I.1. Education and School Dropout

An extensive body of literature identified both internal and external factors linking households to dropouts. On one hand, previous research identified poverty, marginalization and vulnerability that motivate an early start of child labor (^{Román, 2013}) as those internal household conditions that hamper school attendance and performance. In this regard, it is important to note that child labor is not the direct cause of school dropout, because the decision to start working is made when the student has already left school, or when labor and school activities are paired. Family structure and household organization (^{Bulanda and Manning, 2008}; ^{DeLeire and Kalil, 2002}), parent disability conditions (^{Racanello and Garduño, 2018}), in addition to youth issues (alcohol/drug use, as well as pregnancy) and cumulative stress from an early age (^{Ramsdal et al., 2013}) also induce school dropout.

On the other hand, external factors are those inherent to the educational system, and among the most important, we highlight resistance to the socializing codes provided by the education system, interaction among teachers and students, as well as the image teachers create regarding their students and the interaction that follows (^{Román, 2013:38}). Low school and teacher quality and/or motivation is another issue associated with school dropout (^{Dahal et al., 2019}; ^{Lee and Burkam, 2003}; ^{Jensen and Nielsen, 1997}).

Despite internal and external factors intervening simultaneously in the dropout decision (^{Goldschmidt and Wang, 1999}; ^{Rumberger, 1995}), household socioeconomic level is the main factor that determines school abandonment (^{ECLAC, 2008}; ²⁰⁰²). Poverty and vulnerability are closely linked to child labor both outside and inside the household (^{Levinson et al., 2001}; ^{Basu, 1999}) due to the delay in studies that such activities imply, and as long as time allocation prioritizes those that generate income (^{Gunnarsson et al., 2006}; ^{Beegle et al., 2005}; ^{Orazem and Gunnarsson, 2003}). Thus, the needs of families may cause them to choose meager short-term gains that jeopardize the future of the next generations.

I.2. Education and Quality of Life

Quality of life (QoL) is a central issue in human actions. As QoL measurement depends on the dimensions that define this construct, its operationalization is complex and manifold (^{Arenas, 1998}; ^{Nussbaum and Sen, 1996}). A QoL measure is fundamentally a cognitive evaluation expressed as a judgment that includes satisfaction, importance, increased or diminished expectations about the domains of life, as well as their meaning, identified by the person. It also denotes an affective assessment, expressed as a feeling of happiness or sadness as a result of the [person’s] interaction in both material and social environments, including available alternatives for actions, freedom and access to goods and development (^{Garduño, 2002: 232}). QoL also encompasses happiness experienced in several domains, such as family, friends, money available and health status (^{Palomar, 1998}). Although QoL’s subjective and objective dimensions are complementary, the former approach is often preferred, because the latter might misinterpret the link with satisfaction, purpose of life and happiness by focusing on goods people have access to. In this regard, earlier research found that wealth might even undermine happiness (^{Scitovsky, 1976}). Later, in a study about the most important domains of life and their relationship with perceived happiness in a sample of upper-middle-level students in a poor rural community in the state of Puebla (Mexico), ^{Garduño and Raccanello (2009)} found that life domains involving family, friends, inner life and home comforts were significant, while money had a lesser role. Even though research on the connection between income and subjective wellbeing/happiness has mixed results (^{Rukumnuaykit, 2016}; ^{Schneider, 2016}), it is evident that, directly or indirectly, income pervades several domains of life.

II.1. Education Programs in Mexico City and Sustainable Development Goals

Based on the Sustainable Development Goals (SDGS) and the 2030 Agenda promulgated by the United Nations in 2015, the commitment to “ensure inclusive and equitable quality education and promote lifelong learning opportunities for all,” as stated in SDG-4 (Quality Education), was endorsed by the education and social development policies adopted during the 2012-2018 Mexico City Government administration. In this regard, several public programs and actions designed to support education were set up and began operating, based on a transverse perspective, especially for the most disadvantaged and vulnerable population, providing scholarships for several school grades, transportation, food security, school supplies and insurance, among others (^{Evalúa-CDMX, 2018: 111}) with mixed evidence as to impact significance (^{Dustan, 2020}). For the purpose of this paper, the comparison of SDG-4’s indicators between 2012 and 2016 in Mexico City provides evidence of a major step forward toward accomplishing SDG-4, Target 1, as enrollment and completion indicators improved sharply, especially for secondary education⁴ (^{Inegi, 2020}). From this perspective, the Edugar Program design clearly contributes to completion at primary, secondary and high school level by preventing dropout at least in the short run (^{Raccanello and Palacios, 2017: 408}). Even further, as Edugar beneficiaries plan to pursue undergraduate studies in the future (^{Ebrard, Delgado and Carrillo, 2010: 239}), the Program could also contribute to a promising mindset change that might improve beneficiaries’ future.

III.1. Hypothesis

We posit that neither death or total and permanent impairment affects the domain of life perceptions (for children) and satisfaction (for parents/tutors), nor does Program support enable QoL dimension recovery.

Of course, as the event affects households due to the economic downturn, among other effects (^{Ebrard, Delgado and Carrillo, 2010}), we expect a worsening in the domains of life perceptions (for children) and satisfaction (for parents/tutors) immediately after the accident, which could recover (totally, partially or not at all) after joining the Program. Such prediction holds for children and parents/tutors who recently joined the Program as well as for the most experienced ones.

III.2. Sampling

In 2016, the Program supported 9,234 beneficiaries, most of whom attended primary or secondary school, and females comprised slightly more than half of the population served (50.18 percent).

A random sample of beneficiaries (N=619) stratified according to municipality, gender, grade in school⁵ and seniority in the Program, was selected. A threshold of two years of permanence was defined to distinguish between experienced beneficiaries (two years or more in the Program - identified as “old” hereafter) and new entries (less than 2 years - identified as “new” hereafter).

To guarantee beneficiary anonymity and safeguard personal data, the Special Programs Sub-Directorate of the Mexico City System for Comprehensive Family Development - DIF-CDMX (Sistema para el Desarrollo Integral de la Familia de la Ciudad de México) telephoned the parents or tutors of children entitled to the Program, to invite them to participate in the survey, stating that at least one girl/boy per family had to be present, too. To avoid insufficient data due to unresponsiveness, 50% oversampling was requested.

To administer the questionnaires, beneficiaries were summoned to the courtyard of the DIF-CDMX Center of the Venustiano Carranza Municipality on December 10 and 11, 2016, from 09:00 am to 03:00 pm. At the gate, they were identified by verifying their name and Program/id; then, questionnaires were administered simultaneously to adults and children, ensuring that the two groups could not communicate with each other. Previously-trained DIF-CDMX personnel was available to help children/parents/tutors with the questionnaire and to assure that forms were filled out properly.

A greater number of beneficiaries participated than expected, which meant 1,147 children’s questionnaires (863 “old”, 284 “new”), as well as 806 adult questionnaires (599 “old”, 207 “new”), were collected. All beneficiaries who responded to the appointment were surveyed. DIF-CDMX personnel entered the information in an excel spreadsheet and then delivered the questionnaires for any corrections needed in the database.

III.2.1. Children Questionnaire

To determine the QoL perception of the students benefited by the Program, we applied the valid and reliable Kidscreen-27 (Screening for and Promotion of Health Related Quality of Life in Children and Adolescents - 27 item version) questionnaire (^{Vélez et al., 2016}). The domains of life of the Kidscreen-27 are: Physical Wellbeing, Psychological Wellbeing, Autonomy and Parental Relationship, Friends and Social Support, and School Environment. Because of the age of some of the students surveyed, some minor adjustments were made to the instrument text, and two items (#18 and 29) were added, resulting in 29 items in the final version of the questionnaire. We used a Likert five-point scale of agreement for all items: (1) Strongly agree, (2) Agree, (3) Indifferent, (4) Disagree and (5) Strongly disagree.

III.2.2. Adult Questionnaire

Adults were administered a 26-closed-ended-item questionnaire to assess the socioeconomic characteristics of the household (members, sex, age, educational level, and economic dependents), housing services, work activities, expenditures, school dropout, Program Comprehensive Care Model usage and family relationships. We also elicited the level of satisfaction related to inner life, couple relationship and family integration through a 21 -item battery. A Likert five -point scale of agreement was used for all items: (1) Very satisfied, (2) Satisfied, (3) Indifferent, (4) Dissatisfied and (5) Very dissatisfied.

III.3. Testing

To test our hypotheses, we relied on a t-test for group (“new”, “old”) mean comparison on factors obtained through averaging beneficiary responses corresponding to each factor. Stata® software version 15.0 was used in all data analysis.

We carried out the test comparing the mean factor perception (for children) and mean factor satisfaction (for adults/tutors):

IV.1. Children Questionnaire Item Analysis

When comparing “old” (N = 863) and “new” (N = 284) beneficiary average values for each of the 29 items (Table 1), children’s perception on all items, and therefore for all domains of life, worsen immediately after the event (averages increase), supporting the expected behavior. Since items #8, 9, 10 and 18 characterize discomfort, as averages fall we detect greater awkwardness. However, once entitled to Program benefits, perception improves for all items. It is worth noting that we observe the same pattern regardless of beneficiary seniority in the Program. Nonetheless, the (average) perception worsening for “new” beneficiaries - along all items - is greater than the “old” ones immediately after the event (“new”: 0.4931; “old”: 0.3985). Then, once families receive the benefits, the average improvement is also slightly greater for “new” beneficiaries (-0.5438) versus “old” ones (-0.5212). This is probably because the shorter the time since the event, the greater the perception of difficulty. That is to say, as “new” beneficiaries may not have internalized the recent unfortunate event yet, they show higher variance in their perceptions⁶. In this case, the old adage “time is the best cure” fits.

When comparing children’s perception before the event versus that reported after joining the Program, we observe such a positive effect of the support that the latter overwhelms the former one. We note that on average, the change in perception is greater for “old” (-0.1227) than “new” beneficiaries (-0.1047).

It should be noted that Program support by itself could not improve some of the aspects elicited by the adapted Kidscreen questionnaire. A detailed look at the data highlights that before the event children agree on almost all items but those referring to the economic situation. In fact, economic vulnerability comes up when referring to monetary matters (items #19 and 20), as they reflect the lowest agreement level compared to all other items. Although immediately after the event the household economy might worsen even further, it partially recovers once beneficiaries are entitled to Program support. Thus, children seem to be aware of it.

In line with the common expectation, children agree that parents/tutors love them (item #17), but immediately after the event, probably because of the inevitable disequilibrium inside the household, they perceive higher parental detachment that recoups once they join the Program. Such behavior is similar for items regarding the time parents/tutors devote to children, as well as children-parent communication, but works inversely for scolding. Thus, children’s environment seems to recover after accessing Edugar. Children’s perception about their health, mood, spare time, happiness and especially school, due to Edugar, also improves. Above all, beneficiary perceptions about the future take a noticeable upturn, which is worth mentioning when considering their capacity to cope and overcome difficulties, despite the young age.

IV.2. Children Questionnaire Factor Analysis

As previously mentioned, we identify five domains of life throughout the adapted Kidscreen questionnaire: Physical Wellbeing (items #1 to 4), Psychological Wellbeing (items #5 to 11), Autonomy and Parental Relationship (items #12 to 16), Friends and Social Support (items #17 to 24), and School Environment (items #25 to 29). For each domain of life, we obtain the corresponding factor by averaging the answers to all items within the domain.

Before the event, children perceive that School Environment is good, as well as agree that Autonomy and Parental Relationship and Physical Wellbeing are at a positive level. However, the Friends and Social Support factor scores lower on the agreement scale, probably because of the vulnerable environment surrounding the household. The day-to-day struggle and the scarce opportunities available might hamper the youth from accessing a supportive circle of friends, and they could also interfere with some related Psychological Wellbeing items; according to Table 1, children agree less with their good mood and happiness but tend to agree on having fun. The former two features (items #6 and 11) would point to some underlying problems faced by children that might be coped with by engaging in enjoyable activities (Item #7). As shown below, this factor partially improves once beneficiaries receive Program support because the monetary transfer can lessen internal (economic) household pressures and/or provide access to entertainment activities, but it might not have a priori effects on the external variables, like the circle of friends they have access to.

Mean factor comparisons (Table 2) reveal a worsening perception for all factors immediately following the event, but recovery is observed after receiving Program support. It is relevant to observe that in line with previous discussion, besides worsening, perceptions also have higher standard errors, suggesting that there might be different ways (and results) to cope with the loss. It is worth mentioning that perceptions worsen for both “new” and “old” beneficiaries in all domains of life, but the increment is higher for Psychological Wellbeing (“old”: 0.7404; “new”: 0.6968), Autonomy and Parental Relationship (“old”: 0.4607; “new”: 0.5062), and Friends and Social Support (“old”: 0.4777; “new”: 0.4744). Additionally, we acknowledge that mean factor differences are all significant (Table 3) between pre-event and immediately after the event; this is, all children’s domains of life statistically worsen. However, we observe an improvement after receiving support for these same domains. Interestingly, the Program aims at preventing dropout, and the School Environment domain has an above-mean improvement once beneficiaries receive support. Table 3 reports that all mean differences are significant once beneficiaries join the Program.

As all life dimensions improve after receiving Program support, regardless of seniority, it is tempting to state that support allows for perception recovery for children and young people. In particular, the domains with variations above the mean are Physical Wellbeing, Autonomy and Parental Relationship, and School Environment, suggesting that they are the ones most harmed by the event but also mended by Program intervention.

The last row of Table 3 compares mean factor differences before the event vs. after receiving support; according to the t-test. Except for Psychological Wellbeing (“old” and “new”) and Friends and Social Support (“new” only) we detect a significant difference in all other life dimensions. In other words, once they join the Program, children and young people recover their pre-event Psychological Wellbeing perceptions, but only the “new” group does so for Friends and Social Support. As we wondered whether children might even have a better perception after receiving support than the one they had before the event, a one-tailed t-test provides statistical evidence that perception improves for Physical Wellbeing, School Environment (“old” and “new”), as well as Friends and Social Support (“old” only). Thus, the Program does not lead to an improvement in Psychological Wellbeing but helps recover pre-event perceptions, a positive result after all.

IV.3. Parent and Tutor Item Analysis

Item average comparisons before and immediately after the event by “old” (N = 599) and “new” (N = 207) parents/tutors (Table 4), disclose worsening in their satisfaction for each of the 21 items of the questionnaire. On average, the “new” group reports greater worsening than the “old” one (“new”: 0.6007; “old”: 0.4912). As in the previous section, the same interpretation holds: the shorter the time span since the event, the more vivid the perception and therefore higher means and volatility responses. Upon receiving Program support, adults with lower seniority, on average, show heightened perceived satisfaction (“new”: -0.5189; “old”: -0.4560), a result aligned with the above explanation. In fact, people who get through the difficult time, having received the government support, might be more prone to manifest over-improvement in their satisfaction because the Program could ease unsolved (real or perceived) worries. Finally, when comparing satisfaction level after receiving the support with that prior to the event, and although results vary according to item, on average, reduced satisfaction is detected, especially for “new” beneficiaries (“new”: 0.0818; “old”: 0.0353).

Before the event, parents/tutors were satisfied with the support and atmosphere resulting in living with the partner; however, economic vulnerability harms their satisfaction regarding the stability and comfort achieved through the money earned (items #8, 9 and 10).

Despite the capillary local transportation network, in Mexico City people may spend up to two hours reaching the workplace during peak hours because of traffic and public transport crowding; this is why workers living in the suburbs of the capital may allocate up to 12 hours a day for a work shift. Thus, besides economic needs, the stress and rushing they experience undermine time available for themselves (item #13), so they agree less on time left for personal activities like reflecting and relaxing or for vacations (items #14, 15 and 20). Satisfaction level drops immediately after the event for all items but recover once receiving Program benefits except for those referring to the love within the family as well as love, understanding and support from the partner and assessment of inner life and its stability (items #4, 5, 6, 7, 16 and 17). Because of the severity of the event, the Program can improve the adult and family characteristics linked to economic needs. (^{Ebrard, Delgado and Carrillo, 2010: 202}), except those referring to the intimate relationship with the partner hopelessly affected by the event (items #5, 6 and 7).

IV.4. Parent and Tutor Factor Analysis

We focus on three domains of adult life: Family Integration (items #1 to 4); Couple Relationship (items #5 to 7), and Inner Life (items #14 to 17). For each domain of life, we obtain the corresponding factors through the same methodology as above.

Adult factor analysis (Table 5) reveals that prior to the event Family Integration and Couple Relationship were “satisfactory”, while respondents’ perception of their Inner Life was “indifferent”. As for children, the lower the adult’s seniority in the program the higher the factor’s volatility. As well, regardless of seniority, adult/tutor satisfaction worsens after the event, but recovers once adults access the Program and its benefits (Table 6), except for the Couple Relationship factor (“new” only). Thus, although Family Integration and Inner Life mean values return to pre-event levels, Couple Relationship does not, and its mean and standard error increase significantly (Table 5). In fact, items aggregated into this factor refer to specific facets of the relationship that irreversibly changed in the aftermath of the event that involved the partner. Hence, obviously rather than compensate for, the Program can only mitigate such lost.

IV.5. Discussion

According to the analysis of children’s domains of life perceptions before the event, immediately after and once entitled to Program benefits, we found that the support helps make up for perception worsening in the aftermath of the event. However, “new” beneficiaries’ perception has a higher volatility level than the “old” one at both item and factor levels. This is probably due to the shorter time span from the event increasing the perception of being in trouble.

Overall, children’s perceptions improve mostly for Physical Wellbeing, Autonomy and Parental Relationship and School Environment dimensions, but barely for Psychological Wellbeing and Friends and Social Support. Although the Comprehensive Care Model is available to beneficiaries, they seldom rely on it, apparently unaware of the services provided,⁷ except for the entertainment and cultural activities, as 76 percent of parents/tutors are aware of them. According to survey data, less than 5 percent of “old” or “new” beneficiaries⁸ rely on such services, except for entertainment and cultural activities, which is between 42.5 percent (“new”) and 52.6 percent (“old”). Because ^{Ebrard, Delgado and Carrillo (2010: 238)} found higher attendance for psychological support by adults and beneficiaries through the Program, launching a continuous informative campaign about the availability of such services would be a paramount first step toward improving the Psychological Wellbeing dimension.

Parent and tutor satisfaction with Family Integration and Inner Life domains recovers after receiving Program support, whereas Couple Relationship does not, probably due to the nature of the event that disrupts the intimate ties (love, understanding and support) provided by the partner.

We acknowledge that the improvement in life domains for children and parents/tutors points toward suitable household environment reconstruction, which could be a positive basis for attending school.

Even though the number of beneficiaries assisted by the Program has been increasing (from 625 in 2007 to 9,425 in 2017,⁹ and 9,368¹⁰ in 2019), Edugar provides the same cash transfer (832mxn≈44usd) today as in 2007. Thus, transfer impact on day-to-day expense coverage (i.e. food, transportation) has decreased, in real terms. While the decision to increase the transfer amount depends on approved budget, and such analysis is beyond the scope of this paper, between 2007 and 2018¹¹, reduction in transfer purchasing power reached 35.86 percent. Despite the Program’s positive effects on beneficiaries, this should not be underestimated. However, as both children and parents/tutors behave quite similarly, and the time span from event occurrence is inversely related to the variance and means of their responses, we wonder whether beneficiaries are aware of the real (decreasing) transfer support or, as time goes by, life continues and people internalize the event. Accordingly, the question arises: do “old” beneficiaries have a lower variance/mean, versus “new” ones, once receiving the support, because the cash transfer (in real terms) is lower or because their perception/satisfaction fades over time? As children are not aware of economic issues, fading perception over time seems to be the reason for higher (mean/variance) responses. If this interpretation holds, then results fit within the hedonic adaptation framework analysis: as people get used to an event that elicits emotional responses, the wellbeing measure tends to return to the base-line (pre-event) level (^{Frederick and Lowenstein, 1999}). Although we do not have data to test this hypothesis, this might be the reason why the “old” group shows a lesser mean/variance response, a result conceded by ^{Armenta et al. (2014: 71)}, when stating, “Although individuals show some adaptation to negative life changes, they often [emphasis added] never completely return to their initial level of happiness following traumatic events.” Therefore, although we cannot deny that the Program might still play a complementary role in repairing the damage suffered by households, the local government could achieve the goal without increasing the Program budget due to beneficiary adaptation(!) Besides standard indicators, an impact evaluation designed to disentangle such a situation should consider the role of emotional adaptation, an aspect seldom included in such analyses.

Though the Program is widely justified for helping cope with household economic problems, one of the major factors linked to school dropout, preventing abandonment is only the first step in developing skills that lead to reaching a higher QoL. In fact, poor education quality may deny access to necessary academic competencies, jeopardizing children’s QoL in the long run. From a broader perspective, QoL is the result of the interrelationship of endogenous and exogenous factors, as well as the decisions made by individuals (and their families) who live in an environment where public policy can reduce obstacles. That is why policymakers need to have a comprehensive view of the problems and inner consequences they are intended to solve through public budget and targeted expenditures. Recently, the results of the OECD’s Programme for International Student Assessment 2018 (^{OECD, 2019}) were published, and they highlight that “Students in Mexico scored lower than the OECD average in reading, mathematics and science. In Mexico, only 1 percent of students performed at the highest levels of proficiency (Level 5 or 6) in at least one subject (OECD average: 16 percent), and 35 percent of students did not achieved (sic) a minimum level of proficiency (Level 2) in all three subjects (OECD average: 13 percent)” (^{OECD, 2019: 1}), despite a minor improvement in science through the last 12 years (^{OECD, 2019: 4}). This underscores our previous statements: although guaranteeing school attendance is the first piece of the puzzle, greater efforts are needed to achieve skill consolidation by students.