Training, technology upgrading, and total factor productivity improvement of farms: A case of cassava (Manihot esculenta Crantz) production in Dak Lak province, Vietnam

Figures & data

Figure 1. Map of survey sites. Source: Authors’ drawing.

Table 1. Farm holder’s characteristics of Eakar DIstrict, Dak Lak province (2018)

Farming Size	Ethnicity	Household number	Schooling (year)	Labors (labor)	Experience (year)	Cass. Area (ha)	Yield (MT/ha)
Small farms	Minorities	16	7.76	2.38	7.94	0.64	21.38
	Majorities	80	9.60	2.59	8.66	0.66	22.99
Subtotal	-	96	9.33	2.55	8.54	0.66	22.72
Large farms	Minorities	1	5.00	3.00	9.00	2.50	23.00
	Majorities	103	9.39	2.67	7.83	1.57	23.59
Subtotal	-	104	9.36	2.67	7.85	1.58	23.59
All farms		200	9.36	2.62	8.18	1.14	23.17

Source: Authors’ computations.

Table 2. Technical efficiencies calculated in CCR and BCC models in 2015 and 2018

Cassava farming size	Technical Efficiency in 2015 (Geometric Means)			Technical Efficiency in 2018 (Geometric Means)
	Comprehensive efficiency (CRS)	Pure technical efficiency (VRS)	Scale of efficiency (SE)	Comprehensive efficiency (CRS)	Pure technical efficiency (VRS)	Scale of efficiency (SE)
Small farms	0.738	0.831	0.889	0.790	0.867	0.911
Large farms	0.790	0.857	0.921	0.825	0.857	0.963
All farms	0.775	0.852	0.911	0.817	0.870	0.942

Table 3. Technical efficiency, technology, and productivity changes of cassava farms

Farming size	Malmquist Indices	Mean (Geometric)	Max	Min	Range
Small farms	Efficiency change (EFFCH)	0.93	1.58	0.58	1
	Technology change (TECHCH)	1.07	1.78	0.84	0.94
	TFP change (TFPCH)	1.00	2.11	0.62	1.49
Large farms	Efficiency change (EFFCH)	0.96	1.42	0.62	0.8
	Technology change (TECHCH)	1.11	1.31	0.85	0.46
	TFP change (TFPCH)	1.06	1.61	0.67	0.94
All farms	Efficiency change (EFFCH)	0.95	1.58	0.58	1
	Technology change (TECHCH)	1.09	1.78	0.84	0.94
	TFP change (TFPCH)	1.03	2.11	0.62	1.49

Table 4. Dependent and independent variables in logit models

Variable	Type	Descriptions	Expected sign
Dependent variables
TFPCH	Binary	1, if TFPCH greater than 1; 0, otherwise
EFCH	Binary	1, if EFCH greater than 1; 0, otherwise
TECHCH	Binary	1, if TECHCH greater than 1; 0, otherwise
Explanatory variables
GENDER	Binary	1, if the corresponding household’s head is male; 0, otherwise	±
ETHNICITY	Binary	1, if the corresponding household’s head belongs to majority ethnic groups (Kinh or Tay ethnicity; 0, otherwise	±
SCHOOLING	Discrete	Years spent studying at school	+
LABORS	Discrete	Number of labors of the corresponding household (people)	+
DISTANCE	Continuous	Distance from home to the cassava farm (km)	-
EXPERIENCE	Discrete	Cassava farming experience (years)	+
CONTRACT	Binary	1, having a formal contract with cassava buyers.; 0, otherwise	+
TRAINING	Binary	1, attending extension training; 0, otherwise	+
AREACASS	Continuous	Cassava farming area (ha)	±
DIVERSECROPS	Continuous	Measured by the corresponding household’s total arable land area divided by cassava farming area	±
CAPITALSTOCK	Continuous	The total capital stock of the corresponding household	+
ACCESSBANKS	Binary	1, if the corresponding household borrowed from banks; 0, otherwise	+

Table 5. Binary logistic regression models

Explanatory variables	Response variable TFPCH (recoded)		Response variable TECHCH (recoded)		Response variable EFFCH (recoded)
	B	Odds Ratio	B	Odds Ratio	B	Odds Ratio
GENDER	−.184	.832	−.002	.998	.806	2.239
ETHNICITY	.315	1.370	−.411	.663	−.176	.838
SCHOOLING	−.004	.996	−.244	.783	−.104	.901
LABORS	.197	1.218	−.065	.937	.223	1.249
DISTANCE	−.169	.844	−.060	.942	−.022	.978
EXPERIENCE	−.073	.930	−.062	.940	−.034	.966
CONTRACT	.051	1.052	−.142	.867	−.161	.852
TRAINING	.957***	2.603	.812*	2.253	.213	1.237
AREACASS	−.425	.653	.803	2.232	−.718**	.488
DIVERSECROPS	−.320*	.726	−.433**	.648	−.349*	.705
CAPITALSTOCK	.102***	1.108	−.025	.975	.058**	1.060
ACCESSBANKS	1.559***	4.756	−.782	.457	1.162***	3.197
CONSTANT	−1.547	.213	3.812	45.251	−1.341	.262
−2 Log likelihood	202.431		156.771		230.421
Cox & Snell R Square	.302		.091		.169
Nagelkerke R Square	.405		.156		.229
n = 200

*Statistical significance at the 10% level.

**Statistical significance at the 5% level.

***Statistical significance at the 1% level.