Conditional Dispersal Among Primates and Human Property Crime Geoffrey Williams [email protected] September 1, 2016

Abstract Human adolescents, like adolescents in many other species, show an elevated pattern of risky and antisocial behaviors. For decades there has been speculation that this pattern can in part be traced to evolutionary drives; that as individuals reach sexual maturity and their bodies change, their appropriate social identity also changes. A specific version of the general conjecture is that some individuals within a social group will need to disperse to other groups upon reaching sexual maturity; the need to disperse will then lead to greater risk-taking, greater aggression and less attention to social norms during the dispersal period. Dispersal has already been suggested by a number of psychologists as a potential explanation for some criminal behavior. This paper advances the dispersal hypothesis in several ways. First, I review ways that dispersal fits well with certain human behavior patterns, particularly property crime. Second, I propose a simple model of dispersal. Third, I argue that dispersal in primates, both human and non-human, is likely to be hardwired, with predetermined triggers, timing, and resolution and that the neuroendocrinological mechanisms that control dispersal are likely to be fairly similar in both human and nonhuman primates. This suggests that (a) working by analogy across species may allow for important insights into human adolescence, property crime and social structures, and that (b) social scientists studying human property crime look for socioeconomic-based “triggers” of adolescent disengagement, risk-taking and asociality.

1

1

Introduction

Psychologists and other social scientists have begun to explore the idea that there may be strong evolutionary reasons for adolescent risk-taking and for criminal behavior (Ellis et al., 2012; Durrant and Ward, 2015). One idea put forward is that the phenomenon of dispersal may help to explain some forms of criminal behavior (Moffitt et al., 2011). The evolutionary view has been developing for several decades. This paper attempts to push the dispersal conjecture forward in three ways: first, I present theoretical and empirical arguments showing how property crime and other human behavior patterns can be explained by the evolutionary biology of dispersal; second, I propose a simple model of dispersal; third, I argue that dispersal in primates generally is likely to be hardwired, with predetermined triggers, timing, and resolution and that the neuroendocrinological mechanisms that control dispersal are likely to be fairly similar in both human and nonhuman primates. To review dispersal: it is a common pattern among many species, where individuals leave the territory they were born into. Social primates usually live in small groups based in part on family relationships, and thus dispersal requires the individual to shift his or her membership from one group to another. The departure from the natal group is called emigration, and the successful shift from one group to another is called transfer; with regard to primate dispersal I will use dispersal, transfer and emigration interchangeably. This discussion will focus on male dispersal and use the masculine pronoun, as male dispersal is most relevant to property crime, but in many species females are the more likely to disperse. The dispersing individual first disengages or emigrates from his natal group. During periods of emigration, at least in some species, general risk-taking and confrontation of hierarchies is common. He frequently experiences a period either as a singleton or member of a small singlesex group, and then attempts integration into a new group; multiple emigrations are common. For primate species we have data on, emigration probability is highest in the period between sexual maturity and achievement of full size. Crucially, individuals facing poor prospects, such as those lower in the dominance hierarchy, are more likely to emigrate. It is important to note that this entire process is risky and integration into a stable new group is generally not assured; for this reason, some individuals may emigrate multiple times.

2

This pattern of disengagement from the natal group during adolescence, leading to a period of being “unattached” followed by a period of reattachment can help to explain a number of the patterns we see in the data on property crime. There are a number of stylized facts that emerge in the literature on property crime1 : • First, while rates vary between societies, property crime appears to be general and common to all modern societies. • Second, for populations which we have data, property crime activity follows a consistent pattern, peaking around age 17-18, and dropping rapidly after that peak. • Third, among individuals, property crime activity tends to be concentrated in even shorter periods; it is rare for individuals to be active for more than two years. • Fourth, very few acts of property crime seem to earn significant returns; to the degree that earnings from crime serve an economic purpose, it tends to be funding the use of drugs or alcohol. • Fifth, potential costs of property crime to the criminal appear significant, including arrest, injury and death. • Fifth, although property crime does not act as a significant substitute for legitimate work, there is a negative correlation with socioeconomic status. • Sixth, there is strong evidence that property crime activity stops or declines significantly after the individual marries or finds a fulltime, permanent job, and a number of criminologists believe that the stability of marriage and employment cause, at least in part, the desistance from crime. I argue that the general dispersal argument can help us to understand these patterns if we consider that there is a great deal of continuity across species in some of the underlying structures. In terms of the undergirding neurological and endocrinological systems, biologists have documented remarkable similarities in adolescence across a wide range of mammals; lab mice have been useful in helping to understand basic issues of human adolescence, including the specific chemical changes (Spear, 2000). Regarding the social structures that human societies are 1

Which I define as burglary, larceny, theft, motor vehicle theft, arson, shoplifting, and vandalism

3

based on, we can see a close analogue between the small groups in other social primate species and the “band” structure found among human societies without an overarching tribal or state structure (Fukuyama, 2011: pages 53-55), and that vestiges of this type of social group appear in complex societies (Richerson and Boyd, 1999). An important contrast to nonhuman primates is that in humans, when individuals have fully joined a social group, they behave in a highly cooperative way; nonhuman primates show significant cooperation but much less than is seen among humans. Much theoretical and empirical evidence suggests that high levels of cooperation within a human group is maintained in large part by careful monitoring of individual behavior and harsh punishments, even deadly punishments at the extreme, for misbehavior by members (Mirrlees (1974); Merry (1997),Binmore (2007: chapter 5, pages 73-74), ?, ). At the band level, anthropologist Ernest Gellner referred to this as “the tyranny of cousins.” It would then make sense that some human adolescents go through a specific short-lived period of social disconnection from their natal social group, and then over time form new connections with a new social group. Property crime activity by humans would therefore be a byproduct of this period where the individual transfers from his natal group to another social group. Specifically, the disengagement directly around the period of transfer leads the individual to ignore social rules and structures he or she would otherwise respect; this disengagement would drive the most extreme risky and antisocial behavior (such as property crime) among adolescents and young adults. Upon creating social links with a new and stable group, for example with a secure job or romantic relationship, the individual returns to a more staid pattern, avoiding unnecessary risks and abiding by social customs. If this is the case, the patterns of property crime - with generally fairly short duration and rapid onset and cessation - would seem likely to be caused by a strong and detectable sequence of neuroendocrine activity, which radically changes the individual’s attention to and investment in his or her pre-existing relationships and social environment, and may also temporarily reduce the individual’s foresight and/or increase riskseeking. The combination enables both the emigration and the criminal activity. Researchers focusing on humans thus have the opportunity to explore the “triggers” that precede this pattern, most likely with a major socioeconomic status component. Thus, within humans, emigration leads a male to shift from (a) embedded and carefully

4

monitored by fellow group members to (b) isolated and independent to (c) again embedded and monitored. During the period of isolation, the individual is no longer under the group’s strictures, and hence has much greater discretion to commit antisocial and predatory acts with significantly lessened fear of punishment. Under this hypothesis, and following the pattern in other primates, the lower the individual’s social rank, the higher the likelihood of emigrating to another group. As with other primates, the dispersal, while not consciously planned and largely driven by opaque neuroendocrinological mechanisms, is a highly rational tradeoff between the costs of leaving the natal group’s support and the benefits of avoiding inbreeding. As with other primates, not all individuals or individual emigrations would be guaranteed to succeed; on the one hand, some individuals might be less well-prepared to form strong bonds or work in stable positions; on the other, the “secure berths” granted by good jobs and good relationships will not be evenly distributed and equally available among all regions, time periods, and social groups. If, for any of these reasons, an emigration to a new group does not lead to stable connections, the individual would then emigrate again, in the extreme, perhaps throughout his life. As discussed above, such a pattern is found among nonhuman primates. For comparison purposes, the nonhuman primate species I focus on is mountain gorillas Gorilla beringei beringei, as a patrilocal primate species with significant, documented, male emigration. The structure of the paper is as follows. In section 2 I present a simple theoretical model of what dispersal seems likely to have been in simple human band-level societies (i.e. before the development of tribal, religious or state links across groups). In section 3 I review the literature on property crime and other behavior patterns that seem to fit the dispersal story. In section 4 I review the literature on dispersal, concluding with a brief discussion of what is known or believed about emigration in humans. In the concluding section I discuss research that might support or falsify this conjecture.

5

2

Theory

To clarify the specific process I have in mind I present a simple theoretical model of dispersal in a patrilocal primate species with no metagroup social structures. That is to say, a model designed to explain male dispersal in (a) gorilla populations and (b) human populations at the band level of social development. There are two sets of players. The first is comprised of a single male primate, i and the second is G primate groups, indexed g, leading to total player count of 1 + G. Each player is a risk-neutral utility maximizer, all sharing an identical discount factor δ, 0 < δ < 1 In each period t ∈ {0, 1, 2, 3, . . .}, the male primate plays a game against each primate group. The game between the individual and each group looks like this: Group C

D

C

aig , agi

b, d

D

d, b

0, 0

Individual

where C stands for cooperation (it can also be thought of as compliance, by the individual) and D stands for defection. The payoffs to mutual cooperation, aig (Xi , Yg , Zig ), agi (Xi , Yg , Zig ) are each a function of attributes of i, g and interactions between attributes of i and g. For example, the overall strength of i would be part of vector Xi which would likely lead to higher payoffs for the individual in all groups. If g has a territory that is particularly fertile or attractive, this would be a component of Yg increasing the payoff for both g and i. Finally, a high degree of genetic relatedness Zig between i and the adult females of g would reduce the payoffs to cooperation significantly for both g and i. Additionally, the payoffs have the following constraints:

d > arg max(aig ) > arg min(aig ) > (d + b) g

g

d > arg max(agi ) > arg min(agi ) > (d + b) g

g

(d + b) > 0 > b

6

Thus the game between i and any g is a prisoner’s dilemma. Cooperation is costly, so i can only cooperate with one group in any period t. Group g 0 is i’s natal group, and will play C with i in period 0. It should be noted that the strategy of C for the individual includes a wide range of possible behaviors - in his efforts to improve his outcome within any group g, i could choose aggressive confrontation, or extensive networking, or working closely with one or two other group members (some examples of how individual chimpanzees have approached this challenge can be seen in Waal (2007)). For our purposes, all of these different approaches are different ways of cooperating with the group and complying with its rules; they all are identical within the context of the simple model outlined here. The strategy of D for the individual is his simply ignoring the rules of the group; if the group is not attempting to cooperate with him, there is no loss. For the group, the distinction between C and D is the difference between accepting the individual as one of it’s own, or treating him like an outsider. Neither i nor g know the values aig (Xi , Yg , Zig ), agi (Xi , Yg , Zig ) until the first cooperation between them, but all other aspects of the game are common knowledge.

2.1

Equilibrium

I now outline an equilibrium which I believe to be particularly salient, describing each player’s strategy in terms of a finite automota or machine (Osborne and Rubinstein (1994: chapter 9), Abreu and Rubinstein (1988)).

2.1.1

Natal Group g 0

For group g 0 , a machine with two states: 1) Play C. If i plays C, stay in 1. If i plays D, switch to 2. 2) Play D. Stay in 2 regardless of i play.

2.1.2

Other Groups

For all groups g ∈ G, g 6= g 0 , a machine with three states: 1) Play D. If i plays C, switch to 2. If i plays D, stay in 1.

7

2) Play C. If i plays C, stay in 2. If i plays D, switch to 3. 3) Play D. Stay in 3 regardless of i play.

2.1.3

Individual i

For individual i, a machine with G states, where a ¯ is a key parameter determining play: 1) Play C with g 0 , D with all other groups. If aig (Xi , Yg , Zig ) ≥ a ¯, stay in 1. If aig (Xi , Yg , Zig ) < a ¯, switch to 2. 2) Randomly select one group g 00 ∈, g 00 6= g 0 , play C against g 00 , play D against all other groups. If aig (Xi , Yg , Zig ) ≥ a ¯, stay in 2. If aig (Xi , Yg , Zig ) < a ¯, switch to 3.

...

G) Play C with the last group, g. Stay in G. To fully describe i’s strategy we must define a ¯. This is lowest payoff that i will accept and not emigrate to another group. Thus, it must be as great as the individual’s expected payoff from emigrating. It can be seen that if the groups are playing the strategies described above, a ¯ will need to satisfy the following weak inequality

a ¯ 1−δ | {z }

Payoff of staying

3

3 1 − δp (d + b/δ + E(a)/δ 2 ) + δ p ≥ 3 1−δ | {z } |

Cost of further search

Z a ¯

∞

xf (x)dx {z }

Benefit of further search

where f () is i’s belief regarding the probability distribution of aig across all G groups and ∞

Z p=

f (x)dx a ¯

Z

∞

E(a) =

xf (x)dx −∞

2.2

Discussion

It can be seen that, however i forms his understanding of the range and distribution of aig , he then tries to move as efficiently as possible towards a “good” value. In the process, he engages in cooperation with groups. Until the payoff from cooperation

8

attains his goal of a ¯ he will leave by defecting even as they cooperate with him before moving to a new group. The payoff from defection at the point of departure is the analogue to theft.

3

Property Crime Patterns

Property crime appears to be a major phenonomenon across effectively all countries. While data is far from perfect, and difficult to compare easily, and frequencies vary considerably, it appears to be important among every human population and in every region of the world (Lynch and Pridemore, 2011).

3.1

Periods of Activity in Aggregate

Criminal behavior within a large population tends to follow a fairly standard general relationship across different ages: criminal activity increases steeply from early teenage years to late teenage years, and then slowly drops as indivividuals age (Blumstein et al., 1988; Sampson and Laub, 2003). The drop is fastest for property crimes (Blumstein et al., 1988: for example, Figure 1). Laub and Sampson (2003a) found a very rapid decline for property offenses in their study of delinquent boys to age 70, a much slower decline in violence (from a lower peak) and stable frequency of alcohol and drug crimes.

3.2

Periods of Activity at the Individual Level

At the individual level, the pattern of onset and decline in criminal behavior is generally even more rapid. In his study of persistent thieves, Shover (1996) emphasizes that in fact, persistence is very rare among thieves. “Of all male juveniles who ever engage in serious delinquency, most do so infrequently and do not persist at it more than a few months or years. Few are or become chronic offenders. Put differently, the age when they begin committing crime is followed quickly by the age when they cease such activity.” He notes a study of thieves in Canada, where the sample members were active for an average of only 28 months before stopping, and concludes “[t]he overall patterns is clear: By their early to mid-20s, many young chronic offenders cease accumulating serious criminal charges. (pages 119-120)” Sullivan (1989) tracked youths in three

9

different neighborhoods in New York City, and found that most stopped theft and robbery within one or two years of beginning. Williams (2015) looks at self-reported property crime behavior in the National Longitudinal Study of Youth, 1997 Cohort, across fifteen years, and finds that most individuals who report stealing are only active in one year during this period, most theft activity is reported by individuals who are active in only one or two years, and most value is reported stolen by individuals who are active in only three or fewer years.

3.3

Socioeconomic Status and Property Crime

At a high level, there seems to be a general consensus among researchers that there is a negative relationship between socioeconomic status and crime, although exactly how strong is debated (Jr and Rhodes, 1961; Tittle et al., 1978; Braithwaite, 1981). There are a number of explanations provided for the link, many basically being a twist on the elementary, obvious, economic explanation: for people who are poor the need for money is greater and the costs of other considerations (lost time, threat of punishment, damage to future work possibilities) are less. The classic statement of economists on crime is Becker (1968), one of the first articles to detail it is Ehrlich (1973), Grogger (1998) develops and estimates a model at the individual level, and Lochner (2004) develops a multi-stage model of this. Sociological theory has approached the question of the link between socioeconomic status and crime in several ways. One approach is strain theory, whereby individuals who are unable to some of the social and economic ideals of their culture feel a strain, and turn to crime as an alternative approach. Another is social learning, whereby individuals follow the culture around them. A third is social control, which tends to focus on attachment. Each of these theories may predict a different type of relationship between socioeconomic status and crime, and each has different consequences; for a review of these different theories, see Sullivan (1989: pages 3-8) or Lilly et al. (2010).

10

3.4

Earnings or Returns from Property Crime

Although there is a strong belief that low socioeconomic status and crime are linked, it has proved very difficult to effectively document the obvious corollary that property crime provides economic rewards that make it a rational alternative or supplement to legitimate work. A general pattern in much crime research is that earnings from crime are usually fairly low. For example, Reuter et al. (1990) find that one-quarter of active (i.e., reasonably successful and experienced) drug dealers in their sample in Washington D.C. earn only $50 per month. Levitt and Venkatesh (2000) comment on drug gang finances “[t]aken as a whole, our results suggest that even in this financially sophisticated corporate gang, it is difficult (but not impossible) to reconcile the behavior of the gang members with an optimizing economic model without assuming nonstandard preferences or bringing in social/nonpecuniary benefits of gang participation...the possibility of suboptimal decision making cannot be eliminated from consideration.” Getting good data on earnings from crime is difficult, and making an apples-to-apples comparison between different types of crime, across different time periods and populations is all but impossible. However, it can be said that earnings from property crime generally do not appear any better than drug crime or other non-predatory black market activities, and are if anything less remunerative. Williams (2015) finds that self-reported earnings from theft in the NLSY 1997 data average $30 per act of theft (this is for acts stealing items worth MORE THAN $50), with a substantial minority of thieves earnings $0 per act of theft. This is broadly consistent with the best available data sources on reported losses in shoplifting and burglary, as well as the literature on fencing stolen goods. Very few studies using individual level data have found a negative relationship between earnings and crime - the only two I am aware of are Grogger (1998) and Williams and Sickles (2002) - and some have found a positive relationship (Bushway (2011) provides an excellent general review). An important corollary to the low earnings from theft is that ethnographies have found that earnings from property crime are very frequently used for either showy consumption or alchohol and drugs. Some literature refers to this as “life in the fast lane” (for example, Shover (1996) and Gibbs and Shelly (1982)). Wright and Decker (2011) study a group of armed robbers who appear to be unusually persistent and experienced. Their description of the robbers’ motivation

11

is worth quoting at length: A majority of the offenders in our sample spent much of the money they obtained though armed robbery to pursue what was for them an open-ended quest for excitement and sensory stimulation. Forty of the fifty-nine offenders who told us what they did with the proceeds of their stickups said they used most of the cash to initiate or sustain various forms of illicit action, including gambling, drug use and heavy drinking....While the offenders often referred to such activities as partying, there is a danger in accepting this definition of the situation uncritically; the activities were pursued with such an intensity and grim determination that suggest something far more serious was at stake. (p. 35) They term such activity “desperate partying.” Even in ethnographies that look at crime among lower-income youth, the non-economic motivations play a major role, and the income from property crime seems to go largely to entertainment and status goods (see Sullivan (1989: chapter 5, also 6-8)). Economic researchers have tended to work with links between aggregate regional (usually state-level) economic opportunities and criminal behavior. Bushway (2011) provides a good general review of the literature, and notes that effect sizes have been small in the aggregate studies. Raphael and Winter-Ebmer (2001) find a 1 percentage point decline in unemployment leading to 1 to 5 percent drop in property crime, which Zimring (2006: page 66) describes as at the “high end of current estimates.” It is striking that some of the strongest effects of economic incentives have been found at the monthly level: Foley (2011) compares a range of US cities and finds that in cities where welfare payments are grouped at the beginning of the month, property crime rates are at their highest at the end (the highest effect being for robbery, which is 25% higher at the end of the month). That this effect is substantially stronger than the effect of overall wages and unemployment suggests that economic motivations dictate timing for people who are already inclined to commit property crime. There does appear to be evidence of a small group of persistent successful thieves, who acquire enough earnings to “make a living” from property crime. It is important to note that such thieves are extremely rare, at most 0.01% of the population, even in studies that provide evidence of them. Based on the ethnographic evidence about such thieves, they seem likely to be analogous to the “serial emigrators” seen in many primate tribes. While such thieves almost certainly have

12

responsibility for a disproportionate share of total property crime, it is still a minority share.

3.5

Costs of Committing Property Crime

While it is normal to focus on the costs that property criminals inflict on their victims and society at large, many researchers have noted in passing that property criminals incur significant costs themselves, including arrest, incarceration, injury and even death. The probability of being arrested for any individual offense appears to be very low, but because many individuals commit numerous offenses, the odds of arrest for an individual can be quite high. Additionally, there are substantial risks of being injured or killed while committing crime. Burglars, robbers and other criminals show a strong awareness of this (Wright and Decker, 2011; 1994). Tracking over the individual life, differences in delinquent or criminal activity appear to lead to very significant differences in mortality (Sampson and Laub, 2003: figure 5.8). Engaging in property crime, even if the individual avoids violence, incurs significant risks for the offender.

3.6

Lifestyle of Active Property Criminals

It is worth noting that individuals who are actively committing property crime appear unusually disconnected from normal social arrangements, even in comparison with other criminals. The literature on drug dealers presents numerous examples of active drug dealers who are living “normal” lives, such that their neighbors would have no idea of their actual job. Examples include the drug dealers in middle-class Chicago in Pattillo (2013), the suburban California dealers in Adler and Adler (1983), or the successful crack dealer in a middle-class apartment building in Dunlap et al. (1994). In contrast, the examples of property criminals who are living “normal” lives are all but nonexistent - the only example I have come across is the suburban burglar at the opening of Shover (1996). Far more standard are two patterns: first, adolescent males who have either just moved out of their parents or are likely to do so in the near future (the young men studied in Sullivan (1989)), and second, individuals who are living very unsettled lives. A good example of the second type are the armed robbers studied in Wright and Decker (2011: page 38): During our interviews, we asked thirty-two of the offenders to tell us about their living

13

arrangements; twenty-two said that they seldom slept at the same address for more than a few nights in a row, preferring to move from place to place as the mood struck them....In effect, these offenders live as ‘urban nomads,’ ranging across the streets and alleys that connect the high-crime inner-city neighborhoods of Saint Louis....

3.7

Property Crime and Marriage, Work

While much popular discussion (and economic analysis) focuses on unemployment and low wages as a cause of property crime, a number of sociologists and criminologists have found evidence that marriage and work cause desistance from property crime. The mechanism does not appear to be a simple function of the cash increase in opportunity cost, but instead a more complicated and slower mechanism of commitment. Laub and Sampson (2003b: particularly chapters 2, 3 and 6) look at the role of marriage and work in leading to the desistance from crime in a group of delinquent boys tracked to age 70. They present a very interesting discussion of the role of commitments, or “side bets” that lead to a change in behavior over time. Other researchers have argued that in contrast to what might be predicted under some economic models, the causal chain does not seem to go from an increase in wages to maturation and less illegitimate behavior, but instead from maturation and “settling down” to commitment to a regular job, and an increase in wages (Osterman, 1980: chapter 3).

3.8

Discussion

To review some of the basic issues: Property crime is very largely an activity of young males, usually undertaken for very short periods of about a year or less. While frequencies likely vary considerably, it seems to be a significant phenomenon in effectively all societies. The direct monetary gains appear generally very low. Moreover, the criminals who show the most persistence and success spend their earnings largely on gambling, drugs and alcohol. During periods of property crime activity, individuals are generally unusually disconnected from normal social relationships. The activity is risky and costly for the criminals themselves, frequently leading to imprisonment, injury and death. Interpreted in its own terms, then, we have the puzzle of a common and quite risky activity, where the direct material benefit is very low. From an economics perspective, it does not appear to yield a benefit in any way commensurate with the costs. From an evolutionary perspective,

14

the risks seem likely to reduce fitness for no reason. Viewed from the perspective of common sense, it just seems like silly adolescent risk-taking. The dispersal hypothesis provides a potential explanation, suggesting that the material rewards of theft are simply a minor side effect; the real benefit of the period of property crime activity is the disconnection from the natal social environment and the subsequent transfer to a more advantageous location in a group that provides better expectations.

3.9

Related Phenomena in Human Societies (particularly the

United States) The dispersal hypothesis also explains a number of other patterns in modern societies, in particular in census coverage and high school graduation statistics. This section looks at other patterns of costly disengagement among young adult males, particularly of low socioeconomic status. Because some phenomena are more easily or more extensively studied using ethnic categories rather than true socioeconomic categories, I use these as proxies.

3.9.1

Census Undercoverage

The United States Census has had an ongoing issue with undercoverage since the founding. For the past 70 years, the population most significantly suffering undercoverage has been young black males2 . Savage (1982) notes that the 1940 census counted 4.5% fewer white males aged 21-35 than the Selective Service registration rolls, and 18% fewer black males aged 21-35 (page 197). Siegel (1968: table 2, pages 42-43) provides estimates for undercounts in the 1950 and 1960 census, broken out by age, gender and nonwhite/white. The greatest undercoverage in any population is 19.7% among nonwhite males age 25-29 in the 1960 census. The average for nonwhite males is estimated at 10.9%. Nonwhite females show a maximum undercoverage rate of 10.1% at age 15-19, and an average of 8.1%. This can be compared to an average for white males in the same census of 2.8% and a maximum of 4.3% among white males age 20-24. More recent estimates for the 2000 Census suggest that despite decades of work, things have 2

The conditional dispersal story would be supported by a higher rate of undercoverage for males for population with low socioeconomic status than those with higher status. Ideally it would be possible to compare within ethnic group not simply across ethnic groups, but I am not aware of undercoverage estimates broken down by SES alone

15

only improved somewhat. Robinson et al. (2002) in Table 7 pages 31-32, estimate Black male undercoverage as about 4.90-5.81% overall, but perhaps as high as 9.62% for those aged 18-29, and 10.98% for those aged 30-49. Estimates for Black females are generally half that, and for white (nonBlack) populations the central tendency of estimates is close to zero and sometimes negative. Since census data tends to be used as background and sampling frame for many other statistical series, this undercoverage has significant consequences for almost any large scale statistical analysis of the United States population. For example, Cohen and Lynch (2007) estimate that census undercoverage of the black male population drives a substantial part of the divergence between crime estimates in the Uniform Crime Reporting System (UCR) and National Crime Victimization Survey (NCVS). While undercoverage clearly has multiple explanations, the very strong continued pattern for so many decades among young mature males, with a particularly strong pattern in an ethnic group where average socioeconomic status is substantially below the United States average shows a high level of disengagement from mainstream social structures strongly consistent with the conditional dispersal explanation.

3.9.2

School Dropout Patterns

The issue of teenagers dropping out of school is clearly complicated, with numerous factors involved. However, it is worth noting that research strongly shows a major risk factor is low socio-economic status (Hammond et al., 2007; Jimerson et al., 2000). In fact, many studies point strongly to (a) individual school engagement/achievement and (b) family background as the two most important factors in predicting school dropout. In a report by Time magazine reporter Nathan Thornburgh3 the role of socioeconomic status is given pride of place: “The national statistics on the topic are blunt: according to the National Center for Education Statistics, kids from the lowest income quarter are more than six times as likely to drop out of high school as kids from the highest. And in Shelbyville, nearly every dropout I met voiced a similar complaint: teachers and principals treat the ‘rich kids’ better.” One obvious potential counter explanation for the link between SES and dropping out is 3

“Dropout Nation”, Sunday April 9, 2006

16

that school achievement and family background may both be largely driven by genetic factors that heavily determine the student’s achievement at school as well as parental labor market success. The explanation is simple and is likely to play a role. However, it should be noted that there is evidence that among adopted children, dropout rates tend to be significantly lower than the general population, despite evidence that learning disabilities are higher (Silver, 1989). For example, in Bruce Sacerdote’s public use data set on the Holt adoptees (http://www.dartmouth.edu/~bsacerdo/holt_adoption_public_use2006.dta), 99% have at least a 12 years of education. A study of educational attainment among adoptees in the United Kingdom found that adoptees, while not necessarily earning high degrees at a substantially different rate than non-adoptees, were noticeably more likely to attain some qualification than non-adoptees were (Maughan et al., 1998: Table 6, Page 677).

3.9.3

Disengagement Unexplained by Other Measures

A standard explanation for disengagement from the labor market or from education is lack of adequate preparation and skill formation (Heckman, 2006). And indeed, low skill attainment is likely to be a very important explanation for many of the issues discussed above. But there is strong evidence of a post-adolescent increase in disengagement for males in lower SES groups. For example, when a noted scholar of life cycle skill formation analyzed labor market and criminal justice outcomes in the United States, he found that among black males, combining all measures of pre-market factors was not sufficient to explain outcomes. The combination of ability and training could only explain 40 percent of the gap in work and incarceration outcomes relative to white males (Urzua, 2008). While it is almost certain that labor market discrimination plays a central role in this, active disengagement by the individuals themselves, consistent with a conditional dispersal argument, seems like a strong potential explanation.

4 4.1

Condition Dependent Dispersal Background on Dispersal

The general pattern of dispersal is well-established across a wide range of animals; most higher vertebrates show high rates in at least one sex. The following summary of what is known about

17

primate dispersal and emigration owes a great deal to Pusey and Packer (1987) and Handley and Perrin (2007). While there are some consistent patterns, there is also a great deal of variation, both within and across species. The following attempts to focus on the most important patterns, at the risk of oversimplifying. Individual animals follow one of the two strategies of dispersal versus philopatry (staying in the same group and/or habitat). Dispersal has several aspects. There is the actual departure from the group, termed emigration, and transfer, which combines emigration with successful “immigration into a group containing breeding members of the opposite sex”. Another important distinction is between natal emigration, from the natal group or habitat and secondary emigration - any movement after the natal transfer. The timing of emigration is usually between puberty and attainment of full size. Frequently there will be multiple emigrations, presumably in these cases transfer is not successfully achieved for the first emigrations. Sex-biased dispersal (frequently shortened to SBD) is common among many species. Obviously, if a species shows a relative bias towards one sex dispersing, the other sex then shows a bias towards philopatry, thus species which show male-biased dispersal (MBD) are matrilocal, while species with female-biased dispersal (FBD) are patrilocal. Male-biased dispersal is common among mammals, while FBD is more common among birds. Across primate species, males are generally more likely to disperse, but in some (gorillas and chimpanzees, specifically) females appear more likely. It is generally believed that in humans females are more likely to disperse (Seielstad et al., 1998) and that humans are patrilocal. It is worth emphasizing that across all mammal species, the sex bias in dispersal can be strong but is rarely complete - in nearly all species both sexes show at least some dispersal. The patterns vary considerably. In a number of species, individuals are “evicted” by aggressive behavior by conspecifics (In a small minority of primate species, females are actually abducted). Voluntary emigration, either directly to another group, or sometimes as part of a small, usually all-male, group, is very common. All-males bands, apparently a transitional phase from natal group to successful transfer, are frequent. These groups are usually highly affiliative, unlike the competition between males in bisexual groups. Reception in the new group is important and varies considerably. In some cases immigrants are treated aggressively, in some cases there is preferential treatment, for example by members of the opposite sex.

18

The new group that an individual transfers to is frequently one that is adjacent, already containing successful immigrants from the individual’s natal group, offering mating opportunities, and in a good habitat. That is, individuals seem to favor groups that they have some previous links to and information about, and they appear to “trade-up” to a better social setting. The overall pattern of emigration and transfer seems to be fairly risky, with significant costs. Pusey and Packer (1987) review three explanations for dispersal in primates, specifically (a) it allows access to unexploited or underexploited resources (b) it enables individuals to minimize intrasexual competition and (c) it helps individuals to avoid inbreeding. They argue that the first plays a minimal role, the second can be important in some species, but the third appears to be the most important and general cause. The cost and risks of transfer are worthwhile to avoid the increased mortality of inbreeding. Handley and Perrin (2007) suggest that inbreeding avoidance is somewhat problematic as an explanation, and suggest that other issues such as control of resources or cooperation with kin may help to explain the dispersal/philopatry trade-off.

4.2

The Role of Natal Environment

Dispersal has developed rapidly as an area of research over the past few decades. One area that has become more significant in the past 10-20 years of research is the issue of condition-dependent dispersal (i.e., dispersal when only some of the individuals disperse, based on conditions). A key area of this research is analyzing the impact of natal environment on dispersal. Recent surveys include Benard and McCauley (2008) and Bowler and Benton (2005). Research in personality has also begun to play a role in this (Handley and Perrin, 2007). This general work in ecology relates well to a range of studies looking at primate populations. Since most primates are unusually social animals, dispersal is less about location or foraging patch and much more about group membership. Thus, the environment of the natal group, including issues of social hierarchy and relative numbers of breeding adults, become vital. Colvin (1986) looks at timing and circumstances of macaques on the Cayo Santiago preserve. Macaques are a largely matrilocal species, where males are highly likely to emigrate but females rarely do. There are several patterns in Colvin’s work that are of interest. First, social rank is an important predictor of the timing of emigration. Lower-ranked males emigrated earlier than higher-ranked males (effectively all males in his study emigrated). As an important side

19

note, lower-ranked males emigrate in a pattern highly reminiscent of the age-crime curve seen in humans (although macaques migrate soon after puberty and before reaching full size, while crime peaks at or slightly before age 17-18, the point where human males reach full size). Second, Colvin finds that social rank likely affects individuals in one of two ways: One way is that peer relationships show a strong difference by rank, such that lower-ranked males are much more likely to be exposed to peer aggression, much less likely to initiate contact with other peers, and less likely to intercede in conflicts. Generally, lower-ranked males appear to be less active within the group, both by choice and because of pressure by peers. A second potential way that rank affects emigration is via the individual’s relationship with his mother - higher-ranked mothers appear to remain more engaged in their sons’ lives over time. Harcourt and Stewart (2007) survey gorilla society. This is potentially of high relevance for understanding human social relationships, as gorillas are patrilocal: males generally stay while females leave. However (as is generally true of all primate species) the pattern is not absolute, as males frequently do emigrate. Their discussion of the male “decision” to stay or emigrate (pages 281-292) is highly illuminating. There is substantial variation between different gorilla populations in the rate of emigrating vs staying for males. Emigrating is generally very costly for gorilla males as it effectively requires building up a group from scratch, as an emigrating male then needs to persuade females join him. They believe that the key issue is how large a group can be supported within the environment; if large groups can be supported than this is generally optimal for the subordinate males and most females.

4.3

Dispersal in Humans

Figure 1 compares the timing of dispersal in male gorillas with initial property crime activity in humans. To allow for comparison of timing across species, I normalize by age of puberty and age of full size attainment. Although the sample size for gorillas is limited, the similarity in timing is strong. The major avenue of study of dispersal in humans for the past 20 years has been genetic analysis. Seielstad et al. (1998) provided the first global study, looking at differences in variation in mitochondrial DNA (mtDNA) and Y chromosomes. They found that geographically distant populations were much more “different” when comparing Y chromosome genetic structures than

20

when comparing mtDNA structures, suggesting higher female migration rates than male. A later study, Wilder et al. (2004), found a very different pattern, but evidence since then, based on comparing mtDNA and Y chromosome variation, seems to support the pattern that within small regions patrilocality is more important. The lack of a clear consensus on the genetic pattern in humans makes it tricky to predict the exact pattern of dispersal, but the aggressive dispersal of lower-status human males seems perfectly consistent with the range of evidence and explanations.

5

Conclusion

It is worth emphasizing what this conjecture offers that is novel and distinct. There is a significant folk wisdom and social science support for the general idea that low socioeconomic status is causally linked to risky, illegal and antisocial behaviors, and underinvestment in education or legitimate careers. However, most explanations posit fairly transparent, universal, continuous and well-understood mechanisms as the explanation. For example, economists have explained criminal behavior as in part driven by an individual’s low opportunity cost and/or lower lifetime earnings. Other intuitive explanations suggest that inequality may lead to feelings of frustration and resentment, which then lead to disengagement and alienation. The dispersal explanation, in contrast, emphasizes a neuroendocrinological mechanism which is largely subconscious, almost entirely limited to adolescence/young adulthood, and generally discontinuous and unpredictable. It suggests certain areas may be particularly fruitful for research. First, it predicts that property crime and other risky behaviors are likely to be caused by “triggers”, which are closely related to socioeconomic status. There is some hope that the exact nature of the triggers may be determined from human data; recent research that is congruent with this approach, outside the area of property crime, is Kearney and Levine (2014) and Balsa et al. (2014). Second, it suggests that social scientists may want to move beyond simply tabulating risky behaviors (i.e., acts of theft, drug use, etc) and begin to focus more attention on how adolescents change their social identity, residential location, geographic location, and institutional affiliations.

21

The dispersal hypothesis predicts that risky behaviors are going to be associated with substantial shifts in all of these areas. It may be useful to look at how residential moves, for example, correlate with theft. Third, it suggests that there may be strong similarities between dispersal patterns in humans and in the great apes and Old World monkeys, which can guide research in a number of ways. The three part pattern of emigration, movement and immigration is likely to play out in similar ways across these species, and there is likely to be a great deal of recycling of neurological triggers and endocrinal responses across these species. If, for example, the level in the bloodstream of a particular hormone is found to radically increase (decrease) during the emigration period in one species, it becomes quite probable that (a) a similar change may be found in other species and (b) it plays a key role in the reduction (increase) of social bonding.

22

References Abreu, Dilip and Ariel Rubinstein, “The structure of Nash equilibrium in repeated games with finite automata,” Econometrica: Journal of the Econometric Society, 1988, pp. 1259– 1281. Adler, Patricia A. and Peter Adler, “Shifts and oscillations in deviant careers: The case of upper-level drug dealers and smugglers,” Social problems, 1983, pp. 195–207. Balsa, Ana I., Michael T. French, and Tracy L. Regan, “Relative deprivation and risky behaviors,” Journal of Human Resources, 2014, 49 (2), 446–471. Becker, G. S., “Crime and Punishment: An Economic Approach,” Journal of Political Economy, 1968, 76 (2), 169–217. Benard, Michael F. and Shannon J. McCauley, “Integrating across life-history stages: Consequences of natal habitat effects on dispersal,” The American Naturalist, 2008, 171 (5), 553–567. Binmore, Ken, Game theory: a very short introduction, Vol. 173, Oxford University Press, 2007. Blumstein, Alfred, Jacqueline Cohen, and David P. Farrington, “CRIMINAL CAREER RESEARCH: ITS VALUE FOR CRIMINOLOGY,” Criminology, 1988, 26 (1), 1–35. Bowler, Diana E. and Tim G. Benton, “Causes and consequences of animal dispersal strategies: relating individual behaviour to spatial dynamics,” Biological Reviews, 2005, 80 (02), 205–225. Braithwaite, John, “The myth of social class and criminality reconsidered,” American Sociological Review, 1981, pp. 36–57. Bushway, Shawn D., Labor markets and crime Crime and public policy, Oxford University Press New York, NY,

23

Cohen, Jaqueline and James Patrick Lynch, “Exploring differences in estimates of visits to emergency rooms for injuries from assaults using the NCVS and NHAMCS,” 2007), Understanding Crime Statistics: Revisiting the Divergence of the NCVS and UCR, 2007, pp. 183–222. Colvin, John D., Proximate causes of male emigration at puberty in rhesus monkeys The Cayo Santiago macaques: history, behavior and biology., SUNY Press, Albany, 1986. Dunlap, Eloise, Bruce D. Johnson, and Ali Manwar, “A successful female crack dealer: Case study of a deviant career,” Deviant Behavior, 1994, 15 (1), 1–25. Durrant, Russil and Tony Ward, Evolutionary Criminology: Towards a Comprehensive Explanation of Crime, Academic Press, 2015. Ehrlich, I., “Participation in illegitimate activities: A theoretical and empirical investigation,” Journal of political Economy, 1973, 81 (3), 521. Ellis, Bruce J., Marco Del Giudice, Thomas J. Dishion, Aurelio Jos Figueredo, Peter Gray, Vladas Griskevicius, Patricia H. Hawley, W. Jake Jacobs, Jene James, and Anthony A. Volk, “The evolutionary basis of risky adolescent behavior: Implications for science, policy, and practice.,” Developmental psychology, 2012, 48 (3), 598. Foley, C. Fritz, “Welfare payments and crime,” The review of economics and statistics, 2011, 93 (1), 97–112. Fukuyama, Francis, The origins of political order: from prehuman times to the French Revolution, Macmillan, 2011. Gibbs, John J. and Peggy L. Shelly, “NCCD research review: Life in the fast lane: A retrospective view by commercial thieves,” Journal of Research in Crime and Delinquency, 1982, 19 (2), 299–330. Grogger, J., “Market Wages and Youth Crime,” Journal of Labor Economics, 1998, 16 (4), 756–791.

24

Hammond, Cathy, Dan Linton, Jay Smink, and Sam Drew, “Dropout risk factors and exemplary programs: A technical report.,” National Dropout Prevention Center/Network (NDPC/N), 2007. Handley, LJ Lawson and N. Perrin, “Advances in our understanding of mammalian sexbiased dispersal,” Molecular ecology, 2007, 16 (8), 1559–1578. Harcourt, AH and KJ Stewart, “Gorilla society: conflict, compromise, and cooperation between the sexes,” 2007. Heckman, J. J., “Skill formation and the economics of investing in disadvantaged children,” Science (New York, N.Y.), Jun 30 2006, 312 (5782), 1900–1902. LR: 20071114; GR: R01HD043411/HD/NICHD NIH HHS/United States; JID: 0404511; ppublish. Jimerson, Shane, Byron Egeland, L. Alan Sroufe, and Betty Carlson, “A prospective longitudinal study of high school dropouts examining multiple predictors across development,” Journal of School Psychology, 2000, 38 (6), 525–549. Jr, Albert J. Reiss and Albert Lewis Rhodes, “The distribution of juvenile delinquency in the social class structure,” American Sociological Review, 1961, pp. 720–732. Kearney, Melissa S. and Phillip B. Levine, “Income inequality and early nonmarital childbearing,” Journal of Human Resources, 2014, 49 (1), 1–31. Laub, John H. and Robert J. Sampson, Shared beginnings, divergent lives: delinquent boys to age 70, Cambridge, Massachussetts: Harvard University Press, 2003. and

, Shared beginnings, divergent lives: Delinquent boys to age 70, Cambridge: Harvard

University Press, 2003. Levitt, S. D. and S. A. Venkatesh, “An Economic Analysis of a Drug-Selling Gang’s Finances,” Quarterly Journal of Economics, 2000, 115 (3), 755–789. Lilly, J. Robert, Francis T. Cullen, and Richard A. Ball, Criminological theory: Context and consequences, Sage, 2010.

25

Lochner, L., “Education, Work, And Crime: A Human Capital Approach,” International Economic Review, 2004, 45 (3), 811–843. Lynch, James P. and William Alex Pridemore, “Crime in international perspective,” Crime and public policy, 2011, pp. 5–52. Maughan, Barbara, Stephan Collishaw, and Andrew Pickles, “School achievement and adult qualifications among adoptees: a longitudinal study,” Journal of Child Psychology and Psychiatry, 1998, 39 (5), 669–685. Merry, S. E., “Rethinking gossip and scandal,” Reputation: studies in the voluntary elicitation of good conduct, 1997, 47, 333–334. Mirrlees, James, “Notes on welfare economics, information and uncertainty,” Essays on economic behavior under uncertainty, 1974, pp. 243–261. Moffitt, Terrie E., Stephen Ross, and Adrian Raine, “Crime and biology,” Crime and public policy, 2011, pp. 53–87. Osborne, Martin J. and Ariel Rubinstein, A course in game theory, MIT press, 1994. Osterman, Paul, Getting started: The youth labor market, MIT, 1980. Pattillo, Mary, Black picket fences: Privilege and peril among the black middle class, University of Chicago Press, 2013. Pusey, Anne and Craig Packer, Dispersal and philopatry Primate societies., University of Chicago Press, Raphael, Steven and Rudolf Winter-Ebmer, “Identifying the Effect of Unemployment on Crime,” Journal of Law and Economics, 2001, 44 (1), 259–283. Reuter, P., R. MacCoun, P. Murphy, A. Abrahamse, and B. Simon, Money from Crime: A Study of the Economics of Drug Dealing in Washington, DC, Santa Monica, CA: RAND, 1990.

26

Richerson, Peter J. and Robert Boyd, “Complex societies,” Human Nature, 1999, 10 (3), 253–289. Robinson, J. Gregory, Kirsten K. West, and Arjun Adlakha, “Coverage of the population in census 2000: Results from demographic analysis,” Population research and policy review, 2002, 21 (1-2), 19–38. Sampson, R. J. and J. H. Laub, “LIFE-COURSE DESISTERS? TRAJECTORIES OF CRIME AMONG DELINQUENT BOYS FOLLOWED TO AGE 70*,” Criminology, 2003, 41 (3), 555–592. Savage, I. Richard, “Who counts?,” The American Statistician, 1982, 36 (3a), 195–200. Seielstad, Mark T., Eric Minch, and L. Luca Cavalli-Sforza, “Genetic evidence for a higher female migration rate in humans,” Nature genetics, 1998, 20 (3), 278–280. Shover, Neal, Great pretenders: Pursuits and careers of persistent thieves, Boulder, CO: Westview Press, 1996. Siegel, Jacob S., “Completeness of coverage of the nonwhite population in the 1960 census and current estimates, and some implications,” Social Statistics and the City, 1968, pp. 13–54. Silver, L. B., “Frequency of adoption of children and adolescents with learning disabilities,” Journal of learning disabilities, May 1989, 22 (5), 325–327. LR: 20091111; JID: 0157312; ppublish. Spear, L. P., “The Adolescent Brain and Age-Related Behavioral Manifestations,” Neuroscience & Biobehavioral Reviews, 2000, 24 (4), 417–463. Steinberg, L., “Adolescence. 9th international edition,” 2011. Stoinski, Tara S., Veronica Vecellio, Theogene Ngaboyamahina, Felix Ndagijimana, Stacy Rosenbaum, and Kathryn A. Fawcett, “Proximate factors influencing dispersal decisions in male mountain gorillas, Gorilla beringei beringei,” Animal Behaviour, 2009, 77 (5), 1155–1164.

27

Sullivan, Mercer L., ” Getting Paid”: Youth Crime and Work in the Inner City, Ithaca: Cornell University Press, 1989. Tittle, Charles R., Wayne J. Villemez, and Douglas A. Smith, “The myth of social class and criminality: An empirical assessment of the empirical evidence,” American Sociological Review, 1978, pp. 643–656. Urzua, Sergio, “Racial labor market gaps the role of abilities and schooling choices,” Journal of Human Resources, 2008, 43 (4), 919–971. Waal, Frans De, Chimpanzee politics: Power and sex among apes, JHU Press, 2007. Watts, David P., “15 Causes and consequences of variation in male mountain gorilla life histories and group membership,” Primate males: causes and consequences of variation in group composition, 2000, p. 169. Wilder, Jason A., Sarah B. Kingan, Zahra Mobasher, Maya Metni Pilkington, and Michael F. Hammer, “Global patterns of human mitochondrial DNA and Y-chromosome structure are not influenced by higher migration rates of females versus males,” Nature genetics, 2004, 36 (10), 1122–1125. Williams, Geoffrey Fain, “Property crime: Investigating career patterns and earnings,” Journal of Economic Behavior & Organization, 2015, 119, 124–138. Williams, Jenny and Robin C. Sickles, “An analysis of the crime as work model: Evidence from the 1958 Philadelphia birth cohort study,” Journal of Human Resources, 2002, pp. 479– 509. Wright, Richard T. and Scott Decker, Burglars on the Job, Northeastern University Press Boston, MA, 1994. and Scott H. Decker, Armed robbers in action: Stickups and street culture, UPNE, 2011. Zimring, Franklin E., The great American crime decline, Oxford University Press, 2006.

28

Figure 1. A comparison of the life cycle timing of male gorilla dispersal versus male human property crime. Since both gorillas and humans are both viewed as patrilocal species, the cross species comparison is the most apt. Sources: Human male property crime, author’s calculations from NLSY 1997 self-report, Gorilla male dispersal Stoinski et al. (2009: Table 1, page 1158). Ages for puberty and full size are from Steinberg (2011) (ages 11.5 and 18) and Watts (2000) (ages 8 and 15) respectively.

29