The Drivers of Market Integration Among Indigenous

SOCIITT & NATURAL RESOURCES https://dol.org/10.1080/08941920.2017.1331487

[ 11) Check for updates j

The Drivers of Market Integration Among Indigenous Peoples: Evidence From the Ecuadorian Amazon Cristian Vasco•, Grace Tamayob, and Verena Griess' •facultad de Ciencias Agricolas, Universidad Central del Ecuador, Quito, Ecuador; blnstituto de Altos Estudios Nacionales, Quito, Ecuador; 'Department of Forest Resources Management, University of British Columbia, Vancouver, British Columbia, Canada

ABSTRACT

ARTICLE HISTORY

Knowledge of the driving forces behind indigenous participation in the market is essential for practitioners intending to integrate conservation and development policies in indigenous territories. Nevertheless, empirical research on the determinants of market integration among indigenous peoples is still scarce. This article uses household survey data and multivariate techniques to examine the drivers of market integration among indigenous groups in the Ecuadorian Amazon. We use multiple measures of market integration, including the sale of crops, timber, and wildlife; the use of credit; and participation in wage labor. The resu lts show that the way in which indigenous peoples integrate into the market depends on their endowments of human, financia l, and physical capital. More educated households are able to engage in commercial agriculture and nonagricultural wage work, whereas uneducated poor households in communities in conflict with outsiders are pushed to engage in poorly paid agricultural wage work and (often illegal) t imber operations.

Received 4 August 2016 Revised 29 December 2016 Accepted 25 March 2017 KEYWORDS

Amazon; indigenous peoples; lowland Kichwa; market integrat ion; Shuar

Introduction

Given their great biological diversity, tropical forests of lowland Latin America are considered of great importance for conservation. Nevertheless, human activities (e.g., agriculture and exploitation of forest resources) threaten to eliminate some of the world's most biologically diverse forests (Food and Agriculture Organization [FAO) 2005). Most of the remaining forests in Latin America are controlled by indigenous peoples who greatly depend on forest resources for a number of subsistence and market-oriented livelihood activities (Lu, Bilsborrow, and Ona 2012). While it is commonly believed that indigenous peoples use sustainable forest management practices (i.e., long fallow periods, collection of fore.st resources at sustainable levels) (Nuckolls 2010), factors such as rapid population growth and the accelerated integration of indigenous peoples into the market economy threaten not only the rich biodiversity and stability of tropical forests, but also the well-being of those who depend on forest resources for their livelihoods (Zimmerman et al. 2001).

e

CONTACT Cristian Vasco [email protected] ~ Facultad de Ciencias Agricolas, Universidad Central del Ecuador, Gato Sobral y Jeronimo Leyton (Ciudadela Universitaria}, Quito 170521, Ecuador. Color versions of one or more of the figures in the article can be found online at www.tandfon line.com/usnr. © 2017 Taylor & Francis

2

@

C. VASCO ET AL.

A significant body of research has analyzed the impact of market integration of indigenous peoples on deforestation (Godoy et al. 1997; Godoy 2001; Gray et al. 2008), wildlife hunting (Godoy et al. 2010; Vasco and Siren 2016), agricultural patterns (Godoy, Franks, and Alvarado Claudio 1998; Gray et al. 2008), health and nutritional status (Godoy et al. 2005), and reciprocity traditions (Godoy 2001). Although these works have provided important insights on the effects of the market on indigenous peoples' well-being and use of forest resources, the question of why and how indigenous peoples participate in the market economy has received far less attention (Lu 2007). Designing policies to mitigate the (negative) effects of the market on indigenous livelihoods and the environment requires knowledge of the factors driving indigenous peoples into the market economy in the first place. A central part in the debate on the effects of market on indigenous peoples is the definition of market integration. Lu (2007, 594) defines integration into the market as "the commodification of labor, capital, land, and goods and services." Market integration then includes both what is produced for the market and what is consumed from the market. In practical terms, most quantitative studies define market integration as cash income from off-farm work, the sale of agricultural produce, the sale of forest products, and the use of credit (Godoy 2001; Godoy et al. 1997; Godoy et al. 2005; Vasco and Siren 2016). This article analyzes the determinants of integration into the market economy among the Kichwa and Shuar peoples in the province of Pastaza, in the central Ecuadorian Amazon, one of the world's biodiversity hotspots (Myers et al. 2000). The rest of this article is structured as follows: The following section presents the theoretical framework. The subsequent section describes the socioeconomic and cultural context of the Kichwa and the Shuar in the Ecuadorian Amazon and the Pastaza province. Next, the data collection process is described and the statistical methods are defined. The following section presents and discusses the results, while the final section concludes. Theoretical Framework

The livelihood model (Ellis 2000) offers a solid theoretical foundation to analyze market involvement of indigenous peoples. The model posits that livelihood decisions, in this case integration into the market economy, are shaped by different endowments of natural capital (land, water, forest resources), social capital (interpersonal networks, membership in groups and associations), human capital (education, skills, health), physical capital (irrigation canals, implements, tools, roads), and financial capital (cash, savings, and cattle). The type and amount of each capital and the way in which they are combined determine a household's livelihood strategies. To illustrate, households with high endowments of human capital (education) and low endowments of natural capital (land) may decide to engage in off-farm employment as a way of both coping with land scarcity and obtaining higher returns to education (Vasco Perez, Bilsborrow, and Torres 2015). On the other hand, indigenous peoples with abundance of natural capital (land, forest and game) may prefer staying outside the market as they do not perceive resource scarcity and so see no need to engage in the market (Lu 2001). Similarly, households with higher endowments of social capital, in the form relationships and networks outside the community, are more likely to take part in the wage labor market (Vanwey and Vithayathil 2013).

SOCIETY & NATURAL RESOURCES

@

3

As with most livelihood decisions in the developing world, integration of indigenous peoples into the market appears to be shaped by both pull and push factors (Godoy et al. 2005). The first term, pull, refers to the voluntarily decision to join the market economy when it is seen to offer better opportunities to improve welfare. For instance, Godoy et al. (2005) propose that indigenous peoples' participation in the market is mostly motivated by food security (to increase the levels of food consumption and to reduce the effects of seasonality). Others (Reardon et al. 2006) argue that market integration, in the form of participation in the wage labor market, is driven by a cost-benefit analysis in which the returns to working off-farm are higher than those to be obtained otherwise. This is especially likely in areas where agricultural, mining, and tourist sectors make local economies dynamic. An example of this is the employment of the Kichwa and the Waorani peoples by oil companies in the Ecuadorian Amazon (Lu 2001; Gray et al. 2008). Push factors, on the other hand, refer to events that threaten indigenous people's livelihoods. These include population pressure, resource depletion, encroachment by outsiders, and climatic risks (Lu 2001; Godoy et al. 2005). Such threats dramatically reduce resource availability and put indigenous peoples in a position where exchanging goods and labor in the market is one of the few (or only) ways to secure their livelihood. For instance, the Shuar from the southern Ecuadorian Amazon migrated to urban areas to engage in wage labor as a way to cope with population pressure in their home communities (Rudel, Bates, and Machinguiashi 2002). We finish this section with a comment on the limitations of the livelihood model. A main criticism to this approach is that it uses the household as the unit for empirical analysis, assuming that incomes and preferences are shared among all the household members (de Haan 2012). Hence, it may fail to address the effect of individual decision making, and power and gender relations in the social unit (Ellis 1998). On the other hand, some argue that in most rural areas of the developing world the household is the basic locus of decision aking, where "particularly intense social and economic interdependencies occur between a group of individuals" (Ellis 1998, 18). Despite the criticism, the livelihood approach continues to be widely used among both researchers and practitioners (de Haan 2012). The Context: The Kichwa and the Shuar in the Ecuadorian Amazon and Pastaza

In the 1960s, continuous flows of poor peasants from the Coast and the Highlands, the other two geographical regions of Ecuador, migrated to the Amazon in search of cheap land. Although this was mostly a process of spontaneous migration, it was also encouraged by the government as a way both to reduce land pressure and to incorporate the "empty lands" of the Amazon in agricultural production. But far from being empty, the Amazon lands were home to several indigenous peoples, who in many cases were displaced from their lands by the flows of mestizo colonists (Whitten 1976). The Kichwa of Pastaza or the Canelos Runa are the traditional inhabitants of the Pastaza province (Nuckolls 2010). They number about 18,000 and control about 1.4 million ha over common-property regimes (Prefectura de Pastaza 2012). Traditionally, the Kichwa have obtained their livelihood from subsistence-swidden agriculture (plantains and cassava, locally known as yuca) and the collection- at sustainable levels-of forest products, including timber, fibers, game, fish, and wild fruits (Nuckolls 2010). However, others report that the

4

@

C. VASCO ET AL.

Kichwa have increasingly integrated into the market economy, and at present also engage in cash-generating activities, including wage labor, community-based ecotourism ventures (Nuckolls 2010), commercial agriculture (Vasco Perez, Bilsborrow, and Torres 2015), timber logging (Vasco et al. 2017), and game trade (Jacome-Negrete et al. 2013). Although population pressure in Kichwa territories is low compared to that of other ethnic groups (e.g., the Shuar) (Vasco and Siren 2016), competition for resources (agricultural land and game) may exist in communities with higher population densities. The Shuar, also known as the Jibaros, are the traditional inhabitants of the southern Ecuadorian Amazon. As distinct from the Kichwa, who were considered by migrant colonists as "half-civilized" and useful for farm work, the Shuar were widely labeled as "savage" and "dangerous" people (Whitten 1976). In contrast to the Kichwa, who mostly responded to colonist encroachment of their territories by relocating deeper in the forest, away from colonization zones, the Shuar counteracted colonization by claiming property over extensive areas of forest in their original territories in the southern Ecuadorian Amazon. In doing so, they replicated the colonist strategy of clearing a portion of forest, planting pastures, and establishing small herds of cattle (Rudel, Bates, and Machinguiashi 2002). They used this strategy not only in their home territories, but also in provinces to the north where some of them migrated partly because of population pressure in Morona Santiago (Bremner and Lu 2006). The Shuar are highly integrated into the market economy (Lu 2007), with their livelihood activities including cattle ranching and cash cropping (Rudel, Bates, and Machinguiashi 2002), wage work (Vasco Perez, Bilsborrow, and Torres 2015), sale of timber and other forest products (Muzo et al. 2013), and internal and international migration (Mcsweeney and Jokisch 2007). The Shuar of Pastaza number about 5,600 individuals and control approximately 400,000 ha under common-property regimes (INEC 2010; Prefectura de Pastaza 2012). Most of the Shuar communities are located along the route south to Morona Santiago, the province from which they migrated during the 1970s and 1980s. Population densities are higher in Shuar communities (Vasco and Siren 2016), which may have implications for pressure on natural resources.

Data and Methods The Survey

Our data come from a household survey conducted in May- October 2013 in the Pastaza province (see Figure 1). The questionnaire was designed to provide information on household demographic characteristics, household assets, land use, natural resources use, agricultural and nonagricultural income, use of credit, and social capital. The survey was administered to the household head, 1 with assistance of the spouse and children, if available. At the same time, a community survey inquiring about population size, infrastructure, and conflicts with other communities and outsiders was completed by community leaders. Households were selected using a controlled sampling method (Kish 1965). In this framework, 13 communities (see Table 1) were selected using criteria of ethnicity, distance to roads and markets, population size and density, availability of services, and availability of alternative income sources (off-farm employment). Variability in these characteristics across communities improves the reliability and the robustness of


0

@

S

Provincial capital

•

Kidlwa communities

o

Shuar communities

- Roads -

Rivers

Figure 1. The study area.

the data (Cavendish 2003). Afterward, households were randomly selected within each community. In total, 213 households (106 Kichwa and 107 Shuar) were surveyed. While this is not strictly a statistically representative sample, the procedure just described insures a good representation of the different livelihood strategies and levels of market integration of indigenous peoples in Pastaza. For example, larger communities near Puyo, the provincial capital, such as Canelos (Kichwa) and Chubitayu (Shuar), are more integrated into the market economy. Since they concentrate rural public offices, health centers, and schools, the availability of off-farm jobs is higher in these kinds of communities. Residents of these communities are also able to commute to Puyo to engage in (mostly "unskilled") nonagricultural work (e.g., bricklayer, domestic servant). That is not the case for smaller communities relatively close to Puyo, such as Pitirishka and Centro Yu (Shuar), where contact with the market is restricted to the sale of cash crops (mainly naranjilla, a tropical citrus fruit). In Chapintsa and Sharupi (Shuar), and in Santa Cecilia de Villano (Kichwa), where dirt roads were opened recently and forest areas still exist, cash income comes mostly from timber logging. In communities farther away from Puyo, such as Shiram Popunas (Shuar) and Killoalpa, Jaime Roldos, and Nuevo San Jose (Kichwa), there is little contact with the market economy and people rely more on subsistence farming and the collection of forest products.

Outcome Variables and Predictors The dependent variables of interest are six measures of market integration, including the likelihood of selling crops, timber, and wildlife; the likelihood of requesting a loan from a private or public bank; and the fraction of household income from wage labor. In order to have a more accurate picture, we divided wage labor into agricultural and nonagricultural employment, as both differ in terms of human capital requirements and earnings (Laszlo 2005). These measures have been used in prior research (Godoy et al. 1997; Godoy 2001; Godoy et al. 2005; Vasco and Siren 2016) as proxies for market integration since they reflect indigenous people's choice to interact with the market economy (Godoy 2001). In the following, we introduce the dependent variables using the livelihood model as the organizing framework.

6

Kichwa

Shuar

Ethnicity

Table 1.

Centro Yu Chapintsa Chubitayu Pitirishka Shiram Popunas Sharupi Canelos Iskayaku Jaime Roldos Killoalpa Nuevo San José Santa Cecilia de Villano Shiwa Kucha

Community 50 420 1125 250 141 94 1200 60 75 75 150 150 310

Population

Communities included in the sample.

15 40 85 49 12 18 75 14 23 11 31 50 18

Households (total) 7 12 42 24 7 8 39 11 13 4 11 20 15

Surveyed households 80 16 98 55 33 60 300 40 500 200 100 300 40

Area (km2) 0.6 26.2 11.4 4.5 4.2 1.5 4.0 1.5 0.15 0.3 0.5 0.5 7.7

Density (inhabitants/km2)

Dirt road Dirt road Paved road Paved road Trail Trail Dirt road Trail River River River Dirt road Dirt road

Accessible by

1.5 2.0 1.0 0.75 6.0 3.0 1.0 3.5 8.0 8.0 8.0 3.5 2.0

Travel time to Puyo (hours)

60 68 62 50 68 81 44 80 180 180 180 70 80

Distance to Puyo (km)


@

7

Predictors include sets of household and community characteristics that are expected both to influence and to be exogenous to market integration decisions. Human capital is controlled for by the household head's age and years of formal education. The numbers of adults and children (see Table 2 for definitions) in the household are included to assess the effects of labor availability on market integration decisions. A dichotomous variable taking the value of 1 provided the household head recognizes him- or herself as Shuar accounts for the effect of ethnicity on market integration. Individuals of the Kichwa ethnic group are taken as the baseline. We use wealth as a proxy of financial capital. It may be argued that this predictor and market integration are endogenously determined due to reverse causality, or, in other words, that it is not possible to determine from cross-sectional data whether increased market integration is a cause rather than a consequence of more wealth. To address this, we use a wealth index that is a construct based upon assets accumulated over several years prior to the survey, instead of a measure of current income. Thus, households' financial capital endowment is controlled for by a wealth index, 2 which is the first principal component of housing characteristics and household assets. A dichotomous variable, taking the value of 1 if the household head trusts other people in the community, controls for the effect of social capital on market involvement. To control for the natural capital endowment, we include the natural logarithms of the land devoted to crops and pastures, the natural logarithm of the total land allocated to a household by the community asamblea (assembly), and the distance from the home to the nearest river. Additionally, the natural logarithm of the distance to the nearest road is used as a proxy of physical capital. We also include three community-level variables to assess the effect of contextual factors 3 on market integration. The population density (inhabitant per square kilometer) is used as a proxy for availability of resources (i.e., land, timber, game, and fish). The natural logarithm of the distance from the community center to Puyo, the provincial capital, accounts for the effect of proximity to markets at contextual level. Finally, a dichotomous variable taking the value of 1 provided the community had any conflict with encroachers/other communities during the 12 months preceding the survey accounts for the effect of encroachment. Multivariate Approach We use probit models to estimate the likelihood for a household to sell crops, to sell timber, to sell wildlife, and to seek credit. This methodology is used for binary response outcome variables (yes or no response) and estimates the probability for a household with particular characteristics to fall into one of the response categories. For the contribution of agriculture and nonagricultural wage labor, our empirical strategy must account for the fact that an important fraction of the households in the sample had no earnings from agricultural and nonagricultural wage labor (69 and 53% of the sample, respectively). Using an ordinary least squares (OLS) approach under such circumstances would involve a high risk of yielding inconsistent estimators. Alternatively, we rely on a Tobit model approach. This methodology is designed to model continuous variables, which take the value of zero for a non-irrelevant part of the sample (Wooldridge 2002). All the models were estimated with robust standard errors clustered at community level.

8

Note. (0/1) identifies dummy variables.

40.5 8.65 2.00 2.85 0.50 0.49 0.192 0.49 5.63 57.34 2.47 15.14

Age of household head (years) Completed years of formal education of head (years) Household members ages �15 years Household members ages >15 years Household head is Kichwa (0/1) Household head is Shuar (0/1) Wealth index Household head trusts other people in community (0/1) Area under crops and pasture (ha) Total area available for use (ha) Distance to the nearest navigable river (km) Distance to the nearest road (km) Inhabitants by km over which the community holds communal rights Conflict with outsiders/other communities for forest Distance from community to Puyo (km)

8.23 0.43 75.27

0.29 0.41 0.10 0.33 0.15 0.32

Household has sold part of the produce in previous 12 months (0/1) Household has sold timber in previous 12 months (0/1) Household has sold game/fish in previous 12 months (0/1) Household has sought a loan in previous 12 months (0/1) Share of total income from agricultural wage labor Share of total income from nonagricultural wage labor

2

Overall

Description

Descriptive statistics, definitions, and mean values for Kichwa and Shuar households.

Variable Dependent variables Crops Timber Game and fish Credit Agricultural wage labor Nonagricultural wage labor Human capital predictor Age Education Children Adults Ethnicity Kichwa Shuar Financial capital Wealth index Social capital Confidence Natural capital Cultivated land Total land Distance to river Physical capital Distance to road Contextual (community-level) predictors Population density Conflict for forest Distance to Puyo

Table 2.

6.35 0.76 88.45

28.39

3.01 68.84 0.09

0.26

–0.698

– –

41.5 7.66 2.50 2.92

0.17 0.33 0.12 0.18 0.17 0.23

Kichwa

11.58 0.10 63.28

2.87

8.55 47.18 4.84

0.71

0.303

– –

39.7 9.61 1.49 2.79

0.40 0.49 0.08 0.47 0.13 0.40

Shuar


@

9

Descriptive Analysis

Table 2 shows the definitions and descriptive statistics of the outcome and explanatory variables. Twice as many Shuar households sell crops as Kichwa households do. Similarly, more Shuar households (49%) sell timber compared to their Kichwa counterparts (33%). There is little variation in the share of households selling game and fish, with an overall average of 10%. Considerably more Shuar (47%) than Kichwa households (18%) requested a loan during the previous 12 months. There is little difference in the share of income from agricultural wage labor between Kichwa and Shuar households, with an average of 15% for both groups. However, the fraction of income from nonagricultural wage labor is nearly twice as high for Shuar as it is for Kichwa households. Concerning human capital predictors, Shuar household heads are younger and better educated than their Kichwa peers. Kichwa households are larger and comparatively have more adults than their Shuar counterparts. This is probably related to the Shuar moving to urban areas in the Amazon, to other provinces, and even abroad as a result of population pressure and crop failure. Although the mean values of the wealth index are difficult to analyze, the negative sign for the Kichwa may reflect that they are poorer in terms of the assets upon which the index was constructed (Kuntashula et al. 2009). Moving on to social capital, although the Shuar appear to be more integrated into the market economy, they have stronger community ties. The share of Shuar household heads (71%) that trust other people in the community is much higher than that of Kichwa (26%) households. In terms of natural capital predictors, the total land area available for use is considerable larger for Kichwa (68 ha) than for Shuar homes (47 ha). However, the agricultural area of the Shuar (8.5 ha) is nearly three times as large as that of the Kichwa (3 ha), which is consistent with prior research showing that the Shuar have larger agricultural areas as they have specialized in cash cropping and cattle ranching (Gray et al. 2008; Lu et al. 2009). Kichwa households are located much closer (0.09 km on average) to a navigable river than their Shuar peers (4.8 km on average). Moving on to physical capital, the Shuar live much closer to roads than their Kichwa equivalents. The mean distance from Kichwa homes to the nearest road is 10 times as long as it is for Shuar households. Concerning contextual predictors, the mean population density in Shuar communities is nearly twice as large as it is in Kichwa settlements. The risk of encroachment appears to be higher in Kichwa than in Shuar communities. Finally, the Kichwa live farther away from Puyo than their Shuar counterparts.

Results and Discussion

The results of the probit and Tobit models are presented in Table 3. The main findings for each outcome variable are discussed here. Before proceeding, however, we provide a comment on the reliability of self-reported values of timber and game. These are sensitive topics in the Ecuadorian Amazon since harvesting timber requires permission from the Ministry of the Environment and hunting is only permitted for self-consumption. When the topic under investigation is either sensitive or illegal, the data may be affected by "social desirability bias" (Nuno and John 2015). If that is the case, there is the risk that the data collected are inexact and biased. We use several

10

9

C. VASCO ET AL

Table 3. Determinants of market integration among indigenous peoples (marginal effects). Probit Crops Human capital 0.031-* Age Age squared --0.000-* 0,015Education Children 0.055-* Adults --0.007 Ethnicity Shuar 0.160Financial capital --0.034 Wealth index Social capital - 0.018 Confidence Natural capital Cultivated land 0.096*** Total land 0.038 Distance to river 0.037 Physical capital - 0.236*** Distance to road Contextual (cammunity-leve/) predictors Population density 0.005 Conflict for forest 0.069 Distance to Puyo 0.559*** Number of observations

Timber

- 0.0120 0.000 - 0.025*** 0.012 - 0.007 0.331 - 0.006

Tobit Credit

Agricultural wage labor

Nonagricultura l wage labor

- 0.004 0.000 - 0.001 - 0.000 o.oi5

0.029* --0.000* 0.021*0.012 --0.015

0.008 --0.000 --0.0100.034-* 0.010

0.017 --0.000 0.018*** 0.003*** --0.013

- 0.001

0.288*-

--0.040***

0.038*

Game and fish

0.055

0.204***

--0.022-

--0.076

- 0.110

0.046

--0.013

--0.074-

0.082

0.077 - 0.024 0.118

0.004 0.011 --0.055**

--0.044 0.017 --0.042

--0.003 --0.0260.026

--0.050 0.011 --0.027

--0.369***

--0.037**

--0.152*-

--0.043-

0.012

- 0.012** 0.405*** 0.473***

--0.000 --0.003 0.035

0.000 0.050 --0.132 213

--0.001 0.107-* 0.061

--0.008*0.094 --0.102

Nate. Marginal effects for Tobit models are estimated at the unconditional value of y. All the models were estimated with robust standard errors clustered at community level. Asterisks *, **, and -* indicate significance at 10, 5, and 1%, respectively.

strategies to address this potential source of bias. Prior research (Tourangeau and Yan 2007) found that the characteristics of the survey taker may influence responses, with respondents being more likely to divulge/hide sensitive information depending on the characteristics of the person who delivers the survey. In this sense, our survey team was composed of Kichwa and Shuar undergraduate students, who were in all cases native to the area. In distinction from other agents who may require such information (e.g., nonindigenous survey takers, public employees, forest rangers), survey takers with such characteristics were normally not perceived as a threat by interviewees, who, on the contrary, were willing to cooperate with the survey team. In the questionnaire utilized, questions on sensitive topics (i.e., logging and hunting) were asked near the end of the survey. This strategy is expected to reduce the likelihood of false responses to sensitive questions (Brace 2008).

Likelihood of Selling Crops

There is a quadratic relationship between age and the likelihood of selling crops, with the latter increasing with age to a turning point at 55 years. Every year of formal education raises the odds of selling crops by about 1%. A possible explanation is that improved schooling leads to higher levels of specialization, adoption of modern technologies, and higher productivity (Huffman 2001). With improved harvests, educated indigenous households tend to specialize in commercial agriculture. Every child in the household increases the likelihood of selling crops by 5%. A possible explanation is that the production of naranjilla, the most common market crop in the area, is normally undertaken by the children

SOCIE1Y & NATURAL RESOURCES

9

11

of the household. Shuar households are 16% more likely to sell crops than their Kichwa peers. This is consistent with previous research (Gray et al. 2008; Lu, Bilsborrow, and Ona 2012) showing that the Shuar are the most colonist-like indigenous group, exhibiting a strong drive for commercial agriculture. As expected, households with larger agricultural areas are more likely to engage in market-oriented agriculture. Locational predictors offer interesting insights. The likelihood of selling crops is higher for households located in communities far away from Puyo, but is lower for dwellings located far away from roads. In the first case, the result stands in contradiction to prior research (Gray et al 2008) showing that distance to urban areas is a major constraint for indigenous participation in agricultural markets. A possible explanation is that rural people residing closer to urban areas prefer engaging in more lucrative nonagricultural employment rather than in agricultural activities (Vasco Perez, Bilsborrow, and Torres 2015). Concerning distance to roads, the finding is consistent with prior research on tropical forests (Pichon 1997) reporting that cash cropping flourishes near roads, since roads encourage farmers to take advantage of higher prices for agricultural products in urban areas. Nevertheless, caution is needed when interpreting this result, as it may also be likely that households already integrated into agricultural markets chose to settle near roads to facilitate the transport of the produce to markets. Likelihood of Selling Timber

Contrary to crops sales, households with educated heads are less likely to sell timber. Each year of formal education reduces the likelihood of selling timber by about 2%. A possible explanation is found in the work of Salo, Siren, and Kalliola (2014), who argue that new generations of more educated indigenous leaders exhibit more environmental awareness than their predecessors and have designed forest management plans to promote the sustainable use of natural resources. Timber extraction is more likely to occur far away from population centers, where forested areas still exist, and close to roads, which facilitate the transport of timber to markets. Altogether, these results reflect that, at present, distance to roads matters more than distance to urban areas in shaping market integration decisions among indigenous peoples. The higher the population density in a community, the lower is the probability of selling timber. This may reflect that timber is a scarce resource in communities with higher population densities. Households in communities having conflicts with outsiders are 40% more likely to extract and sell timber. While all the communities in the sample hold legal usufruct rights over their lands, conflict with outsiders (i.e., oil and mining companies and colonists) and with other indigenous communities exists in the research area. This finding is consistent with previous research (Godoy et al. 1998) concluding that uncertainty about land rights may increase deforestation by indigenous peoples themselves. In this context, the risk of displacement and uncertainty about the future may encourage indigenous peoples to adopt unsustainable livelihood practices like logging. Caution is needed, however, when interpreting this result, as it is possible that timber sale is a cause rather than a consequence of conflicts with outsiders (i.e., logging companies intending to exploit timber in indigenous territories). While this may happen in other regions of the (Ecuadorian) Amazon, previous research in the study area (Muzo et al. 2013) found that timber markets are still incipient, most timber is directly sold to intermediaries or small-scale sawmills, and only a small fraction of the timber is harvested by intermediaries.

12

9

C. VASCO ET AL

Likelihood of Selling Game and Fish

In distinction from the sales of crops and timber, which are not influenced by wealth, the likelihood of selling game and fish is higher for poorer households. A possible explanation for this finding is that engaging in cash cropping and timber logging entails expenses that poor households are not able to cover (chainsaw operator wages, extra household labor, fuel, seeds, agricultural inputs, tools, among others). On the other hand, hunting and fishing do not demand high levels of investment and therefore can be done by poorer households. The likelihood of selling wildlife and fish is higher in the proximity of roads and navigable rivers. These results may reflect that game and fish are mostly sold along roadsides or in the surroundings of fluvial ports (Jacome-Negrete et al. 2013). Livelihood of Requesting Credit

Everything else held equal, each year of schooling increases the likelihood of requesting a loan by 2%. This may reflect that the process of loan application demands a certain degree of formal education. Furthermore, in the context of rural areas in the developing world, more educated individuals are commonly perceived as more creditworthy (van de Walle 2003). Households with Shuar heads are 28% more likely to ask for a loan than their Kichwa counterparts. A possible explanation is that the Shuar devote larger areas to agriculture and engage in cattle ranching and commercial agriculture more than the Kichwa (Gray et al. 2008) and hence seek credit to buy cattle and agricultural inputs. Since there is no effect of total agricultural area on the likelihood of asking for a loan, there is the possibility that the dummy accounting for Shuar headship is capturing the effect of involvement in agriculture. However, when removing the dummy for Shuar headship from the model, the agricultural area remains nonsignificant, which suggests that there are factors linked to ethnicity that make the Shuar more likely to seek credit. Beyond this, the results also suggest that loans are not necessarily used to finance agricultural production. The likelihood of requesting a loan is higher for households in the proximity of roads, which shows, once again, that access to infrastructure matters more than distance to urban areas when explaining market involvement decisions. Share of Income from Agricultural Wage Labor

Each year formal education reduces the share of income from farm wage labor by 1%. Earnings from nonagricultural work are considerably higher than those from farm wage labor; therefore, the latter may not be attractive for educated individuals (see Table 4). The higher the number of children in a household, the larger is the share of income from agricultural wage work. This likely reflects that adults from households with more children can engage in agricultural wage work as children take care of the family's farm. In terms of social capital, households that trust other community members depend less on agricultural wage labor. A possible explanation is that households in communities with stronger ties meet their extrahousehold labor needs via reciprocal labor. As a consequence, labor markets are limited in these kinds of communities (Vasco 2014). The fraction of income from wage farm work is higher for households residing near roads. Market-oriented agriculture,


9

13

Table 4. Average annual income by type of employment and ethnic group (USS). Agricultural self-employment Wage employment Agricultural wage employment Nonagricultural wage employment Total income

Kichwa

Shuar

All ethnic groups

1160

2016

1300

1593 5902 4065

1453 5060 5634

1522 5352 4890

which is more likely to demand extrahousehold labor, takes place near roads. The fraction of income from agricultural wage labor is higher for poor households with little land, residing in communities that have conflicts over the use of forest. These findings are consistent with previous research (Vasco and Tamayo 2017) and suggest that participation in nonagricultural wage labor is driven mainly by push factors, including poverty, lack ofland, lack of education, and conflicts with outsiders. Share of Income from Nonagricultural Wage Labor

Each year of schooling increases the share of income from nonagricultural labor by 2%. Nonagricultural wage work is the best employment choice in terms of wages, so it makes sense that educated individuals try to maximize returns to education by engaging in nonfarm wage labor. As with agricultural wage income, the share of income from nonagricultural sources is higher for households with more children, probably for similar reasons. The share of income from nonagricultural work is higher for Shuar homes. In contrast, the distance to Puyo and the distance to the nearest road have no effect on income from nonagricultural work. This stands in contradiction to prior literature showing that involvement in nonagricultural work is higher near urban areas (Vasco and Tamayo 2017). As referred to earlier in the text, Shuar households are located closer to roads and urban areas; therefore, it is possible that the dummy controlling for Shuar headship is capturing the effect of locational predictors. To test this possibility further we ran the model without the Shuar dummy. While predictors associated with Shuar households (i.e., population density and confidence) become marginally significant (at 90% probability) (data available upon request), the effect of locational variables remains nonsignificant. This leads us to think that there are other variables linked to ethnicity that drive the Shuar to participate more in the market economy. An alternative explanation for the nonsignificance of locational variables is that recent governmental investment in education and decentralization of public services have opened nonfarm job opportunities (i.e., school teachers, public employees) even in remote areas4 (Vasco Perez, Bilsborrow, and Torres 2015). The higher the population density in a community, the lower is the contribution of nonagricultural work. A possible explanation for this finding is that there is higher competition for nonagricultural jobs in more populated areas.

Conclusion The results presented here demonstrate that market participation of indigenous peoples is principally driven by endowments of human capital (education), financial capital (wealth), and physical capital (access to roads). At a contextual level, distance to markets, population density, and conflicts with outsiders also shape market participation decisions.

14

9

C. VASCO ET AL

Education has mixed effects on market integration, with the direction of the effect depending on the measure of market integration utilized as a dependent variable. Education increases the likelihood for a household to sell crops and seek credit, as well as the fraction of income from nonagricultural work, but reduces the odds of selling timber and the share of income from farm wage labor. Although educated indigenous peoples take advantage of higher market prices of crops, higher wages for nonagricultural work, and a greater share of the credit supply for rural people, the uneducated must interact with the market economy in a less advantageous position- that is, by taking poorly paid agricultural jobs and engaging in often-illegal5 timber operations, which involve the risk of being fined. Integration into the market economy is also determined by push factors, including poverty and encroachment Poorer households tend to engage in the trade of wildlife and agricultural wage work, whereas households in communities in conflict with outsiders are more likely to engage in timber logging and agricultural wage work. These kinds of households lack human (education) and financial (wealth) capitals and hence are pushed to participate in low-paid agricultural wage work and to get involved in timber and wildlife trade. Other than these findings, we provide some policy recommendations to reconcile market integration and environmental conservation. Our results indicate that investing in human capital (education) may contribute to environmental conservation by reducing the likelihood of harvesting timber and smoothing participation in nonagricultural employment, which offers higher income than agriculture while taking household labor away from the farm. Nevertheless, access to education may have negative externalities as it increases participation in cash crop production and access to credit, which may be used for unsustainable economic activities (e.g., cattle ranching). Hence, investments in education should come together with policies to promote the use of more sustainable agricultural practices (e.g., land-intensive agriculture and agroforestry) in indigenous territories. Concurrently, the government should restrict credit for crops/activities with high environmental impact like naranjilla production and cattle ranching and instead should provide cheap credit to more environmentally friendly economic activities (e.g., ecotourism and processing and industrialization of renewable forest products). Finally, we provide a comment on the role of roads on market integration, as availability of roads is highly and positively correlated with five out of the six measures of market integration analyzed here. During the last decade, the Ecuadorian government has carried out significant investments in improving and expanding the road system in the Amazon (Vasco Perez, Bilsborrow, and Torres 2015). While new roads may improve the living conditions of the rural population by facilitating mobility of people, goods, and services, they may also hamper indigenous livelihoods and the fragile Amazon ecosystem, stimulating land clearing for agricultural purposes and the trade of forest products. While some advocate for restricting the construction of new roads in the Amazon (Pichon 1997), the government (national, provincial, municipal) sees the expansion and improvement of the road network as a development priority (SENPLADES 2012; Prefectura de Pastaza 2012). Indigenous peoples themselves have different perceptions concerning access infrastructure, with some preferring staying isolated from markets and rejecting the construction of roads in their territories, and others, principally those living in communities close to already existing roads, demanding the construction of more roads (Tandazo 2009). This poses a big challenge for practitioners seeking to balance rural development and conservation policies in indigenous territories.


9

lS

Notes 1. About 10% of the households in the sample had a female head. 2. The index is the first principal component of the following housing characteristics: number of rooms, availability of bathroom, cement floor, cement walls, piped water in home, electricity, and possession of stove, boat, radio, and shotgun. The first principal component accounted for 28% of the variance in the sample. 3. Another alternative to control for contextual effects is the use of community fixed effects. Nevertheless, given the relatively small sample size, we prefer using community-level aggregate values. 4. For instance, 50% of the Kichwa and 33% of the Shuar nonagricultural workers in the sample are school teachers in indigenous schools. 5. Muzo et al. (2013) showed that more than half of the logging operations by indigenous peoples in the northern Ecuadorian Amazon are done illegally.

Acknowledgments We are grateful to Seraphine Munroe, Jeremy Rayner, and three anonymous referees for valuable comments on earlier versions of this article.

Funding Fieldwork was funded by the Universidad Estatal Amaz6nica at Puyo, Ecuador.

References Brace, I. 2008. Questionnaire design: How to plan, structure and write survey material for effective market research. London, UK: Kogan Page. Bremner, J., and F. Lu. 2006. Common property among indigenous peoples of the Ecuadorian Amazon. Conservation and Society 4 (4):499- 21. Cavendish, W. 2003. How do forests support, insure and improve the livelihoods of the rural poor? A research note. Bogor, Indonesia: CIFOR. de Haan, L. 2012. The livelihood approach: A critical exploration. Erdkunde 66 (4):345- 57. doi: l0.3l12/erdkunde.2012.04.05 Ellis, F. 1998. Household strategies and rural livelihood diversification. Journal of Development Studies 35 (1):1- 38. doi:l0.1080/00220389808422553 Ellis, F. 2000. Rural livelihoods and diversity in developing countries. Oxford, UK: Oxford University Press. FAO. 2005. State of the world's forests 2005. Rome, Italy: United Nations Food and Agricultural Organization. Godoy, R. 2001. Indians, markets, and rainforests: Theory, methods, and analysis. New York, NY: Columbia University Press. Godoy, R., E. Undurraga, D. S. Wilkie, D. Reyes-Garcia, T. Huanca, W. Leonard, T. McDade, S. Tanner, and V. Vadez. 2010. The effect of wealth and real income on wildlife consumption among native Amazonians in Bolivia: Estimates of annual trends with longitudinal household data (2002- 2006). Animal Conservation 13:265- 74. doi: 10.l 111/j.1469-1795.2009.00330.x Godoy, R., F. Franks, and M. Alvarado Claudio. 1998. Adoption of modern agricultural technologies by lowland indigenous groups in Bolivia: The role of households, villages, ethnicity, and markets. Human Ecology 26 (3):351- 69. Godoy, R., K. O'neill, S. Groff, P. Kostishack, A. Cubas, J. Demmer, K. Mcsweeney, and J. Overman. 1997. Household determinants of deforestation by Amerindians in Honduras. World Development 25 (6):977- 87. doi: l0.1016/s0305-750x(97)00007-7

16

9

C. VASCO ET AL

Godoy, R., M. Jacobson, J. D. Castro, V. Aliaga, J. Romero, and A. Davis. 1998. The role of tenure security and private time preference in neotropical deforestation. Land Economics 74 (2):162- 70. doi:l0.2307/3147048 Godoy, R., V. Reyes-Garcia, T. Huanta, W. Leonard, V. Vadez, C. Valdes-Galicia, and D. Zhao. 2005. Why do subsistence-level people join the market economy? Testing hypotheses of push and pull determinants in Bolivian Amazonia. Journal of Anthropological Research 61 (2):157- 78. doi:l0.3998/jar.0521004.0061.202 Gray, C., R. Bilsborrow, J. Bremner, and F. Lu. 2008. Indigenous land use in the Ecuadorian Amazon: A cross-cultural and multilevel analysis. Human Ecology 36:97- 109. doi:l0.1007/ s10745-007-9141-6 Huffman, W. E. 2001. Human capital: Education and agriculture. Handbook of agricultural economics 1:333- 381. doi:10.1016/S1574-0072(01)10010-1 INEC. 2010. Censo de Poblaci6n y Vivienda- 2010. Quito, Ecuador: Instituto Nacional de Estadisticas y Censos. Jacome-Negrete, I., V. Tanchinla, P. Santi, and C. Vargas. 2013. Etnozoologia quichua para la conservaci6n de los mamiferos ungulados en la Amazonia central del Ecuador provincia Pastaza. Revista Amaz6nica- Ciencia y Tecnologia 2 (3):172- 85. Kish, L. 1965. Survey sampling. New York, NY: Wiley. Kuntashula, E., D. van der Horst, and S. Vermeylen. 2009. A pro-poor biofuel? Household wealth and farmer participation in Jatropha curcas seed production and exchange in eastern Zambia. Biomass and Bioenergy 63:187- 97. Laszlo, S. 2005. Self-employment earnings and returns to education in rural Peru. Journal of Development Studies 41 (7):1247- 87. doi:l0.1080/00220380500170915 Lu, F. 2001. The common property regime of the Huaorani Indians of Ecuador: Implications and challenges to conservation. Human Ecology 29 (4):425- 47. Lu, F. 2007. Integration into the market among indigenous peoples: A cross-cultural perspective from the ecuadorian Amazon. Current Anthropology 48 (4):593- 602. Lu, F., C. Gray, R. Bilsborrow, C. Mena, C. Erlien, J. Bremner, L. Barbieri, and S. Walsh. 2009. Contrasting Colonist and Indigenous Impacts on Amazonian Forests. Conservation Biology 24: 881- 885. doi:l0.1111/j.1523-1739.2010.01463.x Lu, F., R. E. Bilsborrow, and A. I. Ona. 2012. Modos de vivir y sobrevivir: Un estudio transcultural de cinco etnias en la Amazonia ecuatoriana. Quito, Ecuador: ABYA-YALA. Mcsweeney, K., and B. Jokisch. 2007. Beyond rainforests: Urbanisation and emigration among lowland indigenous societies in Latin America. Bulletin of Latin American Research 26 (2): 159- 80. doi:l0.1111/j.1470-9856.2007.00218.x Muzo, A., F. d. Gato, P. Pacheco, and B. Torres. 2013. Timber harvesting by smallholders in Napo and Orellana. In Forest use and timber markets in the Ecuadorian Amazon, ed. E. Mejia and P. Pacheco, 47- 68. Bogor, Indonesia: CIFOR. Myers, N., R. Mittermeier, C. Mittermeier, G. Fonseca, and J. Kents. 2000. Biodiversity hotspots for conservation priorities. Nature 403:853- 58. Nuckolls, J. 2010. Lessons from a Quechua strongwoman: Ideophony, dialogue, and perspective. Tucson, AZ: University of Arizona Press. Nuno, A., and F. S. T. John. 2015. How to ask sensitive questions in conservation: A review of specialized questioning techniques. Biological Conservation 189:5- 15. Pichon, F. 1997. Colonist land-allocation decisions, land use, and deforestation in the Ecuadorian Amazon frontier. Economic Development and Cultural Change 45 (4):707- 44. Prefectura de Pastaza. 2012. Plan de desarrollo de la provincia de pastaza al aiio 2025. Puyo, Ecuador: Prefectura de Pastaza. Reardon, T., J. Berdegue, C. B. Barrett, and K. Stamoulis. 2006. Household income diversification into rural nonfarm activities. In Transforming the rural nonfarm economy, ed. S. Haggblade, P. Hazell, and T. Reardon, 1- 29. Baltimore, MD: Johns Hopkins University Press. Rudel, T. K., D. Bates, and R. Machinguiashi. 2002. Ecologically noble Amerindians? Cattle ranching and cash cropping among Shuar and colonists in Ecuador. Latin American Research Review 37 (1):144- 59.


9

17

Salo, M., A. Siren, and R Kalliola. 2014. Diagnosing wild species harvest resource use and conservation. New York, NY: Elsevier. SENPLADES. 2012. Proceso de desconcentraci6n de/ Ejecutivo en los niveles administrativos de planifiaci6n. Quito, Ecuador: SENPLADES. Tandazo, F. 2009. Efectos de construcci6n de la carretera Evenezer-Macuma-Taisha sobre la cobertura

boscosa de/ territorio de la Federaci6n Independiente del Pueblo Shuar de/ Ecuador "FIPSE". Magister Scientiae en Manejo y Conservaci6n de Bosques Naturales y Biodiversidad. Turrialba, Costa Rica: CATIE. Tourangeau, R, and T. Yan. 2007. Sensitive questions in surveys. Psychological Bulletin 133:859- 83. van de Walle, D. 2003. Are returns to investment lower for the poor? Human and physical capital interactions in rural Vietnam. Review of Development Economics 7 (4):636- 53. VanWey, L., and T. Vithayathil. 2013. Off-farm work among rural households: A case study in the Brazilian Amazon. Rural Sociology 78 (1):29- 50. Vasco, C. 2014. Reciprocal and wage labour in rural Ecuador. A quantitative analysis. Journal of Agriculture and Rural Development in the Tropics and Subtropics 115 (1):23- 30. Vasco Perez, C., R Bilsborrow, and B. Torres. 2015. Income diversification of migrant colonists vs. indigenous populations: Contrasting strategies in the Amazon. Journal of Rural Studies 42:1- 10. doi:l0.1016/j.jrurstud.2015.09.003 Vasco, C., and A. Siren. 2016. Correlates of wildlife hunting in indigenous communities in the Pastaza province, Ecuadorian Amazonia. Animal Conservation 19 (5):422-429. doi:l0.1111/ acv.12259 Vasco, C., B. Torres, P. Pacheco, and V. Griess. 2017. The socioeconomic determinants of legal and illegal smallholder logging: Evidence from the Ecuadorian Amazon. Forest Policy and Economics 78:133- 40. doi:l0.1016/j.forpol.2017.01.015 Vasco, C., and G. N. Tamayo. 2017. Determinantes de! empleo no agricola y de Ios ingresos no agricolas en el Ecuador. Revista CEPAL 121:55- 71. Whitten, N. E. 1976. Sacha runa: Ethnicity and adaption of Ecuadorian jungle Quichua. Urbana, IL: University of Illinois Press. Wooldridge, J. M. 2002. Introductory econometrics: A modern approach, 2nd ed. Cincinnati, OH: South-Western College Publishing. Zimmerman, B., C. Peres, J. Malcolm, and T. Turner. 2001. Conservation and development alliances with the Kayap6 of south-eastern Amazonia, a tropical forest indigenous people. Environmental Conservation 28 (1):10- 22.