In plain sight? Reconsidering the linkage between brideprice and violent conflict 1

Abstract

Persistent brideprice inflation has been linked to greater political violence. However, empirically testing this argument is complicated by the paucity of data on brideprice. We argue that despite the lack of over-time brideprice data, one can proxy for variation in marriage markets using changes to population, economic growth, and marriage rates themselves, thereby offering a clearer test of the brideprice–violence relationship. Our analysis suggests that there is little empirical support for such a relationship, and concludes that the previous support was largely due to data limitations and omitted confounds.

Keywords

Brideprice gender political violence terrorism

Introduction

Why men rebel is a classic question in the literature on political violence (Gurr, 1970). Over time, this literature has shifted between structural (e.g. Fearon and Laitin, 2003; Kalyvas and Balcells, 2010; Thies, 2010) and individual or group-level explanations (e.g. Gilligan et al., 2013; Hirose et al., 2017; Wood and Thomas, 2017). The past decade has seen movement toward increasingly micro-level explanations of why individuals engage in rebellion, terrorist, or other antiestablishment activities (Arce et al., 2011; Hatemi and McDermott, 2012; De Juan and Pierskalla, 2016). An intriguing addition to this literature argues that brideprice may be a critical factor in motivating young men to join groups whose purpose is to engage in organized violence. In a widely-discussed article, Hudson and Matfess (2017) suggest that brideprice is so clearly a cause that propels young men toward violence that it has been “hidden in plain sight” all along.

Specifically, Hudson and Matfess (2017) argue that brideprice inflation generates grievances among young males in patrilineal societies—as they are excluded from the increasingly costly marriage market—making them more likely to join rebellion. If correct, this is an important addition to our understanding of the root causes of political violence, as it would allow for effective policy responses. First, brideprice is something that can potentially be managed through policy intervention, unlike many well-known structural factors that move slowly over time. The “going rate” for brideprice is subject to market forces in countries where it is practiced.² Markets can be regulated, thereby offering an opportunity for states, non-governmental organizations (NGOs), and international NGOs to become involved in efforts to manage the price or offset it through other means. Second, such interventions offer the potential to neutralize a seemingly effective rebel and terrorist group recruitment strategy—arranging for wives for men who lack the resources to pay for marriage contracts or providing the resources to pay for the brideprice in their country.

Given these possible benefits, it is important to determine the extent of the relationship between brideprice and political violence. As such, we undertake a thorough examination of Hudson and Matfess’s (2017) argument and empirical analysis. We find that they offer a complex and occasionally confusing series of causal claims, alternating between a supply-side theory and a more norms-based argument. While both could operate simultaneously, the lack of clarity on this point makes it difficult to generate clear empirical expectations. More importantly, Hudson and Matfess (2017) seemingly vascillate between conceptual discussions of brideprice societies (those in which marital norms entail the net transfer of assets from grooms or their families to brides or their families) and brideprice inflation (the amount of assets demanded in such an exchange). Specifically, they offer a theory of violence for the latter, a quantitative empirical test of the former, and qualitative illustrations of the latter in three case studies.

To more directly evaluate the relationship between brideprice and violence, we clarify this distinction theoretically and offer an empirical test of both. First, we assess whether brideprice societies are more violent, extending the analysis of Hudson and Matfess (2017). We find that, after the inclusion of several common predictors of political violence (e.g. development, population, and democracy), there is no longer a clear relationship between brideprice and political violence—suggesting that the prior finding was an artifact of these omitted confounds. Second, while evaluating the brideprice inflation mechanism directly is difficult owing to data limitations, we exploit over-time variation in other factors (e.g. economic growth, population growth, and marriage rates) that influence brideprice markets. For example, when brideprice societies have population bulges, this generates a surplus in the stock of available young brides and drives down prices (Anderson, 2007). As such, this should offer a more direct test of the theory given in Hudson and Matfess (2017). However, this analysis also fails to uncover a robust relationship between brideprice inflation and violence.

In sum, we find little empirical support for a consistent linkage between brideprice and political violence. While this could, of course, be a function of data limitations—such as measurement error in our proxies for brideprice inflation—we argue that until researchers have better data to bring to bear on this question, we should be cautious in our claims of any purported linkage. This is in stark contrast to the current practice in academic and popular literature where, on the basis of the analysis provided by Hudson and Matfess (2017), we have seen articles like “Why polygamy breeds civil war: When large numbers of men are doomed to bachelorhood, they get desperate” (The Economist, 19 March 2018) and “How bride prices drive terrorism recruitment in Middle East, Africa, Asia” (USA Today, 11 August 2017).

Given the policy consequences that might arise from the emergence and diffusion of such beliefs, we need to be especially cautious. As we noted above, if the brideprice–violence link were substantiated it would suggest policy interventions that would be attractive to states, NGOs, and international NGOs. However, as Ish-Shalom (2013) has demonstrated for the democratic peace thesis, scholars must be very careful about the way in which our research may be used to justify policy interventions. This is especially true in emergent issue areas, such as the relationship between brideprice and violence, where a critical consensus has not yet been reached among scholars.

Hidden in plain sight? Brideprice as a cause of violence

Hudson and Matfess (2017: 7) begin their article with an anecdote of a young Indian man involved in the four days of attacks in Mumbai in 2009. When interrogated by the police, he confessed to having joined Lashkar-e-Taiba in order to obtain money so that his family could pay the costs of marriage contracts for his siblings. This motivating example suggests that brideprice is a contributing factor to the material rationale for joining terrorist organizations. Hudson and Matfess (2017: 8) define brideprice as “an overall net transfer of assets from the groom’s family to the bride’s family.” Further, it is paid to offset “the cost to the natal family of raising the bride” (p. 12). Their main argument is that inflationary brideprice leads to marriage market obstruction, diminishing the ability of young men (and their families) to cover brideprice and, in turn, increasing the likelihood that they will join terrorist groups willing to cover such costs. The authors extend this argument, suggesting that inflationary brideprice may lead young men to become engaged in all manner of organized political violence, including “terrorism, rebellion, intergroup aggression, raiding, and insurrection” (p. 8). Thus, brideprice and the resulting marriage market obstruction are critical factors “hidden in plain sight” that connect young men to organized violence.³

In a preliminary test of this argument, Hudson and Matfess (2017) assess whether brideprice societies are more prone to violence, concluding strong support for a linkage between the two. There are two issues with the empirical support they provide. First, there is little correspondence between the theoretical argument and the empirical test. Namely, this analysis offers little leverage over their supply-side mechanism—that aggrieved young men join rebellion owing to an inability to pay increasingly high brideprice. While there may be other reasons to suspect that brideprice societies are more likely to experience violence, which we explore below, these are distinct from the grievance-based arguments at the core of their theory. Second, even if we confine attention to the relationship between brideprice societies and violence, the empirical analysis fails to address many possible threats to inference: only a single year of data is evaluated, no potential confounds are considered, and the dependent variable includes many factors (relations with neighboring countries, size of jailed population, military sophistication) that are unrelated to rebel-based violence.⁴

The quantitative analysis of Hudson and Matfess (2017) essentially consists of two elements: (1) geospatial visualizations of the data and (2) bivariate statistical analyses of brideprice and the Global Peace Index (GPI). While, as admitted by the authors, such analyses are exploratory, they give us reasons for concern. While maps of bridepice (see Figure 1) and GPI (see Figure 2), do indicate that the prevalence of brideprice (darker colors in Figure 1) is highest in regions where violence is highest (darker colors in Figure 2), there are many reasons why this may occur.⁵ In short, the co-location of two predictors does not necessarily imply a relationship between the two, just as co-trending time series do not necessarily imply such a relationship. Of particular concern in this case is whether other determinants of violence may also be abundant in these areas, confounding the seemingly apparent relationship between brideprice and GPI.⁶ Similar issues arise in their subsequent selection of cases for qualitative analysis.

Figure 1.

Brideprice type/prevalence using WomanStats MARR-Scale-3 data, 2016.

Figure 2.

Global Peace Index data by quartile, 2016.

The bivariate analysis that Hudson and Matfess (2017) undertake is unable to offer additional protection against this possibility. While cross tabs are useful as a descriptive exercise, associated statistical tests (i.e. χ² tests) rely on independence assumptions for validity. In Table 1 we replicate Hudson and Matfess’s (2017) crosstab of brideprice on political violence, rounding GPI to the nearest unit value.⁷ From this Hudson and Matfess (2017: 23) conclude that:

The results of this cross tabulation are striking (see table 1): no society with brideprice fell into the most peaceful quartile of this sample of 163 nation-states. No society without brideprice fell into the least peaceful quartile of the sample.

Table 1.

Crosstab of brideprice and political violence (using rounding), 2016.

	Global Peace Index
Brideprice	1	2	3	4
No	16	55	10	0
Yes	0	66	12	4

Note: GPI rounded to the nearest whole number. This replicates table 1 of Hudson and Matfess (2017).

Strictly speaking, this finding is not supported by their results, which instead indicate that the four least peaceful countries have brideprice and the 16 most peaceful countries do not. When we divide the observations into quartiles, however, we instead see that 11 (of 40) countries in the least peaceful quartile do not have brideprice, and seven (of 41) countries in the most peace quartile do.

While Table 2 would still indicate a positive association between brideprice and GPI (χ² = 27.3; p-value ≤ 0.000), this assumes that countries with and without brideprice are otherwise balanced on the other determinants of violence. Given the spatial distribution of each—with both high brideprice and GPI clustering in Africa and the Middle East—this seems unlikely, as these areas also have low levels of development and weak political institutions, both of which are strongly associated with violence (Hegre and Sambanis, 2006). In light of this, we run a series of regression models to assess whether the brideprice–violence link is robust to the inclusion of common country-level predictors of violence (see Table 3). First, in Model 1, we estimate a simple linear regression model of GPI on a binary measure of brideprice (equivalent to an asymptotic t-test), which indicates a strong positive and significant effect, as in Hudson and Matfess (2017). However, in Model 2 when we include a common set of controls—GDP, Population, and Democracy—the relationship between brideprice and violence is now negative and insignificant. Moreover, excluding brideprice in Model 3 actually improves our relative model fit, as indicated by adjusted-R² (from 0.371 to 0.374). These cross-sectional data do not support a robust linkage between brideprice and violence.

Table 2.

Crosstab of brideprice and political violence (using quartiles), 2016.

	Global Peace Index
Brideprice	I	II	III	IV
No	34	19	17	11
Yes	7	22	24	29

Note: GPI divided into quartiles (breaks occur at 1.78, 2.04, 2.29).

Table 3.

Regression(s) of political violence on brideprice, 2016.

	Model 1	Model 2	Model 3
Brideprice	0.381** (0.075)	−0.037 (0.092)
GDP (logged)		−0.159**(0.026)	−0.151**(0.022)
Population (logged)		0.237**(0.033)	0.231**(0.030)
Democracy		−0.021**(0.007)	−0.019**(0.005)
Constant	1.191**(0.053)	1.992**(0.434)	2.130**(0.410)
N	163	154	154
Adj-R²	0.132	0.371	0.374

Note: *p < 0.05; **p < 0.01.

Our re-analysis of Hudson and Matfess (2017) does not support a robust link between brideprice and violence. However, recall that the core theoretical argument they offer is about briedprice inflation and violence, so it may be unsurprising that we (and they) are unable to recover a robust relationship in the above tests. Their case studies do not directly assess the same relationship examined in their cross-tabs (of brideprice society and violence), but instead focus on the use of brideprice inflation as a recruitment strategy by groups like Boko Haram, who arrange for wives for young men who cannot afford the going rate for brideprice. Hudson and Matfess (2017) also discuss similar practices by the Sudanese People’s Liberation Army. They conclude with a description of Saudi Arabia’s attempts to cap brideprice and reduce the cost of weddings. As Hudson and Matfess (2017) note, these attempts seem to have not thwarted the relatively high number of Saudi Arabian men who have joined terrorist organizations. Their defense against this apparently contradictory evidence is the unsubstantiated counterfactual that without bride price caps, the number of young men joining terrorist groups would be much higher.⁸

In the next section, we provide a more thorough discussion of their theory, identifying testable implications that motivate a clearer test of their main logic.

Brideprice inflation as a cause of violence

The theory offered by Hudson and Matfess (2017) is quite complex—indicating that brideprice societies are both causes and consequences of conditions that make violence more likely. They suggest that deeper, historical factors operate in societies with brideprice customs, fostering an environment in which violence is more likely to emerge. For example, the authors argue that poverty and social marginalization alone are insufficient to explain violence, yet may work through brideprice to predispose young men to become involved in organized violence (p. 8). The salience of brideprice as a motivation for participation is rooted, it is argued, in patrilinear culture, in which “the marriage imperative is thus deeply felt among males” as a source of inheritance and lineage giving rise to both brideprice and violence (p. 12). In short, there is an “intensified meaning” of marriage in “more strongly patrilinear societies” (p. 12).

Yet the authors also argue both here and in Hudson et al. (2015: 14) that patrilineality directly produces “an inherently unstable society prone to violent conflict and rentierism”. As a result, the influence of brideprice is not always clear. Does brideprice as a marital custom directly affect violence? Is brideprice exchange a consequence of patrilineality, along with violence, or a conditioning factor in its effect on violence? Failing to clarify these relationships means that, at best, we are likely to underestimate and therefore misunderstand the effect of strong patrilineality, brideprice, and violence. At worst, the brideprice–violence linkage may be spurious, with strong patrilineality producing both.

A clearer argument emerges for the role of persistent brideprice inflation in rebel recruitment and, in turn, violence. In short, Hudson and Matfess (2017: 15) note that the “going rate” for a bride may vary and that “men are highly sensitive to new trends in brideprice.” That is, as brideprice increases, the number of men able to successfully marry decreases. Brideprice inflation is particularly likely “in situations of economic stagnation, rising inequality, or both,” thereby leading to delayed marriage. This is compounded further as an unequal distribution of wealth facilitates polygyny, distorting the marriage market as wealthy men are able to marry multiple wives. Finally, higher female mortality found in brideprice societies is also likely to contribute to marriage market distortion. Taken together, they refer to these three factors as the “patrilineal syndrome” (p. 18).

Therefore, shocks to some feature of this patrilineal syndrome (economic stagnation, polygymy, and/or female mortality) affect brideprice, thereby obstructing marriage markets and heightening grievances (over lack of access to women) among young men. Groups then use this as an opportunity to target and exploit young men who lack sufficient assets to marry. Such groups may either arrange low-cost marriages or simply provide the resources needed for men to pay the brideprice at the going rate. When groups are more successful at recruitment, they are better able to inflict violence in service of their goals (Hudson and Matfess, 2017: 18).

Ultimately, summarizing and clarifying this discussion, we end up with the causal chain given in Figure 3. Strongly patrilineal societies lead to the patrilineal syndrome of inflationary brideprice given situations of economic inequality or stagnation, polygyny, and higher female mortality. These three factors of the patrilineal syndrome produce marriage market obstruction. It is this marriage market obstruction, then, that leads to heightened grievances among young males. Observing such grievances, anti-establishment groups see recruitment opportunities and young males are willing to join for the material resources to secure wives. The added capacity from such recruitment allows groups to engage in greater violence. In sum, Hudson and Matfess (2017) argue inflationary changes to brideprice—not simply its presence—disrupt marriage markets, inducing grievances among young men.

Figure 3.

Marriage markets and political violence in strong patrilineal societies.

How, then, in the absence of brideprice data, can we test these claims? We argue that several empirical implications arise from Hudson and Matfess (2017) that can be tested even in the absence of brideprice data. First, as seen in Figure 3, brideprice inflation is linked to violence via marriage market obstruction, that is, fewer young men are able to secure wives. It is the inability to secure a wife that lies at the core of Hudson and Matfess’s (2017) grievance-based argument. If correct, this suggests that looking at marriage rates offers an alternative and more direct test of their argument. In short, where the proportion of unmarried men outnumbers the proportion of unmarried women in a society in patrilineal societies, we should be most likely to observe violence. Therefore,

H1: Brideprice societies should be more likely to experience political violence when there are higher rates of unmarried young men.

Related to this, research in economics and anthropology indicates that population bulges are likely to induce marriage market distortions in patrilineal societies (Anderson, 2007; Mulder, 1995; Rao, 1993). In short, population bulges create a “marriage squeeze” because, on average, men marry younger women (Rao, 1993). As a consequence, population bulges create a surplus of marriageable age women (prior to their male cohorts obtaining marriageable ages), thereby suppressing brideprice. If correct, Hudson and Matfess’s (2017) argument would indicate that violence should be less likely in patrilineal societies during population bulges. Therefore,

H2: Brideprice societies should be less likely to experience political violence during population bulges.

Finally, changing economic conditions should also affect the brideprice market. During periods of economic stagnation or negative growth, there is a shift in the demand curve—the number of potential grooms (or their families) who can afford brideprice (i.e. demand) diminishes. Moreover, the distributional consequences of negative growth affect lower classes more acutely, as they have less disposable income or savings, which are now required for essential goods. This is exactly the type of aggrieved individual who Hudson and Matfess (2017) argue should be targeted by rebel groups. Therefore,

H3: Brideprice societies should be more likely to experience political violence during economic stagnation or negative growth.

There are other possible determinants of brideprice that have been suggested—both in Hudson and Matfess (2017) and elsewhere. Yet we forgo a discussion of these either because of a lack of available data (e.g. village-level inequality) or because they are almost perfectly correlated with brideprice societies (e.g. polygyny).⁹ In the next section we detail our empirical strategy for evaluating the above hypotheses.

Data and methods

To systematically analyze the brideprice–violence link, we specify a range of models, varying both the measure of violence (i.e. terrorism, civil conflict) and the proxy for brideprice inflation (i.e. marriage difference, youth bulge, and economic growth). Hudson and Matfess (2017) used the GPI as their measure for political violence; however, there are two problems with using this measure in our analysis. First, GPI values only date back to 2007—when the first report was released—limiting its utility in explaining the type of over-time variation that is of interest. Second, there is a problem of construct validity as many of the inputs in the GPI are not germane to the theory offered by Hudson and Matfess (2017), which makes no claims about the relationship between brideprice and the “number of internal security officers and police per 100,000 people” or “financial contributions to UN peacekeeping missions.” This composite measure cannot offer a clean test of the relationship between brideprice and those facets of political violence—“terrorism, rebellion, intergroup aggression, raiding, and insurrection”—identified by Hudson and Matfess (2017: 8).^10,11

Instead, we consider two other measures of political violence that are widely used in conflict studies. First, we use Terrorist Incidents, which is a count of the total number of attacks that take place in a given country-year. The data on terrorist attacks come from the Global Terrorism Database (START, 2018: 9) which defines an attack as “threatened or actual use of illegal force and violence by a non-state actor to attain a political, economic, religious, or social goal through fear, coercion, or intimidation.” Each unique event meeting these criteria enters the dataset, from which we generate yearly totals for each state.¹² In keeping with the existing literature on terrorism, we estimate these models using negative binomial regression to model overdisperion in these data (Li, 2005; Wilson and Piazza, 2013; Walsh and Piazza, 2010).¹³

For our second measure of political violence, we use Civil Conflict incidence from the PRIO/Uppsala Armed Conflict Dataset (Gleditsch et al., 2002). This is defined as the occurrence of a contested incompatibility involving at least two organized parties of which at least one is a recognized government, over a stated political incompatibility, in which at least 25 people are killed in battle-related circumstances. Civil Conflict is a dichotomous variable, coded as “1” during the civil conflict, and “0” for all other years.^14,15

In all models we include Brideprice, a binary measure coded as “1” if net assests move from the groom/groom’s family to the bride/bride’s family according to the WomanStats MARR-Scale-3 data.¹⁶ Given that we lack over-time data on brideprice, we take two steps to allow us to explore the consequences of changes to the brideprice market on violence. First, we treat brideprice societies, Brideprice, as constant across our sample. This seems reasonable as the societal norms that determine such marital customs do not change from year to year (e.g. Anderson, 2007). Second, as discussed in the last section, we interact those societies with brideprice customs with other time-varying factors that affect marriage markets: Marriage Diff, which is the difference between the percentage of unmarried men and unmarried women aged 25–29 (with higher values more likely to induce violence);¹⁷Youth Bulge, which is the percentage of the population between 15 and 24 (with higher values less likely to produce violence);¹⁸ and GDP Growth (with higher values less likely to induce violence). These interactions offer an approximation of the volatility in brideprice markets, offering a further test of Hudson and Matfess (2017). Moreover, our empirical strategy allows us both to compare differences between societies with brideprice customs and those without, and to compare differences within brideprice socieities themselves under different market conditions (via the interaction terms), that is, under brideprice inflation.

To ensure that any brideprice–violence link we uncover is not a function of omitted confounds, we include a series of control variables: logged GDP,¹⁹ logged population,²⁰ democracy,²¹ political instability, ethnolinguistic fractionalization²² and religious fractionalization. In addition to these we also include regional fixed effects (using the classification from Hadenius and Teorel, 2007), helping to ensure that any findings are not the consequence of regionally clustered omitted confounds.^23,24

Analysis

The results from this analysis are presented in Tables 4 and 5. To preview our findings: we find no consistent link between brideprice and violence.

Table 4.

Brideprice and terrorism.

	Model 1	Model	Model 3
Brideprice	−0.143*	−0.513**	−0.200
	(0.069)	(0.076)	(0.122)
Population		0.814**	1.049**
		(0.035)	(0.044)
GDP		0.079**	−0.093**
		(0.026)	(0.034)
Polity		0.013*	0.045**
		(0.006)	(0.006)
Population Growth		0.083**	−0.208**
		(0.023)	(0.027)
GDP Growth		−0.009**	−0.013**
		(0.002)	(0.002)
Instability		0.615**	0.705**
		(0.143)	(0.133)
EthFrac		−0.133	−0.079
		(0.149)	(0.165)
RelFrac		−1.075**	0.802**
		(0.151)	(0.166)
Constant	3.149**	−12.110**	−13.767**
	(0.048)	(0.465)	(0.470)
ln(α)	2.186**	1.727**	1.523**
	(0.019)	(0.022)	(0.022)
N	7460	6066	5915
Regional dummies?	No	No	Yes

Note: *p < 0.05; **p < 0.01.

Table 5.

Brideprice inflation and terrorism.

	Model 4	Model 5	Model 6
Brideprice	−0.032	3.834**	−0.340**
	(0.218)	(0.719)	(0.126)
Marriage Diff	0.013
	(0.009)
Brideprice × Marriage Diff	−0.047**
	(0.010)
Youth Bulge		0.125**
		(0.024)
Brideprice × Youth Bulge		−0.229**
		(0.038)
Brideprice × GDP Growth			0.038**
			(0.009)
Population	1.088**	1.019**	1.041**
	(0.048)	(0.048)	(0.043)
GDP	−0.171**	−0.059	−0.092**
	(0.040)	(0.040)	(0.034)
Polity	0.038**	0.048**	0.046**
	(0.007)	(0.006)	(0.006)
Population Growth	−0.145**	−0.239**	−0.201**
	(0.036)	(0.028)	(0.027)
GDP Growth	−0.048**	−0.013**	−0.048**
	(0.006)	(0.002)	(0.008)
Instability	0.729**	0.633**	0.709**
	(0.154)	(0.135)	(0.133)
EthFrac	−0.473*	−0.337	−0.095
	(0.202)	(0.173)	(0.165)
RelFrac	0.319	0.945**	0.793**
	(0.200)	(0.169)	(0.166)
Constant	−13.029**	−15.934**	−13.554**
	(0.611)	(0.703)	(0.470)
ln(α)	1.324**	1.515**	1.518**
	(0.027)	(0.023)	(0.022)
N	4086	5880	6054
Regional dummies?	Yes	Yes	Yes

Note: *p < 0.05; **p < 0.01.

Table 4 summarizes our findings of the effect of brideprice society on terrorism. When we include no additional covariates, we find a negative and significant relationship between brideprice (societies) and terrorism—the opposite of what is expected by Hudson and Matfess (2017). However, this model is plainly incomplete, so we add a series of controls in Model 2 and regional fixed effects in Model 3.²⁵ We continue to find no support for the idea that bridperice makes terrorism more likely in either model.

However, as noted above, a better test of Hudson and Matfess (2017) is available by considering factors that induce over-time change in marriage markets. In Table 5, we report the results from this series of models. First, in Model 4 we evaluate whether Marriage Diff—the difference between the percentage of unmarried men and the percentage of unmarried women—increases the frequency of terrorism in brideprice societies. This is reflected by the interaction, Brideprice × Marriage Diff, which has a significant but negative coefficient (see also Figure 4, left panel). Again, this is in the opposite direction of our hypothesis based on Hudson and Matfess (2017). Here we see that increasing the proportion of unmarried men (relative to unmarried women) actually makes violence less likely. In Model 4, the coefficient on the interaction Brideprice × Youth Bulge does offer some support of Hudson and Matfess (2017), as we see that youth bulges in brideprice societies are associated with fewer incidents of terrorism (see also Figure 4, middle panel).²⁶ This is consistent with the argument that youth bulges should increase the supply of marriage-aged women, thereby driving down brideprice.²⁷ Finally, in Model 5, we see that GDP growth in brideprice societies actually increases the frequency of terrorist events (see also Figure 4, right panel), which is the opposite of what we should expect. Taken together, the evidence provides at best mixed support for the theory offered by Hudson and Matfess (2017).

Figure 4.

Brideprice and terrorism (conditional relationships).

When instead we estimate models with civil conflict incidence as the dependent variable, we find similar results (see Table 6). Here, we find that in the model without covariates, Brideprice is significant and in the expected direction—that is, societies with brideprice have more incidents of civil conflict. However, when we include common controls, the effect of Brideprice diminishes by 10-fold and is no longer significant (Model 8). Furthermore, when we include regional dummies to account for addition regionally common, time-invariant confounds, the sign on Brideprice switches and is in the unexpected direction, with it now making conflict less likely. This demonstrates how sensitive the finding on Brideprice is and, moreover, how it is probably proxying for some regional effects.

Table 6.

Brideprice and civil conflict incidence.

	Model 7	Model 8	Model 9
Brideprice	0.783*	0.091	−1.997**
	(0.310)	(0.097)	(0.198)
Population		0.828**	1.005**
		(0.040)	(0.056)
GDP		−0.243**	−0.274**
		(0.029)	(0.041)
Polity		0.017*	0.011
		(0.007)	(0.007)
Population Growth		0.144**	−0.042
		(0.035)	(0.045)
GDP Growth		−0.029**	−0.026**
		(0.007)	(0.007)
Instability		0.345*	0.351*
		(0.142)	(0.149)
EthFrac		1.129**	0.747**
		(0.181)	(0.227)
RelFrac		−1.369**	−0.321
		(0.158)	(0.210)
Constant	−2.150**	−9.766**	−11.966**
	(0.255)	(0.512)	(0.629)
N	9312	6370	6345
Regional dummies?	No	No	Yes

Note: *p < 0.05; **p < 0.01.

Turning to the conditional relationships we use to proxy for either marriage market distortion or brideprice inflation (Table 7), we again find no conflict-inducing effects. In Model 10 (and Figure 5, left panel) we find no significant effect of Brideprice × Marriage Diff. Again, this should be the closest proxy to the argument in Hudson and Matfess (2017) as it directly captures marriage market distortions. Turning to Youth Bulge, which the economics literature says should decrease brideprice, we find no significant effect in brideprice societies (see the interaction term and Figure 5, center panel). Finally, in Model 12, Brideprice × GDP Growth is also positive and insignificant.

Table 7.

Brideprice inflation and civil conflict incidence.

	Model 10	Model 11	Model 12
Brideprice	−2.896**	−0.630	−2.053**
	(0.383)	(1.027)	(0.208)
Marriage Diff	−0.042**
	(0.015)
Brideprice × Marriage Diff	0.018
	(0.016)
Youth Bulge		0.115**
		(0.037)
Brideprice × Youth Bulge		−0.070
		(0.052)
Brideprice × GDP Growth			0.013
			(0.015)
Population	1.210**	0.955**	1.004**
	(0.084)	(0.056)	(0.056)
GDP	−0.368**	−0.267**	−0.274**
	(0.063)	(0.042)	(0.041)
Polity	0.007	0.012	0.010
	(0.010)	(0.007)	(0.007)
Population Growth	−0.139*	−0.054	−0.043
	(0.063)	(0.047)	(0.046)
GDP Growth	−0.021*	−0.028**	−0.035**
	(0.010)	(0.007)	(0.012)
Instability	0.213	0.350*	0.351*
	(0.192)	(0.150)	(0.149)
EthFrac	0.074	0.841**	0.753**
	(0.287)	(0.227)	(0.227)
RelFrac	−0.889**	−0.237	−0.322
	(0.281)	(0.210)	(0.210)
Constant	−12.228**	−13.616**	−11.955**
	(0.916)	(0.904)	(0.629)
N	4155	6325	6345
Regional dummies?	Yes	Yes	Yes

Note: *p < 0.05; **p < 0.01.

Figure 5.

Brideprice and civil conflict (conditional relationships).

In sum, we find almost no support for a positive relationship between brideprice and political violence. This is true whether we concentrate on brideprice societies or brideprice inflation. If anything, some of the evidence in Tables 4 –7 indicates a negative effect of brideprice. However, we would encourage readers to not set much store by this seeming pacifying effect. First, in the absence of better data on brideprice inflation any findings should be taken with caution. While, in the absence of such data, we have attempted to find the best available proxies, we are hesitant to make a strong positive advocacy on the basis of these data. Second, the significance of these negative effects is not always robust to alternative model specifications as presented in the Online Appendix. For example, if we account for possible temporal autocorrelation or correlated errors, some of these results fall away as the estimators become less efficient. Finally, even if we were to take these negative effects as meaningful, this does not imply that brideprice is a social good. That is, even if it were to have this positive local effect, the general effect on well-being may be negative as brideprice has been shown to increase domestic violence (Bloch and Rao, 2002)

Conclusion

While more work remains to definitively test the role of brideprice in political violence, we are increasingly skeptical that any such linkage exists. First, our re-analysis of the empirical work in Hudson and Matfess (2017) found that their conclusions were a result of failing to include common control variables. Second, in parsing their theory, we came to believe that a more direct test was possible: using over-time variation in marriage, population, and economic conditions in brideprice societies to proxy for variation in marriage markets. Yet, our empirical analysis still found no robust relationship between brideprice and violence. Obviously, a superior and more direct test of Hudson and Matfess (2017) would require over-time data on brideprice itself, which we do not currently possess. Until data permits such a test, we caution against concluding in favor of a relationship that does not find support in current empirical analyses.

Future work should also focus on clearly identifying the mechanisms through which brideprice and brideprice inflation should produce violence. With additional expectations, other possibilities for empirically testing the relationship should become available. Ultimately, it could be that brideprice inflation itself does not matter directly, but contexts (patrilineality and inequality) and conditions (economic stagnation) produce violence. At present, we believe that this may be likely given findings from the existing literature on brideprice. First, research has suggested that it does not vary greatly over space and time. Second, empirical evidence indicates that bridperice itself is often not very high, with Anderson (2007: 161) noting that “little evidence exists on brideprice escalation in either the historical record or contemporary sources.” If true, the first would suggest that brideprice inflation rarely occurs, and the second would suggest that it is insufficient to crowd out many suitors, much less turn them to terrorism and rebellion.

Until and unless better data on brideprice becomes available, and empirical tests support a stronger association between brideprice and violence, we encourage readers and researchers to be more circumspect. This is especially important given the normative importance of this question and how quickly prior findings in the literature have spread.

Supplemental Material

CT_ReplicationMaterials – Supplemental material for In plain sight? Reconsidering the linkage between brideprice and violent conflict1

Supplemental material, CT_ReplicationMaterials for In plain sight? Reconsidering the linkage between brideprice and violent conflict1 by Scott J Cook and Cameron G Thies in Conflict Management and Peace Science

This article is distributed under the terms of the Creative Commons Attribution 4.0 License (http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/open-access-at-sage).

Supplemental Material

Supplementary_material – Supplemental material for In plain sight? Reconsidering the linkage between brideprice and violent conflict1

Supplemental material, Supplementary_material for In plain sight? Reconsidering the linkage between brideprice and violent conflict1 by Scott J Cook and Cameron G Thies in Conflict Management and Peace Science

Footnotes

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Scott J Cook

Supplemental material

All data, replication materials, and instructions regarding analytical materials upon which published claims rely are available online through the SAGE CMPS website:

Notes

References

Alesina

Devleeschauwer

Easterly

Kurlat

Wacziarg

(2003) Fractionalization. Journal of Economic Growth 8(2): 155–194.

Anderson

(2007) The economics of brideprice and dowry. Journal of Economic Perspectives 21(4): 151–174.

Arce

Croson

RTA

Eckel

(2011) Terrorism experiments. Journal of Peace Research 48(3): 373–382.

Beach

Pederson

(2016) Causal Case Study Methods: Foundations and Guidelines for Comparing, Matching, and Tracing. Ann Arbor, MI: University of Michigan Press.

Beck

Epstein

Jackman

O’Halloran

(2001) Alternative models of dynamics in binary time-series-cross-section models: The example of state failure. Working Paper. Available from: https://doi.org/10.7916/D8HT2W65

Bennett

Checkel

(2015) Process Tracing: From Metaphor to Analytic Tool. Cambridge: Cambridge University Press.

Bloch

Rao

(2002) Terror as a bargaining instrument: A case study of dowry violence in rural India. The American Economic Review 92(4): 1029–1043.

Blundell

Griffith

Windmeijer

(2002) Individual effects and dynamics in count data models. Journal of Econometrics 108(1): 113–131.

De Juan

Pierskalla

(2016) Civil war violence and political trust: Microlevel evidence from Nepal. Conflict Management and Peace Science 33(1): 67–88.

10.

Eisler

(2007) Dark underbelly of the world’s most “peaceful” countries. Available from https://www.csmonitor.com/2007/0726/p09s01-coop.html

11.

Fearon

Laitin

(2003) Ethnicity, insurgency, and civil war. American Political Science Review 97(1): 75–90.

12.

Franzese

Hays

Cook

(2016) Spatial- and spatiotemporal-autoregressive probit models of interdependent binary outcomes. Political Science Research and Methods 4(1): 151–173.

13.

Gilligan

Pasquale

Samii

(2013) Civil war and social cohesion: Lab-in-the-field evidence from Nepal. American Journal of Political Science 58(3): 604–619.

14.

Gleditsch

Wallensteen

Eriksson

Sollenberg

Strand

(2002) Armed conflict 1946-2001: A new dataset. Journal of Peace Research 39(5): 615–637.

15.

Goody

(1973) Bridewealth and dowry in Africa and Eurasia. In: Goody

Tambiah

(eds), Bridewealth and Dowry. Cambridge: Cambridge University Press, pp.1–58.

16.

Gurr

(1970) Why Men Rebel. Princeton, NJ: Princeton University Press.

17.

Hadenius

Teorell

(2007) Pathways from authoritarianism. Journal of Democracy 18(1): 143–156.

18.

Hatemi

McDermott

(2012) A neurobiological approach to foreign policy analysis: Identifying individual differences in political violence. Foreign Policy Analysis 8(2): 111–129.

19.

Hegre

Sambanis

(2006) Sensitivity analysis of empirical results on civil war onset. Journal of Conflict Resolution 50(4): 508–535.

20.

Hirose

Imai

Lyall

(2017) Can civilian attitudes predict insurgent violence? Ideology and insurgent tactical choice in civil war. Journal of Peace Research 54(1): 47–63.

21.

Hudson

Matfess

(2017) In plain sight: The neglected linkage between brideprice and violent conflict. International Security 42(1): 7–40.

22.

Hudson

Bowen

Nielsen

(2015) Clan governance and state stability: The relationship between female subordination and political order. American Political Science Review 109(3): 535–555.

23.

Ish-Shalom

(2013) Democratic Peace: A Political Biography. Ann Arbor, MI: University of Michigan Press.

24.

Kalyvas

Balcells

(2010) International system and technologies of rebellion: How the end of the Cold War shaped internal conflict. American Political Science Review 104(3): 415–429.

25.

(2005) Does democracy promote or reduce transnational terrorist incidents? Journal of Conflict Resolution 49(2): 278–297.

26.

Mulder

(1995) Bridewealth and its correlates: Quantifying changes over time. Current Anthropology 36(4): 573–603.

27.

Rao

(1993) The rising price of husbands: A hedonic analysis of dowry increases in rural India. Journal of Political Economy 101(4): 666–677.

28.

Reiter

(2015) The positivist stsudy of gender and international relations. Journal of Conflict Resolution 59(7): 1301–1326.

29.

Sjoberg

Kadera

Thies

(2018) Reevaluating gender and IR scholarship: Moving beyond Reiter’s dichotomies toward effective synergies. Journal of Conflict Resolution 62(4): 848–870.

30.

START (2018) Global Terrorism Database [Data file]. National Consortium for the Study of Terrorism and Responses to Terrorism. Available from http://www.start.umd.edu/gtd

31.

Thies

(2010) Of rulers, rebels, and revenue: State capacity, civil war onset, and primary commodities. Journal of Peace Research 47(3): 321–332.

32.

Ulfelder

(2012) The trouble with combining, or why I’m not touting the Global Peace Index. Available from http://duckofminerva.com/2012/06/trouble-with-combining-or-why-im-not.html

33.

Urdal

(2006) A clash of generations? Youth bulges and political violence. International Studies Quarterly 50: 607–629.

34.

Walsh

Piazza

(2010) Why respective physical integrity rights reduces terrorism. Comparative Political Studies 43(5): 551–577.

35.

Weidmann

Kuse

Gleditsch

(2010) The geography of the international system: The CShapes Dataset. International Interactions 36(1): 86–106.

36.

Wilson

Piazza

(2013) Autocracies and terrorism: Conditioning effects of authoritarian regime type on terrorist attacks. American Journal of Political Science 57(4): 941–949.

37.

Wood

Thomas

(2017) Women on the frontline: Rebel group ideology and women’s participation in violent rebellion. Journal of Peace Research 54(1): 31–46.

38.

Yair

Miodownik

(2016) Youth bulge and civil war: Why a country’s share of young adults explains only non-ethnic wars. Conflict Management and Peace Science 33(1): 25–44.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

7.26 MB

0.00 MB

0.26 MB