📈 economics

Estimating Social Norm Complementarities

This paper develops and estimates a model of social norm choice incorporating both technological and social complementarities using data from Sierra Leone and Nigeria, revealing that norms like female genital cutting and child marriage exhibit strong complementarities while polygyny and child marriage show substitutability, thereby informing the design of effective legal and social policy interventions.

Original authors: Eliana La Ferrara, Cheaheon Lim, Davide Viviano

Published 2026-05-18

📖 6 min read🧠 Deep dive

CC0 1.0

Original authors: Eliana La Ferrara, Cheaheon Lim, Davide Viviano

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are walking through a village where everyone follows certain unwritten rules, or "social norms." Some people cut their daughters' hair in a specific way, some marry their daughters very young, and some men have multiple wives.

For a long time, economists studied these rules one by one, as if they were isolated islands. They asked, "What makes people follow the rule of early marriage?" or "What makes people follow the rule of cutting?" They assumed that changing one rule wouldn't really affect the others.

This paper argues that these rules are actually more like a bundle of groceries you buy together. Just as you might buy peanut butter and jelly because they taste better together, or because you need both to make a sandwich, people often follow multiple social rules at the same time because they reinforce each other.

The authors, Eliana La Ferrara, Cheaheon Lim, and Davide Viviano, built a mathematical model to figure out how these rules interact. They looked at data from Sierra Leone and Nigeria to see if changing one rule would make another rule stronger or weaker.

Here is how they broke it down using simple analogies:

1. The Two Ways Rules Connect

The authors say two rules can be connected in two different ways:

The "Technological" Connection (The Recipe): This is about the rules themselves. Do they make sense together?
- Analogy: Think of Peanut Butter and Jelly. If you eat peanut butter alone, it's okay. If you eat jelly alone, it's okay. But together, they make a perfect sandwich. The "utility" (happiness) of having both is greater than the sum of having them separately.
- In the paper: If a girl goes through a specific initiation ritual (like the "Bondo" society in Sierra Leone), it naturally prepares her for marriage. The ritual and the marriage are "technological complements" because one leads logically to the other.
The "Social" Connection (The Crowd): This is about what other people think. Do people get more pressure or more reward if they follow both rules?
- Analogy: Imagine a party where you get a prize for wearing a red hat. Now imagine a bigger prize if you wear a red hat and a blue scarf. If the prize for doing both is huge, people will do both. But if the prize for doing both is just the same as doing one, people might just pick the easiest one.
- In the paper: If a community punishes girls who don't follow the rules, do they punish girls who break both rules much more severely than girls who break just one? If yes, that's a "social complement."

2. The Big Findings: Complements vs. Substitutes

The authors found that the relationship between these rules changes depending on where you are and which rules you are looking at.

Case A: The "Best Friends" (Complements)

Where: Sierra Leone.
The Rules: Female Genital Cutting (FGC) and Child Marriage.
The Story: In Sierra Leone, these two rules are like Peanut Butter and Jelly. The authors found that if you try to stop one, the other suffers too.
Why: There is a secret society called "Bondo" that girls join during adolescence. This society initiates them into womanhood. Once they leave the bush, they are considered ready for marriage. The ritual and the marriage are so tightly linked that they reinforce each other. If you lower the value of the ritual, fewer people want to get married early, and vice versa.
The Result: They are Complements.

Case B: The "Rivals" (Substitutes)

Where: Nigeria.
The Rules: Polygyny (men having multiple wives) and Child Marriage.
The Story: In Nigeria, these two rules act like Coke and Pepsi. If you have one, you don't really need the other; in fact, having one might make the other less attractive.
Why: In Nigeria, if a girl marries very young, her parents might want her to be the first wife (the most powerful wife in the house). But if the family is already practicing polygyny, she might be forced to be the second or third wife, which is a lower status. Parents might think, "If we marry her young, we want her to be the main wife. If we marry her into a polygamous family, she won't be." So, the desire for a young marriage and the reality of a polygamous household pull in opposite directions.
The Result: They are Substitutes.

Case C: The "Strangers" (Independence)

Where: Sierra Leone (regarding Polygyny and Child Marriage).
The Story: Here, the rules don't really care about each other. They are like Toothpaste and a Bicycle. Buying one doesn't really change your desire for the other.
The Result: They are Independent.

3. Why This Matters for Policy (The "What If" Scenarios)

The authors used their model to simulate what would happen if the government tried to change these rules. This is like testing a new strategy in a video game before you play it for real.

If rules are Complements (Sierra Leone's FGC & Marriage):
- Strategy: If you want to stop child marriage, you can attack the ritual (FGC) instead of just attacking the marriage law.
- Result: Because they are best friends, hurting one hurts the other. The authors found that targeting the social pressure around the ritual was just as effective as targeting the marriage itself. It creates a "domino effect."
If rules are Substitutes (Nigeria's Polygyny & Marriage):
- Strategy: If you try to stop child marriage, you might accidentally make polygyny more popular, or vice versa.
- Result: Because they are rivals, if you make one option less attractive, people might just switch to the other. The authors found that policies in Nigeria had a much smaller "ripple effect" because the rules weren't helping each other.

Summary

The paper teaches us that social norms are a bundle, not a list.

Sometimes norms are teammates (Complements): Changing one changes the other.
Sometimes norms are rivals (Substitutes): Changing one might just make people switch to the other.
Sometimes they are strangers (Independent): Changing one doesn't matter for the other.

To fix a problem, you have to know which team the rules are on. If you treat rivals like teammates, your policy might fail. If you treat teammates like rivals, you might miss a huge opportunity to fix two problems at once.

Technical Summary: Estimating Social Norm Complementarities

1. Problem Statement

The existing literature on social norms typically analyzes the equilibrium, persistence, and policy responsiveness of individual norms in isolation. This paper argues that norms are often interdependent: the returns from adhering to one norm may depend on the prevalence of another. Specifically, the authors posit that norms can act as complements (where joint adherence yields returns exceeding the sum of individual returns) or substitutes (where adhering to one reduces the returns of adhering to another).

Standard discrete choice models with social interactions (e.g., Brock and Durlauf, 2001) generally treat actions as mutually exclusive or focus on single norms. This paper addresses the gap by developing a model of choice over bundles of non-exclusive norms. The core challenge is to disentangle two distinct channels through which interdependence arises:

Technological Complementarities: Intrinsic utility gains from adopting a bundle that exceed the sum of individual intrinsic utilities (analogous to consumption goods).
Social Complementarities: Returns from conformity that are super-additive when adopting a bundle compared to adopting norms individually.

Understanding the sign and magnitude of these cross-norm spillovers is critical for policy design. If norms are complements, a shock lowering adherence to one norm propagates to the other, amplifying policy effects. If they are substitutes, targeting one norm may induce a reallocation toward the other, potentially offsetting policy goals.

2. Methodology

Theoretical Framework

The authors propose a static choice model where agents select a vector $v \in \{ \emptyset, A, B, AB \}$ , representing adherence to norm A only, norm B only, both, or neither.

Utility Structure: The utility of a choice $v$ consists of an intrinsic term ( $\delta$ ), a social term (conformity returns based on the fraction of the population adhering to specific norms), and an idiosyncratic shock ( $z$ ).
Social Interactions: The model allows for non-linear social sanctions. If an agent adheres to both norms ($AB$), the penalty imposed on non-adherents (or the conformity return) can be super-additive ( $s_{AB} > s_A + s_B$ ) or sub-additive ( $s_{AB} < s_A + s_B$ ).
Equilibrium Characterization: The paper derives conditions for stable equilibria. Theorem 1 establishes that norms are complements if intrinsic joint utility is super-additive ( $\Gamma > 0$ ) or if social returns are super-additive ( $s_{AB} > s_A + s_B$ ) under specific conditions. Conversely, they are substitutes if intrinsic utility is sub-additive and/or social returns are sub-additive. This characterization nests the results of Gentzkow (2007) as a special case where social spillovers are absent.

Empirical Strategy

The model is estimated using repeated cross-sections from the Demographic and Health Surveys (DHS) in Sierra Leone (2008–2019) and Nigeria (2003–2024).

Norms Analyzed: The study focuses on two bundles:
1. Female Genital Cutting (FGC) and Child Marriage (CMA).
2. Polygyny (PGY) and Child Marriage (CMA).
Identification: The authors exploit exogenous shifts in the utility of child marriage to identify the parameters.
- Primary Shifter: The imposition of Child Marriage bans (Child Rights Act of 2007 in Sierra Leone; 2003 in Nigeria). This is assumed to shift the baseline utility of CMA ( $\delta_B$ ) but not FGC or PGY, conditional on covariates and fixed effects.
- Secondary Shifter (Nigeria): Exposure to droughts during teenage years, which increases incentives for child marriage in bride-price societies but is assumed not to affect the intrinsic utility of FGC or PGY.
Estimation: The authors use maximum likelihood estimation with a dynamic specification where utilities depend on lagged adoption shares within ethnic and geographic groups. They employ a smoothing approximation for the indicator functions in the choice probabilities and use numerical integration (quadrature) for the idiosyncratic shocks. Inference accounts for the "generated regressor" problem arising from using estimated group-level adoption shares.

3. Key Results

Sierra Leone: FGC and Child Marriage

Social Multipliers: Both norms exhibit positive and significant social multipliers. The multiplier for FGC ( $s_A \approx 2.46$ ) is substantially larger than for Child Marriage ( $s_B \approx 0.36$ ), consistent with FGC being a highly coordinated practice tied to the "Bondo" secret society.
Complementarity: The estimated technological complementarity parameter ( $\Gamma$ ) is positive and significant ( $\approx 1.05$ ). Combined with linear (or near-linear) social returns, the model identifies FGC and Child Marriage as strong complements.
Mechanism: The authors interpret this via the "Bondo" institution, where initiation into the secret society (FGC) occurs in adolescence and serves as a sequential prerequisite for marriage, creating a natural bundling of the two practices.

Nigeria: Polygyny and Child Marriage

Social Multipliers: Both norms show significant social multipliers ( $s_{PGY} \approx 2.45$ , $s_{CMA} \approx 1.26$ ).
Substitutability: The technological complementarity parameter ( $\Gamma$ ) is statistically indistinguishable from zero. Crucially, the social return to joint adoption is sub-additive ( $s_{AB} < s_A + s_B$ ). According to the theoretical characterization, this implies that Polygyny and Child Marriage are social substitutes.
Mechanism: The authors attribute this to the status dynamics within polygynous unions. While a husband gains status from multiple wives, a young bride entering as a second or third wife (non-first-ranked) faces a lower status and subordinate role. Parents may therefore be less inclined to marry a daughter into a polygynous household, creating a substitution effect.

Policy Counterfactuals

The authors simulate policy interventions to assess equilibrium dynamics:

Sierra Leone (Complements): Targeting the intrinsic utility of FGC (e.g., via a ban or awareness campaign) generates large, cumulative reductions in both FGC and Child Marriage prevalence due to the complementarity. Reducing the social multiplier for FGC is also highly effective. In contrast, targeting Child Marriage alone has a smaller effect and induces a reallocation toward FGC-only adoption.
Nigeria (Substitutes): Interventions targeting either Polygyny or Child Marriage yield smaller equilibrium effects. Because the norms are substitutes, reducing the prevalence of one does not strongly propagate to the other; in some cases, it may induce a reallocation toward the other norm, dampening the overall policy impact.

4. Contributions and Significance

Theoretical Contribution:
The paper extends the literature on discrete choice with social interactions (e.g., Brock and Durlauf, 2001) by allowing for bundled, non-exclusive choices. It formalizes the decomposition of complementarity into "technological" (intrinsic) and "social" (conformity) components, providing a rigorous characterization of when norms act as complements or substitutes in equilibrium.

Empirical Contribution:
By applying this framework to gender norms in West Africa, the paper demonstrates that the interdependence of norms is not uniform. It finds that the same pair of norms (FGC and CMA) can be strong complements in one context (Sierra Leone) due to specific institutional arrangements (Bondo), while other pairs (PGY and CMA) act as substitutes in another (Nigeria) due to status dynamics.

Policy Implications:
The study highlights that the effectiveness of norm-change interventions depends critically on the underlying structure of complementarities.

In settings with complementarities, targeting a "hub" norm with a high social multiplier can trigger a cascade, reducing the prevalence of multiple harmful practices simultaneously.
In settings with substitutabilities, policies must be carefully designed to avoid unintended reallocation effects where reducing one norm inadvertently increases the prevalence of another.

The authors conclude that their framework offers a general tool for analyzing choice settings where individual utility includes both intrinsic and social payoffs, suggesting that future research could extend this to higher-dimensional bundles and a broader range of societal norms.