The changing role of employment and alternative income sources among the urban poor: a systematic literature review

ABSTRACT We perform a systematic review of the literature on the association between income, employment, and urban poverty from a multidisciplinary perspective. Our results, derived from the analysis of 243 articles, confirm the significant role of employment in the urban poor’s lives, highlighting several factors that constrain their ability to improve their labour market outcomes: lack of access to public transport, geographical segregation, labour informality, among others. Furthermore, the paper finds different strategies used by the poor to promote their inclusion in their city’s economy. We found a major bias towards research focused on advanced economies, stressing the need for development studies dealing with the specific challenges of developing economies.


Introduction
From 1981 to 2015, the percentage of the world population living in extreme poverty (less than US$ 1.90 a day, using PPP 2011) went from 42.1% to 10.0% (World Bank, 2019). This remarkable progress was concentrated in a few industrial hubs (mainly located in China and India), and fueled by massive migration from rural to urban areas, meagre employment conditions and growing social and economic inequality (Ravallion 2016). Such changes intensified multiple social problems already existent in the cities (such as overcrowding, pollution, lack of services, crime and high levels of social fragmentation) (Zhao and Li 2016), and magnified the differences between rural and urban areas.
In developed countries poverty in rural and urban areas is usually considered of similar nature (Ravallion 2016), but in most developing countries uncontrolled city growth, high levels of job informality and limited access to certain goods and services have induced new forms of limited wellbeing (Wu 2004;Fulong and Ningying 2007;Linn 2010;Martin and Goodman 2016). Despite the major role of cities in the decline of global poverty, measures based on a single poverty line systemically ignore the remarkable and mounting discrepancies in quality of life, prices and needs between urban and rural areas. In this context, new explorations on the meaning of poverty in urban settings are essential to improve our understanding of this problem and the strategies required for its alleviation.
A key element in the analysis of urban poverty is employment, as it is the main mechanism of access to monetary income (and as a consequence, to goods and services in the market), health insurance, social security, and even identity and social connection (United Nations 2005; Hulme and McKay 2013; Thompson and Dahling 2019). Therefore, occupation is a major component of wellbeing in urban settings, and is usually considered not only the main pathway out of poverty, but also, labour market failures are usually associated with the inability to escape from this condition (Ravallion 2016). However, the complexities of work in the cities -particularly in developing countries with high levels of informality -make it relevant to understand the several forms of work and activities exercised by its inhabitants to gain income to address more effectively their needs and restrictions.
In this paper, we provide a systematic review of the literature on the association between income, employment and urban poverty, to illustrate the changing role of these factors in the wellbeing of the urban population, particularly among its most disadvantaged groups. Our review adopts the steps proposed by Khan et al. (2003) and Brunton et al. (2017), and follows an exhaustive search approach to identify as many relevant concepts as possible.
Unlike previous literature reviews about urban poverty (Sommer 2001;Hasan 2002;Cellini et al. 2008;Linn 2010;Chamhuri et al. 2012), we do not restrict our review to specific disciplines or attributes of urban poverty. Instead, we pursue a multidisciplinary perspective to explore how different social sciences address the experiences of the urban poor in the labour market, as well as the strategies utilised by this group to earn additional income in a context of scarce employment opportunities and poor employment conditions. The previous decision stems from earlier exercises in which selecting more restrictive criteria resulted in a major bias towards studies from developed countries, particularly the United States (US). Thus, to provide a more comprehensive perspective, we relaxed our inclusion criteria, to include both qualitative and quantitative approaches, as well as descriptive and more analytical methodologies.
Our findings confirm the prominent role of employment in the life of the urban poor, highlighting several factors -e.g., lack of access to public transport, geographical segregation, labour informality or inadequate human capital -that inhibit the ability of the poor to improve their labour market outcomes, as well as potential strategies to promote their inclusion into the economy of the cities they inhabit. Despite our comprehensive approach, we still found a major bias towards research in advanced economies (particularly, the US). Such bias, on the one hand, limits the applicability of available research findings towards those contexts where they are most urgently needed; and, by the other, defines a clear research agenda to fill in the existing gaps in the literature about developing countries, as well as to promote studies specific to their challenges.
In the following sections we start by introducing a broad description of the methodology, as well as a characterisation of the articles included in this systematic literature review in terms of the studied countries, their methodologies, and disciplines of origin. Then, we provide a review of the main results of the studies. Finally, we present our conclusions.

Methodology
We follow the steps proposed in Brunton et al. (2017), adopting two main research questions. In the first place, we focus on understanding the main sources of income among the poor in urban areas, which includes the kind of work they usually have, their distribution among household members, and the characteristics of these occupations. In a second stage, we identify policies, strategies and mechanisms that have successfully increased income among the urban poor.
In this last regard, we initially focused on studies providing causal evidence on mechanisms to improve the income of the urban poor, but such approach derived in an extreme concentration of the selected studies on developed countries, especially the United States. Thus, with the purpose of achieving a review as diverse and comprehensive as possible of available studies from different countries and disciplines, we decided to include studies not only focused on causal inference, but also descriptive, as well as from qualitative and quantitative nature.
We employed three main databases to perform our review: EBSCO's Academic Search Complete (which includes EconLit and other relevant data bases), Web of Science and JSTOR. Our search did not imposed restrictions on the period of analysis, but for practical reasons did include only search results from peer reviewed academic journals, available in full text.
Our search used Boolean operators to combine potential alternatives to each of three basic concepts required in our analysis: one for the notion of urban setting, another for the idea of poverty or low income, and finally one regarding employment and income. We performed the search in English and Spanish, but given the large number of results, we further restricted our search results to articles that included the keywords in the abstract (rather than in the whole text) and excluded the word 'income' (and its equivalent in Spanish: 'ingreso'). The basic search algorithm is presented in Table 1.
Finally, we revised the whole set of results to eliminate duplicates and performed two further steps of refinement. Firstly, examining only the abstracts, we excluded those articles irrelevant to our research, i.e., from other fields (e.g., historical studies, health or education), literature reviews, other languages (besides English or Spanish), and so on. Then, after a full-article review, we excluded those unrelated with our research question, as well as those with methodological biases or studying very specific populations (see Figure 1).
Our initial search produced a total of 6811 articles from all the search engines, which were processed using the EndNote Web reference management software to identify duplicates, leading to the exclusion of 4583 search results. Then, we revised the abstracts of the remaining 2245 articles. The abstract analysis led us to further discard 1345 articles from unrelated disciplines and methodological reasons: literary analysis (nine), education research (30), historical studies (54), literature reviews (41), written in other languages (nine), health and medicine (840), social work (279), other topics (66) and other duplicates not identified by EndNote (20).
We skimmed full-text versions of the 897 remaining articles and eliminated those whose research question was unrelated with ours, i.e., that did not address poverty or poor populations (115), that was not focused on urban areas (104), which studied problems unrelated to employment or unemployment (315), or that did not combine the three topics of interest (53). In addition, we omitted those studies analyzing populations too specific to provide generalisable results, such as ethnic minorities or workers in a very specific setting. Finally, we excluded articles with significant methodological biases (e.g., with results unrelated with the objectives of the paper), as well as those purely theoretical (36).
Our final set of articles includes 243 studies (see the entire list in the online supplementary material), 60 qualitative studies, 146 quantitative studies (60 descriptive, 79 using multivariate analysis, and 7 using other methodological approaches), 19 using mixed methods (both quantitative and qualitative), and 18 case studies. Each of the 243 selected studies was systematised separately by two reviewers, each of whom selected among a series of predefined categories to classify the discipline, topic, geographical area of interest, relevant population, methodology and main results. Then, we compared both evaluations and, where significant divergences were noticed, the reviewers discussed to achieve a consensus.
In terms of the disciplines of study, most scientific publications nowadays accept contributions from several disciplines, as long as the scope of the article is within the interest of the publication. Consequently, to provide a general overview of the disciplines from which the studies are drawn, we define for each journal a main discipline of interest using the subdisciplines available in the Web of Science Journal Citation Reports, as well as the SCImago Journal Rankings. Where no clear discipline could be identified, or if a specific journal were not present in those rankings, we used the description of the journal available in their website to define an area and main discipline of interest for each journal.

Results
In order to present the main characteristics of our systematic review, firstly we analyse our resulting set of studies in terms of disciplinary diversity, the year of publication and the geographic area of interest of each article. In the case of disciplinary diversity, we use the journal's main field, as reported in Web of Science's JCR, SCImago, or the journal's description in its own web page.
The articles considered for this systematic review of the literature come mainly from journals within the social sciences (n = 193, 79.4%), as well as from multidisciplinary publications (n = 20, 8.2%), although journals from other disciplines could be identified as well  Figure 2).
In addition, we were interested in the analysis of the disciplinary diversity of our search, so we used a similar approach to that of the main field to identify the main discipline of each journal (see the previous section). The resulting distribution can be observed in Figure 3. Figure 3 shows that most articles in this systematic review come from multi and interdisciplinary journals, mainly from Development Studies (n = 42, 17.3%), Interdisciplinary (n = 29, 11.9%) and Urban Studies (n = 19, 7.8%) publications. Such journals, in conjunction with those from Sociology (n = 36, 14.8%), Geography (n = 32, 13.2%) and Economics (n = 19, 7.8%), concentrate 72.8% of all articles, the rest coming from journals in 15 disciplines, such as Law, Public Policy, Political Science or Anthropology. Regarding the year of publication (see Figure 4), only 20.6% of the articles were published before 2000, while 37% were published from 2010 and onwards. Thus, the largest part of the articles reviewed were published in the last two decades, especially between 2005 and 2014, a period where 46.9% of all articles appeared. In recent years fewer articles were identified, possible due to a lag in the publication of fulltext versions of certain journals in platforms such as JSTOR, one of the main databases use in this analysis.
In relation to geographical distribution (Figures 5 and 6), despite our efforts to reduce the overrepresentation of studies about developed countries, approximately 2 out of every 5 articles in our review analyse populations in the United States (US), Canada or Western Europe (n = 103, 42.4%), and the US alone accounts for 36.6% (n = 89) of all the studies. However, there is as well an emerging literature about the process of urbanisation in Africa, Asia and Latin America, as will be noted in the following sections.
Our systemic literature review led us to identify several mechanisms studied in the literature that may contribute either to the impoverishment of urban households, or to the alleviation of their disadvantages (see Figure 7). In order to organise the discussion, we established a series of analytical categories that describe the main areas of discussion: 1) analysis of the effects of residential segregation and labour market discrimination of minority groups; 2) identification of the factors generating economic gender bias; 3) causes and consequences of urban poverty; and, 4) households' strategies to alleviate poverty and unemployment. In the following sections, each of these topics will be examined. 1

Residential segregation and labor market discrimination
The Spatial Mismatch Hypothesis (SMH) suggests that residential segregation, suburbanisation of employment, and lack of access to adequate transportation limit access to work among minority and disadvantaged groups (Hu 2015). The seminal work of Kain (1965) showed that suburbanisation of low-skill jobs decreased employment opportunities among Black American groups in the US due to residential segregation. As high-income white groups relocate to the suburbs, disadvantaged groups are segregated around city centers. However, such population displacements impose a double burden among the poor (among whom, Black Americans are considerably overrepresented): they are unable to move to the suburbs due to their low-income, but the demand for low-skill jobs they depend on is higher in the suburbs, so that they are forced to spend a considerable portion of their income in transportation. This is particularly burdensome in cities without efficient and accessible transportation, which become a major barrier to take advantage of job opportunities in the suburbs.
The SMH has been widely tested in a variety of settings, although with inconclusive results (Hu 2015). Covington (2009)  metropolitan areas to show that the poor have fewer access to job opportunities than the non-poor, being worse for African American groups. In addition, the evidence from this study confirms that work distribution and housing accessibility are key contributing factors to this disparity. Stoll et al. (2000) study Blacks and Latinos of sub-metropolitan areas of Atlanta, Boston, Detroit, and Los Angeles finding that this groups are highly concentrated in city centers (where lower work availability exists), while higher income groups usually reside in the suburbs. On their part, McLafferty and Preston (1992) focus on females from ethnic-minority groups in north New Jersey, finding that these women have longer transfer-times and localised labour markets, so that they have considerably fewer labour market opportunities than White women.

uses information from US
Pinkster (2009) uses data from two neighbourhoods in the Hague City to tests the SMH, but in contrast to other studies, the author finds that residents of public housing in neighbourhoods with lower income have fewer job opportunities than their counterparts living in mixed-income neighbourhoods, probably due to their limited social networks, which may in turn limit their access to information about employment opportunities.
In contrast, Hu (2015), Hess (2005), and Gottlieb and Lentnek (2001) show conflicting evidence with regard to the SMH. Hu (2015) looks into the poor that reside in urban areas within the Los Angeles region, while Hess (2005) focus on low-income populations in two cities of New York. In both cases, the results suggest that the subgroups under review do not experience spatial mismatch since they have greater access to work than their counterparts who reside in the suburbs. Gottlieb and Lentnek (2001) examine job accessibility in four neighbourhoods in Cleveland differentiated by their location (suburbs or city centre) and colour (Blacks and Whites). As in the case of Hu (2015) and Hess (2005), Gottlieb and Lentnek (2001) do not find evidence of spatial mismatch.
In cases that comply with the SMH, the principal implication is that disadvantaged groups face more restrictions in the labour market, which can be detrimental to their economic well-being due to negative repercussions in their labour supply. However, early studies, such as those by Offner (1972), Bederman and Adams (1974) and Hutchinson (1978), do not show a consensus regarding the effect of work accessibility on employment. Offner (1972) shows that limited job availability in ghettos reduces labour participation of their prime age inhabitants, while Hutchinson (1978) indicates that, with respect to the urban poor, higher job accessibility increases their labour supply. In contrast, Bederman and Adams (1974) prove that variations in underemployment rates cannot be explained by work access differentials using Atlanta census data. More recently, Sanchez et al. (2004) found no association between regional employment and labour outcomes of social programmes beneficiaries.
In terms of public policy, the SMH suggests that increasing vulnerable groups mobility, either by tackling residential segregation or by incentivising public transport, can have significant positive effects on their welfare. An example of this includes commuting subsidies, although existing studies suggest that their benefits could be fewer than the welfare loss, which was generated by residential segregation (Martin 2001). Another example is housing mobility assistance, which may potentially eliminate the restrictions of city centers residents to move to other areas. Popkin et al. (1993) establish that, in the case of lowincome black families that participated in a residential mobility programme, those moving to the suburbs have higher probabilities to find a job than those moving to a city centre. Likewise, the results by Fauth et al. (2008) indicate that low-income families that moved to well-off areas within the same city enjoy better physical health, have a higher probability to work, and a lesser need for welfare than nonmovers. Furthermore, Blumenberg and Pierce (2014) suggest that, to sustain and enhance the positive labour market effects of this kind of programmes, it may be necessary to complement these interventions with improvements in public transport access or even promoting access to vehicles among their beneficiaries.
In addition to the SMH, among the elements explored in the literature to explain the higher rates of unemployment and residential segregation among minority groups, we found discrimination due to hiring practices (Bederman and Adams 1974;Davis 2017) and previous labour market experience (Tienda and Stier 1996). Kasinitz and Rosenberg (1996) study how race stigma and lack of diversity in social connections limit work opportunities in the Red Hook area in Brooklyn. These authors find that, despite the high percentage of Black workers who qualify through ability tests, they are not hired due to lack of experience, as well as racial and class biases. The same phenomenon is observed among firms in Chicago, which seem to prefer white workers whom they hire directly from white-majority neighbourhoods (Neckerman and Kirschenman 1991).
Another element that explains unemployment among Black Americans is the presence of immigrants, which has differentiated effects depending on the workers' skills. On the one hand, the increase in immigrants reduces the labour participation of lowwage and low-skilled blacks due to increased competition (Waldinger 1997;Johannsson and Shulman 2003). But on the other hand, better-educated blacks may benefit as they take advantage of immigrants' contributions to the local and national economy (Adelman et al. 2005).
Lastly, Adelman and Jaret (1999), Santiago and Wilder (1991) and Landale and Lichter (1997) examine the effects of segregation on poverty levels of minority groups. Santiago and Wilder (1991) and Haynie and Gorman (1999) find out that living in the most segregated zones of the US increases poverty risk among Latino communities, and particularly among Latino children. In contrast, Adelman and Jaret (1999) do not find a direct relationship between poverty and segregation of Afro-American groups in the US. In other contexts, such as those studied by Edin et al. (2003), residing in ethnic enclaves increases labour income of low-skill immigrants.

Gender inequalities
Multiple studies in our systematic review identify gender inequalities as a major problem affecting economic wellbeing of the urban poor. Available evidence suggest women receive lower welfare benefits due to heavier credit restrictions (Potts 1995), fewer job opportunities (Farkas et al. 1988;Haynie and Gorman 1999), greater risk of labour exploitation (Oberhauser and Yeboah 2011), higher propensity to carry out lower quality jobs (Cross 2008), or lower salaries (Farkas et al. 1988;Mitra 2005). All of these inequalities result in lower income for female workers, as well as the households with female heads (Cross 2008), as well as higher probabilities of being poor (Gambi 2006;Mberu et al. 2014).
The studies identified in this systematic review stress the role of access to childcare, domestic work, as well as other cultural and contextual factors that inhibit or restricts women's participation in the labour market. Among the factors associated to familial characteristics of women, the most commonly cited factor that affect women's labour market inclusion is the absence of childcare facilities. Studies specify that the time allotted to childcare or its related costssuch as payments for nurseries or full-time schoolshave negative impacts on labour outcomes of female parents. Polgar and Hiday (1974) find that lack of childcare support from relatives reduces work participation of low-income mothers in New York. Hurst and Zambrana (1982) find similar results among Puerto Rican mothers in New York, stressing the lack of emotional connection of non-familial childcare as a potential discouraging factor to use such services in these communities. In the same manner, Lein et al. (2005) suggest that low-income mothers of three US cities would require both family and institutional support for childcare to successfully and permanently remain employed. Further, Hallman et al. (2005) found that costs and travel time to and from childcare facilities are negatively related with working hours of poor mothers in the US. Other studies have documented that the lack of childcare support among low income families compels male parents to shift from formal to low-income informal jobs, negatively affecting the economic wellbeing of female parents, increasing the probability of these families to return to being dependent of social programmes (Anderson et al. 2004;Vera-Sanso 2012).
The relevance of access to childcare in women's labour market participation have motivated multiple studies trying to assess their response to several forms of subsidies and programmes to reduce or eliminate the costs of such services. Anderson and Levine (2000) study the response of female labour supply to the cost of childcare, finding that the elasticity is negative and varies between −0.05 and −0.35 depending on women's skill level. Julie et al. (2006) indicate that childcare subsidies not only incentivise female labour market participation, but also let them work extra hours, increasing family income. With respect to childcare utilisation, Davis et al. (2010) indicate that the factors explaining participation in childcare subsidy programmes heavily depend on the characteristics of the locality, even within urban areas, so that more studies are needed to identify the barriers to programmes participation and promote more equitable access to these subsidies.
Most low income families, however, do not have access to formal childcare services, so they usually make use of informal networks of support with family, friends and neighbours (Morris 1981;Vera-Sanso 2012). This is the case in various poor areas in India where older women not only tend to be self-sufficient, but also contribute in childcare and domestic work helping their daughters and daughters-in-law to participate in extra-domestic work (Vo et al. 2007). Yet, most of this informal provision of childcare and domestic work fall upon female members of the household, reducing their ability to undertake paid work (Reeve 2012) and contributing to sustain the inequalities observed in the labour market among women (Schroeder 2000).
Strategies aimed to provide childcare services not only create work incentives for mothers, but also have repercussions on children's welfare. Engle (1991), in a study of two villas in Guatemala, finds that maternal work does not have any negative effects on the child development if appropriate childcare is provided. Pamela and Karen (2007) demonstrate that nonstandard schedules generate parenting stress in lowincome working mothers, and that this stress, in turn, produce negative behavioural outcomes in their children. Coley et al. (2011) find evidence of greater psychological distress in adolescent children, and that their involvement in criminal activities negatively affects their mothers' participation in the labour market (as shown by lower rates of employment, limited working hours, job instability, lower earnings, and so on).
Lastly, our systematic review revealed the existence of contextual and cultural factors associated with the wellbeing of low-income women, particularly mobility, traditional gender roles and empowerment. Mobility is crucial to promote women's incorporation into the labour market, since they seem to be more likely to face restrictions associated to violence and overcrowding, as well as to decline job offers due to long commuting times (Anand and Tiwari 2006). Venter et al. (2007) document that women in South Africa face larger restrictions in mobility than men, and that such restrictions increase the farther the communities they inhabit are from the city.
Another factor explored in the literature are traditional gender roles, which have consistently been found to have negative effects on female's labour market participation which persist regardless of the work preference or the couple's income (Salway et al. 2005;Kim et al. 2010;Hagelskamp et al. 2011). In a study about low-income urban mothers in New York City, Hagelskamp et al. (2011) observe that mothers from traditionally conservative origins (Mexican, Chinese, as well as Puerto Rican and African-American) do not feel obliged to work due to the traditional gender roles embedded in their culture. In our review, we found only a few examples of policies to address such issues specifically in the case of urban settings. For instance, Al-Mamun et al. (2014) show that microcredit programmes promote empowerment of low-income women in Malaysia on aspects such as economic security, participation in household economic decision-making, and control over their own resources.
Finally, it should be noted that, due to the methodology used, we found few studies that focus on the problems faced by women in urban areas alone. One of them is that of Awumbila and Ardayfio-Schandorf (2008), who analyse the case of young women who migrated from rural to urban areas of Ghana to work as headload porters. Their results indicate that migrants are prone to verbal abuse, robbery, and rape. To avoid it, they form conjugal unions and sexual partnerships, violating their physical and reproductive health and increasing their feelings of powerlessness. On the other hand, another study finds that the individual characteristics of women may have a more significant effect on the risk of poverty in an urban context than in a rural one. Haynie and Gorman (1999) find that young black urban women without high school education are at greater risk of poverty than rural women and rural and urban men with similar characteristics.

Causes and consequences of urban poverty
A significant portion of the reviewed articles focuses on the analysis of how economic hardship resulting from deteriorating employment conditions among urban dwellers affects their well-being, not only in terms of its effects on household income (Ihlanfeldt 2007), but also on others areas such as mental health (Elder et al. 1995), violence (Poveda 2011), crime (McGahey 1986Ihlanfeldt 2006Ihlanfeldt , 2007, and homelessness (Bohanon 1991).
A phenomenon that is gaining relevance in the literature is in-work poverty, conceptually defined as individuals with formal employment who nonetheless fail to earn an income that suffices to keep them out of poverty (Slack 2010). While this phenomenon has become common among developing countries (Han et al. 2011) and in rural areas of developed countries, recent studies indicate that it is increasingly affecting urban areas of developed (Loïc 1996;Slack 2010) and post-socialists countries (Smith et al. 2008) too. For example, Slack (2010) finds that in the US, although the labour supply has consistently provided families outside of metropolitan areas with less protection from poverty than their metropolitan counterparts, this disadvantage has been waning in recent years.
Regarding policies aimed to alleviate in-work poverty, Loïc (1996) suggests that given the inability of the labour market to provide income security, social policy must separate income security from labour market. His proposal, like that of Wacquant (1999), consists of providing a guaranteed minimum income or universal basic income able to provide adequate means of subsistence and social participation to all members of the society. Another proposition centers on the living wage laws that various cities in the US have enacted in recent years. These laws mandate firms working for or receiving assistance from public agencies to pay a wage above the poverty line (Neumark and Adams 2003;Neumark 2004). Neumark and Adams (2003) study the effects of these laws and find that the living wage ordinances increase the earnings of low-wage workers and have a modest reduction in poverty levels, although they also seem to generate small negative effects on employment. Neumark (2004) suggests that living wage laws increase the salary of unionised municipal workers as a result of union's increased bargaining power (given that these laws discourage competition among firms). Other strategies explored in the literature include increasing the minimum wage and expanding income tax credits for low income workers (Cotter 2002).
Education is another variable that plays an important role in people's well-being (Amoo et al. 2013), which has extensively been proved to have a negative relation with poverty levels (Sackey 2005;Gambi 2006;Kohn 2011). However, a growing literature is linking it as well to social mobility (Kohn 2011). In a study of the metropolitan region of Santiago, Chile, Kohn (2011) finds that the poor are considerably more likely to remain in a disadvantaged position because of lack of quality education. Empirical evidence also indicates that household head's education is strongly associated with the well-being of other household members, reducing the probability of the household to be poor (Alem 2015;Manjengwa et al. 2016) and increasing their chances of remaining out of poverty in a given period (Mberu et al. 2014). On the contrary, low educational attainment seem to be associated with a higher risk of children living and working in the streets (Bal Yilmaz and Dulgerler 2011).
Family size can also influence the economic wellbeing of urban households. Polgar and Hiday (1974) show that an additional birth in poor families in New York reduces their income and labour market participation, while Manjengwa et al. (2016) and Mberu et al. (2014) demonstrate that an increase in household size reduces households' per capita consumption and increases their chances of falling into poverty. Other variables related to poverty are the age of the household head (Mberu et al. 2014) and the amount of family savings to finance economic activities (Sackey 2005). Conversely, being a woman or having a female head increases the likelihood of poverty in urban areas, while marital status (being single, separated, divorced or widowed) seem to have an ambiguous association with poverty (Gambi 2006;Mberu et al. 2014).
Another section of the literature identifies macro factors like weather and climate change, macroeconomic indicators, or community and neighbourhood level characteristics and their correlation with the well-being of the vulnerable population in cities. In relation to weather, Baez et al. (2017) show that rain shocks can increase the levels of urban poverty, which is alarming in the face of climate change predictions regarding an increasing likelihood of extreme weather events in several regions of the world (IPCC 2014). Concerning the macroeconomic indicators, existing studies have found that the increase and variability of food prices have adverse effects on the well-being of the urban poor (Bachewe and Headey 2017), while the setting of minimum wage below the poverty line can result in a failure to guarantee that workers meet the basic living conditions (Han et al. 2011). The studies examining the effects of community and neighbourhood characteristics on minority groups wellbeing highlight the role of segregation as a major cause of poverty, although this relationship may disappear among certain groups (as explained in the first analytical category). In addition, in the case of black American groups, other causes can be population growth and the presence of immigrants (Adelman and Jaret 1999). Finally, it has been noted that variables such as local rental costs, unemployment rates, and the percentage of the population in mental health institutions are positively related to the rates of homelessness (Bohanon 1991;Aoki 2003).
A substantial number of articles also focus on investigating the possible impact of poverty on children's welfare, crime, and income dynamics. The studies analyzing the effects of poverty on children's well-being indicate that its main consequences are child labour, behavioural problems, and higher frequency of children in street situation (Janak 2000;Kalff et al. 2001;Dayioğlu 2006;Iqbal 2008;Schmidt and Buechler 2015). For example, Schmidt and Buechler (2015) identify hunger and extreme poverty as the main causes that compel children to work and live on the streets of Tegucigalpa and Comayagua, Guatemala. Janak (2000) analyses the case of Haiti, where children are forced to domestic slavery under the guise of Restavec, a culture where children from mostly black and extremely poor origins are taken by wealthy families to serve as domestic slaves.
Another consequence of poverty is its perpetuation over time. In respect to this, Alem (2015) examine the persistence of poverty in the urban area of Ethiopia and finds that previous experiences of poverty increase the probability of being poor in the future. A limitation of these studies is that most of them rely on short-lived longitudinal data that do not provides a long term perspective, so that it is not known if chronically poor households remain poor in the long term or are able to escape of this situation. A few studies study this phenomena testing the existence of poverty traps. Antman and McKenzie (2007) use rotative panel data to test the existence of a poverty trap in Mexico's urban areas, finding that, although there is low labour mobility in this setting, there is no evidence of a poverty trap.
Lastly, several authors discuss the relationship between unemployment, poverty and crime. Ihlanfeldt (2007) and Ihlanfeldt (2006) analyse the case of Atlanta, finding that in poor neighbourhoods of its inner city better job accessibility among young men reduce drugs related crimes (Ihlanfeldt 2007), while worse job opportunities increase the levels of crime (Ihlanfeldt 2006). Likewise, in a more general study, Poveda (2011) use data from the seven major Colombian cities to show that poverty, inequality, and unemployment increase crime and violence.

Strategies to reduce poverty and unemployment
This section reviews the literature that focuses on mechanisms that the urban poor have embraced to cope with low income and unemployment.
The most recurrent strategy to solve the problem of unemployment among low-income households in urban areas is participation in the informal sector, either as wage-earner or by self-employment in activities that require little or no capital and investment. For example, there is a rich literature on strategies adopted by urban households to complement their income, such as urban agriculture or garbage collection. In the case of urban agriculture, several studies (particularly from African settings) study their potential to boost household income, alleviate unemployment and reduce food insecurity among poor households in fast-growing cities (Flynn 2001;Mosha 2015;Gororo and Kashangura 2016;Omondi et al. 2017). In such settings, urban agriculture seems to be a mechanism to alleviate the effects of economic downturns by creating productive areas from unused land and reducing household food bills (Drakakis-Smith 1991, 1994, although the benefits of such activities can vary depending on the agriculture type developed (Yusuf et al. 2015).
Concerning garbage collection, studies indicate that, on the one hand, it facilitates the employment of people who cannot get a formal job due to lack of education, extreme poverty, criminal records, or immigration status (Gutberlet 2012;Oteng-Ababio 2012;Dinler 2016). However, on the other hand, it can generate negative consequences in terms of health and social inclusion (Gutberlet and Baeder 2008;Vázquez 2016). For example, Gutberlet and Baeder (2008) note that among informal garbage collectors in Santo André, Brazil, many reported diseases such as the flu, bronchitis, body aches, or hand injuries. Hayami et al. (2006) stress that, given the vulnerability of this population and the public service they provide, public policies should be designed in order to improve their well-being. Oteng-Ababio (2012) and Oteng-Ababio et al. (2016) suggests that these groups can lead to sustainable management practices that eventually led them to ameliorate their poverty conditions. However, despite its potential benefits as income generating activities, poor access to capital and financial markets, lack of training, and widespread risk aversion, make informal employment and selfemployment activities unlikely to succeed in the long run (Marc-André and Wheeler 2006). Cheng and Beresford (2012), Mawowa and Matongo (2010), Olufemi (2000), Perlman (2006) and Van Blerk (2008) study how engagement in illegal activities can potentially be seen as an alternative to unemployment and poverty in urban areas. Cheng and Beresford (2012) analyse how the closure of multiple state-owned textile mills in Shaanxi, China, forced the laid-off workers to resort to illicit practices to survive. Similarly, Van Blerk (2008) study how ruralurban migrants in two Ethiopian cities recur to sex work to obtain income and send remittances to their families.
In the case of access to food, Crush and Frayne (2011) document that, in order to reduce expenses, the urban poor tend to consume food from informal food sources (e.g., informal markets, street vendors, and so on). Frayne (2005) and Frayne (2007) highlight the importance of such informal sources of food in the case of rural-urban migrants, who often rely on food transfers from their rural areas of origin as informal sources of food while they find employment or when employment sources are limited. Frayne (2007) demonstrates that 66% of migrants from the Windhoek city of Namibia receive food from their relatives and friends in the rural area, and that in more than 90% of the cases, they are intended for household consumption.
Another strategy used by low-income households to boost their incomes is income diversification. Income diversification has been widely studied in rural areas (Ellis 1998), but in urban areas existing studies provides conflicting results. Ebenezer and Abbyssinia (2018) suggest that income diversification is not a common strategy, as they estimate that only about 5% of households in Eastern Cape, South Africa, diversify their income sources. In contrast, Bigsten and Makonnen (1999) find that income diversification is a typical characteristic among the poor in Ethiopia. Ersado (2005) provides a possible explanation for such differences, pointing out that among Zimbabwean households income diversification is greater among rural households and decreases with the level of urbanisation, so that the differences between the results in Ebenezer and Abbyssinia (2018) and Bigsten and Makonnen (1999) may be associated with the differences in the level of urbanisation between Ethiopia and South Africa.
One of the most studied strategy of poverty alleviation among low-income rural households is domestic and international migration. There are numerous studies on this subject analyzing different aspects of the phenomenon: its pull and push factors (Marzan 2009;Gries et al. 2016); its socio-economic effects in the areas of origin (Baye 2006;Greiner 2011); the choices, opportunities, and housing conditions of migrants (Gilbert and Ward 1982;Greene 1997;Englund 2002;Junhua and Fei 2010); its impact on family relationships (Locke et al. 2012); and the barriers to migration (Gelderblom 2007). In addition, results suggest that the urban poor also resort to migration as a survival strategy. There is evidence of migration of low-income households from four capital cities of the Caribbean Basin to different developed countries (Itzigsohn 1995), but the most interesting finding is the presence of reverse migration, that is, the migration from urban to rural area in the same country. Clark (2012) and Fitchen (1995) stress that, unlike rural-urban migration, reverse migration is not only generated by labour factors, but rather it is influenced by the lack of affordable housing, domestic violence and low quality schools (Fitchen 1995).
Finally, as noted by Awumbila and Ardayfio-Schandorf (2008), rural-urban migration can have adverse effects on vulnerable groups such as women, since instead of strengthening women agency, migration can worsen their position, as well as increase their risk and vulnerability to sexually transmitted infections, rape, and other health problems. On the other hand, the lack of family and social networks in the destination can worsen the situation of rural-urban migrant women. For example, Sundari (2005) and Zhang (2006) find that searching for jobs is more difficult for women with poor social capital since regularly employees hire them based on recommendations. Furthermore, family and social networks can drastically reduce the risks and costs of migration (Zhang 2006), facilitate the housing search, and even provide food and housing during their labour insertion (Eklund 2000;Sundari 2005).

Concluding remarks
This article presents the results of a systematic review of the literature on the association between income, employment and urban poverty from a multidisciplinary perspective. The analysis shows that there is a vast literature on the subject, most of it concentrated on the experiences and circumstances of developed countries (particularly the US), but with a growing number of studies from developing countries. Unlike previous reviews of the literature on this area, we explicitly decided to expand the scope of our review to multiple disciplines and methodologies, so that we were able to identify the main areas of research explored in the literature on urban poverty. In spite of such a comprehensive approach, most of the literature we found was derivative and few seminal studies were identified. Our hypothesis is that this is because most of the studies on urban areas are usually considered an application of more general approaches, so that most models rarely apply specifically to cities or metropolitan areas. Another possibility is that the restrictions imposed on the systematic search may have excluded relevant articles by requiring that the article had at least one word of the three concepts of interest.
In spite of the previous considerations, our results let us identify four major nodes or areas of study in the literature about urban poverty and income. First, the consequences of residential segregation and labour discrimination within vulnerable populations. Second, the economic and cultural factors associated to the existence of well-being differentials between men and women. Third, the micro and macro factors associated with poverty outcomes, and the consequences of poverty itself. Finally, a fourth node regarding the strategies developed by low-income households to reduce their vulnerability and improve their wellbeing.
The literature review enabled the identification of relevant results related to urban poverty, as well as to establish new subjects and research areas. One of these is the lack of agreement over the effects of residential segregation on employment and poverty within minority groups. Such disagreement has important repercussions on research outcomes and makes it difficult to trust these studies for policy making (Jencks and Mayer 1990;as cited in Kain 1992). Existing studies in this area are largely descriptive, and few present causal estimations of such effects, so that more evidence on this regard is required in order to surpass several methodological limitations of previous studies: the endogeneity of residential location and the lack of accuracy on employment access measures. In addition, most of the available studies concern developed countries in Europe and the US, although this is a growing problem in many developing countries with the fastest growing cities. In addition, more research on the subjacent causes of minorities segregation and its dynamics over time must be thoroughly conducted (Ihlanfeldt and Sjoquist 1998).
A series of individual, family, and cultural factors that affect the well-being of women were identified: lower returns to human capital, domestic shores, infant care, and lack of mobility. We also identified both formal and informal strategies that have allowed women to improve their living conditions. These include childcare subsidies, credit access, the assistance of relatives with childcare and domestic work, and family and social network support in the search for employment and housing for migrant women. As in the case of the residential segregation hypothesis, we found few causal studies, particularly on the mitigation capacity of the mentioned strategies to improve women's wellbeing. For example, even if the support provided by a member of the family for childcare renders it easier for the mother to be employed, the reciprocity needed to sustain the support may reduce women's incentives to participate in the labour market actively.
A significant amount of studies focus on the role of education in individual wellbeing and poverty reduction. The articles found in this literature review are consistent with the predictions of the human capital theory that indicates that skills and productive knowledge acquired during education increase people's productivity thus increasing their income (Tilak 2002), although most studies in this area are largely descriptive. The results sustain the existence of public policies aimed towards increasing the levels of education of vulnerable populations, with the aim of not only alleviating monetary shortages but also, in macroeconomic terms, fostering economic growth and development (Tilak 2002).
The literature review showed that an element related to household´s well-being is their size, and that there exists an area of opportunity in the study of the subject specifically in urban areas. Moreover, the need for research with a long-term focus that allows to study the effects of fecundity in poverty dynamics was identified. However, as shown by Anyanwu (2014), this effect can be ambiguous: it increases the well-being of the household by increasing its endowment of labour, but it can also create a poverty trap due to the larger amount of resources needed to sustain a growing offspring.
Another element identified in this review is that the two most common strategies used by households to overcome poverty are informality and migration. The first one allows the urban poor to solve problems like unemployment and low income, but it is not a long-term solution since a job on the informal sector can induce a poverty trap, acting as an obstacle towards the transition to better employment (Bacchetta et al. 2010). Migration represents another natural response of households with an excess of labour supply in urban areas; however, even if it is a decision taken at the microeconomic level, it has important macroeconomic repercussions for the expelling countries. On one hand, it can foster economic growth as remittances are used for financing expenditures on education, health and increasing investments of recipient households; on the other hand, they can have negative impacts on economic growth as far as an important quantity of the expelling country's workforce is sent away (World Bank 2005).
In the light of all the studies analysed in this systematic review, we identified several limitations in the existing research. The first is the bias toward research focused on advanced economies (particularly the US), limiting the generalisation of the available results towards environments where research is urgently needed. This bias shapes a research agenda towards filling out the existent gaps in the literature about developing countries. In the second place, there is a lack of studies that look forward to establishing causal relations between poverty and its determinants and consequences. Finally, as mentioned above, there is a need to deepen research on the existence and causes of urban poverty traps to advance public policies that allow households to overcome them.
We also found limitations in the methodology. That is, papers that focused on urban informal economy and women living in the latter were excluded. There is a need to analyse the extent to which the bias found in advanced economies is due to the databases and the search words used.
The systematic review of the literature did not focus on analyzing public policies to combat poverty; however, we identified and documented those that could promote sustainable urban development. For example, the case of commuting subsidies and housing mobility programmes centers on solving the problem of limited access to work among minority and disadvantaged groups due to residential segregation, suburbanisation of employment, and lack of access to adequate transportation. Other policies that prioritise women are childcare support, which seeks to encourage their participation in the labour market, and credit programmes that focus on empowerment and provide them economic security.
Although these are short-term policies, we also identified medium-and long-term proposals. One of them consists of supporting informal workers dedicated to garbage collection and recycling to organise themselves into micro-enterprises, associations, or cooperatives. This strategy would substantially improve the well-being of garbage collectors by allowing them to access health services, savings schemes, and a decent income, as well as improve their working conditions and self-esteem, and at the same time would promote waste recycling (Gutberlet 2012;Oteng-Ababio 2012). More universal strategies are to establish a guaranteed minimum income, universal basic income, or living wage to provide adequate means of subsistence and social participation to all members of society. Note 1. In order to keep the article's length manageable, we cited the most relevant studies to each topic. The complete list of references can be found in the online supplementary material.

Disclosure statement
No potential conflict of interest was reported by the author(s).