Exposition: The Regulation of Recycling

In dealing with the recycling challenge, regulators have used a diverse mix of strategies, including economic instruments (deposits and taxes), criminal sanctions (fines), technology standards, and voluntary programs.Footnote ² Economic instruments, in particular deposit-refund systems, are used widely in the recycling context. Such programs were introduced in various countries including the United States, Denmark, Germany, Israel, and Australia (Reference Croci and CrociCroci et al. 2005; EPA 2001: 57–64; Reference GardinerGardiner et al. 2009; Reference GrothGroth 2008). Taxes are also used to encourage recycling. In Switzerland, most of the cantons introduced a variable charge on bags used for waste collection. The household collection system is supported by a “bring system” for recyclable materials (Reference FogartyFogarty 2008: 35).

There are also various examples of waste programs using a combination of “command and control” techniques and technology standards. For example, during 2011 a high-tech trash-collection system was implemented in Cleveland, Ohio, with radio frequency identification chips and bar codes embedded in curbside trash and recycling carts that enable city workers to monitor how often residents roll the carts to the curb for collection. If a chip shows that a recycling cart has not been brought to the curb in weeks, a “trash supervisor” sorts through the trash for recyclables. If the trash cart constitutes more than 10 percent recyclable material, the resident can be fined up to $100.Footnote ³ Another example of the use of fines to encourage recycling is the New York City 2010 Electronic Equipment Recycling and Reuse Act, which introduced a disposal ban on discarded electronics, together with a $100 fine on citizens who violate it.Footnote ⁴ Voluntary mechanisms are also widely used. In Israel, the recycling of bottles larger than 1.5 liters is based on a voluntary program.Footnote ⁵ In the United States, the recycling of “E-waste” is still based largely on voluntary mechanisms (Reference McCarthyMcCarthy 2002; Reference WernerWerner et al. 1995).

The Framing Effect of Regulatory Variance

Framing refers to the behavioral and attitudinal effects associated with the form of presentation, rather than the substantive meaning of the legal norm. As such, the framing phenomenon reflects a departure from the predictions of expected utility theory. The idea of framing has been explored in diverse contexts, including the framing of jury instructions (Reference McCafferyMcCaffery et al. 1995) and tax instruments (Reference McCaffery and BaronMcCaffery & Baron 2006) and the study of framing as a de-biasing tool (Reference Jolls and SunsteinJolls & Sunstein 2006).

Applying the theoretical insights developed by the framing literature to the recycling dilemma highlights several differences between the deposit and fine instruments, and these differences could have important implications for the way the instruments are perceived. First, the literature points out that the framing of outcomes as losses or as gains generally affects people's responses (Reference Jolls and SunsteinJolls & Sunstein 2006: 205). Furthermore, loss aversion is generally reference dependent: people tend to be significantly more averse to losses (relative to the status quo) than they are attracted by gains; that is, losses with respect to a current endowment loom larger than gains (Reference Ariely, Huber and WertenbrochAriely et al. 2005; Reference Novemsky and KahnemanNovemsky & Kahneman 2005). Conceiving of loss aversion as reference dependent suggests that the deposit instrument would have stronger public appeal than the fine because the two instruments construct different reference points (or asymmetric futures). Although both the deposit and the fine involve immediate or expected payments (i.e., losses) and have a similar economic effect of putting a price tag on nonrecycling, the fine emphasizes the loss, whereas the deposit emphasizes the gain received by returning the bottle. McCaffery and Baron's claim about penalty aversion provides further support for this argument. Penalty aversion refers to situations in which people prefer bonuses over penalties, when the policies differ only in the form of presentation (Reference McCaffery and BaronMcCaffery & Baron 2004).Footnote ⁶ The fine scenario parallels the penalty in McCaffery and Baron's setting, with the reference point being the“pre-fine” baseline. By contrast, the deposit scenario parallels the bonus setting, with a “post-deposit” as the reference point.Footnote ⁷

Another possible explanation for this prediction is that the deposit seems to give people a stronger sense of control over the outcome. Whereas the fine presupposes the existence of a third party, the enforcement authority, the deposit leaves the decision to the individual, thus contributing to people's perceived self-determination (Reference De YoungDe Young 1993). Finally, the distinction between fines and deposits also brings to the fore the question of people's attitudes toward uncertainty, especially uncertainty that involves low probabilities (Reference Tversky and FoxTversky & Fox 1995). The preference for the deposit over the fine may reflect the probabilistic nature of the fine. It could be a reflection of probability neglect, which occurs when people are highly sensitive to the outcome and are relatively indifferent to significant variations in its likelihood (Reference SunsteinSunstein 2002: 62–63, Reference Sunstein2003: 1298).Footnote ⁸ This neglect means that the expected value of the fine is perceived to be larger than that of the deposit, and given the large nominal gap between the fine and the deposit, this is likely to generate a preference for the deposit. Based on the foregoing theoretical reflections, we hypothesize that deposit schemes enjoy greater public support than fine schemes.

Crowding Out and the Optimal Size of Legal Incentives

The crowding out hypothesis is based on a distinction between “extrinsic” and “intrinsic” motivations (Reference DeciDeci 2000). Extrinsic motivation refers to performance of an activity in order to attain some separable outcome (Reference Ryan and DeciRyan & Deci 2000: 71). By contrast, intrinsic motivation does not depend on external incentives and is therefore self-sustained (Reference Frey and JegenFrey & Jegen 1999: 4; Reference Osterloh, Frey and FrostOsterloh et al. 2001: 233). The literature distinguishes between two types of intrinsic motivation: enjoyment-based and obligation-based (Reference Osterloh, Frey and FrostOsterloh, Frey, & Frost 2001: 233; Reference Ryan and DeciRyan & Deci 2000: 70). Obligation-based intrinsic motivation, which is more applicable to recycling, reflects a sense of moral or civic duty that is not driven by instrumental considerations (e.g., commitment to environmental ethics) (Osterloh, Frey, & Reference Osterloh, Frey and FrostFrost 2001: 233–234; Reference ThØgersenThøgersen 2003: 200), and is related to what has been defined in the psychological study of compliance as morally driven legal compliance.Footnote ⁹ A vast body of literature demonstrates that moral sentiments (intrinsic motivation) and extrinsic incentives (economic or others) are not necessarily synergetic (Reference BowlesBowles 2008; Reference Frey and StutzerFrey & Stutzer 2006). By suggesting that in some cases raising monetary incentives reduces rather than increases socially desirable behavior, the crowding out effect constitutes an intriguing anomaly that works against the classic economic behavioral model (Reference Frey and JegenFrey & Jegen 2001; Reference Mellstrom and JohannessonMellstrom & Johannesson 2008). The tension between intrinsic and extrinsic motivation has been shown to be relevant to the question of compliance inx different regulatory contexts, including tax evasion (Reference WenzelWenzel 2005) and labor law (Reference Feldman and Lobelwhistle blowing) (Feldman & Lobel 2010).

The crowding out literature suggests several behavioral explanations for the potential erosion of internal motivation by external incentives. First, cognitive evaluation theory (CET) stresses the need for individuals to feel competent and in control with respect to a particular task (Reference Festre and GarrousteFestre & Garrouste 2007: 9–10; Reference Ryan and DeciRyan & Deci 2000; Reference Weibel, Rost and OsterlohWeibel et al. 2007: 7–8). External incentives may erode people's motivation by undermining their conceptions of competence and autonomy. Command and control regulation, which leaves little room for discretion, negatively influences the self-conception of the regulated individuals as having reliable decision-making ability and damages their sense of autonomy.Footnote ¹⁰ Bruno Frey explores this theoretical path in several empirical studies (Reference FreyFrey 1999; Reference Frey and JegenFrey & Jegen 2001).

A second explanation is based on the “overjustification” theory, which suggests that when people derive pleasure from an action per se, and would have taken that action even in the absence of external rewards, the introduction of explicit incentives may “overjustify” the activity and undermine the person's self-interest in this behavior (Reference BowlesBowles 2008: 1607; Reference ThØgersenThøgersen 1994). A third explanation focuses on the framing effect of external intervention. Intervention in the form of economic incentives can change people's perceptions of the social contexts in which their actions are embedded, thus shifting the reference frame from the moral plane to the economic one. This frame shifting can crowd out intrinsic motivations by corrupting a purely social act with economic considerations (Reference AnikAnik et al. 2009). Such crowding out can be manifested either in a refusal to contribute to a public good without pay or in an increase in the level of asocial behavior, as suggested by Gneezy and Rustichini's “fine is a price” hypothesis (Reference Gneezy and RustichiniGneezy & Rustichini 2000). Gneezy and Rustichini argue that the introduction of fines can be interpreted as placing a price tag on asocial behavior, thus leading—contrary to the traditional deterrence model—to an increase in the asocial behavior. Another potential explanation associates the crowding out effect with an image motivation—i.e., a desire to signal altruism. Reference Bénabou and TiroleBénabou and Tirole (2006) argue, for example, that if blood donors are paid, the value of blood donation as a sign of generosity is weakened, leading to reduced motivation to donate blood. It should be noted that image motivation, although nonmaterialistic, is extrinsic (Reference Ariely, Bracha and MeierAriely et al. 2009; Reference Bergstrom, Garratt and Sheehan-ConnorBergstrom et al. 2009).

Finally, a recent paper by Reference Goeschl and PerinoGoeschl and Perino (2009: 30) offers a different account of the crowding out phenomenon based on affect-based rather than cognitive-driven behavior. Goeschl and Perino argue that different forms of regulation may differ by how well they map into the subjects' moral categories. A mismatch between the subjects' moral views and a certain regulatory instrument can generate negative feelings (associated, for example, with unfair treatment) and therefore lead to behavior reversal (Reference Koenigs and TranelKoenigs & Tranel 2007; Reference Pillutla and MurnighanPillutla & Murnighan 1996).

If we take seriously both the idea that intrinsic and extrinsic compliance motivations are not necessarily synergetic and the idea that society is comprised of people with varied levels of intrinsic motivation, the question of the overall effect of a certain regulatory initiative becomes sensitive to the distribution of different “types” of individuals within the targeted community. This social variability raises the question of how different “types” of people react to different regulatory incentives. Replacing the assumption of uniformity with one of heterogeneity is therefore critical for the design of optimal social policies. This view challenges the conventional one-size-fits-all model of regulation and suggests a different regulatory paradigm, “differentiated regulation,” which aspires to match different regulatory strategies to different types of individuals.

The crowding out paradigm also raises the question of the optimal size of monetary regulatory incentives. Reference FreyBruno Frey (1999) suggests, in this context, the “U-shape hypothesis,” arguing that the crowding out effect has its strongest “bite” at some intermediate level. This hypothesis is based on the potentially contradictory impact of the crowding out effect and the price effect. First, the crowding out effect increases with the size of the monetary incentive. The relationship between the size of the incentive and the crowding out effect is not linear, however. Low monetary incentives may have a positive expressive message, which is diluted and becomes negative as the incentive is increased. Furthermore, it seems reasonable to assume that when the incentives reach a certain threshold, the crowding out effect ceases to increase. The crowding out effect is countered by the price effect, which increases as the monetary incentive goes up. For example, in the case of environmental taxes, a low tax rate does not generate a perception of control, and this leads the crowding out effect to be small or to turn into a “crowding in” effect owing to the expressive aspects of the tax. The relative price effect of a low tax rate, however, is small. When taxes are raised to a high level, they generate a strong price signal (polluting becomes very costly) and the relative price effect is likely to dominate the crowding out effect.

An intermediately sized environmental tax induces undesirable effects on both accounts. Environmental morale is crowded out as individuals' self-determination becomes noticeably impaired, and at the same time the tax rates are insufficient to induce pro-environmental behavior due to extrinsic motivation. The increase in tax levels can also shift the frame of observation: it can legitimize the antienvironmental behavior and attenuate its negative moral image by offering a price tag for misbehavior (repositioning the behavior as being governed by market rationality) (Reference BowlesBowles 2008: 1606). The question whether an environmental tax rate is to be considered by the social agent as “low” or “high” can be answered only through specific, contextualized analysis. Our study examines this issue by presenting subjects with two levels of fines and deposits (high and low).

Based on the crowding out hypothesis, we project that the introduction of monetary incentives produces the largest reduction in recycling intentions among subjects with higher levels of intrinsic motivation.

The notion of framing suggests, however, that different monetary instruments (in our experiment, fine or deposit) can generate varied crowding out effects. In other words, we expect to find an interaction among the type of monetary incentive (fine or deposit), the level of intrinsic motivation, and the behavioral effect of the regulatory intervention on recycling intentions. There is support for two contrasting predictions regarding the direction of this interaction. On one hand, because fines are more coercive than deposits, they are more likely to enhance the adverse effect of the economic incentive on subjects' autonomy, leading to a stronger crowding out effect. On the other hand, deposits may convey a stronger message of marketization or commodification, generating a stronger crowding out effect because of corruption of the moral motivation for recycling with economic considerations.

Environmental Earmarking as Framing

An important insight of both the framing and crowding out literature is that words matter. Our research design sought to examine the extent to which “environmental” earmarking (ensuring that the money collected under a deposit/fine framework is transferred to a dedicated environmental trust fundFootnote ¹¹ rather than to the general state budget) influences people's attitudes toward the regulatory intervention.

The economic approach to earmarking depends upon the structure of the earmarking system. In the context of tax payments, there is theoretical support for earmarking when there is a close link between the tax payment and the use of the revenues to finance additional expenditures. Such benefit-related earmarking reveals taxpayer preferences for public services and sends a clear message about the amount of public service that should be provided (Reference Bird and JunBird & Jun 2005: 8). Earmarking is considered inefficient, however, when there is no direct link between those who pay the tax and those who benefit from the expenditure, or in cases in which the level of expenditure is not related to the actual tax revenue (Reference Bird and JunBird & Jun 2005: 3). Economists have also analyzed earmarking from a political economy perspective. The government may use earmarking to legitimize a tax increase even if the new tax is channeled into the general budget and has nothing to do with how much is spent on the earmarked target. In such cases, the earmarking is chimerical. Earmarking can also be motivated by “rent-seeking” groups looking to secure portions of the budget to serve their causes (Reference Bird and JunBird & Jun 2005: 11–12; Reference Kimenyi, Lee and TollisonKimenyi et al. 1990).

There is, however, relatively little research on the question of how earmarking influences people's attitudes and whether such influence varies for different regulatory instruments. In our study we examined the differences in the (expressive) effect of earmarking on the perceived fairness of the deposit/fine schemes. Fairness was shown to contribute to the efficacy of the regulatory scheme through its effect on legitimacy (Reference Gibson, Caldeira and SpenceGibson et al. 2005; Reference MearesMeares 2000; Reference RainesRaines 2003). In the fine context, we hypothesized that the addition of earmarking would have a positive effect on the perceived fairness of the regulatory program, by moderating people's basic aversion to the fine, as noted by Reference McCaffery and BaronMcCaffery and Baron (2006). In the deposit context we hypothesized a contrary effect: the introduction of earmarking may be seen as undermining the ability of recyclers to make a complete contribution to the environment by forcing recyclers to collect the deposit (which would otherwise go to the environmental fund), leading to a reduction in the perceived fairness of the deposit program. Overall this leads to the following hypothesis:

Regulation in a Multi-institutional Context

We believe that exploring the instrument-choice dilemma requires sensitivity to the institutional context. One of the hallmarks of modern environmental regulation is the proliferation of private governance programs associated with diverse institutional sources. The institutionalist approach highlights the intricate linkage between individual decisionmaking and the institutional structure in which this action is embedded, considering the institutional context as an independent causal variable (Reference Ghoshal and MoranGhoshal & Moran 1995: 15; Perez, Amichai-Hamburger, & Reference Perez, Amichai-Hamburger and ShterentalShterental 2009). To explore the extent to which the institutional setting influences participants' attitudes, we added two institutional settings, involving organizations with more than ten employees and academic institutions.

Using the institutional perspective as a starting point, we focus on two strands of research. First, there is a wide body of literature, starting with the writings of Robert Ellickson and Elinor Ostrom, demonstrating the capacity of close-knit groups to develop highly effective webs of social norms (Reference NeeNee 1998; Reference OstromOstrom 1999: 506–507; Reference VandenberghVandenbergh 2005). Most of these studies focus on communities that are not formally organized, from city “blocks” (Reference EllicksonEllickson 1998) to Maine lobster fisheries (Reference OstromOstrom 1999: 528–529). Similar social capabilities, which can support the generation of social norms, exist in formal organizations (Reference ArielyAriely 2008: 80–83; Perez, Amichai-Hamburger, & Reference Perez, Amichai-Hamburger and ShterentalShterental 2009). Our study explores one aspect of this question by examining the self-reported efficacy of recycling norms across different institutional settings. In particular, we wanted to examine the effectiveness of organizational norms compared with that of state norms. We hypothesized that because of the lower scale and greater cohesiveness of organizations than of the state, the fine and ethical code instruments are likely to be more effective in encouraging recycling behavior in the organizational setting than in the state setting.

Our research considers the increasing role of organizations as norm setters (Reference LockeLocke et al. 2007; Reference Long and DriscollLong & Driscoll 2008; Reference RilesRiles 2008) and also explores a second question focusing on the limits of organizational norm setting as far as (internal) social legitimacy is concerned. In particular, we wanted to examine the extent to which organizations can use forms of legal intervention that tend to be associated with the state (fines and deposits) in a field that lies outside the usual reach of organizational norms. We hypothesized that such intervention can generate resentment among members of the organization (employees and students). This inquiry explores the boundaries of organizational authority (Reference Santos and EisenhardtSantos & Eisenhardt 2005), focusing in particular on the ways in which such boundaries are reflected in members' attitudes toward organizational norms. We argue that individuals would be less likely to accept controlling measures directed to protect public (environmental) resources (e.g., fines and deposits) within the organizational context than measures focusing on the organization's assets and economic interests.

Participants

The study is based on a panel of 1,842 subjects, purchased from a leading Israeli survey firm.Footnote ¹² A representative sample of the adult Jewish population in Israel was assembled by including participants in the panel based on their demographic background.Footnote ¹³ Forty-two subjects were excluded from the research because of their low degree of understanding of the legal settingFootnote ¹⁴ (N = 34) or missing values on most variables (N = 8). The final sample, therefore, included 1,800 participants (865 males and 935 females) with a mean age of 37.15 years (SD = 14.05 years), 63.4 percent of whom reported having completed some form of higher education (above high school).Footnote ¹⁵ The data were collected using a Web-based survey platform.Footnote ¹⁶

Experimental Design

The procedure was based on an experimental survey approach that allowed us to compare various regulatory mechanisms and to study a multi-instrument regulatory environment.Footnote ¹⁷ Before introducing the participants to the manipulation, we evaluated their degree of environmental commitment, pro-environmental behavior, and willingness to bear personal costs for recycling. The environmental commitment measure was based on three questions that attempted to measure people's attitudes toward the environment. A sample item was as follows: “To what extent are you worried about the situation of the environment?”Footnote ¹⁸ The pro-environmental behavior measure was similarly based on three questions, seeking to assess the extent of people's engagement in pro-environmental behavior. A sample item was as follows: “To what extent do you recycle newspapers and other printed matter?” The willingness to bear personal costs is a related measure that sought to examine the extent to which people's recycling intentions are sensitive to the effort associated with that activity; in our context, it was measured by the distance one must walk to the recycling point.

The survey was based on the following basic hypothetical scenario: Participants were told that, “… we are conducting a survey that seeks to explore recycling behavior among the Israeli population.” They were further told that we would present them with details of the proposed laws and would thank them if they agreed to answer some questions about it. The new laws targeted family-size plastic bottles (1.5 L) that are exempt from current recycling regulation under Israeli law.

The main methodology of the study is based on a between-subject design, distinguishing among three regulatory techniques used to encourage recycling activity: deposit, fine, and ethical code (an additional control group read a scenario that did not contain a description of the law). To further refine the instrument choice study we added two manipulations: the first referred to the size of the monetary mechanisms (high and low fine and deposit), and the second referred to the institutional source and context of the regulatory mechanism. In addition, we supplemented the between-subject design with a within-subject design, in which participants were asked to revisit their responses to certain questions based on new information provided to them.

Below is a detailed description of the three regulatory models that were introduced to the participants as possible “new laws” for encouraging recycling.Footnote ¹⁹

1. 1. Deposit: The law imposes a mandatory deposit on 1.5 L plastic bottles. The consumer can get the deposit money back by returning the bottles to recycling points located in various business outlets (e.g., supermarkets). The deposit was framed as either high (1 NIS) or low (0.25 NIS)Footnote ²⁰ (High Deposit and Low Deposit, respectively).

2. 2. Fine: The second model of regulatory intervention was based on the imposition of a fine on the dumping of 1.5 L bottles into regular waste bins: “The new law requires citizens to separate 1.5 L plastic from the rest of their household waste and throw them into separate recycling containers. According to the law, the number of recycling containers will be increased to improve their accessibility. Special inspectors from the environmental protection ministry will have the authority to impose a fine of 250/1,000 NIS on citizens who fail to comply with the requirement. Based on past experience, it is estimated that the likelihood of detection is 1 per 1000 cases.”Footnote ²¹ The fine was framed as either high (1,000 NIS) or low (250 NIS) (High Fine and Low Fine, respectively). Given the probability of detection (1/1,000), the expected value of the fine was identical with the deposit (1 or 0.25 NIS).

3. 3. Ethical Code: The third generic model was based on an environmental ethical code. This instrument was described as follows: “Recently, some leading Israeli environmental NGOsFootnote ²² have published an environmental ethical code seeking to encourage recycling. The code states, among others, that: ‘1.5 L plastic bottles ought to be recycled, using separate recycling containers located in the streets.’ Based on the code, the environmental NGOs have collaborated with local municipalities and the Ministry of the Environment to increase the number of recycling containers in order to improve their accessibility.” Unlike the foregoing two models, this instrument does not use monetary incentives (henceforth: Ethical Code).

4. 4. Control Group: Finally, participants in the control group were not presented with any legal instrument (henceforth: Control Group). After measuring their level of environmental commitment and willingness to bear costs, participants in the control group answered questions about their appraisal of recycling behavior.

Source and Context of the Law

The foregoing basic scenarios yielded six subgroups: high deposit, low deposit, high fine, low fine, ethical code, and control group. A common feature of the four fine/deposit conditions was the assumption that the source of the legal instrument was the state. But in light of the increasing importance of nonstate sources in the regulatory arena, we wanted to examine the influence of this variable as well, and therefore, we added a variable reflecting the possibility that the source of the law would be an organization. We distinguished between two organizational contexts: an organization (workplace) and a university, targeting two population types, employees and students, respectively. Within each institutional context, we presented participants with three different instruments: High Deposit, High Fine, and Ethical Code.

In the context of the organization (workplace) we used the following text to introduce the legal instrument: “The management of the organization in which you are employed considers the possibility of adopting a policy to encourage recycling. It has been proposed that a deposit of 1 NIS will be collected for every bottle sold in the organization's cafeteria. This deposit will be returned to the employee upon returning the bottle to a recycling point located throughout the workplace.” In addition to the high-deposit mechanism, the general organization vignettes included two other scenarios: high fine and ethical code (using a similar text). These three scenarios are designated deposit-org, fine-org, and ethical-org.Footnote ²³ The university scenario included similarly worded vignettes, generating three parallel conditions: high deposit, high fine, and ethical code, which were designated as deposit-edu, fine-edu, and ethical-edu.

Measures

The various manipulations described above generated 12 subgroups (conditions).Footnote ²⁴ To ensure the randomness of subgroup composition, each participant had an equal chance of being allocated to each of the subgroups.Footnote ²⁵ In addition, we have confirmed that no significant differences were found in the main demographic factors.

After reading their allocated scenarios, participants were presented with a series of questions about their attitudes toward the “new law,” their views on the act of recycling from several perspectives, and their perception of the efficacy of recycling. Based on theoretical considerations and in accordance with statistical testing of the fitness of the scale, the items were aggregated into scales that served as dependent variables.

We first examined the extent to which people thought that the legal mechanism presented to them should be adopted by the state (or the organization/academic institution). This variable was termed the desirability of the law. The next factor focused on the perceived effect of the proposed legal framework on the level of recycling. This factor was divided into the (projected) effect of the targeted instrument on the behavior of others (henceforth, general efficacy) and the (projected) effect on one's own recycling practices (henceforth, personal efficacy). Finally, we presented people with a series of questions focusing on the influence of the law on their level of social appraisal of recycling behavior. Social appraisal of recycling behavior was measured by three items: (a) evaluation of recycling behavior in general (to what extent they considered recycling a deserving practice), (b) concrete evaluation (how they would consider, or appreciate, an individual undertaking recycling action), and (c) willingness of the participant to make a critical comment to a neighbor (or colleague or student) who “breaches” the recycling norm.

In addition, we introduced participants to the possibility of earmarking the money collected in the context of the eight deposit/fine scenarios for a special environmental fund, and examined the effect of the new data on participants' evaluation of the legal instrument. In particular, we examined the extent to which such earmarking would influence people's perception of the fairness of the legal scheme. Finally, we measured and accounted for various demographic factors (such as gender, age, income, and level of education).

General Observations

We started by investigating the way in which different legal frameworks influenced people's attitudes toward the law in terms of its desirability, general and personal efficacy, and social appraisal. At this stage, we focused on five legal settings: High Deposit, Low Deposit, High Fine, Low Fine, and Ethical Code.Footnote ²⁶ To examine how each of the outcome measures changed according to these legal settings, we conducted a series of one-way analyses of covariance (ANCOVAs or multivariate analyses of covariance for multiple subscales) in which the outcome measures were dependent variables and the legal instruments were independent factors. The analyses of covariance were followed by Bonferroni post hoc analyses.Footnote ²⁷

First, we looked into the way in which the different legal instruments affected participants' views of the desirability of each legal instrument. A univariate analysis of covariance (ANCOVA) confirmed that the different subgroups differed significantly in their perceived desirability (F(4,732) = 8.15, p < 0.001, η² = 0.04). As indicated by the post hoc analyses, the two high (M = 55.87, SD = 18.97) and low (M = 56.49, SD = 18.60) deposit instruments, as well as the Ethical Code program (M = 53.87, SD = 12.39), were perceived to be significantly more desirable than the two fine schemes: high (M = 46.82, SD = 21.06) and low (M = 48.11, SD = 21.75). These results are consistent with Hypothesis 1.Footnote ²⁸

At the second stage we considered the question of efficacy. The multivariate analysis of covariance (MANCOVA) indicated that the distinct subgroups differed significantly on the measures of efficacy (F(8,1460) = 2.16, p < 0.05, η² = 0.01). The subsequent univariate analyses indicated that the effect of the legal instrument was marginally significant for the expected change in recycling at the state level (general efficacy) (F(4,731) = 2.13, p = 0.08, η² = 0.01), but not for the expected change in recycling at the personal level (personal efficacy) (F(4,731) = 1.56, p > 0.1, η² = 0.01). With regard to general efficacy, only the high deposit condition (M = 77.08, SD = 20.86) was found to be significantly different from the ethical code condition (M = 67.80, SD = 20.14), as indicated by the post hoc analysis.Footnote ²⁹

Finally, we considered the issue of social appraisal.Footnote ³⁰ The MANCOVA analysis revealed that, overall, the legal instrument subgroups differed significantly on the variables of social appraisal (F(15,2421) = 4.61, p < 0.001, η² = 0.03). As indicated by the subsequent univariate analyses, the effect was significant for all tested variables: evaluation of recycling behavior (F(5,879) = 3.46, p < 0.01, η² = 0.02), appreciation of people who recycle (F(5,879) = 3.54, p < 0.01, η² = 0.02), and willingness to comment to others (F(5,879) = 4.70, p < 0.001, η² = 0.03). The effect of the instruments' variance on the social appraisal factor was rather complex. Specifically, participants in the control and ethical code conditions showed higher appreciation of both recycling in general and people who recycle than did participants in the fine and deposit conditions (both high and low).Footnote ³¹ Whereas participants in the control condition had the highest appreciation for recycling and recyclers, they had significantly lower willingness to scold others who do not recycle (M = 42.26, SD = 27.21) than did participants in each of the other conditions (as revealed by the post hoc analyses).Footnote ³²

The Interplay between People's Intrinsic Motivation and the Choice of Legal Instrument

We argue above that the existence of individuals with varied levels of intrinsic environmental motivation could yield significant differences in the way people react to different legal instruments. As elaborated in the Method section, we used three measures to capture people's intrinsic motivation. To examine the relationship between the measures we conducted a series of hierarchical regressions. First, we conducted two regressions to examine whether participants' levels of environmental attitudes predicted their pro-environmental behavior and their sensitivity to distance from the recycling bin.Footnote ³³ As expected, environmental attitudes were strongly related to self-reported pro-environmental behavior (R² = 19.4 percent, β = 0.45, p < 0.01). Environmental attitudes were also negatively related to sensitivity to distance, as participants who valued the environment to a larger degree were less sensitive to their distance from the recycling bin (R² = 5.2 percent, β = −.23, p < 0.001). Both findings highlight the linkage between environmental values and pro-environmental behavior, consistent with the findings of the psychological literature on pro-environmental behavior (Reference KahnKahn 2007; Reference Nordlund and GarvillNordlund & Garvill 2002).Footnote ³⁴ Based on additional regression, we also found a negative association between pro-environmental behavior and sensitivity to distance (R² = 15.0 percent, β = −0.40, p < 0.001). Participants who reported acting in an environmentally friendly way were also less sensitive to their distance from the recycling containers.

Because environmental attitudes, pro-environmental behaviors, and sensitivity to distance were significantly interrelated, we decided to focus on a single measure, sensitivity to distance, as a proxy for intrinsic environmental motivation in order to avoid repetition. We believe that sensitivity to distance is a better proxy for intrinsic motivation in the context of this study than the other two measures are, because it focuses on the personal effort associated with recycling rather than on values or more general forms of pro-environmental behavior. Other studies demonstrate the critical role of perceived inconvenience in people's willingness to participate in recycling programs (Reference CarlsonCarlson 2001; Reference HansmannHansmann et al. 2006). Therefore, we report our findings based only on the sensitivity to distance factor. People with low sensitivity to distance are referred to as people with high intrinsic motivation; people with high sensitivity to distance are referred to as people with low intrinsic motivation.

To examine the combined effects of intrinsic motivation and legal instrument on desirability of the law, we conducted a two-way univariate analysis of covariance (ANCOVA), in which desirability was a dependent variable whereas the legal instrument and intrinsic motivationFootnote ³⁵ were independent factors. We found a significant main effect for intrinsic motivation (F(1,727) = 15.39, p < 0.001, η² = 0.02): across all conditions people with higher levels of intrinsic motivation considered the new recycling law more desirable (M = 55.29, SD = 18.65) than did the low intrinsic motivation group (M = 48.93, SD = 19.32). We also found a marginally significant interaction between the type of the legal instrument to which the respondents were exposed and the respondents' levels of intrinsic environmental motivation (F(4,727) = 2.27, p = 0.06, η² = 0.01). Among the participants with low intrinsic motivation, the legal instrument had a significant effect on the measure of desirability (F(4,727) = 8.03, p < 0.001, η² = 0.04),Footnote ³⁶ with both deposit scenarios generating significantly higher desirability than the two fine scenarios.Footnote ³⁷ But in the highly motivated group the legal instrument had no significant effect on desirability (F(4,727) = 1.95, p = 0.10, η² = 0.01). This finding suggests a refinement of Hypothesis 1, because it indicates that the regulatory variance influences only individuals with low-intrinsic motivation. The desirability of the legal instruments as a function of intrinsic motivation and of the legal instrument is presented in Figure 1.

Figure 1. Adjusted mean scores for the desirability of the law as a function of the legal instrument and of intrinsic motivation (N = 738).

Intrinsic Motivation and the Effects of the Different Regulatory Instruments

To examine the combined effects of intrinsic motivation and legal instrument on perceived efficacy of the law, we conducted a two-way multivariate analysis of covariance (MANCOVA), in which general and personal efficacy were designated as dependent variables, and the legal instrument and intrinsic motivation were designated as independent factors. The multivariate analysis yielded a significant interaction between the level of intrinsic motivation and the type of instrument (F(8,1450) = 3.42, p < 0.001, η² = 0.02). Based on univariate tests, we found that the interaction was significant only for the perceived efficacy of the law at the personal level (F(4,726) = 6.27, p < 0.001, η² = 0.03), but not for the perceived efficacy at the state level (F(4,726) = 1.29, p > 0.05, η² = 0.01).

Overall, the interplay between the legal instrument and intrinsic motivation, with regard to personal efficacy, was consistent with Hypothesis 2A.Footnote ³⁸ We found that among participants with high intrinsic motivation, the Low Fine instrument generated a higher effect than did the High Deposit, High Fine, and Low Deposit instruments. Ethical Code was ranked second to Low Fine. By contrast, among participants with low intrinsic motivation, the High Deposit instrument generated a greater effect than did the Ethical Code, Low Fine, and Low Deposit scenarios. High Fine was ranked second to High Deposit. As shown in Figure 2, all of the regulatory measures were associated with some increase in the level of self-recycling (a value higher than 50, which designates a lack of influence of the legal measure on personal recycling).Footnote ³⁹ The difference between the conditions was therefore in the strength of the effect.

Figure 2. The adjusted mean scores for the personal efficacy of the law as a function of the legal instrument and of intrinsic motivation (N = 737).

The Interplay among Intrinsic Motivation, Type of Instrument, and Level of Incentive in Determining Personal Efficacy

Consistent with Hypothesis 2B(1), we found differences between the fine and deposit conditions with respect to the interaction between the size of the legal incentive and the level of intrinsic motivation in predicting the behavioral effect of the regulatory intervention.

We conducted two-way univariate covariance analyses (ANCOVAs) separately for the fine and deposit subgroups, in which the personal efficacy was designated as the dependent variable, and the magnitudes of the incentive (high, low) and intrinsic motivation were designated as the independent factors. Within the fine condition we found a significant interaction between the magnitude of the fine and the level of intrinsic motivation, based on ANCOVA analysis (F(1,288) = 12.73, p < 0.001, η² = 0.04). The effect of the magnitude of the fine was reversed for individuals with high and low intrinsic motivation. In the high intrinsic motivation subgroup, Low Fine generated a greater effect on the measure of personal efficacy than did High Fine (F(1,288) = 7.87, p < 0.01, η² = 0.03)Footnote ⁴⁰. By contrast, in the low intrinsic motivation subgroup, High Fine generated a greater effect than did Low Fine (F(1,288) = 4.94, p < 0.05, η² = 0.02). The results of this interaction effect are presented in Figure 3.

Figure 3. Adjusted mean scores of personal efficacy as a function of intrinsic motivation and the magnitude of the monetary incentive (fine).

As shown in the graph in Figure 3, within the class of people with high intrinsic motivation, the behavioral effect of the fine decreased as the level of the fine increased. By contrast, within the class of people with low intrinsic motivation, the effect of the fine increased with the increase in the level of the fine.

The interaction between the size of the deposit and the level of intrinsic motivation yielded a different pattern.Footnote ⁴¹ Within the low intrinsic motivation class, the pattern of results was similar to the Fine context: the behavioral effect of the deposit increased as the size of the deposit increased (F(1,293) = 8.34, p < 0.01, η² = 0.03). But within the high intrinsic motivation group, raising the deposit size was not significantly associated with a behavioral effect (F(1,293) = 0.21 p > 0.05, η² = 0.00). In other words, highly motivated participants reacted to deposits differently than to fines. The results are illustrated in Figure 4.

Figure 4. Adjusted mean scores of personal efficacy as a function of intrinsic motivation and of the magnitude of the monetary incentive (deposit).

The Expressive Effect of Green Earmarking

In a further exploration of the expressive effect of the law, we exposed the respondents in all the deposit/fine subgroups (across all the institutional dimensions) to the possibility that the funds collected under the deposit or fine programs would be transferred to an environmental fund that invests in pro-environmental projects. We examined the extent to which this information changed people's perception of the fairness of the regulatory scheme by comparing participants' perceptions in all the fine and deposit conditions before and after they were told about the environmental fund. To this end, we conducted a repeated measures test to examine the fairness of the program before and after exposure to the earmarking. We found a significant interaction effect between earmarking and legal instrument (F(1,1156) = 128.94, p < 0.001, η² = 0.10).Footnote ⁴² But whereas before the earmarking was announced, the deposit instruments were ranked higher in fairness than were the fine instruments, after the announcement a reverse pattern was found: the deposit instruments were ranked lower in terms of their fairness than were the fine instruments. In other words, “coloring” the money had a positive influence on the perceived fairness of the fine instruments (both high and low) and a reverse effect for the deposit instruments (both high and low). The interaction effect is illustrated in Figure 5.

Figure 5. Estimated means of fairness as a function of earmarking (before and after the announcement) and of the legal instrument within the entire sample (N = 1,159).

Regulation in a Multi-institutional Environment

We conducted a comparative institutional study of desirability and efficacy of the different legal instruments. Specifically, we examined whether the desirability and efficacy of the legal instrument (High Fine, High Deposit, or Ethical Code) changed according to the institutional source of the legal norm (state, organization, or academic institution).

First we conducted a two-way univariate analysis of covariance (ANCOVA) for the desirability of the law, in which the legal instrument and the institutional source of the legal norm were designated as independent factors. We found a significant main effect for the legal instrument across the three institutional settings (F(2,1344) = 114.95, p < 0.001, η² = 0.15).Footnote ⁴³ The Ethical Code (M = 56.38, SD = 11.83) was perceived as more desirable than the Deposit scheme (M = 48.93, SD = 17.47), which in turn was perceived as more desirable than the Fine scheme (M = 39.62, SD = 19.17) (as indicated by post hoc analyses). We did not find a statistically significant main effect for institutional setting (F(2,1344) = 0.50, p > 0.05, η² = 0.00).Footnote ⁴⁴ Furthermore, the interaction effect between the source of the legal norm and the legal instrument was not significant (F(4,1344) = 1.65, p > 0.05, η² = 0.01), indicating that the effect of the legal instrument on the measure of the desirability of the law did not change according to the source of the legal norm.Footnote ⁴⁵ In other words, we found no indication of extended resentment of fines and deposits within private organizations.

Second, to examine the effect of the various legal instruments and institutional settings on perceived personal and general efficacy, we conducted a two-way MANCOVA test in which the legal instrument and the institutional source of the legal norm were designated as independent factors.Footnote ⁴⁶ We did not find a significant main effect for the legal instrument, but we did find a significant main effect for the institutional setting (F(4,2684) = 8.35, p < 0.001, η² = 0.01).Footnote ⁴⁷ Within the Fine and Ethical Code subgroups, the institutional setting had significant effects on general efficacy,Footnote ⁴⁸ but within the Deposit subgroup, the perceived general efficacy was not influenced by institutional setting.Footnote ⁴⁹ In other words, the Fines and Ethical Code programs were perceived as having a greater effect on others within the private organization or academic setting than in the state setting, but the Deposit program was not perceived as such.

Intrinsic Motivation and the Choice of Legal Instrument

Our findings also demonstrate the importance of intrinsic motivation for the perception of different regulatory instruments. We found that participants with low intrinsic environmental motivation demonstrated higher resentment to fines than did those with high intrinsic motivation (who, given their greater belief in the value of recycling, were indifferent to the variance in the legal mechanisms that enforced it). Another surprising result was the preference for the High Deposit regime over the Low Fine regime among the low intrinsic motivation group, despite the reduced expected utility associated with the former regime. A possible explanation to this puzzle could be related to people's inability to understand the true monetary difference between two instruments due to framing and probability neglect (Reference McCaffery and BaronMcCaffery & Baron 2006; Reference SunsteinSunstein 2002).

Consistent with our initial predictions, we also found an indication of a crowding out effect within individuals with high intrinsic motivation. Among participants with high intrinsic environmental motivation, the Low Fine scenario generated a higher perceived (personal) efficacy of the law than did the High Deposit, High Fine, and Low Deposit scenarios. By contrast, among participants with a low intrinsic environmental motivation, the High Deposit scenario generated a higher perceived (personal) efficacy than did the Ethical Code, Low Fine, and Low Deposit scenarios. These differences in the effect of the larger monetary incentives confirm our prediction about the crowding out effect.

We are left with an intriguing question: why did individuals with high intrinsic motivation react more strongly to a Low Fine? One explanation is that the fine program provided some assurance to the highly motivated group that they would not be exploited by free riders. The free-riding argument raises a puzzle, however, because it seems to provide support for a reverse result, with a High Fine generating stronger reaction than a Low Fine owing to the stronger deterrence effect associated with a High Fine. A possible answer is based on Frey's U-shape hypothesis, according to which low monetary incentives may have a positive expressive message, which is diluted and becomes negative as the incentive is increased (Reference FreyFrey 1999; Reference ThØgersenThøgersen 1994). It is possible that the Low Fine manipulation triggered a similar effect to the low tax condition reported in Frey's paper and the High Fine condition triggered the greatest reduction in motivation, similar to Frey's intermediate level of tax.

Consistent with our initial hypothesis, we found that the crowding out effect was influenced by framing. In other words, although we found an antisynergetic effect between the extrinsic monetary incentive and the intrinsic motivation in both the fine and the deposit conditions, this effect was much stronger within the fine context. Whereas in the fine context raising the level of the fine was associated with reduced behavioral effect in the high intrinsic motivation group, raising the level of the deposit in this subgroup led to no significant effect. By contrast, the effect of raising the economic incentive within the low intrinsic motivation group was consistent with economic theory in both the fine and the deposit settings (leading to an increase in recycling intentions).

Earmarking

Examining the effects of environmental earmarking also revealed interesting differences between the legal instruments. “Coloring” the money had a positive influence on the perceived fairness of the fine instrument (both high and low) but a reverse effect on deposit instruments (both high and low). Furthermore, before the earmarking announcement the deposit instruments were perceived to be fairer than the fine instruments, whereas after the announcement the fine instruments were regarded as fairer.

The positive effect in the context of the fine can be explained by the assumption that the addition of earmarking to the regulatory scheme moderates the basic aversion to a fine. The negative effect we found in the case of deposits reflects the way in which earmarking diffuses the distinction between good and bad behavior. The difference we found between deposits and fines reflects the probabilistic nature of fines. In the case of deposits, individuals making a decision to return bottles for recycling know with certainty that they will be paid with money that otherwise would go to the environmental fund. By contrast, in the context of a fine, there is no such effect because nonrecycling is only probabilistically related to the fine, and therefore there is no parallel perception of preventing money from being transferred to the environmental fund by the act of recycling. This difference undermines the ability of people to make a complete contribution to the environment through the deposit program, and reduces its perceived fairness. An alternative explanation may be that earmarking highlights the immoral aspect of nonrecycling, thus bringing to the fore the negative externalities associated with it, which are masked by the neutral term deposit. This moral uneasiness may be projected onto the fairness factor.

An important policy question that we wish to explore in future research is whether the negative effect of earmarking in the case of deposits can be countered by allowing recyclers to waive their right to receive the deposit. There are some indications in the literature that this may be the case. Mellstrom and Johannesson raise a similar argument in the case of blood donation (2008), which they claim is partially driven by people's attempt to signal their altruism to others. Paying a donor reduces the effect of the contribution as a signal, and as a result payments to contributors may reduce contributions from those who are willing to do it for free. Based on their blood-donation experiment, the authors show that although at times payment deters donations, offering subjects the opportunity to donate their payment to charity fully counteracts its deterrent effect.

Comparative Institutional Analysis

The comparative institutional analysis produced several findings with respect to the capacity of private organizations to act as sources of governance. These findings are particularly important in light of the increasing role of the civic society and of private organizations in the regulatory realm. First, we found no evidence of resentment to fines and deposits within organizations, despite the relatively intrusive nature of these instruments. This finding was particularly surprising with respect to fines, an otherwise radical regulatory instrument associated with criminal law. Second, we found that the Ethical Code and Fine instruments were perceived as having a stronger effect on others in the private institutional setting than in the state setting. In the case of the Fine program, this result may reflect the perception that organizations have increased enforcement and monitoring capacities relative to the state. In the case of the Ethical Code program, it may reflect the possibility that within relatively small social environments voluntary codes may have greater effect because of the possibility of increased social (peer) pressure, both negative (social sanctions) and positive (peer praise); higher visibility of antienvironmental behavior; and organizational loyalty (Reference ElaEla 2009).

Limitations and Future Research

The methodology used in the study has certain intrinsic limitations. Measured items were self-reported attitude scales. Mindful of the extensive literature on the complexity of the attitude-behavior relationship (Reference Feldman and ClarkFeldman 2007), we must be careful in interpreting these findings, especially given the fact that recycling is a costly behavior. We cannot dismiss the possibility that participants may have engaged in “cheap talk” in reporting on their intention to recycle. Nevertheless, numerous studies have documented the validity of using attitudes as a proxy for behavior (Reference AjzenAjzen 1991). Furthermore, the focus of experimental surveys is not on the absolute numeric results they produce, but rather on the way in which they allow the researcher to differentiate among subgroups that were randomly assigned to the distinct settings. And although intention to behave is a controversial measure, measures of perceived fairness and social desirability are widely recognized in the literature as important both theoretically and practically (Reference Carlsmith, Darley and RobinsonCarlsmith et al. 2002; Reference TylerTyler 2006). Future research using lab-based methods could extend our understanding of the phenomena described in this article.

Another question concerns the generalizability of our findings given the fact that the study was carried out in Israel. We believe that the structure of contemporary Israeli society allows us to apply the results of the study to other countries, with similar socioeconomic profiles.Footnote ⁵⁰ There has been a marked change in the approach of Israeli society to environmental issues over the last decade, and this change is reflected both in the emergence of strong environmental civic activity and in a wave of new pro-environment legislation. Although security concerns still occupy an important place in everyday life in Israel, we believe that in terms of the exposure to ecological values and norms, the gap between Israeli society and the societies of other high-income countries has been substantively reduced (Reference TalTal 2002). At the same time, it is important to recognize that Israeli society has some unique cultural features. Historically, it was characterized by a strong sense of collective responsibility and a more pluralistic attitude toward authoritative norms (Reference Drory and Vigoda-GadotDrory & Vigoda-Gadot 2010: 199; Reference Smith, Peterson and SchwartzSmith et al. 2002: 203). Although we do not think that these differences affected significantly the results of our study given the close similarities between Israel and other developed countries, we recognize that further research should be conducted to fully support the generalizability of our results.