Shadow Wages and Child Labour Supply in Agricultural ... Wages and Child Labour Supply in Agricultural Households in Nepal DPhil Conference University of Sussex, Brighton Marinella

Shadow Wages and Child Labour Supply in AgriculturalHouseholds in Nepal

DPhil ConferenceUniversity of Sussex, Brighton

Marinella Leone

12 November 2010

Shadow Wages and Child Labour Supply in AgriculturalHouseholds in Nepal

1 Aim and Motivation

2 Literature review

3 Theoretical Framework

4 Empirical Strategy

5 Empirical Analysis

6 Conclusion

Marinella Leone () Shadow Wages and Child Labour Supply in Agricultural Households in Nepal12 November 2010 2 / 22

Aim and Motivation

Aim

Analysis of child labour in the agricultural sector.

Estimate the monetary contribution of child labour to family farms:shadow wage.

Assess whether children work on the farm because they are propelledto do so by poverty or for other reasons: income or substitutioneffect?

Analysis conducted on agricultural households in Nepal using NepalLiving Standard Measurement Survey (NLSS 2003/04).


Aim and Motivation

Motivation

Limited attention to child labour in agriculture, though 132 millionchildren 5-14 years old around the world work in agriculture (70% oftotal child labour, ILO).

Child labour in Nepal is mostly prevalent in the agricultural sector,mainly as self-employed on households’s farm.

Agriculture in Nepal accounts for 40% of GDP and 80% of populationwork in some agricultural activities.

Understand precise role of poverty: recent evidence shows thatchildren work on family farms not only for poverty reasons but, underspecific conditions, due to existence of market imperfections.

Knowledge of whether and how children contribute to family incomeand assessing how their supply reacts to changing economicconditions becomes particularly relevant from a policy perspective.


Literature review

Literature review

Mixed evidence on the role of poverty in child labour (Basu and Van,1998 ⇒ seminal paper; Patrinos and Psacharopoulos, 1997; Ray,2000; Sedlacek et al., 2005).

More recent studies focused on identifying demand/supply factors andincome/substitution effects (Kambambhati and Rajan, 2006;Edmonds, 2007; Kruger et al, 2007; Bhalotra, 2007).

Child labour in agricultural households (Mueller, 1984; Bhalotra andHeady, 2003 ⇒ wealth paradox).

Test of poverty hypothesis on farm child labour (Dumas, 2007; Basuet al., 2010) ⇒ reduced form models.

Shadow wages and labour supply estimated only for adults (Jacoby,1993; Skoufias, 1994; Barrett et al., 2008).

Attempt of estimation of child shadow wage (Menon et al., 2006 ⇒dual side).


Theoretical Framework


General agricultural household model (Jacoby, 1993).

Farm household solve following maximisation problem:

max U(C , li ;Z) (1)

subject to

C = pΘ(Fi ,H,S;A)− whH + wiMi + V (2)

Ti = li + Mi + Fi

0 ≤ Mi ≤ Mi (3)




From FOCs we get:

∂U/∂li∂U/∂C

= wi −µ1iλ

+µ2iλ

= w∗i (4)

p∂Θ

∂Fi= wi −

µ1iλ

+µ2iλ

= w∗i (5)

p∂Θ

∂H= wh (6)

where w∗ is the shadow wage.




If 0 ≤ Mi ≤ Mi then µ1i = 0, µ2i = 0 and wi = w∗i ⇒ model is

separable.

If either Mi = 0 or Mi = Mi ⇒ model is non-separable:

∂U/∂li∂U/∂C

= wi +µ2iλ

= p∂Θ

∂Fi= w∗

i > wi if Mi = 0 (7)

∂U/∂li∂U/∂C

= wi −µ1iλ

= p∂Θ

∂Fi= w∗

i < wi if Mi = Mi (8)




In presence of labour constraint, budget constraint is non-linear;Replace it with a non linear one: at the optimum the slope of thebudget constraint is the marginal product of family farm labour whichis equal to the shadow wage.

The household maximisation problem under the linear budgetconstraint becomes:

max U(C , li ;Z) s.t. C + w∗i li = V ∗ (9)

where V ∗ = maxH,Fi

{pΘ(Fi ,H,S;A)} − whH − w∗i Fi + V + w∗

i T (10)




The solution to this problem yields Marshallian demand functions forleisure and hence the corresponding labour supply functions for eachmember of the household i is:

Ti − li = hi = Fi + Mi = hi (w∗i ,w

∗j ,V

∗;Z) (11)

Exploit duality theory and equate Hicksian and Marshallian laboursupply equations.

Differentiate with respect to shadow wages and obtain Slutskyequations:

∂hi (w∗i ,V

∗)

∂w∗j

=∂hi (w

∗i , u)

∂w∗j

+ hj∂hi (w

∗i ,V

∗)

∂V ∗ (12)


Empirical Strategy

Empirical Strategy I : the production function

Estimate an agricultural production function at household level andobtain marginal product of labour inputs on the farm.

Sample of households that own and operate some land and in whichat least one adult family member works on the farm.

Estimate the following general functional form:

Y = f (Fa,Fc ,Hh,Hx ,S,A;β; ε) (13)

Calculate marginal products of family adult and child labour. Theshadow wages are then used to estimate shadow income.

Assume heterogeneous labour inputs but some variable inputs canassume zero value (e.g. child labour) as not all inputs are used on alloutputs.

Endogeneity of variable inputs: IV approach.

Selection bias in sample used.


Empirical Strategy

Empirical Strategy II: the labour supply

Estimate child labour supply of the following general functional form:

hi = g(w∗c , w

∗a , V

∗,Z ,Xi ) (14)

Compute income and substitution effects.

Endogeneity of shadow wages and shadow income (given thenon-separable nature): IV approach.

Hours of work of children censored at zero: Tobit and Heckmanspecifications.


Empirical Analysis

Empirical Analysis: the production functionData Description

Data: Nepal Living Standard Measurement Survey 2003/04 (NLSSII).

Sample: 2404 households (61% of surveyed hh).

Among these the 45% have children 5-14 years old working in someactivities.

Over one-third of farm households in the sample use their children onthe farm.


Empirical Analysis

Empirical Analysis: the production functionData Description

Description of production function variables Mean Standard deviation

Proportion of input

use

Value of crop and livestock production 36123.2 44943.6

Hectares of land owned or cultivated for someone else 0.9 1.1

Hours worked on farm by family adult (15+) 3437.7 2439.0

Hours worked on farm by family children (5-14) 324.9 686.5 38%

Hours worked on farm by hired labour (man-days * 8 hours) 131.9 422.5 40%

Hours worked on farm by exchange labour (man-days * 8 hours) 132.4 220.5 64%

Expenditures on seeds 264.6 1186.3 41%

Expenditures on fertilisers 1342.8 2718.5 67%

Value of farm equipment 4157.5 28564.5 98%

Other expenditures (includes livestock expenditures) 1769.0 5121.3 81%

Share of land irrigated 0.445 0.403

Share land sharecrop out (dry) 0.034 0.150

Share land sharecrop out (wet) 0.039 0.160

Share land fallow (dry) 0.148 0.280

Share land fallow (wet) 0.019 0.095

Share land sharecropped in 0.175 0.315

Rural West Mountains/Hills 0.290 0.454

Rural East Mountains/Hills 0.307 0.461

Rural West Tarai 0.150 0.357

Rural East Tarai 0.253 0.435


Empirical Analysis

Empirical Analysis: the production functionEstimates

(1) (2) (3) Cobb-Douglas(I) Cobb-Douglas (II) Translog (II) L Land hectares 0.4043*** 0.3649*** 0.5635*** (0.0187) (0.0186) (0.1432) L Adult labour 0.2752*** 0.2604*** -0.0904 (0.0208) (0.0200) (0.1723) L Child labour 0.0102** 0.0458** 0.0110** (0.0037) (0.0157) (0.0035) L Hired labour 0.0466*** 0.1031*** 0.1569* (0.0056) (0.0178) (0.0652) L Exchange labour 0.0152** 0.0246 0.0897 (0.0055) (0.0183) (0.0670) Observations 2404 2404 2404 r2 0.6991 0.7186 0.7351


Empirical Analysis

Empirical Analysis: the production functionEstimates - Shadow wages

Marginal products Estimates

Cobb-Douglas (I) Cobb-Douglas (II) Translog (II)

N Mean sd Mean sd Mean sd

Adult 2404 3.19 3.35 3.09 3.66 3.41 7.49

Children 921 1.44 3.28 6.24 13.68 1.57 3.67


Empirical Analysis

Empirical Analysis: the production functionInstruments

Adult and child farm labour : number of household members 5-10,11-15, 16-17, 18-59 in the household;

Hired and exchange labour : village level variables ⇒ dummy forvillage temporary migration; dummy for households that hire labour inthe village; number of landless households in the village that providelabour on other farms; daily market wage at village level for adultsand children;

Seeds and fertilisers expenditures: market prices of seeds andfertilisers at village level; input sold on the market dummy; distanceto closest market for agricultural input purchases.

Instruments are weak and often violation of exlusion restrictions.


Empirical Analysis

Empirical Analysis: the labour supplyData Description

Sample: Children 5-14 years old living in sample of farm households→ 1494 children.

Among these the 50% have worked in the past 12 months in someactivity.

Among these the 74% work on their family farm, therefore hours ofwork censored at zero for children that do not work.


Empirical Analysis

Empirical Analysis: the labour supplyData Description

Variable description mean sd

Total hours worked by children 5-14 in past 12 months 792.3 776.5

Est shadow wage children (5-14) 1.185 2.805

Est shadow wage adults(15-99) 2.944 2.122

Est shadow income 48674.6 63144.9

Sex child - Male=1 0.484 0.500

Age child 10.975 2.290

Rural East Mountains/Hills 0.379 0.485

Rural West Tarai 0.137 0.343

Rural East Tarai 0.221 0.415

Upper caste/ethn 0.454 0.498

Lower caste/ethn 0.509 0.500

head has primary education 0.203 0.403

head has secondary education 0.134 0.341

head age 43.95 11.11

head has chronic illness 0.108 0.310

head migrated 0.361 0.480

# hh members age 0-4 0.732 0.867

# hh members age 60+ 0.321 0.589

# adult working off-farm 5.159 2.449

subsistence household (crop and livestock sales=0) 0.462 0.499

number of observations 1494


Empirical Analysis

Empirical Analysis: the labour supplyEstimates

Dep var: (1) (2) (3) Annual hours worked From CB(I) From CB(II) From TL(II) Child shadow wage -0.6007*** -0.6122*** -0.6064*** (0.0208) (0.0209) (0.0206) Adult shadow wage 0.1008* 0.1271** 0.1005* (0.0426) (0.0417) (0.0420) Hh shadow income 0.1470*** 0.1575*** 0.1595*** (0.0258) (0.0251) (0.0265) Observations 1494 1497 1497 r2 0.5086 0.5049 0.5089


Empirical Analysis

Empirical Analysis: the labour supplyInstruments

Child shadow wage: children market wage at village level (inputingmissing wages); number of schools in the village;

Adult shadow wage: adult market wage at village level; dummy forwhether household uses hired or exchange labour on the farm;

Household shadow income: dummy for electricity and water in thevillage; dummy for household with a loan; market price of rice atvillage level.


Conclusion

Conclusion

Signs on wage and income elasticities not consistent with the theory.Income and substitution effects have opposite signs from whatexpected. New finding or weakness in estimation procedure?

Given the complexities behind a household’s decision making processand the underlying driving forces, reconsider the appropriateness ofthe empirical methodology.

Is there a more robust way to establish the monetary value of childlabour in farm households?


Shadow Wages and Child Labour Supply in Agricultural ... Wages and Child Labour Supply in Agricultural Households in Nepal DPhil Conference University of Sussex, Brighton Marinella

Documents