See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/281736245

Is vegetable farming technically efficient in Marilog, Davao City, Philippines?

Parametric and non-parametric approaches

Article  in  Acta Horticulturae · September 2013

DOI: 10.17660/ActaHortic.2013.1006.40

CITATIONS READS

0 598

4 authors:

Jon Marx P Sarmiento Glory Dee Antero Romo

University of the Philippines Mindanao University of the Philippines Mindanao
17 PUBLICATIONS   2 CITATIONS    4 PUBLICATIONS   12 CITATIONS   

SEE PROFILE SEE PROFILE

Rachel Ann Quinineza Vlademir Arnibal Shuck

Lapanday Foods Corporation University of the Philippines Mindanao
2 PUBLICATIONS   0 CITATIONS    3 PUBLICATIONS   0 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Reaping Golden Cacao Pods View project

Economic Impacts of New Technologies and Policy Constraints in the Production of Fruits and Vegetables in Southern Philippines and Australia View project

All content following this page was uploaded by Vlademir Arnibal Shuck on 28 September 2016.

The user has requested enhancement of the downloaded file.

Is Vegetable Farming Technically Efficient in Marilog, Davao City,
Philippines? Parametric and Non-Parametric Approaches
J.M.P. Sarmiento, G.D.A. Romo, R.A.D. Quiniñeza and V.A. Shuck
University of the Philippines Mindanao
Davao
The Philippines

Keywords: Cobb-Douglas production function, data envelopment analysis, stochastic

frontier analysis, technical efficiency, translog production function

Abstract
Vegetable production in the Philippines has increased to an average of 7.9 t/ha,
well below the world average of 17.24 t/ha and the South East Asian average of 9.81
t/ha. In Davao City, productivity continues to average just 4.1 t/ha. Marilog district is
both the largest producer of fresh vegetables and is home to the largest number of
vegetable producers (1,366 farmers). Based on a sample of 118 farmers, the most
popular crops cultivated were tomato, squash, eggplant, chayote and bitter gourd.
More than half (65%) of the farmers cultivated 0.01-0.5 ha of land to produce 945 kg
(4.3 t/ha) of vegetables. This level of production can be readily increased by: (1)
expanding the farm area; (2) adopting better technologies; and (3) encouraging the
more efficient use of resources. This study focuses on the third approach. To assess the
technical efficiency of smallholder vegetable farmers in Marilog, a non-parametric
approach using data envelopment analysis with input and output-oriented assump-
tions, and a parametric approach using stochastic frontier analysis with Cobb-
Douglas and translog production functions were employed. Results suggest that
vegetable farming in the area has considerable potential to improve, since average
technical efficiency ranges from 0.29 to 0.64. To improve efficiency, (1) extension
programs should focus on the less experienced farmers; (2) other livelihood programs
should be designed and introduced to the farmers to augment the capital required for
vegetable farming; and (3) model farms should be established in villages such as
Lower Marilog, Datu Salumay and Dalag-lumot.

INTRODUCTION
In 2007, the population of the Philippines reached 88 million (NSO, 2012). As the
population has grown, so too has the rate of urbanization and the purchasing power of the
emerging middle class, stimulated by rising income levels and increasing remittances
from overseas Filipino workers. Collectively, these trends have increased the demand for
food and a greater variety of food. Fresh vegetables are not only rich in vitamins and
minerals, but are much cheaper to buy than meat and fish. However, domestic vegetable
production is struggling to meet the increasing demand, having increased from 3.6 million
t in 1990 to 4.5 million t in 2010.
In 2009, the Philippines was one of the top vegetable producers in the world,
ranking fifth after China, India, Vietnam and Nigeria. Worldwide, vegetable production
achieves an average yield of 17.24 t/ha and 9.81 t/ha in South East Asia. However, produc-
tion in the Philippines is just 9.37 t/ha, about half of the world average (FAOSTAT, 2012).
This implies that productivity is relatively low suggesting inefficiencies at the farm level.
With relatively low productivity, a closer examination of the Philippines vegetable
industry is warranted, since this has been a vital component for rural development. The
Philippine vegetable industry accounts for more than 30% of the total production of agri-
cultural crops in the country (UNDP, 2006). It ranks sixth in terms of the volume of
production after sugarcane, rice paddy, coconuts, bananas and maize, with some 575,000
ha dedicated to vegetable production (BAS, 2012). However, the average vegetable farm
is just 0.5 ha and thus the industry is dominated by smallholder producers (Johnson et al.,
2008).

Proc. IVth IS on Improving the Performance of Supply Chains

in the Transitional Economies 317
Ed.: P.J. Batt
Acta Hort. 1006, ISHS 2013
 While the national figures suggest that the average yield per hectare is increasing,
production in Davao City has fallen from 5.5 t/ha in 1990 to 4.1 t/ha in 2010 (BAS,
2012). The top three vegetable producing districts in Davao City are Marilog District
(1,366 farmers), Toril District (941 farmers) and Paquibato District (283 farmers).
Marilog District also has the biggest volume of production and the largest area planted
with vegetables (Davao City Agriculturist Office, 2010).
Numerous problems continue to constrain the supply of fresh vegetables. These
include production being concentrated during the dry season, the excessive use of
pesticides and high postharvest losses (Johnson et al., 2008), an insufficient supply of
good quality seeds, high input costs, limited access to credit, and inadequate irrigation
(Lantican, 2000). It is crucial, therefore, to measure the extent of inefficiency associated
with vegetable farming in the Philippines.
While there are three possible ways to increase production: (1) to increase the
farm area allocated for vegetable farming, (2) to develop and adopt new technologies, and
(3) to use resources more efficiently (Bakhsh, 2007), this paper will focus only on the
third option. The study aims to identify the factors affecting efficiency and to assess the
level of technical efficiency of vegetable farming in Marilog District, Davao City.

MATERIALS AND METHODS

Data were gathered from October-November 2010 using a structured question-
naire which helped farmers recall the latest cycle of production. Vegetable production was
aggregated and included four root vegetables, 13 fruit vegetables, 6 leafy vegetables, and
11 other vegetables. Due to issues of accessibility and security concerns in the area, only
10 out of 12 barangays (villages) were interviewed in this study.
Several factors affect productivity at the farm level. Some of these factors include
farm area, transportation, irrigation technology, fertilizer, pesticides, hired labour used,
and family labour used (Bakhsh, 2007; Khai and Yabe, 2011). Productivity is also affected
by the technical efficiency of the farmers (Figure 1).

Data Envelopment Analysis and Tobit Models

Technical efficiency is the ability of a firm to maximize output, given a set of
inputs. It is one of the components of overall economic efficiency (Kumbhaker and
Lovell, 2000). Several alternative methods may be employed to estimate technical
efficiency. These are through the non-parametric and parametric frontiers. Non-parametric
frontiers do not impose functional form on the production frontiers and do not make
assumptions on the error term. It makes use of Linear Programming, particularly, Data
Envelopment Analysis (DEA). It has the ability to identify fully efficient DMUs within
the sample size.
It has extensive application in agriculture (Gul et al., 2009; Guyader and Daures
2005; Thiam et al., 2001) and in the Philippines, having been utilised by Tapia in 1994.
The equation is given below.
,
. . 0,
0,
0, (1)
where: is a scalar; is a Nx1 vector of constants; X is a KxN input matrix; Y is an MxN
output matrix; and xi and yi are the ith DMU inputs and output respectively. The value of
 obtained is an efficiency score of the ith DMU which satisfies 0 0 (Coelli,
1996a).
Using DEA methodology, two approaches are possible; an input-oriented or an
output-oriented. The former minimizes the inputs while seeking to produce the same level
of output, while the latter seeks to maximize the output using existing inputs. Adjustments
are possible due to slack inputs/output. Furthermore, two assumptions are possible;
Constant Returns to Scale (CRS) and Variable Returns to Scale (VRS). In this analysis,

318
VRS were employed since there is perfect competition in the market. DEAP v2.1, written
by Coelli (1996a), was used.
One of the approaches that could be used to analyze the sources of technical
efficiency is the two-stage approach, wherein the efficiency score is first derived from the
production frontier and regressed against various farm and farmer specific variables to
explain the variations in technical efficiency. This approach can use either the ordinary
least squares (OLS) method or tobit regression (Thiam et al., 2001). In this study, tobit
analysis is employed using Equation 2 (Sigelman and Zeng, 1999):
yi*  x i    i
yi  yi* if yi*  0 (2)
yi  0 if y  0
*
i
where: y is the latent dependent variable; yi is the observed dependent variable; x i is the
*
i
vector of independent variables;  is the vector of coefficients; and the  i ’s are assumed
to be independently normally distributed ~ 0, which implies that ~ , is
normally distributed.
The Gretl software (Cottrell, 2005) was used.

Stochastic Frontier Analysis

Stochastic frontier analysis (SFA) is a parametric approach in estimating frontiers.
It imposes a functional form on production and makes assumptions on the data. SFA
measures technical efficiency through the use of error term on the production frontier.
This error term is classified into two components: (1) statistical noise and (2) the factors
related to the farmer (Bakhsh, 2007). The second component forms the technical
inefficiency model. However, SFA is unable to handle zero input or output and multiple
outputs (FAO, 2011). Despite this, SFA is preferred because of the intrinsic random
process which is useful in the efficiency evaluation of any agriculture-related studies
(Ekunwe and Emokaro, 2009).
Thus, the stochastic production function is written as:
Y ; (3)
where Yi represents the output of ith farm, xi represents the input, β represents the vector
of unknown parameter, vi represents the random error, and μi represents the error
associated with technical inefficiency.
As argued by Battese and Coelli (1995), the two-stage approach is not coherent in
its supposition on the technical inefficiency effect. As a result, a one-stage approach was
formulated where both the stochastic production function and technical inefficiency effect
in the frontier are simultaneously estimated through maximum likelihood estimation. The
Cobb-Douglas and translog production function were considered in this study through
FRONTIER v4.1 (Coelli, 1996b).

Data Treatment and Limitation

The production analysis was simplified by including four major factors; land,
labour, fertilizer and pesticide use. Other crop protection measures such as herbicide and
fungicide use were minimal, if not zero. Due to the inability of SFA to handle zero input/
output values, fertilizer and pesticide use were imputed as 0.01 kg so as (1) not to reduce
the sample size and at the same time (2) account for residues from previous crops and/or
from adjacent farms.

RESULTS AND DISCUSSION

The sample of 118 farmers produced mostly tomato, squash, eggplant, chayote and
bitter gourd. More than half of the farmers (65%) produced 945 kg (4.3 t/ha) from 0.01-
0.5 ha of land (Table 1).

319
Data Envelopment Analysis
Suppose the farmer’s goal is to maintain the same level of production, but in so
doing, he wishes to minimize input use (cost). An input-oriented DEA suggests that, on
average, land use will be reduced by 63%, labour use by 55%, fertilizer use by 73%, and
pesticide use by 63%. However, using an output-oriented DEA, the ideal production of
6.3 t/ha will be achieved through decreasing the land use by 32%, labour by 33%,
fertilizer by 41%, and pesticide by 60%. However, in reality, this will be very difficult to
achieve due to different farm specific attributes such as terrain, soil fertility, irrigation,
rainfall distribution and farm management factors.

Stochastic Frontier Analysis

Results for the stochastic frontier analysis using Cobb-Douglas, assuming normal
distribution and translog production function and a truncated normal distribution are
presented. To compare the SFA estimates with the inefficiency effect against the Ordinary
Least Square (OLS) estimates without inefficiency, a log-likelihood test was imposed.
The null hypothesis is that there is no technical inefficiency effect for both production
functions. The reported values were -176 and -171 respectively. This implies a likelihood
ratio (LR) of 48 and 59 respectively. With 5% alpha, using mixed chi-square distribution
(Kodde and Palm, 1986), the Cobb-Douglas, with 3 degrees of freedom, had a critical
value of 7.045 and the translog, with 13 degrees of freedom, had a critical value of
21.742. With an LR greater than the critical values, the null hypotheses were rejected.
Thus, the SFA estimates were favoured over the OLS estimates.
The next test involves the presence of a constant term in the technical inefficiency
model. The null hypothesis was that there is no constant in the technical efficiency
measures for both models. The Cobb Douglas had a log likelihood value of -153 with LR
= 48, while the translog had a log likelihood value of -144 with LR = 52. Using the
critical values 7.045 and 21.742 respectively, the null hypotheses were rejected for both
models. Thus, the model will include technical inefficiency effect with constant term.

Sources of Efficiency/Inefficiency
As noted by most authors (Chakraborty et al., 2002; Zaimova, 2010), technical
efficiency scores differ using different assumptions. The mean technical efficiency of the
vegetable farmers in Marilog District ranged from 0.29 to 0.64 (Table 2). This implies
that there is a lot of potential to improve. Among the barangays considered, Dalag-lumot,
Lower Marilog and Datu Salumay have the highest technical efficiency. Lower Marilog
produced a higher output utilizing lower inputs compared with the average. Datu-
Salumay produced much higher output using less land and fertilizer, while Dalag-lumot
utilized very low amounts of fertilizer (Table 3). There is a potential to design model
farms in barangays Lower Marilog, Datu Salumay and Dalag-lumot.
To determine the sources of efficiency, two methods were used: (1) a technical
efficiency model using tobit for the DEA results; and, (2) a technical inefficiency effect
model using SFA. For the results for both methods to be consistent for each variable, a
positive value in tobit must be complemented with a negative value in SFA. Only seven of
the 15 variables exhibited consistent signs: (1) increasing land area and (2) farmers who
own their land contributed to inefficiency; while (3) increasing farming experience, (4)
continuing contact with extension workers, (5) farmers own source of capital, (6) farmers
practicing mono-cropping, and (7) farms located farther from the road contributed to
efficiency. In addition only the third, fifth and seventh variables were found to be signif-
icant at 1%, 5%, and 10% respectively (Table 4).

CONCLUSION
Vegetable farming in Marilog district, Davao City, has considerable potential to
improve, since the average technical efficiency ranged from 0.29-0.64. Efficiency can be
improved by intensifying extension programs by concerned government agencies and
NGOs on less experienced farmers and by introducing other livelihood programs to help

320
farmers generate more sources of capital for vegetable farming. Model farms may be
established in barangays Dalag-lumot, Lower Marilog and Salaysay to promote practices
to improve efficiency.

Literature Cited
Bakhsh, K. 2007. An Analysis of Technical Efficiency and Profitability of Growing
Potato, Carrot, Radish and Bitter Gourd: A Case Study of Pakistani Punjab. University
of Agriculture, Department of Environmental and Resources Economics, Faisalabas.
http://prr.hec.gov.pk.
BAS. 2012. www.countrystat.bas.gov.ph.
Battese, G.E. and Coelli, T.J. 1995. A Model for Technical Inefficiency Effects in
Stochastic Frontier Production Function for Panel Data. Empirical Economics 20.
Chakraborty, K., Misra, S. and Johnson, P. 2002. Cotton Farmer’s Technical Efficiency:
Stochastic and Nonstochastic Production function Approaches. Agricultural and
Resource Economics Review 31(2):211-290.
Coelli, T.J. 1996a. A Guide to DEAP Version 2.1: A data envelopment analysis (computer)
program. Center for Efficiency and Productivity (CEPA) working papers, University
of New England, Australia.
Coelli, T.J. 1996b. A Guide to FRONTIER Version 4.1: A computer program for Stochastic
Frontier Production and Cost Function Estimation. Center for Efficiency and
Productivity (CEPA) working papers, University of New England, Australia.
Cottrell, A. 2005. Gnu Regression, Econometrics and Time-series Library.
http://www.gnu.org/licenses/fdl.html.
Davao City Agriculturist Office. 2010. List of vegetable farmers in Davao City.
Ekunwe, P. and Emokaro, C. 2009. Technical Efficiency of Catfish Farmers in Kaduna,
Nigeria. http://www.aensionline.com.
FAO. 2011. Measuring and assessing capacity in Fisheries. FAO Fisheries technical paper.
www.fao.org
FAOSTAT. 2012. www.faostat.fao.org.
Gul, M., Koc, B., Dagistan, E., Akpinar, M.G. and Parlakay, O. 2009. Determination of
technical efficiency in cotton growing farms in Turkey: A case study of Cukurova
region. African Journal of Agricultural Research 4(10):944-949.
Guyader, O. and Daures, F. 2005. Capacity and scale inefficiency: An application of data
envelopment analysis in the case of French seaweed fleet. Marine Resource
Economics 20:347-65.
Johnson, G.I., Weinberger, K. and Wu, M.H. 2008. The Vegetable Industry in Tropical
Asia: An overview of production and trade, with a focus on Thailand, Indonesia, the
Philippines, Vietnam and India. AVRDC – The World Vegetable Center, Shanhua,
Taiwan.
Khai, H.V. and Yabe, M. 2011. Technical efficiency analysis of rice production in Viet-
nam. Journal of International Society for Southeast Asian Agricultural Sciences 17(1):
135-146.
Kodde, D.A. and Palm, F.F. 1986. Wald Criteria for jointly testing equality and inequality
restriction. Econometrica 54:1243-1248.
Kumbhaker, S.C. and Lovell C.A.K. 2000. Stochastic Frontier Analysis. Cambridge
University Press, Cambridge.
Lantican, F. 2000. Vegetable Program Area Research Planning and Prioritization. Discus-
sion Paper Series No. 2000-07. Philippine Institute for Development Studies. Makati,
Philippines. http://dirp3.pids.gov.ph.
National Statistics Office. 2012. www.census.gov.ph.
Sigelman, L. and Zeng, L. 1999. Analysing censored and sample-selected data with Tobit
and Heckit models. Political Analysis 8(2):167-182.
Tapia, C.G. 1994: Data Envelopment Analysis Part I: basic theory and mathematical
programming models. Matimyas Matematika 1:1-12.

321
Thiam, A., Bravo-Ureta, B.E. and Rivas, T.E. 2001: Technical efficiency in developing
country agriculture: a meta-analysis. Agricultural Economics 25:235-243.
UNDP. 2006. A logistical evaluation of the vegetables sub-sector. Globalization and
Corporate Citizenship (Project ID00014496) From Seed to Shelf, a Logistic Evalua-
tion of the Vegetables Sub-Sector (Unpublished Report). United Nations Development
Programme. www.undp.org.ph.
Zaimova, D. 2010. Industries and Sectors: Issues and Policies. Perspective of Innovations.
Economics and Business 4(1):41-44.

Tables

Table 1. Production and productivity of vegetable farmers.

Volume Productivity
Land area Count (kg) (kg/ha)
0.01-0.5 65% 945 4320
0.51-1.0 23% 2916 3200
1.01-1.5 6% 4682 3235
1.51-3.0 6% 4583 2102

Table 2. Summary of average technical efficiency scores.

Non-parametric Parametric
Input- Output-
oriented oriented Cobb-Douglas Translog
0.01-0.20 8% 50% 14% 55%
0.21-0.40 19% 15% 20% 18%
0.41-0.60 25% 9% 20% 14%
0.61-0.80 7% 4% 32% 5%
0.81-1.0 41% 21% 14% 8%
Mean technical
0.64 0.39 0.53 0.29
efficiency

Table 3. Output and input level of technically efficient barangays.

Volume Land area Labor Fertilizer Pesticide

(kg) (ha) (man-days) (kg) (L)
Dalag-lumot 1331 0.8 733 0.71 0.97
Lower Marilog 2028 0.56 530 4.51 0.3
Datu Salumay 2163 0.54 721 6.65 0.92
Others 1809 0.6 717 14.2 0.73
Average 1834 0.6 697 12.1 0.7

322
Table 4. Technical efficiency/inefficiency models.

Technical efficiency model Technical inefficiency model

Input-oriented Output-oriented Cobb-Douglas Translog
coeff p-val coeff p-val coeff t-ratio coeff t-ratio
Constant 0.31 0.32 0.65 0.00 *** 1.91 1.70 * 1.91 1.70 *
Land -0.04 0.60 -0.12 0.11 0.35 0.88 0.35 0.88
Age 0.00 0.92 0.00 0.56 0.03 1.93 * 0.03 1.93 *
Gender -0.03 0.66 -0.05 0.45 -0.39 -1.13 -0.39 -1.13
Experience 0.01 0.11 0.00 0.50 -0.04 -2.82 ** -0.04 -2.82 ***
Education 0.01 0.60 -0.01 0.27 0.02 0.41 0.02 0.41
Training -0.16 0.08 * -0.11 0.19 -0.32 -0.97 -0.32 -0.97
Family size 0.00 0.94 0.00 0.76 0.07 1.01 0.07 1.01
Tenancy -0.12 0.17 -0.05 0.47 0.44 1.20 0.44 1.20
Extension 0.04 0.57 0.06 0.43 -0.24 -0.68 -0.24 -0.68
Membership 0.05 0.69 -0.03 0.70 -0.13 -0.32 -0.13 -0.32
Access to credit -0.07 0.60 0.07 0.42 0.47 1.08 0.47 1.08
Source of capital 0.07 0.52 0.16 0.06 * -0.39 -0.89 -0.39 -0.89
Farm road 0.02 0.12 0.01 0.22 -0.07 -1.65 * -0.07 -1.65 *
Number of Veges 0.01 0.86 -0.02 0.48 -0.31 -2.69 ** -0.31 -2.69 ***
Crop Pattern 0.02 0.83 0.05 0.59 0.12 0.29 0.12 0.29
t-test, 2-tail, df=120-13; *10% 1.65; **5% 1.98; ***1% 2.62.

Figurese

Fig. 1. Factors affecting the technical efficiency of vegetable farms in Marilog, Davao
City.

323
 324

View publication stats