Introduction

Homicide rates across the US have been trending precipitously upward in the last few years (Lopez, Reference Lopez2022). Recent studies seeking to explain variation in homicide rates nominate diverse causes, including ambient temperature (Xu et al., Reference Xu, Xiong, Abramson, Li and Guo2020), city greenness (Sanciangco et al., Reference Sanciangco, Breetzke, Lin, Wang, Clevenger and Pearson2021), firearm ownership (Studdert et al., Reference Studdert, Zhang, Holsinger, Prince, Holsinger, Rodden and Miller2022), firearm laws (Crifasi et al., Reference Crifasi, Merrill-Francis, McCourt, Vernick, Wintemute and Webster2018), structural racism (Unnever et al., Reference Unnever, Stults and Messner2021), income inequality (Daly, Reference Daly2016), the interaction between absolute income and income inequality (Burraston et al., Reference Burraston, Watts, McCutcheon and Province2019; Gobaud et al., Reference Gobaud, Mehranbod, Dong, Dodington and Morrison2022), and more. While the homicide rate is clearly driven by multiple factors, many correlational studies lack the systematic theory needed to identify the underlying factors that structure individual motivations which may lead someone to kill another. Identifying causal factors is critical as interventions designed to reduce homicide rates need to target the underlying cause, rather than correlated effects. Here, we take an a priori approach by testing predictions derived from the evolutionary social sciences, rather than a purely empirical approach that attempts to infer causal links through model selection. Our analysis focuses on a set of predictor variables that have been hypothesised in evolutionary theoretical models to structure the underlying motivations for homicidal behaviour. While this approach may overlook proximate mechanisms and confounding variables, our goal is to draw attention to the ultimate factors influencing the individual payoffs for homicidal violence.

Researchers drawing on evolutionary theory propose the causal argument that absolute and relative resource scarcity structures the costs and benefits for violent interpersonal competition (Allen et al., Reference Allen, Bettinger, Codding, Jones and Schwitalla2016; Daly, Reference Daly2016, Reference Daly2023; Daly & Wilson, Reference Daly and Wilson1988; Daly et al., Reference Daly, Wilson and Vasdev2001; McCool et al., Reference McCool, Vernon, Yaworsky and Codding2022). Specifically, when resources are unequally distributed, individuals with limited wealth and income may have incentives to undertake high-risk activities – including those that lead to lethal violence – in order to access material and social capital. In this sense homicide represents a ‘manifestation of the local level of competition for scarce resources’ (Daly, Reference Daly2016: 2). This theory articulates with core predictions from life history theory, namely, that variation in resource availability and extrinsic mortality risk structure the costs and benefits of key life history trade-offs. In environments with high extrinsic mortality risk, fast life history strategies are favoured, which can include increased aggression, including homicide (Hackman & Hruschka, Reference Hackman and Hruschka2013; Wilson & Daly, Reference Wilson and Daly1997). Where poverty and inequality elevate extrinsic mortality risk, fast life history strategies may also promote future-discounting, where individuals place a premium on the present over the future (Wilson & Daly, Reference Wilson and Daly1997). These studies and others (e.g. Allen et al., Reference Allen, Bettinger, Codding, Jones and Schwitalla2016; de Courson et al., Reference De Courson, Frankenhuis, Nettle and Van Gelder2023; Daly & Krupp, Reference Daly, Krupp, Eppard and Giroux2022; Daly, Reference Daly2023) develop a general evolutionary theory of lethal violence, which links material conditions necessary for survival and reproduction to the variable costs and benefits of homicide. Importantly, this theory recognises that (1) homicide cannot be explained with reference only to pathologies that produce maladaptive behaviours (Daly & Krupp, Reference Daly, Krupp, Eppard and Giroux2022; Hackman & Hruschka, Reference Hackman and Hruschka2013), (2) most homicide is, rather than a manifestation of mental illness, a complex, facultative capability sensitive to the conditional payoffs for competition over limited resources (Daly & Krupp, Reference Daly, Krupp, Eppard and Giroux2022) and (3) life history trade-offs may partially explain persistent cycles of killings (also termed enduring neighbourhood effects; de Courson et al., Reference De Courson, Frankenhuis, Nettle and Van Gelder2023), whereby homicide itself increases extrinsic mortality risk, which positively feeds back on life history strategies, and in addition, (4) we find it important to bring up the naturalistic fallacy, namely, that while homicide may be adaptive by providing Darwinian fitness benefits under certain conditions, this does not suggest that homicide is beneficial to society or morally appropriate.

This evolutionary theory, originally outlined by Daly and Wilson (Reference Daly and Wilson1988), strengthens causal inferences that economists, sociologists and others have made for decades (see below), and can explain why resource scarcity and inequality so often correlate with harmful societal outcomes, including homicidal behaviour.

Building on the evolutionary theory outlined above, we recently showed that the link between resource limitations and violence is not unique to Western industrialised countries: generalised violence varies systematically across human societies as a function of relative and absolute resource availability (McCool et al., Reference McCool, Vernon, Yaworsky and Codding2022). Specifically, our anthropological cross-cultural analysis shows that among delayed-return economies, a characteristic of agricultural as well as post-industrial societies, generalised violence representing homicide and warfare increases in environments where resources are scarce and heterogeneously distributed across space. In line with Daly et al.'s hypothesis, we argue that these results indicate that lethal violence is most often a manifestation of social competition for scarce, fitness-linked resources. In other words, the benefits for lethal violence and other risk-taking strategies should be sensitive to resource thresholds, below which social and economic standing – or survivorship in subsistence societies – is in jeopardy (de Courson et al., Reference De Courson, Frankenhuis, Nettle and Van Gelder2023). This theory thus predicts that homicide in delayed-return societies should peak when and where resources are scarce and/or unevenly distributed, and that the conditions that promote lethal violence may arise well before resource scarcity threatens basic survival. If the theory is correct, it may help explain the underlying causes of correlations between violence and many other factors including climate change (Hsiang et al., Reference Hsiang, Burke and Miguel2013), population pressure (Oka et al., Reference Oka, Kissel, Golitko, Sheridan, Kim and Fuentes2017), agricultural productivity (Zhang et al., Reference Zhang, Brecke, Lee, He and Zhang2007) and more.

While the evolutionary social sciences provide a general theory grounded in Darwinian logic to elucidate why certain conditions promote homicide, there is an expansive literature in economics, psychology, political science and other sciences linking economic conditions and violence, much of which overlaps with evolutionarily motivated studies. While this literature is too vast to summarise in detail, we outline several important trends in the field pertaining to empirical approaches that establish links between economic conditions and homicide and studies which hypothesise explicit causal relationships. A more thorough review of non-evolutionary approaches can be found in Daly (Reference Daly2016, Reference Daly2023).

A large amount of empirical research stems from the economics, sociology and political science literature, which has long proposed a positive relationship between various forms of inequality (e.g. ethnic inequality) and violent intergroup conflict (Cederman et al., Reference Cederman, Nils and Gleditch2011; Horowitz, Reference Horowitz1985), including homicide (Aransiola et al., Reference Aransiola, Ceccato and Justus2021; Burraston et al., Reference Burraston, Watts, McCutcheon and Province2019; Gobaud et al., Reference Gobaud, Mehranbod, Dong, Dodington and Morrison2022). Causal arguments are varied, but often propose violence to be a reaction to structurally imposed constraints on material and social capital leading to deprivation in certain communities (e.g. McDaniel et al., Reference McDaniel, Ge and Yuan2022; Daly, Reference Daly2023; Vásquez et al., Reference Vásquez, Alvarado, Tillaguango, Işık and Murshed2023). While this research programme has produced numerous empirical insights, studies often find conflicting results (e.g. Alesina et al., Reference Alesina, Michalopoulos and Papaioannou2016; Cederman et al., Reference Cederman, Nils and Gleditch2011; Esteban & Ray, Reference Esteban and Ray2011; Huber & Mayoral, Reference Huber and Mayoral2013) and some studies have called for a greater focus on theorising causal links between competing variables (Burke et al., Reference Burke, Hsiang and Miguel2015).

Prominent theories are too numerous to list here, but several influential models include the ‘social dominance’ hypothesis (Sidanius & Pratto, Reference Sidanius and Pratto2001), which proposes that societies create explicit hierarchies based on factors that privilege dominant groups, disadvantage subordinate groups and use physical force and ideological indoctrination to enforce norms of inequality, which in turn may spark violence within disadvantaged groups. Indeed, there is some evolutionary research suggesting ethnic dominance hinders contributions to public goods and may foment inter-ethnic conflict (Waring & Bell, Reference Waring and Bell2013). The ‘social disorganisation theory’ suggests that local socio-economic conditions are more important than individual characteristics when explaining criminal activity and propose homicide to be a reaction to the breakdown of social cohesion in disadvantaged communities, which fosters feelings of resentment, frustration, hopelessness and alienation (Blau & Blau, Reference Blau and Blau1982; Gobaud et al., Reference Gobaud, Mehranbod, Dong, Dodington and Morrison2022). These theories certainly differ from evolutionary models in their ultimate explanation of violent behaviour and focus more on the drivers of socially imposed constrains and the proximate phycological mechanisms that facilitate violent acts. Nonetheless, they articulate with evolutionary models in positing a causal relationship between resource deprivation and homicidal behaviour (Blume, Reference Blume2021; Burraston et al., Reference Burraston, Watts, McCutcheon and Province2019; Gobaud et al., Reference Gobaud, Mehranbod, Dong, Dodington and Morrison2022; Vásquez et al., Reference Vásquez, Alvarado, Tillaguango, Işık and Murshed2023).

Here we further evaluate this general evolutionary theory by systematically analysing data from the FBI Uniform Crime Reporting database on homicide rates, and the American Community Survey (ACS) on the proportion of households with income under the poverty level (proportion in poverty) and income inequality (Gini Index) across all 50 US states for the years 1990, 2000 and 2005–2020. Importantly, the 2020 data allow us to evaluate whether this general theory also partially accounts for the increase in violence during the COVID-19 pandemic. We also produce a novel analysis that examines how the effects of inequality vary across relatively rich vs. poor states and how the effects of poverty vary across relatively equal versus unequal states. These interactive effects should result from poverty and inequality targeting different aspects of resource availability. Poverty may push an entire population below a destitution threshold (de Courson et al., Reference De Courson, Frankenhuis, Nettle and Van Gelder2023), while inequality may drive resource scarcity in a subpopulation even as average holdings increase for the population at large (Daly, Reference Daly2016). Thus, we expect homicide rates to peak when a meaningful proportion of a population is below a destitution threshold and existing resources are unequally distributed.