Abstract
I provide an alternative explanation for why societies exhibit varied environmental behaviours. I use a Kantian moral approach at a microeconomic level. I show that two identical societies (in terms of income level and political system) might follow different paths with respect to their “green” behaviour. Additionally, I identify tipping points that could nudge a society from a polluting behaviour to a green one. I find that the perception of environment within the society can be an important factor in this shift.
This is a preview of subscription content, log in to check access.
Notes
 1.
The EKC is similar to a traditional Kuznets curve. As a country gets richer and more developed, it begins to pollute more (as measured on a percapita basis). After reaching a certain developmental level, society begins to grant more importance to the environment; therefore, it starts to pollute less as it becomes richer. Hence we observe an inverted Ushape illustrating the relationship between pollution and income per capita. This literature goes back to the early 1990s, including works by Grossman and Krueger (1991), Shafik (1994), and Panayotou (1993), among others. Although the EKC was used to analyse local pollution, it could also be used to compare countries concerning global pollutants.
 2.
For the year 2011, the emissions per capita (in metric tons of CO\(_2\)) were: USA, 17; Canada, 14.1; Australia,16.5; Denmark, 7.2; Finland, 10.2; Germany, 8.9; Netherlands, 10.1; Sweden, 5.5; and UK, 7.1. Average of each group, respectively: 15.9 and 8.2. The GDP per capita (in thousands of current US$) were: USA, 49.8; Canada, 52; Australia, 62.1, Denmark, 61.3; Finland, 50.8; Germany, 45.8; Netherlands, 53.5; Sweden, 59.6; and UK, 41. Average of each group, respectively: 54.7 and 52. Source: World Development Indicators, The World Bank.
 3.
This assertion is considering a public good setting where economics incentives to behave green are few or nonexistent.
 4.
Sweden was the first country to introduce a carbon tax in 1991, and they have been increasing it over time. They are currently aiming to increase it again in order to “point out how to achieve the 2050 vision of zero net GHG emissions” in their Climate Roadmap. (Energy Policies of IEA Countries, Sweden, 2013 review).
 5.
At least not in the short term. In the long term, green countries could push an international environmental agreement, induced by a moral motive, which would lead to an increase in welfare.
 6.
From Kant’s work on moral philosophy (Kant 1785), we know that the categorical imperative formulates the equality postulate of universal human worth. Hence, while a Kantian attitude and an egalitarian attitude are not the same, they are closely related and we could proxy one to the other.
 7.
 8.
 9.
 10.
Kuran talks about the collapse of Eastern Europe’s communist regimes. He divides the society into 10 types of people, ranging from those who are more in favour of a communist government, to those completely opposed to it. He shows that if some sort of threshold is crossed, protests can begin, which can encourage those initially less likely to go against the incumbent regime to join in protesting. This process can lead to a cascading effect, which can in turn trigger the collapse of the whole regime.
 11.
Persson et al. (2000) used a theoretical model to show that presidential regimes should produce an underprovision of public goods (thus leading to a dirtier environment). On the other hand, Bernauer and Koubi (2004) found the opposite result. They use an econometric study to find evidence that presidential democracies provide more public goods than do parliamentarian democracies. More recently, Saha (2007) tested the previous hypotheses empirically. She finds that the electoral system has no effect on any of the environmental public good supply indicators, and that the nature of the political regime also has no significant impact.
 12.
This means that green people prefer others to behave in a green way as well, and grey people prefer others to behave in a grey way. Of course, it might be the case that being grey is always considered ‘bad’, even for grey people, as in the case of smoking.
 13.
If there were a green good that was cheaper than its grey counterpart, agents would automatically choose that good instead of the grey one for purely economic reasons. In the case with multiple products, if some green products were cheaper than their grey counterparts, we could recalculate the pollution produced by a new representative grey good and return to the setup presented here.
 14.
The agent’s binary choice is used for its simplicity: It results in closed forms. Also, a model with mixed choice was developed and tested with simulations, arriving to equivalent results. The behaviour functions and the resulting dynamics turn out to be the same.
 15.
From the budget constraint we have that \((1+\rho )x_{it} + y_{it} = 1\). Since the agent only chooses the green or grey product, we have the case with \(y_{it}=0\) and hence, \(x_{it}=1/(1+\rho )\); or \(x_{it}=0\) with \(y_{it}=1\).
 16.
From the previous formulation it follows that emissions from the rest of the world are not taken into account (if it were the case that we were considering global pollutants). I proceed in this manner because I want to focus on the country’s own drivers to change, not on those coming from the international community. Taking this caveat into account, it is clear that the moral driver will be focused on national behaviour, the aim of this paper, rather than international behaviour (emissions).
 17.
As will be made clear in the following pages, the distribution of \(\alpha \) will not change the main results, but assuming a uniform distribution will certainly ease the subsequent calculations and simulations. The analysis will be performed using a uniform distribution. On the other hand, different distributions will simply change the place of the tipping point and the conditions needed to tip, as in Kuran (1991).
 18.
Since the agent has only two options for \((x_{it},y_{it})\), being these \((\frac{1}{1+\rho },0)\) or (0, 1), I simplify notation by just referring to \(y_{it}\), which can just take the values of 0 and 1. Moreover, from the budget constraint and products’ perfect substitutability, we have that the argument of the function \(\tilde{u}(\cdot )\) is \(x+y=(1+\rho y)/(1+\rho )\).
 19.
The original categorical imperative (or one of the original versions) was: “So act as if the maxim of your action were to become through your will a universal law of nature” (Kant 1785).
 20.
For a wellwritten essay on the relationship between the Kantian imperative and climate change, see Rentmeester (2010).
 21.
There is also another way of tackling this diversity: We might think that each person has the same Kantian attitude, but that the parameter \(\alpha _i\) instead reflects how ‘optimistic’ each person is. Although this is not the same idea stated here, a development and a proof of its equivalence is given in the Appendix A.
 22.
As stated in the previous pages, if we are considering global pollutants, the raise of pollution level could be a result of more than that particular country’s emissions. On the other side, the impact of (local) emissions \(\gamma \) is directly linked to the choices of the agent in situations in which he or she has control.
 23.
An interesting feature to note is that if we consider consumption levels as proportional to some income level w (i.e. comparing \(\tilde{u}(w)\) and \(\tilde{u}(\frac{w}{1+\rho })\)), the private cost of behaving in a green way, \(H(\rho )\), could be increasing, decreasing, or independent of the income level w depending on the functional form of \(\tilde{u}(\cdot )\). Since in this formulation I intentionally leave out the income effect, I use the case where \(\tilde{u}(w) = \ln (w)\) for simulations, which results in a \(H(\rho )\) that is independent of w. For details see Appendix C.
 24.
For the present and following definitions, I use a uniform distribution of \(\alpha _i\). If this were not the case, we would have a different function \(\theta (\cdot )\), but it would still retain the subsequent properties and results.
 25.
I use the usual linear natural absorption, as in Calvo et al. (2012) and Breton et al. (2010). A different form for this absorption could been used, but as it can be observed in Fig. 2, we will have in general only one crossing with \(\theta \). Hence, the addition is a more complex form does not add significant new results.
 26.
We can notice that the behaviour of each agent, and hence of the whole society, is determined by the pollution level, through the function \(\theta (p_{t1},\rho )\) (which also depends on \(\rho \)). Therefore, finding an equilibrium point for the pollution level gives us the equilibrium situation of the whole society.
 27.
In other words, a tax and subsidy will be chosen in order to achieve a new differential in green and grey prices equal to \(\rho _2\). Details on the implementation of the policy (tax and subsidy chosen) can be found in Appendix E, where I show that the policy is equivalent to just changing \(\rho \) in the \(\theta (\cdot )\) function; this will shift the curve to the right, as depicted in Fig. 1.
 28.
This could be due to the fact that there are more than two political actors in place, or only two actors with multiple political dimensions to bargain over, etc.
 29.
We can see from this reasoning that it is not necessary to know exactly where the thresholds are, but only to be aware of the ability to nudge the system toward a greener equilibrium given these thresholds.
 30.
I therefore disregard the case of negative social pressure from people not behaving as the agent does. It can be quickly derived that including this second effect does not change the results.
 31.
This is done for simplicity. Alternatively  and more generally  specifying \(u^a_{it}\) as a function of \(v(\alpha ^*_t)\) (\(1  v(\alpha ^*_t)\)) when the agent is behaving in a grey (green) way, with \(v(0) = 0\) and \(v(1) = 1\) for normalization and \(v'>0\), does not change the coming results.
 32.
This was named in the Introduction as an ‘ideological’ peer pressure. Also refer to footnote 12.
 33.
Since \(v^\prime > 0\) (for the general case with \(v(\cdot )\)), the point when \(S(\alpha ^*_t) = 0\) (neutral social approval: \(v(1  \theta ) = v(\theta )\)) has to be when \(\theta = 1/2\). Hence if \(\theta < 1/2 \rightarrow S(\alpha ^*_t) > 0\) and viceversa. Another way of modelling the peer pressure would be to say that the agent reacts to \(\theta _{t1}\), the share of grey people in the previous period. This would introduce a difference equation into the system where lags would also play a role. In order to keep the model simple, I use the case where the agents instantly respond to society’s behaviour.
 34.
To see what stability means, consider \(K(\Omega p_{t1}) < 2\beta \) and \(\alpha ^*_t = \alpha ^{int}_t\). Suppose the “almost indifferent agents”, i.e. the agents i with \(\alpha ^{int}_t< a_i < \alpha ^{int}_t + \epsilon \) for \(\epsilon \) arbitrarily small and positive, make a mistake (they “tremble” in game theoretic parlance) and choose to be grey instead of green. Then \(\alpha ^*_t = \alpha ^{int}_t + \epsilon \) and \(A(\alpha ^*_t) = \alpha ^{int}_t + (2\beta / K(\Omega p_{t1})) \epsilon > \alpha ^{int}_t + \epsilon \) This implies that all agents i for which \(\alpha _i < \alpha ^{int}_t + (2\beta / K(\Omega p_{t1})) \epsilon \) want to act grey, thus including the trembling almostindifferentagents. They are not willing to revert to green but instead keep behaving grey, while in addition some other agent have now switched to grey. This process will continue until everybody behaves grey. Similarly, if the tremble is in the green direction, we end up having everybody behaving green.
 35.
This follows the same intuition as the result obtained in Nyborg et al. (2006). In their model there are two extreme equilibria, completely green or completely grey, with a third unstable equilibrium in between.
 36.
\(\overline{\theta }\) has to be bigger: it has to get to B, which is further left than A.
References
Andreoni J (1990) Impure altruism and donations to public goods: a theory of warmglow giving. Econ J 100(401):464–477
Baumgärtner S, Petersen T, Schiller J (2014) Bringing norms into action—the concept of responsibility. Working paper, University of Lünenburg
Bernauer T, Koubi V (2004) On the political determinants of environmental quality. Annual meeting of the American Political Science Association, Hilton Chicago and the Palmer House Hilton, Chicago, IL
Bisin A, Verdier T (1998) On the cultural transmission of preferences for social status. J Pub Econ 70(1):75–97
Bisin A, Verdier T (2000) A model of cultural transmission, voting and political ideology. Eur J Polit Econ 16(1):5–29
Bisin A, Verdier T (2001) The economics of cultural transmission and the dynamics of preferences. J Econ Theory 97(2):298–319
Bowles S (1998) Endogenous preferences: the cultural consequences of markets and other economic institutions. J Econ Lit 36:75–111
Brekke KA, Nyborg K (2008) Attracting responsible employees: Green production as labor market screening. Resour Energy Econ 30(4):509–526
Brekke KA, Kverndokk S, Nyborg K (2003) An economic model of moral motivation. J Pub Econ 87(9–10):1967–1983
Breton M, Sbragia L, Zaccour G (2010) A dynamic model for international environmental agreements. Environ Resour Econ 45(1):25–48
Bruvoll A, Halvorsen B, Nyborg K (2000) Household sorting of waste at source. Econ Surv 4(2000):26–35
Buenstorf G, Cordes C (2008) Can sustainable consumption be learned? A model of cultural evolution. Ecol Econ 67(4):646–657
Calvo E, Rubio SJ et al (2012) Dynamic models of international environmental agreements: a differential game approach. Int Rev Environ Resour Econ 6:289–339
Comin D, Rode J (2013) From green users to green voters. Working Paper 19219, National Bureau of Economic Research
Corbett JB, Durfee JL (2004) Testing public (un) certainty of science media representations of global warming. Sci Commun 26(2):129–151
Dunwoody S (2007) The challenge of trying to make a difference using media messages. In: Creating a climate for change, pp 89–104
Extreme Weather and Climate Change in the American Mind April 2013. http://environment.yale.edu/climatecommunication/article/extremeweatherpublicopinionApril2013
Gallup World Poll. http://www.gallup.com/strategicconsulting/enus/worldpoll.aspx
GlobeScan Radar. http://www.globescan.com/expertise/trends/globescanradar.html
Grossman GM, Krueger AB (1991) Environmental impacts of a north American free trade agreement. Technical report, National Bureau of Economic Research
Hollander H (1990) A social exchange approach to voluntary cooperation. Am Econ Rev 80(5):1157–1167
Kant I (1785) Groundwork of the metaphysics of morals
Krosnick JA, Holbrook AL, Lowe L, Visser PS (2006) The origins and consequences of democratic citizens’ policy agendas: a study of popular concern about global warming. Clim Change 77(1–2):7–43
Kuran T (1991) Now out of never: the element of surprise in the east european revolution of 1989. World Polit 44(10):7–48 ISSN 10863338
Laffont JJ (1975) Macroeconomic constraints, economic efficiency and ethics: an introduction to kantian economics. Economica 42(168):430–437
Lee TM, Markowitz E (2013) Disparity in the predictors of public climate change awareness and risk perception worldwide
Leiserowitz A (2006) Climate change risk perception and policy preferences: the role of affect, imagery, and values. Clim Change 77(1–2):45–72
Nyborg K (2011) I don’t want to hear about it: rational ignorance among dutyoriented consumers. J Econ Behav Organ 79(3):263–274
Nyborg K, Rege M (2003) Does public policy crowd out private contributions to public goods. Pub Choice 115(3–4):397–418
Nyborg K, Howarth RB, Brekke KA (2006) Green consumers and public policy: on socially contingent moral motivation. Resour Energy Econ 28(4):351–366
Panayotou T (1993) Empirical tests and policy analysis of environmental degradation at different stages of economic development. Technical report, International Labour Organization
Persson T, Roland G, Tabellini G (2000) Comparative politics and public finance. J Polit Econ 108(6):1121–1161 ISSN 00223808
Rege M (2004) Social norms and private provision of public goods. J Pub Econ Theory 6(1):65–77 ISSN 14679779
Rentmeester C (2010) A Kantian look at climate change. Essays Philos 11(1):76–86
Roemer JE (2010) Kantian equilibrium. Scand J Econ 112(1):1–24
Saha S (2007) Democratic institutions and provision of public good. University of California at Santa Barbara, Economics Working Paper Series qt55f3c17g, Department of Economics, UC Santa Barbara
Schumacher I (2009) The dynamics of environmentalism and the environment. Ecol Econ 68(11):2842–2849
Schumacher I (2014) An empirical study of the determinants of green party voting. Ecol Econ 105:306–318
Shafik N (1994) Economic development and environmental quality: an econometric analysis. Oxford Economic Papers, pp 757–773
Ulph DT, Daube MPK (2014) Moral behaviour, altruism and environmental policy
Wirl F (2011) Global warming with green and brown consumers. Scand J Econ 113(4):866–884
Zahran S, Brody SD, Grover H, Vedlitz A (2006) Climate change vulnerability and policy support. Soc Nat Resour 19(9):771–789
Acknowledgements
I want to thank Bertrand Wigniolle, Eugenio Figueroa, and Scott Barrett for their invaluable help and support. I would also like to thank Maria Kuecken and Alessandra Pizzo for their thorough revision, as well as the various conference and seminar participants, two anonymous referees and the editor who have helped me formulate and articulate these ideas.
Author information
Affiliations
Corresponding author
Appendices
Appendix
A ‘Kantian’ Index and ‘Optimistic’ Index Equivalence
Here I give an alternative interpretation of the modelled the diversity among agents. In particular we start from the case that each person cares about social wellbeing with the same intensity, but that the parameter \(\alpha \) instead reflects a person’s optimism. From this idea, we can use \(\alpha \) as the share of individuals within society that each person thinks (or hopes) will behave as he does. I show here that this approach is equivalent to the approach outlined in the main section of the paper. In the main text, the original utility function is:
Now we might think of \(\alpha \) as a ‘optimism’ measure instead of a ‘Kantian attitude’ measure. This means that \(\alpha \) will now reflect which proportion of the society the agent is expecting (or hoping) to behave as he does. In this new case, we get the following formulation, where \(V_i\) is the utility function in this case:
Following again the same reasoning of the main section, I posit that the agent will behave in a green manner if \(V_i(0) \ge V_i(1)\). We proceed again in the same fashion by rearranging terms and getting for both cases (original and new) the following conditions for green behaviour:
We can again define an \(\alpha ^*\) that divides the society into those whose behaviour is green and those whose behaviour is grey. The only thing left to do is check if this new function \(\theta _2(\cdot )\) has the same properties as the original one \(\theta (\cdot )\). Since the righthand sides of the inequalities are the same, I will only focus on the lefthand sides. In the original version, we had the following properties:
It is easy to verify that the same properties will hold for the case of \(K_2(\alpha _i, p_{t1})\). Hence we arrive at a new \(\theta _2(\cdot )\) with the same properties of \(\theta (\cdot )\) (although not the same function).
We can finally verify the two remarks made about \(\theta (\cdot )\) for \(\theta _2(\cdot )\):

There is a level of \(\Omega \, p_{t1} \; (\Omega p_{min})\) below which everyone exhibits grey behaviour:

Setting \(\alpha = 1 \quad \rightarrow \quad d(\Omega p_{min} + \gamma )  d(\Omega p_{min}) = H(\rho )\)

This will actually gives us the same level of \(\Omega p_{min}\) as before.


There will be always some people who exhibit grey behaviour:

We can again find some positive \(\alpha _i\), such that \(K_2(\alpha _i, \Omega p_{t1}) < H(\rho )\), for any given \({\Omega p_{t1} > 0}\) and \(\rho > 0\). In the same manner, we can see that since \(K_2(\cdot )\) is continuous in \(\alpha _i\) and that \(K_2(\alpha = 0, \Omega \, p_{t1}) = 0\), then there exists an \(\epsilon \) such that \(K_2(\epsilon , \Omega \, p_{t1}) < H(\rho )\) for given \(\Omega p_{t1} > 0\) and \(\rho > 0\) (Fig. 8).

We can finally observe a graph showing both versions of the function \(\theta (\cdot )\):
A third and final option might be to suggest that both mechanisms (how Kantian each person is and how optimistic they are) are in place. For simplicity, I skip this alternative. However, if an individual’s Kantian tendency is positively correlated to their optimism, the present results should hold.
B Equivalence of Utility Function with Daube and Ulph (2014)
The utility function used in the present paper is equivalent to the one used in Ulph and Daube (2014) (henceforth, DU14). To verify this equivalence, we can define the following:
where B and C are Benefits and Costs, respectively, as defined by DU14, but applied to damage and utility function in the present model. \(y^{REF}\) takes the role of \(\tilde{z}\) in DU14: it is some reference level of dirty consumption. U is the utility function from DU14, where the parameter \(\mu \) has a similar role as the parameter \(\alpha \) in the present model. DU14 maximize U with respect to y, taking \(y^s\) and \(y^{REF}\) as given. This gives not necessarily a corner solution with \(y=0\) or \(y=1\), since they allow agents to buy a mix of products. In the present work, I compare the binary option the agent has, with an wage equal to 1: \(U_{y=w=1}\) to \(U_{y=0,w=1}\). This gives:
which is the condition when the agent behaves green, as in inequality (2.7).
C The Cost of Behaving Green Under Other Consumption Utility Functions
Depending on the functional form of the consumption utility function, the cost of behaving in a green fashion \(H(\rho )\) can be an increasing, decreasing, or constant function of the income level w. To illustrate this feature, we can observe how \(H(\rho )\) behaviour changes by using a Constant Relative Risk Aversion (CRRA) utility function, as the following one:
where \(\epsilon \) measures the degree of relative risk aversion. Setting the grey and green consumption levels to w and \(w/(1+\rho )\) respectively, we get, for each part of the function:
And hence,
We can observe that \(H(\rho )\) is constant when \(\epsilon = 1\), meaning it is independent of the value of w. In the case where \(\epsilon \ne 1\), we have that if \(\epsilon < (>) 1\), then \(H(\rho )\) is increasing (decreasing) in w.
D Different Dynamics with Social Approval
As stated in Sect. 4, adding a social approval gain to the agent’s utility function does actually change (structurally) the function \(\theta (\cdot )\), and therefore the dynamics. Adding social approval to the utility function will lead to a different functional form in the sense that the new \(\tilde{\theta }(\cdot )\) function cannot be found by modifying the original parameters and/or by changing the (shape of the) consumption part \(\tilde{u}(\cdot )\).
In order to prove this, we recall that the original function \(\theta (\cdot )\) comes from Inequality 2.7. There, I proved that for a given value of \(\rho \) and a pollution level \(p_{t1}\), there will always be some people exhibiting grey behaviour. I will prove now that this is not the case with social pressure, since it can happen that for given values of \(\rho \) and \(p_{t1}\) (and \(\beta \)), the society can become completely green. Recall the new condition for green behaviour, as in Inequality 4.3 (page 27):
Let us verify whether this hypothesis of everyone behaving in a green manner (\(\alpha ^*_t = 0\)) is true. In this case, the social approval will be equal to: \(S(\alpha ^*_t = 0) = v(1)  v(0) = 1\). Therefore, the previous condition becomes:
Now, to verify that everyone is behaving in a green way, we can simply verify this condition for the least green person, the one purely motivated by economics, who has \(\alpha _i = 0\). Therefore we get:
which is just the minimum weight of the social approval parameter in the agents’ utility function. In other words, with this level of influence of peer pressure in agents’ utility, even the purely homooeconomicus person bears the cost of green behaviour; this is only because of social pressure, not because of a Kantian incentive, since he does not have one.
E Green Policy Details
In the case without policy, we have that the consumption levels of grey and green products per person are equal to 1 and \(1/(1+\rho )\), respectively. This is the direct result of having an income fixed to 1 and prices of 1 and \((1+\rho )\), respectively. When in the presence of a green policy, the grey product is taxed with a tax \(\tau \) and the green one is subsidized with a subsidy \(\sigma \). Hence the new consumptions (per agent) become \(1/(1+\tau )\) and \(1/(1+\rho \sigma )\). Assuming a balanced budget, and with \(\alpha ^*_t\) as the proportion of grey behaving people, we get:
We can see that the new value of \(\rho \) is \(\hat{\rho } = \rho  \tau \sigma \). Therefore, for a desired level of \(\hat{\rho }\) we can calculate \(\tau \) and \(\sigma \). Using equation (E.1.1) to substitute \(\sigma \) along with the previous equation, we get a quadratic equation for \(\tau \):
The remaining question is if the new curve \(\hat{\theta }(\cdot )\) (the one with the policy implemented) will be less than the original one, \(\theta (\cdot )\) (except for \(p_t\) values where \(\hat{\theta }(\cdot ) = 1\)). Recall that \(\theta (\cdot )\) comes from the division of the private cost of behaving green, \(H(\rho )\), and the social cost of behaving grey, K(p) (Eq. (2.8)). \(H(\rho )\) is the difference of utility of consumption when behaving grey and green. A direct result of the policy is a decline in this private cost, which can be verified using the following inequality:
Finally, K(p) is the social cost of behaving grey from the Kantian perspective (where the agent assumes that everyone behaves as he does). In particular, if the grey behaviour is chosen, the agent evaluates this term using the assumption that everyone is buying the grey product. This translate to have an infinitesimal value of \(\tau \) (using the limit yields to zero) and \(\sigma \approx \rho  \hat{\rho }\). Therefore, in the Kantian view, the grey consumer still consumes one unit of grey product and hence K(p) is unchanged. Therefore, \(\hat{\theta }(\cdot ) < \theta (\cdot )\).
Rights and permissions
About this article
Cite this article
Cerda Planas, L. Moving Toward Greener Societies: Moral Motivation and Green Behaviour. Environ Resource Econ 70, 835–860 (2018). https://doi.org/10.1007/s106400170167y
Accepted:
Published:
Issue Date:
Keywords
 Green behaviour
 Kantian morale
 Moral motivation
 Tipping points
JEL Classification
 C62
 D64
 D72
 H41
 Q50