Field Work? An amazing learning opportunity for aspiring development economist.

by Cécile Pierre

In January 2022 I started working as RA and Field Coordinator for Prof. Lucia Corno, on the Project Breast-Ironing Breastfeeding and Child Mortality. This study investigates a harmful tradition common in Cameroon called “Breast Ironing”. It is a traditional practice that involves scarifying, massaging or pressing the breast of adolescent girls and is usually performed by close relatives such as mothers, grandmothers and aunts in order to reverse and thwart the development of the breast. As part of this position, I moved to Yaoundé, Cameroun for 6 months. My job was to set up and supervise the intervention phase of the project, it followed a baseline data collection, and another data collection would come afterward.  

This was not my first job as a RA, but it was my first time in the field and it has probably been the most enriching experience of my life. First of all, it gave me an unrivaled opportunity to understand the issue I was working on. I met victims and perpetrators of breast ironing, I had the chance to witness their testimonies and grasp better the logic behind this practice. Secondly, this project brought me to Central Africa, with all the adaptive challenges this entails, but the core value of this experience is, I think, present in any kind of fieldwork, no matter how far from your home and your culture this field is. What I mean by this is that, yes, a big share of what made this experience so salient for me is linked to being in a completely new country, new culture, and learning to adapt to new standards of life. But what I‘ll remember and carry with me for the rest of my working life are the skills I acquired during this time.

As an aspiring development economist, I have worked with complex datasets, learned to code, and followed advanced theoretical classes during my studies and previous RA position. But even though I thought I already had a clear understanding of the challenges and logistics embedded in the rollout of an RCT, it was not until I found myself actively in the middle of one, that I could properly measure the extent of it. You’re taught about the rules of a good experiment, you probably also can imagine that data collection is not an easy task, and that there is a practical side to consider in every project. But it is also so easy to forget that this line in your dataset is a person, with a life, a job, a family to feed and only 24 hours in her day. I have learned that a perfect design in your office has no value if it does not take into consideration the reality of the people and the contexts in which you are working. This may sound evident, I also thought I knew this already, but there is nothing like a first-hand experience to really understand something.

In a nutshell, the main takeaways for me have been the following: The importance of timing, especially in agricultural settings. People have rhythms, work, and cultural rhythms, that you cannot ignore. This also concerns weather, being in the field during dry or rainy seasons is not the same, this affects your budget and the quality of your output. One week later, or one week earlier is not the same. Secondly, there is a delicate balance between knowing when to listen to your local team and when to push for your way of doing things. Especially when you work in a context like I did, where cultural norms are very strong and different from yours. Naturally, you care for the quality of data you’re collecting, or of the intervention you are rolling out, so you always have in mind all the criteria that should be met for them to be theoretically valid. You should push for such criteria to be respected and know that you’ll face opposition because sometimes your collaborators do not understand the value of doing things a certain way. But you should also value their inputs, they know the people and the culture you are working with and really often there are things you would have never imagined that can have an enormous impact on your work, no matter how open-minded you are or how aware you think you are, you’ll be surprised. To give you an example, there is a part of Cameroun in which we collected data were giving 1 item of something, in our case 1 piece of soap, is a sign of witchcraft. It is hard to take this as a serious issue, however, it is. Communication is also a key element of course, you have to adapt to the communication standard of where you are. In my case it was the speed of speech, people in Cameroun no matter how educated and smart they are, speak slowly, much slower than what I do. I realized that the speed at which I was speaking was an obstacle to efficient communication with my colleagues. Moreover, and a guess this is true for every partnership, speak to your collaborators in terms that make sense to them, and emphasize what has value for them, and sometimes you’ll have to explain things that seem so evident for you.

I believe this is addressed to recent graduates that, as I did, contemplate the idea of applying for a field RA position. My advice to you is: do it. It might be the hardest thing you’ll ever do or the most amazing time of your life, but the amount of knowledge you will retrieve from it is invaluable. For me, it has been such a great time. On top of the learning opportunity it has represented, my stay in Cameroun was a blast. Of course, it has been hard, I had my share of difficulties, but nothing compared to the good side of it. I met amazing people, learned so much about myself, and discovered that I was much stronger than I thought I was. I also witnessed that you really don’t need much to be happy, cliché I know, but so true.

As I am writing this, from my apartment in Kigali, Rwanda, where I just started a new field coordinator job, I realize how much I have learned and grown over the past 6 months, both as a young adult and as an aspiring development economist. I am deeply grateful to have the opportunity to do what I do.

Cécile Pierre

If you want to connect with me write me an email at I’ll be happy to have a chat with you!

A demain, inch’Allah. Fieldwork in a not-so-easy country

During the last two months, as a LEAP student, I had the privilege to follow the first steps of the project Peers in Action in Ouagadougou, the capital city of Burkina Faso.

The project’s aim is to study the impact of peer pressure in changing attitudes towards harmful traditional practices among adolescent girls, with a specifical focus on female genital mutilations (FGM) and child marriage. The intervention, which will start in September 2022, is in partnership with the local branch of the NGO Children Believe, who will be in charge of setting up girls’ clubs, in order to create a safe space for them to share their thoughts and also to learn new skills that will boost their self-confidence.

Burkina Faso is not an easy place to work in. The country’s history has been marked by frequent episodes of political instability (the last significant political event is a coup d’etat at the start of this year, which has left the power in the hands of the military), and the central government is not currently able to control vast areas of the country, leaving the local population prey of continuous terrorist attacks perpetrated by jihadist groups.

Here in Ouaga I mainly worked to coordinate the works of the baseline data collection, based in the offices of the local data collection firm IHfRA (Innovative Hub for Research in Africa). Unfortunately, I was not able to personally go in the field (it is highly discouraged for foreigners to leave the capital) due to the security risk, but I participated in all the other steps of the data collection, such as programming the questionnaires and other materials, training and selecting of enumerators,  checking the quality of the data that we collected and cleaning it.

 Although I already worked as a research assistant, and even did fieldwork in Italy, taking part to this project in the last few months taught me many important lessons about what conducting research in a developing country feels like, and about what actually goes on in the field, with its challenges and its issues.

One of the things that I learned is the importance of being flexible, and of being able to come up rapidly with alternatives to what was previously planned. Countless times we had to rearrange aspects of the research once we faced the reality of the field. Even after discussing for weeks, or even months, central points of the projects such as which should be the relevant population to be surveyed, or the list of villages in which to conduct the analysis, these aspects needed constant readjustments, due to the ever-changing security situation, or more in general as we started to get a better idea of the reality of the villages in which we were working in.

At times we also had to adopt a flexible approach due to the lack of data available that we had. Even basic information such as rough estimations of the population living of an area, or the position of villages in which we planned to do the data collection were not always available, or had not been updated in so many years that they were of no use for the design our analysis.

An aspect that I thought was particularly interesting was the importance of a good knowledge of the context in which we were going to work. Information such as in which season it would be better to conduct the intervention in order to find more people available to attend the clubs (it is autumn, since in summer, which is the rainy season here in Burkina, many people are too busy with work in the fields), or which ones of the local languages the enumerators to be sent in a certain area should be able to speak, are all essential components that will guarantee a good output for the research, that are lost by looking at data alone.

Understanding that the researchers’ culture might be different from the local one is another thing that should be taken into account when preparing the materials that will be used in the data collection. Concepts and words that to us might have a clear meaning might not have the same one for the population that will be surveyed. For example, while discussing with the enumerators about the questionnaire translations, we discovered that the question we asked about how many times the interviewee played with her friends in the last week hid an unintended sexual double meaning, which could lead respondents to give us a different answer to the one we actually wanted to know. Similarly,we realized that we had to explicitly state in the questionnaires to list in the same household children from different marriages of fathers in polygamous relationships if they were living together, or else enumerators would have listed them as belonging to different households.

Furthermore, culture can change across different groups of people even in the same country. This became apparent when we had to identify objects or colors related to the FGM ceremony, and every person we spoke to gave us a different answer, since each culture had its own symbols that were used in the ceremony!    

Being able to coordinate this data collection in Burkina Faso has been a great opportunity to learn new things and to broaden my horizons both from an academic and personal point of view, and I would strongly encourage everyone considering a career as a researcher in development economics to engage in something similar. I believe that getting a first-hand idea of the necessary practical actions that need to be undertaken and of all the issues that might show up in the field will be crucial when planning to do your own research, and that this experience can not be replaced by coursework alone.

I conclude by wishing the best of luck for the implementation phase of Peers in Action, and for the future of Burkina Faso in general! These months have been so formative and I will always bring memories of them with me.

Alessandro Palucci

If you want to connect with me write me an email at, I’ll be happy to have a chat with you!

Can a 1-minute self-recorded video boost children’s aspirations?

By Raffaella Dimastrochicco

Aspiration trap

There is large evidence in the literature that low aspirations are a common issue among people coming from fragile socio-economic backgrounds. This tendency to under-aspire is detrimental for poor people, as it prevents them from investing in education and ultimately condemns them to lower wages, thereby reinforcing their poverty status (Appadurai, 2004; Ray, 2006; La Ferrara, 2019).

Several interventions have been tested and implemented to break this vicious circle – known as “aspiration trap” – and increase aspirations, which range from organizing tutoring programs and academic counselling (Carlana et al., 2017) to institutional changes in the political rules (Beaman, 2012) and the provision of statistical information on the benefits of investing in education (Nguyen, 2008). One further option involves the exposure of children to role models, with both in-person interventions (Porter and Serra, 2019) and showcasing inspirational movies (Riley, 2017). This latest option, in particular, is a cheap and easy to replicate treatment. To which extent can the length and complexity of the video be reduced while still generating a significant treatment effect?

Experimental setting

To answer this question, we conducted a RCT in Naples (Italy) in May 2021, on a sample of 295 primary and secondary school students from fragile families. We showed a very short (1 minute) self-recorded motivational video, in which a young immigrant adult who grew up in Naples told the story of how, starting from a situation of difficulty, they eventually managed to find their passion and this led to happiness and satisfaction with their lives, against all odds. The video was shown on a tablet during face-to-face interviews conducted by trained enumerators. The main goal of the video is to boost self-confidence and encourage students to find their passion, by aiming at what they really wanted to do in life regardless of their precarious conditions.

This picture represents one enumerator while interviewing a primary school student and reporting the answers on a tablet.
Face-to-face interview

Aspirations were measured by asking the students two open questions: “Which school, if any, would you like to attend when you grow up?” and “What would you like to become when you grow up”. Self-confidence was measured through a series of closed questions on a 4-level agreement scale. Data on students’ self-confidence and school and career aspirations were collected right after they watched the video.


Results from the experiment are promising: we detect a significant increase in the self-confidence measure among the treated students by 29% s.d.. When looking at school aspirations, the video treatment increases the likelihood the respondent chooses the “academic track” by 28.5% a s.d., as hoped for. At the same time, there is also a positive effect of the video on career aspirations: among treated students, significantly more children aspire to the most prestigious of the 7 categories of jobs we identified; this category includes jobs as sportsmen, politicians, artists, etc. These results are robust to controls on gender, school level, and nationality.

Limitations and conclusion

There are two main limitations. First, by design, collected data allow us to measure only the short-term effects of the treatment; it would be interesting to measure whether these effects persist over the medium- to long-term.

Second, there may be a form of Hawthorne effect. One other interpretation suggests that children update their aspirations based on the role model experience, without tailoring it to their own situation and passion. Both qualitative and quantitative data collected during the interviews point out in this direction: children seem to lack information to make informative choices, and when provided with an example of a career path that goes beyond their everyday experience, they tend to follow it. In light of these results, it emerges the importance of providing children with a variety of examples about available career paths that are not those they can encounter in their everyday life.

Nonetheless, results suggest that the treatment is effective in boosting aspirations and self-confidence.

I am a MSc student in Economic and Social Sciences, and this article draws from a larger study constitutes my master thesis. If interested in the topic, do not hesitate to get in touch at, I will be happy to have an exchange with you.


Appadurai, A. (2004). “The Capacity to Aspire: Culture and the Terms of Recognition” . Culture and Public Action, edited by V. Rao and M. Walton. World Bank, pp. 59-84.

Beaman, L., E. Duflo, R. Pande, and P. Topalova (2012). “Female Leadership Raises Aspirations and Educational Attainment for Girls: A Policy Experiment in India.” Science, 335, 582–586.

La Ferrara, E. (2019). “Aspirations, Social Norms and Development”. European Economic Association Presidential Address. Cologne.

Nguyen, T. (2008). “Information, Role Models and Perceived Returns to Education: Experimental Evidence from Madagascar”. MIT Job Market Paper.

Porter, C. and D. Serra (2019). “Gender Differences in the Choice of Major: The Importance of Female Role Models”. American Economic Journal: Applied Economics, 12(3): 226-54, 2020

Ray, D. (2006). “Aspirations, Poverty and Economic Change”. Understanding Poverty, edited by A. Banerjee, R. Benabou, and D. Mookherjee. Oxford University Press.

Riley, E. (2017). “Increasing students’ aspirations: the impact of Queen of Katwe on students’ educational attainment”. In CSAE Working Paper WPS/2017-13.

The Impact of Networks on Integration and Social Norms of Migrant Women in Denmark

Intimate partner violence (IPV) is prevalent worldwide even when measures of gender disparity are considered. For example, despite Denmark having the second-lowest Gender Inequality Index score in the UNDP ranking, IPV is still prevalent: in 2014, national rates were not distinctly different from global proportions of 1/3 (FRA, 2014). Immigrant women disproportionately composed this statistic. While only 11.8% of women were classified as immigrants nationally, 42% of women’s shelter stays were non-Danish in 2019 (LOKK, 2020). In March 2021, we are awarded the LEAP student grant to explore these differences further and understand the role of norms in driving these statistics.

The research that motivates our proposal is by Alesina, Brioschi and La Ferrara (2021). Specifically, this paper showed that contemporaneous rates of IPV acceptance were higher amongst women that descended from historically patrilocal ethnicities, suggesting information and social protection influence incidences of violence. Our proposal aims to build on the emphasized persistence of IPV in Alesina et. al (2021) to understand whether attitudes, and subsequently incidences, are malleable when exposed to a contrasting perspective. The aim of our research is to develop on the economic and anthropological literature to understand to what extent social networks influence the norms of native and non-native women victim of IPV.

Hypotheses on the benefits of support networks are manyfold. Native-to-native and migrant-to-migrant bonds could help connect women with someone who has similar experiences, and background, which they can relate to and share. For the latter, if these women are less integrated in Danish society, there is likely a higher risk of being marginalized, leaving fewer means to cope with and overcome domestic violence. The relationship between migrants and natives could help migrants integrate into Denmark, providing legal, economic and social guidance to help them assimilate with the local culture, markets and processes. On the other hand, natives could also benefit by diversifying their social networks outside of those that her perpetrator is familiar with. We then plan to study the impact of such networks on a multitude of socio-economic outcomes, such as employment and job search, social security and health, both mental and physical and aim to conduct a survey through the registry to collect information on norms.

In order to analyse the causal effect of interactions among women of a matching or a different ethnicity, we exploit several sources of exogenous variation in the ethnic composition of the shelter population. At the national level, a variety of measures have been imposed to increase contact between natives and non-natives, and, in particular, we focus on a refugee dispersal policy introduced in the 1980s that allocated migrants as evenly as possible throughout the country to reduce formation of “ethnic enclaves”. At the shelter level, we plan to leverage the fact that women are randomly assigned to different floors, so that the “ethnic composition” of each floor is exogenous. Hence exposure to co-nationals or to women of different ethnicity, and consequently the likelihood of bonding with them, is also randomly determined. As there may be endogenous room switches, we have reached out to several shelters across the country about the possibility of a more robust experimental design to explore how contact with different perspectives can shape social norms.

Given shelters collect the social security numbers of residents, the shelter stays can be combined with the national administrative data. This has provided us access to a rich dataset, which we have purchased through our LEAP grant, containing information on a plethora of dimensions, including health-related information, criminal records, job market outcomes, and social security benefits, about all women between 18 and 65 years old living in Denmark for the last 15 years, along with eventual partners and children.

We have recently received the data and are in the process of cleaning them. We have started conducting a preliminary analysis to understand the socio-economic correlates of shelter stays. We look forward to disseminating our results in due course.


European Union Agency for Fundamental Rights, 2014. Violence against Women: An EU-wide Survey. Luxembourg: Publications Office of the European Union.

Alesina, A., Brioschi, B., & La Ferrara, E. (2021). Violence against women: a cross‐cultural analysis for Africa. Economica, 88(349), 70-104.

The psychological dimension of gender inequality

By Sveva Vitellozzi

In March 2021 I was awarded with the LEAP Student Grant to conduct a lab-in-the-field experiment in Kenya, with the aim of assessing the impact of women’s mental load on cognitive functions and labor productivity.

The idea for this experiment came during my first year of Ph.D. during the class on Behavioral and Experimental Economics while we were asked to write a research proposal of a hypothetical lab experiment. Throughout this course, we talked about the strand of academic literature on the “psychology of poverty”, according to which poverty reduces cognitive functions as it causes daily stress about financial needs (Mani et al. 2013; Schilbach, Schofield, e Mullainathan 2016). The idea of writing my research proposal on women’s mental load, cognitive functions, and productivity was inspired by a comic by Emma, a French author that explains in a very enjoyable and accessible way what mental load is and what it entails (click here to read the article).

Why should we care about mental load?

Mental load refers to the total sum of responsibilities related to the management of household activities. Consequently, women spend not only more “physical” time than men in household’s chores, but also more “mental” time, which constitutes an important psychological burden for them. In developing countries, this burden risks being even more pronounced because of the strict gender roles defined in many traditional societies.

Being an invisible phenomenon, mental load has been long neglected. Only in recent years social media and the press, mostly in Western societies, have started paying more attention to it but it has not been properly addressed in the economic literature yet, nor have its consequences  on economic outcomes. Even though the nature of mental load may differ across countries, as the household’s needs women take care of change across contexts and cultures, its burden is carried mostly by women worldwide.

Hypothesis and design

The main hypothesis of the study is that mental load, by inducing daily and pressing thoughts about household management, negatively affects women’s labor productivity by reducing their attention, which is a limited cognitive resource. In developing countries, and especially in Africa, important gender productivity gaps are observed (Kilic, Palacios-López, e Goldstein 2015; Backiny-Yetna e McGee 2015), but we still do not know whether and to what extent the psychological dimension plays a role in widening this gap. In contexts where the informal labor market is well established, workers are usually paid with a piecework scheme: the more you produce, the more you are paid. The basic assumption is that having the mind occupied by other thoughts can reduce a subject’s attention at the workplace and, consequently, their productivity. This can in turn entail a series of other important consequences for women’s empowerment and for gender equality more broadly.

To test this hypothesis, I plan to run a lab-in-the-field experiment in Kenya in February 2022. Even though the design of the experiment still needs to be finalized, the aim is to trigger thoughts related to household management among participants within the treatment group. They will then be asked to perform an effort task that requires both care and attention, to control for the causal mechanism being tested. The task needs to be sufficiently unrelated to those daily activities usually carried out by the participants, to isolate at best the confounding effects of individual ability. Women in the treatment group are expected to exhibit a lower score in the effort task than those in the control group. 

Pathways for the future

Mental load represents just the tip of the iceberg of the psychological dimension of gender (in)equality. While mental load risks entailing a series of negative effects, such as increased stress and anxiety, it is not the only psychological factor that can contribute to widening gender inequalities worldwide. For instance, women are more likely than men to suffer from depression (Nolen-Hoeksema 2001), which can in turn affect a series of economic outcomes such as labor supply and employment, saving and investment decisions, and labor productivity (Ridley et al. 2020). Especially in developing countries, studies focusing on gender inequalities do not pay particular attention to these psychological dimensions. However, psychological well-being constitutes a crucial dimension of women’s empowerment and is essential to understand to what extent it shapes women’s daily lives and decisions. Further research is needed in this direction to inform at best policymakers and development practitioners to help them design more effective gender-driven programs and policies.


Backiny-Yetna, Prospere, e Kevin McGee. 2015. Gender Differentials and Agricultural Productivity in Niger. Policy Research Working Papers. The World Bank.

Kilic, Talip, Amparo Palacios-López, e Markus Goldstein. 2015. «Caught in a Productivity Trap: A Distributional Perspective on Gender Differences in Malawian Agriculture». World Development 70 (giugno): 416–63.

Mani, Anandi, Sendhil Mullainathan, Eldar Shafir, e Jiaying Zhao. 2013. «Poverty Impedes Cognitive Function». Science 341 (6149): 976–80.

Nolen-Hoeksema, Susan. 2001. «Gender Differences in Depression». Current Directions in Psychological Science 10 (5): 173–76.

Ridley, Matthew, Gautam Rao, Frank Schilbach, e Vikram Patel. 2020. «Poverty, Depression, and Anxiety: Causal Evidence and Mechanisms». Science 370 (6522): eaay0214.

Schilbach, Frank, Heather Schofield, e Sendhil Mullainathan. 2016. «The Psychological Lives of the Poor». The American Economic Review 106 (5): 435–40.

Accelerating poverty elimination through the Poverty Stoplight program — Experience at Fundación Paraguaya

By Fabrizio Cabrera

Last winter, I did an internship at the Fundación Paraguaya, where I collaborated within the department of research and methodology. My role was to support the senior researcher Katharina Hammler, with quantitative and qualitative analysis in order to write a report on the preliminary evaluation findings of an ongoing study: the “add-on” impact of the Poverty Stoplight (PS) program for Fundación Paraguaya’s microfinance clients.

The PS program

The Poverty Stoplight (PS) is an interactive survey and coaching model that complements Fundación Paraguaya’s microfinance program, broadening the focus to multidimensional poverty. The Stoplight is characterized by three key features:

  1. Multidimensional snapshot of lived experience: Program staff work directly with participants to complete an easy-to-use, picture-based survey to represent their quality of life across six dimensions (Income & Employment, Health & Environment, Housing & Infrastructure, Education & Culture, Organization & Participation, and Interiority & Motivation). These indicators are self-assessed by clients as red (severe poverty), yellow (moderate poverty), or green (out of poverty). The approach is designed to center the lived experience of participants, creating data from the bottom up.
  2. Solutions that start from the participant: After the survey is facilitated, participants choose which specific indicators of poverty they want to change from red or yellow to green, as well as the action they think is most likely to produce change. Sometimes this involves taking action as individuals; sometimes it means utilizing community resources or peer-to-peer support; in other cases, it involves accessing government programs.
  3. Personalized coaching to support solution implementation: Program staff take an individualized approach to support participants as they pursue change. Supports include collaborative identification of core challenges, as well as reflection exercises to support continuous improvement of poverty alleviation approaches. 

The PS is applied across a broad range of contexts, ranging from poverty alleviation programs to assessments of quality of work life in professional contexts. The report focused on the application to microfinance, seeking to elucidate the “add-on” impact generated on top of the microfinance program. As a growing number of ever more diverse organizations implement the Poverty Stoplight model around the world, the need for robust evidence on the model’s impact is increasing, and so is the need for evidence on how to best implement the program.

Critical questions of the RCT

To support evidence-based scaling of the model, we deployed a rigorous randomized controlled trial design to engage three critical questions:

  1. What is the impact potential for this model?
  2. What types of participants are most likely to benefit?
  3. What programmatic features are most likely to optimize impact?

Summarized findings of the report

Our results show that the PS accelerated multidimensional poverty reduction by about half of a standard deviation, which corresponds to turning two or three PS indicators from red or yellow to green. While financial indicators showed the greatest poverty reduction, benefits also materialized in non-financial dimensions of poverty.

There were important nuances in program effects for participants with different baseline incomes. While we observed reductions in multidimensional poverty for participants across the income spectrum, suggestive evidence indicates that the microfinance program alone drove the lionshare of multidimensional poverty reduction for lower-income participants, while the PS survey and coaching model drove impact for higher-income participants. 

The results also revealed the impacts of mentoring could be increased slightly by 0.05 standard deviations (or about half a PS indicator) by providing coaches with explicit contact targets that guided how often they contact families). Qualitative follow-up suggests that the regular contact may have contributed to a critical trust-building process between coaches and participants. Notably, the study did not find evidence of impact for a group that just received the PS survey (without follow-up coaching).

Even though the findings are specific to the study context, some general recommendations arise, including (a) targeting participants across the spectrum of multidimensional and monetary poverty; (b) considering  the potential of attending to a broad range of multidimensional poverty indicators, even outside of an organization’s core area of competence; (c) providing follow-up support to participants; and (d) investing in relationship building, and considering setting explicit targets or guidelines around regularity of communication.

Reflections on the experience

My experience within this institution suggests that this program represents real hope for the less disadvantaged people in our community. By doing the field work and monitoring the progress of households in their way out of poverty, the institution is able to add value to design better policies in favor of the people whose voice we do not usually hear. Thus, the quantitative and qualitative analysis of this program not only gave us tools to help the households, but the analysis also empowered the households themselves. The recurrent meetings were a proof for this improvement. From the experience of visiting the households at their homes, I can tell that the program not only made them aware of the type of poverty they were having, i.e., their red dots, but it got them motivated to improve their quality of life by tackling specific aspects of their poverty. At the end of the day, their dedication was something so contagious that fueled the passion of the team  inside and outside the project. With projects like this, the Fundación helps more than 86.000 families and as households that join the different programs keep growing, the challenge is to meet the large scale demand without sacrificing the unique add-ons that the institution offers: providing follow-up support to participants; investing in relationship building; considering setting explicit targets or guidelines around regularity of communication; etc.

Fabrizio is a Fulbright Fellow that is currently doing an MA program in economics at New York University. Please feel free to contact him if you need any further information or you have any questions:

Leveraging pro-social behavior to tackle educational poverty: some takeaways and future perspectives

By Gaia Gaudenzi

There is evidence around the world (two examples here and here) telling us that COVID-19 and the decision to close schools during the lockdown had negative effects on pupils’ educational attainments and well-being. In most cases, students who are in “cognitive” educational poverty (meaning that they are not achieving minimum proficiency levels in math and reading) come from socio-economically and culturally disadvantaged backgrounds (For more information and data about educational poverty in Italy, please the Save The Children’s report here).

The program:
The Tutoring Online Program, designed by Prof. Eliana La Ferrara (Bocconi University) and Prof. Michela Carlana (Harvard University), was born during the 2020 lockdown precisely to mitigate the negative effect of distance learning on middle school students, particularly targeting the disadvantaged ones. The program consisted of assigning a tutor (a volunteer university student) to a student in need. The tutors provided weekly individual online tutoring to the middle school students they were assigned to. The formula has been piloted during the first lockdown in Italy in 2020, where 530 students received tutoring from April to May. The program had impressive results, especially considering its short duration. Participating students showed substantial and significant improvements along four dimensions: educational performance, aspirations, socio-emotional skills, and psychological well-being (a draft of the paper is available here).

Looking back:
I had the opportunity to be the Project Manager of the second edition of the program, during which tutoring has been offered to 810 middle school students from November 2020 until May 2021.

The big challenge of this second edition was the length of the program, which would make it harder for university students to commit to the level of effort observed during the short pilot. I suspected the altruistic behavior by tutors observed in the pilot phase was the result of temporary enthusiasm, driven both by a genuine desire to help those most in need during an unprecedented crisis and by the sudden increase in free time due to home confinement. In a situation where the emergency becomes the norm (and therefore the sense of urgency decays), it would have been very hard to find enough university students  willing to give away their time for free. I was wrong. In a matter of weeks, we received enough applications to start off.

Some reflections and takeaways:
I believe this program is an example of how pro-social behavior can be leveraged to improve public service delivery (I leave the discussion on whether this might be a viable way or not for another time). I also believe that the power of the universities’ brand (Bocconi and Harvard) played a role in the success of the recruitment campaign. However, I definitely believe that the latter is unlikely to be enough to prompt university students to embark on a 7-month program that requires a tutors’ constant effort and presence. It is probable that highly motivated people, with a genuine willingness to help, were more prone to self-select into the program. At the moment, there is no solid evidence that this was the case because there was no variation in the way tutors were recruited. This would have helped us to better understand why tutors decided to participate. Some qualitative and anecdotal evidence, coming from the personal interaction of the team1 with the tutors throughout the year, suggests that in the majority of the cases tutors experienced a boost in motivation when reminded they were doing something meaningful, even if, at that very moment, they were not getting any kind of gratitude in return. This was even more true for tutors who found it difficult to motivate their tutee to take part in the tutoring every week, as required by the program. I gladly admit that the interaction with tutors turned out to be a powerful source of inspiration also for me, managing this project 100% remotely. Their dedication in helping a student who was a stranger to them has been a continuous source of hope in these difficult and unprecedented times.

Moving forward:
The second edition of the program is officially at the end and many questions now arise about its future and the feasibility of a further scale-up. First of all, does an online tutoring program still have a role to play in a post-pandemic world, where lockdowns will (hopefully) be a distant memory and students will be sick and tired of following classes online, being traumatized by months of distance learning? My personal opinion is yes.

First of all, one might argue that the distribution of high-level tutors is very uneven across Italy, with most of them concentrated in large cities where there are more job opportunities. Thus, the supply of skilled tutors in peripheral areas could be particularly scarce. In this case, a Tutoring Online Program would allow also those living in more remote areas to have access to a larger pool of highly qualified tutors.

Second, even if there were a market for quality tutors available pretty much everywhere, some people might still find it difficult to get access to it because of their limited native network (i.e. immigrant families), lack of resources (cognitive and financial), or a mix of both. The Tutoring Online Program might be the best available option for students coming from families that cannot assist their children through their education, for instance because they have a limited amount of human capital accumulation or because they cannot afford (or are willing to pay for) a tutor.

Moreover, sometimes middle schools are not able to provide families and pupils with the appropriate information they need to take informed decisions about their future, namely the choice of which high school to attend, whether to attend university or not, and eventually which job might be right for them. Or, students might be victims of framing and prejudices within the school, and these might affect their performance and aspirations. A tutor also acts as a mentor, both by helping students understand what their interests and aspirations are and by filling the “information gap” on how to fulfil them.

For these reasons, as long as there are high-quality tutors available and willing to volunteer to help a student overcome the difficulties they encounter at school, and as long as it is possible to target the students who would benefit most from it, a Tutoring Online Program might be a relatively simple and inexpensive solution to help several students who struggle to keep up with school and are living in a context of educational poverty.

1The amazing team of Research Assistants that helped either with the monitoring of the intervention or with the implementation of the data collection throughout this year is composed of: Alessandro Palucci, Angelica Bozzi, Antonio Cappucci, Claudio Giambrone, Cristina Perricone, Diego Faurès, Diva Barisone, Emanuele Clerico, Francesca Colombi, Gabriele Todesca, Giulio Radaelli, Matteo Fossi, Michael Massaro, Rosangela D’Erchia, Rossella De Sabbata, Simone Maria Parazzoli.

If you have comments or questions about this blog post please leave a comment here! Alternatively, you can email me at

Tips and suggestions for data collection

Last year I received a LEAP student grant to study the development of socio-emotional skills through interactive English training classes in Quang Nam, Vietnam. The grant has allowed me to have firsthand experience in managing data collection from the field while collaborating with a local NGO, Teach for Vietnam (TFV). In this blog piece, I share some thoughts I had during this rewarding and formative experience, also in light of the COVID-19 situation.

Preparation is essential

Preparations begin way in advance of the data collection start date – they include RA recruitment, survey design, obtaining necessary authorization and clearances from the local authorities and ethics committees.

With the uncertainty of COVID-19, it is important to make sure that local policies are in line with the needs of the institutions involved in the project such as travel restrictions, or the allowed degree of in-person interactions. When possible, in order to keep track of the situation and verify the activities in the field, official documents from the local authorities about the suspension or opening of local activities (e.g. school openness) are important to have.

Ensure effective communication with field research assistants (RAs)

RAs are your arm in the field, step in their shoes and engage them to make this a rewarding experience on both sides. Some of the RAs may not think of research as their long-term career plan, it could even be their first experience ever with research. However, they could be interested in the project because of its social impact, or they are interested in learning new skills (e.g. working with data, or coding). Passionate and responsible RAs go a long way. While rolling out one of the subsequent surveys, the region was affected by bad weather. To avoid the delay in the timing of the survey, one of the RAs in my team volunteered to assist a teaching fellow to distribute and collect the surveys throughout the week when it rained heavily from morning to night (and heavy rain in Vietnam is no joke).

The RAs are often eager to see the initial results of the project. Therefore, starting to share with them the findings as soon as the data comes in can be valuable. Additionally, discussing with the RAs how the data is processed or how other works have been done in this field helps to paint a clearer picture of the project impact and helps the RAs gaining some practical knowledge or data skills.

Things change quickly

When running a data collection in the field, flexibility is key. For example, during COVID-19, local activities change on a daily basis. For this reason, local news is very important to follow the changing situation. Like many other countries, the number of COVID-19 infection cases receives a lot of attention in Vietnam, therefore, when there are new cases, different news outlets will cover them with the description of the situation and responses from the local authority. Read widely. Some news outlets may focus on some occurrences only.

Local responses may be even more local than expected. In Vietnam, the decision of postponing exams or delaying teaching activities comes from the sub-regional authority. On top of that, new cases from a specific neighbourhood can halt some activities in one school, while the others in the same region are not affected.

This is an additional reason to keep close contacts with the RAs and the NGO on the field. While the RAs can quickly report changes in school activities, the NGO can give a heads-up on how the school might respond. For instance, TFV always keeps me up to date on any local development, which gives me more time to prepare contingency plans.

Surveying in the local languages

In different parts of Vietnam, and many other countries in the world, the same object can be described with different words, depending on the local context. Navigating this linguistic ambiguity is important, especially when the respondents are young children. For instance, “bố” and “ba” both mean father in Vietnamese, however, the first is more commonly used in the north, while the latter is more used in the central part of the country. In the first trial of the survey, when first encountered the use of “bố”, some students were confused about which family member the question referred to, leading to a substantially longer time for the students to complete the survey.

When the survey includes questions relating to the effect of COVID-19 on young participants, the uses of language should be taken with extra care. Like all of us, the experience with COVID-19 among young children is unprecedented. Different from us, at a young age, children may not have a clear understanding of the circumstance. This could mean that certain scientific words are very unfamiliar to them, or certain emotions are vague for them to imagine and thus to give a response. 

If the budget and timing allow, always run a pilot-of-pilot of the survey to get a sense of how the respondents interpret the questions. Otherwise, have some local friends, or in my case, one of the RAs coming from the same region, read through the survey. Make sure to go through every question to clarify even the slightest ambiguity. Additionally, concrete examples that are close to the daily activities of the respondents can help them better grasp the meaning of the questions, and therefore, better answer them.

Embrace the learning opportunity!

Running a field data collection, especially during a pandemic, is a bit like being regularly on pins and needles. Uncertainty is unavoidable, problems come in at unexpected times, and they are different each time. Nevertheless, the experience is totally worth the cost. Every challenge requires creativity and careful understanding of the local context. Problem solving is rewarding and informative. In my case, having the opportunity to work closely with a local NGO helps me learn a lot about the educational system in Vietnam, both at the national level and at a more granular local level.

Jacqueline Nguyen

Apart but connected

By Chiara Soriolo

In this article, I want to describe my experience as a member of the research team for the Tutoring Online Project (TOP), organized by LEAP.

The Tutoring Online Project, as the name suggests, consisted in providing a tutor from 3 to 6 hours per week to 500 middle school students in Italy, chosen among those who were more in need of tutoring. Tutors were assigned to students in a randomized fashion, exploiting the fact that there were over 1000 students, but only 500 tutors available. Thanks to this experimental design it is possible to correctly identify the effect of the project on students. The tutoring took place via video-calls, during the period of lockdown that followed the coronavirus outbreak. Baseline and endline surveys were collected to understand the impact of this project.

I start saying that being part of this project has been especially meaningful to me as I have always been captured by the “social” part of economics and by the field of education in particular. Also, I think that almost six months after the major health crisis of the last fifteen years, in Italy at least, not enough attention was given to the needs of students.

After the nationwide lockdown measures, students stopped going to school and switched to distance learning. This created different types of problems, especially for elementary and middle school students. In some cases, students suffered teachers’ lack of familiarity with modern technologies and even families were challenged by this new teaching modality. In fact, given the exceptionality of the situation, ensuring that children attended their online lessons, studied and did their homework was a more demanding task, especially in the case of children who were already struggling at school.

Indeed, this project also aimed at reducing inequality in education: we know that from an educational point of view this health crisis widened the already existent gap between children coming from poor and from wealthy families (Chetty et al., 2020). On average students coming from rich families were less likely to struggle at school in the first place moreover they probably received better support during the lockdown period. The tutoring online project was designed to provide support to students and families who were most in need of this kind of help. We saw that by having a one to one interaction with university students, children were better able to remain on track, study, and learn despite the difficult situation.

My tasks as a RA were mainly related to the data collection process, so I gained the first-hand experience on what fieldwork can entail. I learnt to perform managerial tasks that I was not previously familiar with: I had to keep the contacts with the school and the parents, then I was involved in the creation of one of the surveys and towards the end, I administered a final test to students myself.

Specifically, before the beginning of the tutoring, I contacted many of the schools that showed an interest in participating and explained in further detail the project as well as answer their questions. Most of the headmasters recognized the value of the project and tried their best to comply with the tight deadlines that we gave them in order to start with the online tutoring as soon as possible.

After receiving the list of students from the school, I contacted the families in the list, explained the project, and asked whether they wanted to have the chance of being assigned a tutor to their child. Most of the parents were enthusiastic about the initiative. However, some of them could not participate since they did not have an internet connection at home, or they had neither a computer nor a tablet to use for tutoring. Given that one of the goals of this project was to help reducing inequality in education during this crisis, the fact that probably the most vulnerable families were not able to participate in is something to keep in mind.

A point worth mentioning is that some students (roughly 20%) ended up doing their tutoring using only their smartphone anyway. The main reasons were unexpected issues with the internet connection, or other family members needing the computer at the same time of the online tutoring. Despite this fact, from the analysis of the data no systematic difference could be found between the outcomes of students using their smartphone and the outcomes of all the other students involved in the project. This is very reassuring especially in view of a future scale-up of the project, since that also students owning only a smartphone could participate.

I would say that, apart from the task themselves, in this phase, good communication and coordination with the other team members was really important. Most of the time there was more than one of us on the same job. Working in a team has been something I highly valued: many times parents or headmasters had asked me questions that I was not sure how to answer to, nevertheless, I always had someone in my team solving to my doubts and explaining to me how to do things in the best way possible.

After the project finally began and students were able to meet their tutors, I was involved in the creation of one of the endline surveys. Even if I was not in charge of formulating the survey, I could see how phrasing the survey becomes crucial to obtain good quality data and how concisely conveying precise messages or questions can be trickier than it seems.

After this phase in which the job was mainly in the hands of tutors and students, I had to contact some parents again to increase the response rate of endline surveys, especially in the case of people who were not assigned a tutor. This part was probably the less pleasant one. Even if they were made aware that participation was not ensured, I felt somehow guilty to ask these parents, who were not assigned a tutor, time and attention to fill in the survey. Luckily a very high number of people took their time to complete also this last survey.

After all the effort exerted by so many people, seeing that the project arrived more than successfully at the end of the data collection process gave me a feeling of true satisfaction and made me look forward to the results of the analysis. More than everything however I am happy to have been part of a project that could help 500 students with their academic path, and that could pave the way for other valuable initiatives based on tutoring.

The Potential of Machine Learning for Development Economics Research Data

By Armin Satzger

This short article aims to provide a few pointers for fellow students and applied researchers working on topics related to development, e.g. in the realms of agriculture, environment, health or infrastructure, that are interested in deepening their knowledge of how machine learning (ML) can be useful for the construction of valuable datasets for use in economics research projects. Please note that the linked articles provide for a far deeper dive into the topics mentioned than is possible in this rather brief and therefore necessarily superficial introduction to the subject.

How can machine learning contribute to economics?

Slowly but surely, ML methods and theory are starting to be adopted in mainstream economics research in fields such diverse as financial economics (Gu et al., 2019) and real estate (Glaeser et al., 2018); human capital selection (Bajari et al.,2015; Chalfin et al., 2016); and mechanism design (Dütting et al., 2019). In the following, I would, rather than focusing on ML methods, like to describe some of the new data sources that have now become accessible to economists and, among those, focus on the ones of potential relevance to (aspiring) development economics researchers. For those wishing to learn more about the similarities and differences between traditional econometric and machine learning methods, I would suggest taking a look at Athey & Imbens (2019) and Mullainathan & Spiess (2017), articles intended to supplement graduate-level introductory courses on ML methods for an audience already familiar, to some extent, with econometric theory.

What are the kinds of new data that are becoming accessible?

In general, a major advantage of new machine learning methods is the breadth of accessibility and useability of data that comes with it. This particular concerns areas where ML solves classification problems.

Natural-language processing may serve to make large corpora of texts more accessible to researchers, by allowing to identify certain themes, categories and events from the texts. The sources of text may include things such as financial statements, administrative data, party manifestos, or text from news sites. Last year, the Association for Computational Linguistics also started hosting an annual workshop series on economics and natural language processing (ECONLP; here the link to the proceedings of the first and the second workshop on economics and natural language processing).

Computer vision, on the other hand, may help to correctly classify large numbers of photos in a variety of areas. Glaeser et al. (2018), for example, use Google Street View images and computer vision techniques to investigate the impact of the appearance of houses and their neighbourhoods on real estate prices. Satellite imagery is, however, a particularly promising instance of a new data source. Satellite data may not only be used to construct night-time light intensity proxies for economic activity as has been done before by researchers in development economics (see, e.g., Michalopoulos & Papaioannou (2013)) but also to estimate crop yields, air pollution, and land cover change, amongst other things. A more extensive review of the possible uses of satellite data in economics may be found in Donaldson & Storeygard (2016). To the benefit of economists, remotely sensed and thus already useable data on a wide variety of topics is often already freely available online, as is the case with high-quality LANDSAT satellite data, for example. A great example of the use of such remotely sensed data in a political economy / environmental economics context is Burgess et al. (2012) who study the impacts of institutional redistricting reforms in Indonesia on deforestation / forest cover change by using a raster-level (250m x 250m cells) satellite imagery dataset from MODIS sensor data.

Why does this hold any relevance for the development field?

In development in particular, many non-experimental studies still rely to a large degree on household-level or individual-level surveys, such as the Demographic & Health (DHS) series of household surveys, conducted by organisations like the World Bank in a largely standardized manner in most developing countries in relatively regular multi-year intervals. ML techniques may allow researchers to either expand the range of accessible data by constructing novel datasets themselves or to rely on prior work by other researchers such as Jean et al. (2016) who estimate poverty proxies on a very granular level using satellite imagery in combination with ML algorithms. Compared with more developed economies, developing countries thus provide a particularly attractive setting for new kinds of data as the ones described above.

Last but not least, it shall also be mentioned that these advances in the development area also provide for interesting opportunities for collaboration with researchers from other disciplines / departments, in particular computer scientists working on ML topics. De-Arteaga et al. (2018) discuss a number of research areas where they see potential for making ML techniques more useful for overcoming the challenges typically associated with developing-country data, including improving the robustness of ML algorithms to small and/or messy datasets; introducing decision support systems to battle, e.g., corruption and support health services provision; and improving transfer learning for natural-language processing for low-resource languages to reduce obstacles to information and knowledge flow.

This shall already conclude this short piece hopefully shedding some light on the new data use enabled by machine learning for development economics. My motivation for writing this piece stems from my personal interest in the subject and I would, of course, be delighted to discuss the topic further with anyone from within the Bocconi community who shares this interest.

Armin Satzger is a student in the M.Sc. Economics and Social Sciences programme at Bocconi University. The topic of the blog post was also discussed at a recent session of the regular development coffee meetings between LEAP-affiliated Bocconi faculty members and Bocconi students interested in development.