Local Competition Amplifies the Corrosive Effects of Inequality

Abstract

Inequality is widely believed to incite conflict, but the evidence is inconsistent. We argue that the spatial scale of competition—the extent to which individuals compete locally, with their interaction partners, or globally, with the entire population—can help settle the question. We built a mathematical model of the evolution of conflict under inequality and tested its predictions in an experimental game with 1,205 participants. We found that inequality increases conflict, destroys wealth, and engenders risk taking. Crucially, these effects are amplified by local competition. Thus, inequality is at its most damaging when it arises between close competitors. Indeed, at the extremes, the combined effects of inequality and the scale of competition are very large. More broadly, our findings suggest that disagreements in the literature may be the result of a mismatch between the scale at which inequality is measured and the scale at which conflict occurs.

Keywords

inequality scale of competition conflict relative deprivation tragedy of the commons risk taking open data open materials preregistered

There is growing concern that inequality is socially corrosive. Differential payoffs—of wealth, resources, and power—are believed to cause physical and mental illness (Wilkinson & Pickett, 2009), destabilize economies (Cynamon & Fazzari, 2016; Jayadev, 2013; Kumhof, Rancière, & Winant, 2015), and undermine democratic institutions (Bartels, 2010; Boix, 2003; Muller, 1995). They are also thought to be responsible for violence across the levels of social organization, from homicide to civil and international warfare (Betzig, 1986; Cederman, Gleditsch, & Buhaug, 2013; Daly, 2016; Hensel, McLaughlin Mitchell, & Sowers, 2006; Muller & Seligson, 1987; Østby, 2008; Rider & Owsiak, 2015).

It is surprising, then, that the causal effects of inequality on conflict have not been firmly established. While there is evidence from experimental studies that inequality reduces cooperation (Anderson, Mellor, & Milyo, 2008; Barclay & Benard, 2013; Buckley & Croson, 2006; Burton-Chellew, May, & West, 2013; Côté, House, & Willer, 2015), inconsistent findings from observational studies have led to disagreement over the relationship between inequality and conflict. For instance, income inequality is a reliable predictor of the homicide rate in cross-sectional analyses (Daly, 2016) but not in longitudinal analyses (Neumayer, 2005). Likewise, inequality measured between groups predicts the onset of civil war (Cederman et al., 2013; Østby, 2008), but inequality measured over the entire population does not (Collier & Hoeffler, 2004; Fearon & Laitin, 2003).

We propose that these inconsistencies can be resolved through an appreciation of the spatial scale of competition, the probability that an individual will both interact and compete for reproductive success, or fitness, with the same partners (Taylor, 1992; West et al., 2006). Individuals tend to interact directly with a limited number of partners but may nevertheless compete for fitness with a broader swath of the population. For instance, students may interact with their classmates for grades but compete for jobs with other students drawn from different schools in different cities, states, or countries. When competition is global, interaction partners disperse to compete with the rest of the population. This makes fitness a non-zero-sum property: Both partners can gain (or lose) together. Conversely, when competition is local, interaction partners compete exclusively with one another. This makes fitness a zero-sum property: One individual’s gain is his or her partner’s loss.

Local competition has powerful effects on human behavior. In experimental games, it reduces cooperation (West et al., 2006), increases monetary demands (Barclay & Stoller, 2014), and increases harm to in-group, but not out-group, members (Barker & Barclay, 2016). Some degree of local competition is implied by the hypothesis that relative deprivation underlies the effect of inequality on conflict: By comparing ourselves with “similar others” who are better off than we are, we develop feelings of resentment that foment conflict (Crosby, 1976). If the psychology of relative deprivation has been shaped by natural selection, however, it should be sensitive to the probability that a source of social comparison is also a close competitor. Thus, “similarity” may simply serve as a cue that a comparator is a competitor. Two individuals who share the same interests, the same talents, and the same haunts will more likely be competitors for mates and position than two individuals who share none of these things.

Here, we built an evolutionary model to study the effects of inequality and the scale of competition on conflict. Although this model is not human specific, it provides a formal, functional basis for a psychological theory of inequality. We then tested the model’s predictions in an experiment with human players. In both the model and the experiment, individuals played a single round of the Hawk-Dove game (HDG) with a genetically unrelated partner (Maynard Smith, 1982). In the HDG, individuals play either the belligerent “Hawk” strategy or the peaceful “Dove” strategy to secure a resource of value v. Doves split the resource with other Doves but relinquish it entirely to Hawks, whereas Hawks fight other Hawks for the resource, effectively splitting it but also paying a total cost k for the fight. This gives the general payoff matrix in Figure 1a. The HDG is useful for our purposes because it creates the opportunity for inequality of magnitude v between individuals playing different strategies. It also places a constraint on conflict: The cost of a fight exceeds the value of the resource (k > v), so fights are counterproductive.

Fig. 1.

Payoff matrices for the Hawk-Dove game from the perspective of the player in each row. The general payoff matrix is presented in (a): A Dove playing a Dove earns v/2, a Dove playing a Hawk earns 0, a Hawk playing a Dove earns v, and a Hawk playing a Hawk earns (v – k)/2. Experimental payoff matrices, where k = 100, are shown for (b) the no-inequality conditions (v = 0), (c) the low-inequality conditions (v = 10), and (d) the high-inequality conditions (v = 90).

As a corollary to our study of conflict, we measured the effects of inequality and the scale of competition on the expected wealth of pairs of players and the riskiness of the Hawk and Dove strategies. Research has shown that inequality exacerbates the “tragedy of the commons,” the willingness of individuals to reap personal gains at the expense of group success (Anderson et al., 2008; Barclay & Benard, 2013; Buckley & Croson, 2006; Burton-Chellew et al., 2013; Côté et al., 2015). Moreover, inequality promotes risky behavior in a variety of forms (Daly, 2016; Daly & Wilson, 2001; Mishra, 2014; Mishra, Son Hing, & Lalumière, 2015; Payne, Brown-Iannuzzi, & Hannay, 2017). These secondary analyses were not preregistered but follow logically from the hypothesis that inequality causes conflict.

Evolutionary Model

Method

We built a standard infinite-island model with discrete, nonoverlapping generations (Krupp & Taylor, 2015; Taylor, 1992). We assumed an infinitely large population of asexual individuals distributed over a series of islands. There are two individuals and one breeding vacancy on each island, and the life cycle consists of an interaction stage followed by a competition stage.

At the interaction stage, partners on the same island play an HDG. Strategies are controlled by two alleles at a single, haploid locus: One causes its bearers to play Hawk and the other to play Dove, with population frequencies q and 1 – q, respectively. At the competition stage, individuals compete with their partner from the interaction stage with probability a for the breeding vacancy on their own island. This is local competition, because an individual’s HDG partner and his or her rival for the breeding vacancy are one and the same. Alternatively, both individuals migrate (separately) to different, random islands with probability 1 – a and compete for the breeding vacancy on their respective islands with one of the other individuals in the population at large. This is global competition, because an individual’s HDG partner and his or her rival for the breeding vacancy are different, the latter being sampled from the global population. In either case, the party at the competition stage with the largest payoff is awarded the breeding vacancy: Successful breeders produce two offspring, and after this, all adults die and all offspring migrate to a random island to interact. As a result, partners in both stages are unrelated to one another.

Results

Evolutionary stability of Hawks

The expected payoff to a Hawk is p_H = qp_H,H + (1 – q)p_H,D, the payoff to a Hawk playing against a Hawk (weighted by the frequency of Hawks in the population) plus the payoff to a Hawk playing against a Dove (weighted by the frequency of Doves in the population). Likewise, the expected payoff to a Dove is p_D = qp_D,H + (1 – q)p_D,D, the payoff to a Dove playing against a Hawk (weighted by the frequency of Hawks in the population) plus the payoff to a Dove playing against a Dove (weighted by the frequency of Doves in the population). A focal individual wins the breeding vacancy if his or her payoff is greater than his or her rival’s. With probability a, that rival is his or her partner in the HDG and, with probability 1 – a, that rival is an individual drawn at random from an infinitely large population. Consequently, the expected fitness of a Hawk is

\begin{matrix} w_{H} = a (q p_{H, H} + (1 - q) p_{H, D} - q p_{H, H} - (1 - q) p_{D, H}) + (1 - a) \\ (q p_{H, H} + (1 - q) p_{H, D} - (\begin{array}{l} q^{2} p_{H, H} + q (1 - q) p_{H, D} + \\ q (1 - q) p_{D, H} + {(1 - q)}^{2} p_{D, D} \end{array})), \end{matrix}

and the expected fitness of a Dove is

\begin{matrix} w_{D} = a (q p_{D, H} + (1 - q) p_{D, D} - q p_{H, D} - (1 - q) p_{D, D}) + (1 - a) \\ (q p_{D, H} + (1 - q) p_{D, D} - (\begin{array}{l} q^{2} p_{H, H} + q (1 - q) p_{H, D} + \\ q (1 - q) p_{D, H} + {(1 - q)}^{2} p_{D, D} \end{array})) . \end{matrix}

We find the evolutionarily stable frequency q* of playing Hawk by setting w_H = w_D. After simplification, substitution of payoffs, and rearrangement, we obtain

q * = \frac{1 + a}{1 - a} * \frac{v}{k} .

Thus, our model predicts an interaction between inequality (v) and the scale of competition (a) in the evolution of conflict: The frequency of Hawks grows with the value of the resource, and the rate of change in this relationship increases with the probability of local competition (Fig. 2). Under strictly global competition (a = 0), we expect conflict to persist at a frequency of q* = v/k, the classic result for the HDG (Maynard Smith, 1982). However, under strictly local competition (a → 1), any degree of inequality (v > 0) will cause conflict to dominate.

Fig. 2.

Theoretical effects of inequality (v) and the scale of competition (a) on the evolution of conflict. Holding the cost of fighting constant at k = 100, the evolutionarily stable frequency of Hawks (q*) increases with inequality between winners and losers in the Hawk-Dove game. Lines represent three spatial scales of competition: a = 0 (dotted line), a = 1/3 (dashed line), and a = 2/3 (solid line). They show that the effect of inequality on conflict becomes more severe as competition becomes increasingly local.

Effects on wealth

To quantify the effects of inequality and the scale of competition on the wealth of pairs in the model, we calculate the expected payoff of an interacting pair when the strategies in the population come to equilibrium. Fights arise when two Hawks interact; thus, on average, there are q² fights per interaction. Under these circumstances, pairs can expect to acquire the resource of value v but also to pay a cost k for fighting over it. Under all other circumstances, pairs acquire the resource but pay no cost. Consequently, a pair’s expected wealth is v – kq², where kq² is the amount of the resource the pair is expected to lose. With this, we can determine the expected wealth and losses of a pair at equilibrium by setting q = q*. We simulate this numerically in Figure 3. As can be seen, wealth declines with the probability of local competition. In addition, more wealth is lost as inequality increases, and the rate of change in this relationship increases with the probability of local competition.

Fig. 3.

Theoretical effects of inequality (v) and the scale of competition (a) on expected wealth and losses. We assume that q = q* and k = 100. Curves represent a = 0 (dotted curves), a = 1/3 (dashed curves), and a = 2/3 (solid curves). The left graph displays the expected wealth per pair in the model (v – kq²). It shows that pairs earn less as competition becomes increasingly local, because local competition increases the frequency of Hawks and therefore the chance of a costly fight. The right graph displays the expected losses per pair in the model (kq²). It shows that more wealth is lost as inequality grows and that this effect is more severe as competition becomes increasingly local.

Riskiness of Hawk versus Dove

As is common, we measure the riskiness of a strategy by the variance in the distribution of its outcomes (Daly & Wilson, 2001; Mishra, 2014; Payne et al., 2017): When two strategies have the same expected return, the strategy with the higher outcome variance is the riskier one. In the current study, the outcomes are payoffs, and the expected returns to a Hawk and a Dove are identical when the population frequency of Hawks is at equilibrium (q = q*).

Variance in payoffs p is defined as σ² = E(p²) – E(p)². To find the outcome variance associated with playing Hawk, we simply rewrite this as

{σ_{H}}^{2} = q {p_{H, H}}^{2} + (1 - q) {p_{H, D}}^{2} - {(q p_{H, H} + (1 - q) p_{H, D})}^{2} .

Likewise, to find the outcome variance associated with playing Dove, we rewrite this as

{σ_{D}}^{2} = q {p_{D, H}}^{2} + (1 - q) {p_{D, D}}^{2} - {(q p_{D, H} + (1 - q) p_{D, D})}^{2} .

With substitution of the payoffs from the general game matrix and rearrangement, these become

{σ_{H}}^{2} = \frac{(q - q^{2}) {(k + v)}^{2}}{4}

and

{σ_{D}}^{2} = \frac{(q - q^{2}) v^{2}}{4} .

Because (k + v)² > v², it follows that σ_H² > σ_D². This is true for any value of q, including q = q*. Consequently, Hawk is always riskier than Dove.

Behavioral Experiment

Method

Assuming that we are equipped with psychological mechanisms to process (a) the potential degree of inequality between interaction partners and (b) the extent to which a partner is also a competitor, the model predicts that inequality and the scale of competition will interact to increase conflict in humans. To test this, we recruited participants via Amazon Mechanical Turk to play a single round of a one-shot, anonymous HDG. Past research has shown that the scale of competition has a medium to very large effect size (Barclay & Stoller, 2014; Barker & Barclay, 2016; West et al., 2006); using a Cohen’s d set conservatively to 0.5 and statistical power set to .95, analysis suggested a minimum of 88 participants per condition under a null-hypothesis-significance-testing approach (which we did not use). We recruited 1,205 participants in total, and the smallest sample size of any condition was 183 participants, well above this minimum. Note that we also conducted a series of pilot experiments in which we modified the instructions to help the participants understand the design (e.g., we removed biased language; we made it explicit that there was no communication between players and that the games were one shot). The materials, data, and a meta-analysis of all studies are available at http://osf.io/ycren.

As in the model, the experiment had two stages. At the interaction stage, participants played one of three versions of the HDG (with no, low, or high outcome inequality) with a randomly selected partner drawn from a larger group of 200. At the competition stage, participants were paid on the basis of their performance against their HDG partner (strictly local competition) or on the basis of their performance within the larger group of 200 (strictly global competition). Prior to making their HDG decisions, all participants were told that they had been paired with a partner drawn from a larger group of 200 and were given instructions informing them of the payoff structure of their game and the scale of competition.

The Queen’s University General Research Ethics Board approved the research protocols. Participants were recruited via advertisement on Amazon Mechanical Turk and provided a letter of information. Once they gave consent, they were randomly assigned to six experimental conditions and given a set of task instructions. Following this, they were asked to complete a comprehension test. If they successfully completed the comprehension test, they were allowed to proceed to the HDG. They then chose to play either Hawk or Dove and were paid according to their performance. Participants were excluded from the analyses if they did not pass the comprehension test, did not complete the task, were not located in the United States at the time of participation, or were younger than 18 years of age. One hundred fifty-six individuals did not pass the comprehension test, 42 did not complete the survey, and 3 were excluded because they did not submit a response to the game (i.e., they passed the manipulation check but did not continue on to make a choice for the game).

We manipulated inequality by varying the value of the resource while holding the cost of fighting fixed at k = 100 across all conditions. Participants played an HDG with no (v = 0, Fig. 1b), low (v = 10, Fig. 1c), or high (v = 90, Fig. 1d) levels of inequality. We used neutral language in the instructions, labeling the two strategies as “Option 1” and “Option 2” and the participant’s partner as “Person A.” For example, in the instructions for the low-inequality conditions, participants were informed, “If you choose Option 2 and Person A chooses Option 1, you will gain 10 points and Person A will gain 0 points.” We also made it clear that payoffs were equal when pairs played the same strategy, rather than having 1 participant earn the resource at random. For instance, in the low-inequality conditions, participants were informed that they and Person A would both earn 5 points if they both played Dove and –45 points if they both played Hawk.

We awarded money to players according to the scale of competition to which they had been assigned (Barclay & Stoller, 2014; Barker & Barclay, 2016; West et al., 2006). Under strictly local competition, players were informed that they would be paid $1.00 if they earned more points than their partner in the HDG and nothing if they earned less than their partner. Under strictly global competition, however, players were informed that they would be paid $1.00 if they earned one of the top 100 scores in their group of 200 and nothing if they earned less than the top 100th score. Hence, the chances of earning a monetary reward were the same in both the local and global competition conditions, but a player’s performance was relative only to his or her HDG partner in the local competition conditions and relative to all players in his or her group of 200 in the global competition conditions.

The six experimental conditions were thus local competition and no inequality (local no-inequality condition), local competition and low inequality (local low-inequality condition), local competition and high inequality (local high-inequality condition), global competition and no inequality (global no-inequality condition), global competition and low inequality (global low-inequality condition), and global competition and high inequality (global high-inequality condition). Although we did not expect participants to play the evolutionarily stable frequencies (Burton-Chellew, El Mouden, & West, 2016; Kümmerli, Burton-Chellew, Ross-Gillespie, & West, 2010), we could use the model to predict the relative frequencies of Hawks by substituting the values of a, v, and k for each condition. Doing so gave the following relative frequencies of Hawks: local high-inequality = local low-inequality > global high-inequality > global low-inequality > global no-inequality. In the local no-inequality condition, all decisions resulted in a tie between competitors, and so there was no evolutionarily stable frequency of Hawks. Because there were no consequences to any decision, the local no-inequality condition can be thought of as a window onto participants’ expectations about the “default” strategy to play in our experiment (Yamagishi, Hashimoto, & Schug, 2008). That is, it can be used as a benchmark of what participants believed was the appropriate strategy to play under ambiguous circumstances. If participants hold no expectations about what they “ought” to do, then their decisions should be random. If, however, participants rely on outside information (e.g., social norms), then their decisions will be nonrandom. Thus, we simply predicted that behavior in this condition would range between local high-inequality and global no-inequality. We applied an estimation approach to analyze our data, supplying point estimates and 95% confidence intervals (CIs) for each condition. We also calculated differences in proportions between conditions to estimate effect sizes.

Results

Frequency of Hawks

The effects of inequality and the scale of competition on the frequency of Hawks are shown in Figure 4, and the differences in the proportion of individuals playing Hawk between each pair of conditions are presented in Table 1. As can be seen, our findings are consistent with each of the predictions (local high-inequality = local low-inequality > global high-inequality > global low-inequality > global no-inequality), and the differences between the extremes are, by conventional standards, very large: There were over 45% more Hawks in the local low-inequality and local high-inequality conditions than in the global no-inequality condition. Overall, conflict increased with inequality and, as expected, peaked when competition was local. Moreover, participants typically played Dove in the local no-inequality condition, though slightly less often than in the global no-inequality condition.

Fig. 4.

Actual effects of inequality and the scale of competition on conflict. Circles represent the frequency of Hawks in each condition of the experiment (error bars show 95% confidence intervals). The dotted line and white circles depict results for the global competition conditions, and the solid line and black circles depict results for the local competition conditions. They show that the frequency with which participants played Hawk increased with inequality and that this effect was more severe under local than global competition.

Table 1.

Differences in the Proportion of Hawks Between Conditions

Condition	Global no-inequality	Global low-inequality	Global high-inequality	Local no-inequality	Local low-inequality
Global low-inequality	.21 [.14, .28]
Global high-inequality	.33 [.25, .40]	.12 [.03, .21]
Local no-inequality	.08 [.03, .14]	−.13 [−.21, −.05]	−.25 [−.33, −.16]
Local low-inequality	.47 [.40, .54]	.26 [.17, .35]	.15 [.05, .24]	.39 [.31, .47]
Local high-inequality	.46 [.38, .53]	.25 [.15, .33]	.13 [.03, .22]	.38 [.29, .45]	−.02 [−.11, .07]

Note: To calculate each difference, we subtracted the proportion of Hawks in the condition shown in a given column from the proportion of Hawks in the condition shown in a corresponding row. Values in brackets are 95% confidence intervals.

Effects on wealth

To quantify the effects of inequality and the scale of competition on wealth in the experiment, we calculated the mean payoff (95% CIs) of an interacting pair in each condition as well as mean losses: If the value of the resource and the mean payoff per pair in a given condition are v_c and p_c, respectively, then the mean losses per pair are v_c – p_c. These results are presented in Figure 5. They are based on a dichotomous variable: whether or not both participants in a game played Hawk and therefore fought. To compute a measure of effect size, then, we calculated the differences in the proportion of fights between each pair of conditions (Table 2). As can be seen, wealth was lower under local than global competition. In addition, more wealth was lost as inequality increased, and the rate of change in this relationship was greater under local than global competition. Again, the effects were very large at the extremes (e.g., global no-inequality and local no-inequality vs. local low-inequality and local high-inequality), with a difference of over 40% more fights.

Fig. 5.

Actual effects of inequality and the scale of competition on wealth and losses. Circles represent mean wealth (left graph) or losses (right graph) per pair for the global competition conditions (dotted lines and white circles) and the local competition conditions (solid lines and black circles). Error bars show 95% confidence intervals. The left graph shows that pairs earned lower payoffs under local competition than under global competition because local competition increased the frequency of Hawks and therefore the number of costly fights. The right graph shows that more wealth was lost as inequality increased and that this effect was more severe under local than global competition.

Table 2.

Differences in the Proportion of Fights Between Conditions

Condition	Global no-inequality	Global low-inequality	Global high-inequality	Local no-inequality	Local low-inequality
Global low-inequality	.13 [.06, .21]
Global high-inequality	.18 [.11, .27]	.06 −.05, .16]
Local no-inequality	.01 −.03, .06]	−.12 −.20, –.05]	−.17 −.26, –.10]
Local low-inequality	.43 [.34, .52]	.30 [.19, .41]	.25 [.13, .36]	.42 [.33, .51]
Local high-inequality	.44 [.34, .54]	.31 [.19, .42]	.26 [.13, .37]	.43 [.33, .53]	.01 −.12, .14]

Note: To calculate each difference, we subtracted the proportion of fights in the condition shown in a given column from the proportion of fights in the condition shown in a corresponding row. Values in brackets are 95% confidence intervals.

Discussion

To generate predictions about the effects of inequality and the scale of competition, we built a model of the evolution of conflict under inequality of outcomes. Our model predicted that the potential for inequality causes conflict and that the rate of change in this effect increases with the probability of local competition. Indeed, when competition is strictly local, any amount of inequality drives the Hawk strategy to fixation. This suggests that the psychology of relative deprivation should be sensitive to inequality between interaction partners and the probability that these partners are also competitors.

We tested the predictions of our model with human players and found that conflict increased with inequality and local competition: Each rank-order prediction of the model was mirrored by the frequency of Hawks in each of the experimental conditions. Nearly all of the participants in the global no-inequality condition played Dove, whereas approximately half of the participants in the local high-inequality and local low-inequality conditions played Hawk, a very large effect by conventional standards. The latter finding seems even more remarkable when we consider that the default strategy was to eschew conflict almost entirely: Only 11% of the participants in the local no-inequality condition played Hawk, even though there were no negative consequences for doing so. Thus, our results likely underestimate the true effect of inequality on conflict.

Our findings are also striking when we consider that Hawks pay a cost k to fight other Hawks. Because of this, an interacting pair can expect to earn v – kq². As the frequency of Hawks grows, then, pairs earn less than they otherwise could had they settled their disputes peaceably. Evaluated at equilibrium, losses of potential wealth in our model can be dramatic: With even modest amounts of inequality and local competition, the typical pair pays large costs and leaves the interaction stage with negative payoffs (Fig. 3). Of course, pairs in the global high-inequality, local low-inequality, and local high-inequality conditions of our behavioral experiment performed better than the model forecasts (Fig. 5) because they played Hawk at frequencies below equilibrium and so fought less often than theory predicts. Nevertheless, the average pair destroyed a nonnegligible amount of wealth in all but the global no-inequality and local no-inequality conditions, and losses were most pronounced under local competition. This suggests that individuals become increasingly concerned with the prospect of inequality among their partners as the probability of local competition increases: They would rather be better off than their partners even when they are worse off in absolute terms (e.g., Frank, 2012). Hence, inequality and local competition compound the tragedy of the commons.

An alternative to the interpretations above might be that individuals are swayed by the absolute value of the resource, not the inequality that this value engenders between winners (Hawks playing Doves) and losers (Doves playing Hawks). However, this hypothesis fails to explain the effect of the scale of competition—which has no impact on the absolute value of the resource—on the frequency of Hawks. Moreover, a small adjustment to the payoffs also illustrates the effect of inequality, independent of the absolute value of the resource. Consider the high-inequality payoff matrix (Fig. 1d). Given these payoffs, the theoretical model shows that 90% of individuals should play Hawk under global competition (q* = 0.9). However, if we raise the payoff to a Dove playing a Hawk (p_D,H; upper right cell) from 0 to 40, we can lower the inequality between winners and losers from 90 to 50 while still preserving the absolute value of winning the resource (p_H,D = 90; lower left cell) and the rank order of the payoffs (p_H,D > p_D,D > p_D,H > p_H,H). Applying this modified matrix to the model, we now find that only 50% of individuals should play Hawk under global competition (q* = 0.5). So, reducing inequality in the model reduces the frequency of Hawks, even when the absolute value of the resource remains the same. Thus, individuals are affected not by the absolute payoffs but by the inequality imposed by those payoffs.

Our results also lend support to the broader hypothesis that inequality encourages risky behavior, of which conflict is a manifestation (Daly, 2016; Daly & Wilson, 2001; Mishra, 2014; Mishra et al., 2015; Payne et al., 2017). Risk is widely defined as outcome variance and, as we have shown, playing Hawk always entails greater outcome variance than does playing Dove. Hence, Hawk is the riskier strategy. Because individuals played Hawk more often as inequality and the probability of local competition increased, in both the theoretical model (Fig. 2) and the behavioral experiment (Fig. 4), our study provides further evidence that inequality influences the taste for risk.

A clear implication of our work is that measures of inequality ought to reflect the relevant scale of competition. Indeed, a failure to do so may explain why observational studies sometimes do not find a positive relationship between inequality and conflict (e.g., Collier & Hoeffler, 2004; Fearon & Laitin, 2003; Neumayer, 2005). For instance, it has been argued that inequality is not a plausible explanation for homicide because, throughout the 1990s, income inequality increased while the homicide rate decreased in the United States (Pinker, 2011). However, this argument is based on national statistics aggregated over the population as a whole. Consequently, they can tell us only the excess chance that two people selected at random from a population of 250 million and scattered over an area of approximately 3.8 million square miles are likely to kill each other, given the extent of inequality between them. But homicide does not occur on so vast a stage; it occurs between individuals who are, by necessity, in close physical proximity and who are typically acquainted with one another (Daly, 2016).

We can make sense of the lack of a temporal association between inequality and homicide in this argument by recognizing two facts. First, the increase in inequality in the United States over the 1990s took place mainly in the upper tail of the income distribution (Piketty & Saez, 2003). Second, this appears to have led to the spatial segregation of wealthier households from the rest of the population (Reardon & Bischoff, 2011). Thus, inequality at fine scales of spatial analysis (e.g., neighborhoods) may have decreased at the same time that inequality at coarse scales of analysis (e.g., nationally) increased. This suggests that there has been a mismatch between the spatial scale at which inequality is measured and the spatial scale at which homicides occur. In other words, the study of homicide requires a considerably more local measure of inequality than has typically been used. By extension, our results imply that interpersonal conflict will be best predicted by inequality between competing individuals (Betzig, 1986; Daly, 2016; Wilkinson & Pickett, 2009), intergroup conflict will be best predicted by inequality between competing groups (Cederman et al., 2013; Muller & Seligson, 1987; Østby, 2008), and international conflict will be best predicted by inequality between competing nations (Hensel et al., 2006; Rider & Owsiak, 2015).

Supplemental Material

KruppOpenPracticesDisclosure – Supplemental material for Local Competition Amplifies the Corrosive Effects of Inequality

Supplemental material, KruppOpenPracticesDisclosure for Local Competition Amplifies the Corrosive Effects of Inequality by D. B. Krupp and Thomas R. Cook in Psychological Science

Footnotes

Acknowledgements

We are grateful to Curtis Bell, Martin Daly, Lindsay Heger, Adam Sparks, and Peter Taylor for comments. The views expressed in this article are those of the authors. They do not necessarily reflect the views of the Federal Reserve Bank of Kansas City or the Federal Reserve System.

Action Editor

Steven W. Gangestad served as action editor for this article.

Author Contributions

D. B. Krupp developed and analyzed the model, designed the experiment, and wrote the manuscript. T. R. Cook designed the experiment interface, collected and analyzed the data, and maintained the materials on the Open Science Framework. Both authors discussed the data analyses and commented on the manuscript, and both authors approved the final manuscript for submission.

ORCID iD

D. B. Krupp

Declaration of Conflicting Interests

The author(s) declared that there were no conflicts of interest with respect to the authorship or the publication of this article.

Open Practices

All data and materials, including those pertaining to the pilot experiments, have been made publicly available via the Open Science Framework and can be accessed at http://osf.io/ycren. The model, hypotheses, methods, and analysis plans were preregistered on the Open Science Framework prior to data collection and can be accessed at http://osf.io/ycren. The complete Open Practices Disclosure for this article can be found at https://journals-sagepub-com.web.bisu.edu.cn/doi/suppl/10.1177/0956797617748419. This article has received badges for Open Data, Open Materials, and Preregistration. More information about the Open Practices badges can be found at .

References

Anderson

L. R.

Mellor

J. M.

Milyo

(2008). Inequality and public good provision: An experimental analysis. Journal of Socio-Economics, 37, 1010–1028.

Barclay

Benard

(2013). Who cries wolf, and when? Manipulation of perceived threats to preserve rank in cooperative groups. PLOS ONE, 8(9), Article e73863. doi:10.1371/journal.pone.0073863

Barclay

Stoller

(2014). Local competition sparks concerns for fairness in the ultimatum game. Biology Letters, 10(5), Article 20140213. doi:10.1098/rsbl.2014.0213

Barker

J. L.

Barclay

(2016). Local competition increases people’s willingness to harm others. Evolution and Human Behavior, 37, 315–322.

Bartels

L. M.

(2010). Unequal democracy: The political economy of the new gilded age. Princeton, NJ: Princeton University Press.

Betzig

(1986). Despotism and differential reproduction. New York, NY: Aldine Transaction.

Boix

(2003). Democracy and redistribution. Cambridge, England: Cambridge University Press.

Buckley

Croson

(2006). Income and wealth heterogeneity in the voluntary provision of linear public goods. Journal of Public Economics, 90, 935–955.

Burton-Chellew

M. N.

El Mouden

West

S. A.

(2016). Conditional cooperation and confusion in public-goods experiments. Proceedings of the National Academy of Sciences, USA, 113, 1291–1296.

10.

Burton-Chellew

M. N.

May

R. M.

West

S. A.

(2013). Combined inequality in wealth and risk leads to disaster in the climate change game. Climatic Change, 120, 815–830.

11.

Cederman

L.-E.

Gleditsch

K. S.

Buhaug

(2013). Inequality, grievances, and civil war. New York, NY: Cambridge University Press.

12.

Collier

Hoeffler

(2004). Greed and grievance in civil war. Oxford Economic Papers, 56, 563–595.

13.

Côté

House

Willer

(2015). High economic inequality leads higher-income individuals to be less generous. Proceedings of the National Academy of Sciences, USA, 112, 15838–15843.

14.

Crosby

(1976). A model of egoistical relative deprivation. Psychological Review, 83, 85–113.

15.

Cynamon

B. Z.

Fazzari

S. M.

(2016). Inequality, the Great Recession and slow recovery. Cambridge Journal of Economics, 40, 373–399.

16.

Daly

(2016). Killing the competition: Economic inequality and homicide. Piscataway, NJ: Transaction.

17.

Daly

Wilson

(2001). Risk-taking, intrasexual competition, and homicide. In French

J. A.

Kamil

A. C.

Leger

D. W.

(Eds.), Nebraska Symposium on Motivation: Vol. 47. Evolutionary psychology and motivation (pp. 1–36). Lincoln: University of Nebraska Press.

18.

Fearon

J. D.

Laitin

D. D.

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

19.

Frank

R. H.

(2012). The Darwin economy. Princeton, NJ: Princeton University Press.

20.

Hensel

P. R.

McLaughlin Mitchell

Sowers

T. E.

II . (2006). Conflict management of riparian disputes. Political Geography, 25, 383–411.

21.

Jayadev

(2013). Distribution and crisis: Reviewing some of the linkages. In Wolfson

M. H.

Epstein

G. A.

(Eds.), The handbook of the political economy of financial crises (pp. 95–112). Oxford, England: Oxford University Press.

22.

Krupp

D. B.

Taylor

P. D.

(2015). Social evolution in the shadow of asymmetrical relatedness. Proceedings of the Royal Society B: Biological Sciences, 282, Article 20150142. doi:10.1098/rspb.2015.0142

23.

Kumhof

Rancière

Winant

(2015). Inequality, leverage, and crises. American Economic Review, 105, 1217–1245.

24.

Kümmerli

Burton-Chellew

M. N.

Ross-Gillespie

West

S. A.

(2010). Resistance to extreme strategies, rather than prosocial preferences, can explain human cooperation in public goods games. Proceedings of the National Academy of Sciences, USA, 107, 10125–10130.

25.

Maynard Smith

. (1982). Evolution and the theory of games. Cambridge, England: Cambridge University Press.

26.

Mishra

(2014). Decision-making under risk: Integrating perspectives from biology, economics, and psychology. Personality and Social Psychology Review, 18, 280–307.

27.

Mishra

Son Hing

L. S.

Lalumière

M. L.

(2015). Inequality and risk-taking. Evolutionary Psychology, 13(3), Article 1474704915596295. doi:10.1177/1474704915596295

28.

Muller

E. N.

(1995). Economic determinants of democracy. American Sociological Review, 60, 966–982.

29.

Muller

E. N.

Seligson

M. A.

(1987). Inequality and insurgency. American Political Science Review, 81, 425–451.

30.

Neumayer

(2005). Inequality and violent crime: Evidence from data on robbery and violent theft. Journal of Peace Research, 42, 101–112.

31.

Østby

(2008). Polarization, horizontal inequalities and violent civil conflict. Journal of Peace Research, 45, 143–162.

32.

Payne

B. K.

Brown-Iannuzzi

J. L.

Hannay

J. W.

(2017). Economic inequality increases risk taking. Proceedings of the National Academy of Sciences, USA, 114, 4643–4648.

33.

Piketty

Saez

(2003). Income inequality in the United States, 1913–1998. The Quarterly Journal of Economics, 118, 1–39.

34.

Pinker

(2011). The better angels of our nature: Why violence has declined. New York, NY: Penguin.

35.

Reardon

S. F.

Bischoff

(2011). Income inequality and income segregation. American Journal of Sociology, 116, 1092–1153.

36.

Rider

T. J.

Owsiak

A. P.

(2015). Border settlement, commitment problems, and the causes of contiguous rivalry. Journal of Peace Research, 52, 508–521.

37.

Taylor

P. D.

(1992). Altruism in viscous populations—An inclusive fitness model. Evolutionary Ecology, 6, 352–356.

38.

West

S. A.

Gardner

Shuker

D. M.

Reynolds

Burton-Chellow

Sykes

, . . . Griffin

A. S.

(2006). Cooperation and the scale of competition in humans. Current Biology, 16, 1103–1106.

39.

Wilkinson

R. G.

Pickett

(2009). The spirit level: Why greater equality makes societies stronger. New York, NY: Bloomsbury Press.

40.

Yamagishi

Hashimoto

Schug

(2008). Preferences versus strategies as explanations for culture-specific behavior. Psychological Science, 19, 579–584.

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.

0.00 MB

0.78 MB