UNICEF Working Paper Series

WP/2018/002

WE MUST DO BETTER: A CLOSER LOOK AT THE CONTEXTUAL FACTORS THAT DRIVE AND DISCIPLINE IN NEPAL NEPAL MULTIPLE INDICATOR CLUSTER SURVEY (MICS) 2014 FURTHER ANALYSIS REPORT

© United Nations Children's Fund (UNICEF) Nepal, 2017 PO Box 1187 United Nations House Nepal May 2018

This report gives the results of further analysis of the Nepal Multiple Indicator Cluster Survey (MICS), 2014, which was carried out on behalf of UNICEF Nepal by consultant Samik Adhikari. The four reports in this series are working documents: • Working paper 2018/001: Letting Children Flourish From an Early Age: Early Childhood Care and Development In Nepal • Working paper 2018/002: We Must Do Better: A Closer Look at the Contextual Factors that Drive Child Labour and Discipline in Nepal • Working paper 2018/003: Access to Communication Media and the Acceptance of Violence Among Adolescents Female in Nepal • Working paper 2018/004: Water, Sanitation and Hygiene (WASH) and Nutrition in Nepal with a Focus on Children Under Five

The four reports are available at www..org/nepal

The reviewer and editor of this series of UNICEF Nepal working papers is the Chief of the Planning and Monitoring Section, UNICEF Nepal. Please contact [email protected] or [email protected] for more information on UNICEF Nepal working papers.

The designations in this publication do not imply an opinion on the legal status of any country or territory, or of its authorities, or the delimitation of frontiers.

Suggested citation: United Nations Children’s Fund Nepal, We Must Do Better: A closer look at the contextual factors that drive child labour and discipline in Nepal: Nepal Multiple Indicator Cluster Survey (MICS) 2014 further analysis report, UNICEF Nepal Working Paper Series WP/2018/002, UNICEF Nepal, Kathmandu, 2018.

UNICEF Nepal Working Paper Series

WP/2018/002

WE MUST DO BETTER: A CLOSER LOOK AT THE CONTEXTUAL FACTORS THAT DRIVE CHILD LABOUR AND DISCIPLINE IN NEPAL NEPAL MULTIPLE INDICATOR CLUSTER SURVEY (MICS) 2014 FURTHER ANALYSIS REPORT

EXECUTIVE SUMMARY

This further analysis report on child labour and child discipline uses data from the fifth round of the Nepal Multiple Indicator Cluster Survey (the 2014 NMICS) to shed light on a number of key domains for children aged 1–17 years in Nepal. The main purpose is to highlight statistics and trends on child labour and discipline, including the extent of the performance of economic activities and household chores by children beyond the acceptable number of hours, how many of them work in hazardous conditions, and the subjecting of children to violent discipline. Sub-group analysis by geographical area and socioeconomic characteristics helps identify the most marginalized children who suffer from exploitation and violence.

This further analysis shows that over one in three children aged 5–17 years in the NMICS 2014 sample were engaged in at least one form of child labour. When disaggregated by age group, children in the younger 5–11 years cohort were more likely to be engaged in economic activities above the age-acceptable number of hours, possibly because of the justifiably lower thresholds for younger children, whereas children in the 12–17 years cohort were substantially more likely to be working in hazardous conditions. On child discipline, the further analysis found that more than 80 per cent of the sampled children aged 1–14 years had been subject to some form of violent discipline in their households in the month preceding the survey. Most had been subjected to psychological aggression, but more than 50 per cent had suffered some form of physical violence. Worse still, the younger children aged 1–4 year had suffered physical punishment to a larger extent than the children aged 12–14 year.

The child labour and child discipline indicators were disaggregated by geography, sex, wealth quintile and caste and ethnicity. This showed child labour to be more prevalent in the hill and mountainous areas compared to the Tarai plains for the sample children, whereas every geographic area except for the had a high prevalence of some form of violent disciplining of children.

Household wealth was strongly correlated with the extent of child labour and certain child discipline indicators. While in the case of child labour, moving from the poorest to subsequently higher wealth quintiles lowered the percentage of children working as child labourers; while in the case of child discipline, the jump from the poorest to the second and third quintiles did not produce as significant a decrease in the prevalence of violent discipline the children faced. The children in hill Chhetri households were more likely to be engaged in child labour whereas children in the Madhes-based social group households were more than twice as likely to be subject to physical punishment than children in the hill Brahmin households. The logistic regressions found that the children whose fathers were working outside Nepal were both more likely to be engaged in child labour and to be subject to violent discipline.

An important contributory factor to the continuing prevalence of child labour and violence against children in Nepal is that, despite the introduction and strengthening of rules and regulations to prohibit the exploitation of and violence against children, inconsistencies remain between international and national norms on child . The new modules on child labour and child discipline introduced in the fifth round of the NMICS have allowed for the deeper exploration of the contextual elements that drive child labour and discipline, as presented in this report.

i CONTENTS

Executive summary ...... i Contents ...... ii Tables ...... iii Figures ...... iv 1 Introduction ...... 1 1.1 Background ...... 2 1.2 National policies and programmes addressing child labour and violence against children . 3 2 Data and methods ...... 4 2.1 Data ...... 4 2.2 Methods ...... 4 2.3 Definitions of indicators ...... 5 2.4 Data limitations ...... 6 3 Results ...... 7 3.1 Profile of child labourers in Nepal ...... 7 3.2 Child labour ...... 10 3.3 Child discipline ...... 14 3.4 Logistic regression analysis of determinants of child labour and violent child discipline .. 18 4 Discussion and conclusions ...... 23 References ...... 25 Appendix 1: Map of the NMICS ecozones and Nepal’s former development regions ...... 26 Appendix 2: Additional maps showing results of the NMICS 2014 further analysis ...... 27

ii TABLES

Table 1: Religions, languages, and caste and ethnicity of household heads in households with child labourers aged 5–17 years (NMICS 2014) ...... 8 Table 2: Average hours worked by children by age point in the week preceding the survey above the acceptable child labour thresholds by type of activity (NMICS 2014) ...... 10 Table 3: Percentage of children aged 5–17 years engaged in child labour – by type and age category (NMICS 2014) ...... 10 Table 4: Percentage point difference among children aged 5–17 years engaged in different forms of child labour – by type and sex (NMICS 2014) ...... 12 Table 5: Percentage point differences among children aged 5–17 years engaged in different aspects of child labour – by type and urban/rural residence (NMICS 2014) ...... 12 Table 6: Percentage of children aged 1–14 years subjected to violent discipline – by type and age- category (NMICS 2014) ...... 15 Table 7: Percentage difference among children aged 1–14 years who had been subjected to violent discipline in their households – by type and sex (NMICS 2014) ...... 16 Table 8: Percentage points difference among children aged 1–14 years who had been subjected to violent discipline in their households – by type and urban/rural residence (NMICS 2014) ...... 16 Table 9: Determinants of child labour among children aged 5–17 years using a logistic regression framework (NMICS 2014) ...... 19 Table 10: Determinants of violent child discipline among children aged 1–14 years using a logistic regression framework (NMICS 2014) ...... 21

iii FIGURES

Figure 1: Equity tree of percentage of children aged 5–17 years working under hazardous conditions in Nepal (NMICS 2014) ...... 2 Figure 2: Distribution of household sizes of children aged 5–17 years who were engaged in child labour in the NMICS 2014 sample ...... 7 Figure 3: Percentage of children aged 5–17 years engaged in child labour – by age and educational attainment status (NMICS 2014) ...... 9 Figure 4: Percentage of children aged 5–17 years engaged in child labour – by NMICS 2014 sampling cluster ...... 11 Figure 5: Percentage of children aged 5–17 years engaged in different forms of child labour – by type and wealth quintile (NMICS 2014) ...... 13 Figure 6: Percentage of children aged 5–17 years engaged in the different forms of child labour – by type of child labour and caste and ethnicity (NMICS 2014) ...... 14 Figure 7: Percentage of children aged 1–14 years subjected to any form of violent discipline method – by NMICS 2014 sampling cluster ...... 15 Figure 8: Percentage of children aged 1–14 years subjected to violent discipline methods in their households – by type and wealth quintile (NMICS 2014) ...... 17 Figure 9: Percentage of children aged 1–14 years subjected to violent discipline in their households – by type and caste/ethnicity (NMICS 2014) ...... 18 Figure A1: Economic activity among children aged 5–17 years, NMICS 2014 sampling cluster ...... 27 Figure A2: Household chores carried out by children aged 5–17 years, NMICS 2014 sampling cluster27 Figure A3: Hazardous working condition among children aged 5–17 years, NMICS 2014 sampling cluster ...... 28 Figure A4: Psychological aggression suffered by children aged 1–14 years, NMICS 2014 sampling clust ...... 28 Figure A5: Physical punishment suffered by children aged 1–14 years, NMICS 2014 sampling cluster 29 Figure A6: Severe physical punishment suffered by children aged 1–14 years, NMICS 2014 sampling clusters ...... 29

iv 1 INTRODUCTION According to a 2012 estimate by the International Labour Organization (ILO), around 168 million children worldwide, or 11 per cent of all children aged 5–17 years, were engaged in activities deemed as child labour.1 About half of those children (85 million children) were working in hazardous conditions. And although the overall percentage of children engaged in child labour and hazardous working condition has decreased since 2000, far too many children’s futures are compromised in this way. Furthermore, many children aged 1–17 years regularly face physical and psychological violence. While hard data on the exact nature and frequency of violence against children is hard to obtain, the Multiple Indicator Cluster Surveys (MICS) administered by UNICEF generate comparable data on child discipline-related indicators for more than 70 countries.2

The situation of labour and violence against children is particularly grim in South Asia with an estimated 17 million child labourers aged 5–17 years.3 The absolute number of child labourers is highest in India (5.8 million), but Nepal fares worse in percentage terms. Around 36 per cent of Nepal’s 7–14 year-olds are estimated to be child labourers, followed by 12 per cent in Bangladesh and 11 per cent in Sri Lanka.4 This shows the relative lack of progress in Nepal on abolishing child labour. Additionally, many Nepali children face mental and physical abuse that puts them in harms way and prevents them having normal childhoods.

This further analysis report uses data from the fifth round of the Nepal Multiple Indicator Cluster Survey (the 2014 NMICS)5 to shed light on child labour and child discipline in Nepal. This report presents findings on the three forms of child labour of economic activity, household chores and hazardous work. It also presents the findings on child discipline concerning psychological aggression, physical punishment, severe physical punishment and adults’ perceptions on the use of violence to discipline children.

The 2014 NMICS 2014 found wide variations in the situation of children aged 5–17 years working under hazardous conditions by rural/urban residence and by ecozone (Figure 1).

The main purpose of this further analysis is to identify the most marginalized and vulnerable groups of children based on sub-group analysis by individual and household characteristics including sex, geography, wealth and caste and ethnicity. Finally, this analysis tested for associations between individual and household characteristics and indicators in the two domains of interest – child labour and child discipline, using a logistic regression framework.

1 International Labour Organization, Marking Progress against Child Labour: Global estimates and trends 2000–2012, International Labour Office, IPEC, Geneva, 2013, , accessed March 2017. 2 United Nations Children’s Fund, ‘UNICEF Data: Monitoring the Situation of Women and Children’, 2017, , accessed 25 March 2017. 3 International Labour Organization, South Asia – Fact Sheet: Children in Labour and Employment, ILO, New Delhi, n.d. , accessed 25 March 2017. 4 Sherin, K. and Scott Lyon, Measuring Children’s Work in South Asia: Perspectives from national household surveys, International Labour Organization for South Asia, ILO Country Office, New Delhi, 2015, , accessed 25 March 2017. 5 Central Bureau of Statistics, Nepal Multiple Indicator Cluster Survey 2014: Key findings report, CBS and UNICEF, Kathmandu, Nepal, 2014, , accessed 2 March 2018.

1 Figure 1: Equity tree of percentage of children aged 5–17 years working under hazardous conditions in Nepal (NMICS 2014)

1.1 Background Although there is no universally agreed definition of child labour,6 UNICEF defines child labour taking into account the different activities that children may be involved in that could harm their overall well- being. The details of the definition used by the MICS are explored in Section 2.3. What makes it difficult to design policies on child labour and discipline is that several types of child exploitation and abuse remain within households and communities and their social mores, making it difficult to bring the subject onto the policymaking radar.

The Sustainable Development Goals (SDGs) for 2030 have the following targets to address child labour and violence against children: • Target 8.7: Take immediate and effective measures to eradicate forced labour, end modern slavery and human trafficking and secure the prohibition and elimination of the worst forms of child labour, including recruitment and use of child soldiers, and by 2025 end child labour in all its forms. • Target 16.2: End abuse, exploitation, trafficking and all forms of violence and torture against children.

There is a clear need for coordinated action against these harmful practices to achieve the targets. This further analysis report provides further evidence of the situation of child labour and violence against children in Nepal by presenting disaggregated findings from the 2014 NMICS, which for the first time included modules on child labour and child discipline. By shedding light on these domains, this report hopes to influence policymaking and programmatic agendas in Nepal in order to achieve progress on the SDGs towards the target date of 2030.

6 Edmonds, Eric V, Defining Child Labour: A review of the definitions of child labour in policy research, Working Paper, Geneva, International Programme on the Elimination of Child Labour (IPEC), Geneva, 2008, accessed March 2018.

2 1.2 National policies and programmes addressing child labour and violence against children Among the laws and regulations designed to prevent the exploitation and abuse of children, the Convention on the Rights of the Child (1989) is the most rapidly and widely ratified instrument.7 According to UNICEF, this convention “changed the way children are viewed and treated – i.e., as human beings with a distinct set of rights instead of as passive objects of care and charity.”8 Some of its articles directly pertain to child labour and abuse. For example, Article 32 mandates “Protection from economic exploitation and from performing any work that is likely to be hazardous or to interfere with the child's education, or to be harmful to the child's health or physical, mental, spiritual, moral or social development.”

Nepal ratified the convention in 1990 and has subsequently enacted laws and programmes to combat child exploitation and various types of abuse. In line with the convention the Interim (2007) said: “Every child shall have the right not to be subjected to physical, mental or any other form of exploitation. Any such act of exploitation shall be punishable by law and any child so treated shall be compensated as determined by law… [and that] no minor shall be employed in factories, mines or in any other hazardous work nor shall be used in the army, police or in conflicts.” 9

Other pertinent laws that have been enacted in Nepal since 1990 include: • Children’s Act (1992) • Child Labour (Prohibition and Regulation) Act (2000) • Labour Rules (1993) • Bonded Labour (Prohibition) Act (2002) • Human Trafficking and Transportation (Control) Act (2007) • National Master Plan on Child Labour (2004-14) • National Master Plan on the Elimination of (the Worst Forms of) (2011–2020) (interim draft).

Although Nepal has ratified a number of international instruments on child labour and protection, inconsistencies remain between Nepal’s legislation and international child labour norms. Of particular concern is that the family and the informal work sector are not covered by Nepal’s child labour legislation, and the age of admission to hazardous work is only 16 years. UNICEF is supporting the government to revise and implement the National Master Plan on Child Labour (2011–2020). UNICEF is also advocating for gaps in the existing legislation to be filled and for stronger law enforcement.

This further analysis report provides evidence on critical information gaps on child labour and discipline including issues that may not have been salient until now in the Nepali context. The analysis is based on the responses to the newly included child labour and child discipline modules in the fifth round NMICS. The analysis presented here attempts to identify the most vulnerable children by presenting disaggregated analysis by type of child labour and discipline, age category, geographical area, sex, wealth quintile and caste and ethnicity.

7 United Nations Treaty Collection, ‘Chapter IV: Human Rights, 11 – Convention on the Rights of the Child, New York, 1989’, , accessed 6 November 2016. 8 United Nations Children’s Fund, ‘Convention on the Rights of the Child’, , accessed March 2018. 9 The Government of Nepal, Interim Constitution of Nepal, 2007, , accessed 25 March 2017.

3 2 DATA AND METHODS

2.1 Data The 2014 NMICS covered 12,405 households and provides a comprehensive picture on women and children across all of Nepal’s 15 sub-regions, covering the mountain, hill and southern Tarai plain areas of Nepal’s five development regions (see Appendix 1).10 Of particular interest to this further analysis are the sections of the survey on child labour and discipline, individual and household characteristics such as sex, age, caste and ethnicity, wealth quintile, rural/urban location, as well as the sections on level of education and family composition.

The sample of interest for this analysis is children aged 1–17 years, who made up 21,857 of the 56,539 household members surveyed in the 2014 NMICS. Note that the children in the NMICS sample were the children and child relatives of the heads of the sample households plus other under 18 year-olds (including domestic workers) who had been normally resident in the household for the previous six months. In this context it is important to note that sons of the heads of households are likely to be treated differently to, for example, child domestic workers. No analysis of such differences was carried out.

On child labour, one child aged between 5 and 17 years old was randomly selected per household where there was more than one child in this age group. Also, for the analysis of child discipline, one child between 1–14 years was randomly selected per household where there was more than one child in the age group. The section on child labour thus presents data on 7,146 children aged 5–17 years, while the section on child discipline presents data on 7,554 children aged 1–14 years.

2.2 Methods This study analysed the basic profile of child labourers identified in the 2014 NMICS (covering demographic indicators, educational attainment, household composition, and number of hours worked). The following two domains and sub-domains were selected based on their relevance for the current study:

Child labour, which includes: • economic activity classified as child labour • household chores classified as child labour • hazardous working conditions for child labourers.

Violent discipline, which includes: • psychological aggression • physical punishment • severe physical punishment.

Each of the two domains was explored in further detail by disaggregating relevant indicators and analysing differences in the mean values by sex, urban/rural location, wealth quintile, caste and ethnicity and, in some cases, by educational and marital status. Confidence intervals were used to assess

10 Note that until 2017 Nepal was divided into five development regions – Far Western, Mid-Western, Western, Central and Eastern Development regions (see Appendix 1). These regions have been superseded by the division of the country into seven provinces under the new federal system of governance. The country’s three ecozones (mountains, hills and Tarai) are also shown in Appendix 1.

4 the significance of differences. Finally, associations between the two domains and household and individual characteristics were explored using a logistic regression framework.

The statistical software Stata (version 14), was used for the analysis. Elements of sample design were taken into account by using Stata’s ‘svyset’ command (including information on sample weights, clusters, and strata). Most graphs were made in Tableau and the maps were made using R software.

2.3 Definitions of indicators 1. Child labour

Four categories of child labour were explored: a) Economic activity was classified as work performed on a household’s plot, farm or home garden, looking after animals, helping in the family or relatives’ businesses (either with or without pay) or buying and selling articles, handicrafts, clothes, food and agricultural products. Children were deemed to be engaged in child labour if: • 5–11 year-olds had worked at least an hour on these activities in the preceding week • 12–14 year-olds had worked at least 14 hours on these activities in the preceding week • 15–17 year-olds had worked at least 43 hours on these activities in the preceding week. b) Household chores were classified as fetching water, collecting firewood, shopping for the household, repairing household equipment, cooking and cleaning household utensils, cleaning the house, washing clothes, caring for children, caring for the old or sick or being engaged in other household tasks. Children were deemed to be engaged in child labour if: • 5-11 year-olds had worked at least 28 hours on these activities in the preceding week • 12-14 year-olds had worked at least 28 hours on these activities in the preceding week • 15–17 year-olds had worked at least 43 hours on these activities in the preceding week. c) Hazardous working conditions were defined as i) work performed carrying heavy loads, working with dangerous tools, operating heavy machinery, or performed in an environment that exposed children to dust, fumes, gas, extreme cold, heat, humidity, loud noise or vibration; and ii) working at heights, working with chemicals such as pesticides, glues or explosives, and other things that can be bad for children’s health and safety. A child performing any of these activities, regardless of age category was deemed to be engaged in child labour. d) ‘Overall child labour’ is a composite indicator where a child deemed to be engaged in child labour in any of the above activities is classified to be engaged in overall child labour.

2. Child discipline

The following four subjects were explored on child discipline: a) Psychological aggression – A child was deemed to have experienced psychological aggression if any member of their household had shouted, screamed, or yelled at them, or called them dumb, lazy, or another pejorative term in the month preceding the survey. b) Physical punishment – A child was deemed to have experienced physical punishment if a member of their household had inflicted any of the following things on them in the month preceding the survey — shaken, spanked, hit or slapped them on the bottom with bare hands; hit them on the bottom or elsewhere on the body with a belt, hairbrush, stick, or other hard object; hit or slapped them on the face, head or ears; hit or slapped them on the hand, arms or legs; or beat them repeatedly and forcefully.

5 c) Severe physical punishment – A child was deemed to have suffered from several physical punishment if any member of their household had hit or slapped them on the face, head or ears, or beat them repeatedly as hard as they could in the month preceding the survey. d) Any violent discipline – A child was defined as having suffered from violent discipline if they had been subject to psychological aggression or any sort of physical punishment, including severe physical punishment, by any member of the household in the month preceding the survey.

2.4 Data limitations The NMICS is a comprehensive tool to assess the well-being of women and children globally. However, for the purposes of this report, the NMICS questionnaire has its limitations: • The analysis is constrained to children in the 1–17 age group. Furthermore, the section on child labour includes data for children aged 5–17, while for child discipline, the analysis includes data for children aged 1–14. It could be argued that both these domains are relevant to children across the entire 1–17 year-old cohort. • The NMICS produces cross-sectional data and so all the analysis in this report should be taken as descriptive rather than causal. • Certain useful information is not collected including the time children spent working in hazardous conditions (in the week preceding the survey) and which member of the household had used violent discipline against them.

The analysis that follows makes the best use of the available data to shed light on child labour and discipline in Nepal.

6 3 RESULTS

3.1 Profile of child labourers in Nepal The 2014 NMICS sample included 21,857 children aged between 1 and 17 years (or 38.7 per cent of the 56,539 household members sampled). Of these, 17,295 (or 30.1 per cent of the total) were 5–17 year- olds. And 8,194 or 66.1 per cent of the 12,405 households sampled had at least one child aged between 5 and 17 years. The NMICS child labour module was only administered for one randomly selected child aged 5–17 years in each household, with the questions asked of any adult household member who responded to the questionnaire.

Overall, 37.4 per cent of the children aged 5–17 years surveyed in the NMICS were found to be engaged in child labour. There was, however, considerable variation in the percentage of children deemed to be working as child labourers when the numbers were disaggregated by type (economic activity, household chores, hazardous child labour) and by the three age categories (5–11, 12–14 and 15–17 years). According to the NMICS data, 19 per cent of children aged 5–17 years were involved in economic activity above the age-specific number of acceptable hours, 3.8 per cent of children aged 5–17 years were involved in household chores above the age-specific threshold, and 30 per cent of children aged 5– 17 years were working in hazardous conditions.

This section presents the basic profile of the children aged 5–17 years who were recorded as being engaged in some form or aspect of child labour by the 2014 NMICS.

Demographics Among the NMICS sample children who were engaged in child labour, 52.2 per cent were females. The average household size with children engaged in one of the three aspects of child labour was 6.4 with a 95 per cent confidence interval of household size ranging from 6.0 to 6.8. Although the average household size of children aged 5–17 years not engaged in child labour was lower at 6.1, the difference is not statistically significant to the average size of households with children who were engaged in child labour.

Figure 2: Distribution of household sizes of children aged 5–17 years who were engaged in child labour in the NMICS 2014 sample

Close to 60 per cent of children aged 5–17 years in the NMICS sample and engaged in child labour lived in households with 4–6 members.

7 Close to 58 per cent of the children aged 5–17 years and engaged in child labour lived in a household with 4 to 6 members, which was not statistically different to the proportion of children aged 5–17 years who were not engaged in child labour (Figure 2).

Table 1 summarizes the religions, languages, and caste and ethnicity of household heads in households where children aged 5–17 years were engaged in child labour. The majority of child labourers lived in Nepali-speaking Hindu households. Maithili, Bhojpuri and Tharu were other prevalent languages. There were significant differences in the caste and ethnic identity of child labourers compared to Nepal’s overall population composition. Children aged 5–17 years from hill Chhetri, hill , and hill Janajati (ethnic group) households were significantly more likely to be engaged in child labour. On the other hand, 5–17 year-olds from hill Brahmin households were significantly less likely to be engaged in child labour. Table 1: Religions, languages, and caste and ethnicity of household heads in households with child labourers aged 5–17 years (NMICS 2014) Religion Language Caste and ethnicity

Hinduism 85.9% Nepali 44.4% Hill Chhetri 23.2% Buddhism 5.8% Maithili 14.5% Hill Janajati 21.3%

Islam 3.7% Bhojpuri 6.6% Madhesi (non- 13.1% Brahmin, Chhetri) Kirat 3.0% Tharu 5.4% Hill Dalit 12.3% Christian 1.6% Tamang 3.5% Tarai Janajati 7.8%

Other 0.01% Limbu 2.3% Hill Brahmin 7.3% Newar 1.1% Madhesi Dalit 4.5% Gurung 1.0% Muslim 3.8% Other 21.2% Other 6.7% Sample size (N) = 2,919

Level of education of children engaged in child labour Figure 3 shows the level of education of the 5–17 year-olds engaged in child labour in the NMICS sample. It shows that most children, even when they were engaged in activities deemed as child labour, could read and write and had attended school for at least a year. There was no statistical difference in the latter two indicators among children engaged in child labour and those who were not.

The proportion of the children aged 5–17 year who were engaged in child labour and attending school dropped off considerably from 13 years of age. Overall, 82 per cent of the children aged 13–17 years who were engaged in child labour were attending school, while almost 90 per cent of the children aged 13–17 years who were not engaged in child labour were attending school. This difference is statistically significant at the 99 per cent significance level. There was no statistically significant difference in any of the three indicators explored in this section by sex among child labourers aged 5–17 years in the sample.

8 Figure 3: Percentage of children aged 5–17 years engaged in child labour – by age and educational attainment status (NMICS 2014)

Most children aged 5–17 years and engaged in child labour could read and write and had attended school for at least for a year. However, the number of children engaged in child labour and currently attending school dropped off after the age of 13.

Family composition of children engaged in child labour Among the 5–17 year-olds children engaged in child labour, 90.5 per cent lived with their natural mothers and 71.7 per cent with their natural fathers. When disaggregated by age, 83.2 per cent of the 15–17 year-olds child labourers lived with their biological mothers compared to 93.2 per cent of the 5– 14 year-olds engaged in child labour. There was no statistical difference in the percentage of children who lived with their biological fathers by age and being a child labourer.

There was no statistically significant difference in the child labour status of the children aged 5–17 years whose mothers or fathers lived abroad. However, more fathers of 5–14 year-olds child labourers were working abroad (17.3 per cent) than fathers of 15–17 year-olds child labourers (9.2 per cent).

Average hours worked by children engaged in child labour Table 2 presents the average number of hours worked by the children above the acceptable number of hours (see Box 1) on economic activities and household chores for each age point (year) between 5 and 17 years. Note that the threshold has to be at some point, but the division means, for example, that 11 year-olds can carry out only one hour while 12 year-olds can carry out 14 hours of economic activity to remain just under the threshold.

Box 1: UNICEF’s child labour thresholds — i.e. number of hours work that are acceptable Involvement in economic activities: Involvement in household chores: • aged 5–11 years: 1 hour or less per week • aged 5–14 years: 28 hours or less per week • aged 12–14 years: 14 hours or less per week • aged 15–17 years: 43 hours or less per week • aged 15–17 years: 43 hours or less per week

9 The data shows that across the thirteen age points that the children in the sample were on average working between 5.7 and 15.8 hours more per week on economic activities and between 6.7 and 19.2 hours more on household chores than the threshold number of hours (Table 2). Table 2: Average hours worked by children by age point in the week preceding the survey above the acceptable child labour thresholds by type of activity (NMICS 2014) Age Economic activity Household chores Age Child labour Child labour 5 6.22 6.72 6 5.68 7.12 7 6.92 8.93 8 6.15 9.15 9 8.20 10.89 10 8.71 11.28 11 8.84 10.96 12 9.93 13.83 13 11.59 16.17 14 10.72 13.57 15 12.52 15.63 16 14.27 15.28 17 15.84 19.22

Sample size (N) 2,919 2,919

And if the same age-specific thresholds are used for household chores as are used for economic activities (see Box 1), then the overall extent of child labour according to the 2014 NMICS findings would be much greater than is shown in Table 3. It would mean that more than 30 per cent of the children would be deemed as child labourers working more than the age-specific hours on household chores, and 54 per cent of the 5–17 year-olds would come under the overall child labour category.

3.2 Child labour Table 3 presents the percentage of children who were working as child labourers by type of labour and age category, along with 95 per cent confidence interval estimates. Table 3: Percentage of children aged 5–17 years engaged in child labour – by type and age category (NMICS 2014) Age category Type 5 to 11 12 to 14 15 to 17 Economic activity 27.9 14.9 2.6 95% confidence interval [25.1, 30.7] [12.1, 17.7] [1.6, 3.7]

Household chores 2.7 7.0 2.8 95% confidence interval [1.8, 3.5] [5.3, 8.6] [1.7, 3.9]

Hazardous working conditions 17.4 42.0 45.7 95% confidence interval [15.1, 19.8] [38.1,45.9] [41.6,49.8]

Sample size(N) 3,496 1,858 1,792

10 Although the children aged 15–17 years were the least likely among the three age categories to be involved in economic activity above the acceptable number of hours, they were the most likely to be engaged in hazardous work. The NMICS sample children were more likely to be involved in hazardous work as they moved from younger (5–11 years) to older ages (12–17 years). The 12–14 year-olds were most likely to be performing household chores above their age-specific thresholds compared to the 5– 11 and 15–17 year-olds.

Geographic variation in overall child labour Figure 4 shows the geographic variation for the overall NMICS 2014 child labour indicator among the 5– 17 year-olds.11 Each sampling cluster covered in the NMICS is represented by a dot on the map. Blue dots represent the lower prevalence of child labour in clusters while red dots represent a higher prevalence along the spectrum from blue to red. The indicator used to measure overall child labour is coded as 1 if children aged 5–17 years in the concerned clusters had carried out economic activities or household chores above the maximum number of allowable hours, or were working in hazardous conditions.

For example, the value is 100 per cent for a cluster if all children aged 5–17 years in that cluster were involved in at least one of the above three aspects of child labour or, 50 per cent if a half of the children aged 5–17 years were involved in at least one of the three aspects, and so on. Therefore, what the red dots indicate are the clusters where most children were engaged in at least one form of activity deemed as child labour. Figure 4 shows that, on average, child labour was more prevalent in the hills and mountains than in the Tarai plains. However, wide variations existed, even within sub-regions.

Figure 4: Percentage of children aged 5–17 years engaged in child labour – by NMICS 2014 sampling cluster

On average, children aged 5–17 years living in the hills and mountains were more likely to be involved in child labour than children in the Tarai.

11 See Appendix 2 for similar maps of the results of other child labour and discipline indicators.

11 Variation by sex, urban/rural location, wealth quintile, and caste and ethnicity This section presents findings on the different forms of child labour disaggregated by sex, urban/rural residence, wealth quintile and caste and ethnicity among the children aged 5–17 years to identify which groups were most vulnerable to being child labourers.

Table 4 presents percentage point differences in the different forms of child labour by sex among the 5– 17 year-olds. While there was no statistical difference in the overall child labour statistics and economic activity above the age-specific number of hours between males and females, the females were significantly more likely to be engaged in household chores above the threshold acceptable hours, and to be working in hazardous condition compared to their male counterparts. The females aged 5–17 years were 3.3 percentage points more likely to be engaged in household chores above the acceptable number of hours and 3.0 percentage points more likely to be engaged in activities deemed hazardous for children. Table 4: Percentage point difference among children aged 5–17 years engaged in different forms of child labour – by type and sex (NMICS 2014)

Economic Household Hazardous Overall child activity chores working labour conditions Females -0.002 0.033*** 0.030* 0.019 (0.901) (0.000) (0.075) (0.270) Observations 7,127 7,146 7,146 7,127 p-values: * p < 0.1; ** p < 0.05; *** p < 0.01

Table 5 presents the percentage point differences in the different aspects of child labour by whether the children lived in an urban or rural area. The results show that across Nepal, children living in urban areas were less likely to be engaged in all three aspects of child labour. The children in the urban areas were 13.4 percentage points less likely to be involved in economic activities above the age-specific threshold, 3.8 percentage points less likely to be involved in household chores above the age-specific threshold, 19.8 percentage points less likely to be involved in hazardous work, and were overall 24.7 percentage points less likely to be engaged in any of three child labour activities. Table 5: Percentage point differences among children aged 5–17 years engaged in different aspects of child labour – by type and urban/rural residence (NMICS 2014)

Economic Household Hazardous Overall child activity chores working labour conditions Urban -0.134*** -0.038*** -0.198*** -0.247*** households (0.000) (0.000) (0.000) (0.000) Observations 7,127 7,146 7,146 7,127 p-values: * p < 0.1; ** p < 0.05; *** p < 0.01

Figure 5 shows the results for the percentage of children engaged in the three aspects of child labour by wealth quintile. It shows a significant variation in the prevalence of child labour by wealth quintile, particularly for working in hazardous conditions. More than 50 per cent of the children aged 5–17 years from the poorest wealth quintile were engaged in hazardous work. The children who were engaged in household chores for more than the acceptable number of hours showed the least variation in terms of wealth quintile; but even so, children in the poorest quintile were significantly more likely to be involved in more than the acceptable number of hours on household chores. Finally, the wealth quintile of the children showed a linear relationship with the percentage of them involved in economic activities for

12 more than the acceptable number of hours, pointing to the fact that moving from one wealth quintile to another produces significant decreases in the proportion of children involved in economic activities for more than the acceptable number of hours.

Figure 5: Percentage of children aged 5–17 years engaged in different forms of child labour – by type and wealth quintile (NMICS 2014)

There was a 50-percentage point difference in the percentage of children aged 5–17 years from the richest and poorest wealth quintiles who were engaged in child labour. And finally, Figure 6 presents the results on the percentage of the children engaged in the different aspects of child labour by caste and ethnicity of the household heads.

13 Figure 6: Percentage of children aged 5–17 years engaged in the different forms of child labour – by type of child labour and caste and ethnicity (NMICS 2014)

The children in the hill Dalit and hill Chhetri households were statistically and significantly more likely to be engaged in child labour than the children of other ethnicities.

The prevalence of child labour varied considerably with a 20 percentage points difference between the two groups most likely to be engaged in child labour (hill Dalit and hill Chhetri household children) and the two groups least likely to be engaged in child labour (Newar and hill Brahmin household children). The Muslim household children were least subjected to hazardous working conditions. Overall, the differences in the prevalence of the different aspects of child labour by caste and ethnicity showed less variation than the differences by wealth quintile except for the case of household chores where variation by caste and ethnicity and wealth quintile was considerably less than for the other aspects of child labour.

3.3 Child discipline The 2014 NMICS sample included 18,295 one to fourteen year-olds (or 32.4% of the total sample of children aged 1 to 17 years). A very large proportion of the NMICS sample of 1–14 year-olds (81.7%) had been subjected to some form of violent discipline in the month preceding the survey. And similar to the child labour findings, there are important variations in the child discipline statistics when disaggregated by type and age. Overall, 77.6 per cent of the 1–14 year-olds had been subjected to psychological aggression in their households, 53.3 per cent of them to some form of physical punishment and 14.3 per cent to severe physical punishment. The latter can be particularly damaging to young children’s physical and psychological well-being.

Table 6 presents the child discipline statistics by three age categories and the three types of violent discipline alongside 95 per cent confidence intervals of the estimates. The results show that the 12–14 year-olds were significantly less likely to be subjected to all three types of violent discipline.

The 5–11 year-olds were statistically more likely to suffer from psychological aggression with a significant difference between the first two age categories (1–4 and 5–11 year-olds) in terms of being subjected to physical punishment, including severe forms. A disturbing finding is that more than a half of

14 the 1-4 year-olds had been subjected to physical punishment and 14.5 per cent to severe physical punishment, which can damage young children’s physical and psychological well-being in many ways. Table 6: Percentage of children aged 1–14 years subjected to violent discipline – by type and age- category (NMICS 2014) Age categories* Type 1 to 4 years 5 to 11 years 12 to 14 years Psychological aggression 68.4 82.3 75.8 95% confidence interval [65.4, 71.4] [80.4, 84.1] [72.4, 79.2] Any physical punishment 56.5 58.6 35.7 95% confidence interval [53.4, 59.7] [56.1, 61.1] [31.6, 39.9] Severe physical punishment 14.5 16.2 9.1 95% confidence interval [11.9, 17.1] [13.9, 18.5] [6.7, 11.5] Sample size (N) 2,200 3,496 1,262 Note: * The three age categories were defined arbitrarily for this table although the 5–11 and 12–14 year categories are consistent with the age categories used for the above analysis of child labour.

Geographical variation in overall violent child discipline Figure 7 shows the geographical variation in the extent to which the 1–14 year-olds were subjected to violent discipline. Each cluster is represented by a dot on the map with blue dots representing the lowest prevalence of violent disciplining of the children and red dots the highest prevalence across the spectrum from blue to red. The indicator used to measure violent child discipline was coded as 1 if the 1–14 year-olds in a sampling cluster had been subjected to psychological aggression or some form of physical punishment, including severe physical punishment. For example, the value is 100 per cent for a cluster if all children aged 1–14 years in that cluster were subjected to at least one form of violent discipline in their household, 50 per cent if a half had been subjected to violent discipline, and so on. Therefore, the red dots show the clusters where most children were subjected to some form of psychological aggression or physical punishment in their households. The map, which is covered with red dots except for the area surrounding the Kathmandu Valley, presents a stark message for all Nepal’s child protection stakeholders.

Figure 7: Percentage of children aged 1–14 years subjected to any form of violent discipline method – by NMICS 2014 sampling cluster

The 1–14 years in every geographic area covered in the NMICS sample, except around the Kathmandu Valley, were subject to at least one form of violent discipline in their households.

15 Variation by sex, urban/rural residence, wealth quintile and caste and ethnicity This section presents findings on the proportion of 1–14 year-olds in the NMICS 2014 survey who had been subjected to violent discipline in their households, disaggregated by sex, urban/rural residence, wealth quintile and caste and ethnicity.

Table 7 presents the percentage point differences across the three specific indicators and the one overall indicator by sex. It shows that girls aged 1–14 years were slightly, but significantly less likely, to be physically punished including severe physical punishment, compared to boys of the same age. The girls were 3.8 percentage points less likely to suffer physical punishment of any kind and 2.2 percentage points less likely to suffer severe physical punishment. There is no statistically significant difference by the sex of children who suffered psychological aggression in their households. Overall, female children aged 1–14 years were 2 percentage points less likely to suffer any form of violent discipline in their households. Table 7: Percentage difference among children aged 1–14 years who had been subjected to violent discipline in their households – by type and sex (NMICS 2014)

Psychological Any physical Severe physical Any violent aggression punishment punishment discipline Females -0.020 -0.038** -0.022* -0.020* (0.125) (0.019) (0.056) (0.086) Observations 7,554 7,554 7,554 7,554 p-values: * p < 0.1; ** p < 0.05; *** p < 0.01

Table 8 presents percentage point differences in the different types of violent discipline suffered by the 1–14 year-olds in the 2014 NMICS sample, based on whether they resided in urban or rural areas. Universally, the children in urban areas had been less subjected to violent discipline than their rural counterparts. The 1–14 year-olds in urban areas were 8.1 percentage points less likely to have suffered psychological aggression, 10.5 percentage points less likely to have suffered any physical punishment and 5.6 percentage points less likely to have suffered severe physical punishment than the children in rural areas. Overall, the children in urban areas were 8.1 percentage points less likely to have been subjected to violent discipline in their households than the rural children. Table 8: Percentage points difference among children aged 1–14 years who had been subjected to violent discipline in their households – by type and urban/rural residence (NMICS 2014)

Psychological Any physical Severe physical Any violent aggression punishment punishment discipline Urban households -0.081*** -0.105*** -0.056*** -0.081*** (0.000) (0.000) (0.000) (0.000) Observations 7,554 7,554 7,554 7,554 p-values: * p < 0.1; ** p < 0.05; *** p < 0.01

Figure 8 shows the NMICS 2014 findings on the percentage of children aged 1–14 years who had been subjected to the different types of violent discipline disaggregated by wealth quintile. There was considerably less variation in the child discipline results by wealth quintile across the forms of violence compared to the child labour results. In terms of overall violent discipline, the children in the two richest wealth quintiles were significantly less likely to have been subjected to violent discipline than children in the lowest two quintiles. However, only looking at absolute numbers suggests that the situation of children, even in the richest quintile, is not good as more than 70 per cent of children in all five wealth quintiles had suffered from some form of violent discipline in the month preceding the NMICS. The proportion of children experiencing psychological aggression and some sort of physical punishment

16 followed a similar pattern to the overall violent discipline indicator in terms of variation by wealth quintile. Children in the richest quintile were also statistically less likely to have suffered from severe physical punishment in their households compared to children from the lowest four quintiles.

Figure 8: Percentage of children aged 1–14 years subjected to violent discipline methods in their households – by type and wealth quintile (NMICS 2014)

Children in the wealthiest and second wealthiest quintiles were less likely to have been subjected to violent discipline in their households compared to children in the lowest three wealth quintiles. Finally, Figure 9 shows the percentage of the children aged 1–14 years who were subjected to the three violent discipline methods and the overall measure disaggregated by the caste and ethnicity of the households. Among the four indicators, the percentage of 1–14 year-olds subjected to some sort of physical punishment showed the greatest variation by caste and ethnicity. The children in the hill Brahmin households fared significantly better than children from all the other groups except for Newar children. The hill Dalit household children were significantly more likely to be have been subjected to psychological aggression in their households compared to hill Brahmin, hill Chhetri and Madhesi (non- Brahmin/Chhetri) household children. Overall, children in the hill Brahmin households were statistically less likely to have experienced violent discipline in comparison to all other children except for those in Newar households.

17 Figure 9: Percentage of children aged 1–14 years subjected to violent discipline in their households – by type and caste/ethnicity (NMICS 2014)

The children in the hill Brahmin households were statistically and significantly less likely to have been subjected to physical punishment and overall violent discipline than children in all the other ethnic and caste groups except for children from the Newar households.

3.4 Logistic regression analysis of determinants of child labour and violent child discipline The domains of child labour and violent discipline were further examined using a logistic regression framework with ten individual and household level characteristics used as explanatory variables to test the association between them and variables in the two domains.

The reference categories of the ten characteristics and the four variables for each domain were selected based on the above descriptive analysis and the literature on the subjects. For example, in terms of geography, the Mid-Western Development Region was selected as the reference category as this region performs poorly on child labour and violent child discipline. The hill Brahmins were selected as the reference category for caste and ethnicity as they tend to be the most privileged social group in Nepal. The other characteristics used in this analysis are sex, wealth quintile, urban/rural residence, current school enrolment, whether a child was the eldest child in the household, household size, whether the child’s father lived abroad and age. While the ten variables were selected based on suitability of context and the preceding descriptive analysis, these characteristics are not necessarily a comprehensive list of the factors that affect the two domains. As such, this section should only be treated as further exploratory analysis.

Child labour This section explores ten factors that determine overall child labour and the three aspects (economic activity and household chores worked above the maximum acceptable number of hours and hazardous working conditions), among children aged 5–17 years in the 2014 NMICS sample. A logistic regression framework was used to determine associations with the dependent variables. Reference categories were taken for each characteristic and the odds ratios calculated for the non-reference categories of how much they varied from the reference category on the four determinants of child labour.

18 The results in Table 9 show the factors that are strongly associated with the three aspects of child labour and child labour: • Wealth has a strong and linear relationship with the different child labour indicators. The children in the richest quintile were almost one-tenth as likely as children in the poorest quintile to be engaged in economic activity above the weekly hours limit, and one-fourth as likely to be engaged in household chores above the weekly limit, one-twelfth as likely to be engaged in hazardous working conditions and one-sixth as likely to be engaged in any form of child labour. • The children who were currently attending school were about a half as likely to be engaged in overall child labour as children not attending school and about two-fifths as likely to be engaged in economic activities above the weekly threshold as the children who were not attending school. • The female children were about 1.4 times more likely to be engaged in household chores above the age-specific thresholds and 1.2 times more likely to be engaged in overall child labour than their male counterparts. • The children whose fathers were living abroad (i.e. outside Nepal) were 1.2 times more likely to be engaged in overall child labour than the children whose fathers were not living abroad. • The hill Chhetri children were about 1.2 times more likely than the hill Brahmin children to be engaged in overall child labour and more than twice as likely to be engaged in economic activities above the threshold number of hours. • Compared to the reference category of children in the Mid-Western Development Region, children in the Central Development Region were less likely to be engaged in any form of child labour. Children in the Eastern Development Region were less likely to be engaged in hazardous working condition and overall child labour whereas children in the Far-Western Development Region were less likely to be engaged in economic activity and household chores above the thresholds than children in the Mid-Western region.12 • Finally, the 12–17 year-olds were less likely to be engaged in economic activities, household chores, and overall child labour compared to the 5–11 year-olds, possibly due to the higher weekly hour limits for the older age groups. However, the 12–17 year-olds were four times as likely to be working in hazardous conditions than the younger children.

The female children were about 1.4 times more likely to be engaged in household chores above the acceptable number of hours and 1.2 times more likely to be engaged in overall child labour.

Table 9: Determinants of child labour among children aged 5–17 years using a logistic regression framework (NMICS 2014) Odds ratios

Background characteristics Economic activity Household Hazardous work Overall child chores labour 1. Region Ref. category: Mid-Western Eastern 0.879 0.792 0.642** 0.778* Central 0.366*** 0.453*** 0.514*** 0.592*** Western 0.917 0.594*** 1.028 0.912 Far Western 0.579*** 0.430*** 1.206 0.930

2. Household size Ref. category: Small (1–4) Medium (5–8) 1.029 1.222** 1.189* 1.228*** Large (8 or higher) 0.996 0.965 1.315 0.974 3. Urban/rural

12 See map of the former development regions at Appendix 1.

19 Odds ratios

Background characteristics Economic activity Household Hazardous work Overall child chores labour Ref. category: Rural Urban 0.730* 0.863 0.727* 0.740** 4. Current schooling status Ref. category: Out of school Currently attending school 0.397*** 0.699 0.501*** 0.548***

5. Ethnicity and caste Ref. category: Hill Brahmin Hill Chhetri 2.239*** 1.388** 1.511*** 1.256* Newar 2.588*** 1.043 1.471 1.235 Madhesi Brahmin/Chhetri 0.834 1.080 0.374 0.802 Hill Dalit 1.493* 1.708*** 0.968 1.142 Hill Janajati 2.030*** 1.284 1.195 1.086 Tarai Janajati 1.442 1.208 0.734 1.021 Other 2.463*** 1.572** 1.610 1.771** Muslim 1.349 1.566 0.499** 0.973 Madhesi Dalit 0.930 1.489 1.107 1.440 Madhesi (non-Brahmin/Chhetri) 1.894** 1.412* 1.148 1.323*

6. Wealth quintile Ref. category: Poorest Second 0.663*** 0.659*** 0.609*** 0.638*** Middle 0.401*** 0.392*** 0.378*** 0.370*** Fourth 0.334*** 0.474*** 0.219*** 0.315*** Richest 0.126*** 0.270*** 0.0841*** 0.174*** 7. Father abroad Ref. category: Father not abroad Father abroad 0.943 1.176 1.092 1.230*

8. Eldest child Ref. category: Not eldest child Eldest child 0.893 1.175 1.074 1.117 9. Sex Ref. category: Male Female 1.034 1.456*** 1.100 1.184** 10. Age category Ref. category: 5-11 years 12–14 years 0.377*** 0.0368*** 4.038*** 0.439*** 15–17 years 0.0421*** 0.0138*** 4.954*** 0.448*** Sample size (N) 6,857 6,873 6,873 6,857 p-values: *p<0.1; ** p<0.05; ***p<0.01

Violent child discipline This section explores the factors that determine children being subjected to violent discipline and its components of psychological aggression, physical punishment and severe physical punishment, among children aged 1–14 years in the NMICS sample. A logistic regression framework was applied to 10 household and individual characteristics to determine associations with the dependent variables. Reference categories were taken for each characteristic and the odds ratios calculated.

The results in Table 10 show the level of association of the characteristics with the three types of violence and overall violent discipline faced by the children in the NMICS 2014 sample: • There was a strong relationship between the level of wealth and the different violent child discipline indicators. Compared to every ten children in the poorest quintile, only three in the richest quintile were likely to be subjected to psychological aggression and overall violent discipline. Three-fifths of all the children were likely to have been subjected to at least one form of physical punishment including severe physical punishment and there was no statistical difference in the extent of overall violent discipline methods faced by children in the poorest two quintiles.

20 • The female children were less likely to have been subjected to physical punishment than the male children. Similarly, the eldest children in a household were about 1.3 times more likely to have been subjected to violent discipline than children who were not the eldest. • The children whose fathers were living abroad were 1.4 times as likely to have been subjected to violent discipline in their households and 1.5 times as likely to have been subjected to physical punishment compared to children whose fathers were not living abroad. • The children in hill Janajati households were about 1.7 times more likely than children in hill Brahmin households to have been subjected to violent discipline, while children in the Muslim, Madhesi Dalit and Madhesi households were more than twice as likely to have been subjected to physical punishment, including severe physical punishment, than children in the Muslim households. • The children in the Central and Western regions were more likely to have been subjected to violent discipline than children in the Mid-Western region. However, this result seems to be driven by children in the Central and Western regions being subjected to more psychological aggression than children in the Mid-West. Indeed, children in both the Central and Western regions were about half as likely to have been subjected to physical punishment, including severe physical punishment, than the children in the Mid-West. • Finally, and most worryingly, the 12–14 year-olds were universally less likely to have been subjected to any form of violent discipline than infants aged 1–4 years! This finding is the most worrying and needs bringing to the attention of concerned policymakers along with the fact that the incidence of violent aggression faced by children in all the age groups is very high. Table 10: Determinants of violent child discipline among children aged 1–14 years using a logistic regression framework (NMICS 2014) Odds ratios

Background characteristics Psychological Any physical Severe physical Overall child aggression punishment punishment discipline 1. Region Ref. category: Mid-Western Eastern 1.218 0.541*** 0.804 1.110 Central 1.427** 0.595*** 0.458*** 1.456** Western 1.671*** 0.617*** 0.586** 1.598** Far Western 1.050 0.741** 1.371* 1.048

2. Household size Ref. category: Small (1-4) Medium (5-8) 0.811** 0.928 1.091 0.846 Large (8 or higher) 0.924 1.027 1.433 0.889 3. Urban/rural Ref. category: Rural Urban 0.964 0.932 0.927 0.947 4. Current schooling status Ref. category: Out of school Currently attending school 1.480 1.438 1.773 1.836*

5. Ethnicity and caste Ref. category: Hill Brahmin Hill Chhetri 1.185 1.486*** 1.139 1.341 Newar 1.085 1.929** 1.453 1.097 Madhesi Brahmin/Chhetri 0.574 1.799 0.741 0.577 Hill Dalit 1.543** 1.473** 1.344 1.573* Hill Janajati 1.650*** 1.919*** 1.400 1.763*** Tarai Janajati 1.071 1.622** 1.179 1.195 Other 0.738 2.101*** 2.560** 1.179 Muslim 1.845 2.030*** 2.496** 1.683 Madhesi Dalit 1.140 4.093*** 2.632*** 1.790 Madhesi (non-Brahmin/Chhetri) 1.483* 2.322*** 2.696*** 1.663**

6. Wealth quintile Ref. category: Poorest Second 0.798 0.914 1.063 0.781

21 Odds ratios

Background characteristics Psychological Any physical Severe physical Overall child aggression punishment punishment discipline Middle 0.792 0.664*** 0.733 0.710* Fourth 0.426*** 0.626*** 0.634** 0.410*** Richest 0.321*** 0.589*** 0.598* 0.300*** 7. Father abroad Ref. category: Father not abroad Father abroad 1.256 1.525*** 1.346* 1.402**

8. Eldest child Ref. category: Not eldest child Eldest child 1.137 1.276*** 0.991 1.278** 9. Sex Ref. category: Male Female 0.868 0.797*** 0.818 0.843 10. Age category

Ref. category: 1–4 years 5–11 years 1.053 1.648*** 1.266 1.250 12–14 years 0.687** 0.669** 0.676* 0.706** Sample size (N) 5,127 5,127 5,127 5,127 *p<0.1; ** p<0.05; ***p<0.01

Finally, and most worryingly, the 12–14 year-olds were universally less likely to have been subjected to any form of violent discipline than infants aged 1–4 years. This finding is the most worrying and needs bringing to the attention of concerned policymakers along with the fact that the incidence of violent aggression faced by children in all the age groups is very high.

22 4 DISCUSSION AND CONCLUSIONS This further analysis report highlights several findings on child labour and child discipline in Nepal. Overall 37.4 per cent of the children aged 5–17 years were engaged in child labour in the NMICS 2014 sample. The rate of violence against children aged 1–14 years, as measured by violent discipline suffered by them in their households, was high with 81.7 per cent of them having experienced at least one form of violent discipline from adults members in the month preceding the survey.

This further analysis presents a basic profile of children aged 5–17 years who were classified as child labourers in the NMICS sample. Most of them could read and write and had attended school for at least one year. However, they were significantly less likely to be currently attending school compared to the children who were not engaged in child labour, especially after the age of 13. In terms of family composition, there was no statistically significant difference by child labour status between children aged 5–17 years who had their mothers or fathers living abroad.

The topics of child labour and child discipline were further analysed by disaggregating data on the two domains by age category, sex, wealth quintile and caste and ethnicity. On child labour, the 5–11 year- olds were more likely to be engaged in economic activities above the age-specific threshold than children in the older cohort (12–17 years), whereas children in the older cohort were more likely to be engaged in hazardous work. These findings were corroborated by results from the logistic regressions that controlled for socioeconomic and individual characteristics.

Additionally, female children were more likely to be engaged in overall child labour, primarily because of their engagement in household chores above the acceptable number of hours. Wealth has a strong and linear relationship with the child labour-related variables. A jump from a poorer to a richer wealth quintile, including in the lower wealth quintiles, produced a highly significant drop in the likelihood of children aged 5–17 years being engaged in child labour. In terms of caste and ethnicity, children in hill Chhetri households were significantly more likely to be engaged in all forms of child labour. And finally, the children whose fathers were living abroad were slightly more likely to be engaged in child labour after controlling for household and individual characteristics.

On child discipline, the results from the descriptive analysis and the logistic regressions found that children in the 12–14 year-olds age group were universally less likely to be subject to physical punishment than 1–4 year-olds. This finding should be worrying for concerned policymakers as the absolute incidence of physical punishment faced by children of all age-groups is already very high in Nepal. On other indicators, the analysis found that girls were less likely to be subject to physical punishment than boys, but not other forms of violent discipline; and children whose fathers were living abroad were more likely to have been subject to physical punishment by members of their households in the month preceding the survey that children whose fathers were not abroad.

Wealth had a strong relationship with the likelihood of children being subject to violent discipline. However, the relationship was not linear as it is for child labour. Children in the richest and second richest wealth quintiles were much less likely to have been subjected to violent discipline, and especially severe physical punishment, than children in the poorest quintile. Finally, children in Madhes-based households including Madhesis, Madhesi and Muslims were more than twice as likely to have been subjected to physical punishment than children in hill Brahmin households.

These findings help fill the large information gaps on child labour and child discipline in the Nepali context. Policymakers should identify which are the most marginalized group of children and focus on policies and interventions to prevent exploitation and violence against such children. Even though Nepal

23 has enacted laws and regulations to combat violence against children and their exploitation, a lack of proper understanding of contextual elements such as children being expected to carry out household chores and the widespread psychological aggression that is also socially acceptable and goes unreported, may have, until now, hampered the ability to instigate policy measures against child labour and violent child discipline. This further analysis fills in a number of information gaps by presenting detailed disaggregated findings on the two domains. It is hoped that policymakers and practitioners will use the findings to reduce the incidence of child labour and the violent discipline of Nepal’s children.

24 References Central Bureau of Statistics, Nepal Multiple Indicator Cluster Survey 2014: Key findings report, CBS and UNICEF, Kathmandu, Nepal, 2014, , accessed 2 March 2018. Edmonds, Eric V, Defining Child Labour: A review of the definitions of child labour in policy research, Working Paper, Geneva, International Programme on the Elimination of Child Labour (IPEC), Geneva, 2008, accessed March 2018. International Labour Organization, South Asia – Fact Sheet: Children in Labour and Employment, ILO, New Delhi, n.d. , accessed 25 March 2017. International Labour Organization, Marking Progress against Child Labour: Global estimates and trends 2000–2012, International Labour Office, IPEC, Geneva, 2013, , accessed March 2017. Sherin, K. and Scott Lyon, Measuring Children’s Work in South Asia: Perspectives from national household surveys, International Labour Organization for South Asia, ILO Country Office, New Delhi, 2015, , accessed 25 March 2017. The Government of Nepal, Interim Constitution of Nepal, 2007, , accessed 25 March 2017. United Nations Children’s Fund, ‘Convention on the Rights of the Child’, , accessed March 2018. United Nations Children’s Fund, ‘UNICEF Data: Monitoring the Situation of Women and Children’, 2017, , accessed 25 March 2017. United Nations Treaty Collection, ‘Chapter IV: Human Rights, 11 – Convention on the Rights of the Child, New York, 1989’, , accessed 6 November 2016.

25 Appendix 1: Map of the NMICS ecozones and Nepal’s former development regions

26 Appendix 2: Additional maps showing results of the NMICS 2014 further analysis

1. Child labour Figure A1: Economic activity among children aged 5–17 years, NMICS 2014 sampling cluster

Figure A2: Household chores carried out by children aged 5–17 years, NMICS 2014 sampling cluster

27 Figure A3: Hazardous working condition among children aged 5–17 years, NMICS 2014 sampling cluster

2. Child discipline Figure A4: Psychological aggression suffered by children aged 1–14 years, NMICS 2014 sampling cluster

28 Figure A5: Physical punishment suffered by children aged 1–14 years, NMICS 2014 sampling cluster

Figure A6: Severe physical punishment suffered by children aged 1–14 years, NMICS 2014 sampling clusters

29