Child labour and youth employment as a response to household vulnerability: evidence from rural Ethiopia
Introduction Growing literature of the effect of household vulnerability on children’s work and youth employment; Idiosyncratic shocks and natural disasters apparently lead households to use children as a risk copying instruments There is robust evidence that shocks do in fact matter for housheold decision concerning children’s work and education; But shocks experienced by household can take a variety of forms and their consequences may depend on their specific nature; As a result, the policies required to help cope with risk might also vary depending on the type of shock;
Data and variable definition
The Ethiopia Rural Household Survey (ERHS) is a longitudinal household data set covering households in a number of villages in rural Ethiopia. Data collection started in 1989; In 1994, the survey was expanded to cover 15 villages across the country. An additional round was conducted in late 1994, with further rounds in 1995, 1997, 1999, 2004, and In addition, nine new villages were selected giving a sample of 1477 households We use the 2004 and 2009 round The EHRS round 2004 and 2009 collectes informationon children involvememnt in employment starting from the age of 5 years Data and variable definition
The two rounds of the Ethiopia Rural Household Survey (ERHS) collect also information on occurence of shocks during the 5 years prior to the survey; Children’s work appears to be substancially higher for children belonging to household hit by a shock; Data and variable definition Percentage of children (5-14) in employment, belonging to household experiencing shocks by type of shock, and year Year 2004Year 2009 Type of shockNoYesNoYes Natural disaster Economic Other Lack demand/input Note: Natural disaster (drought, pest-desease on crops, pest or desease on livestock); Economic shocks (input price increase, output price increase=; Other (land redistribution in PA, confiscation of assets); Lack demand input (lack of demand of agricultural products, lack of access to inputs). Source: Author’s calculations based on Ethiopia ERHS
Percentage of children (5-14) attending school, belonging to household experiencing shocks by type of shock and year Year 2004Year 2009 Type of shockNoYesNoYes Natural disaster Economic Other Lack demand/input Data and variable definition On the contrary, the effect of shocks on children’s school attendance is not well defined; Note: Natural disaster (drought, pest-desease on crops, pest or desease on livestock); Economic shocks (input price increase, output price increase=; Other (land redistribution in PA, confiscation of assets); Lack demand input (lack of demand of agricultural products, lack of access to inputs). Source: Author’s calculations based on Ethiopia ERHS
Percentage of youth (15-21) in employment, belonging to household experiencing shocks by type of shock, and year Year 2004Year 2009 Type of shockNoYesNoYes Natural disaster Economic Other Lack demand/input Percentage of youth(15-21) attending school, belonging to household experiencing shocks by type of shock and year Year 2004Year 2009 Type of shockNoYesNoYes Natural disaster Economic Other Lack demand/input Source: Author’s calculations based on Ethiopia ERHS Effect of shocks on youth employment and school attendance are also not well defined;
Children’s work and school attendance in rural Ethiopia
Children’s work and school attendance in Ethiopia Source: Author’s calculations based on Ethiopia ERHS Involvement in economic activity of Ethiopian children remain one of the highest in Africa region Child activity status (age 5-14), by year Activity status MaleFemaleTotalMaleFemaleTotal Employment only School only Employment and school Neither Total Employment Total schooling
Employment rate, by age and years Employment rate Source: Author’s calculations based on Ethiopia ERHS
School attendance rate, by age and years School attendance rate Source: Author’s calculations based on Ethiopia ERHS
Theoretical Model
Optimal labour supply and consumption: perfect capital markets
Child Labour supply: imperfect capital markets
Consumption: imperfect capital markets
Elasticity of child labour supply First best solution Borrowing constraints: no corner solution for child labour supply Borrowing constraints: corner solution for child labour supply Subjective expectations of income risks : 00 Adverse realization of exogenous income shocks: 00
Econometric analysis Preliminary Results
Two approaches to assess the impact of shocks on household behaviour Non-Linear model : by regressing the outcome variable “employment” at time t on the employment at time (t-1), a set of individual and household characteristics at time (t), shocks experienced by the household; Non-Linear model with IV Using past shocks and individual and household characteristics as instruments
(1)(2) Variablesemployment (t) Employment (t-1)0.564*** (7.86)(7.85) Shocks drought0.185**0.187** (2.23)(2.25) pest or desease on crop0.281***0.282*** (3.23)(3.24) Lack of access to inputs (0.63)(0.62) input price increase0.141** (1.99) output price increase (-0.88) lack demand agricultural product (-0.64) land redistribution in PA-0.574**-0.569** (-2.01)(-2.00) confiscation of assets (-0.22)(-0.23) pest or desease on livestock (-1.25)(-1.26) dummy: zero per capita consumption in Kcal (cereals) (0.88)(0.89) Log per capita consumption in Kcal (cereals) (0.39)(0.40) variance ratio deficiency0.541** (2.04) variance per capita consumption in Kcal (cereals) ** (2.06) Constant (0.13) Obs. 1,732; z-statistics in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Regression analysis on employment at time t, without instrumental variable Source: Author’s calculations based on Ethiopia ERHS
(1)(2) Variablesemployment (t) Employment (t-1) (-0.83)(-0.81) Shocks drought 0.187**0.189** (2.40)(2.41) pest or desease on crop 0.231*** (2.72) Lack of access to inputs (0.65)(0.63) input price increase 0.153**0.155** (2.30)(2.33) output price increase (-0.82)(-0.83) lack demand agricultural product (-0.64)(-0.65) land redistribution in PA *-0.507* (-1.88)(-1.87) confiscation of assets (-0.23)(-0.24) pest or desease on livestock (0.38)(0.37) dummy: zero per capita consumption in Kcal (cereals) (0.81)(0.83) Log per capita consumption in Kcal (cereals) (0.43)(0.45) variance ratio deficiency 0.515** (2.07) variance per capita consumption in Kcal (cereals) ** (2.08) Constant (-0.88) Obs. 1,732; z-statistics in parentheses; *** p<0.01, ** p<0.05, * p<0.1 IV Regression analysis on employment at time t Source: Author’s calculations based on Ethiopia ERHS
(1)(2) VariablesSchool attendance(t) School attendance (t-1) 0.580*** (6.84) Shocks drought (0.80)(0.76) pest or desease on crop (-0.42)(-0.51) Lack of access to inputs 0.411*** (3.02) input price increase 0.154**0.155** (2.03)(2.04) output price increase (-0.50)(-0.53) lack demand agricultural product * (-1.69) land redistribution in PA (-0.22) confiscation of assets (-1.04)(-1.03) pest or desease on livestock * (1.63)(1.65) dummy: zero per capita consumption in Kcal (cereals) (0.02)(0.08) Log per capita consumption in Kcal (cereals) (0.46)(0.54) variance ratio deficiency (1.09) variance per capita consumption in Kcal (cereals) (0.67) Constant (-1.19)(-1.20) Obs. 1,675; z-statistics in parentheses; *** p<0.01, ** p<0.05, * p<0.1 Regression analysis on school attendance at time t, without instrumental variable Source: Author’s calculations based on Ethiopia ERHS
(1)(2) VariablesSchool attendance(t) School attendance (t-1) (1.21)(1.11) Shocks drought (1.23)(1.18) pest or desease on crop (-0.50)(-0.59) Lack of access to inputs 0.407*** (2.99)(3.00) input price increase 0.164**0.165** (2.15) output price increase (-0.60)(-0.63) lack demand agricultural product * (-1.83) land redistribution in PA (-0.22) confiscation of assets (-1.04)(-1.03) pest or desease on livestock (-1.49)(-1.48) dummy: zero per capita consumption in Kcal (cereals) (0.19)(0.26) Log per capita consumption in Kcal (cereals) (0.66)(0.73) variance ratio deficiency (1.25) variance per capita consumption in Kcal (cereals) (0.83) Constant (-1.32)(-1.38) Obs. 1,675; z-statistics in parentheses; *** p<0.01, ** p<0.05, * p<0.1 IV Regression analysis on school attendance at time t Source: Author’s calculations based on Ethiopia ERHS