SMS GIRL DATA INSIGHTS: 2 As schools reopened after 18 months, are adolescent girls and boys back in school and engaged with learning? Evidence from four surveys in Punjab, Pakistan. February 2022 Koen Geven, Sheena Fazili, Ayesha Tahir, and Tazeen Fasih SMS GIRL: DATA INSIGHTS | 2 KEY TAKE-AWAYS 1.  Punjab is losing adolescent children from the school system This brief summarizes the main findings from four consecutive at faster rates than initially expected, and adolescent boys are dropping out at higher rates than adolescent girls. Post school reopenings in September 2020–October 2021, boys in our sample 8% OF GIRLS ARE telephone surveys of families in re-enroll at low rates for all groups. There is more variation for DROPPING OUT Punjab, Pakistan. The surveys girls by social background: girls from poor households, rural areas, were conducted as part of a study and those enrolled in public schools and schools supported by the Punjab Education Foundation (PEF) seem to be re-enrolling at 21% OF BOYS ARE assessing the effectiveness of sending lower rates than girls from richer households, urban areas, or those DROPPING OUT attending private schools. phone-based text messages to parents to ensure that adolescent girls The most common cited reasons for not re-enrolling include 2.  (typically aged 10–14 years) continue financial difficulties, pursuing religious education, and parents’ views that their child would be unable to catch up with to learn during school closures and learning.1 Health issues of the child or a family member also FINANCIAL RELIGIOUS re-enroll in school once schools reopen seem to be impacting girls’ re-enrollment in school. While some after the COVID-19 pandemic. of these reasons may not necessarily be directly related to the pandemic, dropouts are much higher than in a regular year. TOO BEHIND HEALTH REMOTE  ngagement with remote learning (offered online, on television, 3. E LEARNING or on radio) was low,2 and has decreased substantially as schools IS NOT BEING reopened. Fewer girls and boys are taking advantage of remote UTILIZED This is the second brief in a series of research learning than at the start of the pandemic; boys’ engagement is briefs, which aims to draw a picture of how the lower than that of girls. That said, remote learning still has an pandemic effects are evolving in the school audience and may fill demand for additional learning support. system over time. The data used in this brief describes the experiences of 5,898 families in TUTORING AS Families are hiring tutors as an alternative to in-person learning, 4.  AN ALTERNATIVE Punjab using four rounds of data collected in but tutors are hired much more regularly by richer families. This TO IN-PERSON (1) August–October 2020, (2) November 2020– creates inequalities in access to additional learning support. LEARNING January 2021, (3) February–March 2021, and (4) September–October 2021. The data reveal GIRLS ARE the following take-aways. Girls still spend considerably more time on household chores than 5.  DOING MORE boys. Over time, boys and girls both reduced time spent on family WORK AROUND care as schools reopened. Since school reopenings in September THE HOME 1 The respondents were given a list of potential reasons for not re- enrolling children to school which were both pandemic related as 2021, girls spent about twice as much time as boys on family well as general, including the option to state any other reasons. care (46 minutes for girls, 22 minutes for boys, daily). 2 Hasan, A., Geven, K., & Tahir, A. (2021). SMS Girl Data Insights. SMS GIRL: DATA INSIGHTS | 3 THE STUDY SETTING Schools in Punjab were closed initially on was collected during second period of school The survey sample includes 5,898 families, March 15, 2020, to prevent further spread of closures (November 2020–January 2021), from which we track either a girl or a boy the COVID-19 virus. Since then, schools have third round of data (Midline 2) was collected in aged 10–14 and those who reported being alternated between opening and closing, with (February–March 2021), approximately a month enrolled in grades 5–7 before the pandemic. the total period of school closures lasting after schools reopened again in late January Out of the total sample, 5,352 families have approximately 10.5 months. We have now 2021, and fourth round of data (Endline) was an indexed girl, and 536 families had an tracked the impact of school closures on a collected in September–October 2021, right indexed boy.3 The survey respondents are sample of families and children since August after schools reopened again in mid-September from 6 districts in Punjab.4 The students 2020. The first round of data (Baseline) 2021 (figure 1). Data on learning levels about whom we present information below was collected during August 2020–October will be collected in the next round of data are predominantly enrolled in public, private, 2020, the second round of data (Midline 1) collection (planned for November 2021). and Punjab Education Foundation schools. Figure 1: A Timeline of School Closures and the SMS GIRL Evaluation SMS GIRL Evaluation Milestones SMS GIRL intervention TV for starts remote learning starts 26 02 Baseline Midline R1 Midline R2 Endline Qualitative R1 Qualitative R2 Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 2021 2021 2021 2021 2021 2021 2021 2021 2021 2021 2021 2021 Summer break 15 01 09 15 26 18 01 15 26 07 01 11 16 Total School School high primary School School School School lockdown reopenings closures schools & middle closures reopenings reopenings reopenings schools (select districts) School Countrywide School School School closures lockdown reopenings closures closures ends (additional districts) School Openings School Closures 3 The sample of girls is much larger to create enough statistical power to estimate the effects of the impact evaluation (not included in this brief). 4 Attock, Faisalabad, Lahore, Muzaffargarh, Rahim Yar Khan, and Sargodha. SMS GIRL: DATA INSIGHTS | 4 THE STUDY SETTING The sample comprises girls and boys who and that families could be tempted to make the pandemic primarily as an income and a were previously sampled for the SABER these new roles permanent by keeping their schooling shock. The first brief also showed Service Delivery Survey conducted in girls at home. We were also concerned about that a substantial share of students was 2018 and was a representative sample of unequal intra-household expenditures, and at risk of dropping out, particularly children Grade 4 students in the Punjab province. that in a situation where families were forced in their adolescence. For instance, parents The sample was complemented with an to decide between their sons and daughters, reported that about 1 in 10 children aged additional sample from a census frame they would invest more in their sons’ 14–17 years were at risk of dropping out, of public schools in two districts in the educations. versus about 1 in 40 children aged 10–12 south of Punjab.5 In each survey round, In our first brief,7 we documented how years. Now, we present the experience of the questionnaire was completed by families were coping with the closures. It these families with subsequent school parents of the children in our sample. became clear that most families experienced reopenings and closures. The analysis uses baseline weights to account for survey design and attrition over various survey rounds. Round 2 (November 2020–January 2021) was Figure 2: Study Sample Population answered by 3,317 households, Round 3 (February–March 2021) was answered Round 1 by 4,482 households, and Round 4 was (August–September 2020) 5,898 households answered by 3,857 households. Our research questions stemmed out of a concern for a potential decline in girls’ Round 2 enrollment due to the COVID-19 pandemic. (November 2020–January 2021) 3,317 households Earlier studies had pointed out that girls faced many risks during pandemics, for instance during the Ebola outbreak in Sierra Leone in 2016.6 We hypothesized that Round 3 4,482 households girls were more likely to take up household (February–March 2021) Grades 5–7 tasks (such as caring for younger siblings), 5 Muzzafarghar and Rahim Yar Khan. Round 4 3,857 households 6 Bandiera et al, 2020. Do School Closures During an Epidemic (September–October 2021) have Persistent Effects? Evidence from Sierra Leone in the Time of Ebola. Working Paper. https://www.homepages.ucl. ac.uk/~uctpimr/research/ELA_SL.pdf 7 Available at: https://openknowledge.worldbank.org/ handle/10986/35477 SMS GIRL: DATA INSIGHTS | 5 5 KEY FINDINGS How many adolescent Figure 3: Enrollment by Time Periods 1 children are we losing from 100 49 the school system? 100 100 94 92 92 Girls As schools progressively reopened from Boys 90 September 2020 onwards, parents are pulling 80 a large number of adolescent students out 81 79 of school. By the first school reopenings Percentage re-enrolled (15 September–November 2020), 6 percent of girls and 10 percent of boys who were enrolled 60 in school prior to the pandemic were no longer enrolled (figure 3). Following the rise in COVID-19 cases, schools were closed again and by the time they reopened again (February–March 2021), 40 8 percent of girls and 19 percent of boys were no longer in school (figure 3). By the third school reopenings (September–October 2021), dropout for girls remained the same, while dropout for 20 boys increased to 21 percent (figure 3). Of those who did return to school, boys spend an average of 4.3 days in school in a 6-day week compared to 3.9 days attended by girls. 0 R1: Mar R2: Sep–Nov R3: Feb–Mar R4: Sep–Oct Substantially more adolescent boys than girls 2020 2020 2021 2021 are not going back to school. This finding is Time period confirmed in other surveys, such as a random- number dialing survey conducted at national level.8 At the start of our surveys, we expected Notes: that adolescent girls would drop out in higher a) Data for March 20 are based on parents’ reporting whether their child was enrolled in school before the first school closure in March 20, collected in the first round of data collection (August–October 8 This recent Gallup random-number dialing survey conducted 2020). September–November 2020 data denote actual enrollment figures while schools were open between December 2020 and June 2021 found that by December from 15 September–November 2020. February–March 2021 data denote actual enrollment figures in ‘20–January ‘21, re-enrollments were 94% for children aged 8-15 in the months of February–March 2021. September–October 2021 data denote actual enrollment figures Punjab and by June 2021, that number was 90%. The same survey in the months of September–October 2021. also confirmed that drop-outs for boys are somewhat higher than for b) The figure shows weighted estimates. girls. The ASER 2021 survey, conducted in four districts in Punjab with a much younger study population found a lower dropout rate. SMS GIRL: DATA INSIGHTS | 6 numbers than boys, given that there are pressures on girls to take on household roles. Figure 4: Expected versus Actual Dropout While the data shows that adolescent girls are in fact dropping out in large numbers, adolescent boys are leaving the school system in even larger numbers than adolescent girls. Expected dropout rate before schools reopened Over time, adolescent children are dropping out 25 (Aug–14 Sept 2020) at much higher levels as compared to parental Dropout rate during first reopening expectations early into the pandemic (figure 4). (15 Sept–Nov 2020) 21 Parents in our first survey wave expected a 3 Dropout rate during second reopening percentage point dropout among boys and a 20 19 (Feb–Mar 2021) 1 percentage point dropout among girls (see Dropout rate during third reopening figure 4). The share of students who ended up (Sept–Oct 2021) not returning to school during the first period of school reopenings (15 September–November 15 Percentage 2020) is substantially higher, with 6 percent boys and 4 percent girls dropping out. The gaps between expectations and reality grew worse over time as there is a drop in enrollment of 10 19 percent for boys and 8 percent for girls by 8 8 the time schools reopened again in February– March 2021 (figure 4). By the third school 6 reopenings (September–October 2021), the gap 5 4 between girls and boys remained the same, 3 while the gap between expected and actual re-enrollment grew as 21 percent of boys in 1 our sample dropped out of school (figure 4). 0 We can triangulate this finding with household Boys Girls survey data in a regular school year. Household Gender survey data tells us that the dropout rates for adolescents are 2 percentage points (p.p.) on average from grade to grade in the period grade 6–8 (1 p.p. are lost in the transition between grades 6 and 7, 1 p.p. between grades 7 and 8, Note: The figure shows weighted estimates and 4 p.p. by the end of grade 8, respectively) for both girls and boys (MICS, 2017–18). SMS GIRL: DATA INSIGHTS | 7 Dropout rates seem to be more severe for girls Figure 5: Changes in Enrollment 18 months into the Pandemic by School Type, in rural areas, and those enrolled in public and Location, Region, and Wealth Status PEF schools than girls from other backgrounds; nonetheless, enrollment trends are downward sloping across groups. By the time schools By Location By Wealth reopened for the third time (September– 100 100 Rural Poorest October 2021), 10 percent of girls in rural 83 quintile 55 Boys Boys areas were no longer enrolled in schools as 100 100 Urban Richest compared to 5 percent of girls in the urban 68 quintile 91 areas (figure 5). Looking at differences by 100 100 Rural Poorest school type, there was a 23 percent drop in 90 quintile 88 Girls Girls enrollment for boys and 9 percent for girls in 100 100 Richest public schools by the third period of school Urban 95 quintile 96 reopenings (September–October 2021). Similar 0 20 40 60 80 100 0 20 40 60 80 100 trends are observed for both boys and girls Percentage Percentage in PEF schools. This is in sharp contrast with the situation in private schools, where there By School Type is a zero drop in enrollment for girls and a Public 100 77 17 percent drop in enrollment for boys by the 100 third period of school reopenings (September– Boys Private 83 October 2021) (figure 5). PEF 100 76 Public 100 91 Girls Private 100 100 PEF 100 89 0 20 40 60 80 100 Percentage March 2020 October 2021 Notes: a) In the wealth disaggregation figure, rich represent the top quintile in the wealth distribution whereas poor represent the bottom quintile. b) The figures show weighted estimates. SMS GIRL: DATA INSIGHTS | 8 Why are families not Figure 6: Reasons for Not Re-enrolling Children in School 2 re-enrolling their children 30 in school? Financial difficulties 23 19 Religious Learning (Hifz) 15 Families cite several reasons for not re- 10 enrolling their children, including financial Child cannot catch up with learn 5 difficulties, religious learning,9 and parents’ 7 Lockdown 5 concern that children would not be able to Child/family member 7 catch up with learning. The latter point is facing health issues 12 particularly pronounced as a difference between 6 Work 0 boys and girls, with families much less likely to believe that boys can catch up (possibly 6 Child lacks interest in school 7 due to their involvement in income generating 3 Reasons Don't know activities). The most common reason parents 2 cite for not re-enrolling their children in schools Other 3 5 is financial difficulties (figure 6). Health issues 3 of the child or a family member seems to have Low quality of education 2 a greater impact on girls not returning to school Admission issues 3 6 (12 percent) relative to boys (7 percent). The 1 absence of the right level of schools in proximity Marriage 5 Boys and marriage are also some of the reasons for 1 B-Form unavailable 0 parents not re-enrolling their children in school, 1 Girls particularly for girls.10 Distance to school 8 0 Parents unwillingness 2 0 COVID-related health concerns 2 0 5 10 15 20 25 30 Percentage Notes: 9 Religious learning might be seen as an alternative to formal a) 413 families (9.2 % of the sample) reported not having re-enrolled their children back to school in schooling. March 2021, whereas 481 families (12.47 of the sample) reported the same in October 2021. 10 The respondents were given a list of potential reasons for not b) Admission issues include unavailability of B-form, of schools in proximity and schools not accepting re-enrolling children to school which were both pandemic related new admissions. as well as general, including the option to state any other reasons. SMS GIRL: DATA INSIGHTS | 9 How much time do children Figure 7: Share of Children Engaged in Remote Learning during School 3 spend learning, including Day by Gender remote learning? Self-study at home increased for both girls and 25 Round 1 boys during the first school reopenings (15 (14 Aug–Sept 2020) September–October 2020) and second school Round 2 closures (November 2020–January 2021), with (Nov 2020–Jan 2021) girls spending over half an hour more than boys. 20 Round 3 By the second period of school closures both (Feb–Mar 2021) girls (139 minutes) and boys (95 minutes) were spending more time on average on self-study. Round 4 (Sept–Oct 2021) This is higher from the average time spent by 15 14 14 Percentage both girls (102 minutes) and boys (66 minutes) during first period of school closures (August– October 2020). However, by the time schools reopened in September–October 2021, both 10 10 girls (107 minutes) and boys (76 minutes) 8 8 were spending less time on self-study at home 7 (figure 8a). Student engagement with remote learning 5 4 has gone down substantially, even when most 2 children have access to either TV-based or online remote learning programs in principle.11 During the first school closures (August–14 0 Boys Girls September 2020), 14 percent of girls’ and boys’ parents report that their children engaged Gender/Rounds with a remote learning activity at home (figure 7). For the second school closures 11 Remote learning includes learning time spent on national and local TV stations, YouTube, and mobile learning apps. 66.7% of girls’ households and 59.3% of boys’ households in our sample Note: The figures show weighted estimates. have access to a television, and 34.9% of girls’ and 27.8% boys’ households have access to the internet. SMS GIRL: DATA INSIGHTS | 10 (November 2020–January 2021), engagement with remote learning decreased significantly, with only 7 percent of girls and 10 percent of boys participating in any remote learning at home. By the time schools reopened in September–October 2021, 8 percent of girls and 2 percent of boys were participating in some remote learning at home (figure 7). For both boys and girls, coming from a poor household means spending less time on remote learning relative to a child coming from a rich household. During the first lockdown, girls from rich households spent on average 15 minutes on remote learning in a day, whereas those from poorer households spent 10 minutes on the same. A similar disparity lies in the average time a female student from a public school (10 minutes) and PEF school (11 minutes) spends on remote learning relative to one who attends private school (16 minutes). Girls in urban areas also spend slightly more time on remote learning (19 minutes) relative to girls in rural areas (8 minutes). SMS GIRL: DATA INSIGHTS | 11 Are private tutors filling the Figure 8a: Typical Student’s Time Use in a Day by Gender 4 learning gaps left by the Remote Learning Self Study Tutor pandemic and for whom? 150 150 150 139 Time spent studying with a private tutor is high, especially during periods of exams. By the second 112 school reopenings (February–March 2021) girls 102 107 Minutes (Mean) spent a daily average of 47 minutes studying 100 100 99 100 with a tutor, whereas, boys spent 56 minutes. 95 Girls’ time spent with a tutor decreased during 68 the initial period that schools were open but 75 76 66 54 56 equalized again in February–March 2021. 50 50 50 52 42 However, by the third school reopenings, the time 47 40 spent with tutor in a day decreased (girls spent: 36 42 minutes and boys: 40 minutes). The most 12 12 28 9 6 5 time spent by girls when schools were open was 12 0 0 0 6 4 near the end of the academic year, potentially in 2 anticipation of end-of-the-year exams (figure 8a). 0 21 21 21 21 20 –O 020 20 21 0 20 1 21 21 21 02 02 02 20 ep r 20 20 eb n 20 20 20 20 20 20 20 20 p2 Se ep 2 t2 n2 Engagement with remote learning and private 20 Oct ct ar an ct ct p ct ar a Se Oc a Se –Ja –O –O –M –M :1 4S –O –J R3 0–J –M p– tutors is higher among economically advantaged p– 14 20 p ep eb 14 20 ep Se R1 g–1 2 eb Se :S :S g– 20 :F :F – :S 20 groups. Students who come from households that :F 5 5 ug R4 Au u R4 R3 15 :1 ov ov R4 :A R3 :A ov : :N :N R1 R1 are wealthier, located in urban settings, and those R1 : R1 :N R1 R2 R2 R2 who attend private schools are significantly more Time periods likely to have access to and engage in remote learning. This is consistent across both genders Girls Boys as well as time periods. Notes: When looking at time spent with private tutors, a) Time Periods: a similar and sharper inequality is observed. Round 1: Schools closed (Aug–14 Sept 2020) Both girls and boys in urban areas spent close Schools open (15 Sept 2020–Oct 2020) to twice the amount of time with private tutors Round 2: Schools closed (Nov 2020–Jan 2021) relative to those in rural areas (rural girls: 36 Round 3: Schools open (Feb–March 2021) minutes versus urban girls: 73 minutes). A more Round 4: Schools open (Sept–Oct 2021) severe disparity is seen between students from b) All boys were surveyed in the first part of the baseline. the poorest versus wealthiest households (22 c) The figures show weighted estimates. minutes and 48 minutes for girls respectively). SMS GIRL: DATA INSIGHTS | 12 How are household Figure 8b: Typical Student’s Time Use in a Day by Gender 5 chores and family care Family Care Household Chores Paid Work Outside Household distributed? 150 150 150 Girls spent more time on household chores, such as cooking and cleaning, relative to boys consistently across periods and spent over Minutes (Mean) 100 100 100 twice the amount of time as boys on household 89 chores during the first school closures 67 70 (August–14 September 2020). Time spent by 80 64 66 girls on household chores peaked during school 65 47 50 54 41 50 44 46 50 closures and decreased when schools reopened (figure 8b). During the first school reopenings 41 29 and second period of school closures, boys 29 30 27 26 16 18 22 11 were spending more time, on average, on 0 0 0 9 family care relative to girls. However, when 2 1 6 1 schools reopened in February–March 2021, Ma 21 –O 020 0 21 –O 021 21 0 1 1 20 R3 0–Ja 020 –M 021 –O 020 21 21 p– 202 02 02 2 boys and girls spent almost the same time on eb n 20 20 20 20 20 20 20 r2 p2 n2 2 t2 p2 n2 p ct ct ct ar ct ar family care. This trend changed when schools ct c Se a Se Se –Ja O –O –O –M R3 0–J – R1 g–14 4 ep ep ep 14 ep 20 ep eb reopened for the third time in September– eb Se –1 2 2 :S S S 20 :S 20 :F g– 20 :F :S :F ug 5 5 5 u R4 R4 Au :1 :1 November 2021, and girls (47 minutes) were R4 ov :1 :A R3 :A ov ov :N :N R1 R1 R1 R1 : :N R1 R2 spending considerably more time than boys R2 R2 Time periods (29 minutes) on family care. During the same time period, boys spent a lot more time working outside the household (11 minutes) relative to Girls Boys girls (1 minute) (figure 8b). However, the time Notes: both boys and girls spent working outside the a) Time Periods: household decreased considerably as compared Round 1: Schools closed (Aug–14 Sept 2020) to the second school reopenings in February– Schools open (15 Sept 2020–Oct 2020) March 2021 (figure 8b). Round 2: Schools closed (Nov 2020–Jan 2021) Round 3: Schools open (Feb–March 2021) Round 4: Schools open (Sept–Oct 2021) b) All boys were surveyed in the first part of the baseline. c) The figures show weighted estimates. SMS GIRL: DATA INSIGHTS | 13 CONCLUSION This brief presents key takeaways from four consecutive telephone surveys that have been conducted to assess the impact of school closures on girls’ education in Punjab, Pakistan. The brief highlights the need to focus on both boys and girls to ensure continued enrollment in schools. It also documents a reduction in time spent on remote learning, although engagement was low to begin with. Whereas families appear to be resorting to private tutoring for additional support as schools continued a closing and reopening pattern during the pandemic, this option appears to be positively correlated to wealth, leaving the children from poorer families at a greater disadvantage. In the subsequent and final round of data collection (planned for November 2021), we seek to collect student assessments data to understand the pandemic effects on the learning levels in Punjab. SMS GIRL DATA INSIGHTS This brief is part of a series produced by the team managing the SMS Girl Impact Evaluation. This includes Koen Geven, Tazeen Fasih, Amer Hasan, Rabea Malik, Javaeria Qureshi, Kevin MacDonald, Ayesha Tahir, Sheena Fazili, Najaf Zahra, and Naveed Hussain. Data has been collected by RCons (quantitative) and CERP (qualitative). This impact evaluation is supported by the School Education Department (SED) and the Punjab Information Technology Board (PITB), under the Third Punjab Education Support Project (PESP-3). Funding from the Strategic Impact Evaluation Fund and the South Asia Gender Innovation Lab is gratefully acknowledged. Figures Authors’ calculations based on SMS Girl survey data. Photo credits cover: STARS/Kristian Buus page 10: STARS/Kristian Buus page 13: STARS/Kristian Buus