This article appeared in a journal published by Elsevier. The attached copy is furnished to the author for internal non-commercial research and education use, including for instruction at the authors institution and sharing with colleagues. Other uses, including reproduction and distribution, or selling or licensing copies, or posting to personal, institutional or third party websites are prohibited. In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier’s archiving and manuscript policies are encouraged to visit: http://www.elsevier.com/copyright
Author's personal copy The Journal of Socio-Economics 39 (2010) 447–454
Contents lists available at ScienceDirect
The Journal of Socio-Economics journal homepage: www.elsevier.com/locate/soceco
Appeals to social norms as policy instruments to address consumption externalities Steffen Kallbekken ∗ , Hege Westskog, Torben K. Mideksa CICERO Center for International Climate and Environmental Research – Oslo, P.O. Box 1129, Blindern 0318, Oslo, Norway
a r t i c l e
i n f o
Article history: Received 25 May 2009 Received in revised form 4 December 2009 Accepted 11 February 2010 JEL classification: D03 H20 Q58 Z13
a b s t r a c t We consider appeals to social norms as a policy instrument to address consumption externalities. We explore whether appeals to social norms can be an efficient policy instrument and compare the efficiency of such appeals to the efficiency of taxation in addressing consumption externalities. We find that when the existing norm helps to shift consumption towards the socially optimal level of consumption, taxation welfare dominates appeals to social norms as a policy tool. While previous studies have found that economic instruments are superior to information in other contexts, we arrive at a different conclusion for situations where the norm shifts behavior away from what is socially optimal. In such cases appeals to social norms can be a better policy instrument than taxation. © 2010 Elsevier Inc. All rights reserved.
Keywords: Social norms Policy instruments Externality
1. Introduction During the last few decades, we have seen an increasingly interdisciplinary approach in economics to understanding decisions and human behavior. Although economic incentives remain an integral element, the importance of emotions, moral sentiments, and social norms in shaping decisions have been explored, and testable implications have been drawn. This paper is a theoretical exploration of the potential use of providing information which appeals to social norms as a policy instrument. We define social norms as “customary rules of behavior that coordinate our interactions with others” (Young, 2007). Social norms are not outcome oriented, have indiscriminate application, and are enforced by internalized emotions and/or sanctions imposed by other individuals (Elster, 1989). The enforcement characteristic differentiates social norms from moral considerations. One suggestion from the economics literature on social norms is that social norms not only shape social interaction, but that in conjunction with economic incentives, they can explain certain
∗ Corresponding author. Tel.: +47 22 85 87 58; fax: +47 22 85 87 51. E-mail addresses:
[email protected] (S. Kallbekken),
[email protected] (H. Westskog),
[email protected] (T.K. Mideksa). 1053-5357/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.socec.2010.02.007
economic outcomes that are difficult to explain using the standard neoclassical approach. Several studies show how social norms are important for economic outcomes (e.g., Akerlof, 1980; Bernheim, 1994; Kreps, 1997). Akerlof (1980) examines adherence to social norms and the implications for economic outcomes. He shows that social customs that are disadvantageous to the individual may still persist if individuals are sanctioned by loss of reputation for disobeying the custom. Bernheim (1994) analyses the implications of the desire for “esteem” judged against the background of social norms. He shows that this desire might generate conformity in choice even when preferences are heterogeneous. Many studies have focused on the importance of social norms for outcomes in the labor market. Lindbeck and Snower (1988) and Lindbeck (1997) find that social norms that discourage unfair wages can explain widespread involuntary unemployment. Lindbeck and Nyberg (2006) show that disincentives to work as a result of a generous welfare system might materialize only with a substantial time lag because of work norms that parents have instilled in their children. Finally, studies have also focused on social norms connected to consumption externalities. Focusing on the case of “considerate” smoking behavior, Nyborg and Rege (2003) analyze the formation of social norms which reduce negative externalities. They show that the introduction of a smoking regulation can influence the social norm concerning considerate smoking behavior, so that there is
Author's personal copy 448
S. Kallbekken et al. / The Journal of Socio-Economics 39 (2010) 447–454
a shift from a no-consideration equilibrium to an equilibrium in which a large share of smokers practice considerate behavior. This paper focuses on the use of normative information measures to address consumption externalities. Some studies within the field of environmental economics have suggested that information provision in general might be an effective way to promote a shift towards more environmentally friendly consumption. Most of these studies, however, focus on the provision of descriptive information rather than on the provision of normative information (such as appeals to social norms). For example, Petrakis et al. (2005) examine a model that considers descriptive information provision as an environmental policy instrument. In the context of information asymmetry, market imperfections, and a consumption externality, the authors conclude that information provision welfare dominates taxation if information provision costs are sufficiently low. Asheim (in press) shows that a benevolent environmental agency may have incentives to withhold information about environmental consequences in order to avoid crowdingout of intrinsic motivation, and argues that if perfect economic instruments are available, they should be used. In terms of using normative information as a policy tool, some useful insights can be gained from empirical studies from other disciplines. Although these studies show that providing information appealing to social norms might be effective in changing behavior, they do not necessarily imply that it is an efficient policy instrument. Goldstein et al. (2008) analyze the consequences of invoking social norms to encourage hotel guests to reuse towels. In their experiment, they used messages that ranged from informing people about the environmental benefits of towel reuse to telling them about the behavior of their fellow hotel guests. Appealing to social norms (referring to what their fellow hotel guests did) is the most effective information strategy when it comes to changing behavior in this context. Their results are confirmed in another study by Nolan et al. (2008). This study focuses on electricity conservation. Participants were given information via door hangers, and their meters were monitored. A message about the proportion of neighbors’ who undertake energy conservation was found to induce more savings than messages providing other reasons to save energy (to protect the environment, to benefit society, or to save money). However, Schultz et al. (2007) point out that appealing to social norms might lead certain groups to use more energy than before—the so called “boomerang effect”. They find that messages referring to what others do are effective only as long as the individual or household receiving the appeal uses more energy than other households. If a household uses less energy than others, an appeal such as this might actually lead to increased energy-use unless it is followed by a message of support for the environmentally friendly behavior. Thus they argue that “for individuals who already abstain from the undesirable behavior, such normative information can produce unintended and undesirable “boomerang effects” (Schultz et al., 2007, p. 430). To avoid this potential “boomerang effect”, it might be useful to focus on injunctive norms rather than descriptive norms. Reno et al. (1993) define descriptive norms as specifying “what most people do in a particular situation,” while injunctive norms specify “what people approve and disapprove within the culture and motivate action by promising social sanctions for normative or counternormative conduct.” Schultz et al. (2007) show that the use of injunctive norms (i.e., using approval or disapproval of energy-use behavior) as a supplement to the descriptive norm avoids the boomerang effect. Most of the economics literature on social norms has focused on the role of norms in influencing behavior. The use of normative information as a policy tool has largely been neglected. To our knowledge, no study has addressed the central research questions in this paper:
Is the provision of information which appeals to social norms an effective policy instrument to address consumption externalities, and if so, is it efficient? How does the efficiency of such appeals to norms compare to the efficiency of taxation as a policy instrument? The rest of this paper is organized as follows. We discuss how to model social status based on adherence to a norm in Section 2 and then present our model and compare the efficiency of appeals to social norms to taxation in Section 3. Finally, in Section 4, we discuss the results and draw some wider conclusions. 2. Modeling social status based on adherence to a norm Using a simple choice model, we analyze the use of providing information which appeals to social norms as a policy instrument. For reasons of simplicity and tractability, we have attempted to keep the model as standardized as possible. Our model does, however, diverge from standard models in one key respect: it includes social norms in the individual’s utility function. More specifically, we assume—like, for example, Lindbeck (1997)—that there is a disutility from deviating from the social norm. This disutility can be interpreted in two ways, both of which are consistent with our models. (1) There is an enforced social norm. If an individual deviates from the norm, social sanctions that reduce the individual’s utility will be imposed. (2) The individual has internalized the norm such that the she suffers “feelings of embarrassment, anxiety, guilt and shame” at the prospect of violating the norm (Elster, 1989). We introduce a social status function S(Y,N), that depends on a consumption norm N, which describes the socially desirable quantity of consumption of good Y, and the actual consumption of a good Y. The mechanism through which norms can influence behavior—that there is a disutility in deviating from the norm—is the key driver of our results.1 It is therefore critical that we model this mechanism in the simplest and most general manner possible. We assume that deviating from the norm reduces utility. This interpretation is consistent with the view of norms as an ideal, standard, focal point, or point of conformity (“everyone should do the same”). It is important to note that, in general, individuals might gain social status both by conforming to social norms and by deviating from them (see Nail et al., 2000; Brewer, 1991; Snyder and Fromkin, 1980). Going against the grain can demonstrate independence and self-sufficiency to others. However, when looking at a representative consumer, it is a natural assumption that utility decreases in deviation from the norm, and that on aggregate the effect of a norm is to shift behavior in the direction of the norm. This does not rule out that certain groups or certain individuals derive greater utility from deviating from the norm than they do from adhering to the norm.2 In addition to assumptions concerning how the norm influences behavior, we also need to make assumptions about how the norm itself—or the perception of the norm—can be influenced. For simplicity we refer to this information provision as making “appeals to a social norm.” In line with the findings by Goldstein et al. (2008), we assume that appealing to a social norm has an impact on behavior
1 Individuals might also obtain positive utility from deviating from the social norm (see for instance Brewer, 1991). The idea is that people need to both differentiate themselves from others and to be similar to others. In our model we study a representative consumer, which is supposed to represent a typical individual. We assume that this representative individual obtains a greater utility gain from adhering to the norm than he/she would obtain by deviating from it. 2 Using appeals to a social norm might make certain groups or individuals more aware of the norm, and increase their desire to deviate from the norm. Whether this result in a shift towards a more socially optimal consumption level or not for such groups, depends on whether the existing norm shifts consumption towards or away from the socially optimal consumption level.
Author's personal copy S. Kallbekken et al. / The Journal of Socio-Economics 39 (2010) 447–454
by either shifting the perception of what the norm is or by changing the norm itself. Shifting the perception can imply either updated beliefs or that the information succeeds in making a norm salient in a new context (for example, that a norm for pro-environmental behavior at home also applies at the workplace). We make no specific assumptions about how much information must be supplied, and at what cost, to shift the norm, only that some amount of information provision will shift the norm. We further assume that the norm concerns what the consumption share of a specific good with a consumption externality ought to be. Finally, we assume that the norm itself, or adherence to the norm, is not affected by taxation. There is a significant literature suggesting how taxes can influence intrinsic motivation, such as the willingness to adhere to a norm (see for instance Frey and Oberholzer-Gee, 1997; Brekke et al., 2003; Gneezy and Rustichini, 2000; Nyborg, 2009). Taxes can both reinforce (crowd in) and undermine (crowd out) intrinsic motivation. If external incentives are viewed as controlling, the instrument might crowd out intrinsic motivation, whereas if external incentives are viewed as acknowledging the opposite might be the case. Finally, several experiments have shown that individuals are more willing to cooperate if others also cooperate (see for instance Fischbacher et al., 2001). Introducing an external incentive in such a setting might result in more cooperative behavior (i.e. adherence to the norm). The assumption that there is no interaction between taxation and the norm is thus non-trivial, and the impact of taxation on the norm could run in both directions (crowding-out making taxation relatively less attractive, and crowding-in making taxation relative more attractive compared to the use of appeals to norms). However, we choose to maintain this assumption as the opposite would imply a great expansion of the scope of our paper, and in order to allow a clear focus on one issue at a time. 3. The model We explore a situation in which individuals make choices over goods that provide them with private satisfaction while also causing external effect on other agents. Specifically, we use the model to address the two research questions outlined in the introduction. While the analysis holds for both positive and negative external effects we refer only to negative externalities below. Furthermore, to make the analysis more intuitive, we specifically refer to negative environmental effects. Later we will use other examples. Individuals weigh the marginal benefits from consumption and the marginal disutility of deviating from the norm. We suppose that there is a large number of agents that sum to one, each consuming two bundles of goods—clean goods and polluting goods. The social planner has two policy instruments at her disposal: Pigouvian taxes and information provision which appeals to a descriptive or injunctive social norm. In this discussion, we do not differentiate between descriptive and injunctive norms. We simply assume that the regulator or social planner would use the most effective appeal in any given context. A representative agent is assumed to have a concave utility function W (X, Y ; N) = U[Xi , Yi , S(Yi , N)] − D(˙i Xi )
(1)
from the consumption of a clean good Y and a polluting good X, with utility from social status/self image S(Yi ,N), which depends on the consumption of the clean good and a social norm concerning the socially acceptable level of consumption of the clean good.3 Utility
3 We abstain from using the subscript i that designates individual i in a set whenever this does not cause confusion.
449
is increasing in all three arguments. The agent takes the environmental damage, D(X)=D(i Xi ), as given. The damage depends on the total amount of the polluting good consumed. We also assume that D(X) is strictly increasing with X and convex in X. The social status function reflects the utility to the individual of adhering to the social norm. In the model specification below, social status is given by—(Yi − N)2 , which implies that any deviation from the norm reduces utility.4 Such a status function is called a strict social status function. We use this model to address the two research questions presented at the end of Section 1. 3.1. Are the policy instruments effective? The agent is subjected to a budget constraint: XPX + YPY = M,
(2)
where PX represents the price of the polluting good X, PY represents the price of the clean good Y, and M is the income of the agent. Moreover, let (3) represent the individually optimal consumption level, while (4) represents the socially optimal level of consumption (given prices and an income constraint). {X ∗ , Y ∗ } ≡ argmaxX,Y U[X, Y, S(Y, N)] st XPX + YPY = M
(3)
{X ∗∗ , Y ∗∗ } ≡ argmaxX,Y [U[X, Y, S(Y, N)] − D(X)] st XPX + YPY = M (4) The consequence of (3) and (4) is the well-known result that W(X* ,Y* ;N) < W(X** ,Y** ;N). This suggests that, in the presence of an externality, the individually optimal consumption level, (3), is suboptimal compared to the socially desirable choice, (4). A corollary that emerges from this is that W(X* ,Y* ;N) is locally decreasing in X* and increasing in Y* , implying that all agents could be better off if all reduce the consumption of the polluting good and increase that of the clean good. Moreover, we assume that the social planner can affect the social norm by providing information, I but that the marginal effect of information provision is decreasing. Formally, N = N(I) where N ′ (I) ≥ 0 and N ′′ (I) < 0.
(5)
Conditions (3), (4), and (5) lead to the conclusion summarized in Proposition 1. Proposition 1. In a setting where an agent obtains utility from the consumption of goods and from social status, changes in consumption norms can be effective in addressing a consumption externality problem. Conditions (3) and (4) both suggest that the distances d(X* ,X** ) and d(Y* ,Y** ) depend upon both economic incentives, namely prices and income, as well as the social norm for the clean good. Formally, d(X ∗ , X ∗∗ ) ≡ f [PX , PY , M, N(I)] ∗
∗∗
d(Y , Y ) ≡ g[PX , PY , M, N(I)]
(6) (7)
Both (6) and (7) suggest that changes in economic and/or social incentives might enhance social well-being. An alternative way to interpret these equations is to define the welfare difference as d(W(X* ,Y* ;N) − W(X** ,Y** ;N)) ≡ h[PX ,PY ,M,N(I)]. The function h(.) indicates that, in principle, a tax on the polluting good, a subsidy
4 Brekke et al. (2003) use a similar specification for self image in their model of moral motivation, where an individual’s self image is determined by a comparison between the actual effort and the morally optimal effort of contributing to a public good.
Author's personal copy 450
S. Kallbekken et al. / The Journal of Socio-Economics 39 (2010) 447–454
Table 1 Combinations of behaviors and socially optimal changes in the norm. Consume more of a good that decreases a negative externality, ∂N >0 ∂I
Consume less of a good that increases a negative externality, ∂N 0 ∂N ∂I Case 2: Extreme anti-pollution norms ∂U ∂S ∂N 0 and ∂W (X ∂I,Y ;N) ?0. ∂t X
Proof.
∂U ∂S
∂U ∂S
Consumption is less than the norm
See Appendix A.
The implication of Proposition 2 is that, in the presence of consumption externalities, both taxes and appeals to social norms can be used as instruments to improve social welfare. Whereas taxes unambiguously improve welfare, the results for appeals to social norms are ambiguous. We cannot, however, determine whether the two instruments are complements or substitutes or whether one of them is more efficient than the other, without making additional assumptions. 3.2. Which is the most efficient policy instrument?
improvement in environmental quality by using information as a policy instrument versus using tax as the policy instrument? To proceed further, we must make the simplifying assumptions that providing information is free and that the revenue is fully recycled (i.e., there are no transaction costs, but also no double dividend). These assumptions ensure that the budget constraint is the same in both cases, so that there are no changes in the marginal utility of income. To achieve a one-unit improvement in environmental quality, you need a decrease in the consumption of X of ␦X. With a constant budget constraint, this implies some given change ␦Y in the consumption of Y, regardless of which instrument is used. For simplicity, we call this a one-unit increase in Y and a one-unit decrease in X. It does not matter whether the shift is achieved by changing relative prices or by changing the norm—exactly the same shift from polluting to clean consumption is required to achieve a one-unit improvement in environmental quality. Using (8) and (9), combined with the assumptions on zero cost of information and full tax revenue recycling, we see that we need ∂X ∂X ∂N ∂Y ∂Y ∂N = ∂N and ∂p = ∂N to achieve the same shift in envi∂p ∂I ∂I X
Marginally shifting the social norm through providing information triggers changes towards the socially optimal level of consumption.6 The marginal changes in welfare in response to marginal changes in information provision can be expressed as:
X
ronmental quality. However, this is a weak assumption. Taking the budget identity, and differentiating with respect to I, we have ∂Y ∂N ∂X ∂N p + ∂N p ≡ 0. Again, differentiating the budget identity ∂N ∂I X ∂I Y with respect to pX and noting that full recycling implies a lump ∂X ∂Y sum rebate of , we have X + pX ∂p + pY ∂p ≡ . But, if we assume X
∂N ∂W ∗ = ∂I ∂I
∂U ∂X ∂U ∂Y ∂U + + ∂X ∂N ∂Y ∂N ∂S
∂S ∂Y ∂S + ∂Y ∂N ∂N
−
∂D ∂X ∂X ∂N
(8)
At the social optimum, this expression should be equal to the cost of providing the information multiplied by the shadow price of the government budget constraint. The net marginal benefit of using the social norm as a policy instrument should thus be equal to the cost of shifting the norm (by providing information). The marginal effect on welfare of changing the tax on the polluting good is given by: ∂U ∂X ∂U ∂Y ∂U ∂S ∂Y ∂D ∂X ∂W ∗ = + + − . ∂PX ∂X ∂pX ∂Y ∂pX ∂S ∂Y ∂PX ∂X ∂PX
(9)
This expression is somewhat easier to interpret. The net benefit of taxation is the difference between the marginal changes in consumption, status, and the externality. Is the use of information to shift the social norms Pareto inferior or superior to taxation as a policy instrument to address a consumption externality? To answer this question, we must first assume that we start from some consumer equilibrium. Pose the following question: What is the utility change if we achieve one-unit of
5 This is assumed to be an excise tax, not an ad valorem tax. This assumption prevents the use of unnecessary double concepts of tax and price elasticity, and allows us to focus only on price elasticity per se. 6 There is a potential issue of endogeneity in this problem as what is socially optimal depends on the norm. This is an issue which deserves further attention, but which does not cause further problems for our analysis as we focus on the welfare changes produced by the marginal application of each instrument, and do not require a precise technical definition of the social optimum in order to do so.
X
∂X ∂Y p + ∂p pY = 0, which inturn imply ∂pX X X ∂Y ∂Y ∂N = . With these assumptions, we can ∂pX ∂N ∂I
X ≡ , we then have
that
∂X ∂pX
now
=
∂X ∂N ∂N ∂I
and
remove elements in (8) and (9) from the comparison of appeals to social norms with taxation because they are either identical or equal by assumption:
−
∂U ∂Y ∂N ∂U ∂S ∂Y ∂N ∂U ∂S ∂N ∂D ∂X ∂N ∂U ∂X ∂N + + + − ∂X ∂N ∂I ∂Y ∂N ∂I ∂S ∂Y ∂N ∂I ∂S ∂N ∂I ∂X ∂N ∂I
∂U ∂Y ∂U ∂S ∂Y ∂D ∂X ∂U ∂X + + − ∂X ∂pX ∂Y ∂pX ∂S ∂Y ∂pX ∂X ∂pX
=
∂U ∂S ∂N ∂S ∂N ∂I
(8–9)
We are left with only one term that differs between the two expressions: the utility change from changes in status resulting from the change in the social norm. Because everything else is the ∂S ∂N same under our assumptions, the sign of the expression ∂U ∂S ∂N ∂I determines whether appeals to social norms or taxation is the most efficient policy instrument. , is always positive because we assume utilThe first term, ∂U ∂S ity to be increasing in social status. The other two terms can be either positive or negative, depending on the specific situation: whether actual consumption is greater than or less than the norm and whether the socially optimal policy is to increase or decrease the current norm (by which we mean to shift the ideal level of consumption up or down).7 This produces a 2 × 2 matrix of possible cases, which are shown in Table 1. To provide good illustrative cases, we no longer limit ourselves to environmental externalities.
∂S ∂N means that appeals to social norms ∂N ∂I may entail a welfare cost, as Bruvoll and Nyborg (2004) also argue. However, as our analyses will show, this welfare change can also be positive under certain circumstances. 7
The ambiguity of the equation
∂U ∂S
Author's personal copy S. Kallbekken et al. / The Journal of Socio-Economics 39 (2010) 447–454
In order to provide more intuition, we provide examples of which situations the different cases might represent.
451
Following Dixit (1979), the utility function of the representative agent is specified as8 : W (X, X, S) ≡ aX + aY − 0.5(X 2 − Y 2 + 2 XY ) + C − b(Y − N)2
Case 1—There is a negative environmental externality caused by the consumption of a polluting good. There is a clean substitute for this good, and a norm concerning the socially acceptable level of consumption of this good exists. The norm, however, implies less consumption of the clean good than what is socially optimal, and actual consumption is smaller yet than the norm. Case 2—There is a negative externality from consuming a good, but the norm is to use even less of this good than what is socially optimal. An example might be a norm not to use any of the pesticide DDT. The zero-use norm might be stricter than what is socially optimal. For example, the WHO (2006, p. 6) recommends the use of DDT for indoor spraying, stating that the “general disapproval of DDT use, due to fears of its harmful effects on the environment and on human health” are “unjustified when DDT is used appropriately.” Case 3—There is a negative health externality (assume a universal health care system) from consuming either too little food (anorexia nervosa) or too much food (obesity). Some people hold themselves to a norm which implies very low food consumption or are part of a group with an extreme norm for thinness. Because the norm is difficult to live up to, they consume more food than their norm implies. A norm implying greater food consumption would have yielded higher welfare (because of the health externality). Case 4—There is a negative environmental externality caused by the consumption of a polluting good, and there is a norm concerning the socially acceptable level of consumption of this good. The norm is, however, for a larger consumption share of the polluting good than what is socially optimal. Actual consumption is greater yet than the norm.
We now have a Proposition 3. Proposition 3. Assume that providing information is free and that tax revenues are fully recycled, then taxation Pareto dominates infor∂S ∂N mation provision if ∂U < 0, and information provision Pareto ∂S ∂N ∂I dominates taxation if
∂U ∂S ∂N ∂S ∂N ∂I
> 0.
The four different cases in Table 1 therefore yield only two different policy implications, making it clear that we are dealing with only two fundamentally different situations. The current level of the norm is either welfare enhancing, in the sense that it shifts consumption towards the socially optimal level of consumption (Cases 1 and 4), or it is “counterproductive,” in the sense that it shifts consumption away from the optimal level (Cases 2 and 3). To be able to compare appeals to norms and taxation in a more general case than the highly restrictive case shown above (i.e., when the budget constraint is the same in both cases), we need to employ a specified model to analyze each of the two situations: • The norm enhances welfare, corresponding to ∂U ∂S ∂N < 0. ∂S ∂N ∂I • The norm is counterproductive, corresponding to ∂U ∂S ∂N > 0. ∂S ∂N ∂I
(10′ )
− D(X),
where social status is given by S = −b(Y − N)2 . The parameter ∈[0,1] defines the degree of substitutability of X and Y. The two goods are independent of each other when = 0 and perfect substitutes when = 1. We assume, in the rest of the paper, that the two goods are close, but not perfect, substitutes to each other; that is, ∈(0,1), and the damage function is convex and quadratic, D(X) = 0.5mX2 , where m is the slope of the marginal damage function. The agent’s utility from social status depends on the amount of good Y consumed and the social norm N, which represents the ideal amount of good Y implied by the norm. The demand functions that characterize the solution to the problem are: Y∗ =
[(1 − )a + PX − PY + 2bN] (1 − 2 )
(10)
X∗ =
[(1 + 2b − )a − (1 + 2b)PX + PY − 2b N] . (1 − 2 )
(11)
These demand functions represent the reaction of the agent to changes in both the economic incentives and social considerations a la the behavioral approach of, for example, Elster (1989) and Lindbeck (1997). The economic incentives are channeled through prices while the normative motivations are channeled through the social norm. ∂X ∗
∂Y ∗ = = and ∂PX ∂PY (1 − 2 )
(12)
∂X ∗ −(1 + 2b) = ∂PX (1 − 2 )
(12′ )
The slope of the demand for X with respect to the social norm has a negative sign throughout. −2b ∂X ∗ = ∂N (1 − 2 )
(13)
The partial slope of the demand for Y, however, is positive throughout. ∂Y ∗ 2b = ∂N (1 − 2 )
(14)
The social norm for the clean product affects the demand for the polluting good as long as the two goods are substitutes. Thus, equations (13) and (14) imply that an increase in the norm for the clean good raises its demand while also reducing the demand for the close substitute, the polluting good. To compare taxes and information provision with respect to social welfare of (X,Y;N) ≡ U[Xi, Yi ,S(Yi ,N)] − D(i Xi ), suppose the planner imposed an optimal tax tX on the consumption of the polluting good. This would raise the price of the polluting good to PX ≡ tX + PX , which would change the consumption of X, from X* to X. Imposing the optimal tax tX on the consumption of the polluting
3.2.1. Taxes versus information with a welfare-enhancing social norm In the spirit of the above general model, we specify the utility function by characterizing choice in a situation where individuals have to choose the consumption that involves a dirty good X, a clean good Y, and other consumption goods C, which are used as the numeraire. Social status is based on the consumption of good Y.
good has three effects: it reduces utility by damage by D′(X)[X
− X ∗ ];
∂U(·) [X ∂X
− X ∗ ]; it reduces
and since the relative price between Y
8 This utility function has been employed without the term for social status, S = −b(Y − N)2 , by, among others Petrakis et al. (2005), in evaluating the importance of descriptive information as a policy instrument in the context of information assymetry, environmental damage, and market imperfections.
Author's personal copy 452
S. Kallbekken et al. / The Journal of Socio-Economics 39 (2010) 447–454
Table 2 Welfare changes from taxation and appeals to social norms. Policy instrument
Changes in welfare
Monetary change
Taxes (A)
∂U(·) [X ∂X
Information (B)
∂U(·) [X − X ∗ ] − D′(X)[X ∂X ∂U(·) ∂Y + N ′ (I) − ∂Y ∂S ∂PX ∂I
Net = H = (A − B)
− X ∗ ] − D′(X)[X − X ∗ ] + − X∗] +
∂U(·) ∂Y ∂S ∂PX ∂U(·) ∂Y ∂S ∂I
tX X −
∂U(·) ∂S
−R tX X + R
and X has changed, the agent also consumes more of good Y, result∂Y . Moreover, this policy results in ing in a welfare change of ∂U(·) ∂S ∂P X
revenues of tX X. Because the optimal Pigouvian tax internalizes an externality, the tax involves no distortion. Let us also suppose that there is a total cost of information R that also reduces the consumption of X from X* to X. The first effect would be to leave the agent with income from reduced con-
In this problem, we observe that the individual choice deviates from the social optimum, and that both changes to the norm and a subsidy to commodity Y can reduce the gap between the individual choice and what is socially optimal. The interesting question is whether information which appeals to the norm is a more efficient instrument than a subsidy from the perspective of social welfare. For a unit change in the subsidy for Y, , we have social welfare 1 ∗S2 2 Y + C − b(Y ∗S − N) − D(Y ∗S ). 2
sumption of X amounting to PX [X ∗ − X], which can be used for consumption of the clean good. This, however, changes utility from
W ∗ = aY ∗S −
[X − X ∗ ] and damage by D′(X)[X − X ∗ ]. Moreover, since X by ∂U(·) ∂X the social norm has changed as a result of the information provision, there are two other effects to consider, namely the effects of increased Y and increased N. These two effects are equal to ∂U(·) ∂Y − ∂U(·) N ′ (I). ∂S ∂I ∂S Let H express the excess benefit of the tax policy over the information provision policy. Table 2 illustrates the welfare and monetary changes resulting from changes in each policy.
The response to changes in the subsidy and change in the social norm is given by:
The value of
∂Y ∂PX
+ N′(I) −
∂U(·) ∂S
∂Y ∂I
∂Y ∂PX
+ N ′ (I) −
∂Y ∂I
is nonnegative as long as
≥ 0. From equation (9), we have N ′ (I) −
∂Y ∂I
=
∂Y ′ ∂Y ∂Y N ′ (I) − ∂N N (I) = N ′ (I) 1 − ∂N . Since ∂N < 1 for the specific util
ity function, we have
∂Y ∂PX
+ N ′ (I) −
∂Y ∂I
≥ 0. This proves that the
social planner would always be better off using taxes than information to address a negative externality when the social norm is welfare-enhancing. 3.2.2. Taxes versus information with a counterproductive social norm Consider now the case of the eating disorder anorexia nervosa. This is a situation where individuals consume fewer calories than what is socially desirable. We assume that there is a negative externality with respect to (low) calorie consumption, which the social planner wants to reduce. Let Y be food and assume that the actual consumption of Y is greater than what the individual perceives the norm to be. The perceived norm might be more stringent than the actual social norm is, or the norm might be particularly strict within a smaller (peer) group. In this situation, there is a norm for thinness that is counterproductive in the sense that it shifts behavior away from what is socially optimal. In economic jargon, greater consumption of food, in such a group, would reduce the negative externality, D(Y) such that ′ D (Y) < 0. We continue to assume a status function that depends on the social norm. For simplicity, we assume utility to depend on Y, the numaire good C, and social status, −b(Y − N)2 subject to the budget constraint C + YPY = M. For b ∈(0,1), the utility function is given by U = aY −
1 2 Y + C − b(Y − N)2 . 2
(15)
The optimal choice of Y is Y ∗S = [(˛ − PY ) + 2bN]/[1 + 2b], which is greater than what the social norm prescribes. For the social planner, the optimal Y would maximize W = aY − 21 Y 2 + C −
b(Y − N)2 − D(Y ), with the socially optimal choice of Y being Y ∗∗ = [(˛ − PY ) + 2bN − D′(Y )]/[1 + 2b], which is strictly greater than Y*S .
(10′′ )
∂Y ∗S ∂W ∗ = [PY − D′(Y ∗S )] ∂ ∂
(16)
∂Y ∗S ∂W ∗ = [PY − D′(Y ∗S )] + 2b(Y ∗S − N). ∂N ∂N
(17)
Since Y*S > N, if the cost of a unit change in the norm (through information provision) is equal to or less than the cost of a unit change in the subsidy, then we have it is always the case that ∗ ∂W ∗ > ∂W can be, dependent on the cost of providing informa∂N ∂ tion relative to the cost of the subsidy. This shows that, with a counterproductive norm and depending on the cost of the information provision, information provision can be a more efficient policy instrument than the economic instrument.9 A fundamental driver of the problem seems to be unhealthy body ideals. The fact that some people diet in order to bring their actual body shape closer to their ideal, when they are already thinner than what most people (and certainly medical professionals) would consider healthy, could also be caused by perceptual distortions (that people believe themselves to be fatter than they actually are). Although descriptive information could potentially address the problems of misperception of relative body mass, Cash and Deagle (1997) state that, “Compared to perceptual distortion effects, self-ideal discrepancy effects were considerably larger.” In other words, people with anorexia have rather extreme body image ideals for weight and shape. For the case of anorexia, our model suggests that using normative information to try to establish a healthier body image would bring a double utility gain. If you succeed in shifting the norm (the body ideal) among people who suffer from anorexia, the result should be increased food consumption, which would reduce external costs. In addition, by shifting the norm towards an ideal body image that is more easily attainable, the deviation between actual and ideal food consumption would likely decrease, and utility would thus increase. People who suffer from anorexia would not only consume more food and thereby become healthier, they would also have a more positive self image. Whether or not this produces an overall welfare gain depends on the cost of providing information which appeals to the norm (either shifting the norm as such or the perception of the norm).
9 In this particular case, using a subsidy would be socially inefficient also independent of this result. The share of overweight and obese people is much higher than the share of people with anorexia. Under normal assumptions, a subsidy would increase the damages from overweight and obesity more than it would reduce the damages from anorexia.
Author's personal copy S. Kallbekken et al. / The Journal of Socio-Economics 39 (2010) 447–454
4. Discussion In this paper we have addressed two research questions. (1) Is the provision of information which appeals to social norms an effective policy instrument to address consumption externalities, and if so, is it efficient? (2) How does the efficiency of such appeals to norms compare to the efficiency of taxation as a policy instrument? We have proposed a mechanism through which appeals to social norms can influence behavior and found that whether this instrument is more or less efficient than taxation depends on the characteristics of the relevant good, the norm, and how the norm affects behavior. More specifically, when the existing norm helps to shift consumption towards the socially optimal level of consumption, taxation Pareto dominates information provision as a policy tool (see Section 3.2.1). In the opposite case, when the norm shifts behavior away from what is socially optimal, information provision can be a better policy instrument than taxation (see Section 3.2.2). Our results might seem at least partially contradictory to the results of Petrakis et al. (2005). However, they compare descriptive information measures to taxes with the perspective that people might lack information on certain issues, for example, on the health effects of paints and possible protective measures. They find that descriptive information measures are welfare superior to taxes if information can be provided at low costs. Our study differs by analyzing the role of normative information as a policy instrument. The results are therefore not contradictory, but they do point to the importance of analyzing different types of information measures separately. Model results are necessarily categorical, and there are some important qualifications that need to be made. One important qualification is that it is not always possible to impose a socially optimal tax. For example, the tax levels required to achieve intended environmental effects are often not politically feasible, and the resulting tax level may end up inefficiently low (Daugbjerg and Svendsen, 2001). In these cases, using normative information measures to change behavior in a pro-social direction may be the best option because the optimal tax is not feasible. The case for using appeals to social norms might also be stronger in cases where there are interaction effects between the tax and the norm, such as crowding-out effects. The effects are, however, ambiguous in general as there are both cases where taxation could decrease and cases where taxes could increase the intrinsic motivation to act according to a norm. The key driver of our results is the mechanism through which information (i.e., appeals to the social norm) can influence behavior. We have assumed that there is a social norm, which is costly to deviate from (in terms of individual utility) and which can be shifted through providing information which appeals to that norm. Choosing a different mechanism would have produced a different result. One natural candidate would be to allow the normative information to enter directly into the utility of consuming the clean good, that is, that the normative information increases the marginal utility of consuming the clean good. With such an approach, changing the preference would not produce a similar (partial) utility loss/gain to that associated with shifting a norm. However, we consider such an approach to be problematic for two reasons. The first is that it would imply that preferences are not given, which would be contrary to the conventional approach since Stigler and Becker (1977). The second reason is that it would provide only trivial insights into how information can influence behavior (i.e., if you have a stronger preference for something you consume more of it). To many economists, the issue of paternalism is a grave concern when it comes to using normative information measures as a policy instrument. Although this is a relevant issue, we do not discuss it because we have chosen to limit the scope of our paper to the analysis of whether or not appeals to social norms can be efficient as a policy instrument when compared to taxation. However,
453
if paternalistic use of information measures is an important concern, this might limit the use of normative information measures. In any case, because normative information measures are used in several areas to change behavior, it is important to understand how they work and whether or not they can be used as an efficient policy instrument. We have investigated this issue in our paper, and found that taxation welfare dominates appeals to social norms when the norm helps to shift consumption towards the socially optimal level of consumption, but that appeals to social norms can be a better policy instrument than taxation when the existing norm is “counterproductive”. Acknowledgements We would like to thank Geir Asheim, Kjell Arne Brekke, Karine Nyborg, participants at the Third Nordic Conference on Behavioral and Experimental Economics, participants at EAERE 2009, as well as one anonymous reviewer for valuable comments. The research was funded by The Research Council of Norway. Appendix A. Proposition 2. Let tX be a tax on the polluting good. Then information provision has a locally uncertain effect, while the tax on X* unambigously improves social welfare. Specifically, we have ∗ ∗ ∂W (X ∗ ,Y ∗ ;N) > 0, and ∂w(X ∂I,Y ;N) ?0. ∂t X
Proof (:). With an ideal social norm (deviations reduce utility), we have W(X* ,Y* ;N) = U(X* ,Y) − a(Y* − N)2 − D(X* ). We also assume that Uij > 0 ∀ i = / j. For situations where Y* < N, to know the sign and value of ∂Y ∗ and ∂P , note that (3) and (4) are solutions to
∂X ∗ ∂X ∗ ∂Y ∗ , ∂N ∂N , ∂PX
X
U1 (X ∗ , Y ∗ ) ≡ PX ∗
∗
(A1) ∗
U2 (X , Y ) − 2a(Y − N) ≡ PY
(A2)
X ∗ PX + Y ∗ PY ≡ M.
(A3)
Differentiating (A2) and (A3) with respect to PX and noting that X* > 0, we have Y21 PX
∂X ∗ ∂Y ∗ + (U22 − 2a) =0 ∂PX ∂PX
(A4)
∂X ∗ ∂Y ∗ = −X ∗ . + PY ∂PX ∂PY
(A5)
Applying Cramer’s rule to (A4) and (A5), it is straightforward to ∂X ∗ ∂Y ∗ note that ∂P < 0 and ∂P > 0. Therefore, X
X
Since W(X* ,Y* ;N) is locally decreasing in X* and increasing in Y* we have W1 (X* ,Y* ;N) ≡ U1 (X* ,Y* ) − D′ (X* ) < 0 and W2 (X* ,Y* ;N) ≡ U2 (X* ,Y* ) − 2a(Y* − N) > 0, suggesting that Y* ≤ N. This completes the proof that welfare is locally increasing with ∗ ∗ tax, ∂W (X∂P,Y ;N) > 0. X
∂W (X ∗ , Y ∗ ; N) ∂X ∗ = [U1 (X ∗ , Y ∗ ) − D′(X ∗ )] ∂N ∂N
∂Y ∗ + [U2 (X , Y ) − 2a(Y − N)] −1 ∂N ∗
∗
∗
Moreover, differentiating (A2) and (A3) with respect to N, we have U21 PX
∂X ∗ ∂Y ∗ + (U22 − 2a) = −2a ∂N ∂N
∂X ∗ ∂Y ∗ + PY =0 ∂N ∂N
(A6) (A7)
Author's personal copy 454
S. Kallbekken et al. / The Journal of Socio-Economics 39 (2010) 447–454
Again, applying Cramer’s rule to (A6) and (A7), it is straightfor∗ ward to note that ∂X < 0, and assuming that each of the goods ∂N ∂Y ∗ > 0. This suggests that ∂N ∗ ∂W (X ∗ ,Y ∗ ;N) ?0 when ∂Y < 1. ∂N ∂N
are complimentary in utility, we have ∂W (X ∗ ,Y ∗ ;N) ∂N
∂Y ∗ > 1 and ∂N ∂Y ∗ = 2aPX /[2aPX ∂N
> 0 when
But the value of + PY U21 − PX U22 ], which is less than 1, in which case the proof is complete. For situations where Y* > N, to realize the sign and value of ∂X ∗ ∂X ∗ ∂Y ∗ ∂Y ∗ , ∂N ∂N , and ∂P , note that (3) and (4) are solutions to (A1), ∂P X
X
(A2′ ), and (A3) where (A2′ ) is given by U2 (X* ,Y* ) − 2a(N − Y* ) ≡ PY . What remains is to follow the same steps as in the case for situations where Y* < N. References Akerlof, G.A., 1980. A theory of social custom, of which unemployment may be one consequence. The Quarterly Journal of Economics 94 (4), 749–775. Asheim, G., in press. Strategic use of environmental information. Environmental and Resource Economics. Bernheim, B.D., 1994. A theory of conformity. Journal of Political Economy 102 (5), 841–877. Brekke, K.A., Kverndokk, S., Nyborg, K., 2003. An economic model of moral motivation. Journal of Public Economics 87 (9–10), 1967–1983. Brewer, M.B., 1991. The social self: on being the same and different at the same time. Personality and Social Psychology Bulletin 17, 475–482. Bruvoll, A., Nyborg, K., 2004. The cold shiver of not giving enough: on the social cost of recycling campaigns. Land Economics 80 (4), 539–549. Cash, T.F., Deagle, E.A., 1997. The nature and extent of body-imagedisturbances in anorexia nervosa and bulimia nervosa: a meta-analysis. International Journal of Eating Disorders 22 (2), 107–126. Daugbjerg, C., Svendsen, G.T., 2001. Designing Green Taxes in a Political Context: From Optimal to Feasible Environmental Regulation. University of Aarhus, Aarhus School of Business, Department of Economics, Aarhus. Dixit, A., 1979. A model of Duopoly suggesting a theory of entry barriers. Bell Journal of Economics 10 (1), 20–32. Elster, J., 1989. Social norms and economic theory. Journal of Economic Perspectives 3 (4), 99–117. Fischbacher, U., Gächter, S., Fehr, E., 2001. Are people conditionally cooperative? Evidence from a public goods experiment. Economics Letters 71, 397–404.
Frey, B.S., Oberholzer-Gee, F., 1997. The cost of price incentives: an empirical analysis of motivation crowding-out. American Economic Review 87 (4), 746– 755. Gneezy, U., Rustichini, A., 2000. A fine is a price. Journal of Legal Studies 29, 1–17. Goldstein, N.J., Cialdini, R.B., Griskevicius, V., 2008. A room with a viewpoint: using social norms to motivate environmental conservation in hotels. Journal of Consumer Research 35, 472–482. Kreps, D.M., 1997. Intrinsic motivation and extrinsic incentives. American Economic Review 87 (2), 359–364. Lindbeck, A., 1997. Incentives and social norms in household behavior. American Economic Review 87 (2), 370–377. Lindbeck, A., Snower, D.J., 1988. Cooperation, harassment, and involuntary unemployment: an insider–outsider approach. American Economic Review 78 (1), 167–188. Lindbeck, A., Nyberg, S., 2006. Raising children to work hard: altruism, work norms, and social insurance. The Quarterly Journal of Economics 121 (4), 1473– 1503. Nail, P.R., MacDonald, G., Levy, D.A., 2000. Proposal of a four-dimensional model of social response. Psychological Bulletin 126, 454–470. Nolan, J.M., Schultz, P.W., Cialdini, R.B., Goldstein, N.J., Griskevicius, V., 2008. Normative social influence is underdetected. Personality and Social Psychology Bulletin 34 (7), 913–923. Nyborg, K., 2009. Will Green Taxes Undermine Moral Motivation? Public Finance and Management, forthcoming. Nyborg, K., Rege, M., 2003. On social norms: the evolution of considerate smoking behavior. Journal of Economic Behavior & Organization 52 (3), 323– 340. Petrakis, E., Sartzetakis, E.S., Xepapadeas, A., 2005. Environmental information provision as a public policy instrument. Contributions to Economic Analysis & Policy 4 (1), 1365–1465. Reno, R.R., Cialdini, R.B., Kallgren, C.A., 1993. The transsituational influence of social norms. Journal of Personality and Social Psychology 64, 104–112. Schultz, P.W., Nolan, J.M., Cialdini, R.B., Goldstein, N.J., Griskevicius, V., 2007. The constructive, destructive, and reconstructive power of social norms. Psychological Science 18 (5), 429–434. Snyder, C.R., Fromkin, H.L., 1980. Uniqueness: The Human Pursuit of Difference. Plenum, New York. Stigler, G.J., Becker, G.S., 1977. De Gustibus Non Est Disputandum. American Economic Review 67 (2), 76–90. WHO, 2006. Indoor Residual Spraying – WHO Position Statement on IRS. World Health Organization, Geneva. Young, H.P., 2007. Social Norms. University of Oxford, Department of Economics, Oxford.