Overcoming Poverty in Multidimensional Poverty Interventions through Self-assessment and Mentoring

A microfinance organization in Paraguay has developed the “Poverty Stoplight” (PS), an innovative technological tool that allows families to self-assess their level of multidimensional poverty and start an integrated mentoring process whose goal is to empower families to eliminate multidimensional poverty based on what the participants value the most. During the program, a mentor works with participants to design a customized family plan to identify their most significant challenges in order to overcome their deprivations. The PS places human development and poverty elimination as the main objectives for the intervention. This paper has three elements. First, it presents the methodology for poverty intervention and the detailed tool (i.e. a multidimensional metric) used to encourage reflecting and promoting agency. Second, it uses the Capability Approach to explore the potential of the PS intervention to increase agency and decrease multidimensional deprivations. Third, it presents results from an ongoing research project that evaluates the program’s effectiveness in helping participants overcome poverty. This empirical part is based on data collected between August 2015 and June 2017 from over 9,000 microfinance clients and analyzed using the technique of OLS regressions. The results indicate that participation in the program is indeed associated with a higher probability of overcoming multidimensional poverty. 1 He is a doctoral researcher at the Institute of Development Studies (IDS) at the University of Sussex. His research interests include power and empowerment, poverty, and international development. He has a Master’s degree in Political Science from the University of Missouri (USA) and a Master’s degree in Social Research from the University of Edinburgh (UK). He has extensive working experience as a development practitioner. Correo electrónico: juanp@ids.ac.uk 2 She is the Director of Monitoring and Evaluation (M&E) of Fundación Paraguaya. She was born in Austria where she studied Economics, Socioeconomics, and Political Sciences before earning a Ph.D. in Global Development from Tulane University in New Orleans, USA. She has worked as a consultant in several education development projects and taught poverty measurement at the University of Vienna. Correo electrónico: khammler@fundacionparaguaya.org.py

The survey is done through a digital platform that can be used online or via an App. A PS mentor guides the participant through the questions, and clarifies concepts or probes answers where necessarybut importantly, it is the respondent who in the end chooses which of the three levels best reflects her reality, and who actively picks the respective option by touching the screen (Burt, 2013;Fundación Paraguaya, 2014).
Immediately after the survey is completed, the software presents the participant's poverty dashboard that summarizes the Greens, Yellows, and Reds. The PS defines those living in poverty as the main stakeholders of their data, and as the most important decision makers when it comes to eliminate their poverty. As such, they need dis-aggregated data that allows them to analyze the possibly overwhelming problem of their own poverty in a granular way, taking stock of both the resources that are already available in the family (the Greens), and the areas of moderate and extreme deprivations (Yellows and Reds). For other stakeholders, such as the implementing organization, the PS provides options of aggregating the datafrom simple metrics such as the percentage of Reds and Yellows in a given indicator in a community, to more technically advanced metrics, such as an Alkire/Foster-type Multidimensional Poverty Index (Alkire & Foster, 2011).

The Poverty Stoplight methodology: self-assessment and mentoring
The mentoring methodology starts during the surveying process and builds directly upon the poverty dashboard. As all 50 multidimensional indicators are designed to be achievable, according to the PS program theory the presentation of a Green in the surveying stage already serves as a first demonstration that a situation of "non-poverty" can be attained by the participant. During the self-diagnosis, participants are supported by FP mentors through a dialogue with the aim of identifying where they are and where they could be. After the selfassessment, the participant, together with the PS mentor, starts elaborating what has been identified as a 'life map': Out of all her Yellows and Reds, she picks five priority areas in which she wants to improve first. These are tagged in the Software and at the same time written into a cardboard map that will stay with the participant. The participant, together with the PS mentor, then reflects on the potential causes of the deprivations and formulates concrete steps that she will take to overcome them.
While FP provides some problem solutions (for instance, if a lack of funding is identified as a roadblock, a family might receive a loan), the overall goal is to mobilize all resources that are availableat the level of the family, from neighbors, in the municipality or from state or national government agencies, for instance. All solutions are registered in the PS platform and build an ever-increasing solutions database that participants and mentors can consult.
After a given time-typically after around half a year to a year, with variations among different program adaptations-participants do a follow-up PS survey. Together with their mentor, they reassess their deprivations in all indicators and use that information to reflect on their progress, and to choose their next improvement priorities. These follow-up surveys also provide valuable information for the implementing organization, as the data can be analyzed in conjunction with the attempted solutions that were registered in the platform.

State of the Art: Overcoming poverty through aspiration-mentoring
There is some evidence that the PS program helps families overcome poverty. Apart from anecdotal evidence brought forward by FP, there are two quantitative studies that are based on administrative data from FP clients who were purposefully selected to participate in the program (Budzyna & Magnoni, 2013;Burt, 2014). Both conclude that program participation is associated with a decrease in deprivations, but the data does not allow to assume any causal relationship. It is possible to look at the wider literature, however, to understand whether a program that focuses on expanding people's capabilities through aspirationincreasing mentoring may have positive effects. pág. 115 ISSN 2362-194X Journal de Ciencias Sociales Año 8 N° 15 A growing body of literature suggests that increasing people's aspirations can help them overcome poverty. Based on the theoretical work on aspirations, empirical researchers have gathered evidence on the social formation of aspirations, and that they do indeed matter in practice. An often-cited example is a study from India (Beaman et al., 2012) that shows that in villages that were randomly selected to have a female leader, the gender gap in aspirations narrowed considerably, and the gender gap in adolescent educational attainment was completely closed. The authors attribute these changes to increased aspirations of the girls. Bernard et al. (2014) show that much less is required to raise aspirations and see real changes. In recent years, poverty mentoring has gained importance (Sheldon, 2017). This is a promising development, given that research has shown that a good mentoring relationships can lead to higher levels of positive motivation and social-emotional and cognitive resources (Eby et al. 2008;Fletcher & Ragins, 2007;Rhodes & DuBois, 2008). Directly in the context of poverty elimination, researchers have found highly promising results regarding the potential benefits of mentoring, for instance in evaluating BRAC's Graduation Approach (Bandiera et al. 2017;Banerjee et al, 2015;Banerjee et al. 2018). However, the evidence base is still rather small, and more research is needed to better understand the mechanisms and effects.
Put together, the literature suggests that aspirations and mentoring are promising tools in the pathway out of poverty. However, our understanding on how these processes work, and what types of interventions can be successful, is still limited.

Theoretical Perspective
The Capability Approach (CA) is, in essence, a people-centered approach for development. It is a normative framework to evaluate individual well-being and social arrangements (Clark, 2005;Robeyns, 2017;Sen, 1999). The CA is "focused on what people can do and be (their capabilities) and on what they are actually achieving in terms of beings and doings (their functionings)" (Robeyns, 2017, p. 36). The CA is multidisciplinary, mostly applied in development studies, social policy, and welfare economics, and defines and conceptualizes notions such as poverty, inequality, and well-being (Alkire, 2005;Sen, 1999). Sen (2009, p. 16) defines the CA as "an intellectual discipline that gives an essential role to the evaluation of a person's achievements and freedoms in terms of his or her actual ability to do different things a person has reason to value doing and being".
This paper uses the CA as a theoretical framework for various reasons. First, the CA allows us to put the individual at the center of our research. Second, the CA does not limit its informational basis by focusing exclusively on utilities, happiness or rights and this helps us to cover the individual as a whole considering, for example, people's interpersonal differences, context, and aspirations (Robeyns, 2003). Third, the CA allows us to assess the effects of the PS in terms of achieved wellbeing and wellbeing freedoms (Sen, 2009).
There is a distinction between the term Capability Approach, referring to the general, open, and underspecified framework, and the term Capability Application, referring to a specific use of the Capability Approach (Robeyns, 2017). In this section, the paper will refer to the Capability Approach (or CA) by briefly presenting its main concepts. In the next section it will introduce the specific use of the CA for the PS intervention, that is, our Capability Application.

The Main Components of the Capabilities Approach
The CA in its more general sense is based on eight core concepts that Robeyns (2017) calls "A-modules". The first of these A-modules are functionings and capabilities. Functionings are things that a person actually manages to do or be, in contrast to capabilities that are about what is possible to achieve depending on the freedoms or valuable options from which the person can choose (Nussbaum, 2000;Sen, 1999). A central part of Sen's definition of functionings is the concept of 'value and have reason to value'. According to this, an activity or situation 'counts' as a valuable functioning for that person only if that person values it (Sen, 2009). This encourages the participation and engagement of those people whose lives are at stake. Functionings are related to different dimensions of life such as health, education, relationships, culture and so forth (Alkire & Deneulin, 2009). Capabilities are a person's freedom to achieve various functionings. This is the set of choices that are open to the person (Alkire & Deneulin, 2009;Sen, 1992Sen, , 1999 learnt to ride a bike). The fourth central module of the CA is the distinction between means and ends. This distinction requires the researchers to evaluate whether it values something as an end or as a mean to an end. For example, being healthy could be an end whereas the means necessary for this end are clean water, access to doctors, and so forth (Robeyns, 2017).
The fifth key module is the use of functionings and/or capabilities as the evaluative space, i.e., the space where interpersonal comparisons and personal evaluations can be made to evaluate the wellbeing of people (Robeyns, 2017). The sixth module concerns the dimensions of ultimate value different from functionings and/or capabilities. Sen (2004) argues that the CA does not capture the procedural aspect of freedom but only the opportunity aspect of it, hence it needs to be complemented with other dimensions such as fair process. The seventh key module is related to the value of pluralism within the CA. Pluralism in the sense that users of the CA should accept other dimensions of ultimate value as mentioned in the sixth concept as well as the multidimensional nature of the CA (Sen, 2004). Finally, the eighth module is related to what Nussbaum (2000) called 'the principle of each person as an end'.
This concept implies that any policy intervention, program, or analysis should look at every individual as an end, and neither look only at the average individual, nor treat individuals as pure means to reach aims defined at the level of the group or society.

The Poverty Stoplight and the Capability Approach
This section introduces the PS as a Capability Application. In addition to the A-modules outlined above, Robeyns (2017) proposes that applications of the CA need to show how they respond to a range of basic characteristics (or B-modules). This section attempts to showcase the alignment mentioned.
The purpose of the PS is twofold; On the one hand, the PS is an attempt to measure multidimensional poverty, providing data on deprivations that is useful for those carrying out the measurement (generally, families living in poverty), but also for other stakeholders, such as NGOs or even government agencies. Hence, as a measurement tool the PS aims to meet the needs of a wide field of audiences. On the other hand, as a self-diagnosis survey and mentoring methodology the PS also aims to help participants to reflect on their own situation, increase their aspirations and enhance their agency, thus increasing their capabilities. It is important to note that the CA was not the starting point for designing the PS program; rather, this paper aims to ex-post embed the PS into the CA, arguing that the PS can be understood as one Capability Application.

The PS metric and the CA
The indicators of the PS were developed based on a broad review of the literature on multidimensional poverty measurement, on expert consultations, and on focus groups with the communities in which the tool is being used. This three-part process generated a set of 50 indicators that cover a wide range of topics, including conventional topics like health and education, but also some more unconventional topics such as having a high self-esteem or having regular recreational activities.
Overall, the selection and definition of indicators can best be described as a combination of a bottom-up process and expert consultations. However, the intention of the PS is to present a list of poverty indicators that are relatable to survey-takers so that the process of self-assessing one's level of poverty encourages reflection and thereby beginning a shift in aspirations. Hence, the question arises whether it introduces noise if external experts add indicators or make changes to the list generated in the bottom-up process? Both from FP's viewpoint, and from the perspective of the CA, the answer is 'no'. FP explains that "it is not only possible but advisable to simultaneously consider multiple viewpoints and perspectives inherent in any debate" (Burt, 2013, p. 54), and defines itself as the "mapmaker" who draws a poverty map based on the inputs from those experiencing poverty. From the perspective of the CA, including the perspective of outsiders makes sense because of the phenomenon of adaptive preferences: A group that is systematically socialized to have low aspirations and ambitions will perhaps not put certain capabilities on its list [of important dimensions], thereby telling themselves that they are unachievable, whereas objectively speaking they are achievable, albeit perhaps only after some social changes have taken place (Robeyns, 2017, p. 139).
The levels of each indicator of the PS were intended to be defined so they can be reached if a family shows a certain behavior or achieves a certain state (Burt, 2013). As a result, the evaluative spaces are mostly functionings or sometimes capabilities. There are also a few indicators that remain on the level of resources. These resources typically have high instrumental value, or a wide range of capabilities or functionings can be associated with them.
In a recent round of revisions, PS indicators were adjusted to be aligned with existing multidimensional poverty metrics that are explicitly based on the framework of the CA: The 3.4. The PS self-assessment and mentoring methodology and the CA While the PS as a metric operationalizes the way the CA might describe multidimensional poverty, the PS methodology operationalizes the way that the CA might see a pathway out of poverty. And just as for the former, the latter component of the PS was not explicitly developed based on the CA, but can in its claims and strategies be understood and evaluated in this framework, as the following discussion will show.

Self-assessment and Reflection
Borrowing from the way Freire (1970) characterizes traditional education, many conventional poverty interventions can be described as treating people like empty repositories that the government or other institutions have to fill with skills, assets, and other resources. In order to decide on the type of intervention to carry out or to evaluate program outcomes, conventional programs extract information from those experiencing poverty using traditional surveys. This process does not allow dialogue and reflection because the individual is subjected to the simple role of data surveyor; the process does not recognize the role of reflection as a step to overcome poverty. One of the innovations the PS introduces to poverty programs is that participants self-diagnose their level of poverty creating a self-reflection process.
The Poverty Stoplight helps families reflect on and analyze their current situation as a step to recognize deprivations and take actions to overcome them. The concept of 'reflection' used by the PS is related to what Freire called 'conscientisation'. In fact, this paper uses the term conscientisation as a synonym of reflection. Conscientisation represents "the development of the awakening of critical awareness" (Freire, 1973). Freire proposed that to reach 'conscientisation' an external intervention is necessary, in which individuals experiencing poverty have participatory consultations to develop a dialogue with the objective of awakening critical thinking and analyzing their situation, thus learning and reflecting from that experience. During the process of dialogue, the individual gradually will gain her ability to choose and more importantly to "become responsible for her own development and to act towards addressing the constraining social structures that oppress her and others around her" (Poveda, 2015, p. 33).
The poverty measurement process employed by the PS is designed to start this process of reflection. People experiencing poverty are no longer docile recipients of poverty programs but ready to critically engage and co-produce knowledge and actions in dialogue with a mentor. Through the self-diagnosis survey, the PS provides participants an opportunity to reflect, question, and assess their own deprived situation as a prerequisite to act upon this to improve their lives. By introducing reflection as a required step for individuals who participate in the PS, the CA is inherently being introduced.

Agency to overcome poverty
All 50 indicators of the PS are designed to be actionable and achievable by participants. This means that the active involvement and action of participants of the PS make them agents of change of their own lives. An agent is "someone who acts and brings about change, and whose achievements can be judged in terms of her own values and objectives, whether or not we assess them in terms of some external criteria as well" (Deneulin & Shahani, 2009, p. 31). Concerns for people's agency and empowerment play a key role in the CA since "greater freedom enhances the ability of people to help themselves, and also to influence the world, and these matters are central to the process of development" (Sen, 1999, p. 152).
Agency and empowerment, like poverty, are plural concepts as well as measurements.
They are sometimes treated as synonymous, although in a strict sense they are related but different in their definitions (Ibrahim & Alkire, 2007;Kabeer, 1999;Narayan, 2005). Agency is about the ability of an individual to make choices about her goals and take action to achieve them; empowerment is the expansion of agency (Ibrahim & Alkire, 2007;Kabeer, 1999;Narayan, 2005). Empowerment is considered the expansion of agency because it is comprised by a) agency and b) the institutional environment, as a pre-condition for agency, which allows individuals to exercise agency (Alsop et al. 2006;Ibrahim & Alkire, 2007;Narayan, 2005). These two elements are not considered mutually exclusive and the process from disempowered to empowered could be considered incomplete unless they are both present.
Additionally, beliefs of personal efficacy constitute the key factor of human agency. If people believe they have no power to produce results, they will not attempt to make things happen (Bandura, 1997). However, people have the power and motivation to change their situation (Chambers, 1997;Narayan, 2005;Sen, 1999). This is why "an empowering approach to poverty reduction is grounded in the conviction that poor people themselves are invaluable partners for development, since they are the most motivated to move out of poverty" (Narayan, 2005, p. 3).

Frame of reference/adaptive preferences
Adaptive preference refers to the negative impact that adjusting to negative circumstances may cause to the individual's freedom (Nussbaum, 2000;Sen, 1999 instance, the CA's rejection of a utilitarian approach to welfare measurement is partly based in the realization that "[o]ur desires and pleasure-taking abilities adjust to circumstances, especially to make life bearable in adverse situations. The utility calculus can be deeply unfair to those who are persistently deprived" (Sen, 1999, p. 62). Due to the existence of such adaptive preferences, people who suffer from deprivations might actually report a level of wellbeing that is higher than what they would report were they not suffering from such deprivations.

Increasing aspirations, increasing capabilities
The value of the PS intervention for people experiencing poverty lies in its ability to induce them to connect themselves to a world out of their situation of deprivation (to "become Green"). Through a self-diagnosis and dialogue with a mentor, the individual will acquire a new vision that will allow them to free themselves from what Freire (1973) defined as the state of oppression.
Aspirations are hopes or ambitions to achieve something in life (Ibrahim, 2011). Appadurai (2004, p. 59) indicates that "in strengthening the capacity to aspire […] the poor could find the resources required to contest and alter the conditions of their own poverty". The way individuals understand the world and how they select their preferences and aspirations for the future is a socially constructed process. Only when individuals consider their own capacity to aspire, do these processes and constraining cultural norms become addressed and revealed. For example, Conradie and Robeyns (2013, p. 565) argue: Thinking about, talking about, and reflecting upon aspirations, especially when this is part of a group process that creates a supportive and encouraging atmosphere, motivates people to use their latent agency to make changes in their lives, which will expand their capabilities.
To achieve their aspirations, individuals experiencing poverty have to exercise their voice and their human agency (Appadurai, 2004), and having these aspirations is crucial in order to work towards a positive change. Conversely, not having "adequate" aspirations may inhibit people from achieving a life they would value: As first argued by Ray (2006), and later formalized by Ray (2011, 2017) and Dalton et al. (2016), people can get stuck in poverty due to aspiration failures, that is, due to their aspirations not being ambitious enough, or due to the feeling that there is just no way to achieve them. As aspirations are formed in the social context in which a person lives, a poor individual is likely to adjust her aspirations to what others have achieved who live in a similar situation as herself. This might well mean that a life out of poverty seems unachievable and is thus not aspired to. Hart (2012;2016) embeds the concept of aspirations into the Capabilities Approach, showing that aspirations are an important step in the realization of many functionings.
The PS addresses this situation, starting with the measurement process: For each indicator, the level Green is presented as an achievable situation of non-deprivation. Thus, Green is not an abstract answer category among others but helps families to imagine themselves as not deprived. Hence, already in the measurement phase, families are encouraged to rethink what they see as possible and start adjusting their aspirations.

Defining priorities: improve on what you value and have reason to value
One of the important contributions of the CA is the focus on the things that a person values and has reason to value. This is closely related to the principle of each person as an end: if the focus lies on how particular arrangements or programs affect individuals, and it has to be assured that the interests of each person are served, the specific objectives of a program necessarily have to be defined by those affected by it. This is what the PS mentoring program aims to accomplish: participating families identify which of the poverty indicators matter most to them, as a starting point for the mentoring process. Based on these priorities, the mentors then work with the families to analyze potential causes of the problem and to identify and develop solutions. In the PS mentoring framework, the "right" solution for a given indicator might look very differently for one family compared to another.

Method and Data
The present analysis uses administrative data from over 9,100 FP microfinance clients who were randomly selected to participate in the PS program at any point between August 2015 and June 2017. New participants entered the program every month, and this paper refers to the program entry survey of each individual as "baseline" and to the subsequent surveys as "follow-up". Follow-up data from within the study period is available from around 2,600 participants.

PS data consists, for each participant, of a brief socioeconomic survey and the 50
Stoplight indicators coded as Red, Yellow, or Green. This paper aggregated the 50 Stoplight indicators using the Alkire-Foster (AF) methodology (Alkire & Foster, 2011), an axiomatic approach to multidimensional poverty measurement whereby the number of deprivations each participant suffers is counted, and those above a defined threshold are identified as poor. The AF methodology is useful for the purpose of this paper, as it allows aggregation of multiple deprivations into a single multidimensional poverty score, thus providing a succinct metric to evaluate the program's effect. The PS itself uses a dashboard approach, and does thus not provide a summary statistic for a person's multidimensional poverty. pág. 123 ISSN 2362-194X Journal de Ciencias Sociales Año 8 N° 15 This paper created a metric for overall poverty (using the Yellows as the deprivation cut-offs ) and for extreme poverty (using the Reds as the deprivation cut-offs ). In each case, the union criterion for poverty identification is used, meaning that a participant is considered "poor" if she is deprived in at least one indicator (Alkire & Foster, 2011  The program effect is estimated for the pooled database as well as separately for semesters two through four (no follow-up data is available for the first study semester). Estimates are based on OLS with loan office-level fixed effects, which account for the substantial differences in economic and social conditions between the different regions of Paraguay. The model is defined as where is the deprivation count for deprivation cut-off (extreme or overall deprivation within each indicator), is a dummy variable with the value 1 if the survey is a follow-up and 0 if the survey is baseline, is the number of days since the first observation in the entire sample (August 7 th , 2015); is the monthly family per capita income level, in 10,000 current Paraguayan Guaraní (at the time approximately USD1.8); and is the residency area (which is a dummy taking the value 1 if the client lives in a rural area). are the loan office-level fixed effects; the coefficients on these loan office dummies are not of interest for this study and thus not reported, but they are included to control for heterogeneities between loan offices.  Table 2 shows, for each study semester and for both overall and extreme poverty, the observed poverty headcount ratio and the adjusted headcount ratio 0 for baseline and follow-up surveys. The data reveals a negative correlation between program participation and multidimensional poverty. In each semester and for both overall and extreme poverty, the percentage of poor survey-takers ( ) is larger among those doing their baseline compared to those doing a follow-up survey. These differences are statistically significant, providing a first indication for the effectiveness of the PS. A similar picture arises for the adjusted headcount ratio ( 0 ). However, in this case, the difference between baseline and follow-up surveys is statistically significant only for semester two and three.   The size of the effect for the overall deprivation count is a reduction of between 0.02 and 0.05, corresponding to a reduction of between 1 and 2.5 Yellows or Reds (given the total number of 50 indicators). For the extreme deprivation count, the reduction amounts to between 0.006 and 0.02, corresponding to a reduction of between 0.3 and 1 Reds.

Results
For all models, higher income is significantly correlated with lower multidimensional poverty, while living in a rural area is only consistently associated with higher levels of extreme multidimensional poverty. Furthermore, there is a clear negative time trend: other things equal, the expected multidimensional poverty level is lower the later the survey is taken. Office-level fixed effects included but not reported. Standard errors clustered at office level.

Discussion
The results are encouraging and support the conclusion of the previous studies that participation in the PS program helps families reduce their deprivations. On average, study participants spent about six months in the program, over which time they could decrease their Reds by between 0.3 and 1, and their Yellows or Reds by between 1 and 2.5, in addition to a favorable time trend of a decreasing overall and extreme deprivation count. Note that the entire study population are microfinance clients, which might explain at least part of the overall decrease; the results suggest that participating in the PS program in addition to the microfinance program increases participants' welfare even further. Overall, the results support the notion that a poverty self-assessment and mentoring program can help people overcome poverty, as the CA would suggest. In other words, this paper provides evidence that suggests that participating in the PS, a multidimensional poverty self-assessment and mentoring intervention, might increase participants' agency and thus reduce their deprivations.
As discussed, the PS indicators measure well-being mostly on the level of functionings, or resources, and not on the level of capabilities. Therefore, the evidence brought forward in the empirical analysis does on allow to conclude that participants managed to expand their capabilities. However, as the PS program theory is in line with the CA, from the moment of reflecting on the status quo (changing frames of reference), to defining priorities for improvement (working on improving what a client values), to building of agency to act on these priorities, it seems at least possible that such a capability expansion has indeed occurred.
However, this analysis has some important limitations, most prominently the lack of a control group, which makes it harder to isolate the program effect. Furthermore, as follow-up data is available only for about 35% of program participants, systematic program attrition is a serious concern. While a robustness analysis suggests the results don't change qualitatively if the analysis is limited to those clients for whom follow-up data is available (results not reported), a potential bias from systematic attrition cannot be ruled out. Finally, the validity of our results might also suffer because the average time difference between baseline and follow-up surveys was only around 6 months. This time difference may well be too short for sustainable changes in multidimensional poverty, especially changes that are related to an increase of agency and capabilities, which tend to be long-term processes. Finally, as argued in the previous paragraph, based on our results a capability expansion can only be assumed, not evidenced.

Conclusions
This paper presents evidence on the question whether a multidimensional poverty selfassessment and mentoring intervention can increase participants' agency and thus reduce poverty. The theoretical part of this paper developed the PS as a Capability Application, using the CA to explore the potential of the PS intervention to increase agency and decrease deprivations, while the empirical part presented the results of a program evaluation. Both the theoretical analysis and the results of the empirical study conclude that multidimensional poverty self-assessment and integrated mentoring is a promising approach to support families on their pathway out of poverty. Specifically, the empirical evidence showed that participating in the PS program in addition to the microfinance program increases participants' welfare and reduces their deprivations.
In relation to the limitations of this paper, more research is needed to truly understand the effects of the PS on multidimensional poverty and capability expansion, and the mechanisms through which agency might be enhanced. As summarized in the discussion section, the results should be validated in further studies that include a control group, account for attrition rates, and take a longer time horizon. In addition, the analysis should be expanded to learn how various dimensions of multidimensional poverty are affected by the program, or whether the results can be replicated by using other conceptualizations of multidimensional poverty. However, it is important to note that this paper summarized a part of a larger research project that uses mixed methods to answer some of these questions. In particular, this larger project explores local notions of empowerment and the PS's perceived effect on it through participatory qualitative as well as quantitative methods; and this research started a new randomized evaluation in 2017 that not only adds a control group to better understand the program effect, but also allows to compare outcomes of different mentoring models and intensities.
In relation to possible extensions to this paper, some important open questions remain about the program with regards to the CA. Without any claim to comprehensiveness and in no particular order, these include the following. First, the pronounced objective of the PS is to help participants become Green in everything. Considering that a number of the indicators are to be located in the space of resources or functionings (as opposed to capabilities), it is not clear how to reconcile this prescribed objective with the claimed focus on agency and the areas a participant values. pág. 129 ISSN 2362-194X Journal de Ciencias Sociales Año 8 N° 15 Second and related, one might argue that in many cases the PS does not promote capabilities, but specific functionings and thus specific lifestyle choices. Given the inherent difficulties in measuring "choice", it becomes a common problem of multidimensional poverty metrics based on the CA. However, the PS is not just a pure metric that needs to resort to proxy measures where the real concept of interest is not available. Rather, its indicators are developed with the explicit purpose of encouraging reflection and promoting agency.
Therefore, one might argue that resorting to the space of functionings or resources, rather than capabilities, is a bigger problem for the PS than it is for other metrics.
Third, the CA emphasizes the individual as the center of development, not the groups an individual might belong to. However, the PS, while typically working with individuals, has a focus on families: According to FP, the family is the central unit of analysis, and in fact most indicators are formulated at the level of the family. FP argues that families generally share resources, and it will be impossible to eliminate the poverty of, say, the family's children without at the same time eliminating the poverty of the mother. A scholar developing an instrument strictly from the perspective of the CA would most likely not make that choice, and its implications for the interventions from the perspective of the CA deserve further scrutiny.