The impact of the productive safety net program (PSNP) on food security and asset accumulation of rural households’: evidence from Gedeo zone, Southern Ethiopia

Figures & data

Figure 1. Map of the study area: (a) map of Ethiopia, (b) map of SNNPR, (c) map of Gedeo zone.

Table 1. Sample households

Woreda	Kebele	Total households		Sample households
		PSNP	Non-PSNP	PSNP	Non-PSNP	Total
Wonago	Tokicha	331	561	21	20	41
	Balebukisa	312	642	30	30	60
	Jemjemo	328	482	35	35	70
	Karasoditi	389	893	37	36	73
Dilla Zuriya	Chichu	523	538	34	34	68
	Andida	295	462	24	24	48
	Gola	373	348	17	18	35
Total		2,551	3,926	198	197	395

Source: Own Survey, 2021

Table 2. Variable description, measurement, and expected signs

Variables Code	Type and definition		Measurement	Expected Signs
Dependent variable
	Dummy, participation in PSNP (Treatment)		1 if yes, 0 otherwise
(II) Outcome variables
KCAL	Continuous, Daily Caloric Intake per adult equivalent		KCAL	+
(2)  CONSEXP	Continuous, Consumption Expenditure of the HH		Measured in Ethiopian Birr (ETB)	+
(3)  FOODSHRT	Dummy, HH faced a food shortage		1 if the household faced a food shortage in a year and 0 otherwise	-
(4)  TOTINC/LVSK	Continues, Total income earned from the farm and non-farm sources		ETB	+
(5)  TLU	Continuous, Livestock Holding or owned		(TLU) Tropical livestock units	+
(6)  ROOF	Dummy, Roofing type of the HH		1 if the house type is iron corrugated and 0 otherwise	+
(III) Independent variables
SEX	Dummy, Sex of Household Head,		1 if male, 0 otherwise
(2)  AGE	Continuous, Age of the household age		Years completed	±
(3)  MARSTAT	Dummy, Marital Status of Household Head		1 if married, 0 otherwise
(4)  FAMSZ	Continuous Family Size of the Household		Number of household members	+
(5)  EDUST	Continuous, Education status of household head		Years of schooling	-
(6)  DISFMRKT	Continues, Distance from nearest market center		Kilometers	-
(7)  ACCEXTN	Dummy, Access to Agricultural Extension service		1 if exposed to extension service, 0 otherwise	-
(8)  CREDIT	Dummy, Access to Credit Service		1 if the household received credit, 0 otherwise	-
(9)  LAND	Continuous, Size of landholding of the Household		Hectare	-
(10)  SAVING	Dummy, Saving engagements by the Household		1 if the household engaged in saving, 0 otherwise	-
(11)  PARTOFFARM	Dummy, Participation in off-farm activities		1 if the HH participates in off-farm activities and 0 otherwise.	-
(12)  MODTECH	Dummy, Use of Modern Farm Technologies		1 if the HH use improved seed and fertilized and 0 otherwise.	-

Table 3. Sample of the study

Category	Frequency	Percent	Cumulative
Non PSNP Beneficiary	197	49.87	49.87
PSNP Beneficiary	198	50.13	100.00
Total	395	100.00

Source: Own survey result, 2021

Table 4. Descriptive statistics of household characteristics (Continuous variables)

Pre-intervention Variable	Total N = 395		PSNP N = 198		Non PSNP N = 197		Difference in mean	t-value
	Mean	Std	Mean	Std	Mean	Std	Mean
Age	35.66	7.87	36.07	8.02	35.26	7.72	0.81	−1.025
Family size	5.88	1.27	6.26	1.27	5.48	1.14	−0.78***	−0.694
Land holding	0.35	0.18	0.32	0.15	0.38	0.19	0.06***	3.509
Educational level	3.59	2.52	3.16	2.64	4.03	2.52	0.869***	3.341
Distance from market	2.73	1.56	2.73	1.63	2.72	1.48	0.0059	−0.0059

Table 5. Descriptive statistics of household characteristics (dummy variables)

Pre-intervention variable	Category	PSNP HH		Non-PSNP HH		Total		χ^2
		N = 198		N = 197		N = 395
		Freq.	percent	Freq.	percent	Freq.	percent
SEX	Male	153	77.27	141	71.57	294	74.43	1.685
	Female	45	22.73	56	28.43	101	25.57
MARSTAT	Married	196	98.99	194	98.48	390	98.73	0.2077
	Not married	2	1.01	3	1.52	5	1.27
ACCEXT	Accessed	155	78.28	133	67.51	288	72.91	5.7996
	Not accessed	43	21.72	64	32.49	107	27.09
CREDIT	Received	167	84.34	49	24.87	216	54.68	140.93***
	Not received	31	15.66	148	75.13	179	45.32
SAVE	Engaged	177	89.85	94	47.72	280	71.07	67.592***
	Not engaged	21	10.15	103	52.28	114	28.93
MODTECH	Use	94	47.72	68	34.52	162	41.12	7.086***
	Not use	103	52.25	129	65.48	232	58.88

Source: Survey data, 2021

Table 6. Logistic regression results

Variables	Coefficient	Std. Err.	P> |z|
SEX	0.129	0.108379	0.232
AGE	0.00068	0.0129287	0.958
MARSTAT	−0.456	2.868908	0.874
FAMSZ	0.577	0.1112052	0.000***
EDUST	−0.145	0.0249571	0.000***
DISMRKT	0.074	0.0513963	0.147
ACCEXT	−0.432	0.0896483	0.000***
CREDIT	2.66	0.3305974	0.000***
LAND	−2.096	0.9010802	0.020**
SAVE	1.396	0.2800141	0.000***
OFFARM	−0.303	0.23309	0.193
MODTECH	.2421357	0.378018	0.522
CONS	−4.189	3.430441	0.222

Number of obs. = 395 LR chi²(12) = 221.16 Prob. > chi² = 0.000

Pseudo R² = 0.3968 Log likelihood = −159.7445

***, and ** means significant at 1 percent and 5 percent probability levels, respectively.

Table 7. Distribution of estimated propensity scores

Groups	Mean	Min	Max	S.D.	Observations
Control Households	0.2648357	0.0009565	0.9430374	0.2745777	197
PSNP Households	0.6976765	0.0692172	0.9410604	0.2204312	198
Total	0.4665935	0.0009565	0.9430374	0.330882	395

Table 8. Matching algorithm result

Matching Estimator	Performance Criteria
	Balancing test*	Pseudo-R^2	Matched sample size
Caliper
0.1	11	0.017	269
0.25	10	0.033	269
0.5	10	0.022	272
Kernel Matching
Band width of 0.1	5	0.328	373
Band width of 0.25	6	0.328	373
Band width of 0.5	5	0.235	373
Nearest Neighbor matching
1 neighbor	5	0.328	373
2 neighbor	9	0.053	373
3 neighbor	11	0.033	373
4 neighbor	10	0.025	373
5 neighbor	11	0.027	373

* Number of explanatory variables with no statistically significant mean differences between the matched groups of the program and non-program households

Table 9. Propensity score and covariate balance test

Variables	Matching sample	Mean		percent reduction		t-test		V(T)/V(C)
	Sample	Treated	Control	% bias	Bias	T	P>|t|	0.68*
	Unmatched	0.73551	0.26583	187.2		18.61***	0.000
	Matched	0.70695	0.70774	−0.3	99.8	−0.03	0.974	0.98
SEX	Unmatched	0.77273	0.71574	13.1		1.30	0.195	-
	Matched	0.77841	0.81108	−7.5	42.7	−0.76	0.449	-
AGE	Unmatched	36.071	35.259	10.3		1.02	0.306	1.08
	Matched	35.636	34.207	18.2	−76.0	1.77*	0.077	1.35*
MARSTAT	Unmatched	0.9899	0.98477	4.6		0.45	0.650	-
	Matched	.98864	0.98011	7.6	−66.2	0.64	0.521	-
TOTFAMSZ	Unmatched	6.2677	5.4873	64.4		6.39***	0.000	1.24
	Matched	6.0511	5.9489	8.4	86.9	0.82	0.415	0.99
EDUST	Unmatched	3.1616	4.0305	−33.6		− 3.34***	0.001	1.09
	Matched	3.375 -	3.5838	−8.1	0.77	−76.0	0.439	1.21
DISMRKT	Unmatched	2.7293	2.7234	0.4		0.04	0.970	1.21
	Matched	2.6018	2.6691	−4.3	−1023	−0.42	0.675	1.19
ACCEXT	Unmatched	0.78283	.67513	24.3		2.42*	0.016	-
	Matched	.80682	.78267	5.5	77.6	0.56	0.576	-
CREDIT	Unmatched	0.84343	.24873	148.6		14.77***	0.000	-
	Matched	0.82386	.8125	2.8	98.1	0.28	0.783	-
LAND	Unmatched	0.32271	.38548	−35.3		−3.51***	0.000	0.60*
	Matched	0.32606	.27223	30.3	14.2	2.72	0.007	0.57*
SAVE	Unmatched	0.89394	.52284	89.2		8.87***	0.000	-
	Matched	0.88068	.88068	0.0	100	−0.00	1.000
OFFFARM	Unmatched	0.46465	0.44162	4.6		0.46	0.647	-
	Matched	0.47159	0. 41,761	10.8	−134.5	1.02	0.310
MODTECH	Unmatched	0.4798	0.34518	27.5		2.74***	0.006	-
	Matched	0.48295	0.4375	9.3	66.2	0.85	0.394	-

***and * means significant at the 1 percent, and 10 percent probability levels, respectively.

Figure 2. The propensity score distribution for balancing tests before and after matching.

Table 10. Chi-square test for the joint significance of variables

Sample	Pseudo R^2	LR chi^2	p> chi^2
Unmatched	0.398	217.73	0.000
Matched	0.025	12.34	0.500

Table 11. Average treatment effect on the treated (ATT) results

Outcome Variable	Treated	Controls	Difference	S.E.	t-value
Total annual income(ETB)	12,396.19	7859.6	4,536.8	695.1	6.52***
Annual food shortage experienced	0.61	0.51	0.10	0.09	1.17
Livestock holdings(TLU)	0.68	0.57	0.10	0.08	1.30
Housing condition	0.5	0.41	0.09	0.08	1.15
Annual consumption expenditure (ETB)	2,949.52	2,365.61	583.92	351.44	1.66*
Daily calorie consumption (Kcal/AE)	2,305.85	2,013.15	292.7	133.31	2.20**

***, **, and * stand for significant at 1 percent, 5 percent, and 10 percent, respectively.

Table 12. Rosenbaum bound for outcome for daily calorie consumption/AE

Gamma	sig+	sig-	t-hat+	t-hat-	CI+	CI-
1	0	0	2336.49	2336.49	2223.89	2412.19
1.25	0	0	2239.45	2362.32	2156.71	2504.33
1.5	0	0	2216.5	2436.92	2088.63	2577.19
1.75	0	0	2160.84	2498.25	2046.99	2632.33
2	2.2e-16	0	2108.64	2561.72	1994.07	2688.34
2.75	1.9e-12	0	2014.45	2638.27	1929.67	2775.6
3	1.5e-11	0	1991.56	2690.01	1909.19	2823.75
3.75	1.4e-09	0	1940.55	2732.83	1821.33	2908.62
4	4.3e-09	0	1937.9	2746.67	1802.13	2930.48
4.25	1.2e-08	0	1929.67	2779.73	1771.81	2970.07
4.75	6.3e-08	0	1889.85	2829.55	1738.5	2986.14
5	1.3e-07	0	1876.02	2848.63	1738.5	2986.14
6	1.3e-06	0	1827.72	2888.37	1716.92	3078.17
6.25	2.1e-06	0	1820.85	2913.11	1709.8	3086.84
6.75	4.7e-06	0	1799.47	2944.09	1676.11	3102.3
7	6.7e-06	0	1779.87	2967.05	1633.48	3120.99
7.25	9.5e-06	0	1752.56	2984.48	1630.82	3156.14
7.75	.000018	0	1738.5	2986.14	1606.47	3193.44
8	.000023	0	1738.5	2986.14	1606.47	3195.53

* gamma—log odds of differential assignment due to unobserved factors

sig+—upper bound significance level

sig-—lower bound significance level

t-hat+—upper bound Hodges-Lehmann point estimate

t-hat-—lower bound Hodges-Lehmann point estimate

CI+—upper bound confidence interval (a = .95)

CI-—lower bound confidence interval (a = .95)