Ever-more sophisticated studies on the methodological approach and the conceptual scope of poverty have led to a consensus among scholars on the dynamic characteristic of this phenomenon – in other words, the existence of an in-and-out of privation movement of individuals and families. Within this context, Hulme and Shepherd (2003) defined five groups according to the location of the punctual and average indicators of poverty vis-à-vis the poverty line. This paper's objective is to adapt this typology to Brazil, using PME (Monthly Job Survey) micro-data for the 2002–2011 timeframe and the six Brazilian Metropolitan Regions covered by PME as well as, by estimating a multinomial logit, investigate which family characteristics relate to a greater or lesser chance of belonging to each of the chronic and transitory poverty categories. Categorization allows observation that, despite a sweeping across-the-board decline in the percentage of families in all poverty categories in the past decade, those families always or usually poor display demographic, socioeconomic, access to and insertion into the labor market categories which differ from families in transitory poverty or classified as never poor. Moreover, Northeastern metropolitan regions (Salvador and Recife) have higher percentages of chronic or transitory poverty. Multinomial logit estimates make it possible to verify that families whose members have completed secondary schooling or college or hold a higher-qualified occupation stand lesser chances of entering into or remaining in poverty. Results call for differentiating among poor families, as they enter into or leave poverty – which is tantamount to saying that the dynamics of poverty must be taken into account as public policies are drawn up.

Com o desenvolvimento de estudoscadavezmaissofisticadosemtermos da abordagemmetodológica e da abrangência do conceito de pobreza, pesquisadoreschegarama um consensosobre a característicadinâmica do fenômeno, ouseja, a existência de um movimento de saída e entrada de indivíduos e famíliasnasituação de privação. Nessecontexto, Hulme e Shepherd (2003) definemcincogrupos de acordo com a localização do indicadorpontual e do indicadormédio de pobrezaemrelação à linha de pobreza. O objetivo do presentetrabalho é adaptartaltipologia para o Brasil, pormeio da utilização dos microdados da Pesquisa Mensal de Emprego (PME), para o período de 2002 a 2012 e as seisregiõesmetropolitanascobertas pela PME, e, pormeio da estimação de um logit multinomial, investigarquaissão as características das famíliasrelacionadas a maioroumenor chance de pertencer a cadauma das categorias de pobrezacrônica e transitória. A categorizaçãopermiteobservarque, apesar da quedageneralizada da porcentagem de famíliasemtodas as categorias de pobrezanaúltimadécada, as famíliassempreouusualmentepobresapresentamcaracterísticasdemográficas, socioeconômicas e de acesso e inserção no mercado de trabalhodistintas das famílias com pobrezatransitóriaouclassificadascomonuncapobres. Além disso, as regiõesmetropolitanas do Nordeste (Salvador e Recife) apresentamporcentagensmaisaltas de pobrescrônicos e transitórios. Osresultadosindicam a necessidade de diferenciação das famíliaspobres, de acordo com o movimento de entrada e saída da pobreza, ouseja, que a dinâmica da pobrezasejaconsiderada no desenho de políticaspúblicas.

Family privation is a phenomenon which has long interested researchers and public managers, but has become the core of academic debate and public policy formulation since the economic crisis in the 80s, as the Human Development Index (HDI) was created, the 90s and particularly in this century's opening decade, as income transfer programs aimed at poorer families were implemented and consolidated. In this time period, monetary privation, the various dimensions of poverty and poverty dynamics itself are measured by indicators not so much to quantify and focus on the issue but rather to formulate policies seeking to reduce privation.

Literature has advanced in distinguishing absolute from relative poverty; monetary from multidimensional poverty ( Soares, 2009a , de Codes, 2008 and Rocha, 2003 ) and in identifying the factors related to such phenomena ( de Barros et al., 2006 , Ribas et al., 2011 and Machado et al., 2014 ).

In the case of socioeconomic phenomena associated to family privation, by using a methodological approach and different concepts, researchers have come to a consensus on the multi-dimensions and complexity of this social problem. However, something inherent to any social phenomenon (its dynamics) is, to this day, very little present in Brazilian literature on poverty. Even though Ribas et al. (2011) has sought to distinguish transitory from chronic poverty in Brazil, through cross-section data, and Machado and Ribas (2010) have shown factors associated to getting into or leaving poverty and, especially, how the market place as a whole affects the time families remain in poverty, such works are still few and far between. One reason for this is that we have only one longitudinal database for secondary data, namely, PME/IBGE's Jobs Monthly Survey, (PME/IBGE), as mobility studies of analysis unit require a long follow-up.

There are two distinct methodological ways to deal with poverty dynamics. First, through a priori probability calculations, known in literature as vulnerability to poverty ( Jalan and Ravallion, 1998 , Dercon, 1999 , Chaudhuri, 2003 , Ligon and Schechter, 2004 , Ribas, 2007a , Gonçalves and Machado, 2014 and Gonçalves, 2015 ). The second methodological approach seeks to break down transitory poverty's chronic dimension, while keeping open the possibility of approaching both aspects simultaneously.

It is possible in this context to highlight White and Killick's work (2001) , developed to deal with this phenomenon's various forms of manifestation. These authors stress the importance of distinguishing between chronic poverty – understood as the state of poverty due to a convergence of factors such as the absence of political influence, few goods/assets, and lack of market access – from transitory poverty, that must be related to the incidence of shocks or greater vulnerability to shocks, such as problems in farm harvests, price fluctuations and diseases ( White and Killick, 2001 , p. 15).

Hence, Hulme and Shepherd (2003) , aware of the need to distinguish the different forms of poverty, define five groups, according to the location of poverty's punctual and average indicators vis-à-vis the poverty line. This paper seeks to adapt such typology to Brazil's case, using PME (Monthly Job Survey) micro-data, from 2002 through 2011, and via income imputation from PNAD's (National Survey through Household Sampling) ¹ non-work, for the six Brazilian metropolitan regions covered by PME. Moreover, estimating a multinomial logit leads to an investigation of socioeconomic factors of families related to a greater or lesser chance of their belonging to one of the categories of chronic and transitory poverty.

This paper is organized into four sections, following this Introduction. Section 2 briefly summarizes the literature on categorizations in the context of poverty dynamics. Section 3 presents the methodologies applied. Section 4 describes the results and establishes a comparison with the results found by Hulme and Shepherd's article (2003) and Section 5 has some closing remarks.

The chief result of the extensive literature on poverty dynamics is to acknowledge the need of a time reference for this phenomenon and to emphasize it as the output of a dynamic process ( Alwang et al., 2001 , p. 7). Most studies on poverty dynamics consider the absolute poverty line, using lines based on family consumption. In the Brazilian case, even though Rocha's lines (1997) refer to consumption reported by the Family Budget Survey (POF/IBGE), studies use absolute lines defined according to family income ( Ribas, 2007a , Ribas, 2007b , Soares, 2009b , Ribas et al., 2011 and Osorio et al., 2012 ).

Using these lines is justified because it is precisely the distinction between individuals or families always poor or poor in a given timeframe which allows a deeper dwelling on the poverty cycle buttressing this phenomenon ( Bane and Elwood, 1986 ). These authors define poverty as income below a pattern of needs, whose calculation is based on family size, and define poverty cycle as a period starting when income drops below the poverty line, after having been above this same line and ends when income climbs above the poverty line, immediately after having been below it.