World Development 176 (2024) 106518 Contents lists available at ScienceDirect World Development journal homepage: www.elsevier.com/locate/worlddev Regular Research Article Changes in adult well-being and economic inequalities: An exploratory observational longitudinal study (2002–2010) of micro-level trends among Tsimane’, a small-scale rural society of Indigenous People in the Bolivian Amazon Ricardo Godoy a, *, Jonathan Bauchet b, Jere R. Behrman c, Tomás Huanca d, William R. Leonard e, Victoria Reyes-García f, Asher Rosinger g, Susan Tanner h, Eduardo A. Undurraga i, Ariela Zycherman j a Heller School for Social Policy and Management, Brandeis University, Waltham, MA 02453, USA b Department of Agricultural Economics and White Lodging-J.W. Marriott, Jr. School of Hospitality and Tourism Management, Purdue University, West Lafayette, IN 47906, USA c Departments of Economics and Sociology, School of Arts and Sciences, University of Pennsylvania, Philadelphia, PA 19104-6297, USA d Centro de Investigación y Desarrollo Socio-Integral (CBIDSI), San Borja, Beni, Bolivia e Department of Anthropology and Program in Global Health Studies, Northwestern University, Evanston, IL 60208, USA f Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain and Institute of Environmental Science and Technology, Universitat Autònoma de Barcelona, 08193, Cerdanyola del Valles, Barcelona, Spain g Department of Biobehavioral Health and Department of Anthropology, Pennsylvania State University, University Park, PA 16802, USA h Department of Anthropology, University of Georgia, Athens, GA 30605, USA i Escuela de Gobierno, Pontificia Universidad Católica de Chile, Santiago 7820436, Chile, and CIFAR Azrieli Global Scholars Program, CIFAR, Toronto, ON M5G 1M1, Canada j Climate Program Office, Oceanic and Atmospheric Research, National Oceanic and Atmospheric Administration, 1315 East-West Highway, Silver Spring, MD 20910, USA A R T I C L E I N F O A B S T R A C T Keywords: Knowing what happens over time to the lifeways of people in contemporary small-scale non-industrial societies Panel of the rural Global South matters because it helps assess changes in the quality of life of underrepresented groups. Hedonic It has been hard to answer the question because longitudinal information is rarely collected in such settings. A Nonindustrial society longitudinal dataset of nine years (2002–2010) from a horticultural-foraging society of Indigenous People in the Quality of life Gini coefficient Bolivian Amazon (Tsimane’) is used for an exploratory analysis of micro-level trends in indicators of well-being Deprivation and economic inequalities. We selected 13 Tsimane’ villages (from ~ 100) that varied in proximity to town and surveyed all households in each village. ~ 240 households were followed yearly to estimate trends of 21 out­ comes (e.g., income, sociality, macronutrients). For each economic outcome, annual and all-years-combined Gini coefficients were estimated for the entire sample across the 13 villages. We show a rise in total asset wealth, a change in asset composition (less traditional wealth, more commercial wealth), higher monetary value of foods eaten, and better-perceived health, but a decline in caloric and protein consumption and no marked gender differences in objective or hedonic measures of well-being. Economic inequalities were non-trivial and showed no marked trend but varied between years; asset inequality varied less than income inequality. We document the value of longitudinal, locally grounded indexes of well-being to obtain a granular view of micro-level changes in well-being and the possible use of inequality in the consumption of calories and macronutrients as a valid proxy for income inequality in rural areas of the Global South with tenuous links to the market economy. * Corresponding author. E-mail addresses: rgodoy@brandeis.edu (R. Godoy), jbauchet@purdue.edu (J. Bauchet), jbehrman@econ.upenn.edu (J.R. Behrman), thamazonia@gmail.com (T. Huanca), w-leonard1@northwestern.edu (W.R. Leonard), Victoria.reyes@uab.cat (V. Reyes-García), arosinger@psu.edu (A. Rosinger), stanner1@uga.edu (S. Tanner), eundurra@uc.cl (E.A. Undurraga), ariela.zycherman@noaa.gov (A. Zycherman). https://doi.org/10.1016/j.worlddev.2023.106518 Accepted 24 December 2023 Available online 4 January 2024 0305-750X/© 2024 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). R. Godoy et al. World Development 176 (2024) 106518 1. Introduction (including ours) lack a control group. Although one could build on Western intellectual contributions to The past two decades have seen a sharp increase in the number of test the hypotheses that greater participation in the market economy or quantitative studies about quality of life of people in small-scale, non- modernization increases sociality (Hume, 1793, p. 121), anomie (Dur­ industrial societies of the rural Global South (hereafter, small-scale so­ kheim, 2005 [orig. 1897]), consumerism (Veblen, 1899, pp. 68-101), cieties), including hunter-gatherers, pastoralists, fishers, and horticul­ the division of labor (Smith, 1884 [orig. 1776], p. 5), or economic turists, because the studies can help reveal what is universal and inequality (Marx, 1920 [orig. 1867], pp. 823–826, 838–848; Vol. 1), idiosyncratic to each society (Henrich, 2022). We use the term well- research results would be questionable for the reasons noted. Faced with being as an umbrella term for quality of life, a good life, or life satis­ these challenges, a prudent way forward is to explore micro-level trends, faction without implying the specialized meaning the term well-being leaving for future researchers the task of identifying the aspects of the has acquired in different academic disciplines (Section 1.2). In this outside world that changed well-being and social exclusion of people in article, we report the results of an exploratory study of micro-level small-scale rural societies of the Global South (Marlier & Atkinson, trends in quality of life in one such society over a nine-year period. 2010). Among empirical researchers of small-scale societies, the interest in change has taken the form of quantitative analysis of how aspects of the 1.2. Well-being: Definitions and measurements outside world, such as globalization, modernization, development, or the market economy, correlate with changes in aspects of human well- The second concern is how to define and measure well-being. being. For example, using cross-sectional data on time allocation, re­ Inherent disagreement exists on these topics because a) well-being as searchers have documented a decline in leisure time between hunter- a stand-alone construct is a recent field of study (Kahneman et al., 1999), gatherer and industrial societies (Bhui et al., 2019; Sahlins, 1972). In b) it carries different meanings across academic disciplines, and c) it another cross-sectional study, researchers found that sociality increased spans many dimensions (Benjamin et al., 2020). We next address five with greater participation in the marketplace (Henrich et al., 2010) related questions about well-being that inform our approach. while others have found that participation in the market economy correlated with a poorer diet, worse health, and greater inequality in [i] Approaches to defining and measuring well-being in anthropol­ material wealth (Donders & Barriocanal, 2020; Mattison et al., 2023). ogy, economics, and psychology With few exceptions dealing with changes in local ecological knowledge (reviewed in Aswani et al., 2018) or health (Shephard & Anthropology. With few exceptions (Fischer, 2014; Mathews & Rode, 1996; Weitz et al., 2014), most quantitative studies have relied on Izquierdo, 2008), anthropology has avoided studying well-being as a cross-sectional information to infer change, and have done so owing to comprehensive construct or acknowledging its subjective side (other the scarcity of panel (longitudinal) datasets from small-scale societies. than emotions), and focused instead on specific aspects of quality of life, Two reasons explain the scarcity. First, small-scale societies are such as diet, sociality, health, and leisure time. Although there is still no numerically small, mobile, and live in hard-to-reach places. National theory or methods in anthropology to study well-being in its totality, statistical agencies ignore them. Even the World Bank’s Living Standards there is growing recognition in anthropology that well-being requires a Measurement Study, a leading international study established in the multidimensional perspective (Page-Reeves, 2022). early 1980 s to collect comparable household socioeconomic data on Economics. Starting in the 1970’s, interest in economics began to shift human welfare from the Global South, has only once (to our knowledge) from equating well-being with GDP or income to including other aspects collected information from a small-scale society, probably due to the besides GDP or income (Oman, 2021, pp. 35-47). Thanks to the work of high costs of collecting such data and the difficulties of applying a Sen (1999), economists began pairing well-being with people’s subjec­ money metric to measure welfare in societies that often lack formal tive value of their functioning (Sen’s term), or what people think most markets or prices (Jerven, 2011; Tucker et al., 2015). Second, most important in life, plus quantifiable measures of material conditions such research on contemporary rural small-scale societies comes from bio­ as income, even if material conditions play a small or no role in peoples’ cultural anthropology, a discipline with a weak tradition of collecting views of the good life. Material conditions count because they provide panel data (Gravlee et al., 2009). Cross-sectional data are adequate to the “general-purpose means” ─ capabilities or freedoms in Sen’s scheme ─ compare end-states, but the analysis of change among living human deployed to meet loftier ends (Robeyns, 2005; Sen, 1999, p. 14). To populations requires using the same protocol to collect information from obtain a fuller grasp of well-being, economists have added non- the same entities over time (Deaton, 1985). monetary aspects such as leisure and mortality (C. Jones & Klenow, Here, we start filling the gap by using a yearly panel dataset 2016) and some have borrowed measures of subjective well-being from (2002–2010) from a highly autarkic society of Indigenous People in the psychology (DeNeve & Sachs, 2020; Stone & Krueger, 2018). Unlike Bolivia Amazon (Tsimane’) to undertake an exploratory analysis of anthropologists who analyze well-being at the level of the individual, micro-level trends in 21 indicators of well-being. The broad aim raises economists often aggregate data and analyze it at the country level. two questions: [1] what do we expect to learn from the analysis? and [2] Psychology. Whereas anthropologists and economists entered the how do we define and measure well-being? study of well-being indirectly through specific aspects like leisure, diet, or income, psychologists entered the field by directly asking people to 1.1. Expectations rate their happiness (see review below). Over the years, psychologists have gone beyond happiness to add satisfaction with various domains in Most past quantitative research on the effects of the outside world on life and other indicators of subjective well-being (Oishi & Westgate, small-scale rural societies has focused on the effect of participation in 2022) while acknowledging that material conditions overlap with, or the market economy on selected outcomes, like sociality, health, or predict, subjective well-being (Diener, Oishi, & Tay, 2018). economic inequality. Results from this line of research are problematic In sum, beginning from separate places, the three disciplines have because individual, household, and community participation in the converged on the need to take a fuller view of human well-being marketplace reflect endogenous choices and the role of other con­ composed of subjective and material aspects. The UNDP’s latest devel­ founders, which means that inferences about causality are not trust­ opment agenda reflects this trend by acknowledging the importance of a worthy. In addition, to assess changes in well-being from exposure to the comprehensive set of indicators to assess progress in quality of life market economy ─ or from larger processes like modernization, glob­ (UNDP, 2023). At present, the UNDP comprises 17 indicators bearing on alization, or development ─ requires a control group that remained at Sen’s capability approach, including such things as monetary poverty, arms’ length from the marketplace; to our knowledge, prior studies food, health, inequality, schooling, natural resources, consumption, and 2 R. Godoy et al. World Development 176 (2024) 106518 climate change. In spirit, though not in detail, our approach mirrors the [v] Where do economic inequalities fit in? growing trend of conceptualizing well-being as an expanding bundle of traits but differs from the UNDP and other approaches in development Economic inequalities are an indirect indicator of well-being because by being location specific. This said, not all development programs stress they tend to predict lower life satisfaction (Cheung, 2018) and worse subjective well-being (Venkataraman, 2013). subjective and objective health among individuals across many soci­ eties, including small-scale rural societies (Jaeggi et al., 2021; Pickett & [ii] Who defines well-being and identifying its determinants Wilkinson, 2015; Undurraga et al., 2016; Undurraga et al., 2010), with some notable caveats (Ngamaba et al., 2018; Reyes-García et al., 2019). Depending on the question posed, researchers from outside the We limit the analysis of inequality to economic outcomes rather than to community, study participants or both play a role in defining well-being. social or health outcomes because inequality of economic outcomes such The definition of researcher dominates if the interest lies in assessing as income or wealth have taken a central place among academics and well-being cross-culturally. Marlier and Atkinson (2010, p. 288) say that policymakers. in such cases the criteria should have an “agreed normative interpreta­ In sum, there is still debate about what is well-being, who defines it, tion” and be clearly and easily “interpretable in an international how to measure it, and whether to use an index or use its parts context.” At first sight, examples of such criteria would include years or separately. schooling or levels of monetary income (e.g., monetary poverty) because more schooling or more monetary income could be taken as signs of 2. Specific aims improved well-being. However, these canonical indicators would have little meaning to people in highly autarkic settings whose human capital We use a yearly panel dataset (2002–2010, inclusive) from a highly takes the form of knowledge of the local environment (rather than years autarkic society of Indigenous People in the Bolivian Amazon (Tsi­ of schooling or academic skills) and who do not need to rely as much on mane’) undergoing increasing exposure to the market economy, glob­ money because they consume what they produce. On the other hand, the alization, Westerners, modernization, development initiatives, and definition of study participants dominates if one wishes to understand governmental programs (hereafter the six processes are referred to quality of life from their perspective. An indicator of well-being mean­ collectively as the outside world) to achieve three specific aims: ingful to local people such as social drinking would lack a clear normative interpretation since too much drinking could lead to alco­ [1] Offer an embracive portrait of well-being or quality of life that holism; more is not necessarily better as it is with human capital. Irre­ includes (a) subjective and objective dimensions of well-being spective of who defines well-being, the choice of predictors typically measured with (b) survey questions and physical measure­ relies on theories from the behavioral or natural sciences and the ments. To identify perceived predictors of subjective well-being, judgment of researchers (Das et al., 2020; Diener et al., 2018) though we asked study participants to indicate what had made them some researchers ask study participants to identify perceived causes of happy or sad (see below). Answers were used to identify the hedonic well-being (Reyes-García et al., 2021). objective measures of well-being that predicted or overlapped with subjective well-being. For instance, food was identified as an [iii] How to measure subjective and objective well-being? evaluative measure of subjective well-being, and through sur­ veys, we measured many aspects of household food consumption. Currently, the most comprehensive method for evaluating subjective For one evaluative dimension, health, we use survey responses well-being relies on survey questions that ask study participants to rate and objective measurements identified by our research team. themselves in three areas: [a] evaluative measures ask people to judge [2] Present a novel approach to measure income inequality in small- their entire life or aspects of their life (e.g., income), [b] experiential or scale rural societies increasingly engaged with the outside world hedonic measures ask people about their positive and negative feelings as an indirect indicator of well-being. Biocultural anthropologists (e.g., happiness, anger) at the time of the interview or in the recent past, have been gaining a better understanding of the material in­ and [c] eudaimonia measures ask people how far they feel their life has equalities and diet of people in small-scale societies. However, meaning or purpose. Answers are rated with a numeric scale (e.g., 0 = little is known about income due to the complexity of defining not satisfied; 10 = very satisfied) (Stone & Krueger, 2018). All this in­ and measuring income in societies without formal prices. We formation provides a comprehensive measure of subjective well-being, argue that inequality in calories and macronutrient (proteins, fat, but the collection of such data in non-Western societies is challenging carbohydrates) intake could be a promising approach of because of translation (Lomas, 2019). Objective measures of well-being measuring objective income inequality because (a) in small-scale like income or wealth come from survey responses, while measures of rural societies, food consumption, expenditures, and income health come from survey responses and direct measurements by re­ correlate highly due to Ernst Engel’s Law that food consumption searchers. Economic outcomes are typically estimated with monetary dominates aggregate household consumption among money-poor values. people (Clements & Si, 2018; Kaus, 2013; Seale & Regmi, 2006; Zimmerman, 1932) and (b) the approach sidesteps the difficulties [iv] Should the indicators of well-being be combined or treated of imputing prices to foods in societies without a price-setting separately? market. Measures of inequality in the calories and macronutri­ ents people eat in a highly autarkic rural society should allow one Indicators of subjective well-being are analyzed separately when to get near the true distribution of unseen, hard-to-measure in­ indicators cannot be collapsed into one variable (Decancq et al., 2009). come of the society and could be added to other objective in­ This type of analysis allows researchers to see how different indicators dicators of well-being. (For brevity, we sometimes use the term change. In contrast, other researchers prefer to combine answers into an macronutrients for calories, proteins, fat, and carbohydrates). To index (Benjamin et al., 2020) if there is evidence that the separate di­ be clear and not oversell our point: There is a large literature on mensions of well-being reflect a latent trait (Winther et al., 2015), which country income inequality and nutritional status of individuals is more likely to happen when indicators have an “agreed normative (reviewed in Alao et al., 2021) and a literature on inequality in interpretation”. By aggregating, a summary index reduces measurement monetary food expenditures (reviewed in Sassi et al. (2022), but a errors (Krueger & Schkade, 2008) and false positives, streamlines the slimmer literature (most of it in India) on formal measurements of description of results, and enhances the use of results in policy (Marlier nutritional inequalities per se, such as estimates of Gini co­ & Atkinson, 2010; Alkire et al., 2022; UNDP 2021). efficients of energy or macronutrient intake (Bell et al., 2021; 3 R. Godoy et al. World Development 176 (2024) 106518 Kharkwal & Malhotra, 2020; Liu et al., 2023; Singh, Kumar, & health services increased. Vaccination campaigns in villages became Singh, 2016), and none of these studies underscore the role of routine, and the government, non-government institutions, and foreign energy and macronutrient inequality as proxies for hard-to- institutions began to supply other health services to Tsimane’ besides measure income inequality in rural settings of the Global South. vaccinations. Government officials visited Tsimane’ villages to issue [3] Make the clean and coarse datasets and documentation of the birth certificates and national identification cards, which Bolivian citi­ panel and ancillary studies freely available to the public to re­ zens need to access governmental services. Around 2001, the Bolivian dress the glaring gap in longitudinal datasets from small-scale government began to build schools in Tsimane’ villages that previously societies of the Global South (Appendix A). had no schools; the intervention increased the tendency to adopt a sedentary lifestyle. Representatives of political parties and local officials 3. Tsimane’ Amazonian panel study (TAPS) and the setting visited villages to garner support for the municipal and national elec­ tions of 2002, 2005, and 2009. Road infrastructure improved, and solar The culture, human biology, health, psychology, and economy of panels were placed in village schools to provide light and run electronic Tsimane’ have been analyzed by researchers from the Tsimane’ Health teaching equipment. Toward the end of the study, the national electric and Life History Project (Gurven et al., 2017) and by researchers from grid reached villages near towns while telephone towers were erected in the Tsimane’ Amazonian Panel Study (TAPS) (Leonard et al., 2015; parts of the Tsimane’ homeland so people could use cellular telephones Mulligan et al., 2022). This study centered on Tsimane’ living along the in villages within the covering radius of the towers. During the last year river Maniqui, department of Beni, an area we call the homeland. Prior of the study (2010), 3 % of adults in the sample owned telephones and 3 studies with all nine years of TAPS data have focused on a few outcomes % owned televisions. Real (inflation-adjusted) fares to towns fell yearly and show (a) increased consumption of cooking oil and pasta, adult by 4 % during 2002–2010, making it easier for Tsimane’ to travel to body mass index (BMI; body weight in kg/standing height in m2), and town and for merchants to stop by villages to trade.1 During the study, prevalence of overweight and obesity (Bethancourt et al., 2019), (b) less adult Tsimane’ men increased the number of days working for loggers, convivial traditional drinking (Rosinger & Bethancourt, 2019), and (c) cattle ranchers, and Andean highlanders who had encroached on the catch-up growth among stunted children (Undurraga et al., 2018; Zhang homeland of Tsimane’, often dwelling in settlements next to Tsimane’ et al., 2016). villages. Real cash earnings from these jobs rose by 8 % per year while We depart from these studies by putting well-being at the center rice and maize sales fell, each by ~ 2.5 %/year. The frequency of pur­ stage of the analysis and by examining time trends of many indicators at chases also rose; every year Tsimane’ adults bought 3 % more durable the same time, particularly foundational indicators of objective well- assets and were 2.4 percentage points more likely to buy apparel. being like income, macronutrients, and material inequalities that so Starting in 2004, the chances an adult would not spend cash the past far have received no attention in longitudinal studies of small-scale so­ fortnight fell by 5.8 percentage points per year. In the past three de­ cieties. Furthermore, we go beyond earlier studies using TAPS data by cades, 10 % of Tsimane’ emigrated from the department of Beni, some to providing a panoramic instead of a narrow view of well-being limited to settle in cities and rural towns, some to settle in distant villages with a single indicator, like convivial drinking or food consumption. And more wildlife in other departments. what makes the endeavor important is how it relates to the polarized The sketch highlights two points relevant to this article. It provides views about changes in the quality of life of people in highly autarkic evidence of the many concurrent events engulfing the sample of Tsi­ small-scale rural societies from increasing dealings with the outside mane’ studied and underscores the difficulties of isolating the effect of world. As we show elsewhere (Bauchet et al., 2020), some observers markets, outside institutions, migrations, environmental degradation ─ from the Enlightenment to the present have seen greater anomie, as shown by Tsimane’ moving farther away to have access to more consumerism, inequality, and squalor while others have seen material wildlife ─ or development on well-being in our study. prosperity and greater life satisfaction. Where is the nuanced middle ground? 4. Approach to define and measure well-being At the time of our research, Tsimane’ numbered 17,000 people (INE, 2015) and lived in over 100 villages, mainly in the department of Beni. Building on the previous discussion of well-being or quality of life, Almost nothing is known about Tsimane’ before the twentieth century we follow a pragmatic, eclectic approach to define it. Our approach because they lived in or retreated into the backlands of the Bolivian relies on subjective and objective aspects (predictors) identified by study Amazon beyond the reach of Westerners. At the turn of the twentieth participants and our research team. These dimensions were assessed by century, two foreigners ─ Pauly, an Argentinian engineer and explorer, answers to survey questions and physical measurements by our research and Nordenskiöld, a Swedish anthropologist ─ visited Tsimane’ along team. We analyze indicators separately but test whether using one or the river Maniqui and penned what they saw. In the ethnographic sketch more summary indicators is justified. of the outcomes (Appendix B), we sometimes turn to their works to Before discussing specifics, we note one criterion used to select highlight what has changed over the past century. During the mid- outcomes as it limits what we can say about changes in the main anal­ 1950’s, Protestant missionaries from the United States settled perma­ ysis. The outcomes had to have been measured every year from 2002 nently along the river Maniqui to proselytize Tsimane’ and to set up the until 2010 because this made micro-level yearly trends comparable. first rural schools for them (Kempf & Kempf, 2017). Tsimane’ began to Nevertheless, we bring in information collected occasionally or from settle in villages with schools, a shift from their historical semi-nomadic ancillary studies to complement the main analysis (sections 9, 5.1). way of life. Notwithstanding the shift toward more permanent resi­ dence, Tsimane’ at the time of our research were one of the world’s most economically self-sufficient, small-scale, nonindustrial, contemporary rural societies. A comparative study of 15 such societies ranked Tsi­ 1 mane’ next to last in contact with the market economy, with an average During the 2007 survey, adults had travelled an average and median of 12 and 6 times to the regional towns in the past year; by 2010, mean and median of 7 % of all household calories bought in the market (Henrich et al., values for the same recall period and destination had risen to 19 and 12. Survey 2010). answers about travel frequency collected before 2007 cannot be used because During the study, the influence of the outside world on the homeland they were coded very different and could not be meaningfully recoded to of Tsimane’ deepened. The Bolivian government established national compare frequencies. Before 2007, we asked about frequency of travel to only programs to transfer cash for old-age pensions (2002) and to promote one town (San Borja) and coded answers as “Never”, “Once a year”, or “Once a school attendance (2006) and preventive health care among pregnant month”. Starting in 2007, we asked about the number of times the respondent women and children under two years of age (2009). Delivery of public had travel to any town. 4 R. Godoy et al. World Development 176 (2024) 106518 4.1. Measures of subjective well-being points to perceived morbidity and sociality as crucial components of subjective well-being (Das et al., 2020; Krueger & Stone, 2014; Lee et al., During the first two years of the study (2002–2004), we asked adults 2017; Waldinger & Schultz, 2023). In Appendix B we discuss the (age ≥ 16y) an open-ended question about all the events that had made importance of chicha as a social glue. them happy the past week, and, separately, about all the events that had made them sad in the past week. Respondents could list as many events 4.2. Measures of objective well-being as they wished. Appendix C contains a verbatim summary of what made respondents happy or sad. The initial questions about the causes of Under this rubric, we include objective measures of foods eaten happiness and sadness had strengths and shortcomings. (expressed in monetary value, macronutrients, and calories), privately Strengths. First, the Tsimane’ language has words for happiness and owned physical asset wealth, monetary income, aspects of horticulture sadness, so the use of these words was easily understood by study par­ that capture the amount of food available to the household to eat, barter, ticipants (Godoy et al., 2010), especially when limiting the recall to the sell, or to make chicha, and body-mass index (BMI). past week. Second, the use of an open-ended question posed to each villager (who were often by themselves without bystanders listening) 4.3. Final selection criteria made it easier for respondents to express their true feelings. A more public participatory approach through focus groups could have been Based on our ethnographic understanding of Tsimane’ and the data used but posed its own challenges (Sim & Waterfield, 2019), such as available to estimate yearly trends for all nine years of the panel, we confidentiality and peer pressure. Third, the happiness-sadness di­ include outcomes that met one or two additional criteria. (i) They re­ chotomy resembles the utilitarian pleasure-pain continuum but is a flected an indicator of well-being identified by Tsimane’. For example, convenient albeit blunt entry point into what people value or eschew. As Tsimane’ identified poor health as a reason for sadness, and to capture such, answers could be read as measures of evaluative well-being or health, we measured body-mass index and asked about the number of their predictors. For instance, if food emerges as a salient reason for days respondents, or their dependents had been bedridden the past sadness or happiness, satisfaction with food must be a valid evaluative fortnight. Several aspects of horticultural production were included proxy of subjective well-being, and the measure of foods eaten must be a because Tsimane’ frequently mentioned that the quality of their fields blunt but reasonable correlate of the evaluative measure. This line of (and by implication of crop production and food) made them happy or thinking is in line with Dressler’s (2020) concept of cultural consonance, sad. (ii) The outcomes helped Tsimane’ attain or deal with some of the which has been examined among Tsimane’ (Reyes-García et al., 2010), reasons that made them happy or sad. Cash income and asset wealth are but not with food. Fourth, we found that despite their autarky, Tsimane’ examples. Almost no respondent mentioned cash income as a reason for valued many of the same things valued by people in the rural Global either sadness or happiness, yet Tsimane’ need money to pay for things South (Alkire, 2002, p. 190). Like other people in the rural Global South, they care about, such as food they have learned to value but cannot Tsimane’ cherished social relations, food, shelter, and some material produce (e.g., pasta) and health services (even from local curers). Same prosperity but did not mention the environment, knowledge, supernat­ with asset wealth ─ almost no respondent mentioned an asset surplus or ural beliefs, or religion, all important to Tsimane’ (Huanca, 2006). deficit as a cause of mirth or dejection, but they value dugout canoes, Shortcomings. We did not measure all relevant variables. [1] The livestock, and metal tools; these they use to travel, farm, fish, and hunt. initial approach should have been followed by the same questions every Most assets provide both food (e.g., eggs from poultry) and lump sums of year about the degree to which respondents were satisfied with their cash for emergencies (e.g., from the sale of livestock or rifles). food, shelter, social relation, and all the other principal reasons for their happiness or sadness; the improved approach would have allowed us to 5. Data collection, preparation, and analyses estimate directly changes in evaluative well-being and spot new ex­ pressions of well-being. Although we asked about evaluative measures Before collecting information, we obtained IRB approval from of food, clothing, health, and sociality, and about hedonic measures of Northwestern University (732–002, 732–007, 732–016, STU000007), well-being in the form of experiences with positive (happiness) and the Tsimane’ Council, and study participants. negative emotions (anger, fear, sadness) in the past week, we did not do so every year, so we exclude these measures from the main analysis. [2] 5.1. Data collection Asking people about what made them happy or sad does not get at views about the provision of public services that contribute to well-being. Every dry season (May-September) during 2002–2010 a team of Village political leaders, the municipal government, and the governing Bolivian and international researchers and Tsimane’ assistants gathered body for most Tsimane’ villages (Tsimane’ Council) play a role in information on 21 outcomes among all households of 13 Tsimane’ vil­ defending territorial rights against encroachers, mismanaging natural lages along the river Maniqui (Table 1). We trained Tsimane’ assistants resources by logging firms, crime, and the quality of schools and public in data collection and collaborated with them to ensure they translated health services. [3] Last, we did not ask about eudaemonic measures of the questions accurately when the respondent could not understand well-being. Spanish. Gathering data during the same time of the year erased the Given the open responses and limited by these shortcomings, for the effects of seasonality on survey responses. We selected villages because main analysis we are left with two indirect measures of evaluative well- they varied in proximity to the principal urban hub of the area (the town being: number of reported bed-ridden days the past fortnight and fre­ of San Borja). Villages in the study were one hour to one day away by car quency of convivial drinking of a home-brewed beverage (chicha) that or outboard motor from San Borja and had a median of 19 households Tsimane’ use to display sociality. These are measures of well-being (standard deviation, SD = 6). because Tsimane’ mentioned that convivial drinking of chicha made The selection of villages raises the question of how representative our them happy, while poor health made them sad. Drinking chicha was the sample was of the Tsimane’ population. To answer the question, we turn fourth most important reason for happiness after events like a good fish to a randomized-controlled trial (RCT) done in 2008–2009 in 40 villages catch or being in the company of family, while poor health was the beyond the 13 villages of the panel study. We restrict the comparison to leading reason for sadness (Appendix C). The measures we obtained are data from the baseline survey of the RCT (2008) and the 2008 yearly indirect because they do not capture how satisfied respondents were survey of the panel study because the trial changed outcomes by the end- with their health or with their chicha consumption; instead, they mea­ line survey (2009). We did not measure all the outcomes of the panel sure the frequency of the events. Nevertheless, both measures capture study in the RCT because the aims of the two studies differed, but some canonical aspects of well-being. Indeed, recent international research outcomes were measured in the same way, and those are the ones 5 R. Godoy et al. World Development 176 (2024) 106518 Table 1 other outcomes come from self-reports of adults about themselves, their chil­ Definition and measurement of variables recorded annually among Tsimane’ dren, or their households. Parents provided answers about bedridden days of households used for the main analysis /1/, 2002 (baseline)-2010. children. For data on variables collected at the level of the household, we relied on one household head for the answers. Men typically answered questions about Variable Definition and measurement horticulture and asset wealth and women answered questions about food con­ Horticultural Current conditions of household farmlands sumption. The methods to measure BMI are described in Rosinger et al. (2013). production: (5) BMI. Analysis of BMI excludes pregnant women and is done separately for Forest Hectares (ha) of old-growth and fallow forest cleared for women and men. horticulture in past year Plots Number of forest plots cleared for horticulture in past year Plantains ha of farmlands under plantain cultivation discussed next. Manioc ha of farmlands under manioc cultivation The analysis shows similarities and differences. Demography. Villages Economic flows: Monetary value of inflows to the household in past two weeks in the panel study were larger (19 household/village) than villages in /2/ Income Monetary income from wage earnings and sale of local the RCT (14 households/village), but total household size (~6 people) goods and the percentage of nuclear households were similar in the two studies Barter Monetary value of goods received in barter (~75 %). 23 % of the sample of women 14y ≤ age ≤ 50y in the panel Autarky Binary variable if an adult had not earned money or bartered study reported being pregnant in 2008 compared with 19 % in the RCT. (yes = 1; no = 0) /3/ Economic stocks, Monetary value of 22 physical assets owned by the household at Women in the panel study seemed more vulnerable to environmental wealth: the time of the survey /2/ stresses for they were 6 % more likely than women in the RCT study to Total Total value of traditional, commercial, and livestock assets give birth during the dry season. Horticulture. Here no differences Traditional Goods made by Tsimane’ women or men from local plants: appear. Households in both samples cleared the same number of plots canoe, mortar, cotton bag, wooden quern, set of bows and (~1) from the forest and the same area of forest (1 ha) and put the same arrows Commercial Commercial goods from the market: axe, bicycle, cooking area under manioc cultivation (0.1 ha), the principal ingredient to make pot, cutlass, hook (fishing), knife, mill (hand), mosquito bed chicha. Monetary earnings. Although adults in both samples worked for net, fishing net, radio, rifle, shotgun, watch the same number of employers (1) in the past week and sold many of the Livestock Domesticated animals raised at home: cow (or bull), duck, same farm and forest produce, they differed in the number of days hen (or chicken), pig Sociality: Episodes of sociality among adults in past week /3/ worked, earnings, and what they sold. Adults in the panel study worked Gifts Mean number of gifts given by adults in a household mostly for cattle ranchers, whereas adults in the RCT study worked for a Chicha Mean number of days adults in a household drank chicha mix of smallholders, traders, and loggers. Households in the panel study Food: Monetary value and nutritional composition of 21 own and sold mainly plantains and thatch palm from the forest, whereas house­ market foods consumed by the household in the past week holds in the RCT study sold mainly rice. In the past 1–2 weeks, adults in Value Own: manioc, rice, maize, plantain, wildlife (birds, fish, game [mainly mammals]), eggs, chicken, duck, pork. the longitudinal study worked three times more days and earned three Market: beef, cow head, jerky, canned sardines, oil, lard, times more money than their peers in the RCT study. Villages in the white granulated sugar, flour, pasta, bread. /2/ panel study paid higher real wages. Shocks. Illness was the most common Dietary The number of kilocalories of energy and grams of protein, adverse shock in both studies, but people in the RCT study were five Consumption: fat, and carbohydrates of each of the 21 foods consumed was Energy obtained by using the conversions and methods in Flores percentage points more likely to have experienced any shock the past Protein et al. (1971) and Bethancourt et al. (2019). Macronutrients year and experienced 3.3 % more shocks than study participants in the Fat = protein, fat, and carbohydrates, but in the text, we RCT. In short, the evidence suggests that Tsimane’ in the longitudinal Carbohydrates sometimes use the term macronutrients to cover energy as study had a stronger toehold in the market economy but were like other well Tsimane’ in other areas. Health: Reported health in the past two weeks for all people in the household and body-mass index (BMI) of adults at the time of A household in the panel study had a mean and median of six people the survey (SD = 2.7), evenly split between adults (defined as people ≥ 16 years of Bedridden Number of days the person was bedridden in the past two age) and children or people < 16 years of age. The sample during the weeks baseline year (2002) had ~ 240 households (Appendix D). Other than BMI Measure of body-mass index of adults at the time of the survey /3–5/ the measurement of BMI, the measurement of other outcomes came from answers to survey questions. We followed the protocol of Lohman et al. Table 1. Definition and measurement of variables recorded annually among (1988) to measure BMI. A portable Seca stadiometer was used to mea­ Tsimane’ households used for the main analysis /1/, 2002 (baseline)-2010. - sure standing height to the nearest 0.1 cm and a Tanita BF-522 W scale continued. was used to measure body weight to the nearest 0.1 kg. Notes: We used three surveys to gather information about individuals, (1) Unless indicated otherwise, the value of all variables refers to mean amounts per person in the household. Amounts per person came from dividing the sum of households, and communities (villages and towns). (1) Surveys about the value of the outcome in the household by the number of people living in the individuals centered on monetary earnings (income), barter, sociality, household at the time of the yearly survey. and perceived health. Questions about monetary earnings, barter, and (2) The monetary values in all tables are in bolivianos. Nominal monetary values sociality were addressed to all household members ≥ 16 years old or were converted to inflation-adjusted (real) values using Bolivia’s Consumer younger if they headed a household. Questions about self-perceived Price Index, retrieved from the World Bank on May 1, 2022: https://data.worldb health were addressed to all adults; the female household head was ank.org/indicator/FP.CPI.TOTL?locations = BO. Wealth was measured at the asked to judge the health of dependents under 16 years of age. (2) The individual level among all adults in the household but, in addition, a household- survey about household traits had questions to assess area deforested, level measure was estimated by adding the wealth of individuals, plus the wealth areas and crops under cultivation, asset wealth, and food consumption. of children, who often own poultry. For the main analysis we use household- Questions about food consumption were addressed to the female level measures of wealth, but for the analysis of gender differentials (Table 3) we use individual measures of wealth. We used our prior ethnographic obser­ household head and questions about horticulture were addressed to the vations during 1999–2001 to select the 22 assets that we measured. We tried to male household head. (3) Questions about villages and towns dealt with make sure the assets measured covered goods owned by better-off and less well- prices, which we used to estimate the monetary value of foods eaten and off households, by women and by men. asset wealth. We addressed questions about village prices to one or more (3) Adults were people age ≥ 16 years old at the time of the survey, or younger if village leaders and questions about town prices to shopkeepers in two they headed a household. towns of the study area (San Borja and Yucumo). (4) Except for BMI (body-mass index; weight [kg]/standing height[m2]) all Three recall periods were used: (a) one week for questions about 6 R. Godoy et al. World Development 176 (2024) 106518 foods eaten and sociality, (b) two weeks for questions about monetary Second, we describe the 21 outcomes at baseline. Third, we use bivariate earnings, barter, and perceived health, and (c) the past year for ques­ regressions with survey year as a predictor to explore micro-level time tions about forest clearance. Questions about asset wealth and areas trends of outcomes, aware we are not controlling for confounders that under cultivation with plantain or manioc referred to conditions at the would happen with any nested observational dataset. With many out­ time of the survey. comes analyzed simultaneously, the likelihood of finding false positives rises. To protect against this mistake, we report levels of statistical sig­ 5.2. Data preparation nificance of micro-level time trends after adjusting for false discovery rates (Jones et al., 2019) (Appendix E and Table 2, columns 2 and 4). We transformed data in several ways. Robustness analysis. We did three robustness checks. (1) We controlled for fixed traits of villages to remove some of the noise from (1) Prices from villages and towns were imputed to the quantity of these types of confounders (e.g., village-to-town distance). In addition, assets owned or foods eaten. We first relied on current village we took two steps to assess if the analysis of inequality was sensitive to selling or buying prices. If these prices were missing, we imputed the measurement of inequality. (2) Besides estimating yearly Gini co­ the median price from the sample of villages that had the price, efficients for each economic outcome for all households, we computed a and if the good had not been bought or sold in any village for the grand Gini for each outcome using all households and years combined past three months, we imputed the current price of the good in and did so as a check for the study’s short duration and modest yearly towns to the commodity without a price. sample size of households (~240). (3) We used the coefficient of vari­ (2) To estimate time trends of monetary values, we converted current ation instead of the Gini to judge if results changed by the method to nominal monetary values to inflation-adjusted values using compute economic inequality. Bolivia’s Consumer Price Index (Table 1). To convert the national Extensions. Gender. The main analysis and the robustness analysis currency (bolivianos) into United States dollars (USD), we used gloss over gender disparities in the household, an important aspect of the unofficial foreign exchange rate currency in the two towns of disaggregation in the analysis of well-being (Marlier & Atkinson, 2010). the study. For transparency, we report monetary values in USD To address the shortcoming, we unbundle some household-level data and bolivianos when discussing results. During the study, the and compare micro-level trends of indicators that were measured at the average yearly foreign currency exchange rate was 7.47 bolivia­ individual level among adult women and men but that we aggregated to nos/USD (SD = 0.42). the household level in the main analysis. These indicators included the (3) Excluding BMI and sociality, when the question or physical monetary value of goods obtained in barter, monetary earnings, the measurement referred to an adult, we added the amount reported monetary value of privately owned material assets, sociality (gifts given by each adult in the household and divided the sum by the total and convivial drinking of chicha), bedridden days, BMI, and body-fat number of household members at the time of the survey, which measured with the sum of four skinfolds (triceps, biceps, subscapular, yields an average value per person in the household. BMI values suprailiac), the latter an indicator of energy availability stored in the refer to the average BMI of adult (non-pregnant) women and human body. Hedonic measures. We analyze trends in four emotions adult men in the household at the time of the survey. Sociality (fear, sadness, anger, happiness) that we omitted in the main analysis indicators refer to the average for adults in a household. We did because they were not measured every year. not use adult equivalents to estimate household size because many adult Tsimane’ do not know their age (Gurven et al., 2007). 6. Quality checks and limitations (4) To explore micro-level trends of outcomes which had some zero values, reduce the influence of outliers, and ease the interpreta­ Quality checks. We tried to reduce measurement errors by using the tion of results, we changed the value of household averages with same questions and field staff every year, but errors nonetheless arose. the inverse hyperbolic sine transformation. Some measurements had random or systematic mistakes. There was (5) We do the main analysis using the household as the unit of rounding or digit heaping around multiples of five in the last digit of analysis for three reasons. (a) Two dimensions of well-being whole numbers in answers to questions about cash earnings and the last (horticulture and food consumption) were measured at the digit of whole numbers in answers to questions about the cash value of household level. (b) Tsimane’ subsistence puts the household at goods obtained in barter. There were also biases from forward tele­ the center stage of most daily horticultural activities. (c) The scoping and omissions. When asking about barter, bedridden days, and computation of Gini coefficients is typically done with the monetary income, we first asked for the amounts the week before the household as the unit of observation owing to data availability. interview and, subsequently, for the amounts two weeks before the Therefore, cross-cultural comparisons of economic inequalities interview. Reported values for the past week were higher than reported are facilitated by analyzing inequality at the household level values for two weeks ago. For example, respondents were 13–15 per­ (Hruschka et al., 2017; Killewald et al., 2017). centage points more likely to say they had sold goods the past week than (6) We used the Gini coefficient to measure economic inequalities. As two weeks ago. Reported answers about the area of forest cleared for has long been recognized (Atkinson, 1970), there are several horticulture coincided with field measurements by our research team methods to estimate economic inequality, each with advantages (Vadez et al., 2003). We excluded extreme height, body weight, and and disadvantages (Bowles & Carlin, 2020; Woolard-Mayfour & skinfold values for all ages, following thresholds recommended by WHO Hruschka, 2022). Nevertheless, we used the Gini coefficient (1995). The sample of households remained relatively stable during the because it facilitates cross-cultural comparisons as it has become study, but adults entered and left the panel. Of the 633 adults present in the indicator of choice in many fields. It can range from zero the baseline year (2002), 4.42 % left permanently after 2002 and 38.23 (complete equality) to one (complete inequality); the latter % were surveyed every year until 2010; the rest (57.35 %) were sur­ would happen if, for instance, one household owned all the assets veyed a mean and median of six times during the remaining seven years of the group. (2003–2009). Limitations. (1) In a panel study with branching questions, re­ 5.3. Analyses spondents might have learned that saying “no” to initial forking ques­ tions saved them time from having to provide details about the event. Main analysis. The main analysis proceeds in three steps. First, we For instance, in answer to the question, “How many days were you assess the correlations between outcomes to assess if they reflect latent bedridden the past week?”, saying “none” would have saves time among variables and whether the creation of one or more indexes is warranted. seasoned respondents from having to answer follow-up questions about 7 R. Godoy et al. World Development 176 (2024) 106518 Table 2 Micro-level trends in household levels and economic inequality of well-being indicators: 2002 (baseline)-2010. Outcomes [A] Levels – households [B] Inequality (Gini) Baseline %Δ/year. Village fixed effect: Baseline %Δ/year No Yes (1)/a/ (2A)/b/ (2B)/b/ (3)/c/ (4)/d/ Horticulture: Forest 0.179 0.40 0.40 0.459 − 0.5 Plots 0.298 − 0.60 − 0.60 0.359 0.5 Plantain 0.112 − 0.80*** − 0.80*** 0.512 1.7 Manioc 0.021 0.01 0.10 0.735 1.6 Economic flows: Income 4.650 2.30 2.60 0.670 0.4 Barter 0.337 − 0.40 − 3.00 0.717 1.3 Autarky 0.061 1.00* 1.00 Not applicable Economic stocks, wealth: Total 886 4.20*** 4.30** 0.412 1.1 Traditional 228 − 5.2** − 4.80* 0.419 1.5 Commercial 503 7.60*** 7.70*** 0.452 0.2 Livestock 155 − 6.30** 6.80 0.693 2.1 Sociality: Gifts 0.085 − 0.01 − 0.10 Not applicable Chicha 0.060 − 0.10 − 0.10 Food value: Value 6.860 7.40*** 7.40*** 0.476 − 1.2 Food consumption: Calories 3025 − 1.40* − 1.4** 0.304 − 1.6 Proteins 91 − 2.30* − 2.30** 0.373 − 1.7 Fat 49 0.10 0.10 0.477 − 1.1 Carbohydrates 545 − 1.20 − 1.20* 0.304 − 2.0 Health: Bedridden 0.131 − 0.80*** − 0.80*** Not applicable BMI-women 23.143 0.40** 0.50 BMI-men 23.364 0.20 0.20 Table 2. Micro-level trends in household levels and economic inequality of well-being indicators: 2002 (baseline)-2010. – continued. Notes: Table 1 has definition of variables and how we measured them. For Autarky, the number in column 1 represents the share of adults who were autarkic, and the number in column 2 represents the percentage-point change per year in the likelihood of being autarkic. *, **, and *** significant ≤ 10 %, ≤5%, and 1 %≤. Values have been adjusted for false discovery rates. (a) Values come from column “Mean” of Appendix D. All values refer to household means, except for macronutrients; those values refer to median amounts per household during the baseline year. (b) Values come from Appendix E. Other than Autarky, values capture the percentage change per year. Since Autarky took the value of one or zero, the value is read as a percentage-point change for each additional year. (c) Values come from the year 2002 in Appendix F, Part 1. (d) Values come from column “%Δ/year” in Appendix F, Part 1. Regression coefficients, standard errors, and p values for this column come from Appendix F, Part 3. P values have been adjusted for false discoveries. ailments, symptoms, and cures. Some of the improvements in perceived 7.1. Baseline (2002) conditions – Levels of socioeconomic and health morbidity discussed later could reflect panel conditioning. (2) The brief outcomes duration of this exploratory study does not allow us to estimate non- linear trends confidently. Deforestation. Because Tsimane’ are principally subsistence horti­ culturists, the yearly area of forest a household clears approximates the 7. Main analysis: Descriptive results of baseline area cultivated and is therefore a valid proxy for aggregate yearly agricultural sales and unseen income from consuming what households [7.1] Testing for latent variables. grow. In 2002, 93 % of households had cleared forest for horticulture. Households did not scatter plots to protect crops against localized losses To test for latent variables, we computed Chronbach’s alpha to assess from pests, predators, diseases, or floods. In 2002, a household cleared whether all the socioeconomic and health outcomes or a subset of var­ an average of 1.5 plots and a median of one forest plot (SD = 0.86). The iables captured latent variables (Appendix G). Alphas ranged from a total forest area cleared by a household amounted to 0.9 ha composed of high of 0.64–0.65 for horticultural variables or for all socioeconomic fallow forest plus old-growth forest (SD = 0.8 ha). Adjusted by house­ and health variables (0.63), down to 0.18 (sociality) or 0.15 (health). hold size, deforested areas in 2002 averaged 0.17 ha per person The results fall below the modest reliability threshold of 0.70 recom­ (Table 2). mended in some of the literature (e.g., Nunnally & Bernstein, 1994; Monetary income. Tsimane’ had low monetary income. Table 2 Peterson, 1994) and reflect the reality that most of the items measured suggests that daily cash income per person measured with monetary distinct aspects of well-being (Lance, Butts, & Michels, 2006). However, earnings from wage labor and from sale of farm and forest goods was the four macronutrients reflected a reliable latent variable (alpha = USD 0.65 (4.65 bolivianos), below the World Bank’s daily international 0.90). For these reasons, we provide a granular analysis of change in poverty line for Bolivia of USD 1.9 per person (World Bank, 2022a). To well-being, but occasionally bring in the principal component factor put the figure in context: In 2002, an employer spent 5.8 bolivianos in a index to describe changes in the total socioeconomic and health in­ mid-day meal for a Tsimane’ hired to prepare pasturelands or to cut and dicators or the four macronutrients consumed. transport logs from the forest. Thus, with 4.65 bolivianos a Tsimane’ could have bought 80 % of a lunch meal. The average reported daily 8 R. Godoy et al. World Development 176 (2024) 106518 monetary value of goods received in barter per person was also low, USD groups, we found that 2 % of adult Tsimane’ were underweight (BMI < 0.04 (0.33 bolivianos). During 2002, 67 % of adults (n = 611) had not 18.5), 75 % had a healthy BMI (range: 18.5–24.99), 20 % were over­ bartered during the two weeks before the interview, and 6.1 % of weight (range: 25–29.99), and 3 % had obesity (BMI ≥ 30). households were autarkic, meaning no adult in the household had bar­ tered or earned cash the past fortnight. 7.2. Baseline conditions (2002) - Gini coefficients of economic Asset wealth. Tsimane’ households had modest amounts of privately inequalities at baseline and mean yearly Gini coefficients of economic owned assets; most of these assets were not expensive. Thirty percent of inequalities during 2002–2010 adults reported they did not own footwear, 42 % of women did not own hens, and a quarter of adults did not own fishhooks. When they owned The histograms in the two parts of Figure 1 summarize the Gini co­ an asset, they owned one, as they did in Nordenskiöld’s days a century efficients of economic inequalities at baseline (Figure 1A and Table 2 ago (Appendix B). In 2002, the average monetary worth of wealth in [column B3]) and the average yearly Gini coefficients across the nine privately owned physical assets reached 886 bolivianos per person, years of the study (Figure 1B). The histograms resemble each other, equivalent to USD 126. To ground this figure, consider the following: If a though some outcomes moved a bin or two from Figure 1A to Figure 1B. Tsimane’ were to sell all their assets worth 886 bolivianos they could buy We focus on Figure 1B in the discussion that follows because it comes enough food to eat for 129 days (886 bolivianos/6.86 bolivianos in self- from a larger sample of observations while reflecting the ranking of reported per capita daily value of food consumption). Fifty-seven economic inequalities at baseline. In addition, the mean yearly Gini percent of asset wealth was stored in commercial goods (503 bolivia­ coefficients of Figure 1B resemble the grand Gini (Appendix F, Part 1, nos), followed by wealth in goods crafted from local plants (26 %; 228 penultimate column), so we are safer characterizing economic in­ bolivianos), and livestock (17 %; 155 bolivianos). These estimates do not equalities among Tsimane’ using the mean yearly Gini coefficient of capture net worth because they omit debts. outcomes. Sociality. Sociality was modest. Most adults said they did not bestow Figure 1B shows four noteworthy results. [1] There is much variation gifts or consume chicha in the previous week. Sixty percent of adults said among Gini coefficients for different outcomes, from very high they had not given a gift the past week, and those who did reported inequality with Gini coefficients of 0.75 for the monetary value of goods offering few gifts (mean = 1.5; median = 1 gift). Three-quarters of adults received in barter during the past two weeks, area under manioc culti­ had not consumed chicha the week before the interview, and those who vation, and monetary value of livestock wealth, down to Ginis of about drank it reported doing so on one day. What effect did chicha drinking 0.3 for the intake of three macronutrients (proteins, carbohydrates, have on feeling happy or sad? For the week before the interview, we calories) consumed the past week. [2] Other than livestock wealth and estimated the association between the number of days an adult reported macronutrient intakes, flow variables related to barter and cash income having consumed chicha and whether they felt happy or sad. The esti­ had higher Gini coefficients (~0.6) than stock variables related to asset mates come from the seven years in which we measured the two emo­ wealth (Gini ~ 0.4). [3] Conclusions about the amount of wealth tions (2002–2006, 2009–2010). Drinking chicha was associated with an inequality versus income inequality hinge on how they are measured, increase in the chances of feeling both happy and sad; the likelihood of with the range of Gini values overlapping. Traditional, commercial, and feeling happy increased by 4.6 percentage points per day of drinking total wealth of the privately-owned assets we measured were more while the probability of feeling sad also increased, but by a slightly equally distributed (Gini ~ 0.4) than livestock wealth (Gini = 0.7). In­ lower amount (3.3 percentage points per day of drinking).2 The net ef­ come was more equally distributed if we equated income with the area fect of drinking was positive but small. The mixed effects of chicha of forest cleared for horticulture or with the monetary value of foods drinking could come from (a) alcohol having both stimulant and seda­ eaten (Ginis ~ 0.4) than if we equated income with cash earnings (Gini tive effects on the same person (Hendler et al., 2013), (b) happy and sad = 0.6). [4] The Gini coefficients of two proxies for income (consump­ people drinking for different motives, or (c) both of these reasons. tion) ─ (a) area deforested and (b) consumption of proteins, calories, Food. Each day, an average Tsimane’ reported eating food worth ~ and carbohydrates ─ were similar, with Gini coefficients near 0.4. USD 1 (6.86 bolivianos) in the form of calories (median = 3025 kcal), Variation in Gini coefficients among different outcomes has impli­ carbohydrates (median = 545 g [g]), proteins (median = 91 g), and fat cations for how one characterizes the economic inequality of a small- (median = 49 g). The intakes meet international recommendations scale society. If one uses aggregate asset wealth (grand Gini = 0.42; (Kraft et al., 2018, p. 1187). Measured through the average monetary Appendix F, Part 1), Tsimane’ are more egalitarian than two other value of daily food consumed per person reported in the surveys (6.86 contemporary rural societies where Gini coefficients for material wealth bolivianos; ~1 USD), Tsimane’ were better off than measured through have been estimated: Dominicans (Gini = 0.67) and Pimbwe (Gini = the average daily cash earned per person (4.65 bolivianos; USD 0.65), but 0.56) (Gurven et al., 2010, p. 58). Putting aside the validity of still far below the USD 1.9/day per person international poverty line. comparing Gini coefficients of asset inequality between industrial and Health. In the baseline sample of children and adults (n = 1453), the nonindustrial economies (and the possible fragility of Gini coefficients average Tsimane’ had been bedridden a total of 1.6 days the past two with small samples), Tsimane’ have much less asset inequality than five weeks (median = 0; SD = 3.8). The average BMI of adult men and non- industrial nations reported by Cowell et al., (2018,p. 336) ─ Italy (0.60), pregnant adult women in a household was similar (men = 23.36 [SD = UK (0.66), Finland (0.68), United States (0.80), and Sweden (0.89) ─ 2.1]; women = 23.14 [SD = 2.6]). Using the BMI of all adult men and all and less asset inequality than China (0.81) or India (0.86) (Davies et al., adult non-pregnant women surveyed during each year of the study 2017). These results suggest that Tsimane’ society is egalitarian in pri­ (observations = 4907) and following WHO (2000) definitions of weight vately owned total asset wealth, perhaps because most assets measured were inexpensive and available to most households.3 The story differs with income. The grand Gini coefficient of Tsimane’ 2 income inequality measured with monetary earnings or with area The results come from two separate ordinary least square regressions with deforested reached 0.63 and 0.41, respectively. The World Bank binary outcome variables for whether the respondent had felt happy or sad, computed Gini coefficients of income inequality for 82 countries for and, in both regressions, included as a predictor the number of days the respondent had consumed chicha. The recall period for both outcomes and for 2010, the last year of the panel study. The mean and median Ginis for the predictor referred to the seven days before the interview. The regressions these countries were 0.36 and 0.33. These Gini coefficients of income were run with robust standard errors clustered by village-year. The regression with happiness as an outcome had a p value of 0.001 (n=4198) and the 3 regression with sadness as an outcome had a p value of 0.02 (n=4197). P values In analyses not shown we find a generally positive correlation between the were not adjusted for false discovery. price of an asset and the Gini coefficient of the asset. 9 R. Godoy et al. World Development 176 (2024) 106518 Figure 1. Gini coefficients a) at baseline and b) yearly mean (2002–2010). Figure 1A comes from the column for the year 2002 in Appendix E (Part 1), and the histogram in Figure 1B represents the yearly means in Appendix E (Part 1). The variables are defined in Table 1, with slight alteration in the name of variables in the histograms. For clarity, a “w” was added as a suffix to the types of asset wealth; S = stock. F = flow. inequality would put Tsimane’ ─ if they were a country ─ as one of the many people did not need to barter, work for wages, or sell goods to earn most unequal societies in the world, on a par with South Africa (Gini = cash, probably due to household pooling of resources and governmental 0.63) if one measured income with cash earnings, or as a moderately transfers. Worse. The inflation-adjusted monetary value of wealth in unequal society (Gini ~ 0.41) if one measured yearly income with area goods made from local plants (traditional wealth, as defined in Table 1) deforested. Either way, Tsimane’ would still be in the top 25 % of un­ and wealth in livestock declined by 5.2 % and 6.3 % per year. The result equal societies, with 61 of the 82 countries in the World Bank dataset shows Tsimane’ were changing how they held asset wealth, away from having Gini coefficients below 0.41 (World-Bank, 2022b). The com­ traditional article and livestock toward commercial goods. Per capita parisons are not ideal but underscore the chief point that what economic caloric and protein consumption declined by 1.4 % and 2.3 % per year, inequality we use and how we measure inequality can lead to different as did the area under plantain cultivation (0.8 %/year). conclusions about the economic disparities of a small-scale society. We did not use a summary index to capture socioeconomic aspects of well-being because of the low cutoff value for Chronbach’s alpha (Ap­ 8. Main analysis: Micro-level trends in outcomes measured at pendix G). However, here we report the trend for (a) the index of all the household level socioeconomic and health variables and (b) the index for energy and macronutrients (protein, fat, carbohydrates) consumption. Each of the Table 2 (column 2A) suggests that during the study, 10 of the 21 two indexes is expressed as a principal component factor. Each year saw indicators measured at the household level did not change, and the in­ an improvement of 0.01 SD in the index of all socioeconomic and health dicators that did change showed that quality of life got better in some variables (p = 0.09; n = 1685) and an improvement of 0.02 SD in energy respects and worse in others. Unchanged. Examples of indicators that did + macronutrient intake (p = 0.078; n = 2261). Both total well-being not change include area and number of forest plots cleared, monetary (socioeconomic + health) and macronutrients consumption improved earnings, monetary value of goods received in barter, and sociality. by a small amount; the coefficients suggest that after a decade, total Better. The real value of wealth in commercial assets and total wealth, well-being and macronutrient intake would improve by only 0.1 SD to and the real monetary value of food consumption increased yearly by an 0.2 SD. average of 7.6 %, 4.2 %, and 7.4 % per person. During 2002–2006 and 2009–2010, we asked adults how satisfied they were with their diet and 9. Main analysis: Micro-level trends in economic inequality found that only 4 % of adults felt they had eaten poorly the past week; the share of dissatisfied respondents shrank from 7 % during 2002–2004 We find no evidence of changes in inequality (Table 2, column 4). to 2 % during 2009–2010. The likelihood of being bedridden in the past Although we cannot speak about trends in inequality with confidence two weeks fell by 0.8 %/year. In surveys done during occasional years, because we have only nine data points, we can describe the variability of we found Tsimane’ reported improved self-perceived health. The share Gini coefficients across years. In Figure 2, we show the inter-annual of adults rating their health as “poor” declined from an average of 21 % coefficient of variation (CV) of Gini coefficients for each outcome dur­ during 2002–2003 to 6 % during 2009–2010. Last, each year saw a one- ing 2002–2010. Figure 2 highlights three points. First, Gini coefficients percentage point increase in the likelihood an adult would be autarkic, varied considerably over time, from low CV values of 0.04 for the meaning they had not earned cash or bartered in the past fortnight. number of forest plots cleared, monetary value of goods received in Despite the greater presence of the outside world in the village economy, barter, or total asset wealth to high CV values of 0.8 or above for calories 10 R. Godoy et al. World Development 176 (2024) 106518 traits of villages (Table 2, column 2B). Changes in livestock wealth became statistically insignificant after conditioning for village fixed ef­ fects. [b] Method to compute inequality. The amount of inequality was sensitive to the method used to compute inequality. The ranking of outcomes by their economic inequality using the Gini coefficient (Figure 1B) differed from the ranking when using the coefficient of variation (Appendix F, Part 2), probably because the coefficient of variation is more sensitive to outliers, which is more likely to happen with small samples (Bendel et al., 1989). In the main analysis (Figure 1B), the monetary value of barter transactions, livestock wealth, monetary income, and area under manioc cultivation had the most inequality; none of these outcomes was among the most unequally distributed resources when using the coefficient of variation. 11. Extensions Gender. In Table 3 we compare growth rates of selected well-being indicators between women and men. The results provide some evi­ dence of a gender bias in favor of men. Yearly growth rates in the monetary value of barter transactions, total asset wealth, and traditional wealth were higher for men than women (by 5.7 %, 4.3 %, and 6 %). The number of gifts offered grew by 1.1 % per year more for men than for women. Only with commercial wealth did the yearly growth rate for women meaningfully surpassed the growth rate for men: 3.3 % per year more for women than men. Compared with men, women’s BMI and adiposity grew yearly by 0.3 % and 0.6 % more, but the effect sizes were small. We found no gender differences in the growth rates of monetary income, monetary value of livestock, chicha consumption, or self- reported morbidity. Hedonic measures. In Table 4, we show growth rates of hedonic measures of well-being, including information on differences between adult women and men. The most noteworthy finding is the lower like­ Figure 2. Inter-annual coefficient of variation of Gini coefficients, lihood of feeling negative and positive emotions or a trend toward 2002–2010. The coefficient of variation was computed by dividing the stan­ dard deviation by the mean yearly Gini coefficient of each outcome from Ap­ emotional neutrality. As the variable Year shows, from one year to the pendix E (Part 1). See notes to Figure 1 for the conventions followed to name next, an adult woman reported being (a) less fearful, sad, and angry the bins of the histogram. (5.5, 5.5, and 4.9 percentage points) but also (b) less happy (6.9 per­ centage points). The other noteworthy finding is that the trend in he­ and most macronutrients eaten, monetary value of foods eaten, and donic measures for men was like that of women, as suggested by the wealth in assets Tsimane’ make from local plants. Second, Gini co­ modest statistically significant results for the interaction term between efficients for asset wealth varied less between years (CV range the variables Male and Year. 0.05–0.08) than Gini coefficients for the two forms of income ─ area deforested (CV = 0.08) and food consumption (CVs ~ 0.07–0.09). Third, 12. Discussion the inter-annual variability in inequality in the more inclusive form of asset wealth (total wealth, CV = 0.049) was lower than the inter-annual This study contributes to research on well-being or quality of life, variability in the inequality of smaller asset bundles, such as wealth in and inequality. Before turning to the two topics, we discuss reasons for livestock (CV = 0.069), wealth in goods made from local materials (CV the micro-level trends that were unambiguous; these were trends in (a) = 0.081), or wealth in commercial articles (CV = 0.073). Not all in­ the composition of asset wealth, (b) diet, (c) monetary value of foods equalities changed in the same way; some were more stable. eaten, (d) perceived health, (e) plantain cultivation, and (f) gender A snapshot of inequality could yield an unrepresentative picture of differences. longer-run realities because some Gini coefficients varied significantly over a short time. For example, a study among Tsimane’ of area defor­ 12.1. Reasons for unambiguous trends ested by households ─ as a proxy for income ─ done in 2004 (Gini = 0.44) would find a 29 % larger dispersion in income than a study four (a) Changes in the composition of asset wealth (reduction in tradi­ years later (Gini 2008 = 0.34) (Appendix F, part 1). Using total asset tional wealth; increase in commercial wealth) reflect intentional wealth instead of smaller asset bundles reduces variability and the need choices by Tsimane’ in how they wanted to hold wealth. For for repeated measures, but even then, differences in Gini coefficients example, instead of using bows and arrows to fish or hunt, or between years, though more modest, would still be misleading about the cotton bags to carry goods, Tsimane’ shift to use commercial longer-run situation. For instance, a study of total asset wealth in 2009 fishing nets, metal hooks, rifles, shotguns, or backpacks, in part would find a 16 % higher Gini (0.44) than a study done three years because these commercial goods are cheaper and last longer than earlier (Gini 2006 = 0.38) (Appendix F, part 1). goods made from plants. (b) In energy and macronutrient consumption we see declines in 10. Robustness calories and proteins. The first could reflect a trend toward less physical activity ─ and correlative lower demand for calories ─ [a] Village fixed effects. Besides the growth rate of livestock wealth, from a more sedentary lifestyle (Section 3 and Rosinger et al., results from the main analysis did not change after controlling for fixed 2013) and the second could reflect greater defaunation (Fernández-Llamazares et al., 2017). 11 R. Godoy et al. World Development 176 (2024) 106518 Table 3 Differences in yearly growth rates of well-being indicators between adult (age ≥ 16y) women and men: 2002–2010. (1) (2) (3) (4) (5) (6) Variables: Income Barter Autarky Total Traditional Commercial Male*year − 0.034 0.057** − 0.002 0.043*** 0.060** − 0.033** (0.029) (0.020) (0.003) (0.011) (0.020) (0.016) Year 0.019 − 0.063*** 0.011*** 0.024*** − 0.065*** 0.117*** (0.019) (0.012) (0.002) (0.007) (0.012) (0.014) Male 70.852 − 113.634** 3.258 − 86.056*** − 119.994** 67.863** (58.629) (40.066) (6.289) (21.401) (39.679) (32.799) Constant − 35.669 127.595*** –22.464*** − 40.137*** 136.190*** − 229.223*** (37.374) (24.734) (4.502) (14.638) (23.488) (27.590) Observations 5,397 5,397 5,598 5,376 5,376 5,376 R-squared 0.138 0.012 0.009 0.237 0.012 0.331 (7) (8) (9) (10) (11) (12) Variables: Animals Chicha Gifts Bed BMI Body Fat Male*year 0.040 0.004 0.011* − 0.011 − 0.003*** − 0.006* (0.032) (0.005) (0.006) (0.011) (0.001) (0.003) Year − 0.044 − 0.011*** − 0.010 − 0.034*** 0.005*** 0.014*** (0.021) (0.003) (0.004) (0.008) (0.001) (0.002) Male − 79.476 − 8.019 − 21.913* 22.358 6.187*** 10.709 (64.183) (10.767) (11.718) (22.137) (2.395) (6.674) Constant 91.633** 21.358*** 20.820** 69.720*** − 6.006*** –23.816*** (41.847) (6.768) (8.310) (15.566) (1.918) (5.013) Observations 5,376 5,273 5,397 5,522 4,907 5,198 R-squared 0.001 0.040 0.003 0.012 0.005 0.374 Note: Definition of outcomes are the same as in tables 1-2, with one exception: We added the sum of four skinfolds as a measure of subcutaneous body fat store, necessary for immune function. Other than Autarky (1 = autarkic, 0 = otherwise), all other outcomes are logged. Only indicators measured at the individual level are included. Regressions are OLS with robust standard errors clustered by study participants. P values of variables are adjusted for false discovery rates; *, **, and *** significant ≤ 10 %, ≤5%, and 1 %≤. In addition, the probability of eschewing plantain cultivation Table 4 during this period rose by 2.3 %/year, whereas the probability of Comparison of yearly micro-level trends in hedonic measures of well-being be­ not growing rice remained flat at 0.1 %/year. tween adult (age ≥ 16y) women and men: 2002–2006, 2009–2010. (f) Trends in gender differentials defy an easy explanation. From one (1) (2) (3) (4) year to the next, men were more likely to accumulate traditional Variables: Fear Happy Sad Angry assets and to barter and were less likely to accumulate commer­ Male*year 0.013** 0.004 − 0.007 0.008* cial asset; one might have expected the opposite because men (0.005) (0.005) (0.005) (0.004) have a stronger foothold in the market economy, have more Year − 0.055*** − 0.069*** − 0.055*** − 0.049*** (0.003) (0.004) (0.003) (0.003) schooling, and are more likely to speak the national language Male − 27.131** − 8.608 13.322 − 16.750 (Spanish). It is possible that men were acquiring more household (9.770) (10.236) (10.220) (8.934) commercial articles that got assigned to women. Compared with Constant 111.641*** 138.892*** 111.556*** 99.061*** women, men had barely discernible yearly reductions in BMI and (7.012) (7.368) (6.954) (6.331) Observations 4,077 4,077 4,076 4,077 body fatness and the same likelihood of feeling happy or sad over R-squared 0.092 0.167 0.107 0.095 time. The weak evidence for gender differences in rates of change suggests that Tsimane’ women have maintained strong decision- Note: Outcomes are binary variables, with names = 1 if respondent reported making power in their household. Women are empowered having experienced the emotion in the past week, and zero otherwise. All other notes are the same as in Table 3. through the support of kin, strong respect in Tsimane’ households for the private property each spouse brings to a union, and cooperation in horticulture (Bauchet et al., 2021). (c) The rise in the real monetary value of foods eaten reflects a growing propensity to acquire food. During 2004–2010, the Like other Indigenous Peoples in the Amazon, Tsimane’ have seen chances an adult would buy food in the past fortnight rose by 1.4 their share of intrusion from missionaries, oil firms, gold panners, drug percentage points per year; during 2002–2010, the likelihood an traffickers, loggers, cattle ranchers, and colonist smallholders from the adult would barter for food rose yearly by 1.8 percentage points. Andean highlands (Reyes-García et al., 2012), but not in the extreme From 2002 to 2010, the diet changed away from wildlife and form seen in other parts of the Amazon. Until recently, Tsimane’ have crops grown by Tsimane’ toward processed foods from the been able to protect themselves by living in remote places and, when marketplace, principally cooking oil and pasta (Bethancourt necessary, fighting back or retreating farther into the backlands. Under et al., 2019). Unfortunately, our price data is too spotty to assess these conditions, exposure to the outside world goes along with a small the role of relative food prices in diet trends. bundle of micro-level changes, barely perceptible taken together, as (d) Improvement in perceived health reflects the spread of public shown by the 0.01SD/year change in the index of socioeconomic and health services (Section 3), a rise in real cash expenditures in health dimensions measured (Section 8). medicines and health services by households (5 %/year during 2004–2010), and panel conditioning (Section 6). (e) Many factors explain the reduction in plantain cultivation but a 12.2. Well-being simple one is the growing preference to cultivate rice, a leading cash crop and an appreciated food. During 2002–2010 the total This article contributes to (i) longitudinal studies of well-being or area under plantain cultivation for a household declined by ~ 8 quality of life, (ii) studies of well-being in small-scale societies, and (iii) %/year, while the area under rice cultivation rose by ~ 1 %/year. studies of deprivation in the rural Global South. Longitudinal studies of well-being. The past two decades have seen 12 R. Godoy et al. World Development 176 (2024) 106518 many case studies and cross-cultural studies on longitudinal trends in 12.3. Economic inequality in small-scale societies subjective well-being (Diener & Chan, 2011; Diener et al., 2018). These studies come mostly (but not exclusively) from industrial nations. The We included economic inequalities because studies across societies largest ones rely on nationally representative samples, use the same find that they often correlate with poorer quality of life, principally survey questions across time and place, and analyze information at the health (Section 1.2). Although there is some evidence that economic individual level. Much of that research is sparked by a search for de­ inequalities matter for quality of life in small-scale societies, there is less terminants (Carla & Falco, 2018; Du et al., 2019) or by an interest in agreement on which economic disparities to assess or the appropriate identifying the effects of subjective well-being (Diener & Chan, 2011; methods to measure them. So far, most research has focused on asset Zaninotto & Steptoe, 2019). The use of the same methods of data inequality rather than income inequality and has found that asset collection across sites and time facilitates comparisons and the assess­ inequality varies between villages (Jaeggi et al., 2021), types of sub­ ment of external validity and permits lagging predictors to enhance sistence (Tucker et al., 2015), kinship systems (Mattison et al., 2023), claims about causal effects. The studies have often found positive as­ and generations (Borgerhoff Mulder et al., 2009; Gurven et al., 2010). sociations (but also some heterogeneity) between monetary income and Fewer studies have examined income inequality per se and, except for subjective well-being (Blázquez-Cuesta & Budría, 2012; Das et al., 2020; Tucker et al. (2015) and Reyes-García et al (2019), the other studies Zhang & Hong, 2023). In contrast, our micro-level panel study comes equate income exclusively with monetary earnings and they equate in­ from one place and rests on idiosyncratic indicators of well-being or come inequality with earnings inequality (Godoy et al., 2005; Godoy quality of life relevant to one society (Kaufman et al., 2022). It also et al., 2004; Gurven et al., 2015), clearly inadequate since much of a differs from international panel studies because it is motivated by a household’s income comes from consuming what they produce. The different question and approach. We are interested in a pixelated shortfall needs correction because the defining debate of our time cen­ description of micro-level change in well-being as defined by study ters on income inequality not on wealth inequality (DeLong, 2022; participants in an economically self-sufficient society and avoid identi­ Piketty, 2022), though income and wealth are (far from perfectly) fying determinants (Section 1.1). We supplement the current literature correlated (Keister & Lee, 2017; Piketty, 2014; Saez & Zuckman, 2014). through a quantitative ethnographically grounded understanding of Without or with little monetary savings, monetary investments, or micro-level change. monetary expenditures, as happens in small-scale societies, current Well-being in small-scale societies. Anthropological studies of well- household consumption becomes conceptually equivalent to current being among pastoralists, hunter-gatherers, and horticulturists have household income, even if one cannot see or measure current income. generally eschewed examining well-being comprehensively, have relied Ernst Engel’s Law predicts that household food consumption will mostly on cross-sectional data, and have focused in identifying pre­ dominate current total household consumption in such settings. If so, dictors (not consequences) of well-being. In a comparative study of three then consumption inequality will track unobserved income inequality. horticultural-foraging societies (Tsimane’ in Bolivia, Baka in Cameroon, But how should one aggregate food consumption and use it to measure Punan in Indonesia), researchers asked participants during four consumption inequality without relying on prices? We propose to use consecutive quarters to rate their life satisfaction and the reasons for inequality in energy and macronutrient consumption as a first approx­ their rating and found that health was the leading perceived cause of life imation to consumption inequality (and by implication to income satisfaction (Reyes-García et al., 2021). In a cross-sectional study of inequality) (Bell et al., 2021). The measure is incomplete because it fishing communities in the Solomon Islands and Bangladesh, researchers leaves out the consumption of leisure, shelter, amenities, water, services, found that evaluative and hedonic measures of well-being bore a weak and public goods, but is a sensible start. association with an index of monetization (Minarro et al., 2021). Like The intuition needs validation yet note from figures 1A-1B how near those studies, our study struggled with how to measure well-being in an to each other lie the Gini coefficients of the amounts of protein, energy underrepresented type of society, but our approach differs by doc­ (calories), and carbohydrates eaten, and how near all of them are to the umenting micro-level change and by putting aside concerns with area of forest cleared, a coarse measure of yearly income capturing the determinants. production of all annual crops to eat, swap, store, and sell. Although area Deprivation. This study contributes to the literature on deprivation by deforested can proxy for current income in horticultural societies, it is stressing the difficulties of using the label or its neighbor, poverty not as useful as energy or macronutrient intake to gauge income (Alkire et al., 2015; UNDP 2022). During the study, Tsimane’ were inequality among pastoralists, fishers, or hunter-gatherers because these money poor, had low levels of privately owned asset wealth, had low societies rely on other livelihood activities. Energy and macronutrient levels of schooling, a life expectancy of 53 years (Gurven et al., 2007), inequality serve as a more general currency. In sum, the suggested widespread respiratory and gastrointestinal infections (Appendix B), approach would allow assessing inequality in current household con­ widespread child growth stunting (Undurraga et al., 2018), a diet poor sumption or unseen household current income in a broader range of in key micronutrients and diversity (Appendix B), and non-trivial small-scale societies. amounts of economic inequalities, as we have seen. These indicators would put Tsimane’ into the deprived or poor bin. If so, why have so few 13. Conclusions Tsimane’ emigrated? (Section 3). One can build on Graeber and Wen­ grow (2021) and speculate about what offsets these objective indicators The article points to two future conversations. of deprivations, many of which have an “agreed normative interpreta­ The first is a practical conversation between members of small-scale tion”. The pull to stay might come from the historic freedom Tsimane’ societies, researchers, decision-makers, civil society, and international have enjoyed to pursue a mix way of life combining farming and organizations about how to define well-being so the definition captures foraging (Gehring, 2013), ample wilderness, village and household au­ what is meaningful to people in these societies, the need to collect data tonomy and choice, social support from a dense network of kin reflecting on quality of life in small-scale societies, how to pay for the effort, what a system of preferential cross-cousin marriage, a lifestyle with low stress type of data (if any) to collect, who owns the data (Reyes-García et al., (Appendix B), and the perceived psychic costs of migration (Chen et al., 2022), and how to use it once collected. If the consensus is to move 2019). Living in a remote area, Tsimane’ have been able to seize the forward, the information should be collected over time following the rudder when adopting aspects of the Western world and the same protocol of data collection used in the rest of the country to make it marketplace. possible to track changes, compare quality of life and social exclusion between members of small-scale societies and groups in the rest of country, and decide what to do about gaps or trends in gaps. The second conversation is between nutritionists, economists, and 13 R. Godoy et al. World Development 176 (2024) 106518 biocultural anthropologists about how to best measure food consump­ Declaration of competing interest tion inequality in small-scale societies. The first step would be to vali­ date the idea that in these societies, consumption approximates unseen The authors declare that they have no known competing financial income, and that food consumption dominates aggregate household interests or personal relationships that could have appeared to influence consumption. This would make it possible to test the external validity of the work reported in this paper. prior findings from industrial societies about the effects of income inequality on individual well-being using a valid measure of income for Data availability small-scale societies. Appendix A describes where the public can obtain the coarse and Disclaimer clean datasets of the study + the computer codes used in this article. Data + computer codes are freely available to anyone The scientific results and conclusions, as well any views or opinions expressed herein, are those of the authors and do not necessarily reflect Acknowledgements the views of NOAA or the Department of Commerce. We would like to thank two reviewers and the editor of World CRediT authorship contribution statement Development for several rounds of comments on the initial and two subsequent revisions. Thanks also to Tsimane’ who participated in the Ricardo Godoy: Conceptualization, Formal analysis, Funding study, the Tsimane’ Council, and the TAPS research team. Grants from acquisition. Jonathan Bauchet: Conceptualization. Jere R. Behrman: the following institutions funded this research. [A] Program of Cultural Writing – review & editing, Formal analysis. Tomás Huanca: Method­ Anthropology of the National Science Foundation (NSF) of the USA: ology, Supervision. William R. Leonard: Conceptualization, Data BCS: 0650378, BCS: 0552296, BCS: 0200767, BCS: 0111905, BCS: curation, Funding acquisition, Writing – review & editing. Victoria 0104575, BCS: 0078801. [B] World Bank, Washington DC: Might cul­ Reyes-García: Conceptualization, Data curation, Funding acquisition, ture pay off? Using an experimental design to evaluate the effects of Writing – review & editing. Asher Rosinger: Formal analysis, Writing – farming innovations and cultural empowerment among lowland Amer­ review & editing. Susan Tanner: Conceptualization, Data curation, indians in Bolivia. [C] The John D. and Catherine T. McArthur Foun­ Formal analysis, Writing – review & editing. Eduardo A. Undurraga: dation, Chicago, Illinois: 56633. Encroachers, conservation, and the Conceptualization, Formal analysis, Visualization, Writing – review & welfare of Indigenous People in the rain forest of Bolivia. [D] This work editing. Ariela Zycherman: Conceptualization, Formal analysis, contributes to the ICTA-UAB “María de Maeztu” program for Units of Writing – review & editing. Excellence (CEX2019-000940-M). Appendix A The clean de-identified panel dataset (2002–2010) used for this article is in Stata V17 and is open to the public without restrictions. The dataset can be downloaded from the following two web pages (A and B): A: Electronic library of Brandeis University: https://scholarworks.brandeis.edu/esploro/outputs/dataset/Overview-of-the-Tsimane-Amazonian-Panel/9923926193901921?in stitution ¼ 01BRAND_INST B. Inter-University Consortium for Political and Social Research (ICPSR), University of Michigan: https://www.icpsr.umich.edu/web/ICPSR/studies/37671 or. http://dx.https://doi.org/10.3886/ICPSR37671.v2 The yearly unprocessed de-identified datasets of the panel are also open to the public without restrictions and can be downloaded from the following web page: C: Electronic library of Brandeis University. https://scholarworks.brandeis.edu/esploro/search/outputs?query¼any,contains,godoy% 20coarse&page¼1&institution¼01BRAND_INST&scope¼Research&sort¼rank&mfacet¼facet_rtype,include,datasets,1 An extensive report summarizing aspects of the panel study is open to the public and can be downloaded from the following web page: https://scholarworks.brandeis.edu/permalink/01BRAND_INST/1bo2f6t/alma9923926194001921 The Stata 17 computer codes are available as a supplement to this article and the guidelines on how to replicate the results are described in a supplementary Word document titled, “Guidelines to replicate results”. Last, the unprocessed and processed datasets for the ancillary studies of the panel are available in [A] and [B]. Appendix B. . Ethnographic sketch of outcomes Horticulture: Forest area and number of plots cleared, and hectares under plantains and manioc cultivation. Tsimane’ have been accomplished horti­ culturists since at least the early twentieth century (Nordenskiöld, 1979 [orig. 1924], pp. 34–35). Every year most households clear one plot of old- growth forest or fallow forest to farm. Forest area cleared is a meaningful outcome because it proxies for food availability ─ a frequently mentioned 14 R. Godoy et al. World Development 176 (2024) 106518 reason for happiness and sadness ─ and total yearly income (cash and imputed). In the forest area cleared, households put in their three main annual crops (rice, manioc, and maize) and the main perennial crop (plantain) they will consume, barter, store, and sell for the rest of the year (Vadez et al., 2008). The annual deforested area captures both household consumption from own horticultural production and earnings from selling the four leading crops and is thus a valid proxy for household yearly non-monetary income, plus some monetary income. Though reasonable as a proxy for total household income, area deforested underestimates total monetary income because it excludes monetary earnings from wage labor and from the sale of wildlife, discussed in the next heading. It also underestimates household consumption because it excludes consumption of wildlife. Farm crops provided 86 % of the carbohydrates, 64 % of the calories, and 35 % of the fats in the diet of Tsimane’. Among farm crops, rice and plantain occupy most of a cleared forest parcel because Tsimane’ use the crops as their principal food staple and cash crops (Zycherman, 2013). We focus on plantain because it provides 60 % of cultigen calories (Kraft et al., 2018), is a chief cash crop, and is a safety net as it tolerates neglect (but not flooding) and attacks from pests, diseases, and predators. Manioc is important not as a food (Tsimane’ rarely eat it) but as the principal ingredient to make chicha, the fermented beverage households offer guests to display sociality (Zycherman, 2015). Economic flows: Income, barter, and autarky. We considered two economic outcomes: (a) flows of monetary income (earnings) and the monetary value of goods received in barter and (b) stocks of privately owned physical assets, discussed in the next section. Tsimane’ engage with the marketplace intermittently. To earn cash, men work for cattle ranchers, logger, and for non-Indigenous People dwelling inside or next to the homeland of Tsimane’. Almost no Tsimane’ women work for wages. Women and men earn cash by selling thatch palms from the forest and farm produce (crops, eggs, livestock) to traveling traders who come to the village or directly to consumers and merchants in towns. Tsimane’ also relied on barter to obtain meat and market foods (e.g., refined white sugar, pasta) and, in exchange, supply rice, plantains, and thatch palms. Economic-wealth: Total, traditional, commercial, and livestock. Tsimane’ have clear private property rights to tools, livestock, and household utensils. Parents give chicks as gifts to their children so children learn how to care for animals, and women and men each own assets separately, though adults in a household have usufruct rights to the goods of others. We grouped privately owned physical assets into three categories (Table 1). These forms of storing private wealth go back at least a century. In the early twentieth century, Nordenskiöld noted Tsimane’ built conical coops to guard poultry from night predators, listed the many locally made and commercial goods Tsimane’ owned (Nordenskiöld, 1979 [orig. 1924], 2001 [orig 1924]), and remarked on how Tsimane’ did not hoard assets; they made or acquired an asset as they needed it, not before (Nordenskiöld, 2001 [orig 1924], p. 157). Sociality: Gifts and chicha. The main forms of sociality we measured center on (a) gifts and (b) convivial drinking of chicha. Tsimane’ typically bestow meat and fish as gifts. When they help each other with labor, they help mainly with farm chores. Adverse regional shocks such as floods do not trigger much intra-village or inter-village transfers of goods or aid from other households, in part because outside institutions step in to help. However, bedridden adults under ordinary circumstances were more likely to receive gifts. Of all forms of displaying pro-social behavior, public drinking of chicha in a village is the most ubiquitous (Rosinger & Bethancourt, 2019; Zycherman, 2015). From the time they first set food in the homeland of Tsimane’, Protestant missionaries prohibited the drinking of alcohol (including fermented chicha); it is possible that villagers underreported the amount of chicha drank to avoid potential opprobrium (Godoy et al., 2021). Only women make chicha, which each sex drinks separately. Chicha is served by women in the village and is offered to anyone who stops by. Men sit in a circle as they sip from a gourd, passing it to the person next to them; when the gourd is empty, a woman in the household refills it until the vat is empty. Women drink likewise in a separate circle. Households gain silent praise for sharing chicha. Foods: Value and macronutrients (calories, proteins, fat, carbohydrates). Pauly (1928, pp. 117-118) and Nordenskiöld (2001 [orig 1924], pp. 158–161) noted that Tsimane’ enjoyed a well-balanced diet of wildlife and home-grown crops. Fish eclipsed game animals in their diet and, among foods from livestock, only poultry stood out. Tsimane’ had a profusion of food but did not share it with Nordenskiöld. Instead, Tsimane’ regaled Nordenskiöld with chicha. In summarizing his impressions of the Tsimane’ diet, he wrote that the land of Tsimane’ was “without doubt a good place to live. I don’t think these people are ever hungry” (p. 161). During our study, the diet of Tsimane’ consisted of a mix of crops they grew, fish and game animals they caught, and market foods they bought or bartered for (Rosinger et al., 2013). Their diet, though adequate in energy and macronutrients, lacks diversity (Kraft et al., 2018) and may be limiting key micronutrients (Bethancourt et al., 2019; Foster et al., 2005; Kraft et al., 2018, pp. 1186–1187). Health: Bedridden days and BMI. The health afflictions of Tsimane’ include gastrointestinal (McDade et al., 2007; Tanner et al., 2013) and respi­ ratory infections (Gurven et al., 2017), accidents, and illnesses from natural disasters like floods. In 2009, the Bolivian government started a con­ ditional cash-transfer program for pregnant women and mothers with children under two years of age who met program requirements, such as mothers having pre-natal checkups. Compared with international height references, Tsimane’ adults are short in stature (Godoy et al., 2006). Our data for the last year of the panel (2010) suggests that adults (n = 637) were below international standards in z-scores of weight-for-age (-0.78), age-sex scores of arm muscle area (-0.60), and sum of triceps and subscapular skinfolds (-0.45). Despite the burden of disease, Tsimane’ are physically active and have outstanding cardiovascular health from a way of life amalgamating horticulture, fishing, hunting, and plant collection (Gurven et al., 2013; Kaplan et al., 2017). Bethancourt et al. (2019) found significant increases in the prevalence rates of both overweight and obesity among women and men and Nyberg (2009, 2012) found Tsimane’ had very low cortisol levels (a biomarker of stress). Appendix C. . Textual answers by adults about what made them happy or sad the past week This appendix documents verbatim answers of respondents during 2002–2004. Twenty-five and 24 unique answers accounted for at least 1 % of the reasons for being happy or sad; these are the ones shown in this appendix. Happy. The dataset on happiness had a total of 2831 observations. A person who said nothing had made them happy counted as one observation, but so would a person who said only one event had made them happy. People who said nothing had made them happy are excluded from the tabulation below. Of the total of 2831 observations, 2390 observations had a reason for happiness. 25 reasons were mentioned at least 1 % of the time; they accounted for 84.42 % of the 2390 observations. Verbatim answers N Percent Work well 216 7.6 In company of family in household 207 7.3 Good fish harvest 173 6.1 Drink chicha 150 5.3 (continued on next page) 15 R. Godoy et al. World Development 176 (2024) 106518 (continued ) Verbatim answers N Percent Hunting by self or someone in household 150 5.3 Good work in the fields 123 4.3 Play soccer or do sports 107 3.8 Visit towns 105 3.7 Good food; there was food 103 3.6 In company of non-household family members 96 3.4 Village festivity 91 3.2 Harvest or goods from the field 87 3.1 Visit relatives 87 3.1 Visit by relatives in the village 83 2.9 Good health 73 2.6 Visit by relatives living outside the village 68 2.4 Fields produce well 68 2.4 Eat meat 67 2.4 Sell well 62 2.2 Purchases 58 2.0 Domestic animals 53 1.9 Family has good health 47 1.7 Listen to Church messages 46 1.6 Eat fish 41 1.4 Take a walk /a/ 29 1.0 Sad. There were a total of 2379 observations on the reasons for sadness, including respondents who said nothing had made them sad. Of this total, 2000 observations (84.06 %) of the 24 reasons for sadness accounted for at least 1 % of all reasons for sadness. Verbatim answers N Percent Own illness or accident 325 13.7 Illness of someone in the household 278 11.7 Lack of food, poor-quality food 270 11.3 Insults, swears 213 9.0 Lack of meat 79 3.3 Death, illness, or loss of animals 69 2.9 Theft 67 2.8 Lack of money 65 2.7 Lies 62 2.6 Mad at non-household family members 56 2.4 Poor weather 52 2.2 Crop loss 51 2.1 Inferior quality of farm plots 50 2.1 Bad dreams 47 2.0 Drunkenness 42 1.8 No hunting 38 1.6 Mad at a household member 37 1.6 No fish 36 1.5 Bitten by wasp, sting ray, etc. 31 1.3 Unable to work in the fields 29 1.2 A household member left 29 1.2 Accident in the family 25 1.1 Visitors don’t come 25 1.1 Almost had an accident 24 1.0 Notes: N: Frequency of activity Percent: N/2831 for happiness and N/2379 for sadness /a/ The phrase “take a walk” does not capture the meaning of the Tsimane’ word sóbaqui, which stands for visiting, strolling, or taking a leisurely walk for pleasure (Ellis, 1996). Appendix D. . Part 1. Summary statistics of outcomes among households at baseline (2002): Number of households, mean, and standard deviation /a/ Outcomes: N Mean SD Horticulture: Forest 228 0.179 0.216 Plots 228 0.298 0.205 Plantains 227 0.112 0.131 Manioc 226 0.021 0.051 Economic flows: Income 245 4.65 15.7 Barter 245 0.337 0.561 (continued on next page) 16 R. Godoy et al. World Development 176 (2024) 106518 (continued ) Outcomes: N Mean SD Autarky 245 0.061 0.24 Economic stocks: Wealth Total 243 886 813 Traditional 243 228 263 Commercial 243 503 560 Livestock 243 155 339 Sociality: Gifts 245 0.085 0.089 Chicha 245 0.06 0.084 Food: Value 241 6.86 8.004 Macronutrients /b/ Calories 257 3025 2147 Proteins 257 91 98 Fats 257 49 83 Carbohydrates 257 545 360 Health: Bedridden 245 0.131 0.215 BMI-women 192 23.143 2.692 BMI-men 214 23.364 2.105 /a/ Table 1 has definition and measurement of variables. /b/ For macronutrients, values in the column “Mean” refer to medians instead of averages, which we used as an additional step to redress the influence of outliers. Appendix D. . Part 2. Year-by-year mean of outcomes among households (2002–2010) 2002 2003 2004 2005 2006 2007 2008 2009 2010 Horticulture: Forest 0.179 0.201 0.173 0.184 0.207 0.179 0.205 0.230 0.199 Plots 0.298 0.387 0.298 0.306 0.301 0.292 0.271 0.282 0.290 Plantains 0.112 0.127 0.067 0.085 0.052 0.041 0.052 0.055 0.061 Manioc 0.021 0.025 0.012 0.022 0.019 0.016 0.021 0.030 0.018 Income flows: Income 4.650 3.429 4.231 5.131 4.584 4.774 5.946 5.584 5.454 Barter 0.337 0.412 0.330 0.455 0.408 0.315 0.331 0.377 0.370 Autarky 0.061 0.120 0.076 0.067 0.111 0.164 0.134 0.158 0.123 Wealth: Total 886 867 854 931 813 911 1393 1019 1262 Traditional 228 209 180 208 197 253 237 192 115 Modern 503 524 520 576 504 515 978 651 968 Livestock 155 134 154 146 112 143 178 176 179 Sociality: Gifts 0.085 0.059 0.035 0.075 0.070 0.092 0.058 0.065 0.062 Chicha 0.060 0.054 0.067 0.077 0.061 0.070 0.068 0.046 0.050 Food: Value 6.86 7.30 9.64 8.16 6.49 7.74 12.51 8.28 14.81 Macronutrients: Calories 3025 3714 2629 2851 3056 2275 2624 3116 2817 Proteins 91 118 81 88 90 73 78 90 85 Fats 49 61 42 43 48 47 49 50 53 Carbohydrates 545 642 453 532 564 396 440 549 519 Health: Bed 0.131 0.184 0.064 0.081 0.105 0.134 0.075 0.089 0.053 BMIF-women 23.143 23.446 23.344 23.481 23.762 23.645 23.961 23.574 24.302 BMI-men 23.364 23.486 23.213 23.527 23.658 23.437 23.573 23.655 23.753 For definition of variables see notes to Appendix D, Part 1. Appendix E. . Regression results for column 2A of Table 2: No village fixed effects [1] [2] [3] [4] [5] [6] [7] Forest Plots Plantain Manioc Income Barter Autarky Year 0.004 − 0.006 − 0.008*** 0.0001 0.023 − 0.004 0.010* (0.002) (0.003) (0.002) (0.001) (0.018) (0.007) (0.003) Constant − 7.170* 11.989** 15.971*** − 0.394 − 44.412 7.395 − 19.023*** (4.169) (5.285) (3.223) (1.308) (36.708) (13.229) (6.716) Observations 2,193 2,210 2,183 2,182 2,261 2,261 2,261 R-squared 0.004 0.005 0.039 0.001 0.002 0.001 0.006 [8] [9] [10] [11] [12] [13] [14] Total Traditional Commercial Animals Food Calories Proteins Year 0.042*** − 0.052** 0.076*** − 0.063** 0.074*** − 0.014* − 0.023* (0.009) (0.017) (0.013) (0.021) (0.014) (0.007) (0.009) Constant − 76.457*** 109.899*** − 144.893*** 131.611*** − 145.222*** 35.775*** 51.833*** (18.743) (35.040) (25.233) (41.450) (27.439) (13.556) (18.242) (continued on next page) 17 R. Godoy et al. World Development 176 (2024) 106518 (continued ) [1] [2] [3] [4] [5] [6] [7] Observations 2,243 2,243 2,243 2,243 2,235 2,261 2,261 R-squared 0.021 0.020 0.051 0.008 0.053 0.004 0.008 [15] [16] [17] [18] [19] [20] [21] Fat Carbs Gifts Chicha Bed BMI-W BMI-M Year 0.001 − 0.012 − 0.0001 − 0.001 − 0.008*** 0.004* 0.002 (0.013) (0.007) (0.001) (0.001) (0.002) (0.002) (0.001) Constant 3.093 30.081** 0.590 2.438 15.627*** − 5.153 0.782 (25.767) (15.009) (1.829) (2.458) (3.704) (3.161) (2.232) Observations 2,261 2,261 2,261 2,261 2,261 1,890 2,018 R-squared 0.001 0.002 0.001 0.001 0.020 0.009 0.001 Appendix E. . Regression results for column 2B of Table 2: With village fixed effects [1] [2] [3] [4] [5] [6] [7] Forest Plots Plantain Manioc Income Barter Autarky Year 0.004 − 0.006 − 0.008*** 0.001 0.026 − 0.003 0.01 (0.002) (0.003) (0.002) (0.000) (0.023) (0.005) (0.005) Constant − 7.346 11.723* 15.863*** − 0.137 − 50.185 6.585 − 19.092* (4.897) (6.337) (3.305) (0.841) (46.987) (9.666) (9.050) Observations 2,193 2,210 2,183 2,182 2,261 2,261 2,261 R-squared 0.004 0.005 0.044 0.001 0.003 0.001 0.006 [8] [9] [10] [11] [12] [13] [14] Total Traditional Commercial Animals Food Calories Proteins Year 0.043** − 0.048* 0.077*** − 0.068 0.074*** − 0.014** − 0.023** (0.011) (0.015) (0.012) (0.025) (0.013) (0.004) (0.007) Constant − 78.994*** 102.851*** − 148.370*** 140.599** − 146.767*** 36.366*** 51.978*** (22.980) (29.270) (23.793) (50.186) (25.113) (8.759) (14.280) Observations 2,243 2,243 2,243 2,243 2,235 2,261 2,261 R-squared 0.022 0.019 0.055 0.009 0.057 0.004 0.008 [15] [16] [17] [18] [19] [20] [21] Fat Carbs Gifts Chicha Bed BMI-W BMI-M Year 0.001 − 0.012* − 0.001 − 0.001 − 0.008*** 0.005 0.002 (0.008) (0.004) (0.001) (0.001) (0.001) (0.002) (0.001) Constant 2.696 31.281*** 0.189 2.415 16.100*** − 5.555 0.533 (16.217) (8.805) (1.566) (2.857) (2.593) (3.361) (2.712) Observations 2,261 2,261 2,261 2,261 2,261 1,890 2,018 R-squared 0.001 0.003 0.001 0.001 0.021 0.010 0.002 Notes: Regressions are ordinary least squares with robust standard errors clustered by households in a village each year (column 2A, Table 2) or clustered by households in a village (column 2B, Table 2). For the year coefficient, *, **, and *** indicate significance levels at ≤ 10 %, ≤5%, and ≤ 1 % adjusted by the Westfall and Young method for false discoveries, as developed and adapted to Stata by Jones et al. (2019). Other than Autarky, all other outcomes were transformed using the inverse hyperbolic sine function. The variable Autarky was left as defined in Table 1. Headings in columns 8–11 of this appendix refer to the type of asset wealth, as defined in Table 1. Carbs = carbohydrates. Appendix F. . Part 1: Gini coefficients of economic inequalities Baseline Yearly Grand Outcomes: 2002 2003 2004 2005 2006 2007 2008 2009 2010 Mean Gini %Δ year Horticulture: Forest 0.459 0.386 0.446 0.361 0.420 0.408 0.346 0.423 0.437 0.409 0.414 − 0.5 Plots 0.359 0.376 0.403 0.354 0.376 0.371 0.366 0.384 0.397 0.376 0.380 0.5 Plantain 0.512 0.517 0.611 0.535 0.612 0.627 0.551 0.671 0.553 0.577 0.594 1.7 Manioc 0.735 0.637 0.802 0.725 0.731 0.740 0.733 0.828 0.801 0.749 0.758 1.6 Economic: Income 0.670 0.594 0.646 0.569 0.588 0.692 0.638 0.650 0.637 0.631 0.639 0.4 Barter 0.717 0.732 0.735 0.749 0.692 0.797 0.804 0.792 0.778 0.756 0.759 1.3 Wealth: Total 0.412 0.400 0.403 0.407 0.386 0.393 0.438 0.449 0.432 0.414 0.426 1.1 Traditional 0.419 0.452 0.411 0.410 0.440 0.447 0.443 0.412 0.532 0.441 0.452 0.2 Commercial 0.452 0.407 0.403 0.421 0.392 0.384 0.467 0.468 0.398 0.421 0.446 1.5 Livestock 0.693 0.663 0.726 0.779 0.713 0.746 0.729 0.789 0.842 0.744 0.753 2.1 Food: Value 0.476 0.479 0.430 0.375 0.387 0.406 0.479 0.370 0.449 0.427 0.453 − 1.2 Food: Macronutrients Calories 0.304 0.358 0.295 0.297 0.301 0.295 0.336 0.281 0.269 0.304 0.313 − 1.6 Proteins 0.373 0.428 0.405 0.358 0.344 0.364 0.419 0.334 0.342 0.374 0.383 − 1.7 Fats 0.477 0.475 0.487 0.445 0.449 0.396 0.513 0.437 0.429 0.457 0.463 − 1.1 Carbohydrates 0.304 0.357 0.285 0.302 0.297 0.303 0.312 0.278 0.258 0.299 0.310 − 2.0 18 R. Godoy et al. World Development 176 (2024) 106518 Appendix F. . Part 2: Coefficient of variation (CV) of economic inequalities Baseline Yearly Grand Outcomes: 2002 2003 2004 2005 2006 2007 2008 2009 2010 Mean CV %Δ year Horticulture: Forest 0.890 1.367 1.211 1.496 1.291 1.366 1.549 1.292 1.142 1.291 1.243 2.0 Plots 1.592 1.415 1.348 1.477 1.425 1.448 1.413 1.415 1.333 1.429 1.409 − 1.1 Plantain 0.889 0.942 0.481 0.934 0.802 0.771 0.875 0.619 0.896 0.803 0.711 − 0.4 Manioc 0.416 0.785 0.507 0.561 0.611 0.633 0.656 0.215 0.484 0.541 0.398 − 4.4 Economic: Income 0.305 0.787 0.583 0.833 0.912 0.497 0.616 0.616 0.722 0.656 0.569 3.8 Barter 0.616 0.565 0.558 0.530 0.725 0.504 0.476 0.514 0.502 0.555 0.542 − 2.5 Wealth: Total 1.230 1.301 1.175 1.187 1.332 1.186 1.107 1.013 1.042 1.175 1.088 − 2.6 Traditional 1.191 1.075 1.246 1.235 1.162 1.088 1.046 1.197 0.905 1.126 1.084 − 2.1 Commercial 1.087 1.325 1.316 1.332 1.410 1.394 0.975 1.151 1.349 1.262 1.080 − 0.2 Livestock 0.462 0.496 0.388 0.333 0.400 0.333 0.386 0.293 0.246 0.369 0.332 − 6.9 Food: Value 0.867 0.934 1.124 1.411 1.365 1.199 0.888 1.420 0.974 1.134 0.950 1.8 Food: Macronutrients Calories 1.660 1.159 1.722 1.606 1.494 1.838 1.271 1.720 1.837 1.593 1.443 1.9 Proteins 1.181 0.973 1.150 1.426 1.464 1.312 1.045 1.560 1.367 1.280 1.157 2.9 Fats 0.876 0.868 0.893 1.115 0.761 1.227 0.443 1.095 1.007 0.919 0.768 − 0.1 Carbohydrates 1.730 1.196 1.734 1.537 1.629 1.766 1.609 1.607 1.955 1.646 1.495 2.3 Note. The grand Gini or the grand coefficient of variation is computed by combining all household observations for an outcome across all years. Yearly growth rates for Gini coefficients of Appendix F, Part 2, are in the next table (Appendix F, Part 3). Appendix F. . Part 3: Yearly micro-level growth rates of Gini coefficients of economic outcomes (last column of table in Appendix F. Part 1)(N ¼ 9) [1] [2] [3] [4] [5] [6] [7] Forest Plots Plantain Manioc Income Barter Total Year − 0.005 0.005 0.017 0.016 0.004 0.013 0.011 (0.013) (0.006) (0.011) (0.009) (0.009) (0.005) (0.006) Constant 9.208 − 11.485 − 35.373 –32.682 − 8.630 − 27.167** –23.189* (26.386) (11.382) (22.754) (17.557) (17.569) (10.395) (11.641) R-squared 0.021 0.109 0.251 0.327 0.030 0.489 0.344 [8] [9] [10] [11] [12] [13] [14] Commercial Traditional Animals Food Calories Proteins Fat Year 0.002 0.015 0.021 − 0.012 − 0.016 − 0.017 − 0.011 (0.011) (0.010) (0.006) (0.014) (0.010) (0.011) (0.010) Constant − 4.532 − 31.226 − 42.838*** 22.705 30.978 32.661 22.182 (21.124) (19.403) (11.966) (27.892) (20.720) (22.080) (19.644) R-squared 0.004 0.260 0.644 0.092 0.256 0.249 0.163 [15] Carbs Year − 0.020 (0.010) Constant 39.656* (19.148) R-squared 0.394 Note: P values for the Year coefficient were adjusted for false discovery rates. Carbs = carbohydrates. Appendix G. . Identifying latent variables Chronbach’s alphas were calculated for all or some of the variables in Table 1 to decide if they reflected one or more latent variables and if a principal component factor was warranted. We used principal component factor with varimax rotation to create a score for the variables that seemed to justify it: all socioeconomic variables and all macronutrients. Variables included: Alpha [I]. Socioeconomic and health variables [1] All. Forest, plots, plantains, manioc, income, barter, traditional wealth, commercial wealth, livestock, gifts, chicha, food value, bedridden, and BMI /a/ 0.63 [2] Horticulture. [a] Without food. Forest, plots, plantains, manioc 0.65 [b] With food. Forest, plots, plantains, manioc, food value 0.64 [3] Wealth. Traditional wealth, commercial wealth, livestock 0.53 [4] Economic flows. Income, barter, food 0.26 [5] Sociality. Gifts, chicha 0.18 [6] Health. Bedridden, BMI 0.15 [II]. Macronutrients and calories [7] Macronutrients. Calories, proteins, fat, carbohydrates 0.90 19 R. Godoy et al. World Development 176 (2024) 106518 Note: /a/ Total wealth and autarky were excluded because they were created by combining variables in Table 2; for example, total wealth is the sum of the three types of wealth. References Diener, E., Lucas, R., & Oishi, S. (2018). Advances and open questions in the science of subjective well-ebing. Collabra. Psychology, 4(1)15. Donders, I., & Barriocanal, C. (2020). The influence of markets on the nutrition transition Alao, R., Nur, H., Fivian, E., Shankar, B., Kadiyala, S., & Harris-Fry, H. (2021). Economic of hunter-gatherers: Lessons from the Western Amazon. International Journal of inequality in malnutrition: A global systematic review and meta analysis. BMJ Global Environmental Research and Public Health, 17(17), 6307. https://doi.org/10.3390/ Health, 6, e006906. ijerph17176307 Alkire, S. (2002). Dimensions of human development. World Development, 30(2), Dressler, W. (2020). Cultural consensus and cultural consonance: Advancing a cognitive 181–205. theory of culture. Field Methods, 32(4), 383–398. Alkire, S., Foster, J., Seth, S., Santos, M., Roche, J., & Ballón, P. (2015). Multidimensional Du, H., King, R., & Chi, P. (2019). Income inequality is detrimental to long-term well- poverty measurement and analysis. New York: Oxford University Press. being: A large-scale longitudinal investigation in China. Social Science & Medicine, Alkire, S., Kanagaratnam, U., Nogales, R., & Nicolai, S. (2022). Revising the global 232, 120–128. multidimensional poverty index: Empirical insights and robustness. Review of Income Durkheim, E. (2005 [orig. 1897]). Suicide: A study in sociology (J. Spaulding & G. and Wealth, 68(S2), 5347–5384. Simpson, Trans.). London: Routledge. Aswani, S., Lemahieu, A., & Sauer, W. (2018). Global trends of local ecological Ellis, R. (1996). A taste of movement: An exploration of the social ethics of the Tsimanes of knowledge and future implications. Plos One, 13(4), e0195440. lowland Bolivia. U of Saint Andrews. PhD. Atkinson, A. (1970). On the measurement of inequality. Journal of Economic Theory, 2, Fernández-Llamazares, A., Díaz-Reviriego, I., & Reyes-García, V. (2017). Defaunation 244–263. through the eyes of the Tsimane’. In V. Reyes-García, & A. Pyhälä (Eds.), Hunter- Bauchet, J., Undurraga, E., Zhang, R., & Godoy, R. (2020). Markets and indigenous gatherers in a changing world (pp. 77–90). New York: Springer. peoples: Testing hypotheses from the Western intellectual tradition. Journal of Fischer, E. (2014). The good life: Aspiration, dignity, and the anthropology of wellbeing. Anthropological Research, 76(2), 185–208. Stanford, California: Stanford University Press. Bauchet, J., Undurraga, E., Zycherman, A., Behrman, J. R., Leonard, W. R., & Godoy, R. Foster, Z., Byron, E., Reyes-García, V., Huanca, T., Vadez, V., Apaza, L., … Leonard, W. R. (2021). The effect of gender targeting of food transfers on child nutritional status: (2005). Physical Growth and Nutritional Status of Tsimane’ Amerindian Children of Experimental evidence from the Bolivian Amazon. Journal of Development Lowland Bolivia. American Journal of Physical Anthropology, 126, 343–351. Effectiveness, 13(1), 276–291. Gehring, K. (2013). Who benefits from economic freedom? Unraveling the effect of Bell, W., Lividini, K., & Masters, W. A. (2021). Global dietary convergence from 1970 to economic freedom on subjective well-being. World Development, 50, 74–90. 2010 altered inequality in agriculture, nutrition and health. Nature Food, 2(3), Godoy, R., Byron, E., Reyes-García, V., Vadez, V., Leonard, W., Apaza, L., … Wilkie, D. 156–165. (2005). Income inequality and adult nutritional status: Anthropometric evidence Bendel, R., Higgins, S., Teberg, J., & Pyke, D. (1989). Comparison of skewness from a pre-industrial society in the Bolivian Amazon. Social Science & Medicine, 61 coefficient, coefficient of variation, and Gini coefficient as inequality measures (5), 907–919. within populations. Oecologia, 78(3), 394–400. Godoy, R., Gurven, M., Byron, E., Reyes-García, V., Keough, J., Vadez, V., … Pérez, E. Benjamin, D., Cooper, K., Heffetz, O., & Kimball, M. (2020). Self-reported wellbeing (2004). Do markets worsen economic inequalities? Kuznets in the bush. Human indicators are a valuable complement to traditional economic indicators but are not Ecology, 32(3), 339–364. yet ready to compete with them. Behavioral Public Policy, 4(2), 198–209. https://doi. Godoy, R., Karlan, D., & Zinman, J. (2021). Randomization for causality, ethnography for org/10.1017/bpp.2019.43 mechanisms: Illiquid savings for liquor in an autarkic society. Retrieved from NBER Bethancourt, H., Leonard, W., Tanner, S., Schultz, A., & Rosinger, A. (2019). Working Paper Series, 29566 http://www.nber.org/papers/w29566. Longitudinal changes in measures of body fat and diet among adult Tsmiane’ Godoy, R., Zeinalova, E., Reyes-García, V., Huanca, T., Kosiewicz, H., Leonard, W., & forager-horticulturist of Bolivia, 2002–2010. Obesity, 1–13. https://doi.org/ Tanner, S. (2010). Does civilization cause discontentment among indigenous 10.1002/oby.22556 Amazonians? Test of empirical data from the Tsimane’ of Bolivia. Journal of Bhui, R., Chudek, M., & Henrich, J. (2019). Work time and market integration in the Economic Psychology, 31(4), 587–598. https://doi.org/10.1016/j.joep.2010.04.001 original affluent society. Proceedings of the National Academy of Sciences, 116(44), Godoy, R. A., Leonard, W. R., Reyes-Garcia, V., Goodman, E., McDade, T. W., Huanca, T., 22100–22105. … Vadez, V. (2006). Physical stature of adult Tsimane’ Amerindians, Bolivian Blázquez-Cuesta, M., & Budría, S. (2012). Deprivation and subjective well-being: Amazon in the 20th century. Economics & Human Biology, 4(2), 184–205. https:// Evidence from panel data. Review of Income and Wealth, 60(4), 655–682. doi.org/10.1016/j.ehb.2005.11.001 Borgerhoff Mulder, M., Bowles, S., Hertz, T., Bell, A., Beise, J., Clark, G., … Wiessner, P. Graeber, D., & Wengrow, D. (2021). The dawn of everything: A new history of humanity. (2009). Intergenerational Wealth Transmission and the Dynamics of Inequality in New York: Farrar, Strauss and Giroux. Small-Scale Societies. Science, 326(5953), 682–688. Retrieved from :// Gravlee, C. C., Kennedy, D., Godoy, R., & Leonard, W. (2009). Methods for collecting 000271233200026. panel data. What can cultural anthropology learn from other disciplines? Journal of Bowles, S., & Carlin, W. (2020). Inequality as experienced differences: A reformulation of Anthropological Research, 65(3), 453–483. the Gini coefficient. Economic Letters, 186(108789). Gurven, M., Jaeggi, A., von Rueden, C., Hooper, P., & Kaplan, H. (2015). Does market Carla, S. A., & Falco, P. (2018). Does the risk of poverty reduce happiness. Economic integration buffer risk, erode traditional sharing practices and increase inequality? A Development and Cultural Change, 67(1), 1–28. test among Bolivian forager-farmers. Human Ecology, 43(4), 515–530. Chen, J., Kosec, K., & Mueller, V. (2019). Moving to despair? Migration and well-being in Gurven, M., Jaeggi, A. V., Kaplan, H., & Cummings, D. (2013). Physical activity and Pakistan. World Development, 113, 186–203. modernization among Bolivian Amerindians. Plos One, 8, e55679. Cheung, F. (2018). Income distribution predicts greater life satisfaction across individual, Gurven, M., Kaplan, H., & Supa, A. Z. (2007). Mortality experience of Tsimane national, and cultural characteristics. Journal of Personality and Social Psychology, Amerindians of Bolivia: Regional variation and temporal trends. American Journal of 115, 867–882. Human Biology, 19(3), 376–398. https://doi.org/10.1002/ajhb.20600 Clements, K., & Si, J. (2018). Engel’s Law, dietary diversity, and the quality of food Gurven, M., Mulder, M. B., Hooper, P. L., Kaplan, H., Quinlan, R., Sear, R., … Bell, A. consumption. American Journal of Agricultural Economics, 100(1), 1–22. (2010). Domestication Alone Does Not Lead to Inequality Intergenerational Wealth Cowell, F., Karagiannaki, E., & McKnight, A. (2018). Accounting for cross-country Transmission among Horticulturalists. Current Anthropology, 51(1), 49–64. https:// differences in wealth inequality. Review of Income and Wealth, 64(2), 322–356. doi.org/10.1086/648587 Das, K., Jones-Harrell, C., Fan, Y., Ramaswami, A., Orlove, B., & Botchwey, N. (2020). Gurven, M., Stieglitz, J., Trumble, B., A., B., Beheim, B., Davis, H., . . . Kaplan, H. (2017). Understanding subjective well-being: Perspectives from psychology and public The Tsimane Health and Life History Project: Integrating anthropology and health. Public Health Reviews, 41(25). https://doi.org/10.1186/s40985-020-00142-5 biomedicine. Evolutionary Anthropology, 26(2), 54-73. Davies, J., Lluberas, R., & Shorrocks, A. (2017). Estimating the level and distribution of Hendler, R., Ramchandani, V., Gilman, J., & Hommer, D. (2013). Stimulant and sedative global wealth, 2000–2014. Review of Income and Wealth, 63(4), 731–1359. https:// effects of alcohol. Current Topics in Behavioral Neuroscience, 13, 489–509. https://doi. doi.org/10.1111/roiw.12318 org/10.1007/7854_2011_135. PMID: 21560041. Deaton, A. (1985). Panel data from time series of cross-sections. Journal of Economics, 30, Henrich, J. (2022). The weirdest people in the world. New York: Farrar, Straus & Giroux. 109–126. Henrich, J., Ensminger, J., McElreath, R., & Barrat, A. (2010). Markets, religion, Decancq, K., Decoster, A., & Schokkaert, E. (2009). The evolution of world inequality in community size, and the evolution of fairness and punishment. Science, 327, well-being. World Development, 37(1), 11–25. 1480–1484. DeLong, J. (2022). Slouching towards utopia: An economic history of the twentieth century. Hruschka, D., Hadley, C., & Hackman, J. (2017). Material wealth in 3D: Mapping New York: Basic Books. multiple paths to prosperity in low- and middle-income countries. Plos One, DeNeve, J., & Sachs, J. (2020). The SDGs and human well-being: A global analysis of e0184616. synergies, trade-offs, and regional differences. Scientific Reports Natureresearch, 10, Huanca, T. (2006). Tsimane’ oral tradition. Landscape and identity in tropical forest. La Paz, 15113. https://doi.org/10.1038/s41598-020-71916-9 Bolivia: Wa-Gui. Diener, E., & Chan, M. (2011). Happy people live longer: Subjective well-being Hume, D. (1793). Essays and treatises on several subjects (Vol. 1). Basil: J.J. Tourneisen. contributes to health and longevity. Applied Psychology: Health and well-being, 2(1), INE. (2015). Censo de población y vivienda 2012, Bolivia. Características de la población. La 1–43. Paz, Bolivia: Instituto Nacional de Estadísticas INE. Diener, E., Oishi, S., & Tay, L. (2018). Advances in subjective well-being research. Nature Jaeggi, A., Blackwell, A., von Rueden, C., Trumble, B. C., Stieglitz, J., Garcia, A., … Human Behavior, 2(4), 253–260. https://doi.org/10.1038/s41562-018-0307-6 Gurven, M. (2021). Do wealth and inequality associate with health in a small-scale subsitence society?, eLife, 10:e59437. https://doi.org/10.7554/eLife.59437 20 R. Godoy et al. World Development 176 (2024) 106518 Jerven, M. (2011). Users and producers of African income: Measuring the progress of Oman, S. (2021). Understanding well-being data. Improving social and cultural policy, African economies. African Affairs, 110(439), 169–190. practice and research. Switzerland AG: Springer Nature, Palgrave Macmillan. Jones, C., & Klenow, P. (2016). Beyond GDP? Welfare across countries and time. Page-Reeves, P. (Ed.). (2022). Well-being as a multidimensional concept. Understanding American Economic Review, 106(9), 2426–2457. connections among culture, community, and health. Lanham, Maryland: Lexington Jones, D., Molitor, D., & Reif, J. (2019). What do workplace wellness programs do? Books. Evidence from the Illinois Workplace Wellness Study. Quarterly Journal of Economics, Pauly, A. (1928). Ensayo de etnografía americana. Viajes y exploraciones. Buenos Aires: 134(4), 1747–1791. Talleres s.a. casa Jacobo Peuser, ltda. Kahneman, D., Diener, E., & Schwarz, N. (Eds.). (1999). Well-being: The foundations of Peterson, R. (1994). A meta-analysis of Chronbach’s coefficient alpha. Journal of hedonic psychology. New York: Russell Sage Foundation. Consumer Research, 21(2), 381–391. Kaplan, H., Thomson, R., Trumble, B., Wann, S., Allam, A., Beheim, B., & Frohlich, B. Pickett, K., & Wilkinson, R. (2015). Income inequality and health: A causal review. Social (2017). Coronary atherosclerosis in indigenous South American Tsimane’: A cross- Science & Medicine, 128, 316–326. sectional cohort study. Lancet, 389(10080), 1730–1739. Piketty, T. (2014). Capital in the twenty-first century. Cambridge, MA: Harvard University Kaufman, M., Guest, A., & Mmbaga, B. (2022). What the World Happiness Report doesn’t Press. see: The sociocultural contours of wellbeing in northern Tanzania. International Piketty, T. (2022). A brief history of equality. Cambridge, MA: Harvard University Press. Journal of Wellbeing, 12(4), 27–50. Reyes-García, V., Angelsen, A., Shively, G., & Minkin, D. (2019). Does income inequality Kaus, W. (2013). Beyond Engel’s Law - A Cross-country analysi. The Journal of Socio- influence subjective well-being? Evidence from 21 developing countries. Journal of Economics, 47, 118–134. Happiness Studies, 20, 1197–1215. Keister, L., & Lee, H. (2017). The double one percent: Identifying an elite and a super- Reyes-García, V., Gallois, S., Pyhälä, A., Díaz-Reviriego, I., Fernández-Llamazares, A., elite using the joint distribution of income and net worth. Research in Social Galbraith, E., … Napitupulu, L. (2021). Happy just because. A cross-cultural study on Stratification and Mobility, 50, 1–12. subjective wellbeing in three Indigenous societies. Plos One, 16(5), e0251551, Kempf, D., & Kempf, E. (2017). Journey to Chimane land. Maitland, Florida: Xulon Press. 0252021. https://doi.org/10.1371/journal.pone.0251551. Retrieved from. Kharkwal, S., & Malhotra, R. (2020). Assessment of the food security status of households Reyes-García, V., Gravlee, C. C., McDade, T., Huanca, T., Leonard, W., & Tanner, S. belonging to different regions of Himalayan belt. Economic Affairs (India), 65(4), (2010). Cultural consonance and body morphology. Estimates with longitudinal data 611–617. https://doi.org/10.46852/0424-2513.4.2020.17 from an Amazonian society. American Journal of Physical Anthropology, 143(2), Killewald, A., Pfeffer, F., & Schachner, J. (2017). Wealth inequality and accumulation. 167–174. Annual Review of Sociology, 43, 379–404. Reyes-García, V., Ledezma, J., Paneque-Gálvez, J., Orta-Martínez, M., Gueze, M., Lobo, Kraft, T., Stieglitz, J., Trumble, B., Martin, M., Kaplan, H., & Gurven, M. (2018). A., . . . Huanca, T. (2012). Presence and purpose of non-indigenous peoples on Nutrition transition in 2 lowland Bolivian subsistence populations. American Journal indigenous lands. A descriptive account from the Bolivian lowlands. Society and of Clinical Nutrition, 108, 1183–1195. Natural Resources, 25(3), 270-284. Krueger, A., & Schkade, D. (2008). The reliability of subjective well-being measures. Reyes-García, V., Tofighi-Niaki, A., Austin, B., Benyei, P., Danielsen, A., Fernández- Journal of Public Economics, 92(8–9), 1833–1845. Llamazares, A., … Tengo, M. (2022). Data sovereignty in community-based Krueger, A., & Stone, A. (2014). Measuring subjective well-being: Progress and environmental monitoring: Toward equitable environmental data governance. challenges. Science, 346(6205), 42–43. https://doi.org/10.1126/science.1256392 BioScience, 72(8), 714–717. Lance, C., Butts, M., & Michels, L. (2006). The sources of four commonly reported cutoff Robeyns, I. (2005). The capability approach: A theoretical survey. Journal of Human criteria: What did they really say? Organizational Research Methods, 9(2), 202–220. Development, 6(1), 93–114. Lee, M. T., Kubzansky, L. D., & VanderWeele, T. J. (Eds.). (2017). Measuring well-being. Rosinger, A., & Bethancourt, H. (2019). Chicha as water. Traditional fermented beer Interdisciplinary perspectives from the social sciences and the humanities. New York: consumption among forager-horticulturist in the Bolivian Amazon. In K. Hockings, & Oxford University Press. R. Dunbar (Eds.), Alcohol and humans (pp. 147–162). Oxford, UK: Oxford University Leonard, W., Reyes-García, V., Tanner, S., Rosinger, A., Schultz, A., Vadez, V., … Press. Godoy, R. (2015). The Tsimane’ Amazonian Panel Study (TAPS): Nine years Rosinger, A., Tanner, S., Leonard, W., & Taps.. (2013). Precursors to overnutrition: The (2002–2010) of annual data available to the public. Economics & Human Biology, 19, effects of household market food expenditures on measures of body composition 51–61. among Tsimane’ adults in lowland Bolivia. Social Science & Medicine, 92, 53–60. Liu, M., Fang, C., Bai, Y., Sun, B., Liao, X., & Liu, Z. (2023). Regional inequality and Saez, E., & Zuckman, G. (2014). Wealth inequality in the United States since 1913: urban-rural differences of dietary water footprint in China. Resources, Conservation & Evidence from capitalized income tax data. NBER Working paper 20625. In N. B. o. Recycling, 199, e107236. E. Research (Ed.), National Bureau of Economic Research (Vol. 20625). Cambridge, Lohman, T. G., Roche, A. F., & Martorell, R. (1988). Anthropometric Standardization MA: National Bureau of Economic Research. Reference Manual (Vol. 10). Champaign, Illinois: Human Kinetics Publishers. Sahlins, M. (1972). Stone Age economics. New York: Aldine de Gruyter. Lomas, T. (2019). Happiness. A cross-cultural lexicon on well-being. Cambridge, MA: The Sassi, M., Trital, G., & Bhattacharjee, P. (2022). Beyond the annual and aggregate MIT Press. measurement of household inequality: The case study of Lake Naivasha basin, Marlier, E., & Atkinson, A. (2010). Indicators of poverty and social exclusion in a global Kenya. The European Journal of Development Research, 34(1), 387–408. https://doi. context. Journal of Policy Analysis and Management, 29(2), 285–304. org/10.1057/s41287-021-00380-6 Marx, K. (1920 [orig. 1867]). Capital. A critique of political economy. Chicago: Charles H. Seale, J., & Regmi, A. (2006). Modelling international consumption patterns. Review of Kerr & Company. Income and Wealth, 52(4), 603–624. Mathews, G., & Izquierdo, C. (Eds.). (2008). Pursuits of happiness: Well-being in Sen, A. (1999). Development as freedom. New York: Knopf. anthropological perspective. New York: Berghahn Books. Shephard, R. J., & Rode, A. (1996). The health consequences of ’modernization’: Evidence Mattison, S., MacLaren, N., Sum, C., Mattison, P., Liu, R., Shenk, M., … Wander, K. from circumpolar peoples. Cambridge, UK: Cambridge University Press. (2023). Market integration, income inequality, and kinship system among the Mosuo Sim, J., & Waterfield, J. (2019). Focus group methodology: Some ethical challenges. of China. Evolutionary Human Sciences, 5, e4. Quality & Quantity, 53, 3003–3022. https://doi.org/10.1007/s11135-019-00914-5 McDade, T., Reyes-García, V., Blackinton, P., Tanner, S., Huanca, T., & Leonard, W. Singh, A., Kumar, K., & Singh, A. (2016). Trends in inequality in food consumption and (2007). Ethnobotanical knowledge is associated with indices of child health in the calorie intake in India: Evidence from the last three decades, 1983–2012. Social Bolivian Amazon. Proceedings of the National Academy of Sciences, 104(15), Indicators Research, 128(3), 1319–1346. 6134–6139. Smith, A. (1884 [orig. 1776]). An inquiry into the nature and causes of the welth of nations. Minarro, S., Reyes-García, V., Aswani, S., Selim, S., Barrington-Leight, C., & Galbraith, E. London: T.Nelson and Sons. (2021). Happy without money: Minimally monetized societies can exhibit high Stone, A., & Krueger, A. (2018). Understanding subjective well-being. In J. Stiglitz, subjective wellbeing. Plos One, 16(1), e0244569. J. Fitoussi, & M. Durand (Eds.), For good measure: Advancing research on well-being Mulligan, C., Boyer, D., Turner, T., Delson, E., & Leonard, W. (2022). Data sharing in metrics beyond GDP. Paris: OECD Publishing. biological anthropology. Retrieved from Amerian Journal of Biological Anthropology, Tanner, S., Rosinger, A., Leonard, W., McDade, T., Reyes-Garcia, V., & TAPS.. (2013). 1–28 https://doig.org/10.1002/ajpa.24499. Health and adult productivity: The relation between adult nutrition, helminths and Ngamaba, K., Panagioti, M., & Armitage, C. (2018). Income inequality and subjective agricultural, hunting, and fishing yields in the Bolivian Amazon. American Journal of well-being: A systematic review and meta-analysis. Quality of Life Research, 27(3), Human Biology, 25(1), 123–130. 577-596. doi:doi: 10.1007/s11136-017-1719-x. Epub 2017 Oct 24. PMID: Tucker, B., Lill, E., Tsiazonera, Tombo, J., Lahiniriko, R., Rasoanomenjanahary, L., . . . 29067589; PMCID: PMC5845600. Tsikengo, J. (2015). Inequalities beyond the Gini: Subsistence, social structure, Nordenskiöld, E. (1979 [orig. 1924]). The ethnography of South-America seen from the gender, and markets in southwestern Madagascar. Economic Anthropology, 2, 326- Mojos in Bolivia. New York: AMS Press. 342. doi:10.1002/sea2.12034. Nordenskiöld, E. (2001 [orig 1924]). Exploraciones y aventuras en Sudamérica. La Paz: Undp. (2023). United Nations Development Program, 2030 Agenda. Retrieved from Plural Editores. https://www.undp.org/sustainable-development-goals?utm_source=EN&utm_me Nunnally, J., & Bernstein, I. (1994). Psychometric theory (3rd ed.). New York: McGraw- dium=GSR&utm_content=US_UNDP_PaidSearch_Brand_English&utm_campai Hill. gn=CENTRAL&c_src=CENTRAL&c_src2=GSR&gad_source=1&gclid=Cj0KCQi Nyberg, C. (2009). Market integration, stress, and health: An exploration of hypothalamic- AyKurBhD5ARIsALamXaEjJZ56zn9lW8V512kZwWbT3MtMxkC3yDT-k2KPpuJLu pitutiary-adrenal dynamics among the Tsimane’ of the Bolivian Amazon. Evanston, t8GOjqOPGIaAq4ZEALw_wcB. Illinois: Northwestern University. PhD. UNDP 2022 (United Nations Development Program. 2022 Global multidimensional Nyberg, C. (2012). Diurnal cortisol rhythms in Tsimane’ Amazonian foragers: New poverty index (MPI): Unpacking deprivation bundles to reduce multidimensional insights into ecological HPA axis research. Psychoneuroendrocrinology, 37(2), poverty. Retrieved from New York. 178–190. https://doi.org/10.1016/j.psyneuen.2011.06.002 Undurraga, E., Behrman, J. R., Emmett, S., Kidd, C., Leonard, W., Piantadosi, S., … Oishi, S., & Westgate, E. (2022). A psychologically rich life: Beyonnd happiness and Godoy, R. (2018). Child stunting is associated with weaker human capital among meaning. Psychological Review, 129(4), 790–811. 21 R. Godoy et al. World Development 176 (2024) 106518 native Amazonians. American Journal of Human Biology, 30(1). https://doi.org/ WHO. (1995). Physical status: The use and interpretation of anthropometry. Report of a 10.1002/ajhb.23059 WHO Expert Committee. In WHO Technical Report Series (Vol. 854). Geneva: World Undurraga, E., Nica, V., Zhang, R., Mensah, I., & Godoy, R. (2016). Individual health and Health Organization. the visibility of village economic inequality: Longitudinal evidence from native WHO. (2000). Obesity: Preventing and managing the global epidemic. In WHO Technical Amazonins in Bolivia. Economics and Human Biology, 23, 18–26. Report Series (Vol. 894). Geneva: World Health Organization. Undurraga, E. A., Nyberg, C., Eisenberg, D. T. A., Magvanjav, O., Reyes-Garcia, V., , … Winther, C., Dinesen, S., Sondergaard, S., & Bech, P. (2015). The WHO-5 well-being Huanca, T., & TAPS Bolivia Study Team. (2010). Individual Wealth Rank, index: A systematic review of the literature. Psychotherapy and Psychosomatics, 84, Community Wealth Inequality, and Self-Reported Adult Poor Health: A Test of 167–176. Hypotheses with Panel Data (2002-2006) from Native Amazonians, Bolivia. Medical Woolard-Mayfour, K., & Hruschka, D. (2022). Assessing comparative asset-based Anthropology Quarterly, 24(4), 522–548. measures of material wealth as predictors of physical growth and mortality. Social Vadez, V., Reyes-Garcia, V., Godoy, R. A., Williams, L., Apaza, L., Byron, E., … Wilkie, D. Science & Medicine - Population Health, 17(101065). (2003). Validity of Self-Reports to Measure Deforestation: Evidence from the World-Bank. (2022a). Poverty Data. Retrieved from https://povertydata.worldbank. Bolivian Lowlands. Field Methods, 15(3), 289–304. org/poverty/country/BOL. Vadez, V., Reyes-Garcia, V., Huanca, T., & Leonard, W. R. (2008). Cash Cropping, Farm World-Bank. (2022b). World development indicators. Retrieved from https://datatopics. Technologies, and Deforestation: What are the Connections? A Model with Empirical worldbank.org/world-development-indicators/themes/poverty-and-inequality.htm. Data from the Bolivian Amazon. Human Organization, 67(4), 384–396. Retrieved Zaninotto, P., & Steptoe, A. (2019). Association between subjective well-being and living from ://WOS:000261218600004. longer without disability or illness. JAMA Netw Open, 3(7), e196870. Veblen, T. (1899). The theory of the leisure class: An economic study of institutions. New Zhang, R., Undurraga, E., Zeng, W., Reyes-García, V., Tanner, S., Leonard, W., … York: The Macmillan Company. Godoy, R. (2016). Catch-up growth and growth deficits: Nine-year annual panel Venkataraman, S. (2013). Subjective well-being: A primer for poverty analysts. Journal of child growth for native Amazonians in Bolivia. Annals of Human Biology, 43(4), Poverty and Social Justice, 21(1), 5–17. https://doi.org/10.1332/ 304–315. 175982713X664029 Zhang, Z., & Hong, M. (2023). Research on the heterogeneous effects of resident’s Waldinger, R., & Schultz, M. (2023). The goodlife: Lessons from the world’s longest scientific income on mental health. Mathematical Biosciences and Engineering, 30(3), study of happiness. New York: Simon & Schuster. 5043–5065. Weitz, C., Friedlaender, F., & Friedlaender, J. (2014). Adult lipids associated with early Zimmerman, C. (1932). Ernst Engel’s Law of expenditures for food. Quarterly Journal of life growth in traditional Melanesian societies undergoing rapid modernization: A Economics, 47(1), 78–101. longitudinal study of the mid-20th century. American Journal of Physical Zycherman, A. (2013). The changing value of food: Localizing modernity among the Tsimane’ Anthropology, 153, 551–558. Indians of lowland Bolivia. (PhD). Teacher’s College, Columbia University, New York. UNDP (United Nations Development Program) 2021.2021 Global multidimensional Zycherman, A. (2015). Shockdye’ as world: Localizing modernity through beer in the poverty index (MPI): Unmasking disparities by ethnicity, caste and gender. New Bolivian Amazon. Food, Culture & Society, 18(1), 51–69. York. 22