Recovery from stunting in early childhood and subsequent schooling outcomes: Evidence from NIDS Waves 1–5

Figures & data

Table 1. Sample of children with non-missing data on HAZ at t₁ and t₂ who were re-interviewed in Wave 5.

	W1 2008	W2 2010/11	W3 2012	W4 2014/15	W5 2017	N
2 year-olds in W1	t1 = 2y	t2 = 4/5y			t3 = 10/11/12y	211
2 year-olds in W2		t1 = 2y	t2 = 4/5y		t3 = 8/9/10y	177
2 year-olds in W3			t1 = 2y	t2 = 4/5y	t3 = 6/7/8y	452
Total						840

Source: Own calculations from NIDS Waves 1–5.

Table 2. Summary statistics.

Variable	N	T	Mean (Std. Dev.) or %
Outcome variables
No. of grades completed	838	t3	2.01 (1.78)
Old for grade	837	t3	15.5%
Correct age for grade	837	t3	55.8%
Young for grade	837	t3	28.7%
Enrolled in Gr 1 or higher	839	t3	97.4%
Age first enrolled	763	t3	5.64 (0.82)
Failed in 2016	827	t3	5.6%
Stunting status
Not stunted in t1 and t2	840	t1 and t2	61.7%
Stunted in t1 and t2`	840	t1 and t2	12.0%
Catch up by t2	840	t1 and t2	19.3%
–Incomplete catch up (HAZ < −1)	840	t1 and t2	13.1%
–Complete catch up (HAZ ≥ −1)	840	t1 and t2	6.2%
Late incident stunted	840	t1 and t2	7.0%
Control variables
Age 6 (omitted)	840	t3	13.2%
Age 7	840	t3	38.8%
Age 8	840	t3	6.9%
Age 9	840	t3	15.8%
Age 10	840	t3	1.4%
Age 11	840	t3	19.2%
Age 12	840	t3	4.6%
W1 sample (omitted)	840	t1	25.1%
W2 sample	840	t1	21.1%
W3 sample	840	t1	53.8%
Age in months at t1	840	t1	30.36 (3.37)
Age in months at t2	840	t2	59.02 (4.94)
African	840	t3	88.1%
Female	840	t3	53.2%
Ln (per capita hh income)	840	t3	6.78 (0.81)
Grant received for index child	837	t3	85.0%
No. of children under 15y in hh	840	t3	3.31 (2.20)
Mom deceased	837	t3	5.6%
Mom schooling (years)	786	t3	10.16 (2.91)
Mom schooling missing	840	t3	6.4%
Urban	840	t3	42.1%
Western Cape (omitted)	840	t3	7.3%
Eastern Cape	840	t3	11.4%
Northern Cape	840	t3	6.2%
Free State	840	t3	5.6%
KwaZulu-Natal	840	t3	38.5%
Northwest	840	t3	6.3%
Gauteng	840	t3	8.1%
Mpumalanga	840	t3	6.6%
Limpopo	840	t3	10.1%

Source: Own calculations from NIDS Waves 1–5.

Note: The sample consists of children with non-missing data on HAZ at t₁ and t₂ who were re-interviewed in W5.

Table 3. Regression results using recovery from stunting definition of catch-up growth (OLS coefficients).

	I	II	III	IV	V
	No. of grades completed	Old for grade	Enrolled in Gr 1 or higher	Age first enrolled	Failed in 2016
Stunted in t1 and t2`	−0.279***	0.124***	−0.020	0.133	0.073***
	(0.071)	(0.038)	(0.018)	(0.094)	(0.025)
Catch up by t2	−0.171***	0.079***	−0.026*	0.099	0.054***
	(0.058)	(0.031)	(0.015)	(0.077)	(0.021)
Late incident stunted	−0.054	0.059	−0.020	0.134	0.007
	(0.087)	(0.046)	(0.022)	(0.115)	(0.031)
F test: Stunted in t1 and t2 = Catch up by t2	1.80	1.12	0.10	0.10	0.42
Prob > F	0.18	0.29	0.75	0.75	0.52
N	834	833	835	761	824

Source: Own calculations from NIDS Waves 1–5.

Notes: Standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.10. Omitted category is not stunted at either t₁ or t₂. Controls included: the child’s gender, race, age (in months), whether the mother was alive/deceased, mother’s schooling, whether mother’s schooling was missing, the log of per capita household income, whether a grant was received for the child, the number of children younger than 15 years in the household, urban/rural residence, province of residence, and wave and age fixed effects.

Table 4. Regression results using recovery from stunting definition of catch-up growth, excluding children aged 6 in Wave 5 (OLS coefficients).

	I	II	III	IV	V
	No. of grades completed	Old for grade	Enrolled in Gr 1 or higher	Age first enrolled	Failed in 2016
Stunted in t1 and t2`	−0.301***	0.142***	−0.015	0.128	0.088***
	(0.078)	(0.043)	(0.020)	(0.103)	(0.027)
Catch up by t2	−0.185***	0.090**	−0.033**	0.090	0.032
	(0.064)	(0.035)	(0.017)	(0.086)	(0.022)
Late incident stunted	−0.058	0.074	−0.022	0.131	−0.002
	(0.096)	(0.053)	(0.025)	(0.127)	(0.033)
F test: Stunted in t1 and t2 = Catch up by t2	1.70	1.11	0.60	0.10	3.43
Prob > F	0.19	0.29	0.44	0.75	0.06*
N	723	722	724	652	714

Source: Own calculations from NIDS Waves 1–5.

Notes: Standard errors in parentheses. ***p < 0.01, **p < 0.05, * p < 0.10. Omitted category is not stunted at either t₁ or t₂. Controls included: the child’s gender, race, age (in months), whether the mother was alive/deceased, mother’s schooling, whether mother’s schooling was missing, the log of per capita household income, whether a grant was received for the child, the number of children younger than 15 years in the household, urban/rural residence, province of residence, and wave and age fixed effects.

Table 5. Regression results using stricter definition of catch-up growth (HAZ ≥ −1 in t₂) (OLS coefficients)

	I	II	III	IV	V
	No. of grades completed	Old for grade	Enrolled in Gr 1 or higher	Age first enrolled	Failed in 2016
Stunted in t1 and t2`	−0.280***	0.125***	−0.020	0.133	0.074***
	(0.071)	(0.038)	(0.018)	(0.094)	(0.025)
Incomplete catch up (HAZ < −1)	−0.240***	0.115***	−0.029*	0.110	0.096***
	(0.068)	(0.036)	(0.017)	(0.089)	(0.024)
Complete catch up (HAZ ≥ −1)	−0.031	0.008	−0.021	0.073	−0.030
	(0.092)	(0.049)	(0.024)	(0.129)	(0.033)
Late incident stunted	−0.054	0.059	−0.020	0.134	0.008
	(0.087)	(0.046)	(0.022)	(0.115)	(0.031)
F test: Stunted in t1 and t2 = Incomplete catch up	0.22	0.05	0.16	0.04	0.52
Prob > F	0.64	0.82	0.69	0.84	0.47
F test: Complete catch up = Incomplete catch up	3.77	3.50	0.08	0.06	10.94
Prob > F	0.05**	0.06*	0.78	0.80	0.001***
N	834	833	835	761	824

Source: Own calculations from NIDS Waves 1–5.

Notes: Standard errors in parentheses. ***p < 0.01, **p < 0.05, * p < 0.10. Omitted category is not stunted at either t₁ or t₂. Controls included: the child’s gender, race, age (in months), whether the mother was alive/deceased, mother’s schooling, whether mother’s schooling was missing, the log of per capita household income, whether a grant was received for the child, the number of children younger than 15 years in the household, urban/rural residence, province of residence, and wave and age fixed effects.

Table 6. Regression results using stricter definition of catch-up growth (HAZ ≥ −1 in t₂), excluding children aged 6 in Wave 5 (OLS coefficients),

	I	II	III	IV	V
	No. of grades completed	Old for grade	Enrolled in Gr 1 or higher	Age first enrolled	Failed in 2016
Stunted in t1 and t2`	−0.301***	0.142***	−0.015	0.128	0.088***
	(0.078)	(0.043)	(0.020)	(0.103)	(0.026)
Incomplete catch up (HAZ < −1)	−0.234***	0.127***	−0.035*	0.070	0.070***
	(0.074)	(0.041)	(0.019)	(0.098)	(0.025)
Complete catch up (HAZ ≥ −1)	−0.081	0.012	−0.028	0.142	−0.049
	(0.104)	(0.057)	(0.027)	(0.148)	(0.035)
Late incident stunted	−0.058	0.074	−0.022	0.131	−0.002
	(0.096)	(0.053)	(0.025)	(0.127)	(0.033)
F test: Stunted in t1 and t2 = Incomplete catch up	0.48	0.08	0.65	0.21	0.31
Prob > F	0.49	0.78	0.42	0.65	0.58
F test: Complete catch up = Incomplete catch up	1.62	2.98	0.06	0.18	8.38
Prob > F	0.20	0.08*	0.81	0.67	0.004***
N	723	722	724	652	714

Source: Own calculations from NIDS Waves 1–5.

Notes: Standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.10. Omitted category is not stunted at either t₁ or t₂. Controls included: the child’s gender, race, age (in months), whether the mother was alive/deceased, mother’s schooling, whether mother’s schooling was missing, the log of per capita household income, whether a grant was received for the child, the number of children younger than 15 years in the household, urban/rural residence, province of residence, and wave and age fixed effects.

Table A1. Full estimation results with controls, using recovery from stunting definition of catch-up growth (OLS coefficients).

	I	II	III	IV	V
	No. of grades completed	Old for grade	Enrolled in Gr 1 or higher	Age first enrolled	Failed in 2016
Stunted in t1 and t2`	−0.279***	0.124***	−0.020	0.133	0.073***
	(0.071)	(0.038)	(0.018)	(0.094)	(0.025)
Catch up by t2	−0.171***	0.079***	−0.026*	0.099	0.054***
	(0.058)	(0.031)	(0.015)	(0.077)	(0.021)
Late incident stunted	−0.054	0.059	−0.020	0.134	0.007
	(0.087)	(0.046)	(0.022)	(0.115)	(0.031)
Age 7	0.142*	−0.004	−0.048**	−0.110	−0.042
	(0.081)	(0.043)	(0.021)	(0.106)	(0.029)
Age 8	0.460**	−0.095	−0.073	−0.242	−0.008
	(0.195)	(0.104)	(0.050)	(0.255)	(0.070)
Age 9	0.421*	−0.045	−0.104*	−0.127	−0.035
	(0.240)	(0.127)	(0.062)	(0.315)	(0.086)
Age 10	−0.100	0.605*	−0.057	−0.977	0.829***
	(0.676)	(0.359)	(0.173)	(0.862)	(0.240)
Age 11	0.266	0.781**	−0.066	−1.341	0.922***
	(0.712)	(0.378)	(0.183)	(0.913)	(0.253)
Age 12	0.410	0.761**	−0.059	−1.274	0.844***
	(0.725)	(0.385)	(0.186)	(0.934)	(0.258)
W2 sample	−2.144***	0.668*	0.011	−0.907	0.817***
	(0.658)	(0.349)	(0.169)	(0.842)	(0.234)
W3 sample	−3.818***	0.537	−0.063	−1.056	0.916***
	(0.700)	(0.372)	(0.180)	(0.898)	(0.249)
Age in months at t1	0.016	0.016***	0.003	0.024	0.013***
	(0.011)	(0.006)	(0.003)	(0.015)	(0.004)
Age in months at t2	0.030***	0.003	−0.001	0.017	−0.007**
	(0.009)	(0.005)	(0.002)	(0.012)	(0.003)
African	0.162*	−0.047	0.015	0.031	0.023
	(0.097)	(0.052)	(0.025)	(0.130)	(0.035)
Female	0.222***	−0.106***	0.013	−0.139**	−0.038**
	(0.044)	(0.023)	(0.011)	(0.058)	(0.016)
Ln(per capita hh income)	0.089***	−0.020	0.005	−0.117***	−0.003
	(0.033)	(0.018)	(0.008)	(0.044)	(0.012)
Grant received for index child	0.102	0.027	0.018	−0.260***	−0.008
	(0.068)	(0.036)	(0.018)	(0.091)	(0.025)
No. of children < 15y in hh	0.012	−0.006	−0.001	−0.019	0.000
	(0.011)	(0.006)	(0.003)	(0.015)	(0.004)
Mom deceased	−0.050	−0.011	0.001	−0.156	0.031
	(0.096)	(0.051)	(0.025)	(0.133)	(0.035)
Mom schooling (years)	0.029***	−0.006	0.000	−0.014	−0.009***
	(0.008)	(0.004)	(0.002)	(0.011)	(0.003)
Mom schooling missing	0.347***	−0.082	−0.015	−0.254	−0.122***
	(0.123)	(0.065)	(0.032)	(0.166)	(0.044)
Urban	−0.008	0.041	0.025	0.193**	0.009
	(0.060)	(0.032)	(0.015)	(0.079)	(0.022)
Eastern Cape	0.085	0.133**	−0.063*	−0.062	0.043
	(0.126)	(0.067)	(0.032)	(0.168)	(0.046)
Northern Cape	0.230*	−0.016	−0.005	−0.238	−0.006
	(0.123)	(0.065)	(0.031)	(0.158)	(0.044)
Free State	0.194	0.027	−0.008	−0.109	0.077
	(0.143)	(0.076)	(0.037)	(0.187)	(0.051)
KwaZulu-Natal	0.296**	0.017	−0.028	−0.373**	0.067
	(0.122)	(0.065)	(0.031)	(0.162)	(0.044)
Northwest	0.270*	0.050	−0.022	−0.213	−0.005
	(0.144)	(0.076)	(0.037)	(0.189)	(0.052)
Gauteng	0.457***	−0.014	−0.047	−0.372**	0.003
	(0.136)	(0.072)	(0.035)	(0.178)	(0.049)
Mpumalanga	0.264*	0.030	−0.041	−0.244	0.018
	(0.145)	(0.077)	(0.037)	(0.192)	(0.052)
Limpopo	0.363***	0.002	0.011	−0.314*	0.004
	(0.139)	(0.074)	(0.036)	(0.183)	(0.050)
Constant	0.592	−0.867**	0.980***	6.533***	−0.704**
	(0.780)	(0.414)	(0.200)	(1.010)	(0.278)
N	834	833	835	761	824

Source: Own calculations from NIDS Waves 1–5.

Notes: Standard errors in parentheses. ***p < 0.01, **p < 0.05, *p < 0.10. Omitted categories are not stunted at either t₁ or t₂, Wave 1 sample, age 6, male, rural, Western Cape.