Wealth Classes and Their Importance
After considerable group discussion, each researcher estimated a value for each of the major wealth classes (embodied, material, and relational) in the population she/he studied. These values, which we label α, are defined as the percentage change in a family’s wellbeing associated with a percentage change in the wealth class in question, holding other wealth classes constant at the average level. The α estimates are fractions that sum to unity, reflecting their derivation from the Cobb-Douglas production function (see paper by Gurven et al. in this issue for details). The estimates for our five focal hunter-gatherer populations are provided in . Note that the α values are not statistical estimates based on measurements, but rather judgments of each researcher based on months or years of fieldwork with each respective population.
Estimates of alpha exponents for hunter-gatherer societies in the project sample (see text for further explanation).
The researchers generally agreed that relational wealth is of greatest importance, but beyond that there is no clear consensus (). However, the relative α values for embodied and material wealth roughly correspond to the position of these five societies along the mobility/density continuum, with the most mobile populations (Ache, Hadza) having high values for embodied and very low values for material wealth, the sedentary/high-density populations roughly equal α’s for these two categories of wealth, and the Ju/’hoansi grouping with the sedentary populations despite their low density and relatively high mobility (). Note that Ju/’hoansi place considerable emphasis on ties to particular homelands (n!ore
) to which they have foraging rights based on kinship and residence (Lee 1979
; Wiessner 2002
and above), a form of (communal) material wealth; in contrast, Ache and Hadza practice much more open-access foraging on their lands.
Intergenerational Wealth Transmission
As discussed in the introductory paper in this CA forum, we measure intergenerational wealth transmission with the unit-free coefficient β, which estimates the “elasticity” in wealth for parent-offspring pairs, or the percentage change in wealth in the second generation associated with a one percent change in parental wealth. presents these wealth-transmission measures by wealth type for each of the five focal populations (plus one measure for Tsimane, otherwise treated in the paper on horticultural societies in this forum). Sample sizes vary substantially, tending to be larger for more easily measured variables, and range from 26 parent-offspring pairs to nearly 200. The method of calculating β requires comparable information on parents and offspring, and sometimes sex-specific pairings. For many individuals, censoring due to death prior to study or to emigration from the study area leaves many unpaired individuals.
Wealth transmission and inequality measures for hunter-gatherer populations.
We have eight measures of four types of embodied wealth from five populations: weight, grip strength, reproductive success, and foraging success. In contrast, material wealth is represented by only two measures (shares in whaling-boat harvest, and an index of housing quality), both from Lamalera. Similarly, relational wealth is represented by only two measures, exchange (hxaro) partners among Ju/’hoansi, and food-sharing partners in Lamalera. Given this very uneven representation, it is not reasonable to make statistical comparisons between wealth classes for just the hunter-gatherer cases; instead, see the concluding paper in this forum, where measures from the full set of societies can be compared.
We can, however, offer interpretations of particular wealth measures in individual societies. With regard to intergenerational transmission of hunting skills, the Ache results show no relationship between fathers’ and sons’ age-corrected hunting return rates (). This might seem surprising, since men do inherit the genetic component of their father’s somatic traits (body size, strength, athleticism, intelligence, etc.), which should have some impact on hunting success. However, all Ache boys hunt with many different men, receive little or no formal instruction from their fathers, and often don’t live with biological fathers during teen years. Thus, the lack of a relationship between hunting success of fathers and sons is not a complete surprise. Ache men whom Hill has interviewed suggest that most of their skill came from their own practice during teen years in the forest, and from listening to stories by other men, as well as from observing the hunting process (but not necessarily with father as the model). The similar measure for the Hadza (foraging success) includes gathering as well as hunting, and females as well as males (though mother-daughter pairs are excluded from the calculated β, for reasons explained above); it also is close to zero.
Figure 1 Parent-offspring wealth data for three representative cases: (a) Ache hunting success (β=0.08, embodied wealth); (b) Lamalera boat shares (β=0.12, material wealth); (c) Ju/’hoansi exchange partners (β=0.21, relational wealth). (more ...)
In contrast, among the Tsimane we find a quite high β (0.384) for father-son hunting returns, measured as calories gained per hour spent foraging. Young Tsimane males are likely to go hunting with their fathers (and sometimes other male relatives) up until the point where they can hunt alone, generally in their late teen years. Thus, there is a greater likelihood that a Tsimane hunter had his father as the prime or even only model for hunting skills in his formative years than is the case for the Ache, and fathers who hunt frequently may be more likely to have sons who actively hunt. Dyads come from four interfluvial Tsimane villages where hunting is fairly common due to the low population density and near proximity of remote forest (Gurven, Kaplan, and Gutierrez 2006
); since local animal densities and other conditions will affect hunting return rates, the β for Tsimane hunting returns was calculated with village dummy variables to control for any possibility that we were measuring the effects of correlation in location of father-son pairs rather than transmission of ability or skill.
Neither of the β values for reproductive success rise to high levels. The lack of association between parents’ and childrens’ reproductive success among Lamalerans has several possible explanations. One is that cooperative acquisition of food followed by food-sharing between household tends to reduce differences between households in access to food resources. Another possibility is that the lack of association is due to generational changes in fertility patterns. There is some evidence (Nolin, unpublished) that the age at marriage in Lamalera has increased substantially throughout the 20th
Century. A secular trend towards delayed reproduction and reduced fertility may reduce any association between parents’ and children’s reproductive success. While crewmen, and especially harpooners, exhibit higher reproductive success than non-hunters (Alvard and Gillespie 2004
) any man can participate in the fishery. The role of harpooner is nominally heritable, but in practice can be pursued by any youth who shows promise.
Previous research with the Meriam has shown that being a turtle hunter, and particularly a hunt leader, is associated with greatly elevated age-specific reproductive success (Smith et al. 2003
). This RS differential is not found for Meriam men who have reputations as particularly good fishermen, political leaders, traditional dancers, “ladies men,” or “hard workers.” Among Meriam women, those with reputations as “hardworking” do have somewhat elevated mean RS, but these women are disproportionately married to turtle hunters (Smith et al. 2003
). The estimated transmission coefficient for Meriam parent-child RS is effectively nil (). This agrees with previous findings that pairs of Meriam brothers of roughly the same age where only one is a turtle hunters show the same divergence in RS as the full set of Meriam hunters vs. non-hunters, and that patrilineage does not predict hunting role (Smith et al. 2003
). The conclusion seems to be that, whatever combination of abilities, experience, and motivation lead some Meriam men to succeed in turtle hunting or other RS-correlated attributes does not tend to be passed from parent to offspring. Similarly, differences in female RS in this population do not seem to correlate with parentage. Given the relatively egalitarian nature of Meriam social life, and the emphasis on personal achievement, these results are as might be expected. Although the sample size is small (91 pairs, smallest of the whole set of 12 RS measures in the project), the low β value is also found in most other cases where this wealth type was measured (see other papers in this CA Forum). However, it is intriguing that Hill and Hurtado (1996
:414) found a substantial correlation for RS of sons and their parents (but not for daughters); sex-specific transmission of RS should be explored in future studies.
Our two β measures for body weight show a substantial degree of parent-child transmission (about 0.3 for Hadza, about 0.5 for Ache), the pattern also found in other populations measured in this project. Given the considerable degree of inter-household food-sharing among Ache (Kaplan and Hill 1985
) and Hadza (Sherry and Marlowe 2007
), and the low β for Ache hunting success (<0.1), the parent-child weight relationship in these two populations may be substantially due to genetic inheritance in disease resistance or some other unmeasured variable.
In the category of material wealth, our only two measures come from the same population, the whaling village of Lamalera, Indonesia. Household wealth exhibits moderate association between parents and children. The index of wealth here is based on features of household construction, improvements that generally require materials that must be purchased with cash. However, there is very little opportunity for wage labor in the village (aside from teaching or a few government posts). One way for men (especially young, unmarried men) to acquire money is to take occasional temporary construction jobs in the district capitol. However, another important source of cash may be remittances from other close family members who have found permanent jobs outside the village. Insofar as both parents and children have access to the same source of remittance money (say, from the same son/brother) they may manifest similar levels of household wealth. In this case wealth may be acting as much as a proxy indicator of social capital as of material wealth. In contrast, shares that Lamalerans own in whaling boat (téna) harvests do not show a significant intergenerational elasticity, and the estimated effect size is rather small (). This may be because parental influence is less important than more general support within one’s lineage for securing shares in the lineage’s téna. Other ways of securing boat shares include seeking shares from affinal lineages or through repeated, long-term participation in the crew of another lineage’s boat. Parental influence is likely minimal in either case.
Finally, we have two measures of relational wealth. There is a moderate relationship (β≈0.2) in the size of hxaro
spheres between Ju/’hoansi parents and offspring (). This relationship is dependent on two factors. The first was the size and strength of spheres of hxaro that parents maintained until old age for their children to inherit. Approximately 25% of hxaro partnerships were passed on from parents to children as parents aged or upon their deaths when children took their possessions, gave them to parent’s partners, and asked that the relationship be continued (Wiessner 1986
). The second factor was social competence. Some children were competent to “replace” parents who had an active social sphere; others were not. Few children of parents with narrow hxaro spheres expanded their spheres far beyond those of their parents (i.e., “upward mobility” in hxaro was infrequent).
The highest β value for Lamalera is in food-sharing partnerships. Sharing ties are more common between geographically and genealogically closer households (Nolin 2008
). Because parents and children share a common set of kin they are likely to share many of the same partners, and their total number of partners may vary with the number of close kin in the village. The tendency for children to reside near their parents is mild in Lamalera, but when it does occur parents and children are likely to have sharing ties to the same residentially close neighbors. Even when children establish households further away, they may maintain sharing relationships with neighbors near their natal household. Social relationships such as food-sharing relationships differ from other types of wealth in that possession by the parents is not mutually exclusive of possession by the children. Grown children can have established sharing relationships with the same other households as their parents. In fact, parents may be instrumental in helping their children establish these types of social relationships as they enter early adulthood (Scelza 2009).
Although our primary focus in this project is on intergenerational wealth transmission, we have an interest in the corresponding degree of wealth inequality, and how this might vary by wealth type, production system, and other factors. lists the Gini coefficient for each wealth type in our sample. We use the Gini because of its wide usage, unit-free definition, and intuitive meaning: the coefficient can range from zero (completely equality) to unity (all wealth held by a single household). For comparison, Ginis of monetary income range from about 0.25 in several Scandinavian countries to over 0.6 in some poor countries, with the U.S. at about 0.41 (UNDP 2007). Most Ginis for our eleven wealth types are in the moderate range, few being less than 0.1 or greater than 0.4 (); the lowest Ginis are those for weight (both <0.1), and the highest is for Lamaleran boat shares (0.47). There are too few measures of material and relational wealth to discern a pattern by wealth class (but see the concluding paper in this CA Forum).
To characterize an overall measure of wealth inequality for the sample populations, we weight the average Gini coefficient for each wealth class () by the average α (importance for well-being) of that wealth class (). The resultant estimate equals 0.25, very close to the simple arithmetic average of the Ginis in (0.27).