Effects of fodder production on smallholder farmers’ household income in Homa Bay County, Kenya: An application of propensity score matching

Figures & data

Figure 1. Map of the study area.

Table 1. Distribution of sample size

Sub-County	Households	Proportion	Proportionate per Cluster
Rachuonyo East	1,161	0.59	177
Rachuonyo South	813	0.41	123
Total	1,974		300

Source: KNBS, 2019

Table 2. Covariates used for propensity score matching

Variable	Description	Expected sign
Dependent variable
Participation	Choice decision by the farmer to grow fodder
Independent variables
Age	Age of household head (years)	+
Gender	Gender of household head (male=1, female=0)	+/-
Education	Education level of the household head (years)	+
Household size	Household size (numbers)	+/-
Farm size	Land under fodder (acres)	+
Group membership	Group membership (yes=1, no=0)	+
Herd size	Number of cattle owned by a farmer	+
Volume	Milk Volume (liters)	+
Off Farm Income	Income derived from off-farm activities (Kshs)	+
Manure quantity	Quantity of Manure (number of wheelbarrows)	+
Credit access	Access to credit (yes=1, no=0)	+
Extension contacts	Access to extension services (yes=1, no=0)	+
Farming experience	Experience in fodder production (years)	+
Fertilizer quantity	Quantity of fertilizer (kg)	+
Farming labour	Labour hours (man-hours)	+
Fodder training	Training on fodder production (yes=1, no=0)	+

Table 3. Variance inflation factor (VIF) results

Variable	VIF	1/VIF
Land under fodder (acres)	3.97	0.2517
Fertilizer quantity (Kgs)	3.45	0.2900
Land preparation (man days)	3.41	0.2936
Extension contact	1.75	0.5704
Manure quantity (wheelbarrows)	1.67	0.5977
Group membership	1.64	0.6079
Years of schooling	1.51	0.6607
Age of household head (years)	1.51	0.6618
Total herd size (numbers)	1.36	0.734
Household size (numbers)	1.35	0.742
Access to training	1.28	0.7826
Volume of milk (litres)	1.22	0.8194
Access to credit	1.1	0.9089
Gender	1.09	0.9193
Mean VIF	1.88

Table 4. Categorical variables affecting use of Napier and natural grazing among farmers in Rachuonyo East and South sub-counties

Variables	Category	Percentage		Chi-square	Sig
		Napier	Natural grazing
Gender	Male	69.17	72.22	0.327	0.568
	Female	30.83	27.78
Education level	None	3.33	5.56
	Primary	17.50	32.78	12.227***	0.007
	Secondary	31.67	30.56
	Tertiary	47.50	31.11
Main occupation	Business	28.33	39.44		0.000
	Livestock	18.33	13.33	35.804***
	Crop cultivation	8.33	22.22
	Casual Labour	5.00	10.00
	Formal employment	40.00	15.00
Group membership	Yes	65.00	13.33	85.651***	0.000
	No	35.00	86.67

*** indicates a significance level at 1%

Table 5. Continuous variables affecting use of Napier and natural grazing among farmers in Homa Bay County

Variables	Mean			t-test	p-value
	Napier	Natural grazing	Combined
Household size (Numbers)	4.575	5.027	4.847	−2.070**	0.039
Age (Years)	44.300	47.417	46.17	−2.212**	0.028
Herd size (Numbers)	3.225	3.356	3.303	−0.500	0.618

** indicates a 5% significance level

Table 6. Probit Estimation of factors influencing the decision of farmers to mainly feed their livestock on Napier grass

Variable	Coefficient	Std. Err.	Z	P >/Z/
Gender	−0.5901	0.7567	−0.78	0.435
Age (years)	−0.0131	0.0299	−0.44	0.662
Years of schooling	0.1619*	0.0950	1.70	0.088
Household size (number)	−0.4317*	0.2496	−1.73	0.084
Herd size (number)	−0.3931*	0.2206	−1.78	0.075
Manure quantity(wheelbarrows)	−0.1834	0.1924	−0.95	0.340
Land preparation (man days)	0.6769*	0.3216	2.10	0.035
Land size fodder (acres)	14.3386***	3.8932	3.68	0.000
Volume of milk(litres)	−0.1594	0.1173	−1.36	0.174
Extension access	−2.3994*	1.0995	−2.18	0.029
Fodder training	0.9276	0.6008	1.54	0.123
Access to credit	−0.7982	0.6307	−1.27	0.206
Group membership	−0.4863	1.0402	−0.47	0.640
Fertilizer quantity (Kg)	4.8430	22.5206	0.22	0.830
Constant	−2.0706	1.7689	−1.17	0.242

Number of observations= 300, LR Chi²= 365.25, Pseudo R²= 0.9045, Prob>Chi²=0.0000, Log likelihood= −19.280672

***, **, * represents significance levels at 1%, 5% and 10% respectively.

Figure 2. Region of common support from the propensity score matching.

Table 7. Region of common support

Treatment assignment	Off support	On support	Total
Untreated	0	180	180
Treated	106	14	120
Total	106	194	300

Table 8. Propensity score test for covariates based on nearest neighbor matching technique

Variables	Sample	Mean		% reduction		t	p>|t|
		Treated	Control	% bias	bias
Gender	Unmatched	0.6917	0.7222	−6.7		−0.57	0.569
	Matched	0.7143	0.7857	−15.6	−133.8	−0.42	0.676
Age(years)	Unmatched	44.3	47.417	−26.5		−2.21	0.028
	Matched	40.429	44.286	−32.8	−23.8	−0.99	0.329
Years of schooling	Unmatched	12.417	9.9833	57.2		4.77	0.000
	Matched	12.714	13.857	−26.9	53.0	−1.05	0.302
Household size	Unmatched	4.575	5.0278	−24.7		−2.07	0.039
	Matched	4.6429	4.9286	−15.6	36.9	−0.62	0.538
Herd size (numbers)	Unmatched	3.225	3.3556	−5.9		−0.50	0.618
	Matched	2.9286	3.2143	−13.0	−118.8	−0.54	0.595
Quantity of manure (wheelbarrows)	Unmatched	1.45	1.0417	18.4		1.67	0.097
	Matched	3.25	3.7143	−20.9	−13.7	−0.51	0.611
Labour on land preparation(man days)	Unmatched	5.5417	1.1111	137.3		12.84	0.000
	Matched	2.8571	3.3571	−15.5	88.7	−1.41	0.171
Land under fodder (acres)	Unmatched	0.5388	0.0766	169.5		15.87	0.000
	Matched	0.3009	0.3018	−0.3	99.8	−0.02	0.988
Volume of milk (litres)	Unmatched	3.55	2.3903	43.5		3.87	0.000
	Matched	3.7857	4.7857	−37.5	13.8	−0.72	0.480
Extension contacts	Unmatched	0.3583	0.40	−8.6		−0.73	0.469
	Matched	0.3571	0.2857	14.7	−71.4	0.39	0.699
Fodder training	Unmatched	0.4417	0.4833	−8.3		−0.71	0.480
	Matched	0.5	0.5714	−14.3	−71.4	−0.37	0.717
Access to credit	Unmatched	0.45	0.4611	−2.2		−0.19	0.850
	Matched	0.2857	0.5	−42.9	−1828.6	−1.15	0.262
Group membership	Unmatched	0.3917	0.3056	18.1		1.54	0.124
	Matched	0.2143	0.2143	0	100	0	1.000
Quantity of fertilizer (Kgs)	Unmatched	14.113	0.0056	149		14.14	0
	Matched	0	0	0	100	0
	Sample	Pseudo R2	LR Chi2	P> chi2	Mean Bias	Med Bias
	Unmatched	0.905	365.25	0.000	48.3	21.5
	Matched	1.000	38.82		17.9	15.5

Table 9. Average treatment effects on income

Variable	Sample	Treated	Controls	Difference	S.E	t-stat
Income effects	Unmatched	13130	7381.17	5748.83	1650.62	3.48
	ATT	15707.14	11790.48	3916.67	9045.82	0.08

USD = 152.34 Kshs

Table 10. Mantel- Haenszel for household income

Gamma (Γ)	Q_mh+	Q_mh-	P_mh+	P_mh-
1	4.1325	4.1325	0.000018	0.000018
1.05	4.1696	4.2908	0.000015	8.9e-06
1.1	4.113	4.3499	0.00002	6.8e-06
1.15	4.0597	4.4072	0.000025	5.2e-06
1.2	4.0093	4.4628	0.00003	4.0e-06
1.25	3.9615	4.5169	0.000037	3.1e-06
1.3	3.9161	4.5695	0.000045	2.4e-06
1.35	3.8729	4.6207	0.000054	1.9e-06
1.4	3.8318	4.6707	0.000064	1.5e-06
1.45	3.7924	4.7195	0.000075	1.2e-06
1.5	3.7548	4.7673	0.000087	9.3e-07

Note: MH bounds using STATA 13. Γ = 1≈no “hidden” bias(odds of differential assignment due to unobserved factors). Q_ mh+=Mantel-Haenszel statistic (assumption: overestimation of treatment effect); Q_ mh- =Mantel-Haenszel statistic (assumption: underestimation of treatment effect); P_ mh+= Significance level (assumption: overestimation of treatment effect); P_ mh_=Significance level (assumption: underestimation of treatment effect).

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.