Religious and secular ethics offer complementary strategies to achieve environmental sustainability

The paper frames environmental sustainability as an ethical issue and notes growing attention to ethics in achieving sustainability. It distinguishes two main groups of environmental ethics: secular ethics (SEC) and religious ethics (REL). SEC emphasizes responsibility to nature, duties to future generations, rights of humans and non-humans, and inter- and intra-generational equity. REL varies by religion (e.g., Buddhism on equilibrium, Christianity on love of neighbor, Hinduism on equal dignity of humans and non-humans, Islam on trusteeship and parsimony, Judaism on stewardship). Failures of climate agreements suggest a practical problem: ethical principles must provide feasible and reliable incentives that lead to realistic equilibrium conditions. Empirical models can test consequences of applying ethical concepts. The empirical literature on ethics and actual environmental behavior (as opposed to self-reports) is limited, especially multi-country studies covering entire populations. Prior work by the author showed beneficial impacts of majority religions and of key SEC principles on global sustainability (measured by the ecological footprint) and on local sustainability proxies (waste management, organic food purchases, energy conservation). Environmental ethics affects behaviors through individual feelings and social pressures. Purpose: to derive theoretical insights into the feasibility of REL and SEC for environmental sustainability using a dynamic model (evolutionary game with replicator dynamics) and to test 12 hypotheses with 32 random- and between-effects regressions on a panel dataset covering 23 religions/denominations and 3 SEC indicators across 181–183 countries. Three insights guide the hypotheses: (1) REL more likely to shape many behaviors for global sustainability, SEC some behaviors for local sustainability; (2) REL more likely to work via individual feelings and social pressures from small communities, SEC via social pressures from the whole society; (3) REL better at short-run collective action in small communities, SEC better at long-run society-wide change. These are tested using national-level data and distinguishing individual determinants (precepts and principles) and social determinants (reciprocity, religion majorities, religion shares) across behaviors requiring time, money, or both, and an overall footprint measure.

The introduction surveys work linking ethics and environmental sustainability, noting recent emphasis on ethics (e.g., Menning 2016; Lenzi 2017; Spahn 2018; Whiting and Konstantakos 2019; Batavia et al. 2020). Empirical studies based on religious ethics include Peifer et al. (2016), Arli and Tjiptono (2017), Yang and Huang (2018); on secular ethics include Yuan et al. (2017), Sorkun (2018), Khan et al. (2019). Multi-country, population-wide analyses remain limited. Prior studies by the author found: (i) majority religions can positively affect global sustainability (Zagonari 2020a); (ii) three SEC principles (intra-generational equity, responsibility to nature, responsibility to future generations) can benefit global sustainability (Zagonari 2019a); (iii) both majority religions and SEC principles can aid local sustainability (Zagonari 2020c). Literature on ethical solutions to environmental collective-action problems is nascent; Bolis et al. (2017) found few studies supporting value-oriented (substantive) rationality transitions. The paper positions teleological and deontological approaches, environmental pragmatism, and virtue ethics as conceptual backdrops for the dynamic framework.

Methods comprise a theoretical model for feasibility and an empirical panel-data strategy for reliability. Theoretical framework: A dynamic evolutionary game with replicator dynamics models interactions between pro- and anti-environmental behaviors. Payoffs are defined as Reward (R), Sucker (S), Temptation (T), and Punishment (P), capturing both religious and secular ethical valuations: SEC responsibility to nature increases R; distributive justice reduces T and increases S; responsibility to future generations reduces P. REL analogs: sacredness of nature increases R; concern for current community reduces T; concern for future community increases S; parsimony reduces P. Expected utilities for pro- and anti-environmental actions are U_E = (1−α)R + αS and U_A = (1−α)P + αT, with α the probability of choosing a transform (pro-environmental) strategy. Average utility U = αU_E + (1−α)U_A. Replicator dynamics: ∂α/∂t = α(1−α)[(R−P)(1−α) + α(S−T)]. Equilibria: α1=0, α3=1, and α2=(P−R)/(P−R+S−T), under conditions c1: P−R > T−S, T < S, or c2: P−R < T−S, T > S. In unsustainable societies, T>S initially (c2). Dynamics indicate when α1, α2, α3 are stable/unstable, with basins of attraction depending on R, P, S, T. REL is theorized to more strongly alter S and T (promoting T<S), yielding short-run increases and convergence to pro-environmental behavior (Fig. 2). SEC is theorized to more strongly alter R and P, enlarging α2 and producing long-run, partial adoption (Fig. 1). Insights: (1) REL affects many behaviors relevant to global sustainability; SEC some behaviors for local sustainability. (2) REL works via individual feelings and social enforcement in small and large communities; SEC via secular principles and whole-society pressures. (3) REL solves short-run collective action in small communities; SEC drives long-run increases society-wide. Data and empirical strategy: National panel data (1995–2017) on local behaviors—household waste management (WM), organic food purchases (OF), household energy conservation (EC)—and global behavior—ecological footprint (EF). Dependent variables include levels (short-run, SHO) and yearly changes (long-run, LON). Data sources harmonized across EU, OECD, World Bank, FAO, and Global Footprint Network. SEC variables: Gini coefficient (ine) for reciprocity/inequality, GDP share for environmental protection (cur) for concern for current environment (increasing R), and GDP share for environmental R&D (fut) for concern for future environment (reducing P). REL variables: for minority religions, percentages of adherents (individual-level effects); for majority religions, country-level dummies (>50% adherents; social pressure from large communities). Religion data from worldreligions.org; 23 religions/denominations considered, also aggregated contexts with 14, and majority-only contexts with 9 and 5. Controls include income, education, unemployment, and share over 65 years. Estimation: random-effects regressions for minority religions (emphasizing individual determinants) and for SEC; between-effects regressions for majority religions (social pressures). Reliability defined as statistically significant beneficial impacts at 10% level. Twelve hypotheses (combinations of the three insights) tested using relative success rates and Bernoulli tests comparing frequencies of significant effects.

Analytical findings: REL is overall more feasible than SEC. REL more likely to impact many pro-environmental behaviors (via increasing R and S, decreasing T and P), supporting global sustainability and short-run convergence; SEC more likely to influence some behaviors (via increasing R and decreasing P), supporting local sustainability and long-run trends. Statistical findings (selected results from Tables 4–7 and remarks):

• Global sustainability (EF, ΔEF): For individual-level effects (minority religions), Confucianism and Sikhism (14- and 23-religion contexts) significantly reduced EF; Vaishnavism within Hinduism had significant beneficial EF impacts. For changes in EF, Christianity (5-religion context) and Catholicism (9-religion context) showed significant beneficial effects. SEC variables showed no significant beneficial effects on EF or ΔEF.
• Local sustainability (WM, AWM): Multiple REL variables were significant, including Buddhist denominations (Mahayana, Theravada), Christian denominations (Catholic, Independent, Protestant), Muslim denominations (Shia, Sunni), and Baha’i and Ethnic religions for WM. SEC: fut (environmental R&D) positively affected WM and AWM in all 8 examined cells; cur had positive effects in 4 of 8 WM/AWM contexts. ine (Gini) positively associated with WM/AWM in 3 of 8 cells (interpreted as reciprocity where social pressures matter).
• Organic food purchases (OF, ΔOF): REL effects were sparser; Baha’i and Ethnic religions showed significance for OF in individual-level analyses; some Christian and Buddhist denominations had mixed signs across contexts. SEC: cur positively affected OF and ΔOF in 6 of 8 cells; fut had positive effects in 2 of 8; ine negatively affected OF/ΔOF in 2 of 8 cells (inequality effect consistent with OF as a luxury good).
• Energy conservation (EC, ΔEC): No SEC variables had significant beneficial effects on EC or ΔEC; no REL or SEC had significant effects on ΔEC. REL impacts on EC levels were limited and not beneficial overall.
• Relative success rates and hypothesis tests: Bernoulli tests (significance threshold z>1.64, 10%): Hypothesis set 1 (global vs local): 1b SEC>REL for LOC (z=5.21) supported; 1d SEC LOC>GLO (z=3.18) supported; 1a REL>SEC for GLO not significant (z=1.42); 1c REL GLO>LOC opposite sign (z=-0.31). Hypothesis set 2 (individual vs social): 2b SEC>SOC (z=5.70) supported; 2c REL IND>SOC (z=3.01) supported; 2d SEC SOC>IND (z=1.67) supported; 2a REL>SEC for IND opposite sign (z=-0.72). Hypothesis set 3 (short vs long run): 3b SEC>REL for LON (z=6.18) supported; 3c REL SHO>LON (z=4.32) supported; 3a REL>SEC for SHO opposite sign (z=-0.67); 3d SEC LON>SHO not significant (z=1.11). Overall, hypotheses supported in 2/4 (set 1), 3/4 (set 2), and 2/4 (set 3).
• Additional patterns: Minority religions are more relevant for individual feelings than majority religions are for social pressures (success rates 30/222 vs 5/84). Minority religions are more relevant when more religions are considered (18/138) than when fewer are considered (10/84). No SEC positively affected EC or ΔEC. Religions more often significant when both individual feelings and social pressures are involved (e.g., WM) than where mainly individual feelings matter (e.g., OF). Synthesis: REL and SEC are complementary—REL more reliable for short-run effects and individual feelings; SEC more reliable for long-run effects and social pressures, and more reliable for local than global sustainability.

Findings address the central question by showing that ethical systems influence pro-environmental behavior along different mechanisms and timescales. REL tends to mobilize individual feelings and small-community social enforcement, solving collective-action problems rapidly and improving levels of behaviors across multiple domains, aligning with the theoretical T<S dynamic and convergence. SEC, implemented via societal institutions and policies (e.g., environmental protection budgets, R&D), gradually shifts incentives (increasing R, reducing P), improving changes in behaviors particularly for local actions (WM, OF). The absence of SEC effects on global EF and EC underscores limits of broad societal levers without community-level enforcement or theological motivations. The complementarity across time (REL short-run; SEC long-run), space (SEC more reliable locally than globally), and social structure (REL individual feelings; SEC societal pressures) suggests combined strategies. Education and information campaigns can strengthen SEC pathways at the whole-society level; theological guidance, sermons, and community religious leadership can activate REL pathways in communities of all sizes. The results align with global citizenship perspectives that do not require a single global ideology but leverage diverse ethical traditions to advance sustainability.

This study integrates a dynamic evolutionary model with multinational panel evidence to compare religious and secular ethics as feasible and reliable strategies for environmental sustainability. Methodologically, it links teleological and deontological determinants to replicator dynamics and operationalizes REL and SEC through national indicators and religious composition. Empirically, REL more effectively elevates levels of diverse pro-environmental behaviors in the short run through individual feelings and community pressures, contributing to global sustainability. SEC more effectively drives long-run changes in selected local behaviors via societal levers (education, protection spending, R&D). The two are complementary across time, space, and social scale. Operationally, policymakers should pair whole-society education/information (SEC) with theological development and religious community engagement (REL). Future research should incorporate additional SEC indicators (e.g., concern for future generations via public/private debt measures, consumption attitudes), more control variables (e.g., gender), and finer-grained identification of religious precepts (e.g., parsimony, equilibrium) to refine causal pathways.

Two primary limitations are noted: (1) The number of religions/denominations (5, 9, 14, or 23 across contexts) is modest for asymptotic assumptions in Bernoulli comparisons of success frequencies; finer-grained global data on sects and communities are unavailable. (2) The number of secular ethics indicators is small (three: ine, cur, fut); additional consistent global measures (e.g., debt shares as proxies for concern for future generations) are lacking. Measurement error in religion variables and potential denomination-affiliation biases are discussed; panel methods mitigate but do not eliminate such issues.