in order to achieve policy goals (Howlett 2009). Different “calibrations” of policy designs provide criteria to classify arrangements of policy tools in view of how means are specified in order to achieve policy goals (Howlett 2009).¹ While useful for thinking about the choices made by policymakers, this approach does not comparatively conceptualize different extents to which settings are specified, instruments are calibrated, and goals are defined in the first place.

According to the second, a comparative approach, we can distinguish the density and intensity of policy instruments across time, sectors and countries. Regulatory density tells us about the formal extent and breadth to which governmental activities address a certain policy area (Bauer and Knill 2014). This dimension captures the amount of government activity in terms of the degree of regulatory penetration, complexity and internal differentiation (Knill et al. 2012). Turning to the content (intensity) of government activity, substantial intensity denotes the restrictiveness of policy output in terms of the level (instrument settings) and scope (personal/substantive/temporal) of state intervention. Restrictiveness refers to the substantial freedom left to policy addressees, or the generosity of the services or resources provided (Bauer and Knill 2014: 33; Knill et al. 2012: 430; also discussed as government intervention, authority, stringency, permissiveness or coerciveness by Hood 2007; Howlett and Rayner 2007; Sager 2009; Schneider and Ingram 1990; Vedung 1998).² Finally, formal intensity refers to factors that affect the probability to effectively achieve regulatory requirements, and define the consequentiality of non-compliant behaviour (enforcement, administrative and procedural capacities) (Adam et al. 2015; Knill et al. 2012).

In accordance with this second approach and other than Howlett and Cashore (2009), I separate policy goals from formal and substantial means and label the latter policy instruments (see Schaffrin et al. 2015).³ Schaffrin et al. (2015) identify five types of policy instruments in the current literature, which they differentiate along the restrictiveness dimension using Vedung’s (1998) tripartite classification of sticks, carrots and sermons. To simplify things, I use Vedung’s trichotomy. The three categories distinguish analytically relevant ways of exercising coercion for the purpose of illustrating the notion of explicitness, though they do not capture the more fine-grained degrees of restrictiveness (Sager 2009) or differentiations of different policy instruments (Hood 2007).

Sticks (regulations) are authoritative restrictions on actions or directives as how to act. They limit the choices available to the individuals. Sticks manifest themselves as regulatory measures (Schaffrin et al. 2015) or “authority” (Hood 2007; Howlett and Rayner 2007). Carrots are economic or non-monetary policy instruments. They make an action easier or more difficult by providing or depriving material resources. Incentives make actions less expensive (e.g. subsidies), whereas disincentives make them more expensive (e.g. taxes)—yet the addressees are not obligated to take the measures involved. Both market-based approaches and public investment are carrots, called “treasure” by Hood (2007) and Howlett and Rayner (2007). Finally, sermons are voluntary appeals by means of information or exhortation, characterized by the absence of obligation. Sermons are soft instruments according to Schaffrin et al. (2015) or “nodality” according to Hood (2007) and Howlett and Rayner (2007).⁴ Carrots and sticks can themselves display different degrees of restrictiveness.

I will now define the two aspects of explicitness: goal definition and valence attribution, in a positive or negative direction. Based on the example of organ donation policy, I propose a stepwise procedure of comparatively measuring positive and negative explicitness, first for sermons, then for carrots and sticks, and discuss the relationship between explicitness, density, intensity and nudging.

In its most basic form, explicitness is a dichotomy: either a desired direction of behavioural change is defined, or it is not. Differing degrees of explicitness need to be defined in the specific context of the policy through three consecutive steps: identifying policy outcomes and target groups, goal definition, and valence attribution. Using the example of organ donation policy, I now propose one possible procedure for measurement using these three steps. Neither goal definition nor valence attribution need necessarily be conceptualized in a dichotomous way.

In the case of organ donation policy, the technical problem is clear: more and more patients in need of transplants are waiting for a donor organ (Flückiger 2010; Matesanz and Dominquez-Gil 2007). The initial pool of potential donors consists of patients with a formal brain death diagnosis or non-heart-beating patients. Once a potential donor has been identified, the next of kin are asked for their consent or refusal to donate organs. A potential donor turns into an effective donor as soon as at least one solid organ has been retrieved. The number of refusals as a share of total requests for organ donation is the biggest obstacle to higher donor rates. The technical solution lies in lowering these refusal rates (e.g. Dunkel 2011; Roels et al. 2010; Shafer et al. 2006; Simillis 2010; Simpkin et al. 2009).

Figure 2 illustrates that implementing agents who put policy instruments “on paper” into practice may adapt these instruments (Lipsky 1980/2010). Hence, most instruments target more than just one actor group—with differing degrees of density, intensity and explicitness. For example, the Spanish government obliges all hospitals to run a quality control programme (stick; obligatory). However, this programme is intended as an informational measure (sermon; voluntary) for hospital staff to improve organ donation processes. I classify policy instruments according to the degree of authority and explicitness exercised on the final target population (Sager 2009: 540), namely the medical staff and next of kin involved in the family decision-making process.

Step 2. Goal definition. Framing theory tells us that the way in which policy instruments articulate particular problems is a decisive factor in terms of how citizens react to these outputs (Kahneman and Tversky 2000; Nelson et al. 1997). Based on the definition of the problem above, we can identify whether an instrument indicates the envisaged direction of behavioural change (explicit) or not (non-explicit). In the case of organ donation, the question is whether an instrument specifies as the intended behavioural change an increase in the consent rates to organ donation.⁶

To ensure comparability, we need to distinguish positive explicitness (in favour of a given goal) and negative explicitness (contra a given goal). For example, comparative studies of AIDS prevention programmes would reveal that South Africa’s campaign under president Thabo Mbeki quite explicitly discouraged effective prevention, whereas Western European countries explicitly emphasized prevention as the intended behaviour. Different instruments might target different, contradictory goals; or some instruments might explicitly target a goal in a positive direction, and others, in a negative direction—resulting in incongruent or contradictory policy designs or instrument mixes (Howlett and Rayner 2007; Howlett 2009). An instrument is more explicit if it aims at positively influencing the donor’s decision, as did the Spanish information campaign. Conversely, a non-explicit instrument represents a neutral or absent official position, as did the Swiss information campaign (Dunkel 2011; Flückiger 2010). In such cases, the government avoids taking a position on the morally contested issue of organ donation (Engeli and Varone 2011).

In reality, policy designs are mixes, which can make it conceptually challenging to determine a single policy goal that allows for the operationalization of “directional explicitness”. However, it is still possible to assess “mere explicitness” levels (independently of directionality) between instruments with different goals, or for a given instrument and its several goals.⁷

Step 3. Valence attribution. An additional question is whether the behavioural change is attributed a valence statement: is there also a positive or negative connotation to the goal? Beyond the mere—and already validating (Mooney 1999)—fact that certain behaviour is explicitly intended, valence attribution refers to the positive or negative meaning that policymakers attach to the policy goal in order to encourage addressees to think and act along particular lines (Chong and Druckman 2007). For example, the message “donate your organs” indicates a direction of behavioural change. “Donate life” additionally attributes a strong positive meaning to this behavioural change. Like goal framing, valence attribution can influence how addressees process information and evaluate behavioural change, for example, whether they perceive this change in terms of gain or loss (Kahneman and Tversky 2000; Levin et al. 1998).

While goal definition establishes the presence or absence of explicitness, valence attribution is especially useful in specifying degrees of explicitness. At the minimum both goal definition and valence attribution allow us to detect non-explicitness (no goal definition and no valence attribution), intermediate explicitness (goal definition but no valence attribution) and high explicitness (goal definition and valence attribution), in either a positive or negative direction relative to a given goal.

Not only do non-coercive sermons define goals more or less explicitly, but so do all types of policy instruments. Specifying the explicitness of carrots and sticks highlights the conceptual and empirical relationship between the different dimensions of density, formal intensity, restrictiveness and explicitness. In order to establish explicitness as a separate dimension of policy instruments, I illustrate this relationship and show that explicitness can nevertheless vary at comparable levels of density and restrictiveness.

Sticks. Sticks, too, often define goals since they prohibit behaviour or encourage the same behaviour; or they can be non-explicit, for example, procedural. A procedural stick could create the duty to perform certain formal steps and assign implementing actors, while delegating the choice of the objective of these steps and their concrete design to the implementing actors (Engeli and Varone 2011).⁸ One example of a non-explicit stick is informed consent: it requires a choice to be made, but leaves the direction of the choice completely to the hands of the donor’s relatives. Conversely, the presumed consent model is a more explicit stick than informed consent. It assumes that everyone is a potential donor unless the relatives opt out. It formulates organ donation as the intended behavioural change, even if it does not impose organ donation (Johnson and Goldstein 2003). The way in which valence attribution manifests itself depends on the specific context of each case. Generally, sticks can attribute positive or negative meaning to a certain behaviour. For example, the size of the fine for fast driving can indicate the importance of traffic accident prevention for a government (Adam et al. 2015).

Carrots. Measuring goal definition is trickier for carrots than for sermons and sticks. Other than the latter, incentives and disincentives inherently indicate some direction of behaviour and would ideally only be fully non-explicit when absent (Schneider and Ingram 1990). Arguably, for carrots, the intentionality of goal definition requires attention (Knoepfel et al. 2011; Vedung 1998). For example, relatives of potential organ donors face non-material costs of separation if the body has to be transported to a retrieval hospital. When designed, this organizational measure was not intended as a disincentive to giving consent, but it clearly needs to be classified as such (Simillis 2010). Still, the instrument itself does not explicitly formulate any goal. Hence, non-explicit carrots neither purposefully encourage nor purposefully discourage a given behaviour. In the field of organ donation, we can think of explicit (positive) incentives as carrots encouraging consent: for example, if hospital donor coordinators receive financial rewards for recruiting donors. Explicit negative incentives would be carrots that purposefully discourage giving consent to donate organs. Carrots often assign a concrete value—material or other—to certain outcomes, implying greater costs or rewards for the addressees.

While intertwined, the dimensions of density, substantive and formal intensity, and explicitness refer to different aspects of the quantity and quality of state action: just as body size and body weight are two distinct but interrelated indicators of someone’s physical appearance. This is not a problem of the explicitness concept; it applies equally to the relationship between density and restrictiveness. For example, more restrictive measures can, but do not have to coincide with a higher density of prohibitions; just as heavier people are often, but do not have to be taller than lighter people (Adam et al. 2015). Equally, however, sometimes regulators add density to regulations to create exemptions, leading to less restrictive regulations (Thomann and Zhelyazkova 2017). It would be wrong to say that density and restrictiveness are independent—just as body size and body weight also correlate—but they are clearly distinct. Similarly, explicitness can sometimes coincide with different degrees of density, formal intensity and the ambiguity of goals and tasks, amongst others. Notwithstanding this interrelationship, explicitness is a separate analytic dimension of policy instruments, measured via two questions: first, does an instrument indicate a desired direction of behavioural change? Second, is positive or negative valence attached to this goal? Neither regulatory density (quantity of regulation), regulatory restrictiveness (leeway left to regulatees) nor formal intensity (efforts to achieve compliance) addresses these questions.

Still, these aspects can sometimes inform explicitness. For example, the amount of educational measures can be one indicator that a government attributes high valence to behavioural change (e.g. awareness about organ donation). Yet, two countries can have a similar density of educational campaigns which still differ in the explicitness with which they encourage organ donation. Clearly, the density of output is not identical to the explicitness of an instrument. Similarly, if a government prescribes a certain behaviour, then it formulates this behaviour as a goal. However, the different models of consenting to organ donation also show that instruments can have the exact same level of restrictiveness and still differ in their explicitness. Hence, restrictiveness does not equal explicitness. Finally, the high cost of non-compliance (e.g. a prison sentence) attaches value to certain behaviours. Despite this, this cost could be equally high for two sticks, one being merely procedural and the other, highly explicit. Obviously, formal intensity is distinct from explicitness.

Given its behavioural dimension, one could conclude that explicitness is simply a nudge. A nudge “is any aspect of the choice architecture that alters people’s behaviour in a predictable way without forbidding any options or significantly changing their economic incentives. To count as a mere nudge, the intervention must be easy and cheap to avoid. Nudges are not mandates. Putting fruit at eye level counts as a nudge. Banning junk food does not” (Thaler and Sunstein 2008: 5). Given this broad definition, “the boundaries for qualifying an intervention as a nudge are clearly a little blurred” (Oliver 2013: 694). Prominent examples of nudges include defaults, information nudges and self-control nudges, such as simplification, the use of social norms, increases in ease and convenience, disclosure, warnings, precommitment strategies, reminders, eliciting implementation intention, and information about the consequences of earlier decisions (see John et al. 2011).

Explicitness is an analytic dimension of every policy instrument, regardless of its coerciveness and form. Conversely, nudges are certain non-coercive instruments that work at the psychological level, and whose goal is typically covert (Oliver 2013) rather than explicit. Explicitness is a precise measurable quality of instruments that does not yet refer to their effects. Conversely, the concept of a nudge is much fuzzier and mainly defines itself through its effect on the regulatees. Clearly, then, explicit policy instruments can—but do not necessarily always—have the form or effects of nudges. For example, many argue that an opt-out solution to giving consent to organ donation is a nudge. Yet equally clearly, explicitness (as an analytic feature of every policy instrument) and nudges (as specific, non-binding instruments informed by behavioural considerations) are not the same thing.

In summary, the added value of assessing the explicitness of policy instruments lies in shedding light on an additional aspect of their content: the degree and quality of goal definition. Similar to the notion of nudging, explicitness helps us understand how policy instruments shape the decision environment of target groups.

The degree of goal ambiguity of a policy influences its implementation (Baier et al. 1986; Matland 1995). Accordingly, explicitness should help us understand the consequences of policy instruments on target groups. The emerging literature on behavioural public policy supports this assertion (John 2016). Policy instruments structure the decision environment of target groups (Shafir 2013). Governments communicate messages of different intensities through policy instruments (Cialdini 2003). The behavioural assumption is that individuals are more likely to take action in support of policy goals if the goals are promoted by government officials as important (Schneider and Ingram 1990: 519). The way in which messages are formulated matters, as oftentimes “people do not have well-defined preferences for (…) particular questions, (…). Rather, people may construct preferences at the time the question is asked” (Johnson et al. 2005: S17). For instance, many people prefer to stick to the default set out by governments rather than making an active choice (Johnson and Goldstein 2003). The intensity of the message can greatly affect how people perceive and respond to policy instruments by changing beliefs about the goal, activating, or altering the priority of existing beliefs (Chong and Druckman 2007; Kahneman and Tversky 2000; Nelson et al. 1997; Weaver 2014).

Prior to 2007, organ donation policy in Switzerland was unregulated at the national level. In contrast, Spain issued a national Transplantation Act as early as 1979 and founded the Organización Nacional de Trasplantes (ONT) in 1989. The Spanish model was exported to various countries all over the globe (Matesanz and Dominquez-Gil 2007; Matesanz 2008; Quigley et al. 2008). In light of fast-growing waiting lists, the Swiss Transplant Act “copied” the basic elements of the Spanish model—hence all the sermons, carrots and sticks are present in both countries. The countries are also comparable in organizational terms. Both organ donation sectors are based on a three-tiered coordination system (national, regional, hospital). Each hospital with an intensive care unit (ICU) has at least one person responsible for donation and transplantation activities. In addition, religious values are translated with similar intensity in both countries (Engeli et al. 2013; Engeli and Varone 2012). This similarity, together with identical levels of restrictiveness, facilitates identifying different levels of explicitness.

Given the novelty of the posited argument, this paper has an explorative and in-depth focus. To illustrate the link between the degree of explicitness of the policy instruments and refusal rates, explanatory typologies are used to complement the qualitative case studies to connect different combinations of values of explanatory variables with particular outcomes (Elman 2005). These patterns are then substantiated through the case studies as well as through previous empirical evidence from other studies.

Data sources and measurement. Information regarding policy instruments and their effects were collected in 2010 for the Federal Office of Public Health (FOPH) through a qualitative content analysis of primary and secondary literature, and semi-structured interviews with a total of 28 experts and practitioners in both countries (Manatschal and Thomann 2011; see Table 2). Data on refusal rates stem from the national quality monitoring programmes. The Spanish sample comprises all hospitals in 2009 (71 big, 68 small hospitals), whereas data are only available for 50% of the Swiss hospitals. The participant hospitals are considered representative (Manatschal and Thomann 2011: 33, 64). Finally, the survey question “would you be willing to donate one of your organs to an organ donation service immediately after your death?” serves as a proxy for values on organ donation. In Spain, a special Eurobarometer was conducted in 2006 (N = 749) (Eurobarometer 2007). In Switzerland, a representative health survey was carried out in 2007 (N = 13304) (Balthasar and Müller 2008). Value polarization denotes the difference between the share of positive answers and the share of negative answers of all respondents expressing an opinion (excluding “don’t knows”).

Table 1, which presents the substantial content of the two countries’ regulatory regimes, offers two important insights.

First, both countries have adopted policy instruments with similar degrees of coerciveness, which hence cannot account for the striking differences in refusal rates. Second, Spanish hospitals have the most explicit instrument logic, followed by the Swiss Latin-speaking ones, and finally the German-speaking hospitals. In line with Howlett and Cashore’s (2009) notion of design logics, this pattern begins with the objectives defined by the national transplant acts: the Spanish law explicitly aims to encourage voluntary, altruistic and non-remunerative organ donation to ensure that every citizen in need of a donor organ has optimal chances of receiving one. Conversely, the Swiss law simply seeks to regulate the processes of organ donation.