Effect of market access provided by farmer organizations on smallholder vegetable farmer’s income in Tanzania

Figures & data

Table 1. Description of variables used in the logistic regression model

Code variables	Variable description	Units of measurement	Expected sign
Dependent variable
Market access	Market access through farmer organizations	1 = Yes, 0 = No (dummy)	±
Independent variables
Gd	Gender of household head	1 = male, 0 = female (dummy)	±
Educ	Education level of vegetable farmers	Number of years in schooling (continuous)	+
Hhsize	Farmer’s household size	Number of family members (continuous)	+
Fsize	Farm size	Farm size in acres (continuous)	+
Ext	Extension services	Number of contacts with extension	±
PriceT	Market price of Tomato	Tomato market price in USD (continuous)	+
PriceA	Market price of Amaranth	Amaranth market price in USD (continuous)	+
PriceAF	Market price of African eggplant	African eggplant market price in USD (continuous)	+
TransfreqT	Frequency of transaction for tomato	Number of times a farmer sales tomato to the market (continuous)	+
TransfreqA	Frequency of transaction for amaranth	Number of times a farmer sales amaranth to the market (continuous)	+
TransfreqAF	Frequency of transaction for african eggplant	Number of times a farmer sales african eggplant to the market (continuous)	+
Transpcost	Cost incurred when transporting the produce	Actual amount of money (USD) used during transportation (continuous)	+
Dism	Distance from farm to the nearest market	Distance from the farm to the nearest market.	+
QuantsT	Quantity of tomatoes supplied	Amount of tomatoes supplied to the market in kilograms	+
QuantsA	Quantity of amaranth supplied	Amount of amaranth supplied to the market in kilograms	+
QuantsAF	Quantity of African eggplant supplied	Amount of African eggplant supplied to the market in kilograms	+
MrktI	Access to market information	1 = Household access to market information, 0 = otherwise (dummy)	±

Table 2. Mean of the variables used in propensity score matching regression model

Mean
Characteristics	Group member	Non member	t-Test	p-Value
Gender	0.65	0.71	4.06**	0.03
Education	7.01	6.13	−0.65	0.52
Household size	5.33	5.57	0.73	0.47
Farm size	0.60	0.46	−2.88***	0.00
Extension contacts	84.02	79.30	1.00	0.31
Price of tomato (USD)	0.19	0.16	−0.83	0.41
Price of amaranth (USD)	0.02	0.03	0.23	0.81
Price of African eggplant (USD)	0.03	0.02	−0.64	0.51
Market information	0.75	0.73	0.27	0.79
Frequency of transaction for tomato	3.78	3.82	0.8	0.78
Frequency of transaction for amaranth	4.09	4.08	−0.01	0.99
Frequency of transaction of African eggplant	4.42	4.54	1.02	0.31
Distance	1.08	0.59	−1.84*	0.07
Transportation cost (USD)	0.22	0.68	5.66***	0.00
Quantity of tomato supplied	777.197	760	−0.07	0.94
Quantity of amaranth supplied	341.73	514.75	0.41	0.68
Quantity of African eggplant supplied	262.04	72.148	−1.77	0.08
Tomato grower	0.68	0.65	−0.44	0.15
Amaranth grower	0.56	0.52	−0.62	0.53
African eggplant grower	0.42	0.25	−2.84	0.66

Notes: *, **, *** represent significance levels at 10%, 5%, and 1%, respectively.

Figure 1. Distribution of estimated propensity scores and common support for the treated and control groups.

Table 3. Average treatment effect for smallholder vegetable farmers’ income (USD)

Matching algorithm	Treated	Control	Difference	Standard error^a	t-Statistic
Nearest neighbour matching (NN)	204.29	168.62	35.6679	46.1727	0.57
Radius matching	220.11	176.65	43.5580	36.6476	1.11*
Kernel matching	220.11	177.90	42.2203	36.5760	1.15*

Note: * represent significance level at 10%.

Appendix A. Factor influencing smallholder vegetable farmers’ access to markets

Variable	Marginal effect (dy/dx)	Standard error	p-Value
Gender	0.029*	0.056	0.091
Education	0.035	0.010	0.603
Household size	−0.001	0.000	0.760
Farm size	0.241***	0.097	0.012
Extension services	0.652	0.080	0.930
Price of tomato	−0.021	0.018	0.228
Price of amaranth	−1.311	0.000	0.944
Price of African eggplant	9.206	0.000	0.527
Market information	0.041*	0.000	0.082
Frequency of transaction for tomato	0.024	0.203	0.723
Frequency of transaction for amaranth	−0.011	0.012	0.381
Frequency of transaction for African eggplant	−0.05	0.078	0.488
Distance	0.078***	0.070	0.047
Transportation cost	0.116	0.058	0.267
Quantity of tomato supplied	0.049	0.063	0.438
Quantity of amaranth supplied	0.044	0.014	0.145
Quantity of african eggplant supplied	0.003	0.000	0.723
Tomato growers	0.071	0.034	0.281
Amaranth growers	0.182	−0.320	0.488
African eggplant growers	0.062	0.283	0.426

Notes: *, **, *** represent significance levels at 10%, 5%, and 1%, respectively.

Appendix B. Performance of different matching estimators

	Performance criteria
Matching estimator	Balancing test	Pseudo-R${ }^2$	Matched sample size
Nearest neighbour matching (NN)
NN(1)	6	0.019	70
NN(2)	4	0.025	70
NN(3)	4	0.058	70
Radius matching
0.01	3	0.007	58
0.25	5	0.021	70
0.50	4	0.035	70
Kernel matching
Band width 0.01	6	0.039	70
Band width 0.25	6	0.012	70
Band width 0.50	7	0.006	70

Appendix C. Propensity score, covariate balance, and chi-square test for the joint significance of variables

	Unmatched		% bias	Matched		% bias
Variables	Treated	Control		Treated	Control
Gender	0.650	0.760	0.7	0.650	0.711	0.6
Education	6.8	6.96	−6.9	6.8	6.896	4.1
Household size	5.333	6.356	5.8	5.333	5.570	5.5
Farm size	0.517	0.537	−4.9	0.517	0.508	2.4
Tomato price (USD)	0.151	0.182	−11.7	0.151	0.149	0.9
Amaranth price (USD)	0.023	0.028	−0.2	0.024	0.026	0.5
African eggplant price (USD)	0.028	0.028	−0.3	0.028	0.027	0.9
Market information	0.171	0.578	−1.7	0.171	0.391	0.8
Distance	0.576	0.917	−16.9	0.576	0.625	−2.4
Transportation cost (USD)	0.464	0.460	0.6	0.464	0.462	0.2
Quantity of tomato supplied	826.2	745.02	3.5	826.2	784.78	1.8
Quantity of amaranth supplied	91.614	568.32	−17.4	91.614	87.526	0.1
Quantity of African eggplant supplied	133.1	174.73	−5.5	133.1	113.78	2.6
Tomato grower	0.686	.651	7.3	0.686	0.683	0.5
Amaranth grower	0.529	0.537	−1.7	0.529	0.504	4.7
African eggplant grower	0.343	0.326	3.6	0.343	0.329	2.9
Pseudo-R${ }^2$	0.031			0.001
LR χ${ }^2$	9.07			1.27
p > χ${ }^2$				1.000

Notes: *, **, *** represent significance levels at 10%, 5%, and 1%, respectively.

Appendix D. Result of sensitivity analysis using the mh bounding approach of Mantel-Haenszel (1959) bounds for variable income

Gamma(Γ)	Q_mh^+	Q_mh^−	p_mh^+	p_mh^−
1	0.000	0.000	0.000	0.000
1.05	−0.071	−0.071	0.528	0.528
1.1	−0.071	−0.071	0.528	0.528
1.15	−0.071	−0.071	0.528	0.528
1.2	−0.071	−0.071	0.528	0.528
1.25	−0.071	−0.071	0.528	0.528
1.3	0.325	0.325	0.325	0.325
1.35	−0.071	−0.071	0.528	0.528
1.4	0.325	−0.071	0.325	0.325
1.45	0.325	0.325	0.325	0.528
1.5	−0.071	−0.071	0.528	0.528

Gamma: 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).

Haenszel, Mantel, L. (1959). Statistical aspects of data from retrospective studies of disease. Journal Of The National Cancer Institute, 22, 719–748.

 PubMedGoogle Scholar