Poverty indicators and development alternatives for the Amazon: The use of multidimensional index for the Brazilian Legal Amazon region

Indicadores de pobreza e alternativas de desenvolvimento para a Amazônia: o uso de índice multidimensional para a região da Amazônia Legal brasileira.

Abstract

Measuring poverty requires an understanding of issues that transcend a purely economic character/nature. This article proposes the IPM-Trajectories, a poverty indicator for the Legal Amazon, based on the multidimensional approach proposed by Alkire-Foster. The deprivations for the urban and rural areas are considered in a differentiated way, based on the Census microdata in the dimensions of Health, Education, and Living Conditions. IPM-Trajectories is an initial effort that seeks to dialogue, from another perspective, with narratives that reinforce the idea of persistent and chronic poverty in the Amazon and the invisibility of the economies referred to as the biome.

Keywords:
Poverty; Multidimensional Poverty Index; Development; Regional; Amazon

Resumo

Mensurar a pobreza requer a compreensão de questões que transcendem o caráter puramente econômico. Este artigo propõe o IPM-Trajetórias, um indicador de pobreza para a Amazônia Legal, baseado na abordagem multidimensional proposta por Alkire-Foster. As privações das áreas urbanas e rurais são consideradas de forma diferenciada, com base nos microdados do Censo, nas dimensões Saúde, Educação e Condições de Vida. O IPM-Trajetórias é um esforço inicial que busca dialogar, sob outra perspectiva, com narrativas que reforçam a ideia de pobreza persistente e crônica na Amazônia e a invisibilidade das economias associadas ao bioma.

Palavras-chave:
Pobreza; Índice de Pobreza Multidimensional; Desenvolvimento; Regional; Amazônia

1 Introduction

In September 2030, the United Nations General Assembly published "Transforming our World: the 2030 Agenda for Sustainable Development" (U.N., 2015). In its first paragraph, the resolution: "This Agenda is a plan of action for people, the planet, and prosperity. [...] We recognize that eradicating poverty in all its forms and dimensions, including extreme poverty, is the greatest global challenge and an indispensable requirement for sustainable development.[...] As we embark on this great collective journey, we pledge that no one will be left behind. ". The agenda presents its operationalization in 17 objectives associated with 169 goals. Together, they form the 17 Sustainable Development Goals (SDGs). SDG 1, "End poverty in all its forms everywhere," reaffirms the elimination of poverty as an essential dimension of economics oriented to people and the planet.

Poverty and development have a long tradition in economics (George, 1881; Schultz, 1961; Furtado, 1974; Sen, 1981; Mahbub Ul Haq, 1996; Kabeer, 1994; Banerjee e Duflo, 2011). Furthermore, the SDG agenda arrives while the global emergency consolidates (IPCC-208, 2021, 2022a, 2022b; WMO, 2022). The same pattern of urban-industrial development that sustains the functioning of the global economy feeds and aggravates several reflexes of planetary environmental action and its social and environmental territories. This intersection between the economic development strategies of the needs of people living in different regions and the climatic and ecological balance makes the debate on the options of development models we have in dispute more crucial than ever.

The Legal Amazon comprises 5,217,423 km² in Brazil, about 61% of the Brazilian territory. It encompasses the entirety of the states of Acre, Amapá, Amazonas, Mato Grosso, Pará, Rondônia, Roraima and Tocantins and part of the State of Maranhão (195 municipalities - 90% of the total). As it shelters the entire Amazon biome, it includes the formations of the Amazon Forest and the formations associated with ecosystems subject to the Tropical Forest Protection Law (Law No. 11,428, of 2006). The Legal Amazon constitutes a space of risks and opportunities: territories with development projects are in dispute. Developing policies that address people, the planet, and prosperity in an integrated way (U.N., 2015) requires rescuing the Amazon economies and their protagonists, which must be at the center of a truly sustainable development project (Costa, 2020; Costa et al., 2021).

These disputes between development models in economic studies made the categories and methodologies evolve to establish new poverty measures. The 1990 United Nations Development Program (UNDP) report (UNDP, 1990) launched the Human Development Index (HDI), which included the dimensions of longevity (life expectancy at birth) and education (average years of schooling and predicted years of schooling) in addition to GDP per capita. In 1997, UNDP introduced the Human Poverty Index (HPI), which measures multiple deprivations. In 2010, UNDP decided to introduce a new international measure of poverty - the Multidimensional Poverty Index-MPI (MPI), which measures the combination of deprivations each family experiences using data to observe deprivations overlapping in three dimensions - education, health, and living conditions. The MPI used a multidimensional measurement technique known as the Alkire-Foster (A.F.) method (Alkire and Foster, 2011) and was developed and applied by the Oxford Poverty and Human Development Initiative - OPHI with support from UNDP. In 2018, UNDP and OPHI developed a new version of the MPI adapted to the SDGs (OPHI-UNDP, 2018). Although methodological experiments with multidimensional poverty measures have a pioneering history in Latin America, they have fallen by the wayside (Boltvinik, 2013; Jannuzzi and de Sousa, 2016). Nevertheless, the MPI is a methodology and measure of poverty associated with new global agendas and is a relevant indicator for debates on development in national spaces.

In Brazil, after 2010, a growing number of studies has addressed the construction of MPIs nationally. However, they underestimate the regional specificities (Santos et al., 2016; da Rosa et al., 2019; Rodrigues, 2019). General indices, notably the HDI- Human Development Index (HDI), helped to build narratives of the persistence of poverty in the Amazon associated with the backwardness of its rural economy. The GDP per capita measures aim to understand the stage of development of these territories. In this perspective, economies that rely on the biodiversity of the biome and its protagonists are said to resist progress, modernity, and prosperity. In contrast, commodity-oriented economies support better frameworks for human development. This reading results in statements by public agents such as this: "We have to recognize that where there are many forests, there is much poverty," which echoes a narrative that is more prevalent in the various sectors of Brazilian society than one would think. Even in critical, adequate, and broader readings that use this poverty measure (HDI) in the construction of their argument (Rodrigues et al., 2009), the narrative of the "less economically developed" region appears.

Thus, there is an urgent need to construct and revise poverty measures for the Amazon. The multidimensional approach to poverty measurement, MPI, can be used to construct indicators that can tell us more about the territory's local specificities, such as differences between its urban and rural communities subject to different agrarian economy trajectories and agents, as well as different levels of land use and land cover transformation.

2 The Multidimensional Poverty Index - MPI

Understanding poverty as a multidimensional phenomenon began to be explored in the 1970s. The pioneering work of Amartya Sen (Sen, 1976) highlights two steps for measuring poverty: identifying who is poor and aggregating information about poverty. The definition of the poverty line is the first step. At the same time, the second brings a novelty when trying to aggregate data from poor individuals in a general indicator. In this aspect, indices that begin to disaggregate poverty as a combination of different phenomena, such as inequality and income disparity, stand out (Thorbecke et al., 1984).

The multidimensional indices dealing with poverty have been multidimensional since their identification. In this sense, several methodological advances were grouped into the following approaches: axiomatic and information theory, fuzzy methodologies, and theories of latent variables (Alkire and Santos, 2013). Among the techniques, Dashboard methods and composite indices are widely used as indicators for human development (Alkire et al., 2015). The SDG agenda, for example, is a Dashboard where 49 indicators are used to monitor the 17 objectives and 169 targets. In addition, the HDI and the Human Poverty Index (HPI) add more than one aspect of poverty in their formulation.

Despite being widely used, both methodologies fail to identify simultaneous deprivations. On the other hand, axiomatic measures, such as the Multidimensional Poverty Index (MPI), have several convenient characteristics. First, they fulfill the two steps for measuring poverty proposed by Sen (1985). Second, the portfolio of axiomatic measures includes measures that not only apply when indicators are cardinal but also when indicators are ordinal.

The method for calculating the MPI requires identifying the units of analysis. Usually, the household then simultaneously calculates the indicators they are deprived of. This way, a poverty profile is created in a weighted deprivation score. Units are called multidimensionally poor if their deprivation score exceeds an interdimensional poverty threshold. For the indicator, the proportion of poor people and their average deprivation score (i.e., the "intensity" of poverty or the percentage of concurrent deprivations they experience) become part of the final measure of poverty (Alkire et al., 2018).

The MPI is currently one of the indicators used to understand world poverty. The Global MPI is an application of the Alkire-Foster (A.F.) method. The World Bank used the technique for disseminating a multidimensional poverty measure incorporating five dimensions of well-being: i) monetary poverty, ii) education, iii) access to basic infrastructure, iv) health and nutrition, and v) security. (Alkire et al., 2018). The A.F. method allows the choice of measure, space, unit of analysis, dimensions, deprivation thresholds (to determine when a person is deprived in a dimension), weights or values (to indicate the relative importance of deprivations), and poverty cutoff (to determine when a person is deprived enough to be considered poor). This flexibility allows the methodology to have several applications and be used in different situations, ensuring the unique character for identifying poverty in various conditions (Alkire et al., 2015a).

The A.F. methodology has two phases: identification and aggregation. Intuitively, the phases can be separated into steps, as Alkire et al. (2015a) described. There are six stages in the first identification step. The first is to define the set of indicators that will be considered in the multidimensional measure, considering that the data for the indicators need to be available for the same person or family observed. In the second and third stages, the cutoff points for each indicator and the definition of deficiencies. Finally, each indicator's weight or relative value is selected in the last steps, whose sum is one. The deprivation score is the poverty threshold, the proportion of weighted deprivations a person needs to experience to be considered multidimensionally poor.

3 Multidimensional Poverty Indicators in Brazil

Several research institutions, research groups, and non-governmental organizations, inspired by the developments of the HDI and, in particular, the HDI-M-Municipal HDI, a methodological adjustment to the Global HDI produced by UNDP Brazil, made efforts to develop summary measures of the social situation - the indicators of social reality, in a multidimensional perspective. This issue appears in documents and reports that present the various indicators proposed by planning and statistical institutions, research centers, universities, government agencies, and organized civil society.

In the line of composite indicators, the Living Conditions Index (ICV), Firjan Municipal Development Index (IFIRJAN), Social Progress Index (IPS), Social Vulnerability Index (IVS), Municipal Governance Index (IGM), and Territorial Governance Index (IGT) are some examples for Brazil. These initiatives emerged from the need to build indicators that capture human development beyond economic growth. Among these indicators, the IPS Amazônia is an initiative to address the Brazilian Amazon, using global IPS as a reference. In addition, the consolidation of poverty as a phenomenon of multiple dimensions and the growing presence of the methodological approach associated with the MPI in the international scenario boosted empirical studies in Brazil based on the method of Amartya Sen's capabilities (Sen, 1976).

The Bagolin and Ávila (2006) study developed an index composed of four indicators - nutrition, health, safety, and basic knowledge (schooling) - to analyze multidimensional poverty among Brazilian states. Barros, Carvalho, and Franco (2006) constructed the family poverty index, including a total of 48 indicators and six dimensions, based on data from the National Household Sample Survey (PNAD) from 1993 and 2003: a) vulnerability; b) access to knowledge; c) access to work; d) scarcity of resources; e) child development; and f) housing shortages. This way, the most fundamental dimensions of poverty addressed in the empirical literature were considered, except for health conditions. Each of these six dimensions captures the lack of access to the means necessary for families to satisfy their basic needs and, in part, the existence of unsatisfied basic needs. In turn, Kageyama and Hoffmann (2006) also used the PNAD microdata to analyze poverty in Brazil between 1992 and 2004, combining indicators of income and housing conditions.

Also, based on the PNAD, Buainain, Dedecca, and Neder (2013) constructed a panel of deprivation indicators for rural areas in 2009. Vaz and Jannuzzi (2014) estimated multidimensional poverty in Brazil between 2001 and 2013 using the same criteria as in the Global MPI to evaluate government programs to overcome poverty in the country. Jannuzzi and Fonseca (2017) analyzed trends in monetary poverty, inequality, and multidimensional poverty from 2004 to 2014.

Using Censo microdata, Serra (2017) and Serra et al. (2020) calculated MPI for Brazil, comparing rural and urban areas. Furthermore, Santos et al. (2016) developed an MPI specific for the Amazon region; they analyzed the dynamics of poverty in rural and urban areas in the municipalities of the Legal Amazon. The study developed by Rodrigues et al. (2020) advances the MPI methodology. It analyzes multiple dimensions of intra-urban poverty in the Metropolitan Region of Belém. Finally, Rosa (2021) analyzes the evolution of multidimensional poverty in the Northern Region of Brazil, which accommodates a large part of the Legal Amazon, and its relationship with territorial, economic, and population dynamics in the years 1991, 2000, and 2010 using the MPI that she produced for the North region. Table 1 brings together some of the central studies bringing a multidimensional perspective to poverty indicators in Brazil.

Table 1:
Experiences of Multidimensional Poverty Indicators in Composition and Axiomatic Approaches (MPI) in Brazil.

References	Dimensions	Components/Indicators	Geographic Scope
Bagolin and Avila (2006)	Health	1. hospital beds; 2. supply of doctors; 3. infant mortality; and 4. life expectancy.	National.
	Security	1. Homicides; 2. Police presence
	Education	1. Illiteracy; and 2. schooling.
	Food Security	The proportion of people with food restrictions
Barros, Carvalho and Franco (2006)	Vulnerability	1. fecundity; 2. special attention and care for children, adolescents, and young people; 3. special attention and care for the elderly; 4. demographic aspects and 5. presence of the mother.	National, comparing rural and urban areas.
	Access to knowledge	1. illiteracy; 2. schooling; and 3. professional qualification.
	Access to work	1. availability of work; 2. quality of the job; and 3. remuneration.
	Resource scarcity	1. extreme poverty; 2. poverty; and 3. income generation capacity.
	Infant development	1. child labor; 2. access to school; and 3. school progress.
	Housing shortages	1. property ownership; 2. housing deficit; 3. shelter; 4. access to water supply; 5. access to sanitation; 6. to garbage collection; and 7. access to electricity.
Kageyama and Hoffmann (2006)	Income	Minimum Wage	National, comparing rural and urban areas.
	Lack of basic equMPIent	1. the presence of piped water; 2. the presence of a bathroom or toilet; and 3. household with electricity .
Buainain, Dedecca and Neder (2013)	Income	1. proportion of poor; 2. poverty intensity; 3. aggregate poverty gap value; and 4. FGT(2) index.	National, comparing rural and urban areas.
Vaz and Jannuzzi (2014)	Income	Household Income	National, comparing rural and urban areas.
	Basic needs	1. Adequate housing; 2. access to public infrastructure services; and 3. access to education
Jannuzzi and Fonseca (2017)	Income	Household income	National, comparing rural and urban areas.
	Basic needs	1. Adequate housing; 2. access to public infrastructure services; 3. access to work; 4. access to social protection and 5. access to education
Serra (2017)	Living standard	1. Water pipeline; 2. bathroom for exclusive use; 3. garbage disposal; 4. electrical energy; 5. durable consumer goods; and 6. resident/dorm density.	National, comparing rural and urban areas.
	Education	1. School attendance and literacy; and 2. appropriate age-school grade and education level.
Serra,Yalonetzky and Maia (2020)	Living standard	1. Water pipeline; 2. bathroom for exclusive use; 3. garbage disposal; 4. electrical energy; 5. durable consumer goods; and 6. resident/dorm density.	National, comparing rural and urban areas.
	Education	1. School attendance and literacy; and 2. appropriate age-school grade and education level.
Santos et al (2016)	Health	1. drinking water; and 2. total number of stillborn children (stillbirth).	Legal Amazon, comparing rural and urban areas
	Education	1. incomplete primary education; and 2. school-age children who are not in school.
	Living standard	1. electricity; 2. sanitary sewage; 3. garbage disposal; and 4. if the family has a car, freezer, telephone, and television.
Rodrigues et al (2020)	Health	1. drinking water; and 2. total number of stillborn children (stillbirth).	Metropolitan region of Belém, comparing rural and urban areas.
	Education	1. incomplete elementary education; and 2. school-age children who are not in school.
	Living standard	1. electricity; 2. sanitary sewage; 3. garbage disposal; and 4. if you have a car, freezer, telephone, and television.
Rosa (2021)	Income	per capita household income	The northern region of Brazil, comparing rural and urban areas.
	Life standard	1. access to electricity, 2. density of residents per dormitory, 3. access to sanitary facilities, and 4. assets [television, radio, refrigerator, and telephone]
	Health	1. garbage collection, 2. drinking water, 3. sanitary sewage, and 4. infant mortality
	Education	1. incomplete primary education and 2. whether the head of the household can read and write

Based on the studies surveyed (Table 1), it is possible to identify that there is no specific indicator considering the characteristics of the Amazon region. Furthermore, the impacts of changing variables on the size of poverty are not analyzed. This fact may be of paramount importance considering all the unique aspects of the Brazilian Legal Amazon region.

In this context, the present work proposes the bases for creating a new MPI for the Legal Amazon and the discussion of the impacts on poverty measurements. The MPI- Trajectories sought to (re)define dimensions and variables that could express particularities that impact poverty in the region's rural and urban populations in legal Amazon.

4 The MPI-Trajectories: rationality of dimensions and variables

The MPI-Trajectories include three central dimensions: Health, Education, and Living Conditions (Rorato et al., 2023). The Living Conditions dimension is structured into two sub-dimensions. One of Housing and Collective Consumption Goods (Sanitation and Electricity) and the other of Work and Private Consumer Goods. The production of the MPI-Trajetórias adopted the methodological path that Santos et al. (2016) proposed with adjustments. The indicators associated with each dimension and each stratum (urban and rural) were collectively defined by a broad and interdisciplinary group of specialists, including epidemiologists, economists, ecologists, biologists, geographers, and social scientists. Each indicator was mapped with binary values (0 or 1) where 1 indicates deprivation, while 0 indicates no deprivation. The units of analysis used were rural and urban households included in the 2000 and 2010 demographic census sample questionnaires. By starting from the household to observe situations of deprivation, the MPI approach restricts the databases that can be used to construct indicators for the 2000 and 2010 censuses.

The MPI-Trajectories for 2000 is composed of 16 indicators for the rural and 15 for the urban households, while for 2010, there are 19 and 18 indicators, respectively (Table 2). Some variables that make up the set of indicators differ between the two analyzed dates, according to the information collected in the sample questionnaire for the Census as mentioned above. However, the meaning of the indicator is maintained. This approach seeks to contemplate the particularities in intra-municipal spaces and the changes that have taken place over the decade in the region. Therefore, it was necessary to establish different cuts for the values of the indicators in urban and rural areas.

Table 2:
Variables and weights are used to compute the rural and urban multidimensional poverty indices.

Dimension	Variables		Weight
			Urban PMI		Rural PMI
			2000	2010	2000	2010
Education (¼)	schooling of adults	at least one household > 18 years of age, excluding the head of household, who did not complete elementary school	1/12	1/12	1/12	1/12
	schooling of the head of household	head of the household is illiterate	1/12	1/12	1/12	1/12
	schooling of children	at least one household, aged between 4 and 14, not in the school	1/12	1/12	1/12	1/12
Health (¼)	death of newborn	at least one death of a newborn in the household	1/4	1/12	1/4	1/12
	death of infants	at least one death of infants between 1 and 5 years of age	-	1/12	-	1/12
	early death	at least one death of householders between 6 and 73 years of age (life expectancy)	-	1/12	-	1/12
Living conditions: Housing (¼)	house construction quality	not a permanent house made of timber or brick	-	1/28	-	1/32
	temporary or collective house	temporary, collective or improvised house (p.e., tent)	1/24	1/28	1/28	1/32
	house size	(# householders / # dorms) > 5	1/24	1/28	1/28	1/32
	sewage	house not linked to a sewage network or septic tank)	1/24	1/28	1/28	1/32
	garbage	house without garbage collection services	1/24	1/28	1/28	1/32
	energy	house without electricity	1/24	1/28	1/28	1/32
	water	house without water services, or well/spring in the property, or do not have piped water.	1/24	1/28	1/28	1/32
	land ownership	permanent house in ceded land or other similar condition	-	-	1/28	1/32
Living conditions: income and consumer goods (¼)	social security	the only source of income is social security or unemployment aid, or student aid, or family allowance or other social programs.	1/20	1/20	1/20	1/20
	paid labor	head of the household without payment from main job in the last month	1/20	1/20	1/20	1/20
	food storage	household without a fridge	1/20	1/20	1/20	1/20
	communication	household without computer or cell phone, or T.V. or radio	1/20	-	1/20	-
		household without computer or cell phone with internet access, or T.V. or radio	-	1/20	-	1/20
	consumer goods	household without car for private use and washing machine.	1/20	-	-	-
		household without at least one of these items: car for private use, washing machine, phone.	-	1/20	-	-
	mobility	household without a car for private use.	-	-	1/20	-
		household without at least one of these items: car or motorcycle for private use.	-	-	-	1/20

• Source: Research data.

In the Education dimension (Table 2), it was considered that householders over 18 years of age who have not studied or have not completed elementary school are in a more significant situation of deprivation than those with basic schooling. Moreover, the illiteracy of the head of the household is also an indicator of deprivation. The level of education of the head of the household and adult members indicates the opportunities for inclusion and access to the world of work and access to information. It is assumed that the higher the level of education, the greater the skills and knowledge acquired. Thus, the criterion of not knowing how to read and/or write was adopted for those responsible for the household. A cutoff of 4 to 14 years was established for school-age children who do not attend school, an age group that shows whether school-age children are being exposed to the teaching and learning environment.

In the Health dimension, MPI approaches traditionally consider indirect indicators or even variables such as stillbirth (Table 1), which would be more associated with genetic conditions than access to health. This is due to the lack of health status in most datasets. In the MPI-Trajectories, more direct indicators of the individuals' health conditions were chosen within the constraints posed by the census datasets. For 2000, only one variable in the sample questionnaire was available for the health dimension, while for 2010, there were three variables (Table 2). Households with deaths of a live-born child were considered to be in a situation of deprivation (in 2000 and 2010), as well as those with records of deaths in children between 1 and 5 years old (infant mortality) and between 6 and the average life expectancy, defined as 73 years (only for 2010). Deaths in children can be associated with deficiencies in primary health care and socio-environmental conditions.

The Living Conditions dimension is split into Housing and Collective Consumer Goods and Work and Private Consumer Goods (Table 2). In the former, the indicators reflect the structure, type, and condition of the household, as well as the lack of sanitation and water services, which can indicate extreme poverty, when combined. People in households without access to the sewage public network or without a septic tank, as well as those without access to piped water or spring, are considered in a situation of deprivation in these indicators. Indicators such as electricity, household density, and type of household indicate the quality of housing to which families are subjected, as well as their quality of life. Thus, households without electricity, that are collective or improvised, and with five or more residents per bedroom were classified as deprived in this indicator.

In Work and Private Consumer Goods, the work indicators describe the status of those responsible for households, regarding their participation in income transfer programs, and lack of paid work; private goods refer to lacking a refrigerator, lack of information, and communication apparatus, and lack of private goods for mobility. The income of those responsible for the household indicates the ability to cover the household budget and the acquisition of goods and services, so the absence of income indicates a situation of deprivation. A refrigerator at home indicates the capacity to store food, and a microcomputer, telephone, radio, or T.V. signal inclusion in the digital age. Thus, individuals who do not have these items were considered deprived of communication and basic goods. Due to socioeconomic changes and the popularization of specific technologies in the 2000-2010 decade, the variables that make up some indicators were altered to capture these changes better. Thus, in 2010, the absence of a cell phone, a microcomputer with internet access, and a motorcycle were included as a deprivation situation.

4.1 Poverty indicator and cutoff

The Multidimensional Poverty Index (MPI) developed by Alkire and Foster reflects the share of deprivations that an individual (or household) suffers at a given time. In this framework, individuals are considered multidimensionally poor if their weighted deprivation score exceeds a poverty cutoff. The MPI can be expressed as the product of two components: the incidence of poverty, which captures the share of the population identified as multidimensionally poor, and the intensity of poverty, which captures the average share of deprivations among the poor. The formal algebraic definition of these components, as well as the dual‑cutoff identification strategy, follows the Alkire-Foster methodology and is presented in the next subsection.

4.2 Alkire-Foster (AF) method

The Alkire-Foster (AF) methodology provides an innovative approach to measuring poverty, going beyond traditional metrics such as the Foster-Greer-Thorbecke (FGT) class of measures (Foster, Greer and Thorbecke, 1984). The AF method offers an integrated and transparent way to compute the multidimensional poverty index $M_0$, adopting the individual (or household) as the central unit of analysis. This allows for a more accurate identification of multidimensional poverty and for a decomposition of the index by population subgroups and by dimensions (Alkire et al., 2015).

Formally, consider a population with n households and d dimensions of well-being. The achievements are arranged in an $n\times d$ matrix $Y=\left[y_{ij}\right]$, where $y_{ij}>0$ represents the achievement of household $i=\mathrm{1,}\dots ,n$ in dimension $j=\mathrm{1,}\dots ,d$. Each row vector of Y corresponds to the achievements of one household across all dimensions, and each column vector corresponds to the distribution of achievements across households in a given dimension.

For each dimension j, a deprivation cutoff $z_j$ is defined, representing the minimum level of achievement required for a household not to be considered deprived in that dimension. The vector of cutoffs is denoted by $z=\left(z_1,\dots ,z_d\right)$. Given the achievement $y_{ij}$, a household is considered deprived in dimension j if $y_{ij}<z_j$, and non-deprived otherwise. This leads to the deprivation matrix $g^0=\left[g_{ij}^0\right]$ , defined as:

$g_{ij}^0=\{\mathrm{1,}if{y}_{ij}<z_j\mathrm{0,}if{y}_{ij}\ge z_{j}fori=\mathrm{1,}\dots ,n,j=\mathrm{1,}\dots ,d$ (1)

The matrix $g^0$ summarizes the deprivation status of each household in each dimension: the row vector $g_i^0$ collects all deprivations of household i, while the column vector $g_j^0$ collects the deprivations of all households in dimension j.

Dimensions may differ in their relative importance. To account for this, a vector of weights $w=\left(w_1,\dots ,w_d\right)$ is introduced, where $w_j>0$ denotes the weight attached to dimension j. When all dimensions are equally important, this corresponds to a standard counting approach. When some dimensions are considered more relevant, larger weights are assigned to them.

The deprivation score of each household is then obtained as the weighted sum of its deprivations: $c_i={\displaystyle \sum }_{j=1}^d{w}_j{g}_{ij}^0$,where $c_i$ reflects the breadth of deprivations experienced by household i. If the household is not deprived in any dimension, $c_i=0$; if it is deprived in all dimensions, $c_i$ reaches its maximum value. The vector of deprivation scores for the whole population is $c=\left(c_1,\dots ,c_n\right)$.

In addition to the dimension-specific cutoffs $z_j$, the AF methodology employs a second cutoff to identify multidimensionally poor households: the poverty cutoff k. This cutoff represents the minimum deprivation score required for a household to be classified as poor. The identification function ${\rho }_k\left(y_i;z\right)$ is defined as:

${\rho }_k\left(y_i;z\right)=\{\mathrm{1,}if{c}_i\ge k\mathrm{0,}if{c}_i<k$ (2)

for each household i. Households with $c_i\ge k$ are identified as multidimensionally poor, while those with $c_i<k$ are considered non-poor.

This “dual cutoff” approach lies between the union criterion (which classifies a household as poor if it is deprived in at least one dimension) and the intersection criterion (which requires deprivation in all dimensions), offering a more realistic identification of multidimensional poverty.

Once the poor are identified, the AF method applies a censoring step. The censored deprivation matrix $g^0\left(k\right)$ is obtained by multiplying each element of $g^0$ by the identification function: $g_{ij}^0\left(k\right)=g_{ij}^0{\rho }_k\left(y_i;z\right)$. Thus, if household i is identified as poor $\left({\rho }_k\left(y_i;z\right)=1\right)$, its deprivations remain unchanged in all dimensions. If it is not poor $\left({\rho }_k\left(y_i;z\right)=0\right)$, all its deprivations are set to zero. The censored deprivation score for each poor household is then: $c_i\left(k\right)={\displaystyle \sum }_{j=1}^d{w}_j{g}_{ij}^0\left(k\right)$.

4.2.1 Aggregation: the adjusted headcount ratio $M_0$

The aggregation step builds on the intuition of the FGT class of poverty measures (Foster; Greer; Thorbecke, 1984), interpreting the multidimensional index as the mean of a transformed deprivation vector. The adjusted headcount ratio $M_0$ is defined as the average of the censored deprivation scores:

$M_0\left(Y;z\right)=\mu \left[c\left(k\right)\right]=\frac{1}{n}{\displaystyle \sum }_{i=1}^n{c}_i\left(k\right)$ (3)

This index can be expressed as the product of two components (Alkire et al., 2015): the incidence of poverty H and the intensity of poverty A. Let q denote the number of multidimensionally poor households. The incidence is given by:

$H=\frac{q}{n}$, which represents the proportion of households identified as poor. The intensity A is the average share of deprivations among the poor, defined as:

$A=\frac{1}{q}{\displaystyle \sum }_{i:{\rho }_k\left(y_i;z\right)}^{}{c}_i\left(k\right)$ (4)

Therefore, the adjusted headcount ratio can be written as: $M_0=H\times A$. While H captures how widespread multidimensional poverty is, A reflects its severity by indicating the fraction of weighted deprivations experienced by poor households. The product $M_0=H\times A$ satisfies dimensional monotonicity: whenever a poor household overcomes a deprivation in any dimension, its censored deprivation score decreases, which reduces A and, consequently, lowers $M_0$.

An important advantage of the AF methodology is that $M_0$ is decomposable by population subgroups and by dimensions. The overall poverty level of a society can be expressed as a population-weighted sum of subgroup poverty levels, and the dimensional contributions $M_0$ reveal which deprivations are most responsible for multidimensional poverty. This allows for a detailed diagnosis of poverty patterns and a direct link between the multidimensional index and policy priorities in the regional context analyzed in this article.

5 Analysis and discussion

The MPI-Trajectories for the Legal Amazon is disaggregated into urban/rural areas at the municipal level for the 2000 and 2010 census years, with 792 and 807 municipalities, respectively. This analysis provides more information about the behavior of choices related to dimensions and variables for urban and rural areas. Finally, table 3 presents a descriptive synthesis at the state level.

Table 3:
Descriptive statistics of the MPI-Trajectories for the Legal Amazon, by State, 2000 and 2010.

State	2000			2010
	MPI	MPI Urban	MPI Rural	MPI	MPI Urban	MPI Rural
Acre	0.44	0.33	0.46	0.09	0.02	0.14
	(0.06)	(0.06)	(0.06)	(0.06)	(0.007)	(0.08)
Amazonas	0.46	0.38	0.49	0.1	0.023	0.16
	(0.06)	(0.08)	(0.05)	(0.05)	(0.015)	(0.06)
Amapá	0.35	0.29	0.4	0.04	0.016	0.09
	(0.06)	(0.08)	(0.06)	(0.03)	(0.011)	(0.04)
Maranhão	0.47	0.42	0.46	0.07	0.03	0.09
	(0.06)	(0.06)	(0.05)	(0.03)	(0.014)	(0.03)
Mato Grosso	0.3	0.25	0.28	0.02	0.008	0.05
	(0.07)	(0.05)	(0.08)	(0.03)	(0.004)	(0.05)
Pará	0.44	0.37	0.43	0.07	0.021	0.1
	(0.07)	(0.06)	(0.07)	(0.04)	(0.012)	(0.05)
Rondônia	0.37	0.28	0.36	0.02	0.009	0.04
	(0.07)	(0.05)	(0.05)	(0.01)	(0.004)	(0.02)
Roraima	0.35	0.23	0.37	0.12	0.019	0.16
	(0.07)	(0.07)	(0.05)	(0.09)	(0.031)	(0.08)
Tocatins	0.4	0.34	0.41	0.04	0.016	0.07
	(0.07)	(0.07)	(0.07)	(0.03)	(0.009)	(0.04)

• Source: Research data. Standard deviation in parentheses

The indicators, in this case, are different due to the necessary decision to use different variables to compose the dimensions related to Urban and Rural, and between the years, with changes in the questionnaire of the Demographic Censuses samples. However, to a certain extent, they are comparable indicators, as most of the composition of the indicators comes mainly from the same variables. Therefore, some comparisons will be made, considering the sensitivity in some interpretations.

The main highlight is the weight that the rural environment provides in the composition of the general indicator of the MPI-Trajectories. In the year 2000, the general MPI-Trajectories was 0.41, emphasizing the states of Maranhão, Amazonas, and Pará, which presented the highest average MPIs, as they are the most populous states in the Legal Amazon, which ended up influencing the composition of the indicator. On average, the rural MPI in all states was higher than the urban one. This aspect is due to the punishment that the Census variables generate in the equation between the two environments.

It is observed that the Census is still a base that needs to capture essential aspects of the rural environment adequately. In particular, in the Amazon, the questionnaire does not provide options to better observe the living and consumption patterns in households associated with rural productive systems operated by peasant agents and their microeconomic rationalities (Costa, 2019; Costa et al., 2022). For example, comparing rural and urban areas in 2000, the State that presented the smallest distance was Mato Grosso. That is, the more significant influence of rural areas in the State (where 20.64% of the population lived) indicates that being part of the rural areas increases the possibility of classifying a household as multidimensionally poor. These first results point to a need to rethink the variables for the rural context used in this first experiment.

For the year 2010, with the possibility of using other variables now present in the questionnaire, and the substantial reduction in poverty that occurred between 2000 and 2010, there was a sudden drop in the absolute result of the MPI-Trajectories, indicating that the configuration of the new variables contributed to this reduction. This reduction was observed in other studies comparing censuses using the same variables and dimensions11. The average value for the general MPI in 2010 was 0.06, with the rural having an average value of 0.09 and the urban one of 0.02. Once again, Amazonas, Maranhão, and Pará were the states that presented the highest relative level of multidimensional poverty in the Legal Amazon. However, as the composition of the MPI is based on the product of the incidence (H) of poverty multiplied by its intensity (A), it is interesting to analyze the impact of multidimensional poverty based on these indicators. Table 4 shows the results for the incidence of poverty by State in the Legal Amazon.

Serra (2017) observed that the MPI dropped from a level of 0.388 to 0.232 for the Brazilian Rural from 2000 to 2010 using similar indicators between the two years analyzed.

Table 4:
Multidimensional Incidence of Poverty (% of Poor) in the Legal Amazon (2000-2010) by State.

States	2000			2010
	Incidence	Incidence Urban	Incidence Rural	MPI	Incidence Urban	Incidence Rural
Acre	0.86	0.74	0.93	0.27	0.07	0.4
	(0.09)	(0.11)	(0.08)	(0.18)	(0.02)	(0.24)
Amazonas	0.89	0.8	0.96	0.33	0.07	0.51
	(0.09)	(0.12)	(0.06)	(0.14)	(0.05)	(0.16)
Amapá	0.79	0.69	0.87	0.14	0.05	0.28
	(0.11)	(0.17)	(0.08)	(0.08)	(0.04)	(0.11)
Maranhão	0.92	0.86	0.94	0.21	0.11	0.28
	(0.07)	(0.09)	(0.07)	(0.09)	(0.04)	(0.11)
Mato Grosso	0.72	0.64	0.68	0.07	0.03	0.15
	(0.12)	(0.11)	(0.16)	(0.09)	(0.01)	(0.14)
Pará	0.88	0.81	0.89	0.22	0.07	0.32
	(0.10)	(0.10)	(0.10)	(0.14)	(0.04)	(0.15)
Rondônia	0.82	0.69	0.84	0.07	0.03	0.14
	(0.12)	(0.11)	(0.08)	(0.04)	(0.01)	(0.07)
Roraima	0.75	0.57	0.82	0.35	0.06	0.49
	(0.13)	(0.14)	(0.09)	(0.25)	(0.01)	(0.22)
Tocantins	0.85	0.77	0.9	0.12	0.05	0.22
	(0.10)	(0.11)	(0.09)	(0.08)	(0.03)	(0.13)

• Source: Research Data. Standard deviation in parentheses

When comparing the periods, a substantial reduction in Multidimensional poverty is observed in all states, which, even if they have different compositions in the variables from one year to another, show the substantial reduction that occurred between the analyzed period. In 2000, about 85% of the households in the Legal Amazon were considered poor due to the structure and composition established by the MPI-Trajetórias. For the same year, Serra (2017) obtained a value close to 47% for the North region. This difference occurs because the MPI-Trajectories use more variables and more dimensions. Serra (2017) used only eight variables distributed in 2 dimensions.

Even with the different configurations, rural households weigh more in the composition of the percentage of poor people. While in the urban environment, 77.16% of the households in the Legal Amazon were considered multidimensionally poor, in the rural environment, they were 87.35%. The highlight was the State of Maranhão, where 92% of its population was considered multidimensionally poor, with 86% of the urban population and 94% of the rural population.

In 2010 the reduction was significant. Overall, 17.6% were considered multidimensionally poor, 6.7% of the urban population, and 27% of the rural population. The State of Amazonas was the one that most contributed to the number of poor in the Legal Amazon, with 33% of its population living in multidimensional poverty. The state also showed the largest urban-rural gap: while only 7% of urban residents were multidimensionally poor, 51% of rural residents fell into that category.

Observing the relationship between the incidence and intensity of poverty (Figure 1), the reduction of the MPI-Trajectories was more significant for poorer municipalities. In 2000, the relationship between intensity and incidence indicated a large number of poor people (incidence H) having between 40% and 60% of deprivations (intensity A). In the rural environment, there is a large number of poor with an incidence above 70% and intensity concentrating between 45% to 60% of deficiencies. In urban areas, this logic is reversed. Most of the poor are in municipalities with lower incidence and intensity, around 40%. This situation reflects, in part, the great difficulty in assessing the situation of deprivation and needs in the rural Amazon. Even with the possible adjustments, considering only the information present in the 2000 and 2010 census sample questionnaires, this first experiment indicates that it is urgent to exhaust the exploration of this database and, at the same time, to investigate some methodological aspects related to the sensitivity of the A.F. method with situations of use of dimensions, with a different number of variables, and different variables for urban and rural areas.

Figure 1:
Values of Intensity (A) and Incidence (H) of the municipalities of the Legal Amazon, by U.F., 2000-2010 for the MPI-General Trajectories and its Urban and Rural components.

In 2010, in both environments, the impact that a set of social policies, implemented together in the period (2000-2010), generated in the reduction of multidimensional poverty is evident. Even though our annual indicators are different, that is, they are not directly comparable, it is possible to verify that, in poorer regions, the different policies to combat poverty had a significant impact on the dimensions of the population in poverty and on the intensity of multidimensional poverty in the region. While in 2000 most municipalities had more than 80% of their population classified as multidimensionally poor, by 2010 this share had dropped to around 40%, and in urban areas it rarely exceeded 30%. What is most striking in the results, considering the General Index, is the reduction in poverty intensity, which fell to around 32%, below the level observed in 2000.

Figure 2 presents a set of maps where we can see, for the municipalities of the Legal Amazon, the results obtained in this first version of the MPI-Trajetórias. Each Figure (a), (b), and (c) shows a set of maps with the Multidimensional Poverty Intensity (H) and the Multidimensional Poverty Intensity (A) for the General, Urban, and Rural components, respectively, of the MPI- Trajectories for the years 2000 and 2010. It is necessary to remember that these maps of H and A for the two years are not strictly comparable.

Figure 2:
(a) Incidence (H) and Intensity (A) of Multidimensional Poverty in the Municipalities of the Legal Amazon, 2000-2010 for the MPI-Trajectories General; (b) Incidence (H) and Intensity (A) of Multidimensional Poverty in the Municipalities of the Legal Amazon, 2000-2010 for the MPI-Trajectories Urban component; (c) Incidence (H), and Intensity (A) of Multidimensional Poverty in the Municipalities of the Legal Amazon, 2000-2010 for the MPI-Trajectories Rural component.

The results reported in Tables 5 and 6 show that the MPI is highly sensitive to the choice of variables that enter each dimension. This sensitivity emerges clearly when we compare the “Conventional Indicator” (CI), which keeps the set of variables used in 2000 unchanged for the 2010 exercise, with the IPM‑Trajectories (“Trajectories Indicator”, TI), which allows the set of variables to evolve in order to better reflect the specific living conditions in the Legal Amazon. In the rural case, the CI suggests a reduction of 63.9% in multidimensional poverty between 2000 and 2010, while the TI indicates a much larger decline of 78.9%. A similar pattern is observed for urban areas, where the decline in multidimensional poverty is 87.3% under the CI and 94.4% under the TI.

These differences are not merely statistical. They arise from substantive changes in the way deprivation is captured, especially in the Living Conditions dimensions. In rural areas, the TI incorporates an additional indicator in the “Housing and Collective Consumption Goods” subdimension-permanent private housing on ceded land or in another tenure condition-which is particularly relevant in agrarian contexts marked by customary land tenure and peasant production systems. In the “Work and Private Consumer Goods” subdimension, the TI also refines the indicator of access to information and communication technologies by including the radio as a relevant asset alongside television and microcomputers. These adjustments make the rural MPI less dependent on an urban‑biased bundle of assets and housing standards, and more consistent with the socio‑economic trajectories that characterize the agrarian space of the Amazon.

The contrast between TI and CI is especially striking in 2000. For that year, there is virtually no difference between the two indicators in urban areas, but the rural TI is about 17% lower than the rural CI (Table 5). This suggests that conventional specifications, which rely heavily on variables designed to capture urban living conditions, tend to inflate rural multidimensional poverty when applied mechanically to the Amazon. By contrast, when the indicator is recalibrated to include variables that are more meaningful for rural livelihoods, the level of measured poverty decreases, although it remains substantially higher than in urban areas. In other words, the very narrative that associates “rural Amazonia” with chronic and homogeneous poverty is, at least in part, a product of the way multidimensional poverty is measured.

Table 5:
Average of MPI Indicators and analysis of differences in MPI Urban Index Averages 2000-2010

Indicators	MPI Rural Index
	2000		2010		diff
	TI	CI	TI	CI	TI	CI
mpi	0.41	0.48	0.09	0.18	-78.90%	-63.89%
	(17.664)***		(24.53)***		(21.09)***
h	0.87	0.96	0.27	0.55	-69.79%	-42.79%
	(137.37)***		(53.26)***		(25.38)***
a	0.47	0.51	0.33	0.32	-27.91%	-32.94%
	(24.66)***		(105.88)***		(-22.12)***

• Source: Research Data. Significance codes: 0.000 (***) Significant at 0.01%, 0.001 (**) Significant at 0.1%, 0.01 (*) Significant at 1%, 0.05 (.) Significant at 5%, 0.1 ( ) Significant at 10%. Source: Own elaboration.

Table 6:
Average of MPI Indicators and analysis of differences in MPI Urban Index Averages 2000-2010

12	MPI Urban Index
	2000		2010		diff
	TI	CI	TI	CI	TI	CI
mpi	0,35	0,35	0,02	0,05	-94,43%	-87,32%
	(0.00)		(18.54)***		(22.03)***
h	0,77	0,77	0,07	0,16	-91,67%	-80,07%
	(0.00)		(31.09)***		(22.23)***
a	0,45	0,45	0,3	0,29	-31,71%	-30,93%
	(0.00)		(398.56)***		(-10)

• Source: Research Data. Significance codes: 0.000 (***) Significant at 0.01%, 0.001 (**) Significant at 0.1%, 0.01 (*) Significant at 1%, 0.05 (.) Significant at 5%, 0.1 ( ) Significant at 10%. Source: Own elaboration.

Taken together, Tables 5 and 6 highlight two key implications for the trajectory of the MPI in the Legal Amazon. First, they show that the downward trend in multidimensional poverty between 2000 and 2010 is robust to alternative specifications, but its magnitude depends on whether the measurement strategy is able to incorporate transformations in infrastructure, access to services and diffusion of technology that actually occurred in the period. Second, they underline the importance of explicitly designing rural‑sensitive indicators: when variables that reflect urban lifestyles are used as the implicit norm, rural deprivation tends to be overestimated; when rural‑specific variables and thresholds are introduced, the MPI portrays a more nuanced picture, in which poverty remains severe and widespread in rural areas, but is not immutable.

This sensitivity of the MPI to the underlying set of variables reinforces the need to carefully specify indicators that adequately capture rural specificities in the Legal Amazon. Otherwise, multidimensional poverty in rural areas may be systematically overestimated.

6 Final Considerations

In general, in constructing economic and social indicators, indicators of productivity, consumption, and purchasing power (both evaluated in the short term) are considered sufficient to assess populations' well-being levels. When evaluated in terms of the poverty category, these parameters are associated with the processes that generate economic and social development. The concept of poverty has been thought of as an antithesis category, which aims to demonstrate, by negation, the lack of significant advances in economic development processes. Where there is poverty, in this sense, there is no development.

Precisely for this reason, it is fundamental for the debate on development alternatives for the Brazilian Amazon, the debate on methodologies for the construction of poverty indicators, and the impact of these measures to think about alternatives for the design of policies oriented to the possibilities of certain regional economies referred to the biome.

When the meaning of poverty is established as a parameter for substantive evaluation of the degrees of freedom necessary to estimate people's human development (SEN, 2000), the ideal would be to build several indicators with the potential to assess the continuous and multidimensional process of advancement of the territorial conditions of development of their populations, in an integrated way to the production of their livelihoods. Although limited, the alternative to this "ideal" has been the use of a family of indicators that consider the multidimensional aspects of poverty and development, going beyond components linked only to consumption or individual income levels. In this family, the MPI, whose methodology is based on Sen's (1976) capabilities approach and the A.F. decomposition method (Alkire and Foster, 2011), presents itself with substantial potential in two aspects: the first, by the association with the UNDP, which strengthens its presence as a marker in global development-related agendas, such as the 2030 Agenda of the SDGs and the New Urban Agenda of UN-Habitat. Moreover, the second is due to its methodological aspect, which takes an axiomatic option for constructing the measure and has the flexibility to accommodate different dimensions and variables. This methodology creates conditions for incorporating the debate on the dimensions and variables that need to be present in the information systems that feed national statistics.

The preliminary results of this experiment, which generated the MPI-Trajectories, reveal several critical opportunities and limitations that need to be addressed if we are to continue considering the MPI as a framework for discussing poverty and development in the Brazilian Amazon. An explicit limitation observed relates to the measure for the rural component of the MPI-Trajectories. Assessing this measure, which represents poverty in rural areas of the Amazon, is essential. Although, with a better qualification for the proposed variables that made up the dimensions of rural poverty, it was impossible to go beyond the information available in the sample questionnaires. As the questionnaire does not capture specific parameters necessary for the characterization of peasant households and individuals, very present in the agrarian space of the Amazon (Costa, 2019), the rural poverty measure has a high intensity (value A - poor with many needs).

The limitations are even more remarkable when it is necessary to consider the actual constitution of deprivations in the agrarian space of the Legal Amazon. In the Amazon, there are competing rural techno-productive trajectories. Understanding their dynamics is essential to designing a regional development project that includes poverty reduction and conservation aspects of the forest landscape. The identification and characterization of these rural techno-productive trajectories in the Amazon were carried out by Costa (2009). In a recent study, Costa (2020) presents and discusses agrarian dynamics in the context of technological variants linked to these systems, which depend on land or labor and use mechanical-chemical components or make use of historically accumulated technologies of forest resources management. The results presented covering three periods, accompanied by agricultural censuses, demonstrate the weight that the trajectories based on temporary crops and beef cattle have taken in the region and the risks and structural changes that this has implied for the biome. The results also demonstrate the counterpoint to these trajectories, represented by trajectories based on agroforestry systems and permanent crops in the context of peasant systems.

Even so, this first experiment with the MPI-Trajetórias showed opportunities in this field of narrative dispute for a development model that fully recognizes the Amazon agrarian space and some of its protagonists. In the decade from 2000 to 2010, we observed that the general idea of the direct association of rural Amazonia with the persistence of poverty is dubious. The incorporation of new variables to measure deprivations in rural areas, which took place after a period (the 2000s to 2010s) with the implementation of programs and policies that also looked at the economies of rural techno-productive systems associated with agroforestry systems conducted by peasant agents, shows, in the eastern Amazon region, where these are still the dominant systems, an evolution in the H and A components of multidimensional poverty.

These first results must be better evaluated and analyzed. Furthermore, the limitations regarding the database and the decomposition method, when different compositions are adopted for urban and rural areas and, in some instances, the use of different bases, need to be better clarified. In this way, a new proposal dealing specifically with the rural environment that has just appeared on the global stage through the Food and Agriculture Organization with OPHI (FAO-OPHI, 2022) can help.

However, for a sustainable development agenda in the Amazon, it is essential to take the economies related to the biome seriously as strategies for a proposal for regional development that is socially just and environmentally responsible. It is possible, even with this incipient experiment, to say that looking for better indicators of poverty, in particular of its component in the rural space, more than necessary, is imperative because we can recognize that our poverty indicators contain problems and must be rethought, or we will have to choose between saving the indicators or saving the forest and the people who live in it.

Beyond the methodological debate, the results presented here have direct implications for the design of public policies in the Legal Amazon. The sharp reduction in multidimensional poverty between 2000 and 2010, particularly in the intensity component, is strongly associated with a period of expansion of social policies and income transfer programs in Brazil. However, the IPM‑Trajectories also shows that many of the deprivations that persist are structurally linked to deficits in basic infrastructure, access to quality public services and opportunities for stable and adequately remunerated work, especially in rural areas. In other words, social policies that alleviate immediate privations are necessary, but they are not sufficient to permanently overcome multidimensional poverty in the region.

In the Brazilian case, social programs have often been adopted under the influence or encouragement of international organizations, without being consistently articulated to a broader, sovereign national development strategy. This tends to generate a pattern in which policies focused on social protection coexist with relatively low levels of productive investment and with the maintenance of a primary‑export growth model that reproduces structural inequalities and limits productivity gains. In such a context, the risk of a “middle‑income trap” - in which countries spend long periods with modest productivity growth and limited diversification - becomes more salient. International experience suggests that sustained reductions in poverty have usually combined social policies with high rates of productive investment, structural transformation and massive job creation, rather than relying exclusively on income transfer programs.

From the perspective of Amartya Sen’s capability approach, the deprivations captured by the MPI‑Trajectories - in education, health and living conditions - are manifestations of a broader restriction of substantive freedoms. Overcoming these restrictions in a lasting way requires the integration of social policies with other public policies (industrial, agricultural, environmental, infrastructure and regional policies) within a long‑term national development project. The very logic of the MPI already points in this direction: by combining dimensions that are infrastructural in nature and that cannot be improved without higher levels of productive investment, the index makes explicit that the fight against poverty in the Legal Amazon cannot be reduced to targeted social programs. It depends on a development strategy capable of simultaneously expanding people’s capabilities, diversifying local economies and preserving the forest and the socio‑environmental systems on which these economies depend.

Agradecimentos

Os editores agradecem aos autores pela colaboração com a Revista Nova Economia. Os autores agradecem ao Centro de Síntese em Biodiversidade e serviços ecossistêmicos do CNPq (CNPq-Sinbiose) pelo financiamento (Processo 442357/2019-2). À União Europeia (Ações Marie Sklodowska-Curie, acordo de subvenção 101109642), que financiou a pesquisadora Raquel M. Lana.

References

• ALKIRE, S.; FOSTER, J. Counting and multidimensional poverty measurement. Journal of Public Economics, v. 95, n. 7-8, p. 476-487, 2011.
• ALKIRE, S.; FOSTER, J. E.; SETH, S.; SANTOS, M. E.; ROCHE, J. M.; BALLON, P. Multidimensional Poverty Measurement and Analysis: Chapter 3 - Overview of Methods for Multidimensional Poverty Assessment WORKING PAPER NO. 84. Oxford: Oxford Poverty & Human Development Initiative (OPHI), 2015.
• ALKIRE, S.; FOSTER, J. E.; SETH, S.; SANTOS, M. E.; ROCHE, J. M.; BALLON, P. Multidimensional Poverty Measurement and Analysis: Chapter 5 - The Alkire-Foster Counting Methodology WORKING PAPER NO. 86. Oxford: Oxford Poverty & Human Development Initiative (OPHI), 2015a.
• ALKIRE, S.; KANAGARATNAM, U.; SUPPA, N. The Global Multidimensional Poverty Index (MPI): 2018 Revision OPHI MPI Methodological Notes, v. 46, p. 1-31, 2018.
• ALKIRE, S.; KANAGARATNAM, U.; NOGALES, R.; SUPPA, N. Revising the global Multidimensional Poverty Index: Empirical insight and robustness OPHI Research in Progress 56a. Oxford: Oxford Poverty and Human Development Initiative, University of Oxford, 2020.
• ALKIRE, S.; SANTOS, M. E. A Multidimensional Approach: Poverty Measurement & Beyond. Social Indicators Research, v. 112, p. 239-257, 2013.
• BAGOLIN, I. P.; ÁVILA, R. P. Poverty distribution among the Brazilian states: A multidimensional analysis using capabilities and needs approaches In: ENCONTRO NACIONAL DE ECONOMIA, 34., 2006, Salvador. Anais [...]. Salvador: ANPEC, 2006. p. 39-50. Disponível em: http://www.anpec.org.br/encontro2006/artigos/A06A142.pdf Acesso em: 31 jul. 2019.
  » http://www.anpec.org.br/encontro2006/artigos/A06A142.pdf
• BANERJEE, A. V.; DUFLO, E. Poor Economics: A Radical Rethinking of the Way to Fight Global Poverty New York: PublicAffairs, 2011. ISBN 978-1-58648-798-0.
• BARROS, R. P.; CARVALHO, M.; FRANCO, S. Pobreza multidimensional no Brasil. Rio de Janeiro: Ipea, 2006. (Texto para Discussão, n. 1227).
• BOLTVINIK, J. Poverty Measurement and Alternative Indicators of Development. In: VAN DER HOEVEN, R.; ANKER, R. (eds.). Poverty Monitoring: An International Concern London: Palgrave Macmillan, 1994. https://doi.org/10.1007/978-1-349-23134-8_4
  » https://doi.org/10.1007/978-1-349-23134-8_4
• BOLTVINIK, J. Métodos de medición de la pobreza. Conceptos y tipología. Socialis. Revista Latinoamericana de Política Social, Buenos Aires, n. 1, out. 1999.
• BOLTVINIK, J. Medición multidimensional de pobreza. América Latina de precursora a rezagada. Revista Sociedad & Equidad, n. 5, jan. 2013.
• BUAINAIN, A. M.; DEDECCA, C. S.; NEDER, H. D. Características regionais da pobreza rural no Brasil: algumas implicações para políticas públicas. In: BUAINAIN, A. M.; DEDECCA, C. (Org.). A nova cara da pobreza rural: desenvolvimento e a questão regional Brasília: IICA, 2013. cap. 2. (Série Desenvolvimento Rural Sustentável, v. 17).
• CLACSO/CROP. Multidimensionalidad de la pobreza: propuestas para su definición y evaluación en América Latina y el Caribe 1. ed. Ciudad Autónoma de Buenos Aires: CLACSO, 2014. E-book.
• CODEÇO, C. T. et al. Epidemiology, Biodiversity, and Technological Trajectories in the Brazilian Amazon: From Malaria to COVID-19. Frontiers in Public Health, v. 9, art. 647754, 2021. doi: 10.3389/fpubh.2021.647754.
  » https://doi.org/10.3389/fpubh.2021.647754
• COSTA, F. A. Trajetórias tecnológicas como objeto de política de conhecimento para a Amazônia: Uma metodologia de delineamento. Revista Brasileira de Inovação, Campinas, v. 8, p. 287-312, 2009.
• COSTA, F. A. A brief economic history of the Amazon (1720-1970) Cambridge: Cambridge Scholars Publishing, 2019. 348 p.
• COSTA, F. A. Economia camponesa referida ao bioma da Amazônia: atores, territórios e atributos. Paper do NAEA, v. 1, n. 2, ed. 476, 2020.
• COSTA, F. A. et al. Chapter 15: Complex, Diverse and Changing Agribusiness and Livelihood Systems in the Amazon In: NOBRE, C. et al. (Eds.). Amazon Assessment Report 2021. New York: United Nations Sustainable Development Solutions Network, 2021. Disponível em: https://www.theamazonwewant.org/spareports/. DOI: 10.55161/CGAP7652.
  » https://www.theamazonwewant.org/spareports
• COSTA, F. A. Structural diversity and change in rural Amazonia: A comparative assessment of the technological trajectories based on agricultural censuses (1995, 2006 and 2017). Nova Economia, v. 31, n. 2, 2021.
• DA ROSA, S. S.; BAGOLIN, I. P.; FERNANDES, D. A.; RODRIGUES, D. L. Multidimensional Poverty and its Relationship with Economic, Territorial and Population Dynamics: An Analysis of the Northern Region of Brazil from 1991 to 2010 In: CONFERENCE OF THE LATIN AMERICAN AND CARIBBEAN ASSOCIATION FOR HUMAN DEVELOPMENT AND THE CAPABILITIES APPROACH, 7., 2019, Cholula. Anais [...]. Cholula, 2019.
• DESAI, M.; SEN, A.; BOLTVINIK, J. Social Progress Index. A Proposal Regional Project To Overcome Poverty in Latin America and the Caribbean, (RLA/86/004). DE 138. UNDP, U.N., 1994.
• FAO-OPHI. Measuring rural poverty with a multidimensional approach: The Rural Multidimensional Poverty Index Rome: FAO, 2022. (FAO Statistical Development Series, n. 19). https://doi.org/10.4060/cb8269e
  » https://doi.org/10.4060/cb8269e
• FOSTER, J.; GREER, J.; THORBECKE, E. A class of decomposable poverty measures. Econometrica, v. 52, n. 3, p. 761-766, 1984.
• FURTADO, C. O mito do desenvolvimento econômico Rio de Janeiro: Paz e Terra, 1974.
• GEORGE, H. Progress and Poverty: An Inquiry into the Cause of Industrial Depressions and of Increase of Want with Increase of Wealth: The Remedy New York: D. Appleton and Company, 1881. (1st ed. 1879).
• IBGE. PORTARIA Nº PR-254, DE 25 DE AGOSTO DE 2020 Diário Oficial da União, Brasília, 27 ago. 2020. Edição: 165, Seção: 1, p. 71.
• INE. Boletín Técnico: Índice de Pobreza Multidimensional (MPI) para Paraguay Asunción: Instituto Nacional de Estadística, Gobierno Nacional, 2021.
• IPCC-AR6. Summary for Policymakers. In: MASSON-DELMOTTE, V. et al. (eds.). Global Warming of 1.5°C. An IPCC Special Report on the impacts of global warming of 1.5°C above pre-industrial levels and related global greenhouse gas emission pathways, in the context of strengthening the global response to the threat of climate change, sustainable development, and efforts to eradicate poverty: IPCC, 2018. In Press.
• IPCC-AR6. Climate Change 2021: The Physical Science Basis. Contribution of Working Group I to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change Technical Summary. In: MASSON-DELMOTTE, V. et al. (eds.). Cambridge: Cambridge University Press, 2021. p. 33-144. doi:10.1017/9781009157896.002.
  » https://doi.org/10.1017/9781009157896.002.
• IPCC-AR6. Climate Change 2022: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change In: PÖRTNER, H.-O. et al. (eds.). Cambridge: Cambridge University Press, 2022a. In Press.
• IPCC-AR6. Summary for Policymakers. In: PÖRTNER, H.-O. et al. (eds.). Climate Change 2022: Impacts, Adaptation, and Vulnerability. Contribution of Working Group II to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change Cambridge: Cambridge University Press, 2022b. In Press.
• JANNUZZI, P. M.; FONSECA, J. C. G. Pobreza monetária e multidimensional no Brasil: estimativas aprimoradas para análise do papel das políticas sociais de 1992 a 2014. Revista Ciências do Trabalho, n. 7, 2017.
• JANNUZZI, P. M.; DE SOUSA, M. F. Contribuição para a agenda de objetivos de desenvolvimento sustentável: Indicador de Pobreza Multidimensional para América Latina inspirada em proposta preliminar da Cepal Estudo Técnico n.º 13/2016. Brasília: Secretaria de Avaliação e Gestão da Informação (SAGI), Ministério do Desenvolvimento Social e Combate à Fome-MDS, 2016.
• KABEER, N. Beyond the Poverty Line: Measuring Poverty and Impoverishing Measures In: ______. Reversed Realities: Gender Hierarchies in Development Thought. London: Verso, 1994. p. 136-162.
• KAGEYAMA, A.; HOFFMANN, R. Pobreza no Brasil: uma perspectiva multidimensional. Economia e Sociedade, Campinas, v. 15, n. 1, p. 79-112, jan./jun. 2006.
• MAHBUB UL HAQ. Reflections on Human Development Oxford: Oxford University Press, 1996. 288 p.
• MAXWELL, S. The meaning and measurement of poverty ODI Poverty Briefing, 1999. Disponível em: www.odi.org.uk/briefing/pov3.html
  » www.odi.org.uk/briefing/pov3.html
• NAMIBIA STATISTICS AGENCY. Namibia Multidimensional Poverty Index (MPI) Report 2021. Windhoek, 2021. p. 1-68.
• OPHI; UNDP. Unmasking disparities by ethnicity, caste and gender. Global Multidimensional Poverty Index Oxford: Oxford Poverty & Human Development Initiative; New York: UNDP, 2021.
• OPHI-UNDP. Global Multidimensional Poverty Index 2018: The Most Detailed Picture To Date of the World's Poorest People Oxford: University of Oxford, 2018.
• RAVALLION, M. Issues in measuring and modelling poverty. Economic Journal, v. 106, n. 438, p. 1328-1340, 1996.
• RAVALLION, M. Toward better global poverty measures. Journal of Economic Inequality, v. 14, n. 2, p. 227-248, 2016.
• RODRIGUES, A. S. L. et al. Boom-and-Bust Development Patterns across the Amazon Deforestation Frontier. Science, v. 324, p. 1435-1439, 2009. DOI: 10.1126/science.1174002.
  » https://doi.org/10.1126/science.1174002
• RODRIGUES, D. L.; SANTOS, R. B. N.; BAGOLIN, J. P.; FERNANDES, D. A. Pobreza multidimensional intraurbana na região metropolitana de Belém. Redes, v. 25, p. 2251-2273, 2020.
• RODRIGUES, D. L. Pobreza multidimensional, território e meios de vida na região da Ilha das Onças, município de Barcarena-PA 2019. Tese (Doutorado em Economia) - Programa de Pós-Graduação em Economia, Universidade Federal do Pará, Belém.
• RODRIGUES, D. L.; SANTOS, R. B. N.; FERNANDES, D. A. Pobreza e desenvolvimento humano na região metropolitana de Belém: uma análise comparativa entre os indicadores linha de pobreza, IDHM e MPI para os anos de 2000 e 2010 In: PRÊMIO ATLAS DO DESENVOLVIMENTO HUMANO NO BRASIL: COLETÂNEA DE ARTIGOS. Brasília: PNUD: IPEA: FJP, 2015. p. ISBN: 978-85-88201-30-9.
• ROSA, S. S. Pobreza multidimensional e dinâmica espaço-temporal na Região Norte do Brasil 2021. Tese (Doutorado em Economia do Desenvolvimento) - Programa de Pós-Graduação em Economia do Desenvolvimento, PUCRS, Porto Alegre.
• RORATO, A. C.; DAL'ASTA, A. P.; LANA, R. M. et al. Trajetorias: a dataset of environmental, epidemiological, and economic indicators for the Brazilian Amazon. Scientific Data, v. 10, art. 65, 2023. https://doi.org/10.1038/s41597-023-01962-1
  » https://doi.org/10.1038/s41597-023-01962-1
• SANTOS, M. E. Challenges in designing national multidimensional poverty measures Statistics Series, n. 100 (LC/TS.2019/5). Santiago: Economic Commission for Latin America and the Caribbean (ECLAC), 2019.
• SANTOS, R. B. N.; RODRIGUES, D. L.; FERNANDES, D. A.; SANTOS, F. T. P. Diferenças da pobreza multidimensional urbana e rural na Amazônia Legal entre 2000 e 2010 In: CONGRESSO DA SOCIEDADE BRASILEIRA DE ECONOMIA, ADMINISTRAÇÃO E SOCIOLOGIA RURAL, 54., 2016, Alagoas. Anais [...]. Alagoas: SOBER, 2016.
• SCHULTZ, T. W. Investment in Human Capital. The American Economic Review, v. 51, n. 1, p. 1-17, mar. 1961.
• SEN, A. Poverty: An Ordinal Approach to Measurement. Econometrica, v. 44, n. 2, p. 219-231, 1976.
• SEN, A. Poverty and famines: an essay on entitlement and deprivation Oxford: Oxford University Press, 1981.
• SEN, A. Commodities and capabilities Amsterdam: North Holland, 1985.
• SEN, A. Desenvolvimento como liberdade São Paulo: Companhia das Letras, 2000.
• SEN, A. Desigualdade reexaminada Rio de Janeiro: Record, 2001.
• SERRA, A. S.; YALONETZKY, G. I.; MAIA, A. G. Multidimensional Poverty in Brazil in the Early 21st Century: evidence from the Demographic Census : Springer, 2020.
• SERRA, A. S. Pobreza multidimensional no Brasil rural e urbano 2017. 161 f. Tese (Doutorado em Desenvolvimento Econômico) - Universidade Estadual de Campinas, Campinas.
• THORBECKE, E.; FOSTER, J.; GREER, J. Class of Decomposable Poverty Measures. Econometrica, v. 52, n. 3, p. 761-766, 1984.
• U.N. Transforming our world: the 2030 Agenda for Sustainable Development Resolution adopted by the General Assembly on 25 September 2015. United Nations. A/RES/70/1. Disponível em: https://www.un.org/ga/search/view_doc.asp?symbol=A/RES/70/1& Lang=E . Acesso e: 22 maio 2022.
  » https://www.un.org/ga/search/view_doc.asp?symbol=A/RES/70/1& Lang=E
• UNDP. Human Development Report 1990: Concept and Measurement of Human Development New York; Oxford: Oxford University Press, 1990. Disponível em: http://www.hdr.undp.org/en/reports/global/hdr1990
  » http://www.hdr.undp.org/en/reports/global/hdr1990
• UNDP. Human Development Report 1997: Human Development to Eradicate Poverty New York; Oxford: Oxford University Press, 1997. Disponível em: http://www.hdr.undp.org/en/content/human-development-report-1997
  » http://www.hdr.undp.org/en/content/human-development-report-1997
• VAZ, A. C. N.; JANNUZZI, P. M. Indicador de Pobreza Multidimensional como síntese dos efeitos da abordagem multissetorial do Plano Brasil Sem Miséria. Revista Brasileira de Monitoramento e Avaliação, Brasília, n. 8, p. 32-49, 2014.
• WMO. State of the Global Climate 2021 Geneva: World Meteorological Organization, 2022. WMO-No. 1290.

Disponibilidade dos Dados de Pesquisa

Os dados para o cálculo do IPM podem ser acessados em: GitHub - Trajetorias-Sinbiose/Trajetorias_dataset · GitHub

Códigos JEL:

I32, R11, R20.

JEL Codes:

I32, R11, R20.

Editor responsável

Lucas Resende de Carvalho (Editor Associado).

Centro de Desenvolvimento e Planejamento Regional, Universidade Federal de Minas Gerais, Belo Horizonte, MG, Brasil.

ORCID: https://orcid.org/0000-0002-3618-3967

Publication Dates

• Publication in this collection
  23 Oct 2026
• Date of issue
  2026

History

• Received
  15 Sept 2025
• Accepted
  13 Apr 2026