Cereal production practices and technical efficiency among farm households in major “teff” growing mixed farming areas of Ethiopia: A stochastic meta-frontier approach

Figures & data

Figure 1. Map of the study areas.

Source: Ethio GIS and CSA (2007)

Table 1. Descriptive statistics

Variable name	Adea		Enemay		Overall
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
Sex head (Male Headed HH)	0.94	0.24	0.96	0.20	0.95	0.21
Age head	47	12.71	44	12.77	45	12.79
Family size	5.10	1.77	4.44	1.69	4.70	1.75
Farm size	1.50	1.02	1.02	0.67	1.21	0.86
Extension access (yes)	0.79	0.41	0.69	0.46	0.73	0.45
Membership in organization (yes)	0.61	0.49	0.83	0.38	0.74	0.44
Mobile telephone ownership (yes)	0.91	0.29	0.75	0.43	0.81	0.39
Number of crops	3	1.06	3	1.13	3	1.10
TLU	6.30	3.58	4.18	2.29	5.03	3.05
Input center (km)	2.35	1.96	1.92	1.71	2.08	1.82
Land quality	2.59	1.43	2.87	1.75	2.76	1.64

Source: Authors’ analysis using primary data (2020)

Table 2. Mean values of the cereal output, input variables and stress incidence by crop type

Variables	“Teff”		Wheat		Barely		Maize		Sorghum
	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.	Mean	Std. Dev.
Yield (kg/ha)	1115.32	531.69	1799.80	1022.53	1711.95	1166.18	1349.13	1785.89	2064.14	1108.41
Land (ha)	1.08	0.74	0.50	0.41	0.35	0.20	0.23	0.13	0.38	0.13
Seed (kg/ha)	35.03	22.39	140.09	66.81	97.75	49.84	31.76	37.10	14.85	5.40
UREA (kg/ha)	156.98	112.40	123.97	96.24	101.06	83.43	68.42	300.92	-	-
NPS (kg/ha)	142.19	93.90	128.36	92.12	82.28	85.67	23.12	44.88	-	-
Man days/ha	63.84	29.31	67.56	33.95	54.27	21.77	77.83	90.01	59.51	25.49
Oxen days/ha	16.21	9.84	17.91	12.36	14.17	7.50	24.71	47.04	13.54	7.58
Stress incidence	0.53	0.47	0.37	0.48	0.17	0.38	0.38	0.48	-	-

Source: Authors’ analysis using primary data (2020)

Table 3. Improved technologies adoption by crop type

Practices	“Teff” (N = 374)		Wheat (N = 225)		Barely (N = 66)		Maize (N = 71)		Sorghum (N = 15)		Overall (N = 378)
	No	%age	No	%age	No	%age	No	%age	No	%age	No	%age
Improved seed/HYV (yes)	114	30.48	49	21.78	-	-	20	28.17	-	-	134	35.45
Seed rate (yes)	71	18.98	23	10.22	-	-	2	2.82	6	40	97	25.66
Fertilizer rate/UREA (yes)	126	33.69	87	38.67	18	27.27	6	8.45	-	-	139	36.77
Fertilizer rate/NPS (yes)	165	44.12	90	40	14	21.21	1	4.1	-	-	207	54.76
Row planting (yes)	42	11.23	10	4.4	2	3.03	32	45.07	-	-	75	19.84

Source: Authors’ analysis using primary data (2020)

Table 4. Intensity of adoption by crop type

Practices	“Teff” (N = 374)	Wheat (N = 225)	Barely (N = 66)	Maize (N = 71)	Sorghum (N = 15)	Overall (N = 378)
	(%age/Mean)	(%age/Mean)	(%age/Mean)	(%age/Mean)	(%age/Mean)	(%age/Mean)
Improved seed/HYV (ha)	24.37 (0.32)	22 (0.15)	-	29.58 (0.07)	-	22.62 (0.41)
Seed rate (ha)	11.13 (0.11)	10 (0.06)	-	2.81 (0.01)	40 (0.125)	10.72 (0.15)
Fertilizer rate/UREA (ha)	24.16 (0.25)	36.41(0.17)	27.27 (0.10)	0.8 (0.02)	-	19.73 (0.27)
Fertilizer rate/NPS (ha)	33.62 (0.35)	38.6 (0.19)	21.21 (0.08)	1.4 (0.004)	-	32.3 (0.48)
Row planting (ha)	8 (0.09)	4.44 (0.02)	3.03 (0.02)	44.6 (0.11)	-	10.11 (0.12)

Source: Authors’ analysis using primary data (2020)

Table 5. Maximum likelihoods estimate of the parameters for SFPF model

Stochastic frontier parameters		Adea wereda		Enemay wereda		Metafrontier parameters
		Coeff.	Std. Err.	Coeff.	Std. Err.	Coeff.	Std. Err.
Constant	${\beta }_0$	9.212***	0.569	8.432***	0.331	9.199***	0.060
lnAREA	${\beta }_1$	0.415***	0.103	0.381***	0.075	0.423***	0.014
lnSEED	${\beta }_2$	0.205***	0.066	0.036	0.037	0.167***	0.005
lnFERTILIZER	${\beta }_3$	0.246	0.074	0.294***	0.047	0.248***	0.009
lnLABOR	${\beta }_4$	0.010	0.064	0.115*	0.065	−0.018**	0.009
lnOXEN_DAY	${\beta }_5$	−0.114**	0.047	0.063	0.060	−0.079***	0.008
Sigma2 $\left({\sigma }^2={\sigma }_v^2+{\sigma }_u^2\right)$		0.113***		0.125***		0.034***
Gamma $\left(\gamma ={\sigma }_u^2/{\sigma }^2\right)$		0.837***		0.822***		0.970***
Number of observation		150		222		374
Log likelihood		−75.26		−61.06		396.90

Source: Authors’ analysis using primary data (2020)

Coefficients with ***, **, and * are significant at 1, 5 and 10 percent level of significance, respectively

Table 6. Meta-level technical efficiency by technology adoption status of farm households

Technology	Adopter	Non-adopter	t-value
Use of HYV	0.646 (0.013)	0.543 (0.011)	−5.935***
Recommended seed rate	0.632 (0.015)	0.562 (0.010)	−3.626***
Recommended N fertilizer rate	0.580 (0.015)	0.579 (0.011)	−0.045
Row planting	0.600 (0.010)	0.575 (0.010)	−1.173
HYV with recommended seed rate	0.652 (0.025)	0.571 (0.009)	−2.852***
HYV with recommended N fertilizer rate	0.687 (0.022)	0.564 (0.009)	−4.806***
HYV with row planting	0.644 (0.021)	0.571 (0.009)	−2.670***
HYV with recommended seed rate and N fertilizer rate	0.672 (0.045)	0.576 (0.009)	−2.042**
Participation in cluster farming	0.591 (0.016)	0.576 (0.010)	−0.740
Overall TE	Mean = 0.580 (0.167) Min = 0.131, Max = 0.907		-
Overall TGR	Mean = 0.901 (0.078) Min = 0.638, Max = 0.993		-

Source: Authors’ analysis using primary data (2020)

Coefficients with ***, **, and * are significant at 1, 5 and 10 percent level of significance, respectively

Figure 2. Efficiency distribution by adoption of HYVs.

Source: Authors’ analysis using primary data (2020)

Table 7. Factors explaining efficiency of smallholder farm households

Variables	Model-1		Model-2
	Coeff.	S.E.	Coeff.	S.E.
Constant	0.5020***	0.0609	0.4752***	0.0546
Socio-economic factors
Head Sex (1 if male)	0.1125***	0.0390	0.0996***	0.0350
Head Age (years)	−0.0012*	0.0007	−0.0016**	0.0007
Family size (persons)	0.0007	0.0053	−0.0004	0.0047
Mobile telephone ownership (1 if yes)	0.0759***	0.0223	0.0677***	0.0200
Extension service (1 if yes)	−0.0084	0.0230	−0.0010	0.0208
Membership in Organization (1 if yes)	0.0532**	0.0213	0.0360*	0.0190
Farm size (ha)	−0.0258**	0.0123	−0.0087	0.0110
Access to input center (km)	−0.0091***	0.0036	−0.0083***	0.0032
TLU	0.0029	0.0032	0.0045	0.0028
Number of crops	0.0140*	0.0081	0.0068	0.0072
Land quality index	−0.0048	0.0054	−0.0079	0.0049
Stress incidence (ha)	−0.1163***	0.0204	−0.0919***	0.0182
Improved production practices
Cluster farming	0.0586**	0.0229	0.0170	0.0205
High yielding variety	0.0724***	0.0213	0.0864***	0.0191
Seed rate/planting density	0.0550***	0.0196	0.0504***	0.0176
Fertilizer rate (UREA)	0.0157	0.0176	0.0175	0.0158
Row planting	0.0016	0.0221	0.0003	0.0198
Dependent variable Number of observation	TE scores derived from Group frontiers 372		TE scores droved from Meta-frontier 372
Wald chi2(17)	127.11		133.10
Prob > chi2	0.0000		0.0000
Log likelihood	162.31		201.87

Source: Authors’ analysis using primary data (2020)

Coefficients with ***, **, and * are significant at 1, 5 and 10 percent level of significance, respectively

Table

Variables	Definitions	Hypothesis associates with the inefficiency variables
Dependent variable: Cereal output: the natural logarithm of the aggregate value of cereal production in ETB^10
Conventional inputs for production
Ln land	Continuous	The natural logarithm area planted in ha
Ln seed	Continuous	The natural logarithm seed use in kg
Ln fertilizer	Continuous	The natural logarithm of fertilizer use in kg
Ln labor	Continuous	The natural logarithm of labor use in man-days^11
Ln draught power	Continuous	The natural logarithm in ox-days
Inefficiency variables
Sex head	Dummy: 1 = if household head is male, and zero if female.	Male-headed households has higher level of technical efficiency than their counterpart female-headed households.
Age head	Continuous: age of the household head in years.	Household with older head has lower technical efficiency as compared with households having younger heads
Family size	Continuous: total number of people living in house	Households with higher number of family size more likely to be technically efficient than households having less number of family size.
Farm size	Continuous: area of total cultivated land in hectare	Households having more cultivated cereal land tends to be less technically efficient when compared to those with smaller land holding.
Extension access	Dummy 1 = if the household has access to extension services, 0 otherwise	Households who have access to extension services have higher technical efficiency
Membership in organization	Dummy: 1 if household belong to an association, 0 otherwise	Membership in organization tends to increase the technical efficiency of the households.
Mobile telephone ownership	Dummy: 1 if household has mobile telephone, 0 otherwise.	Mobile telephone ownership increase technical efficiency due to access to information on available technologies
Number of crops	Contentious: the number of crop grown by the HH	The higher number of crops the households cultivating tends to decrease the technical efficiency of the households
Input center (km)	Location of HH relative to input center in km	The longer distance to input market tends to be technically inefficient.
Tropical livestock unit (TLU)	Continuous: total number of livestock in TLU	Households with higher herd size more likely to be technically efficient as compared to those with smaller herd holdings.
Stress incidence	Continuous: Proportion of area of cultivated land affected by stresses	Incidence of biotic and abiotic stress on the farm tends to decrease the technical efficiency of households
Use of improved seed	Dummy: 1 if household used improved variety on some proportion of farm land, 0 otherwise	Use of improved seed increase the technical efficiency of the households
Cluster farming	Dummy: 1 if household participated in cluster farming, 0 otherwise	Participating in cluster farming tends to increase the technical efficiency of households
Seed rate	Dummy: 1 if household applied as per recommended rate on some proportion of farm land, 0 otherwise	The use of seed as per recommended rate increase technical efficiency of households
Nitrogen fertilizer rate	Dummy: 1 if household applied as per recommended rate on some proportion of farm land, 0 otherwise	The use of fertilizer as per recommended rate increase technical efficiency of households
Row planting	Dummy: 1 if household planted in row on some proportion of farm land, 0 otherwise	Planting cereal crops with row expected to have higher level of technical efficiency
Land quality index^12	Continuous: 1 if the soil has best land quality, 9 is lowest quality	Farm households having best land quality become more technical efficient than those who do not have.

Source: Authors’ compilation(2020)

Table

Age groups (Years)	Male	Female
<10	0	0
10–13	0.2	0.2
14–16	0.5	0.4
17–50	1.0	0.8
>50	0.7	0.5

Source: Storck et al. (1991).

Table

Crop	Seed rate (kg/ha) Vertisols		Seed rate (kg/ha) Clay soil		Seed rate (kg/ha) Light soil		UREA rate [Kg/ha]	NPS rate [Kg/ha]
	Broadcast	Row	Broadcast	Row	Broadcast	Row
Tef	20–25	10	25	15	25–30	15–20	100	100
Wheat	125–150	100–125	125–150	100–125	125–150	100–125	100	100–125
Barely	125	85	125	85	125	85	100	100
Maize	-	-	-	23–30	-	-	200	200
Sorghum	-	-	15–20	8–10	-	-	100	100

Source: EIAR (2007); MoA (2010)

Figure D1. Kernel density distribution of technical efficiency.

Source: Authors’ analysis using primary data (2020)

Storck, H., Emana, B., Adnew, B., Borowicciki, A., & Shimelis , W. (1991). Farming system and resouce economics in the tropics. In Farming system and farm management practices of smallholders in the Hararghe. Vol. II. Wissenschaftsverlag Vauk Kiel, Germany, Highland.

 Google Scholar

EIAR. (2007). Crop technology Utilization Manual. Addis Ababa, Ethiopia.

 Google Scholar

MoA, 2010 Crop Production Package. Addis Ababa, Ethiopia

 Google Scholar

Data Availability Statement

The data used for the study are available from the corresponding author upon request.