Citation
Technical efficiency of the dual-purpose cattle system in Venezuela

Material Information

Title:
Technical efficiency of the dual-purpose cattle system in Venezuela
Creator:
Ortega, Leonardo
Place of Publication:
[Gainesville, Fla.]
Publisher:
University of Florida
Publication Date:
Language:
English

Subjects

Subjects / Keywords:
Agricultural economics ( jstor )
Employee efficiency ( jstor )
Farm economics ( jstor )
Farming ( jstor )
Farms ( jstor )
Labor productivity ( jstor )
Production efficiency ( jstor )
Productive efficiency ( jstor )
Productivity ( jstor )
Stocking rate ( jstor )
CATTLE, DUAL, EFFICIENCY, PURPOSE, TECHNICAL
Dissertations, Academic -- Food and Resource Economics -- UF ( lcsh )
Food and Resource Economics thesis, Ph. D ( lcsh )
City of Gainesville ( local )
Genre:
government publication (state, provincial, terriorial, dependent) ( marcgt )
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Summary:
ABSTRACT: The dual-purpose cattle system (DPCS) is the traditional cattle production system in the lowland tropics of Latin America where crossbred cattle are used for the production of milk and beef. This system is based on local and low cost inputs, but has been often considered to be inefficient due to its low partial productivity indices when compared with those used in developed countries. Few data exist about the efficiency of this system but the scant literature available is based on partial productivity indices. These indices provide useful information but do not take into account the effect of total inputs on total outputs as a measure of total efficiency. This study attempts to provide standard measurement of the technical efficiency (TE) of this system based on the concept of total factor productivity. An analysis is conducted to estimate the main determinants TE of the DPCS located in Zulia State, Venezuela.
Summary:
ABSTRACT (con't): A deterministic production frontier model and two stochastic production frontier models (half normal and exponential distribution for the error term) were estimated on a sample of 127 farms. The Cobb-Douglas functional form was used. Average TE values were 0.630, 0.819, and 0.922 for the deterministic, half-normal and exponential models. Higher values were obtained for the stochastic models because these models separate the effect of random noise from the effect of technical inefficiency. However, the ordinal ranking of the farms according to their TE values was similar. In a second stage, TE values were regressed against some socio-economic and technological variables in order to explain the variation on efficiency. A logistic model was used since the dependent variable (TE) is bounded by zero and one. The parameters of this model were calculated using the OLS technique. The significant positive factors were farmer experience, farmer's presence on the farm, location, production system, cow productivity, and frequency of technical assistance. Farm size and labor productivity showed a quadratic effect and credit a negative impact. The simulation model suggests how policies and managerial decisions to address these variables could help improve efficiency of this system.
Thesis:
Thesis (Ph. D.)--University of Florida, 2002.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: World Wide Web browser and PDF reader.
System Details:
Mode of access: World Wide Web.
General Note:
Title from title page of source document.
General Note:
Includes vita.
Statement of Responsibility:
by Leonardo Ortega.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Ortega, Leonardo. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
12/1/2004
Resource Identifier:
029835025 ( ALEPH )
53113120 ( OCLC )

Downloads

This item is only available as the following downloads:


Full Text

PAGE 1

TECHNICAL EFFICIENCY OF THE DUAL-PURPOSE CATTLE SYSTEM IN VENEZUELA By LEONARDO ORTEGA A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2002

PAGE 2

Copyright 2002 by Leonardo Ortega

PAGE 3

TO MY FAMILY DAVID LEONARDO, YOANA VICTORIA, AND YOANA

PAGE 4

iv ACKNOWLEDGMENTS I would like to express my deep gratitude and appreciation to Dr. Chris Andrew, chairman of the supervisory committee, for his advice, outstanding guidance, and encouragement during the development of the graduate program, planning and writing of this dissertation. Very specially, I would like to acknowledge Dr. Ronald W. Ward, whose dedication, assistance, and guidance in the analysis of the data and writing of this dissertation have been exceptional demonstrating his excellence in mentoring. Sincere appreciation is extended to other members of the committee, Dr. T. H. Spreen and Dr. P. E. Hildebrand from the Food and Resource Economics Department and Dr. A. de Vries from the Animal Sciences Department, for their advice and for reviewing the dissertation. I would like to extend my gratitude to other faculty, staff, and students in the Food and Resource Economics Department that made these years very rewarding in knowledge, experience, and friendship. Thanks are addressed to La Universidad del Zulia (LUZ) for the financial support provided. Finally, special thanks go to my wife Yoana, for her support, patience, and encouragement during all these years, and to my two children, David Leonardo and Yaona Victoria, for their understanding and patience.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES..........................................................................................................xii ABSTRACT.....................................................................................................................xi v CHAPTER 1 INTRODUCTION...........................................................................................................1 Problem Statement..........................................................................................................3 Research Objectives........................................................................................................5 2 LITERATURE REVIEW................................................................................................8 Characteristics of the DPCS in Latin America...............................................................8 Characteristics of the DPCS located in Zulia State......................................................11 Agro-ecological Conditions...................................................................................11 General Characteristics of the DPCS.....................................................................13 Resource Management and Animal Performance..................................................14 Technical Efficiency.....................................................................................................15 Determinants of Technical Efficiency..........................................................................23 3 DATA AND DESCRI PTIVE STATISTICS.................................................................27 Survey Methodology.....................................................................................................27 Sample Selection....................................................................................................28 Data Collection......................................................................................................29 Valuation Methods.................................................................................................30 Descriptive Statistics.....................................................................................................31 Farmer Characteristics...........................................................................................31 Owner characteristics.......................................................................................31 Manager characteristics...................................................................................33 Farm Characteristics..............................................................................................35 General characteristics.....................................................................................35 Organization and resource management..........................................................37 Production and productivity.............................................................................42

PAGE 6

vi Production Cost......................................................................................................44 Net Farm Income and Profitability........................................................................47 Financial Aspects...................................................................................................47 Technical Aspect....................................................................................................48 Forage management.........................................................................................48 Animal management........................................................................................52 Technical assistance.........................................................................................58 Management Aspects.............................................................................................59 Organization.....................................................................................................60 Control.............................................................................................................60 Planning...........................................................................................................65 Direction..........................................................................................................65 4 ECONOMETRIC PROCEDURES................................................................................71 Production Frontier Methodology.................................................................................71 Functional Form.....................................................................................................73 Deterministic Production Frontier..........................................................................74 Stochastic Production Frontier Function................................................................74 Normal-half normal distribution......................................................................75 Exponential distribution...................................................................................76 Standardized Coefficient........................................................................................78 Multicollinearity.....................................................................................................78 Technical Efficiency Estimates.....................................................................................79 Determinants of Technical Efficiency..........................................................................80 Defining the Efficiency Variables ( z )....................................................................81 Farmer characteristics......................................................................................81 Farm characteristics.........................................................................................84 Technological variables...................................................................................87 Modeling the Distribution of Efficiency Indices..........................................................89 Heterocedasticity....................................................................................................91 Standardized Coefficients......................................................................................91 5 EFFICIENCY ANALYSIS............................................................................................92 Production Frontier Models..........................................................................................92 Technical Efficiency Values.........................................................................................98 Determinants of Technical Efficiency........................................................................104 Farmer Characteristics Variables.........................................................................109 Farm Characteristic Variables..............................................................................109 Credit (CRED)..............................................................................................109 Farm size (DSUG).........................................................................................110 Location (Z)...................................................................................................111 Production system (PSYST)..........................................................................111 Land tenure (DTEN)......................................................................................112 Technological Variables......................................................................................112 Breeding system (DBRED)............................................................................112

PAGE 7

vii Frequency of technical assistance (DTECHN)..............................................113 Cow productivity (PROD).............................................................................113 Labor productivity (LTMILKKER and LTMILKSQ)...................................114 Stocking rate (CARGANEF).........................................................................114 6 SIMULATION OF THE DETERMI NANTS OF TECHNICAL EFFICIENCY........118 Simulation Model........................................................................................................118 Single Effect of Socio-economic Variables................................................................119 Farmer Characteristics.........................................................................................119 Farm Characteristics............................................................................................121 Credit..............................................................................................................121 Farm size........................................................................................................124 Location.........................................................................................................126 Production system..........................................................................................127 Technological Variables......................................................................................129 Cow productivity...........................................................................................129 Frequency of technical assistance..................................................................131 Labor productivity.........................................................................................133 Analysis of Alternative tw o Variables Simulations....................................................134 Farm Size and Labor Productivity.......................................................................134 Farm Size and Production System.......................................................................135 Farm Size and Cow Productivity.........................................................................136 Farm Size and Producer Presence........................................................................137 Farm Size and Frequency of Technical Assistance..............................................138 7 SUMMARY AND CONCLUSIONS..........................................................................140 Summary of the Findings............................................................................................141 Characterization of DPCS....................................................................................141 Production Frontier and Technical Efficiency.....................................................143 Determinants of Technical Efficiency.................................................................143 Policy and Managerial Implications...........................................................................144 Limitations and Future Studies...................................................................................146 APPENDIX A QUESTIONNARIE....................................................................................................148 B TSP PROGRAM.........................................................................................................169 C SIMULATION OUTPUT / TECHNICAL EFFICIENCY / HALF-NORMAL DISTRIBUTION.........................................................................................................193 D SIMULATION OUTPUT / TECHNICAL EFFICIENCY / EXPONENTIAL DISTRIBUTION.........................................................................................................201

PAGE 8

viii LIST OF REFERENCES.................................................................................................204 BIOGRAPHICAL SKETCH...........................................................................................210

PAGE 9

ix LIST OF TABLES Table page 1.1 DPCS productivity indices............................................................................................2 1.2 National productivity indices and goals of DPCS........................................................4 2.1 Management indices for DPCS in Zulia State............................................................14 2.2 Physical productivity indices for DPCS in Zulia State...............................................15 3.1 Population and sample distribution of DPCS by location..........................................29 3.2 Owner characteristics of DPCS..................................................................................32 3.3 Who is the manager of DPCS ?..................................................................................34 3.4 Manager characteristics of DPCS in Zulia State.........................................................34 3.5 General characteristics of DPCS.................................................................................36 3.6 Land tenure............................................................................................................... ..38 3.7 Use of the land........................................................................................................... .39 3.8 Average cattle distribution per farm...........................................................................40 3.9 Herd management and or ganization indices of DPCS................................................41 3.10 Labor utilization indeces of DPCS...........................................................................41 3.11 Composition and structure of capital per farm (Bs.).................................................42 3.12 Intensity of capital per farm (Bs)..............................................................................42 3.13 Production structure and composition per farm........................................................43 3.14 Productivity indeces of DPCS in Zulia State............................................................44 3.15 Composition and cost stru cture per farm of DPCS...................................................46 3.16 Relationship between cost, output s, and factors per farm of DPCS.........................46

PAGE 10

x 3.17 Profitability and net ma rgin per farm of DPCS........................................................47 3.18 Loan situation per farm surveyed.............................................................................48 3.19 Annual interest rate for loans to surveyed farms......................................................48 3.20 Number and pasture size of DPCS............................................................................48 3.21 Grazing system of DPCS..........................................................................................49 3.22 Rest and utiliza tion periods (Days)...........................................................................49 3.23 Cultural practices of DPCS.......................................................................................50 3.24 Number of farmers that classify the herd..................................................................52 3.25 Methods of animal selection of DPCS......................................................................53 3.26 Cultural practices of DPCS.......................................................................................53 3.27 Zootecnic parameters of DPCS.................................................................................54 3.28 Breeding System of DPCS........................................................................................54 3.29 Milking characte ristic of DPCS................................................................................55 3.30 Supplement feed used by DPCS...............................................................................56 3.31 Characteristic of the herd health plan of DPCS........................................................57 3.32 Mortality and cause of mortality...............................................................................58 3.33 Characteristics of tec hnical assistance of DPCS.......................................................59 3.34 Organization chart of DPCS.....................................................................................61 3.35 Description of records kept by DPCS.......................................................................62 3.36 Use of the record by DPCS.......................................................................................63 3.37 Knowledge of milk production cost by farmers.......................................................63 3.38 Do you compare results among years?.....................................................................63 3.39 Type of corrective used to achieve goals..................................................................64 3.40 Control the machinery operatorsÂ’ activities..............................................................64 3.41 Farmers' objectives of DPCS....................................................................................66

PAGE 11

xi 3.42 Do you have plans?...................................................................................................67 3.43 Type of instruments used by farmers to plan the activities......................................67 3.44 Knowledge by farmers of the possi bilities to increase production levels.................67 3.45 Ways to increase production of DPCS......................................................................68 3.46 Management and problem solution of DPCS...........................................................69 4.1 Socio-economic variables...........................................................................................83 4.2 Farm size................................................................................................................. ....86 4.3 Production per cow.....................................................................................................88 5.1 Box-Cox model for testing nonlinearity.....................................................................93 5.2 Production frontier estimates......................................................................................95 5.3 Standardized coefficients of production frontier estimates........................................96 5.4 Technical efficiency estimates per farm...................................................................100 5.5 Ordinal rank of farm according to the level of technical efficiency.........................102 5.6 Estimates of the determinants of technical efficiency..............................................105 5.7 Standardized coefficients of the determinants of technical efficiency.....................107 5.8 t-statistic for the dummy variables of the determinants of technical efficiency (halfnormal distribution).............................................................................................116 5.9 t-statistic for the dummy variables of the determinants of technical efficiency (Exponential distribution)....................................................................................117

PAGE 12

xii LIST OF FIGURES Figure page 2.1 FarrellÂ’s model........................................................................................................... .17 5.1 Ranking of standardized coefficien t of deterministic production frontier..................96 5.2 Ranking of standardized coefficien t of half-normal production frontier....................97 5.3 Ranking of standardized coefficien t of exponential production frontier . ...................97 5.4 Ranking of standardized coefficients for the determinants of technical efficiency (half-normal distribution).....................................................................................108 5.5 Ranking of standardized coefficients for the determinants of technical efficiency (Exponential distribution)....................................................................................108 6.1 Impact of producer experien ce on technical efficiency............................................120 6.2 Impact of producer experience on average technical efficiency...............................120 6.3 Impact of producer presen ce on technical efficiency...............................................122 6.4 Impact of producer presence on average technical efficiency..................................122 6.5 Impact of credit on technical efficiency....................................................................123 6.6 Impact of credit on aver age technical efficiency......................................................123 6.7 Impact of farm size on technical efficiency..............................................................124 6.8 Impact of farm size on average technical efficiency................................................125 6.9 Impact of location on technical efficiency................................................................126 6.10 Impact of location on aver age technical efficiency................................................127 6.11 Impact of production system on technical efficiency.............................................128 6.12 Impact of production system on average technical efficiency................................129 6.13 Impact of production per co w on technical efficiency............................................130

PAGE 13

xiii 6.14 Impact of production per cow on average technical efficiency..............................131 6.15 Impact of frequency of technical assistance on technical efficiency......................132 6.16 Impact of frequency of technical assi stance on average technical efficiency........132 6.17 Impact of labor productiv ity on technical efficiency..............................................134 6.18 Impact of farm size and labor productiv ity on average technical efficiency..........135 6.19 Impact of farm size and production sy stem on average technical efficiency.........136 6.20 Impact of farm size and cow productiv ity on average technical efficiency...........137 6.21 Impact of farm size and producer pr esence on average technical efficiency.........138 6.22 Impact of farm size and technical assi stance on average technical efficiency.......139

PAGE 14

xiv Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy TECHNICAL EFFICIENCY OF DUAL-PURPOSE CATTLE SYSTEM IN VENEZUELA By Leonardo Ortega December, 2002 Chair: Dr. Chris O. Andrew Major Department: Food and Resource Economics The dual-purpose cattle system (DPCS) is the traditional cattle production system in the lowland tropics of Latin America where crossbred cattle are used for the production of milk and beef. This system is based on local and low cost inputs, but has been often considered to be inefficient due to its low partial productivity indices when compared with those used in developed countries. Few data exist about the efficiency of this system but the scant literature available is based on partial productivity indices. These indices provide useful information but do not take into account the effect of total inputs on total outputs as a measure of total efficiency. This study attempts to provide standard measurement of the technical efficiency (TE) of this system based on the concept of total factor productivity. An analysis is conducted to estimate the main determinants TE of the DPCS located in Zulia State, Venezuela. A deterministic production frontier model and two stochastic production frontier models (half normal and exponential distribution for the error term) were estimated on a

PAGE 15

xv sample of 127 farms. The Cobb-Douglas functional form was used. Average TE values were 0.630, 0.819, and 0.922 for the deterministic, half-normal and exponential models. Higher values were obtained for the stochastic models because these models separate the effect of random noise from the effect of t echnical inefficiency. However, the ordinal ranking of the farms according to their TE values was similar. In a second stage, TE values were regressed against some socio-economic and technological variables in order to explain the variation on efficiency. A l ogistic model was used since the dependent variable (TE) is bounded by zero and one. The parameters of this model were calculated using the OLS technique. The significant positive factors were farmer experience, farmer’s presence on the farm, location, production system, cow productivity, and frequency of technical assistance. Farm size and labor productivity showed a quadratic effect and credit a negative impact. The simulation model suggests how policies and managerial decisions to address these variables could help improve the efficiency of this system.

PAGE 16

1 CHAPTER 1 INTRODUCTION Before 1983, the Venezuelan government carried out a protectionist policy toward agriculture. This policy was intended to create an agrarian structure that would allow the country to achieve self-sufficiency and to decrease the expenditure of foreign currency due to food imports. During this period subsidy policies were established regarding fertilizers, concentrate feed commodities, low interest rates on loans, as well as the establishment of an output price system for the different agricultural products. This period could be described as a period of economic growth in which a horizontal expansion prevailed. The period following 1983 has been characterized by a contraction in the economy during which hyperinflation, a large fiscal deficit, and a sharp decline of gross domestic product took place, a situation that not only in cluded Venezuela but also other Caribbean and Latin American countries. This reduction in economic activity was also felt by the agricultural sector. Subsidies to agricultural inputs and preferential financial conditions were eliminated, and policies to decrease support for extension and agricultural research were adopted in order to mitigate macroeconomic problems. VenezuelaÂ’s milk and beef industries did not escape from this situation. Per capita milk and beef consumption dropped from 120 to 80 liters, and from 20 to 17 kg respectively, due to a decrease in per capita income (Plasse, 1992). Likewise, milk production was reduced from more than 1500 to nearly 1400 million liters per year during 1987-1991 (Castillo, 1992). To a greater extent, this slump was a consequence of

PAGE 17

2 eliminating the economic-competitive ability of intensive dairy farms due to high production costs. The dual-purpose cattle system (DPCS), the traditional cattle production system in the lowland tropics of Latin American in which local cattle of mixed zebu, criollo and European breeding are used for the production of milk and meat (Sere and de Vacarro, 1985), has endured the economic decline. It uses local and low cost inputs as an alternative to the more expensive purebred cow system. This more favorable input situation provides the stability and flexibility necessary to buffer the economic changes and market conditions prevailing in developing countries like Venezuela. The DPCS has become the main alternative to supply the milk requirements of tropical countries, encompassing around 78% of total bovine and 41% of total milk production in the Latin-American tropics. This system, however, is often considered to be inefficient due to its low partial productivity indices as suggested by the indices (Table 1) for eight countries (Honduras, Costa Rica, Panama, Bolivia, Colombia, Brazil, Venezuela and Mexico). Table 1.1. DPCS productivity indices Concept Averages and ranges Milk yield per cow-day (kg) 4.0 (2.8 – 6.5) Milk production per lactating period (kg) 1,180 (749 – 1,584) Duration of lactating period (d) 290 (244 – 311) Calving rate (%) 64 (39 – 81) Age at first calving (mo) 37 (32 – 43) Calf mortality (%) 13 (12 – 24) Stocking rate (AU ha-1) 1.4 (0.72 – 1.9) Annual production of milk per ha (kg ha-1 yr) 476 (182 – 749) Annual production of beef per ha (kg ha-1 yr) 116 (45 – 192) Source: Pearson, 1986. AU: animal unit

PAGE 18

3 Experimental and empirical evidence obtaine d in different areas of the American tropics indicate that it is possible to duplicate, at least, the indices of this system using available technology (Gonzalez, 1992; Holmann and Lascano, 1998; Plasse, 1992; Urdaneta et al., 1992). Therefore, the challenge for DPCS is not only to produce more milk and beef but also to produce these commodities more efficiently (Plasse, 1992). Problem Statement In Venezuela, DPCS contributes approximately 90% of total national milk production (Plasse, 1992) and represents approximately 60% of total milk consumption. Expectations based on national productivity indices indicate that there is room to improve productivity in Venezuela if the technologies developed by the national research centers and universities are applied (Table 1.2). These technologies have been designed to improve animal and land productivity, infrastructure, and managerial problems. Recommended practices include pasture fertilization, adequate pasture size, rotational grazing, strategic supplementation, reproductive control, animal health, use of account and technical record keeping, and technical assistance among others. Empirical evidence from Zulia State where DPCS is most common indicates that farmers receiving technical assistance who are located in humid areas on average triple the average values of productivity. Production levels averaged 1500 L. milk ha-1yr-1, and 150 kg beef ha-1yr-1 and 1.82 AU ha-1 with extreme values of 2.62 AU ha-1, 2500 L milk ha-1yr-1, and 270 kg beef ha-1yr-1 for model farms (Gonzalez, 1992). Empirical experiences in the dry forest are also available. Romero (1995) evaluated the impact of the technology generated by a local program (Laberinto program) sponsored by the state university (University of Zulia). The results from the Laberinto program show the benefits of using the technology. Farmers who did not use the technology had lower

PAGE 19

4 indices (578 L ha-1yr-1, 0.68 Animal Unit (AU) ha-1yr-1, and 327 Bolivares (Bs) ha-1yr-1 of gross margin) than those farms using the technology (1,374 L ha-1yr-1, A U ha-1yr-1, and 3026 Bs ha-1yr-1). Table 1.2. National productivity indices and goals of DPCS Concept Level of production ActualGoal Change (%) Births (%) 45 65 44 Calf Mortality (%) 11 8 -27 Calving percentage (%) 40 60 50 Adult Mortality (%) 10 4 -60 Age at first calving (yr) 4 3 -25 Age of male to slaughter (yr) 4 2.5 -37 Weight of male to slaughter (kg of beef adjusted to 2.5 yr) 281 450 60 Liters of milk per cow – year (L) 1200 2000 67 Liters of milk per cow in herds-year (L) 540 1300 141 Kg of beef per cow in herd-yr 51 131 157 Source: Plasse (1992) The Venezuelan government, through its agricultural research centers, has focused on improving the economic efficiency of this system to make its products more accessible to consumers and to improve profit for producers, while attempting to diminish loss of foreign currency due to import of food. Despite these efforts, productivity continues to be low. It is possible to observe that some farmers seldom adopt a complete technological package or even individual practices, while farmers adopting almost all the technological inputs do not necessarily obtain higher levels of productivity. This indicates that not only the technology inputs but also socio-economic factors, such as managerial skill, access to credit, and contact with extension agents among others, are determinants of the efficiency of this system. Interestingly, the efficiency of this system has been genera lly measured through partial productivity indices which provide useful information but do not take into account the effect of total inputs on

PAGE 20

5 output as a measure of total efficiency. Furthermore, if technical change is to be measured, these indices will be biased because they include the effects of factor substitution together with the effects of advances in production technique (Peterson and Hayami, 1977). A standard for this system is needed that uses the concept of total factor productivity and indicates how efficient these farmers are given a set of inputs and the technology available. Estimating the efficiency of this system through total factor productivity (TFP) indices, instead of partial productivity indices, is essential. Defining and quantifying the socio-economic factors that determine the efficiency of this system will help to address the efforts made by different public and private organizations to improve efficiency of this system, to make farmersÂ’ products more accessible to consumers (lower prices), and to be more competitive in an open economy. Likewise it will contribute to improving the Venezuelan balance of payments. Research Objectives Technical efficiency is the focus of this research with the objective to determine efficiency and to identify and quantify the factors that explain the distribution of efficiency scores of the dual-purpose cattle system in Venezuela. Specific objectives include the following: Characterization of the DPCS, descri bing components, inputs, outputs and technology. Estimation of the production function fron tier and the different partial output elasticities. Determination of the technical efficiency level expressed as an index and the creation of a standard for this system considering the total factor productivity concept. Definition and quantification of the factor s that determine technical efficiency.

PAGE 21

6 Generation of valuable information about the factors that influence technical efficiency to be used by research centers, universities, producers, and policy makers to improve efficiency in the DPCS. Given the above objectives, the following hypotheses guide the research. DPCS presents a low level of technical efficiency, which is or can be influenced by Technological variables such as breeding system, stocking rate, technical assistance, labor productivity and cow pr oductivity (factors that positively affect the efficiency of DPCS). Human capital variables such as educa tion level, experience, and producer residence (contribute to increase the efficiency). Farm characteristics such as size, loca tion, tenure, production system, and access to credit that constrain the efficiency of this system. To achieve the objectives and to test hypotheses, information from a survey conducted in Zulia State in 1994 was used to de velop stochastic and deterministic frontier functions to obtain the technical efficiency values of the 127 farms surveyed. These efficiency values were regressed against selected socio-economics variables in order to define the determinants of the efficiency and to study the managerial and policy implications. This study is divided into seven chapters, Chapter 1 is the introductory chapter in which the problem statement, objectives, and hypothesis are set forth. In chapter 2, a general description of DPCS in Latin America and an overview of the different methodologies used to define and to explain the technical efficiency of the firms are presented. Chapter 3 refers to the data and the descriptive statistic of DPCS used in this study. Chapter 4 describes the production frontier methodology and the model developed to explain the variation on technical efficien cy. The empirical results are presented and discussed in Chapter 5. Chapter 6 includes a simulation model used to measure the impact of the socio-economic variables on technical efficiency. Lastly, Chapter 7 is a

PAGE 22

7 summary of the major findings of this research and some recommendation for further studies.

PAGE 23

8 CHAPTER 2 LITERATURE REVIEW This chapter encompasses three sections: the first section presents a general description of the dual-purpose cattle system in South America, including characterization of the system for Zulia State, Venezuela. An overview of the development of the efficiency analysis is given in the second section, in which different approaches are analyzed and their advantages and disadvantages listed. The third section deals with different methodologies and variables used to explain the variation in the technical efficiency values. Characteristics of the DPCS in Latin America The dual-purpose cattle system is a term that has been utilized to describe the traditional cattle production system in the lowland tropics of Latin American. In this system, local cattle consisting of a mix of zebu, criollo and European breeds are used for the production of milk and meat (Sere and de Vacarro, 1985). Traditionally, cow and calf constitute a biological and natural production unit during the nursing period (Castillo, 1992). The economic importance of DPCS is attributed to the percentage of output that it represents with respect to national production. It encompasses around 78% of total bovine population and 41% of milk production in Tropical America (Fernandez-Bacca, 1995; Estrada, 1993). In the late 1970s and early 1980s, in Colombia, Nicaragua, Panama, Guatemala, and Brazil, DPCS provided 51%, 75%, 67%, 75% and 35% of national milk yields respectively (Guzman, 1995; Sere and de Vaccaro, 1985;Vargas,

PAGE 24

9 1993). In Venezuela, this system contributes approximately 90% of total national milk production. (Plasse, 1992). This system is shown to be more efficient, both economically and productively, than a system based on dairy purebred cows (Holmman et al., 1990). Some of the reasons for the predominance of this system in Latin American include Flexibility and plasticity to modify farmer s’ objectives and technological practices in response to the changes in markets and to the changes in relative prices of inputs and outputs (Estrada, 1993; Morillo and Urdaneta, 1998; Sere and de Vaccaro, 1985). Low requirements of additional infrastructure compared with beef production (Morillo and Urdaneta, 1998; Sere and de Vaccaro, 1985), and use of resources with low opportunity cost (Estrada, 1993). Relatively higher prices of milk compared to beef across the tropical region (Sere and de Vaccaro, 1985). Continuous flow of cash income from milk sales that cover the daily operational expenses and provide milk for household consumption, and the convenient sale of beef animals at the most appropriate moment (Estrada, 1993; Morillo and Urdaneta, 1998; Sere and de Vaccaro, 1985). Climatic conditions and higher production co st which restrict the efficiency of intensive dairy systems (Nicholson et al., 1994). Better reproductive and productive efficien cy (mortality levels in young animals and milk production per lactation) of cr ossbred cows than dairy purebred cows under tropical climate conditions (Plasse, 1992). Generally, DPCS is located between the Tropic of Cancer and the Tropic of Capricorn and approximately 1500m above sea level (Castillo, 1992; Sere and de Vaccaro, 1985; Vaccaro et al., 1993; Vargas, 1993). This system is found in a wide range of agro-ecological conditions. Soils vary from acid soils with low organic matter to basic soils with high levels of organic matter. Topography is variable, but most DPCS farms are located in the low land where the drainage can be deficient, depending on the texture of soil. Rainfall is seasonal and erratic with precipitation oscillating from 800 to 3500 mm per year (Garcia, 1993; Sere and de Vaccaro, 1985; Vacarro et al., 1993;

PAGE 25

10 Vargas, 1993) and a dry season ranging from 2 to 7 months (Ortega, 1996; Sere and de Vaccaro, 1985, Vaccaro et al., 1993). Average temperature oscillates between 20 and 28 C (Vaccaro et al., 1993; Vargas, 1993;). DPCS has been classified under different criteria such as size of the operation (small, medium and large), type of operation (family, commercial, crop-livestock, livestock, etc.), level of technology (low or traditional, intermediate, or high), type of beef product delivered into the markets (cow -calf, cow-yearling, and cow-steers) or any combination of criteria. DPCS traditionally presents the following characteristics: Farm size varies considerably from 20 – 1000 ha (Castillo, 1992). The presence of family farms is common; otherwise a farm may be managed by an administrator and visited occasionally by the owner (Sere and de Vaccaro, 1985). Cows are not specialized in milk production (primarily beef cows) but are used for milk, generally zebu cows or crossbred of zebu and criollo. The breeding system generally is not controlled natural mating. Milking is mainly done manually once a day with the calf present. Male calves are generally sold at weaning. Herd nutrition is principally based on natural pasture grazing. A minimum requirement of milking infrastructure prevails, and Farmers usually do not keep records The characteristics of DPCS vary according to the level of technology found on farms. It is possible to find farms adopting different technology levels such as a rotational grazing system of improved pastur es, forage conservation, pasture irrigation, artificial insemination, automated milking, feed supplementation strategies, use of cross breeding to improve milk production, and farm computerized accounts and technical

PAGE 26

11 (productive and reproductive) records. Likewise, productive and reproductive indices of this system vary according to level of technology applied and to the agro-ecological conditions where farms are located. Generally these indices are considered to be low, inefficient, and representative of developing countries by some technicians, cattlemen, and politicians (Plasse, 1992). It has been demonstrated, however, that to produce milk and beef with one type of animal in the tropics is more efficient than using specialized dairy herds (Plasse, 1992). Holmman et al., (1990) compared the profitability of the dual-purpose system based on Holstein crossbred cattle in low ecozones with that of systems using purebred Holsteins in the highlands. He showed greater profits for crossbred cattle than purebred Holsteins due, mainly, to high production costs in the latter case. Characteristics of the DPCS located in Zulia State Agro-ecological Conditions DPCS in Venezuela is mainly located in Zulia State, which provides more than 70% of milk and 30% of beef production nationally (Castillo, 1992; Delgado, 1989; Fernandez, 1992). This state is located in the northwest of Venezuela, on the border of Colombia, encompassing an area of 6,310,000 ha, and 21% (1,300,000 ha) adjacent to Lake Maracaibo (Fernandez, 1996). DPCS is found in both near rural and urban areas where road infrastructure allows transportati on of daily milk production from the farms to the processing plants. This system is classified mainly according to the beef product that producers deliver to market. Therefore, ther e are cow-calf, cow-yearling, and cow-steer systems with the cow-yearling system as the predominant system (Gonzalez, 1992; Graterol et al., 1987).

PAGE 27

12 The topography in Zulia state is diverse because it is surrounded by the Andes Mountains. Slopes decrease from the mountains to Lake Maracaibo. DPCS is mainly located in the flat land below 1000m above sea level. Climate varies from north to south and from the lake to the west and east. Rainfall constitutes the most variable element and increases from north to south and toward west and east of the lake. The most extreme variation occurs from north to south where rainfall ranges from 387 mm to 2600 mm per year. Average rainfall is approximately 1260 mm per year, and evaporation ranges from 2776 mm in the north to 1500 mm in the southern part of the state. Temperature oscillates from 26 to 29 C, reaching the minimum temperatures in January and the maximum in July. Soils are highly heterogeneous in terms of their physical and chemical characteristics. It is possible to find acid soil with low fertility and soils with good chemistry and physical conditions (Gonzalez, 1992). Vegetation, according to the Holdridge cla ssification, ranges from tropical desert weed in the north part to very humid tropical forest in the southern part of the state (Fernandez, 1996). DPCS is mainly located in very dry tropical forest, dry tropical forest and in humid tropical forest (Urdaneta et al ., 1995). Very dry tropical forest is below 600m above sea level, temperature ranges from 23 to 29 C and average rainfall between 450 to 950 mm per year. Dry tropical forest is between 400 and 1000 m above sea level with an annual average temperature that ranges from 22 to 29 C and an average rainfall of 1000 mm to 1800 mm per year. Tropical humid forest has a rainfall of 1800 mm to 3500 mm per year, with an average temperature of 24 C. Elevation in this area varies from sea level to 1000m above sea level. In the first two zones the rainfall pattern is

PAGE 28

13 bimodal, with two drought periods. The dry season occurs from December to March and from July to August, being more marked in the very dry tropical forest than the dry tropical forest. In humid tropical forest there is no dry season and rainfall is spread throughout the year. General Characteristics of the DPCS In general DPCS located in Zulia State presents the following characteristics: The average farm size is around 288 ha a nd 407 head (cows, heifers, steers, yearlings, calves, and bulls). The owner-manager is a literate person, and in 50% of the cases has an education higher than elementary school, while the chief operator is a trained person, and most of the labor is illiterate. Labor is hired and most of it is from a neighboring country (Colombia), or native Indians called “guajiros.” Labor supply is unstable, and there is high turnover. The type of animal used is a crossbred from crossing Bos taurus (Holstein, Brown swiss), Bos indicus (Brahman, Guzerat, Gyr) and Criollo. Milking is done mainly manually with the calf present, twice a day. The breeding method is commonly natura l uncontrolled, but there is a tendency toward using natural controlled mating and artificial insemination. Periodical vaccinations, to control foot a nd mouth disease, brucelosis, septicemia hemorragic, carbon symptomatic, and endo and ecto parasites, are regularly applied. Herd feeding is based on grazing of improve d species and supplement feed is used strategically. The traditional management of pasture is the non-systematic rotational stocking system. Guinea grass (Panicum maximum) is the main forage in least humid areas, and paragrass (Brachiaria mutica) and alemangrass (Echinochloa polystachia) in more humid areas. Pasture size is variable, making it difficult to manage the herd and pastures. Fertilization is not a common practice. Weed control is done irregularly using manual, chemical, and mechanical controls.

PAGE 29

14 Milk contributes to around 70% of total fa rm revenue. It can be higher or lower depending on whether the system is cow-calf, or cow-steer. Cow-yearling is the predominant system Labor (31%), supplement feed (18%), and machinery and building depreciation (24%) represent approximately 73% of total cost. Few producers keep accounting and technical records. Land and cattle constitute the main capital investment Resource Management and Animal Performance Weighted average ranges for some variables in different geo-political zones of Zulia state provide an idea of how the resources are managed (Table 2.1). These variables are stocking rate (AU ha-1), percentage of milking cows (PMC), relation cowbull (RCB), total cows and animal unit per person-year equivalent (P-YE), and hectares per person-year equivalent (Ha P-YE-1). Table 2.1. Management indices for DPCS in Zulia State SouthEastern CoastWestern CoastNorth-West Stocking Rate 1.2 – 1.80.7 – 0.90.7 – 1.00.49 – 1.1 Cows ha-1 0.720.390.38– 0.50.53 – 0.57 PMC 60606661 –70 RCB 24 – 30302929 Cow P-YE-1 20.016.916.612.7 AU P-YE-1 37.332.434.124.6 Ha P-YE-1 27.843.242.822.3 Source: Urdaneta et al., (1995) Weighted average productivity indices per ha, cow, and labor are shown in table 2.2. Levels of milk production per hectare and per cows range from 0.8 to 4.9 and 2.8 to 5.0 lt./day respectively. As illustrated in both tables there is high variability inside the state and among zones. This variability is due mainly to the different agro-ecological conditions present in the state, and to the level of technology adoption applied by the farmers.

PAGE 30

15 Table 2.2. Physical productivity indices for DPCS in Zulia State SouthEastern CoastWestern Coast North-West L ha-1 d 1.7 –4.90.8 –1.31.3 – 2.20.8 -2.9 Kg ha-1 yr 89.4 –225616499 L milking cow-1 d 5.0 –7.83.55.2 – 8.34.71 – 5.9 L total cow-1 d 3.1 –4.12.0 – 3.473.5 – 4.32.8 – 5.0 Kg beef total cow-1 yr 207159166172 L Person-Year Equivalent-1 26585214702645123701 Kg beef P-YE-1 4136269427492197 Source: Urdaneta et al., (1995) Technical Efficiency In economic terms, a producer is considered efficient if a higher output cannot be obtained from a given set of inputs and t echnologies and if this output cannot be produced at a lower cost. The first part of this concept refers to the technical efficiency concept, and the second part is related to the allocative efficiency or price efficiency definition. Different approaches have been used to measure efficiency. Among the common approaches used are the index approach and Farrell‘s index of efficiency. The index approach is based on the use of partial factor productivity (PFP) and total factor productivity (TFP) indices. The PFP index measures the productivity of an individual input or factor (e.g., labor, capital, land, etc.) and is calculated as the ratio between output and input. These indices measure the intensity of use of individual inputs over outputs but they do not consider the effects of the other inputs. Due to these disadvantages, indices of TFP were developed as the ratio of aggregate output to aggregate inputs. Input-output aggregation can be assessed by linearly weighting inputs and outputs by their prices, or geometrically using the factor shares as the weights criteria. Both indices (linear and geometric) present the “index number problem” (Peterson and Hayami, 1977).

PAGE 31

16 They assume that observed output is the fron tier or best practice ignoring inefficiency, technical and allocative (Grosskopf, 1993). Alternatively, frontier models were introduced to measure efficiency. They are based on the total factor productivity concept and they do not present the “index number problem” (Bravo-Ureta, 1986). Frontier models differ from average production models in that they consider the maximum possible and not maximum average output. The development of frontier models to measure the efficiency begins with the work of Farrell (1957). This author decomposed ec onomical efficiency (EE) into technical efficiency (TE) and allocative efficiency (A E), using unit isoquant and assuming constant returns to scale to characterize the frontier technology (Frsund et al., 1980). Figure 2.1 illustrates Farrell’s model for the given function 2 1, x x f y where UI represents the unit isoquant [ y x y x fi 2, 1 ]; Xa is the actual combination of inputs ( xa1/y, xa2/y ) to produce unit output UI; Xb represents the best combination of inputs ( xb1/y, xb2/y ) to produce unit output UI; Xc and Xd are the minimum cost combination of inputs ( xc1/y, xc2/y ) to produce unit output UI. P2/P1 is the ratio of input prices, representing the isocost line; TE is calculated through the ratio OXb / OXa; AE is defined as the ratio OXd / OXb; and EE is determined through the interaction TE * AE, which is equal to the ratio OXd / OXa. After Farrell’s model, several techniques were suggested to measure the efficiency, such as mathematical programming or data envelopment analysis (DEA), and econometric analysis. The primary advantage of DEA is that it does not require one to impose a functional form for technology on the data, but it has the limitations that the estimates do not have statistical properties, and the analysis is sensitive to extreme

PAGE 32

17 values. The econometric approach overcomes these problems but imposes functional forms on the data, which sometimes are over restricting. (Bauer, 1990). Figure 2.1. FarrellÂ’s model Frontier models have been classified according to the combination of two main criteria. The first criterion classifies the models into parametric and non-parametric according to whether a functional form is or is not imposed on the data. The second criterion classifies the models into deterministic and stochastic according to how the frontier is estimated. The deterministic approach assumes that any deviation from the frontier is due to inefficiency, and this inefficiency is measured through the error term ( u ). The stochastic approach on the other hand uses a composed error ( v and u ), where part of the deviation ( u = one-side disturbance) from the frontier is due to inefficiency and the other part to the statistical noise or random error ( v ) (Bauer, 1990; Bravo-Ureta and Pinheiro, 1993; Bravo-Ureta and Rieger, 1990; Frsund et al., 1980). Given these two criteria, Frsund et al. (1980) classified the models as deterministic non-parametric, deterministic parametric, deterministic statistical frontier models, and stochastic frontier approach.

PAGE 33

18 Deterministic non-parametric approaches, as in FarrellÂ’s model, are estimated through DEA and have the advantage of not imposing a functional form on the data, but the disadvantage of assuming constant returns to scale is difficult to overcome. Another disadvantage of these models is that they are very susceptible to outliers and measurement error, and the parameters do not have statistical properties. Deterministic parametric frontier models are also estimated through DEA. These models generally use a simple functional form with the ability to overcome the problems of constant returns to scale, but the susceptibility to outliers and parameters without statistical properties remains (Bravo-Uureta, 1986; Frsund et al., 1980). Kumbhakar and Lovel (2000) provided a general specification for a linear programming and a quadratic programming model, which are the most popular models. A linear specification is Min i j ji j iy x u ln ln0 (2.1) st. j i ji jy x ln ln0 where yi is output of the ith firm; xji is the amount of input j used in firm i ; j represents the parameters to be estimated, and ui is one-side error of the ith firm. The TE for the ith firm can be estimated by exp-u. The quadratic model is given by minimizing the square of the residual subject to the same constraints. Min i iu2 (2.2) St. j i ji jy x ln ln0

PAGE 34

19 Deterministic statistical frontier models m easure inefficiency through the error term considering only one-side error ( u ). The difference between this approach and the previous is that it follows the econometric framework. Assuming a production frontier like yi = f ( xi, ) TEi, technical efficiency can be expressed as a ratio between yi / f ( xi, ), which is the conventional measure of total factor productivity, where the values of TEi are between 0 and 1, is the vector of coefficient of inputs xi , and “i” is the number (n) of farmers in the sample. When natural log is applied this model becomes linear ln yi = ln f ( xi,) + ln TEi (2.3) = ln f ( xi,) ui where ui should be greater than zero. Since ui = ln TEi, TEi uiexp, “ ui” represents the deviations from the frontier function, which is assumed to be identical and independently distributed with non-negative mean and finite variance. In this approach the distribution form of the error term can be assumed or not. If an explicit form is assumed, the frontier should be calculated through the maximumlikelihood technique; otherwise, OLS can be used because it provides the best linear unbiased estimators. The OLS intercept should be corrected because it is biased downwards. Two approaches have been suggested, one called corrected ordinary least square (COLS), and the other called modified ordinary least square (MOLS). The MOLS is less orthodox than COLS (Greene, 1997), and this is the reason it will not be discussed in this presentation. The COLS is shown in the following equation: yi = (0 + umax) + j xij – ( ui umax) (2.4)

PAGE 35

20 The constant term has been shifted upward by an amount equal to the maximum positive error value ( umax) (Bravo-Ureta and Rieger, 1990), and a proof of the consistency of COLS estimators appears in Greene (1980). One of the disadvantages of COLS is that the asymptotic distribution of the corrected constant term is unknown, and it is not independent of the distribution assumed for the error term (Frsund et al., 1980). If a specific distribution is assumed for the error term, the maximum likelihood technique is applied. Several distributions have been used such as exponential, halfnormal, and gamma (Bravo-Ureta and Rieger , 1990). A Gamma distribution as proposed by Greene has been the most used form (Bravo-Ureta and Rieger, 1990) because it is a more flexible distribution than the others and it has desirable asymptotic properties (Greene, 1980). The advantage of deterministic statistical frontier models over mathematical programming is that the estimates have statistical properties. Nevertheless, in the former all deviations from the frontier are attributed to the inefficiency (Bravo-Ureta and Rieger, 1990; Greene, 1997). The stochastic frontier approach differs from the statistical approach in that the error term is decomposed in to random error ( v ) and one-side error ( u ) terms. One-side error represents the inefficiency of firms, and the random error (both-sides) represents the random effect uncontrolled by the firms. This approach was introduced at the same time by Aigner, Lovel, and Schmidt (ALS; 1977); and by Meeusen and van den Broeck (1977). The general formulation can be expressed as ii ix f y exp ; (2.5)

PAGE 36

21 where iiivu; vi is unrestricted and represents the random error, may take any value, and normal distribution is generally assumed. ui is the same as before. Both components of the error term are considered to be independent and identically distributed (iid) across observations. Equation (5) can be rewritten as i v i iTE x f yiexp ; (2.6) where TEi can be defined as the ratio between yi / iv ix f exp ;. yi represents the observed output, and [iv ix f exp ;] represents the stochastic production frontier, which can be divided in two parts; a deterministic part [ f ( xi, )] common to all producers, and a random effect (ivexp) which is specific for each producer. Taking the natural log of equation (2.5), gives i i i iu v x f y ; ln ln (2.7) The estimation of ui in this case should be made indirectly by considering the mean or the mode of its conditional distribution i i iu v u | (Jondrow et al., 1982). To calculate the one-side component of the error ( ui) several distributions have been assumed such as half-normal, exponential, truncated normal, or gamma distribution. Empirically, each one has lead to different results of sample mean efficiencies (Greene, 1990). There is not concrete evidence indicati ng that the ordinal ranki ng of the efficiency scores is sensitive to the distribution of the one side error according to Kumbhakar and Lovel (2000). These authors, based in the work of Greene (1990), found the lowest rank correlation coefficient between exponential and gamma distributions when they were compared to half normal and truncated normal distributions. If this behavior can be

PAGE 37

22 generalized, the authors argue that there is no reason to use a more flexible distribution than a more simple distribution. Most of the empirical work reported has followed the half-normal distribution. The main advantage of this approach is the introduction of a disturbance term representing the noise, error measurement, and exogenous shock beyond the control of the production unit in addition to the efficiency component (Bravo-Ureta and Rieger, 1990). Theoretically, the stochastic approach should be superior to the other approach, but empirically there is no conclusive evidence indicating this tendency. The stochastic approach should yield a higher efficiency measurement than the deterministic approach, however, there is evidence showing the opposite (Banker et al., 1986 cited by Schmidt 1986, Bravo-Ureta and Rieger, 1990). The correlation between the parameters from different approaches has been high, indicati ng that the ordinal rank from the different approaches is the same and it is independent from that used (Bravo-Ureta and Rieger, 1990). Different functional forms have also been used to analyze farm efficiency, and the Cobb-Douglas functional form is most used. As pointed out by Schmidt (1986) and Greene (1997), more functional forms avoid th e distortion and yield higher efficiency measures. Greene (1997) found collective st atistical significance for 15 coefficients when translog and Cobb-Douglas production f unctions were compared, but the changes in the estimators were relatively minor. Kopp and Smith (1980) concluded that the impact on the efficiency index was small when they compared three different functional forms for the production function (Cobb-Douglass, CES, and translog). Likewise, Bauer

PAGE 38

23 (1990) pointed out that as we move away from the Cobb-Douglas functional form, estimation becomes more complicated. Determinants of Technical Efficiency A survey of agricultural efficiency analysis in developing countries made by Bravo-Ureta and Pinheiro (1993) revealed that human capital is an important factor to explain the variation in farm efficiency. Characteristics such as farmer education and experience have shown a positive and statistically significant effect on farm efficiency. Likewise, contact with an extension agent and access to credit have also had a positive effect. Other variables tested include farm size, farm region, tenure, irrigation, and fertilizer. Traditionally, farmer education and experience have been considered as proxy variables to measure the managerial ability of farmers, where ability is defined as “the capability to perceive, interpret, and respond to new events in the context of risk.” (Schultz 1981, cited by Feder et al., 1985). It is assumed that farmers with formal education have more ability to adopt technology and to reduce the cost of adoption. Experience, measured as number of years in the business, is a variable that can increase the efficiency of resources in the production process and overcome lack of education. It is expected that farmers with more experience are more efficient because they can better evaluate the use of different technologies and are better organized than farmers with less experience. This efficiency, however, can decrease as the producer becomes more risk averse with age, especially when the results of introduction of new technologies are expected to be obtained in the long run. Contact with extension agents is another variable that can explain variation in efficiency. Producers in contact with extension agents have more access to market and

PAGE 39

24 technology information than producers without contact. Access to information helps to reduce the risk and uncertainty of the business. Access to credit has been considered as another importan variable responsible for gain in efficiency. Generally, the development of new technologies to improve the firm efficiency has been associated with the use of new inputs, which require a higher working capital or a large initial investment. Farmers who have had access to credit, to overcome the problems of capital constraints, are expected to be more efficient than those who have not experienced this problem. Some studies, however, have found that rational farmers overcome this restriction. Related to farm size, different measures have been used as indicators for this variable. In the case of dairy farms in developed countries, the number of cows is the variable commonly used. Empirical work conducted in these countries, indicates a positive relationship between farm size and efficiency (Bravo-Ureta and Rieger 1990; 1991; Kumbhakar et al., 1991; Neff et al., 1991), a ttributed to the scale effect (Tauer and Belbase, 1987; Zepeda, 1994). Opposite results have been found in developing countries where smaller farmers tend to be more productive than larger farmers, and where area (ha) is the variable frequently used to measure the size effect. Some reasons accounting for this opposite behavior are imperfections in land, labor and capital markets, and the quality of land owned by different farm size operations (Langedyk, 2001). Land tenure is another factor that has been utilized to explain the variation in efficiency. The relationship between efficien cy and type of tenure has not been made clear. An explanation is that tenants speci alize in technologies that require less fixed capital, while owners use technologies that require more fixed capital. Although these

PAGE 40

25 are the main variables used to explain the differences in efficiency, there is wide room to test different variables to answer the variation found in efficiency. Information about efficiency analysis for livestock farms in developing countries is scarce when the total factor productivity concept is used. Most of the empirical work in livestock farms is based on partial factor productivity indices such as output (yields or gross output) per hectare or per cow. These productivity indices generally have been calculated by using farm size in order to compare the productivity according to different sizes or they have been compared using di fferent production systems. These indices do not indicate the level of farm efficien cy for a given input set and technology. The study of methodologies that help to expl ain variation in efficiency has not been as rich as the study on methodologies developed to estimate efficiency. These methodologies can be classified in two main approaches: the first approach is a two step procedure where indices of technical efficiency are determined in the first step using the method outlined above with production being regressed on identified inputs ( xi). Then in the second step, technical efficiency indices are regressed against a set of socio-economic explanatory variables ( zi) that are expected to influence TE , where ordinary least squares (OLS) is commonly applied. The following equations show the first approach (a two step procedure): i i i iu v x f y ; ln ln (2.8) i i i iz z u (2.9) where all the variables are as previously defined except for zi which represents the vector of socio-economics variables, and is a vector of the coefficients of socio-economic variables.

PAGE 41

26 The second approach, as proposed by Kumbhakar et al., (1991), estimates the coefficients of the production frontier, the efficiency values, and the determinants of TE in one step. In this case, the determinants of the efficiency values are incorporated into the one sided error. i i i iu v x f y ; ln ln, (2.10) i i ie z u (2.11) where ei is the random error. Three main reasons are considered in the use of the second approach. First, if the socio-economic explanatory variables affect e fficiency directly, they should be included in the production frontier because the estimates of the parameters in the production frontier and the technical efficiency could be inconsistent. Second, utilization of OLS in the regression step is not appropriate (unless the dependent variable has been previously transformed or different estimation techniques such as limited dependent variables can be employed) because the dependent variable (TE) is bounded by zero and one. Lastly, the meaning of the residual is not clear in the regression step. Advocates of the first approach indicate that if the socio-economic va riables explain the variation in efficiency and do not have a direct impact on the structure of the production frontier, the two-step approach can be used (Kalirajan, 1991). McCarty and Yaiswarng (1992) compared the two approaches, and they found little difference between them.

PAGE 42

27 CHAPTER 3 DATA AND DESCRIPTIVE STATISTICS Information describing the survey met hodology and the different variables that characterize the farmers, farms, and the techni cal and managerial aspect of the farms is presented in this chapter in two major s ections. The first section is the survey methodology, which is related to the selec tion of the population, sample, and data collection. It also includes the approaches us ed to define the different variables. The second section, divided into farmer-characteris tic variables, farm variables, technical and managerial variables, relates the descriptive analysis of the data. Central tendency measures are used to describe the continuous variables and frequency analysis to the descriptive variables. The information obtained and processed in this chapter is vital to clear understanding of how the system operates. It also will be used as input for the production frontier models and for the model th at will explain variation in efficiency among the farms to be developed in chapter 4. Survey Methodology Data used in this research comes from a survey conducted by the Unidad Coordinadora de Proyectos Conjuntos of the Universidad of Zulia, Zulia State, Venezuela. The survey objective was to determine milk and beef production costs in the dual-purpose cattle system. Information related to economic variables such as gross revenue, input expenses, and capital, plus the questionnaires to process the social, financial, technical, and management information was obtained from this unit.

PAGE 43

28Sample Selection Lack of information about the number of dual-purpose farms in Zulia State justified the utilization of farms delivering milk to the main milk processors in the State as the study population. Dual-purpose farms delivering milk to the cheese industry were not considered for this study. Zulia State was divided into four geo-political areas, representing different agroecological conditions. Milk plants, in each area, supplied the information about the number of farms and the volume of milk delivered daily by each farm. A total of 2,516 farms represented the population and they were distributed by zone as follows: Zone 1 (South) corresponding to the southern area of Zulia State, included three counties (Catatumbo, Colon, and Sucre). A total of 790 DPCS or farms were located in this predominantly humid tropical forest zone. Zone 2 (Eastern coast) characterized as very dry and dry tropical forests included 479 farm representing five counties (Baralt, Valmore Rodriguez, Lagunillas, Cabimas, Santa Rita, and Miranda). Zone 3 (Western), dominated by dry tropical forest, included 604 farms from two counties (Rosario de Perija and Machiques de Perija). Zone 4 (North-West) an area characterized by very dry and dry tropical forests included 643 farms representing four counties (Canada de Urdaneta, Jesus Enrique Lozada, Mara, and Paez). To select the sample, the population was stratified according to milk production per day per farm per zone. Four levels of milk production were established: level 1-between 50 and 100 L per day; level 2-from 101 to 550 L per day; level 3-ranging from 551 to 2900 L per day; and level 4-greater than 2,900 L per day. A random sample of 127 (5%)

PAGE 44

29 from a population of 2,516 farms was selected and distributed by zones for 40, 24, 31, and 32 farms in zone 1, 2, 3, and 4, respectively (Table 3.1). Table 3.1. Population and sample distribution of DPCS by location Location Population (# of farms) Sample (# of farms) Vegetation Zone 1 (South) 79040Humid tropical forest Zone 2 (Eastern Cost) 47924Very dry and dry tropical forest Zone 3 (Western) 60431Dry tropical forest Zone 4 (NorthWest) 64332Very dry and dry tropical forest Total farm 2516127 Source: UCPC survey, Zulia, Venezuela, 1994. Replacement farms were also selected by level of production and by zone because some sample selection restrictions were imposed. These restriction were: a) cattle enterprise as a main activity on the farm; b) willingness of farmers or managers to participate and provide the information required for the study; c) farmers or managers who have managed the farm for at least one year in order to provide the economic information for the year under study; d) the farm had to be consolidated (developed) farms and the typical weather conditions for year the studied had to be typical (farms that suffered flooding, fires, or legal problems were not used); and e) road accessibility to the farm. Data Collection To obtain information for the 1994 economic period, a questionnaire (Appendix A) was developed to explore: i ) owner demographics (sex, age, instruction level, residence, permanency on the farm, experience as producer) and production unit (production systems, years in operation etc.); ii ) production factors (land tenure, inventory of land, pasture and crops, cattle, buildings and installations, and machinery and equipment); iii )

PAGE 45

30 input and output activities ( types and amount of fertilizer, herbicides, concentrate feed, diesel and oil, milk production per day in the dry and rainy season, type and number of animals sold, etc.); and iv ) financial, technological and managerial behavior. Personal interviews by trained personnel were used to apply the questionnaire to farm owners or managers. Prices of assets such as land, pastures and crops, buildings, machinery, cattle, and price of others inputs were obtained from th e database of agricultural inputs stored by UCPC, which maintains and updates these data periodically since most of the farmers do not keep their own accounting records. Valuation Methods An array of methods was used to assess the different capital components. Land was assessed using market prices. Cattle were valued according to their productivity capacity using the UCPC method. Replacement method was used to assess buildings and machinery. Forage and crops were valued using establishment costs. Total capital investment in the farms was calculated disregarding debt in order to compare farms. Likewise, finance costs and the opportunity cost of capital were not included in production cost. Labor expenses were calculated using the salary paid by the farmers plus benefits according to the labor law. Output prices, to calculate revenues from milk and beef, were provided by farmers. Milk consumed by the calves, around 10% to 20% of total milk production, was exclude in the calculation of milk revenue. In summary, the completed survey with data usable for the efficiency analysis included 127 farms. A total of 150 questions were answered, and the overall quality of the information was judged acceptable.

PAGE 46

31Descriptive Statistics The information gathered by the survey wa s classified in four major variables related to farmer and farm characteristics, and technical and managerial aspects. Some of the variables are categorical and frequenc y analysis was applied to describe the distribution of the data across these variables. Other variables are quantitative and central tendency so dispersion measures such as mean, maximum and minimum values, and coefficient of variation were utilized for their analysis. Farmer Characteristics In this section, demographic variables such as sex, age, instruction level, presence of farmers on the farm, and experience are used to describe and to characterize owners and managers of DPCS. Owner characteristics The owners of the dual-purpose cattle farms in Zulia state are mainly Venezuelan citizens (99.21%) where 87.5% are male and 12.5% are female (Table 3.2). More than 50% of the owners are over 50 years old, an important factor to consider for technology adoption. As age increases farmers generally become more risk averse to adoption of new technologies that can improve the farm efficiency. Generally, level of instruction has been considered as a proxy variable for education, which has been linked to the process of technology adoption. High levels of education of farmers is associated with lo wer cost of learning new technology. More than 90% of the farmers surveyed are literate persons, and about one half of them have a level greater than elementary school. The category of “others” (table 3.2) represents farmers with vocational and technical education.

PAGE 47

32 Table 3.2. Owner characteristics of DPCS NumberPercentage Age (years) < 21 10.8 21-30 97.4 31-40 1915.7 41-50 2621.5 51-60 2823.2 > 60 3831.4Total farmers 121100.0 Instruction level Illiterate 108.3 Read and write 108.3 Elementary 4537.5 High school 2420.0 College degree 2117.5 Other 108.4Total farmers 120100.0 FarmerÂ’s Residence Farm 3528.9 Town near to the farm 5343.8 Town far from the farm 3226.5 Other 10.8Total farmers 121100.0 Presence on the farm Permanent 4235.0 Daily visit 4436.7 Every two days 119.2 Twice a week 86.7 Other 1512.5Total farmers 120100.0 Experience as Producer (years) None 00 < 5 43.3 Between 5-10 75.8 Between 11-15 1512.4 Between 16-20 1411.6 > 20 8166.9 Total farmers 121100.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 48

33 Related to the presence of farmers on farms, more than 70% of farmers stay there or visit the farm daily during the week. Approximately 30% of farmers live on the farms and almost 45% live in nearby towns, indicating that farmers closely observe farm activities. This suggests that often farms are more than a business; they are likely to be homes. This characteristic when considered together with age of the farmers plus a low percentage requesting credit suggests that most DPC farmers are risk averse. Evidence of risk aversion may be a constraint for adop tion of new technologies that could improve farm efficiency. Years of experience become an important factor to consider if management is included into the production function. It is assumed that more experimental managers spend less time and less cost in administrative activities. Dual-purpose cattle farmers can be considered as experienced producers because 80% of them have more than 20 years of experience, and only 3% of the farmers have less than five years of experience. Manager characteristics Dual-purpose cattle farms were managed by the owners (66% of the cases studied) or either a close or far rela tive of the owner (34%) (Table 3.3). Close to 90% of the managers were over 30 years old and in mo re than 50% of the cases their age was between 30 and 50 years old. The educati on level of the managers was elementary education (40%), high school (20%) or bachelor diplomas (20%). The majority of the managers were experienced with more than 20 years in 50% of the cases. The residence of these managers is on the farm or in a town nearby for a combined 82% (Table 3.4).

PAGE 49

34 Table 3.3. Who is the manager of DPCS ? Number Percentage Owners 121 100.0 Owner managers 80 66.1 Hired managers 41 33.9 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.4. Manager characteristics of DPCS in Zulia State Number Percentage Age (years) < 21 1 0.8 21-30 15 12.4 31-40 41 33.9 41-50 27 22.3 51-60 17 14.1 > 60 20 16.5Total managers 121 100.0 Education level Illiterate 5 4.1 Read and write 11 9.1 Elementary 46 38.1 High school 24 19.8 Bachelor 23 19.0 Other 12 9.9Total managers 121 100.0 Experience (years) None 1 0.8 < 5 6 4.9 5-10 14 11.6 11-15 22 18.2 16-20 17 14.1 > 20 61 50.4Total managers 121 100.0 Manager residence Farm 45 37.2 Town near from the farm 54 44.6 Town far from the farm 21 17.4 Other 1 0.8Total managers 121 100.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 50

35Farm Characteristics In this section a descriptive analysis of the farms is presented including general characteristics, organization and management of resources, and production and productivity. General characteristics More than 85% of the farms are 20 years or more years old indicating that most of the farms are consolidated and stable (Table 3.5). Seven percent of farms are composed of enterprises besides cattle, including crops such as plantain, corn, cassava, and in some cases, vegetables. Three main systems are considered to classify DPCS; these are cowcalf, cow-yearling, and cow-steer. This classi fication is based on the percentage of total revenue that comes from the sale of milk and beef animals. When revenue from milk represents more than 80% of total sales revenue the system is considered to be cow-calf; if revenue from milk is between 70 and 80%, it is designated as cow-yearling, and when milk revenue represents less than 70% the system is called cow-steer. As can be seen in table 3.5, cow-yearling predominates with 85% of the farms. Most farms present a flat topography (64%) with good drainage (89%). Access to the farms is by roads, which in most cases (94%) were in good condition as were the internal roads (70%). Water came from different sources depending on the farm location, most important being a combination of rainfall, deep wells, and rivers or small streams of water. Rainfall was stored in small ponds locally known as “jaguey.” In general to some degree 88% of the farms depend on rainfall.

PAGE 51

36 Table 3.5. General characteristics of DPCS farms AnswerPercentage Age of Farm (years) Between 15 00.0 Between 6 – 10 43.3 Between 11 – 15 54.1 Between 16 – 20 54.1 > 20 10586.8 Others 21.7Total farms 121100.0 Main activity Cattle 11393.4 Cattle – crops 86.6Total farms 121100.0 Main Cattle Activity Cow – calf 54.1 Cow – yearling 10284.3 Cow – steer 1411.6Total farms 121100.0 Topography Flat 7763.6 Rolling 65.0 Flat-rolling 2924.0 Ravine 00.00 Flat – ravine 43.3 Rolling – ravine 10.8 All 43.3Total farms 121100.0 Drainage Good 10789.2 Deficient 10.8 Mixed 1210.0Total farms 120100.0

PAGE 52

37 Table 3.5. Continued. AnswerPercentage Road conditions External Good 11494.2 Regular 75.8 Poor 00.0 No access 00.0 Total farms 121100.0 Internal Good 8570.3 Regular 3226.5 Poor 10.8 No roads 32.4 Total farms 121100.0 Water available Rainfall 119.1 River 43.4 Rainfall-river 97.4 Well 00.0 Rainfall-well 4133.9 River-well 54.2 Rainfall-river-well 2419.8 Aqueduct 00.0 Rainfall-aqueduct 32.5 Rainfall-river-aqueduct 10.8 Well-aqueduct 10.8 Rainfall-well-aqueduct 10.8 River-well-aqueduct 10.8 Others (jaguey) 10.8 Rainfall-others 129.9 Rainfall-river-others 10.8 Well-aqueduct 10.8 Rainfall--well-others 54.2 Total farms 121100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Organization and resource management Land. Most livestock and agricultural production activities require land. It has an intrinsic potential that can be increased through fixed investment such as deforestation, leveling, drainage, soil amendment, etc., which are difficult to disaggregate once they

PAGE 53

38 have been made (Sposito, 1994). Land is also a scarce factor that has a commercial value. This value will depend on land quality, accessibility to markets, and human endeavor. Farmers regard land as a saving instrument. They expect to receive remuneration equal or greater than equivalent money invested in alternative activities (Guerra, 1992). Therefore, it is important to define tenure, use, and productivity of the land. Land tenure generally conditions the type of investment in the farms and it could be considered a credit constraint depending on the type of tenure. When the land is not owned, farmers limit fixed investment inside their farms. In the case of DPCS, two thirds of the land where the farms are settled belongs to the government, and only one third to farmers (Table 3.6). Table 3.6. Land tenure AnswerPercentage Government 7763.6 Farmers 4033.1 Government-farmers 21.7 County 10.8 Others 10.8 Total farms 121100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Farms had an average size of 322 hectares (ha.) with a coefficient of variation of 139, and a development index of 93.39. This index represents the relationship between the land used in cattle activities and commercial crops with respect to the total acreage (Table 3.7). Most of the land (90%) has been used in cattle activities, suggesting that to increase production, farmers should increase productivity and efficiency of the factors and inputs. Approximately 98% of the grazing lands are planted to improved forages and less than 8% of the grazing land is irrigated. The crop area dedicated to animal feed is

PAGE 54

39 practically marginal, indicating that the main source of feed for the system is based on forages where production and forage availability depends on rainfall distribution. Table 3.7. Use of the land Mean (Ha)C.V.Max Min Percentage Cattle area 287.7139.33538.1 9.0 89.4 Commercial area 2.7299.545.0 0.0 0.8 No used area 31.4414.41200.0 0.0 9.8 Total area 321.7138.03538.1 9.0 100.0 Development index 93.414.5100.0 40.0 Use of Cattle Area Forages area 282.7137.83398.0 7.0 98.3 Irrigated forage area 21.1393.3696.0 0.0 7.3 Crops area 2.0360.050.0 0.0 0.7 Buildings and roads area 5.0278.7140.1 0.5 1.7 Cattle area 287.7139.33538.1 9.0 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Cattle. The animal type used in this system is an animal product of rotational and non-systematic breeding of Bos taurus such as Holteins and Brown Swiss, Bos indicus mainly Zebu, and native animals. This crossbred animal is called Mosaico Zuliano or Mosaico Perijanero (Isea and Rincon, 1992). Under lowland tropical condition this crossbred has demonstrated more efficiency in producing milk than specialized dairy animals (Holmann et al., 1990; Castillo, 1992). Herd composition and structure is expressed in head (number of animals) and animal units (Table 3.8). Animal unit (AU.) is a standard measure that allows comparing different species or types of animals based on the amount of feed that an animal consumes, which is also related to their corporal surface or weight. In this study, a cow (400 kg.) represents 1AU, and other animals receive a value according to this cow equivalent. For example, one bull represents 1.5 AU, a young bull 1 AU, a steer and a heifer are 0.8 AU, a female and a male yearling represent 0.6 AU, and a calf 0.3 AU.

PAGE 55

40 Thus, the distribution of different categories of animals according to the age, sex, and weight of animals describes the average herd (Table 3.8). Table 3.8. Average cattle distribution per farm MeansC.V.Max Min Percentage Average herd structure (Heads) Bulls 4.7126.449 0 1.1 Milking cows 113.0154.61650 9 27.7 Dry cows 43.9152.4624 0 10.8 Heifers 35.8194.5702 0 8.8 Steers 14.5470.1702 0 3.6 Female yearlings 52.8164.0600 0 13.0 Male yearlings 34.8194.2504 0 8.5 Young bulls 1.3336.427 0 0.3 Calves 106.6145.41498 9 26.2Total heads 407.2158.26240 31 100.0 Average herd structure (AU) Bulls 7.0126.473.5 0 2.4 Milking cows 113.0154.61650 9 39.0 Dry cows 43.9152.4624 0 15.2 Heifers 28.6194.5562 0 9.9 Steers 11.6470.1552 0 4.0 Female yearlings 31.7164.0360 0 10.9 Male yearlings 20.9194.2302 0 7.2 Young bulls 1.3336.427 0 0.4 Calves 32.0145.4449 3 11.0Total animal unit 289.9160.34540 21 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Stocking rate (SR) is one of the main indices related to herd management and organization. It indicates the intensity of an imal units per ha of pasture during the year. Farms with high stocking rates generally imply better pasture management than those with low SR. The average SR for these farms was 1.21 AU. Ha-1, with a maximum of 3.86 AU (Table 3.9), which suggests there is a room for improvement.

PAGE 56

41 Table 3.9. Herd management and organization indices of DPCS MeansC.V. Max Min Total cow pasture-ha-1 0.759.0 3.0 0.2 Stocking rate 1.254.7 3.9 0.3 Calves milking cow-1 1.010.8 1.2 0.3 Milking cow total cow-1 0.713.5 100.0 48.8 Total cow farm-1 156.910.8 2274.0 12.0 Source: UCPC survey, Zulia, Venezuela, 1994. Labor. One of the main factors to consider in a farm is labor efficiency because it affects the profitability of the business. La bor efficiency is difficult to compare among farms because it depends not only on the skill and training of the labor used but also on the amount of mechanization, size of farm, class of labor (family, hired), and other factors (Kay 1981). A measure called person-year equivalent (P-YE) has been used to compare labor among farms. This index homogenizes the workday of hired and family labor across the year (Table 3.10). The average person-year equivalent and milkers per farm are 12.39 and 5.56 respectively. Table 3.10. Labor utilization indeces of DPCS MeanC.V. MaxMin P-Y E farm-1 124159.0 200.01.2 Cattle area P-Y E-1 25.557.8 79.83.8 AU P-Y E-1 23.736.4 53.68.6 Total Cow P-Y E-1 13.134.9 29.35.6 Milking Cow milker-1 20.259.0 133.34.8 Milkers farm-1 5.6143.4 74.01.0 Source: UCPC survey, Zulia, Venezuela, 1994. Capital. Based on the difficulty to obtain farmer debt information, capital was considered free of debt. This translates into total capital being equivalent to total assets. Rate of return on capital and other indices based on capital were utilized as measures of profitability and capital intensity to compare the farms. Composition and structure of capital show land and cattle as the main investment representing more than 65% of the

PAGE 57

42 total capital (Table 3.11). A similar tendency was reported by Fernandez 1992 and Sere and de Baccaro 1985. Table 3.11. Composition and structure of capital per farm (Bs.) MeanC.V.MaxMin Percentage Land 33255500206.57430010001280000 38.5 Forages 12933800110.6639389000 15.0 Buildings and installations 9708887122.3115071000698271 11.3 Machinery and equipment 5388942109.347090000206983 6.2 Cattle 25021200157.93852830001851500 29.0 Total assets 86308400152.613496000006658981 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Indices of capital intensity are presented in table 3.12. These indices represent a ratio between capital and production factors such as land, labor, and cattle. They give an idea about the efficiency of the capital invested. These indices could be used to compare the efficiency of capital through the time, among type of farms, or industries. In the case of capital per labor, it will vary by farm type and by the amount of laborsaving technology used (Kay 1981). Table 3.12. Intensity of capital per farm (Bs). MeanC.V.Max Min Total capital Cattle area-1 34856540.9963298 152076 Total capital Total cow-1 59992437.62114495 262963 Total capital AU-1. 33180935.41012575 157702 Total capital P-Y E-1 755820141.319675000 3026809 Source: UCPC survey, Zulia, Venezuela, 1994. Production and productivity The dual-purpose cattle system has been ch aracterized by flexibility to adjust to changes in output and input prices. Farmers ma ke some internal adjustments to maintain farm productivity and profitability according to the changes in output and input prices. For example, changes made by farmers relate d to the herdÂ’s breeding program depend on which product (milk or beef) generated the greatest benefits. Approximately 70% of the

PAGE 58

43 total revenue comes from milk sales, and around 30% from the beef sales (Table 3.13). This is the typical revenue structure of the sy stems located in Zulia state. However, milk percentage ranges from more than 80% to less than 60% of the total revenue depending on the system used: cow-calf, cow-yearling, or cow-steer. Table 3.13. Production structure and composition per farm Milk Beef Total Revenue (L) (Bs) (kg) (Bs) (Bs) Means 2390359801750269833712456 13515000 C.V. 143.6143.6139.3148.5 139.4 Max 254700010697400033300048870000 155844000 Min 219008760001100132000 1230000 Percentage 72.527.5 100.00 Source: UCPC survey, Zulia, Venezuela, 1994. Productivity is the relation between output a nd production factors or inputs. It can be measured in physical or in monetary terms. In this study, total revenue, gross product, and net product were used as monetary measures to calculate farm productivity. Total revenue equals milk revenue plus beef revenue. Gross product was calculated by subtracting the value of physical inputs and services used in the productive process from total revenue. It represents the return to management, labor, and assets. Net product is the difference between gross product and depreciation. Physical and monetary measures by production factor (land, cow, labor and assets) represent partial productivity indices because they measure the productivity of a particular factor or input (Table 3.14). Th erefore, farms can present low productivity in one factor and high in another factor. Average milk production per cow-year is approximately 1556 L., which is equivalent to 4.79 L* cow-day-1. This value is higher than those reported previously (Fernandez-Baca 1995; Holmann and Lascano 1998; Sere and de Vaccaro 1985). However, this level of production was below the potential value

PAGE 59

44 indicated by Holmann and Lascano 1998, and by Urdaneta et. al., when the available technology was introduced to this system. A similar trend is observed when productivity by hectare is analyzed. Table 3.14. Productivity indeces of DPCS in Zulia State MeanC.V.Max Min Land Productivity L. Ha-1 98959.13650 253 Kg. Beef Ha-1 11057.7430 11 Total revenue Ha-1 5541253.3213678 12438 Gross product Ha-1 3888755.9109889 9428 Net product Ha-1 3146762.796307 3853 Cow Productivity L. Milking cow-1 218628.23931 760 L. Total Cow-1 155631.03260 640 Kg. Beef Total cow-1 18242.0531 63 Total revenue Total cow-1 8829925.5156794 49851 Gross product Total cow-1 6127231.5139507 29310 Net product Total cow-1 4895441.8134552 14381 Labor Productivity L. E.H-1. 1962436.339083 7988 L. Milker-1. 4301256.0219000 13140 Kg. Beef P-Y E-1. 233848.66583 500 Total revenue P-Y E-1(Bs) 111976033.42238292 484167 Gross product P-Y E-1 (Bs) 78050838.61654599 287296 Net product P-Y E-1 (Bs) 62954348.11490486 144455 Capital Productivity Total revenue 1000 Bs. capital-1 15934.1357 35 Gross product 1000 Bs. capital-1 11139.2259 26 Net product 1000 Bs. capital-1 9050.7254 12 Net margin 1000 Bs. capital-1 5881.0222 -87.22 Source: UCPC survey, Zulia, Venezuela, 1994. Production Cost Costs were classified as inputs and services , depreciation, and salary (Table 3.15). Debts related to assets were not included because it was difficult to obtain accurate information from farmers. Cost structure and its relation to total revenue is also shown in

PAGE 60

45 table 3.15. Inputs represented 47% of total cost followed by salary with 33%. The main inputs were supplement feed and veterinary medicine (27% of total cost). These inputs were targeted to increase animal productivity; inputs to increase land productivity (seed, fertilizer, and pesticides and herbicides) were very low (no more than 6% of total cost). Total costs represented 57% of total revenue, a number that could decrease substantially if the opportunity cost of capital is included. Average farm cost per factors and per output are listed in table 3.16. Milk and beef cost were determined using the allocation cost methods where beef cost and milk cost are separated according to percentage of participation that each product has in total revenue. Therefore, if 70% of total revenue comes fr om milk, 70% of total cost corresponds to milk production. This method was used because of the lack of accounting records to allow separation of different costs according to outputs. Output cost is an index that shows how efficiently farmers are managing their farms. In this system average milk and beef costs are ~27 and 87 Bs with maximum and minimum values of 68 and 11 Bs for milk, and 226 and 34 Bs for beef respectively indicating great variability among farms.

PAGE 61

46 Table 3.15. Composition and cost structure per farm of DPCS Mean(Bolivares Bs.) C.V.Max. Bs. Min. Bs Total Cost % Total Revenue % Total Revenue 13515000139.41558440001230000 100.0 Cost Structure Input Cost 3601414135.238639700284746 46.626.7 Seed 86935232.914268450 1.10.6 Fertilizer 81654335.922630000 1.10.6 Pesticides and herbicides 237977192.725525500 3.11.8 Supplements feed 1323295208.9254835004800 17.19.8 Veterinary medicine 692876143.1891370644303 9.05.2 Fuel and lube 258915105.622295210 3.41.9 Machinery rent 29789386.010000000 0.40.2 Repairing and parts 296720169.435729990 3.92.2 Building maintenance 168544166.920000000 2.11.3 Taxes and insurance 29506317.66000000 0.30.2 Utilities 264739177.540800000 3.42.0 Technical services 45826188.34800000 0.60.3 Others 84639346.221514710 1.10.6Depreciation 1569533100.1905512539480 20.411.6 Buildings and installations 69825995.75235808.523160 9.05.2 Machinery and equipment 871274122.463449860 11.36.5Salary 254437638948600122720154 33.018.8Total Cost 7715323121.370613600790005 100.057.1 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.16. Relationship between cost, outputs, and factors per farm of DPCS Mean (Bs) C.V.Max Min Cost L-1 273568 11 Cost Kg-1 8736226 34 Cost Cattle area-1 3564773224131 9342 Cost Cow-1 5720940118330 22786 Cost AU-1. 322424576946 13661 Cost P-Y E-1. 701998401654908 265424 Cost Capital-1 0.1420.2 0.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 62

47Net Farm Income and Profitability Net farm income, the difference between total cost and total revenue, represents the return to the owner for personal labor, management, and equity capital used in the farm (Kay, 1981). Some indexes based on net farm income and some indexes of profitability such as net farm income-total capital and revenue-cost relations are shown in table 3.17. The average net farm income-total capital relation is 5.84%, indicating that farmers receive in return 5.84 Bs per each 100 Bs invested in the farm. The revenue-cost relationship indicated that farmers receive a gross income of 1.72 Bs per 1 Bs expended into the productive process, or farmers realize a net income of 0.72 Bs per bolivar spent. Table 3.17. Profitability and net margin per farm of DPCS Mean (Bs.)C.V.Max Min Net farm income farm-1 1027140017785230400 -3273610 Net farm income Cattle area-1 197659276648 -32736 Net farm income Cow-1 3109074112780 -42514 Net farm income P-Y E-1 417762781288770 -687734 Net farm income Total capital-1 5.848122.17 -872180.0 Total revenue Total cost-1 1.7363.6 0.6 Source: UCPC survey, Zulia, Venezuela, 1994. Financial Aspects In the 10 years prior to the survey 35% of farmers requested loans from private and public finance institutions, and 32.77% were bestowed (Table 3.18). Approximately 70% of the credit was granted with variable annual interest rates that ranged from 8 to 60%, while the fixed annual interest rate oscillated from 3 to 19% (Table 3.19). Loans were mostly invested in the farm and used mainly to increase acreage, to establish pasture, to expand buildings, and to purchase machinery and equipment.

PAGE 63

48 Table 3.18. Loan situation per farm surveyed NumberPercentage Requested 4235.3 Bestowed 3932.8 Not bestowed 32.5 Not requested 7764.7 Total farms 119100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.19. Annual interest rate for loans to surveyed farms Type NumberPercentage Variable (8% 60%) 3071.4 Fixed (3% 19%) 1228.6 No. of farm requesting credit 42100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Technical Aspect Forage management The number of pastures per farm generally varies with farm size, ranging from farms with only 4 pastures to a farm with 600 pastures, the average number being 39.2 pastures (Table 3.20). Generally, bigger farms tend to have pastures than smaller farms. The same tendency was observed for pasture size ranged from 1.6 to 43 ha, with an average of 8.3 ha. Table 3.20. Number and pasture size of DPCS. MeanCV.Max Min Number 39.2158.8600 4.0 Size (ha) 8.380.843 1.6 Source: UCPC survey, Zulia, Venezuela, 1994. The predominant grazing method used by DPCS farmers was rotational stocking (92%) (Table 3.21). The data did not allo w differentiation on the basis of rotational stocking method used, i.e. between systema tic rotational or non-systematic rotational stocking. Systematic rotational stocking, introduced in recent years, is a technique to improve the efficiency of grazing by increasing productivity per hectare.

PAGE 64

49 Table 3.21. Grazing systems of DPCS AnswerPercentage Permanent 32.5 Rotational 11092.4 Permanent and rotational 65.1 Total farms 119100.00 Source: UCPC survey, Zulia, Venezuela, 1994. Utilization and rest periods vary according to season and irrigation presence in pastures. Utilization periods of less than three days and rest periods of less than 21 days were the most common for the dry and rainy seasons, respectively (Table 3.22). Rest periods of 21 days or less could compromise pasture persistence depending on the particular case. Table 3.22. Rest and utilization periods (Days) 3>3 and 7>7 and 15 > 15Total Dry season utilization period (# of farms) 72209 1102 Percentage 70.619.68.8 1.0100.0 Rainy season utilization period (# of farms) 682610 2106 Percentage 64.124.59.4 2.0100.0 21>21 and 42>42 and 63 >63 Total Dry season rest period (# of farms) 463711 195Percentage 48.438.911.6 1.1100.0 Rainy season rest period (# of farms) 483714 2101 Percentage 47.536.613.9 2.0100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Farmers apply different cultural practices (Table 3.23) with weed control being the most common practice. A total of 98% of farmers controlled weeds using different methods; the most common practice is a mix of manual, mechanical and chemical control.

PAGE 65

50 Table 3.23. Cultural practices of DPCS Answer Percentage Weed Control None 2 1.7 Control 117 98.3 Manual 15 12.8 Mechanic 6 5.1 Manual and mechanic 26 22.2 Chemical 7 6.0 Chemical and manual 16 13.7 Chemical and mechanic 3 2.6 Chemical, manual, and mechanic 44 37.6Total farms 119 100.0Fertilization None 96 80.0 Use 24 20.0 Urea 12 50.0 Phosphorus 0 0.0 Urea + phosphorus 3 12.5 Complete formula 1 4.2 Urea + complete formula 3 12.5 Phosphorus + complete formula 1 4.2 Urea + phosphorus +cComplete formula 1 4.2 Manure 3 12.4Total farms 120 100.0Insect Control None 100 84.0 Control 19 16.0 Pesticide 4 21.1 Intensive grazing 13 68.4 Pesticide and intensive grazing 2 10.5Total farms 119 100.0

PAGE 66

51 Table 3.23. Continued. Answer Percentage Forages Conservation None 102 85.0 Conservation 18 15.0 Hay 6 33.3 Silage 6 33.3 Hay and silage 3 16.7 Cutting grass 1 5.6 Cutting grass and hay 1 5.6 Cutting grass and silage 1 5.6 Irrigation No Irrigation 90 75.0 Irrigation 30 25.0 Aspersion 9 30.0 Flooding 16 53.4 Aspersion and flooding 3 10.0 Furrow 1 3.3 Unknown 1 3.3 Total farms 120 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Fertilization is not a generalized practice. Only 20% of the farmers applied fertilizer; urea being the most common fertilizer (66.67%). When used urea was applied alone or combined with another type of fertilizer. Pest control is another cultural practice that is not widely used; only 15% of farmers controlled insect attacks to pasture using intensive grazing as the main practice. Forage deficits in the dry season are one of the main problems affecting production of the farms located in Zulia state except for those south of the state. Forage conservation and irrigation are practices used to compensate for this deficit. Only 50% of the farmers use conservation and 25% of them use irrigation. Use of hay and silage (83.33%), and flood irrigation (53.33%) are the most common practices.

PAGE 67

52Animal management Herd classification by sex, production, or age is a common practice among farmers (Table 3.24). Generally, producing cows and bulls, and calves graze in separate groups. While dry cows, heifers and bulls, steers, male and female yearlings are grouped depending on different criteria such as use of artificial insemination, castration, number of pastures, etc. Table 3.24. Number of farmers that classify the herd NumbersPercentage Classify 10992.4 No classify 97.6 Total farms 118100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Animal selection is made through different criterion, the most important being level of production (Table 3.25). Among the practices employed by farmers, the most common are deshorning, belly bottom cure, and classification and branding of the herd, which were indicated in 91.74%, 96.67%, 84.30%, and 93.39% of the cases respectively (Table 3.26). Some parameters related to animal performance were calculated like average birth weight and age, weaning weight and age, first calving weight and age, and sale age and weight of male (Table 3.27.). Calves were born with an average weight of 36 kg, while age and weight at weaning were about eight months and 163 kg re spectively. Age and weight at sale of males had the greatest coefficient of variation, possible due to differences in the production system (cow-calf, cow-yearling, a nd cow steer). The average age and weight at sale of males was 15 months and 258 kg respectively. Heifers were mated when average age or weight was 28 months or 330 kg, and calving was at about 38 months or

PAGE 68

53 395 kg on average respectively. In terms of breeding system, 54.55% of farmers used natural breeding, and only 26.44% of farmers used artificial insemination (Table 3.28). Table 3.25. Methods of animal selection of DPCS Answer Percentage None 1 0.8 Production 22 18.4 Physical condition 9 7.5 Production and physical condition 14 11.7 Phenotype 2 1.7 Phenotype and production 3 2.5 Phenotype and physical condition 6 5.0 Phenotype, production, and physical condition 4 3.3 Parent information 0 0.0 Parent information and production 4 3.3 Parent information and physical condition 1 0.8 Parent information, production, and physical condition 6 5.0 Parent information and phenotype 2 1.7 Parent information, phenotype and production 4 3.3 Parent information, phenotype and physical condition 1 0.8 Parent information, phenotype, production, and physical condition 19 15.9 Others 2 1.7 Others and production 12 10.0 Others and physical condition 2 1.7 Others, phenotype, and production 1 0.8 Others, parent information, production, and physical condition 1 0.8 Others, parent information, and phenotype 1 0.8 Others, parent information, phenotype, production, and physical condition 3 2.5 Total farms 120 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.26. Cultural practices of DPCS Answer Total farm Percentage Yes No Dehorning 111 10 121 91.7 Castrating 69 52 121 57.0 Milk weighing 58 63 121 47.9 Animal weighing 58 63 121 47.9 Belly bottom cure 116 4 120 96.7 Herd classification 102 19 121 84.3 Herd identification 113 8 121 93.4 Separation by sex 80 41 121 66.1 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 69

54 Table 3.27. Zootecnic parameters of DPCS MeanCV MaxMin Weight at birth (Kg) 3617.6 5025 Age at weaning (Months) 818.2 121.5 Weight at weaning (Kg) 16317.5 22560 Male sale age (Months) 1555.0 361.5 Male sale weight (Kg) 25842.6 48060 Age first service (Months) 2821.1 4816 Weight at first service (Kg) 3308.1 400230 Age at first calving (Months) 3815.4 5224 Weight at first calving (Kg) 3957.7 480300 Lactation period (Months) 814.3 126 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.28. Breeding System of DPCS AnswerPercentage Natural breeding 6654.5 Control Breeding 2117.3 Free and Control Breeding 21.7 Artificial Insemination 108.3 AI and free Breeding 119.1 AI and Control breeding 119.1 Total farms 121100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Milking is generally done manually (95%), with the calf present (96%), twice a day (98%) in a milking pen (96%) (Table 3.29). Calf presence is very important in this system because it stimulates milk production, and depending on degree of crossbreeding of animals, lactation can be suppressed when the calf is not present. Approximately 80% of calving takes place in special pasture that fa rmers prepare to assist the cows or heifers when needed. Herd feeding is based on grazing of improved pasture species, usually not fertilized. According to Stobbs 1971, tropical forages have a potential for production of approximately 8 or 9 kg cow-1day-1 when bred animals are used. These levels, however, can drop drastically during the dry season.

PAGE 70

55 Table 3.29. Milking characteristic of DPCS AnswerPercentage Milking type Manual 11294.9 Mechanical 21.7 Manual and mechanical 43.4Total farms 118100.0 Calf presence Present 11096.5 Not present 10.9 Mixed 32.6Total farms 114100.0 Milking frequency Once a day 21.8 Twice a day 11298.2Total farms 114100.0 Milking Place Milking pen 10796.4 Yard 43.6Total farms 111100.0 Calving place Yard 10.8 Special pasture 9781.5 Any place 2117.7Total farms 119100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Levels of production greater than those pr evious mentioned can be obtained when fertilization, irrigation, forage conservation, and animal supplementation are used (Gonzalez, 1992; Urdaneta et. al. 1992). More than 55% of farmers use a supplement feed such as concentrate feed, minerals, salt, molasses or other source (Table 3.30). Generally, concentrate feed is used to feed the production cows and the calves, while minerals and salt are given to the entire herd to compensate for mineral deficiencies. Molasses is also a common supplement given to the entire herd.

PAGE 71

56 Table 3.30. Supplement feed used by DPCS UseNot usedTotalPercentage used Concentrate Feed 665412055.0 Milk Substitute 41171213.3 Minerals 715012158.7 Salt 1061512187.6 Molasses 645712152.9 Others 863512171.1 Source: UCPC survey, Zulia, Venezuela, 1994. Characterization of the health program (Table 3.31), shows 95% of the farmers implement a health program, and about 88% of farmers enforce the complete regional vaccination plan suggested by the Ministry of Agriculture. Likewise, 95% of farmers control external parasites, and a 100% of the farmers control internal parasites. Frequency of parasite control is variable and around 80% of the farmers execute this practice at least every four months. In re lation to mortality, about 45% and 30% of the farmers reported none or little mortality in a dults and calves respectively (Table 3.32). The main causes of mortality in adults resulte d from septicemia and snake bites (19% and 13%, respectively). The main causes of cal f mortality were diarrhea, pneumonia, and septicemia all representing 50% of deaths.

PAGE 72

57 Table 3.31. Characteristic of the herd health plan of DPCS Health Plan AnswerPercentage Yes 11495.0 No 65.0Total farms 120100.0Vaccination Plan Complete 10688.3 Incomplete 1411.7Total farms 120100.0External parasite control No control 54.5 Control 10795.5Total farms 112100.0Internal parasite control No 00.0 Every to weeks 65.4 Every three weeks 65.4 Every Month 2623.2 Every two Months 1916.9 Every three Months 1412.5 Every four Months 1614.3 Every six Months 1614.3 Once a year 21.8 Yes 65.4 Eventually 10.8Total farms 112100.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 73

58 Table 3.32. Mortality and cause of mortality Adults% Calves % No mortality 32.5 0 0.0 Mortality 11597.5 116 100.0 Few mortality 5344.8 33 28.4 Diarrhea 00.0 26 22.4 Pneumonia 00.0 14 12.0 Pneumonia and diarrhea 00.0 8 6.9 Septicemia 2218.6 11 9.5 Anaplasmosis 43.3 2 1.7 Snake bite 1512.7 2 1.7 Piroplasma 10.9 0 0.0 Antrax(C sintomatico) 10.9 2 1.7 Difunsion Birth 00.0 3 2.6 Tripanosomiasis 75.9 1 0.9 Foot and Mouse disease (FMD) 10.9 0 0.0 Calving 21.6 4 3.5 Rabies 10.9 0 0.0 Leptospirosis 10.9 0 0.0 FMD and Rabies 10.9 0 0.0 Snake bite and rabies 10.9 0 0.0 Snake bite and tripanosomiasis 10.9 0 0.0 Septicemia and Neumoenteritis 00.0 3 2.6 Septicemia and diarrhea 00.0 1 0.9 Septicemia and anaplasmosis 00.0 1 0.9 Others 43.4 5 4.3Total farms 118100.0 116 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Technical assistance Providing technical support contributes to increased productivity and efficiency of the productive process because it decreases the learning cost of new technology, making transference and adoption of technology more efficient. A total of 87% of farmers indicated having received technical assistance, private technical service being the most demanded at 70% (Table 3.33). Farmers demanded veterinary services more than any other technical service, indicating that more attention is given to the animal than pasture production. Sixty percent of the farmers recei ved technical assistance at least twice per month.

PAGE 74

59 Table 3.33. Characteristics of technical assistance of DPCS Answer Percentage Technical assistance None 16 13.2 Public 4 3.3 Private 85 70.3 Yes 16 13.2Total farms 121 100.0 Type Veterinary 76 63.8 Agronomist 4 3.4 Agronomist and veterinary 6 5.0 Zootecnist 0.0 Zootecnist and veterinary 2 1.7 Junior college 3 2.5 Junior college and Veterinary. 7 5.9 Junior collage, Veterinary, and Agronomist 1 0.8 Other 4 3.4 None 16 13.5Total farms 119 100.0 Frequency Permanent 5 4.5 Four times per mo 20 17.5 Once per mo 25 21.9 Twice per mo 18 15.8 Every two mo 6 5.3 Every three mo 3 2.6 Every four mo 4 3.5 Once a Year 1 0.9 Occasionally 16 14.0 None 16 14.0Total farms 114 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Management Aspects Management is important to the productive process because the success of a business relies on how scarce resources are combined and used. Farmers with the same amount and quality of resources will probably achieve different results. Many different definitions of management have been given. However, many authors agree to define

PAGE 75

60 management according to which illustrates the broad scope of management and its complexity (Kay, 1981). According to McDermontt and Andrew 1999, management has been defined as assuming responsibility for setting objectives, planning, implementing, and evaluating activities for the success of an organization and its program in both the short and long run. To study management aspects, several questions were developed in order to obtain information about managerial functions such as organization, control, planning, and direction. The results of these questions are shown in the section below. Organization The organizational structure of management of DPCS farms varies according to size and the level of technology of the farm. There were farms with a simple structure where the owner is the principal figure and delegates most or all partial responsibilities to a general manager. The rest of the labor is not classified and does not play an important role into the decision process. A more structured organization was also present in some cases with board of directors and a manager. The board delegates responsibilities to the general manager or the chief, who transfers responsibilities to the others members in the farm organization (Table 3.34). In both cases technical support can be present. Control Dual-purpose cattle farmers tend to keep more technical records than accounting records (Table 3.35) possibly indicating that they are more interested in physical production than economic returns. Records were used by 40% of farmers (Table 3.36) only for technical decision vs. 15% and 34% who used records for either economic and economic plus technical purposes respectively. Similarly, only 12% of the farmers reported knowing the production cost for a liter of milk (Table 3.37). These results suggest that status to be recognized as a producer with the highest levels of production

PAGE 76

61 per cow may be the main objective or very important to these farmers. Only 54% of producers compared results across years, and about 60% indicated corrective actions to adjust any deviation from their goals (Tables 3.38 and 3.39). Table 3.34. Organization chart of DPCS Answer Percentage None 2 1.7 Organizational structure 117 98.3 Owner 3 2.6 Owner-workers 20 16.9 Owner-chief-workers 40 33.6 Owner-barn chief-workers 2 1.7 Owner-veterinary-workers 1 0.8 Owner-family labor 1 0.8 Ownerfamily labor-hire labor 1 0.8 Owner-chief-barn chief –workers 6 5.1 Owner-chief-inseminator-workers 1 0.8 Owner-chief-veterinary-agronomist-accountant-workers 1 0.8 Owner-manager-agronomist-workers 1 0.8 Owner-manager-workers 7 6.0 Owner-manager-chief-workers 16 13.5 Owner-manager-barn chief-workers 3 2.6 Board-manager-chief-workers 5 4.3 Owner-manager-machinery operatorbarn chief-workers 1 0.8 Owner-manager-chief-barn chief-inseminator-workers 1 0.8 Owner-manager-veterinary-chief-workers 1 0.8 Owner-manager-veterinary-inseminator-workers 1 0.8 Owner-manager-veterinary-Inseminator-mach. operator-milkersfield workers 1 0.8 Board-manager-veterinary-chief-barn chief-mach. operator-milkersfield workers 1 0.8 Board-manager-accounter-chief-chief barn-workers 1 0.8 Board-manager-veterinary-accountant-chief-mach. operator-workers 1 0.8 Board-manager-vet-agro-account-chief-Inseminator-mach. operatormilkers-field workers 1 0.8Total farms 119 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Supervision of farm operator activities is conducted by 44% of farmers, while about 24% do not control the operator activities because either: they do not control the activities per se, they do not have machinery operators; or because the operator is a

PAGE 77

62 family member. The rest of the farmers control the activities through chores or by productivity per hour or per hectare (Table 3.40). Table 3.35. Description of records kept by DPCS Answer Total farms YesNo Farmers keepin g records (Percentage) Accounting records Daily book 546612045.0 Mayor book 497011941.2 Small box 427711935.3 Income statement 516811942.9 General Balance 526711943.7 Inventory book 526812043.3 Technical records Cattle inventory 982212081.7 Labor record 912811976.5 Weight milk 625812051.7 Herd history 744411862.7 Inputs control 694811759.0 Vaccination control 117312097.5 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 78

63 Table 3.36. Use of the record by DPCS Answer Percentage None 11 9.7 Technical decisions 46 40.7 Vaccination 16 34.7 Herd information 17 37.0 Herd management 4 8.6 Cattle mobilization 1 2.2 To know production 1 2.2 Herd information and vaccination 5 10.9 Cattle mobilization and vaccination 2 4.4Economic decision 17 15.0 To solicit credit 1 5.9 To know income statement 16 94.1Technical and economic decisions 39 34.5 To control 17 43.6 To take decisions 15 38.4 To know income statement and vaccination 2 5.1 To know income statement and to control 2 5.1 To Know income statement and herd information 1 2.6 To know income statement and to take decision 1 2.6 To know income statement, herd information, and to control 1 2.6Total farms 113 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.37. Knowledge of milk production cost by farmers AnswerPercentage Do not know 10688.3 Know 1411.7 Total farmers 120100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.38. Do you compare results among years? AnswerPercentage No 5445.8 Yes 6454.2 Total farmers 118100.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 79

64 Table 3.39. Type of corrective used to achieve goals Answer Percentage None 43 40.6 To decrease cost 7 6.6 To increase production 11 10.4 To sell assets 2 1.9 To sell assets and decrease cost 2 1.9 To find the problem and fix it 20 18.9 To implement new alternatives 9 8.5 To implement new alternatives and decrease cost 1 0.9 To implement new alternatives, sale assets, and increase production 1 0.9 To solicit credit 0 0.0 To solicit credit and sell assets 2 1.9 To give adequate instructions 3 2.8 To work more 2 1.9 Others (according to the case, circumstantial, etc.) 3 2.8 Total farmers 106 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.40. Control the machinery operatorsÂ’ activities Answer Percentage No control 7 6.2 By chores 13 11.5 Do not have operator 9 8.0 Supervising 50 44.2 Supervising and by chores 3 2.7 Per hr 11 9.7 Per hr and chores 1 0.9 Per hr and supervising 2 1.8 Per hectare 4 3.5 Per ha and per hr 2 1.8 Family 11 9.7 Total farmers 113 100.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 80

65Planning Most farmer objectives were technical (78%), where 60% were related to the adoption of new technologies to improve physical productivity of the farms (Table 3.41). This data highlight the fact that only 4% of farmers indicated having economic goals which corroborate the results pointed out before. The main objective seems to be recognition inside the agricultural community as a successful producer based on physical production level per animal. Many of the farmers (66%) indicated pla nning as an approach to develop their farms, where the budget was the instrument most used to plan activities (Tables 3.42 and 3.43). Similarly, 93% of the farmers manife sted knowledge of how to increase the level of farm production. Farmers considered that the main sources to increase farm efficiency and productivity were improving pasture mana gement and animal management, and the horizontal growth of the production factors (Table 3.44 and 3.45). Direction Close to half of DPCS farmers surveyed (46%) solve labor problems themselves (Table 3.46). Generally, labor problems were solved using dialogue and the labor legal office. In general terms (92%) dual-purpose farmers are open-minded persons that accept suggestions from their personnel. Most farmers (75%) categorize the labor and consider the chief as the key person inside the farm.

PAGE 81

66Table 3.41. Farmers' objectives of DPCS AnswerPercentage None 1613.8 No changes 54.3 Technical objectives 9077.6 To increase production factors (cows, pasture, land,) 910.0 To diversify production (fatten, breeding center,) 77.8 To diversify production and to increase production factors 22.2 To fix or to build buildings, installation, machinery and implements (cowshed, tractor, well) 77.8 To fix or to build buildings, machinery and implements, and to increase production factors 33.3 To implement techniques (improve pasture, AI, irrigati on, animal selection, div pasture, decrease mortality) 1820.0 To implement techniques and increase production factors 1617.8 To implement techniques, to diversify production and to increase production factors 11.1 To implement techniques, to fix or to build buildings, machinery and implements 1617.8 To implement techniques, to fix or to build buildings, machinery and implements, and to increase production factors 22.2 To implement techniques, to fix or to build buildings, mach. and implements, and to diversify production 11.1 To increase or To improve production 88.9 Economic objectives 32.6 To decrease cost 00.0 To sell the farm 3 100.0 Economic and technical objectives 21.7 To decrease cost and to implement techniques 150.0 To consolidate the farm 150.0 Total farmers 116100.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 82

67 Table 3.42. Do you have plans? Plan existence AnswerPercentage No 4034.2 Yes 7765.8 Total farmers 117100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.43. Type of instruments used by farmers to plan the activities Answer Percentage None 45 38.7 Memorandum 5 4.3 Budget 37 31.8 Budget and memo 8 6.9 Project 3 2.6 Project and memo 1 0.9 Project and budget 6 5.2 Project, budget, and memo 4 3.5 Others 3 2.6 Others, and budget 2 1.7 Others, budget, and memo 1 0.9 Others, project, budget, and memo 1 0.9 Total farmers 116 100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Table 3.44. Knowledge by farmers of the possibilities to increase production levels AnswerPercentage No 86.7 Yes 11193.3 Total farmers 119100.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 83

68 Table 3.45. Ways to increase production of DPCS Categories Answer Percentage Do not know 1 0.9 Improving forage management (pasture division, irrigation) 15 14.1 Improving animal management (feeding, selection, AI.) 17 15.9 Improving animal and forage management 10 9.4 Improving buildings and machinery 1 0.9 Improving buildings and machinery and forage management 6 5.6 Improving buildings and machinery, animal and forage management 2 1.9 Increasing production factors (animals, land) 21 19.6 Increase production factors and improving forages management 11 10.3 Increase production factors and improving animal management 3 2.8 Increase production factors, and improving buildings and machinery 2 1.9 Decreasing cost 0 0.0 Decreasing cost and improving animal management 1 0.9 Decreasing cost, and improving animal and forage management 2 1.9 Diversifying production 2 1.9 Diversifying production and improving forage management 1 0.9 Diversifying production, and improving buildings and machinery 1 0.9 Diversifying production, and increasing production factors 1 0.9 Others 9 8.4 Others, animal and forage management 1 0.9 Total farms 107 100.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 84

69 Table 3.46. Management and problem solution of DPCS Answer Percentage Who solves the labor problems? Do not have problems 13 12.3 Owner 48 45.7 Manager 19 18.1 Owner and manager 1 1.0 Chief 11 10.5 Chief and owner 8 7.6 Chief and manager 5 4.8Total farmers 105 100.0 How do the farmers solve problems? Do not have problems 2 2.8 Dialogue 27 38.1 Firing 12 16.9 Firing and dialogue 2 2.8 Labor legal office 18 25.4 Labor legal office and dialogue 4 5.6 Labor legal office and firing 2 2.8 Workers union 3 4.2 Others 1 1.4Total farmers 71 100.0 Do you accept suggestions? No 9 7.8 Yes 107 92.2Total farmers 116 100.0 Do you categorize the labor No 19 25.00 Yes 57 75.00Total farmers 76 100.0

PAGE 85

70 Table 3.46. Continued. Answer Percentage Who is the key person on the farm for the owner? None 1 1.2 Owner 9 10.4 Manager 5 5.8 Owner and manager 1 1.2 Chief 39 45.2 Chief and owner 2 2.3 Chief and manager 2 2.3 Inseminator 0 0.0 Milkers 4 4.7 Milkers and chief 4 4.7 Accountant 0 0.0 Accountant and chief 1 1.2 Tractor operator 0 0.0 Tractor operator and chief 1 1.2 Tractor operator, milkers and chief 2 2.3 Technician 0 0.0 Technician and chief 1 1.2 All 4 4.7 Son 5 5.8 Workers 2 2.3 Wife 2 2.3 Chief and workers 1 1.2 Total farmers 86 100.0 Source: UCPC survey, Zulia, Venezuela, 1994.

PAGE 86

71 CHAPTER 4 ECONOMETRIC PROCEDURES This chapter is organized in two sections, one related to the determination of the production function frontier and the technical e fficiency values of farms, and the other refers to the determinants of technical e fficiency. Different methodologies have been proposed to estimate efficiency as is discussed in chapter 2. Each approach has advantages and disadvantages. The stochastic production frontier approach seems to overcome the limitations of the others to determine the technical efficiency of the DPCS. The deterministic approach is also used in order to compare the two approaches. Details of the estimation of the production frontier, such as the functional form and the distributions for the one-sided error, and technical efficiency values are set forth. The second section describes the model developed to explain variation in technical efficiency where socio-economic variables such as farmer and farm characteristics, and technological variables are used. Likewise a description of the simulation model is provided which is used for policy purposes or managerial decisions in chapter 6. Production Frontier Methodology A production frontier represents the maximum output that can be obtained from a set of inputs or the minimum input set required to produce various outputs given a technology (Kumbhakar and Lovell, 2000). Deterministic and stochastic production frontier approaches have been used to calculate the partial output elasticities and to define technical efficiency of the farms

PAGE 87

72 referred to previously as objectives 2 and 3 in chapter 1. The general models can be expressed as iu i ix f y exp ,, in the case of statistical frontier (4.1) ii ix f y exp ,, in the case of stochastic frontier (4.2) where iiivu; yi is output of the ith farm; xi is a vector of inputs; is the vector of parameters; ui represents one-sided error; and vi is the random error. The dependent variable which represent output ( yi) is measured using gross revenue in order to aggregate the different outputs of farms because the data does not permit the desegregation of inputs according to outputs. Using gross revenue in value terms also introduces the possibility of price differences across the farms, which can reflect economies of size in the sale of outputs. The explanatory variables (vector xi ) in these models follow the inputs present in the cost and capital structure of this system established by UCPC University of Zulia such as: i.e. labor (LE1), capital in pastures (LCP), capital in land (LCL), capital in buildings (LCE), capital in machinery (LCM), capital in cattle (LCC), machinery repairs and parts (LJ3), veterinary medicine (LI6), seeds (LI2), fertilizer (LI3), herbicide and pesticide (LI4), concentrate feed (LI5), ga s and lube (LI7), machinery rental (LJ2), building maintenance (LJ4), taxes and insurance (LJ5), utilities (LJ6), and miscellaneous (LJ7). All these variables, except for labor, are measured in monetary terms (Bolivares) to aggregate some inputs and to consider the quality of the inputs; labor is measured in person-year equivalents.

PAGE 88

73Functional Form A Box-Cox transformation was used in the deterministic model to test the functional form of the production frontier between the linear and double-log model by the following equation: i i iu x f y ) ; ( 1 . (4.3) This equation reduces to a linear model i i iu x f y ) ; ( 1 when = 1 and to a log linear model when = 0 after using a Taylor series expansion. The method of ordinary least squares was applied, and the results indicated that the double-log model was the appropriate specification due to lowest error su m of squares, highest R-square and log of the likelihood. The results are presented in the following chapter. A Cobb-Douglas functional form was select ed over the translog function to avoid the degree of freedom problem. The Cobb-Douglass frontier function after taking the natural logarithm of equations (4.1) and (4.2) are i i iu x f y ; ln ln (4.4) i i i iu v x f y ; ln ln (4.5) The statistical frontier function model (4.4) was calculated using corrected ordinary least square (COLS), while maximum likelihood (ML) was used for the stochastic frontier (4.5) following Aigner, Lovel, and Schmidt methodology. The coefficients for these models measure the elasticity of output with respect to each of the explanatory variables, i.e. the percentage of change in the dependent variable given a percentage of change in each explanatory variable (Gujarati, 1995). Once the production function

PAGE 89

74 frontier was specified and determined, technical efficiency values for the farm were calculated using the Jandrow technique (Jandrow et al., 1983). Deterministic Production Frontier The estimation of the deterministic frontier through the COLS consists in estimating the parameters for the production frontier (4.4) using ordinary least squares. However, the intercept from OLS is biased and must be corrected. The intercept 0 was corrected adding the maximum OLS residuals C 0 = 0 +max (ui) Likewise, the residuals were corrected in the opposite direction adding also the maximum residual estimated as Cui = ui+ max (ui) where: C = corrected 0 = intercept ui = residuals estimated max = maximum The technical efficiency values for all the farms were calculated using this equation as first set forth in eq. (2.3) iu iTE exp (4.6) Stochastic Production Frontier Function To calculate the production frontier function, different distributions for the error term were tested such as: normal-half normal, normal-exponential, normal-truncated, and normal-gamma. However, only the half normal and exponential could be estimated. The

PAGE 90

75 Truncated and Gamma distribution, which are mo re flexible distributions than the Half Normal and Exponential, could not be estimated because convergence of the log likelihood was not achieved. This problem has been attributed to insufficient sample size (Ritter and Simar, 1997). These authors point out that the parameters of these distributions are difficult to estimate for sample sizes less than 200 observations and the current sample is 127 observations. Normal-half normal distribution This distribution is one of the most widely used in empirical work and was proposed almost simultaneously by Aigner et al. (1977) and Battesse and Corra (1977). Three main assumptions were established to estimate this model: 2, 0 N ~v iiid v. 2, 0 N ~u iiid u, and ui 0. ui and vi are independently distributed and are not correlated with the exogenous variables. In order to obtain the likelihood function it is necessary to derive the marginal density function from [ f ()], where the joint density functions of u and v [ f ( u , v )], and the joint density function of u and [ f ( u , )] need to be defined. Following the notation of Kumbhakar and Lovell (2000), the density functions of u and v are 2 22 exp 2 1u i u iu u f (4.7) 2 22 exp 2 1v i v iv v f , (4.8) and the joint density function of u and v is

PAGE 91

76 2 2 2 22 2 exp 2 1 ,v i u i v u i iv u v u f (4.9) Given that iiivu, the joint density function of u and is 2 2 2 22 2 exp 2 1 ,v i i u i v u i iu u u f (4.10) By integrating u out of f ( u,) the marginal density function of is 0,i i iu f f (4.11) 2 22 exp 1 2 2 i i (4.12) i i2 (4.13) where . is the cumulative distribution function for a standard normal random variable, . represents the standard normal density function; ; ln lni i ix f y ; v u ; and 2 2 2v u . Once the density function of is defined, the log likelihood function for a sample of n observation is obtained by the equation ii i in const L2 22 1 ln ln ln (4.14) The coefficients of the production frontier were obtained by maximizing the likelihood function with respect to the parameters. Exponential distribution This distribution has been suggested by Aigner et al. (1977) and Meeusen and van den Broeck (1977). The assumptions for this model are

PAGE 92

77 2, 0 ~v iiid v. iid ui~exponential, ui and vi are independently distributed and are independent from the exogenous variables. To estimate the log likelihood function for this model the same steps of the half-normal model were followed. The marginal density function of was given by du u f fi , (4.15) 2 22 exp 1u v u i u v v i u given that the joint density function of ui and vi [ f(ui, vi) ], and ui and i, and [ f(ui, i) ] are 2 22 exp 2 1 ,v i u i v u i iv u v u f , (4.16) 2 22 exp 2 1 ,v i i u i v u i iu u u f , (4.17) where the density function of ui and vi are given by u i u iu u f exp 1 (4.18) 2 22 exp 2 1v i v iv v f (4.19) The log likelihood function was estimated using the following equation: ii u i u v v i u v un n const L ln 2 ln ln2 2 (4.20)

PAGE 93

78Standardized Coefficient Coefficients obtained from the regression models cannot be directly compared because the explanatory variables have different scales and different variances. In order to compare and rank the relative importance of the exogenous variables, the coefficients were normalized using the ratio of the standard deviation of the explanatory variables to standard deviati on of the dependent variable (Pindyck and Rubinfeld, 1998) y x j jj * (4.21) where: * j= standardized coefficient j = coefficient of the exogenous variable jx = standard deviation of the exogenous variable y = standard deviation of the dependent variable Multicollinearity All the assumptions set for the linear models were satisfied except for the presence of multicollinearity among the independent variables. This correlation was expected because of the many exogenous variables used in the frontier model. The variables showing a high degree of correlation were: la bor, different classifications of capital (pasture, land, machinery, buildings, and cattle), machinery repairs and parts, and veterinary medicine. Principal component techniques were used to deal with multicollinearity problems. This technique uses a set of new variables obtained from linear combinations of the

PAGE 94

79 exogenous variables. These new variables are called principal components and contain information from the exogenous variables that are not correlated. In this research, the first four principal components were used in the regression model (production frontier) because they explained nearly 95% of the variation identified by the original variables. When the coefficients were obtained for the different variables and four principal components in the production frontier model, coefficients for the exogenous correlated variables were rebuilt and the appropriate tests were performed (see program, appendix B for details). The econometric software package Time Series Processor (TSP) version 4.4 (TSP International, 1998) was used to perform the statistical calculations. For comparison purposes, estimation of coefficients for the different stochastic production frontiers was also performed using the econometric software Limdep version 6 (Greene 1991). Technical Efficiency Estimates Technical efficiency values were estimated using the Jondrow et al. (1982) technique. The approach uses the mode or the mean of the distribution as a point estimator of ui. The research uses the mean as a point estimator. Given the conditional density function of ui given i: i i i i if u f u f , | (4.22) the means for the half-normal and exponential distributions were calculated through the following equations respectively: i i i i iu 1 1 |2 (4.23)

PAGE 95

80 u v v i u v v i u v v i v i iu | (4.24) Then, technical efficiency values for both models were calculated with equation 4.6 iu iTE exp Determinants of Technical Efficiency Two approaches were initially considered to explain the variation in technical efficiency. One approach includes a two step procedure with the indices of technical efficiency determined in the first stage using the methods outlined above where production is regressed on the identified inputs ( xi). The second step regressed the technical efficiency indices against a set of socio-economic explanatory variables ( zi) that are expected to influence TE. Here, the method of ordinary least squares (OLS) is commonly applied. The following equations illustrate the first approach i i i iu v x f y , ln ln (4.25) i i iz TE (4.26) where all the variables are as previously defined except for zi representing the vector of socio-economic variables and the coefficients of socio-economic variables. The second approach, as proposed by Kumbhakar et al. (1991), estimates the efficiencies and the determinants of TE in one step. In this case, the determinants of efficiency are incorporated into the one sided error. i i i iu v x f y , ln ln, (4.27) i i ie z u (4.28) where ei is the random error.

PAGE 96

81 This method could not be utilized because it requires a functional distribution of the error term, which could not be tested because of the sample size. The OLS technique was used to run the regression in the first approach, but because the dependent variable (TE) is bounded between zero and one, it was transformed using the logistic distribution as i iz iTE exp 1 1 (4.29) i i iz TE 1 1 log (4.30) where all terms have been described previously. Defining the Efficiency Variables ( z ) The socio-economic variables tested and considered to explain the variation in efficiency are described in this section and a summary of the 13 variables that encompass the z factors with their respective characteristic and expected behaviors is listed in Table 4.1 These variables were classified in three categories: a) farmer characteristics, which include education, experience, and presence on the farm; b) farm characteristics, like tenure, credit, farm size, location, and production system; and c) technological variables such as cow productivity, breeding system, fr equency of technical assistance, labor productivity, and stocking rate. Most of these variables are categorical except for labor productivity and stocking rate, which are included in the models as continuous variables. Farmer characteristics Education (EDU). Education is used as a proxy variable to measure the managerial skill of managers. This variable was introduced as a dummy variable in the model with two levels EDU2 if producers had an elementary education or higher, and EDU1otherwise. Different levels of educati on were also considered but they were not

PAGE 97

82 significant. Each one of these categories takes the value of 1 if the statement is true or zero otherwise. Experience (EXP). Experience is another variable used to measure the managerial ability of the farmers. Like education this is a binary variable, in which EXP2 represents producers with more than 5 years of experien ce, and EXP1 otherwise. Each variable takes the value of 1 if the statement is true otherwise it is zero. Different levels of experience were tested but after 5 years no differences in managerial ability were found. Presence (PER). Presence of the producers on the farms was another variable considered to explain the variation in efficiency. Farmers that spent more time on the farm may closely supervise the different activities conducted. Recently producersÂ’ longevity on the farms is compromised due to insecurity in the countryside. This binary variable has to levels, PER2 for farmers who stay in the farm at least twice a week, and PER1 otherwise. Each level takes the value of 1 is the statement is true, or zero otherwise. This variable is expected to have a positive effect on efficiency.

PAGE 98

83 Table 4.1. Socio-economic variables Variable Description Value (1 = yes; and 0 =otherwise) Hypothesis EDU1 Producers that are illiterate or only read and write 0, 1 a1 > 0 EDU2 Producer with elementary school or higher education 0, 1 a2>a1 EXP1 Farmers with 5 years of exp or less 0, 1 a1 > 0 EXP2 Farmers with more than 5 years of exp 0, 1 a2>a1 PER1 Farmers with presence on the farm less than twice a week 0, 1 a1 > 0 PER2 Farmers with presence on the farm twice a week or more 0, 1 a2>a1 TEN1 County and government land 0, 1 a1 > 0 TEN2 Private land 0, 1 a2>a1 CRED1 Farmer that do not use credit 0, 1 a1 > 0 CRED2 Farmer that use credit in the last 10 years 0, 1 a2 0 SUG2 Farm size from 300 to 400 ha 0, 1 a2 > a1 SUG3 Farm size greater than 400 to575 ha 0, 1 a3 > a2 SUG4 Farm size greater than 575 ha 0, 1 a 4< a3 Z1 Producers located in the south part of Zulia State 0, 1 a1 > 0 Z2 Producers located in th e eastern part of Zulia State 0, 1 a2 0 PSYST2 Cow-yearling 0, 1 a2 > a1 PSYST3 Cow-steer 0, 1 a3 > a2 PROD1 Milk production per cow equal or less than 1000 liters. 0, 1 a1 > 0 PROD2 Milk production per cow higher than 1000 to 1500 liters. 0, 1 a2 > a1 PROD3 Milk production per cow higher than 1500 to 2000 liters. 0, 1 a3 > a2 PROD4 Milk production per cow higher than 2000 to 2500 liters. 0, 1 a4 > a3 PROD5 Milk production per cow higher than 2500 liters 0, 1 a5 < a4 BRED1 Producers natural breeding 0, 1 a1 > 0 BRED2 Producers using artificial insemination 0, 1 a2>a1 TECHN Less than once a month 0, 1 a1 > 0 TECHN Once a month or higher 0, 1 a2>a1 LTMILKER Labor productivity > 0 a>0; a2<0 CARGANEF Stocking rate > 0 a>0; a2<0

PAGE 99

84Farm characteristics Land tenure (TEN). The relationship between efficiency and type of tenure is not clear in empirical work. An explanation is that tenants specialize in technologies that require less fixed capital, while owners use technologies that require more fixed capital. A dummy variable with two categories is introduced to measure the impact of this variable on efficiency, TEN2 representing the farmers owning the land and TEN1 otherwise. Each one of these categories takes the value of 1 if the statement is true or zero otherwise. Credit (CRED). Previous studies indicate no clear relationship between use of credit and efficiency. In this research a negative relationship between efficiency and credit is expected because interest rates in Venezuela had been high and volatile during the 10 years prior to the survey. A dummy variable with two levels was used in the model, CRED2 denotes producers who obtained credit in the last 10 years, and CRED1 otherwise. Each level takes the value of 1 if the statement is true or zero otherwise. Farm size (SUG). Farm size is another factor that has been associated with the farm efficiency. In this research, the relationship between efficiency and farm size is expected to increase up to a certain level, and then decline. This assumption is based in empirical evidence showing medium farmers tending to have a greater level of technology adoption than small and large farmers. Small farmers display more constraints (capital, tenure, etc.) that limit the adoption of new technology. Large farmers seem to be satisfied with their current level of profit, and there is not an incentive to adopt new technology maybe because the cost of adoption is higher than the expected utility. A binary variable with four levels was used to avoid multicollinearity problems and because sometime continuous variables tend to form clusters. The ranges: SUG1 for

PAGE 100

85 farm size less than 300 ha, SUG2 for farm size from 300 to 400 ha, SUG3 for size greater than 400 to 575 ha, and SUG4 for farm size greater than 575 ha were selected to maximize the likelihood function. Each one of these categories takes the value of 1 if the statement is true or zero otherwise. Both farm size (SUG) and cow productivity (PROD) were included in the TE models where initially both were measured as continuous variables. Clearly, one could have included both variables using the continuous distribution. However, a priori there was some expectation that the effects of either or both variables could be non-linear with the possibility that the sign of the slopes could even change over the range of values. To facilitate the potential non-linearity using the origin data some form of polynomial in the variables would be required. That, in turn, can create potential multicollinearity problems, depending on the distribution properties of each variable. Furthermore, if the distribution of values within each variable tend to be clustered around certain size categories, then using the variables as continuous is not as useful. An appropriate alternative would be to take each variable and explore the nature of the distributions and then express the variable in terms of a few meaningful categories that reasonably reflect the distribution. For example, size was readily grouped into four sizes (i.e., less than 300 hectare; 300 to 400 hectare; etc.). Then, these four sizes can be expressed using dummies to reflect each size. With this approach any non-linearity in the response would be detected with the different coefficients associated each binary size (or productivity) variable. Problems associated with the modeling of the original variable(s) for nonlinearity are completely removed when this approach is used. Beyond the statistical considerations, one probably learns more with the binary variables since public policies

PAGE 101

86 are generally directed to groups (e.g. small farms) and not each farm size or production unit. If the original response were highly non-linear then a higher order polynomial would have been required with at least two parameters and possibly more needed to capture the response. With four size categories, only three parameters are required. Hence, there was little or no loss in degrees of freedom using the binary approach. The critical decision is to correctly identify th e groups. Maximum likelihood procedures were used in defining the appropriate groups. See Tables 4.2 and 4.3 for insight into the groups and clusters that existed for farm size and cow productivity. Table 4.2. Farm size Size (ha) No. of farmsPercentage <300 9171.7 >=300 and <=400 118.7 >400 and <=575 86.3 >575 1713.4 Total 127100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Location (Z). To characterize DPCS, Zulia State was divided into four geopolitical zones. Farmers from Zone 1 and Zone 3 are expected to be more efficient than the farmers from other locations because Zone1 presents the best agro-ecological conditions, (best quality of soils and good rainfall), and farmers from Zone 3 are characterized as innovative and progressive. Four classes were established to represent the four different geo-politic areas: Z1 for the south area, Z2 for the eastern coast, Z3 for Rosario and Machiques de Perija counties, and Z4 for counties located in the north-west. Each one of these categories takes the value of 1 if the statement is true or zero otherwise. Production system (PSYST). The dual-purpose cattle system with a dual objective, to produce milk and beef, has been allowed to adapt to market instability and unstable policies. However, three major production sub-systems may be found

PAGE 102

87 depending of the percentage of total revenue coming from milk and beef. These subsystems are: cow-calf, cow-yearling, and cow-steer . It is expected that farmers oriented to beef production (cow-steer) will be more efficient because this sub-system is tailored to the definition of the dual-purpose system. A categorical variable with three levels was adopted: PSYST1 for cow-calf, PSYST2 for cow-yearling, and PSYST3 for cow-steer, taken the value of 1 if the statement is true or zero otherwise. Technological variables Cow productivity (DPROD). This index measured the liters of milk per cow per year as a function of many variables, some intrinsic to the animal and others to herd management by the farmers. Animal variables like animal breeding, and management variables like sanitary plan and feeding will affect the liters per lactation produced, length of lactation, and calving interval, which will influence this index. For the same reasons as the farm size variable, a binary variable with 5 categories was created to measure the effect of production per cow on technical effi ciency (Table 4.3). PROD1 for cow equal or less than 1000 l. of milking; PROD2 for milk production higher than 1000 to 1500 l.; PROD3 for levels higher than 1500 to 2000 l; PROD4 for levels higher 2000 to 2500 l.; and PROD5 for levels higher than 2500 l. A quadratic effect is expected because as the productivity per cow increases, the efficiency of the farm increases until a certain level. Then it starts to decrease because of high production costs due to the high degree of breeding required for this level of production which is not adapted to the tropical conditions. Like others categorical variables these variables take the value of 1 if the statement is true or zero otherwise.

PAGE 103

88 Table 4.3. Production per cow Level ( L ) No. of farms Percenta g e (%) 1000 118.7 > 1000 1500 5240.9 > 1501 2000 4132.3 > 2001 2500 1915.0 > 2500 43.1 Total 127100.0 Source: UCPC survey, Zulia, Venezuela, 1994. Breeding system (BRED). The breeding system commonly used in these farms is natural mating. In recent years, however, several breeding companies have been established in the area, promoting artificial insemination (AI), which has become a popular practice among farmers. The use of AI implies better reproductive control and improvement in animal genetics oriented to increasing the farm efficiency. Nevertheless, if this practice is not well addressed a negative result can be obtained. A dummy variable with two levels (BRED2 for AI and BRED1 otherwise) is used to measure the impact on efficiency. Each one of these categories takes the value of 1 if the statement is true or zero otherwise. Frequency of technical assistance (TECHN). The research centers located in Zulia state have generated new technologies in order to improve the efficiency of the farms. These technologies have increased animal and land productivity through the implementation of new practices such as: artificial insemination, embryo transplant, rotational grazing, fertilization, and strategy supplementation (Hay, silage) among others. The use of these technologies requires a higher level of knowledge, supervision and control that is usually out of reach of the farmers, but can be compensated using technical assistance. Likewise, farmers in contact with extension agents (veterinarians, agronomists, animal scientists, etc.) have more access to market and technical

PAGE 104

89 information, an aspect that very likely will improve the efficiency of the farms. A categorical variable with two categories wa s tested: TECHN2 for frequency of technical assistance equal or greater than once a month, and TECHN1 for less than once a month. Each level takes the value of 1 if the statement is true or zero otherwise. Labor productivity (LTMILKER). As discussed in chapter 3, labor is the main factor in the cost structure for this system, representing 33% of the total cost and approximately 20% of total revenue. It indicates that producers should use this resource as efficiently as possible in order to achieve the greatest impact on efficiency. This variable is expected to show a quadratic effect similar to that of cow productivity. To measure this effect a continuous variable was used in its linear (LTMILKER) and quadratic form (LTMILKSQ). Stocking rate (CARGANEF). This variable shows how the land resource has been managed in relation to the animals because herd nutrition is based on grazing of natural and improved pastures. In the grazing system farm production is given by the interaction of production per cow and stocking rate. There is a negative relationship between these two variables, as the stocking rate increases the productivity per cow decreases. Therefore it is important to determine if this variable has a significant impact on efficiency and what is the optimum stocking rate. A continuous variable in linear and quadratic form was adopted, but only the linear form was included in the model since this variable was not significant, likewise the interaction between cow productivity and stocking rate. Modeling the Distribution of Efficiency Indices Initially, the following model (Equation. 4.31) was proposed to explain technical efficiency

PAGE 105

90i iCARGANEF LTMILKSQ LTMILKER TECHN TECHN TEN TEN BRED BRED PROD PROD PROD PROD PROD PSYST PSYST PSYST Z Z Z Z SUG SUG SUG SUG CRED CRED PER PER EXP EXP EDU EDU y 33 32 31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1 02 1 2 1 2 1 5 4 3 2 1 3 2 1 4 3 2 1 4 3 2 1 2 1 2 1 2 1 2 1 where 1 1 logi iTE y. Since this model includes classifications that are exhaustive, by definition the sum of the variables within a group sum to one and, hence, perfect multicollinearity problems among the categorical variables. To avoid these problems, the following restriction was imposed: k j j i j0. It means that the sum of the coefficient for the different levels of each categorical variable should be zero. Implicitly, then one coefficient can be expressed in term of the other coefficient. This condition generates the following model (Equation. 4.32) i iCARGANEF LTMILKSQ LTMILKER DTECHN DTEN DBRED PROD PROD PROD PROD PSYST PSYST Z Z Z DSUG DSUG DSUG CRED DPPER DPEXP DPEDU y 33 32 31 30 28 26 24 23 22 21 19 18 16 15 14 12 11 10 8 6 4 2 051 41 31 21 31 2 41 31 21 41 31 21 where 1 1 logi iTE y, 0 represent the average technical efficiency , DPEDU = EDU2– EDU1; DPEXP = EXP2–EXP1; DPPER = PER2–PER1; CRED = CRED2–CRED1; DSUG21 = SUG2–SUG1; DSUG31 = SUG3–SUG1; DSUG41 = SUG4–SUG1; Z21 = Z2–Z1; Z31 = Z3–Z1; Z41 = Z4–Z1; PSYST21 = PSYST2–PSYST1; PSYST31 = PSYST3–PSYST1; PROD21 = PROD2–PROD1; PROD31 = PROD3–PROD1; PROD41 = PROD4–PROD1, PROD51 = PROD5–PROD1; DBRED = BRED2-BRED1; and

PAGE 106

91 DTEN = TEN2-TEN1. The coefficients for the different categorical variables represent the deviation of each variable based on the average farm (0), averaged over all of the characteristics. The coefficient for each categorical variable with to levels was calculated as follows: for example for the effect of education (EDU1) is 0-2 and EDU2, 0+2. The impact of variables with more than two levels was calculated as follows: for the case of farm size depicted by three categories, the impact of SUG1 is 0-10-11-12; SUG2, 0+10; SUG3, 0+11; and SUG4 impact is 0+12. A similar procedure was used for the others variables. Heterocedasticity In this model all the assumptions for the linear model were satisfied except for the variance of the error terms i , which were not homoscedastic. To correct this problem the WhiteÂ’s heterocedasticity-consistent variances and standard error approach was applied using the Robustse command in TSP econometric software. This procedure generates standard errors of OLS estimators that are consistent even in the presence of unknown heterocedasticity based on the principle of maximum-likelihood (Gujarati, 1995; Hall and Cummins, 1998; Pindyck and Rubinfeld, 1998). Standardized Coefficients Like the coefficients of the production frontier, the coefficients of socio-economic variables explaining the variation in efficiency were standardized to rank the variables according to the magnitude of their effect using the equation: I IY X J J *, where *J represents the standardized coefficients; j the parameters without standardization, and IX and IY index the standard deviation of the independent and dependent variables.

PAGE 107

92 CHAPTER 5 EFFICIENCY ANALYSIS Measuring efficiency and productivity, and knowing the determinants of these indexes has always been of interest for farmers and government. They are used as indicators to measure and evaluate farmer performance, and to develop and evaluate policy. This chapter presents the: i) production frontier models results using the deterministic and the two stochastic models (half-normal and exponential model); ii) technical efficiency values, contrasting effi ciency values for different frontier models; and iii) determinants of the technical efficiency model, and a discussion of the effect of the different socio-economic variables on the efficiency of the farms. Production Frontier Models Deterministic and stochastic production functions were estimated to determine technical efficiency. Coefficients for the deterministic production frontier were calculated using corrected ordinary least squares, while coefficients for the stochastic production functions were calculated assuming two different distributions for the onesided error term “u,” the half–normal and exponential distribution. The Box–Cox technique was utilized to test the functional form on the deterministic production frontiers between a linear model ( =1) and a double-log model ( =0). Values of lambda and their respective error sum of squares, log of the likelihood function, and R-square from the iterative process are listed in table 5.1.

PAGE 108

93 Table 5.1. Box-Cox model for testing nonlinearity Iteration Lambda Sum of squared residualsLog of Likelihood function R-squared 1 0.0100 7.7771-4.7277 0.9497 2 0.0600 38.3737-102.8936 0.9505 3 0.1100 197.3634-203.6105 0.9495 4 0.1600 1057.6607-306.8547 0.9467 5 0.2100 5882.4742-412.3837 0.9421 6 0.2600 33742.3960-519.8105 0.9355 7 0.3100 198214.4833-628.7021 0.9270 8 0.3600 1184684.5762-738.6568 0.9167 9 0.4100 7165593.6012-849.3454 0.9044 10 0.4600 43684894.7342-960.5196 0.8903 11 0.5100 267670608.7271-1072.0041 0.8745 12 0.5600 1645226674.9714-1183.6805 0.8571 13 0.6100 10131395810.0042-1295.4728 0.8383 14 0.6600 62460005052.2159-1407.3343 0.8182 15 0.7100 385331987793.9280-1519.2384 0.7970 16 0.7600 2378323412050.0000-1631.1712 0.7749 17 0.8100 14684874348518.8000-1743.1272 0.7521 18 0.8600 90704542699658.5000-1855.1058 0.7289 19 0.9100 560484327790145.0000-1967.1092 0.7055 LAMB=0.01 TO .95 BY 0.05; ZY=((Y**LAMB)-1)/LAMB; OLSQ (ROBUSTSE) ZY C PZ1 PZ2 PZ3 PZ4 LI2 LI3 LI4 LI5 LI7 LJ2 LJ4 LJ5 LJ6 LJ7; The Cobb – Douglas functional form (double log model, =0) presented the lowest error sum of squares (7.77) and the highest R – square (0.950). Since the model explains 95% of the total variation in total revenue, use of double-log model is adequate. Findings for the different production frontiers, their respective coefficients, standard errors, and t-statistics are shown in table 5.2. Coefficients represent the output elasticity with respect to independent variables. They measure the percentage change in output related to percentage change of the independent variables. Variables such as labor (LE1), capital in land (LCL), capital in machinery (LCM), capital in cattle (LCC), veterinary medicine (LI6), supplement feed (LI5), building maintenance (LJ4), and taxes and insurances (LJ5) were statistically different from zero at the 95% or higher level for all three production frontiers. All these coeffi cients were positive. Among the inputs that did not have a significant effect on production stands fertilizer (LI3), which has been

PAGE 109

94 promoted by the extension service in order to improve the production of the dual-purpose cattle system. Variance parameter estimates, u and v, for the half-normal and exponential models, showed opposite results and different values (Table 5.2). For the half-normal distribution, the one sided error ( ui) dominates the symmetric error ( vi) suggesting that deviation from the frontier is mainly due to some factors that farmers can control. However, the exponential distribution indicates that the gap between the observed output and maximum output is mainly due to random effects. Standardized coefficients (listed in Table 5.3) allow comparison of the coefficients numerically to each other because the scale effect is eliminated. Ranking of the coefficients did not change substantially in all three models; however, small variations were detected mainly between the half-normal and the other two models (Figures 5.1, 5.2, and 5.3). Supplement feed, capital invested in cattle and labor were inputs with the more significant effect on the dependent variable in all three models and the hierarchy of the effects was in the same order.

PAGE 110

95Table 5.2. Production frontier estimates Corrected ordinary least squaresHalf normal dist ribution (Eq. 4.14)Exponentia l distribution (Eq. 4.20) Variable Estimate Std errort-statisticEstimate Std error t-statisticEstimateStd errort-statistic C 3.4046 3.6533 3.4828 LE1 0.2302a 0.013517.05120.2306a0.0119 19.39620.2303a0.012218.8653 LCP -0.0219 0.0306-0.7145-0.01840.0258 -0.7153-0.0213 0.0268-0.7940 LCL 0.1174a 0.01229.60290.1198a0.0112 10.65000.1178a0.011310.3820 LCE 0.0253 0.03200.79160.01630.0326 0.50020.02390.03160.7590 LCM 0.0543a 0.01703.18770.0495a0.0173 2.86380.0536a0.01683.1903 LCC 0.2598a 0.015317.02160.2612a0.0140 18.66790.2601a0.014218.3634 LJ3 -0.0040 0.0040-1.0181-0.00380.0040 -0.9391-0.00400.0040-0.9963 LI6 0.2995a 0.022613.24090.3016a0.0204 14.80060.2998a0.020714.4850 LI2 0.0035 0.00301.16350.00310.0028 1.11730.00340.00291.1913 LI3 -0.0037 0.0038-0.9743-0.00360.0031 -1.1479-0.00360.0033-1.1101 LI4 0.0018 0.00290.61450.00190.0025 0.75410.00190.00250.7379 LI5 0.0709a 0.01644.32540.0715a0.0158 4.51550.0712a0.01644.3402 LI7 0.0045 0.00540.83250.00570.0083 0.67840.00470.00850.5536 LJ2 0.0049 0.00351.39320.00360.0037 0.95760.00460.00371.2450 LJ4 0.0072b 0.00322.24310.0080b0.0033 2.40360.0072b0.00322.2900 LJ5 0.0096b 0.00481.98990.0106b0.0043 2.46250.0099b0.00452.2117 LJ6 0.0061 0.01850.33080.00590.0153 0.38350.00650.01380.4682 LJ7 -0.0044 0.0031-1.4371-0.0055c0.0033 -1.6813-0.00470.0032-1.4702 Lambda 1.69261.1397 1.4851 Sigma 0.2975a0.0566 5.2591 Theta 12.227811.86501.0306 Sigma V 0.1989a0.03266.0949 Sigma U R-square 0.9494 Adj. R-square 0.9428 Log likelihood 14.6138 15.1381 14.7144 a= P 0.01; b= 0.01 < P 0.05; c= 0.05 < P 0.10

PAGE 111

96 Table 5.3. Standardized coefficients of production frontier estimates Variables VariablesÂ’name DeterministicHalf-normal Exponential LE1 Labor 0.21650.2168 0.2166 LCP Capital in pasture -0.0288-0.0243 -0.0281 LCL Capital in land 0.12580.1284 0.1263 LCE Capital in building 0.02050.0132 0.0194 LCM Capital in machinery 0.05740.0524 0.0567 LCC Capital in cattle 0.25620.2576 0.2565 LJ3 Machinery repairs and parts -0.0260-0.0243 -0.0257 LI6 Veterinary medicine 0.32110.3234 0.3214 LI2 Seeds 0.02940.0264 0.0287 LI3 Fertilizer -0.0259-0.0251 -0.0255 LI4 Herbicide 0.01570.0165 0.0166 LI5 Concentrate feed 0.11630.1172 0.1167 LI7 Gas and lube 0.01280.0160 0.0134 LJ2 Machinery rental 0.03050.0223 0.0286 LJ4 Building maintenance 0.05460.0607 0.0550 LJ5 Taxes and insurance 0.05450.0599 0.0557 LJ6 Utilities 0.01160.0111 0.0122 LJ7 Miscellaneous -0.0343-0.0432 -0.0363 -0.10-0.050.000.050.100.150.200.250.300.35 LJ7 LCP LJ3 LI3 LJ6 LI7 LI4 LCE LI2 LJ2 LJ5 LJ4 LCM LI5 LCL LE1 LCC LI6VariablesEstimates Figure 5.1. Ranking of standardized coe fficient of deterministic production frontier

PAGE 112

97 -0.10-0.050.000.050.100.150.200.250.300.35 LJ7 LI3 LJ3 LCP LJ6 LCE LI7 LI4 LJ2 LI2 LCM LJ5 LJ4 LI5 LCL LE1 LCC LI6VariablesEstimates Figure 5.2. Ranking of standardized coe fficient of half-normal production frontier -0.10-0.050.000.050.100.150.200.250.300.35 LJ7 LCP LJ3 LI3 LJ6 LI7 LI4 LCE LJ2 LI2 LJ4 LJ5 LCM LI5 LCL LE1 LCC LI6VariablesEstimates Figure 5.3. Ranking of standardized co efficient of exponential production frontier

PAGE 113

98 Technical Efficiency Values Values of the “u” term were estimated using the Jondrow, et. al. (1982) methodology for the half-normal and exponential models. Estimates of the technical efficiency for all farmers were calculated using the equation i iu TE exp and each of the three production frontiers. These estimates varied notably across models (Table 5.4). The average estimates of technical efficiency were 0.630, 0.819 and 0.922 for the deterministic, half-normal, and exponential models respectively. As might be expected, the highest technical efficiency values we re obtained from the exponential distribution because the gap between observed and maximum output was mainly due to random effects. The lowest estimated values came from the deterministic estimates. According to this model, deviations (random or not) from the frontier are due to inefficiency. Zellner and Revankar (1970) cited by Greene (1997) reported similar results using the half-normal and the exponential distribution to measure efficiency in the transportation equipment manufacturing industries. The authors concluded that even though they seemed to be similar the distribution of inefficiency values from the two models was not the same. Information is lacking about the efficiency of dual-purpose cattle systems in tropical regions so comparison of results cannot be made. The available information relates to dairy intensive systems in developed countries, where average technical efficiency values rank from 0.45 to 0.85 depending on the method used and distribution assumed for the one-sided error “u” (Bravo-Ureta, 1983; Bravo-Ureta and Rieger, 1990; Bravo-Ureta and Rieger, 1991; Kumbhakar et . al., 1991 Tauer and Belbase, 1987;). The average technical efficiency values obtained in this research from the half-normal and

PAGE 114

99 exponential distribution could be considered high, discarding statements that the dualpurpose cattle system is inefficient as has been indicated by some policy makers and researchers. The ordinal ranking of the farms according to the technical efficiency estimates for the three models is similar (Table 5.5). The deterministic and the exponential models presented the closest ranking but also had the closest ranking for standardized coefficients. This result is similar to that observed by Bravo-Ureta and Rieger (1990) where the ordinal ranking was independent of the method used. (Deterministic or stochastic).

PAGE 115

100Table 5.4. Technical efficiency estimates per farm Farm id Deterministic Half-normal Exponent ialFarm idDeterministicHalf-normalExponenti al Farm idDeterministicHalf-normalExponent ial 0 0.4702 0.7079 0.8877410.70570.88850.9413 870.69550.87900.9399 1 0.7002 0.8823 0.9406420.61720.83880.9293 881.00000.94340.9597 2 0.7804 0.9057 0.9478430.67470.86760.9375 890.76290.90250.9464 3 0.5245 0.7610 0.9085440.49920.72370.8990 900.59210.82320.9247 4 0.5418 0.7851 0.9137460.71260.88800.9418 910.51540.75010.9052 5 0.6215 0.8359 0.9297470.72910.88770.9432 920.51420.75530.9052 6 0.7815 0.9106 0.9482480.64650.85240.9335 930.77320.90410.9473 7 0.5629 0.7852 0.9179490.62550.84920.9310 940.56420.80890.9197 8 0.5025 0.7423 0.9010500.68340.87120.9386 950.53510.76170.9108 9 0.4443 0.6769 0.8750510.92540.93700.9568 960.42630.65400.8635 10 0.5928 0.8268 0.9251520.82260.91600.9507 970.71650.88820.9423 11 0.4196 0.6468 0.8588530.64640.86010.9339 980.72520.88990.9430 12 0.5750 0.8058 0.9210540.67310.86220.9369 990.98320.94230.9590 13 0.5169 0.7505 0.9055580.44220.66190.8715 1000.69310.89220.9405 14 0.7623 0.8991 0.9462590.74100.89780.9445 1010.78470.91270.9484 15 0.8357 0.9214 0.9516600.52180.74750.9067 1020.45200.67960.8781 16 0.7636 0.9031 0.9465620.77480.90490.9473 1030.64520.84510.9329 17 0.6816 0.8711 0.9383630.89210.93120.9550 1040.66620.85820.9361 18 0.3519 0.5667 0.7921640.66260.86650.9361 1050.73780.89130.9441 19 0.5234 0.7633 0.9080650.64940.84580.9334 1060.83280.92470.9519 20 0.3715 0.5882 0.8163660.63850.85480.9330 1070.70280.88330.9407 21 0.6852 0.8766 0.9387670.62660.84130.9307 1080.43220.64820.8656 22 0.7043 0.8831 0.9409680.86050.92740.9532 1090.86100.92520.9533 23 0.6578 0.8565 0.9350690.51320.74540.9041 1100.84430.92030.9522 24 0.4570 0.6933 0.8821700.45450.68410.8798 1110.52790.75470.9087 25 0.5667 0.7863 0.9187710.51400.75090.9051 1120.67750.87430.9380 26 0.6068 0.8256 0.9271720.57140.80380.9205 1130.44170.65960.8716 27 0.5650 0.7898 0.9186730.54190.76580.9135 1140.43750.65280.8684 28 0.5413 0.7733 0.9128740.58340.80090.9224 1150.53510.76650.9108 29 0.4531 0.6855 0.8794750.71530.88400.9420 1160.64160.85460.9333 30 0.6183 0.8381 0.9294760.57910.81420.9224 1170.55530.79050.9170 31 0.5909 0.8183 0.9245770.69960.88560.9407 1180.79120.91080.9486

PAGE 116

101 Table 5.4. Continued. Farm id Deterministic Half-normal Exponent ialFarm idDeterministicHalf-normalExponenti al Farm idDeterministicHalf-normalExponent ial 32 0.4876 0.7191 0.8952780.57330.80240.9206 1190.77520.90590.9474 33 0.6047 0.8212 0.9265790.58820.81060.9234 1200.70670.88000.9412 34 0.6618 0.8632 0.9358800.44530.66850.8740 1210.64570.85340.9336 35 0.8975 0.9320 0.9553810.75530.90010.9459 1220.79250.91610.9491 36 0.5060 0.7369 0.9016820.81300.91950.9504 1230.58940.81750.9240 37 0.6153 0.8302 0.9287830.62910.84290.9310 1240.60390.81940.9267 38 0.6473 0.8500 0.9335840.56290.80380.9185 1250.50090.72410.9000 39 0.7643 0.9036 0.9466850.37640.59580.8214 1260.52890.77320.9104 40 0.4910 0.7182 0.8962860.69010.87780.9394 1270.48820.73090.8963 Mean0.63020.81940.9220 Std Dev0.13410.08710.0297 Minimum0.35190.56670.7921 Maximum1.00000.94340.9597

PAGE 117

102Table 5.5. Ordinal rank of farm according to the level of technical efficiency Ranking Farm Id. Deterministic Farm Id. Ha lf-normal Farm Id. Exponential Ranking Farm Id. Deterministic Farm Id. Half-normal Fa rm Id. Exponential 1 88 1.0000 88 0.9434 88 0.9597 42 21 0.6852 21 0.8766 21 0.9387 2 99 0.9832 99 0.9423 99 0.9590 43 50 0.6834 112 0.8743 50 0.9386 3 51 0.9254 51 0.9370 51 0.9568 44 17 0.6816 50 0.8712 17 0.9383 4 35 0.8975 35 0.9320 35 0.9553 45 112 0.6775 17 0.8711 112 0.9380 5 63 0.8921 63 0.9312 63 0.9550 46 43 0.6747 43 0.8676 43 0.9375 6 109 0.8610 68 0.9274 109 0.9533 47 54 0.6731 64 0.8665 54 0.9369 7 68 0.8605 109 0.9252 68 0.9532 48 104 0.6662 34 0.8632 64 0.9361 8 110 0.8443 106 0.9247 110 0.9522 49 64 0.6626 54 0.8622 104 0.9361 9 15 0.8357 15 0.9214 106 0.9519 50 34 0.6618 53 0.8601 34 0.9358 10 106 0.8328 110 0.9203 15 0.9516 51 23 0.6578 104 0.8582 23 0.9350 11 52 0.8226 82 0.9195 52 0.9507 52 65 0.6494 23 0.8565 53 0.9339 12 82 0.8130 122 0.9161 82 0.9504 53 38 0.6473 66 0.8548 121 0.9336 13 122 0.7925 52 0.9160 122 0.9491 54 48 0.6465 116 0.8546 48 0.9335 14 118 0.7912 101 0.9127 118 0.9486 55 53 0.6464 121 0.8534 38 0.9335 15 101 0.7847 118 0.9108 101 0.9484 56 121 0.6457 48 0.8524 65 0.9334 16 6 0.7815 6 0.9106 6 0.9482 57 103 0.6452 38 0.8500 116 0.9333 17 2 0.7804 119 0.9059 2 0.9478 58 116 0.6416 49 0.8492 66 0.9330 18 119 0.7752 2 0.9057 119 0.9474 59 66 0.6385 65 0.8458 103 0.9329 19 62 0.7748 62 0.9049 62 0.9473 60 83 0.6291 103 0.8451 49 0.9310 20 93 0.7732 93 0.9041 93 0.9473 61 67 0.6266 83 0.8429 83 0.9310 21 39 0.7643 39 0.9036 39 0.9466 62 49 0.6255 67 0.8413 67 0.9307 22 16 0.7636 16 0.9031 16 0.9465 63 5 0.6215 42 0.8388 5 0.9297 23 89 0.7629 89 0.9025 89 0.9464 64 30 0.6183 30 0.8381 30 0.9294 24 14 0.7623 81 0.9001 14 0.9462 65 42 0.6172 5 0.8359 42 0.9293 25 81 0.7553 14 0.8991 81 0.9459 66 37 0.6153 37 0.8302 37 0.9287 26 59 0.7410 59 0.8978 59 0.9445 67 26 0.6068 10 0.8268 26 0.9271 27 105 0.7378 100 0.8922 105 0.9441 68 33 0.6047 26 0.8256 124 0.9267 28 47 0.7291 105 0.8913 47 0.9432 69 124 0.6039 90 0.8232 33 0.9265 29 98 0.7252 98 0.8899 98 0.9430 70 10 0.5928 33 0.8212 10 0.9251 30 97 0.7165 41 0.8885 97 0.9423 71 90 0.5921 124 0.8194 90 0.9247 31 75 0.7153 97 0.8882 75 0.9420 72 31 0.5909 31 0.8183 31 0.9245 32 46 0.7126 46 0.8880 46 0.9418 73 123 0.5894 123 0.8175 123 0.9240 33 120 0.7067 47 0.8877 41 0.9413 74 79 0.5882 76 0.8142 79 0.9234 34 41 0.7057 77 0.8856 120 0.9412 75 74 0.5834 79 0.8106 76 0.9224 35 22 0.7043 75 0.8840 22 0.9409 76 76 0.5791 94 0.8089 74 0.9224 36 107 0.7028 107 0.8833 107 0.9407 77 12 0.5750 12 0.8058 12 0.9210 37 1 0.7002 22 0.8831 77 0.9407 78 78 0.5733 84 0.8038 78 0.9206 38 77 0.6996 1 0.8823 1 0.9406 79 72 0.5714 72 0.8038 72 0.9205 39 87 0.6955 120 0.8800 100 0.9405 80 25 0.5667 78 0.8024 94 0.9197 40 100 0.6931 87 0.8790 87 0.9399 81 27 0.5650 74 0.8009 25 0.9187 41 86 0.6901 86 0.8778 86 0.9394 82 94 0.5642 117 0.7905 27 0.9186

PAGE 118

103Table 5.5. Continued. Ranking Farm Id. Deterministic Farm Id. Half-normal Farm Id. Exponential 83 84 0.5629 27 0.7898 84 0.9185 84 7 0.5629 25 0.7863 7 0.9179 85 117 0.5553 7 0.7852 117 0.9170 86 73 0.5419 4 0.7851 4 0.9137 87 4 0.5418 28 0.7733 73 0.9135 88 28 0.5413 126 0.7732 28 0.9128 89 115 0.5351 115 0.7665 115 0.9108 90 95 0.5351 73 0.7658 95 0.9108 91 126 0.5289 19 0.7633 126 0.9104 92 111 0.5279 95 0.7617 111 0.9087 93 3 0.5245 3 0.7610 3 0.9085 94 19 0.5234 92 0.7553 19 0.9080 95 60 0.5218 111 0.7547 60 0.9067 96 13 0.5169 71 0.7509 13 0.9055 97 91 0.5154 13 0.7505 92 0.9052 98 92 0.5142 91 0.7501 91 0.9052 99 71 0.5140 60 0.7475 71 0.9051 100 69 0.5132 69 0.7454 69 0.9041 101 36 0.5060 8 0.7423 36 0.9016 102 8 0.5025 36 0.7369 8 0.9010 103 125 0.5009 127 0.7309 125 0.9000 104 44 0.4992 125 0.7241 44 0.8990 105 40 0.4910 44 0.7237 127 0.8963 106 127 0.4882 32 0.7191 40 0.8962 107 32 0.4876 40 0.7182 32 0.8952 108 0 0.4702 0 0.7079 0 0.8877 109 24 0.4570 24 0.6933 24 0.8821 110 70 0.4545 29 0.6855 70 0.8798 111 29 0.4531 70 0.6841 29 0.8794 112 102 0.4520 102 0.6796 102 0.8781 113 80 0.4453 9 0.6769 9 0.8750 114 9 0.4443 80 0.6685 80 0.8740 115 58 0.4422 58 0.6619 113 0.8716 116 113 0.4417 113 0.6596 58 0.8715 117 114 0.4375 96 0.6540 114 0.8684 118 108 0.4322 114 0.6528 108 0.8656 119 96 0.4263 108 0.6482 96 0.8635 120 11 0.4196 11 0.6468 11 0.8588 121 85 0.3764 85 0.5958 85 0.8214 122 20 0.3715 20 0.5882 20 0.8163 123 18 0.3519 18 0.5667 18 0.7921

PAGE 119

104Determinants of Technical Efficiency To understand the causes of variation in efficiency among producers, a logistic model (Equation 4.30) was used to regress technical efficiency values from the halfnormal and exponential distribution against selected socio-economic variables. This model was chosen because the technically efficient estimates were bracketed from zero to one. The independent variables were classified as: a) farmer characteristics such as education, experience, and frequency of visit by the owners; b) farm characteristics such as size, location, tenure, production system, and credit, and c) technological variables as in breeding system, milk production per total cows, milk production per milker, stocking rate, and frequency of contact with extension agents. Most of these variables are dummy variables, and they will cause multicollinearity problems if all are included in the model simultaneously. To avoid this problem the restriction of the summation of coefficient of dummies variables equaling zero ( k j j i i0) was imposed. The intercept term from this model represents an average value over all dummies variables. Thus, the estimates of the independent variables measure the deviation of each dummy variable from the average technical efficiency for all farms. Results from the half-normal and exponential distribution can be seen in table 5.6. All estimates were simultaneously different from zero at a high significance level (see F test). The model explained approximately 50% of total vari ation in technical efficiency for both distributions, which is considered adequate for cross sectional data. Because of the use of the logistic model, the sign of the coefficients should be read in opposite direction to know the effect on the dependent variable. The coefficients showed the expected sign

PAGE 120

105 except for education (DPEDU) and breeding system (DBRED), but they were not significant. Similarly, coefficients for land tenure (DTEN) and stocking rate (CARGANEF) were not different from the av erage technically efficient farm in both models. Table 5.6. Estimates of the determinants of technical efficiency Half – normal distributio n Exponential distribution Variables Variables’ namesEstimatet-statist icP-valueEstimatet-statisticP-value C Intercept -0.2296-0.74270.4597-1.6587-8.92780.0000 DPEDU Education 0.03840.44390.65820.02860.56470.5737 DPEXP Experience -0.2826-2.01140.0474-0.1729-1.80850.0741 DPPER Presence -0.1695-2.14970.0344-0.1255-2.07580.0409 CRED Credit 0.08081.51570.13330.06111.79000.0770 DSUG21 Size 0.46872.76330.00700.27312.71710.0080 DSUG31 Size -0.6261-3.73320.0003-0.3685-3.83940.0002 DSUG41 Size 0.26891.76800.08060.15321.67740.0971 Z21 Location 0.10640.93730.35130.06931.04100.3008 Z31 Location -0.2601-1.99060.0497-0.1594-2.15410.0341 Z41 Location 0.01820.19940.84240.00180.03240.9742 PSYST21 Production system0.07851.12930.26200.05721.27590.2055 PSYST31 Production system-0.2892-2.80290.0063-0.1852-2.76910.0069 PROD21 Cow productivity0.25412.31480.02300.14792.10570.0382 PROD31 Cow productivity-0.0894-1.00740.3166-0.0652-1.19010.2373 PROD41 Cow productivity-0.2543-2.02720.0458-0.1516-2.21340.0295 PROD51 Cow productivity-0.3843-1.78720.0775-0.2370-1.96030.0532 DBRED Breeding system 0.04890.69210.49080.04000.92530.3574 DTEN Land tenure 0.05541.05480.29450.02480.75520.4522 DTECHN Technical assistance -0.0956-1.70720.0914-0.0560-1.57230.1196 LTMILKER Labor productivity -0.0289-4.29980.0000-0.0173-4.12320.0001 LTMILKSQ L. productivity squared 0.00013.59060.00060.00013.75470.0003 CARGANEF Stocking rate -0.0004-0.34440.7314-0.0003-0.46040.6464 F 3.8477 3.8363 R-squared 0.4990 0.4982 Adj R-squared 0.3693 0.3684 Log likelihood -61.1130 -7.6162 N of Obs 108 108 Although the coefficient’s behavior was similar, the numerical value and the level of statistical significance varied depending on whether half-normal or exponential technical efficiency values were used. Variables including experience (DPEXP),

PAGE 121

106 presence (DPPER), farm size (DSUG), location (Z), production system (PSYST), milk production per cow (PROD), and liter per milker were significant at the 95% or higher level with respect to the average technically efficient farm in both models, while credit (CRED) and frequency of technical assi stance (DTECHN) were only significant for a one-tail test. The standardized coefficients for both distributions allow ranking of the independent variable according to relative impact on the dependent variable because the scale effect is removed (Table 5.7). Rankings were very similar with little variation. Productivity per milker (LTMILKER), farm medium-high size between 400 and 575 ha (DSUG31), and production system cow-steer ( PSYST31) affected in a greater magnitude the average technical efficiency of the farms (Figs. 5.4 and 5.5). Dummy variable t-statistic values for the half-normal and exponential distribution are listed in tables 5.8 and 5.9 respectively. They allow detection of statistical differences for the different dummy variable levels, in contrast to the coefficients presented in table 5.6 where they are compared against the average technically efficient farm. An analysis of each variable will follow:

PAGE 122

107 Table 5.7. Standardized coefficients of the determinants of technical efficiency Half-Normal distribution Exponential distribution Variables EstimateEstimate DPEDU 0.04830.0587 DPEXP -0.1708-0.1701 DPPER -0.1896-0.2285 CRED 0.12770.1571 DSUG21 0.50500.4789 DSUG31 -0.6282-0.6017 DSUG41 0.32480.3013 Z21 0.12560.1331 Z31 -0.3272-0.3264 Z41 0.02300.0037 PSYST21 0.10190.1209 PSYST31 -0.3883-0.4047 PROD21 0.26880.2547 PROD31 -0.0899-0.1067 PROD41 -0.2071-0.2010 PROD51 -0.2203-0.2212 DBRED 0.07300.0972 DTEN 0.08960.0652 DTECHN -0.1587-0.1511 LTMILKER -1.1705-1.1412 LTMILKSQ 1.11541.0983 CARGANEF -0.0408-0.0580

PAGE 123

108 -0.6-0.4-0.20.00.20.40.60.81.01.21.4DSUG21 DSUG41 PROD21 CRED Z21 PSYST21 DTEN DBRED DPEDU Z41 CARGANEF PROD31 DTECHN DPEXP DPPER PROD41 PROD51 Z31 PSYST31 DSUG31 LTMILKERVariablesEstimates Half-Normal distribution Figure 5.4. Ranking of standardized coefficients for the determinants of technical efficiency (half-normal distribution) -0.6-0.4-0.20.00.20.40.60.81.01.21.4 DSUG21 DSUG41 PROD21 CRED Z21 PSYST21 DBRED DTEN DPEDU Z41 CARGANEF PROD31 DTECHN DPEXP PROD41 PROD51 DPPER Z31 PSYST31 DSUG31 LTMILKERVariablesEstimates Exponential distribution Figure 5.5. Ranking of standardized coefficients for the determinants of technical efficiency (Exponential distribution).

PAGE 124

109Farmer Characteristics Variables To classify experience two levels were esta blished, EXP1 for farmers with less than 5 years of experience and EXP2 for farmers with more than 5 years of experience. Experience variable DPEXP, which represents the difference of farmers with more than 5 years of experience with respect to farmers having average experience, shows the right sign and was significant at the 95% level for a one-tail test in both models. Farmers with more than five years of experience were more efficient than those with the average experience (Table 5.6) and younger farmers (Tables 5.7 and 5.8). Possible years of experience compensate the lack of education, which was not significant. To study the presence of the owner in the farm two levels were also established, PER1 for farmer with less than two days of presence per week, and PER2 = 1 for producers that visited or stayed in the farm twice or more times a week. Farmers with two days of presence or more were more efficient than farmers with lower on-farm frequency. This result suggests that managerial functions such as direction, supervision, and control, are executed more efficiently with a higher visit frequency or farmer presence on the farm. Farm Characteristic Variables Credit (CRED) The credit variable showed a significant negative impact on average technical efficiency at the 95% level for a one-tail test in the exponential model (Table 5.6). The half-normal model showed the same but with a lower level of significance (90%). This result was expected because of the increment in interest rate reported during recent years, which reached levels higher than 60%. The same response was obtained when the two levels of this variable (CRED1 for no credit and CRED2 for credit) were compared.

PAGE 125

110 Producers who obtained credit in the most recent 10 years were less efficient than farmers without credit at the 95% level (Tables 5.8 and 5.9). Producers who obtained loans in previous years had to sell part of their actives in order to cancel their debts because of high financial cost. Positive effects of credit over farm efficiency have been reported in the literature. However, for DPCS a review or reformulation of the credit program for this sector is necessary. As the results in this research indicate, it is difficult to improve the efficiency of this system with such poor credit conditions. Few agricultural businesses can support high interest rates. Farm size (DSUG) To measure the effect of farm size on techni cal efficiency, farms were classified in four categories: i ) small farms (< 300 ha SUG1), ii ) medium-small farms (between 300 and 400 ha SUG2), iii ) medium-large farms (between 400 and 575 ha SUG3), and iv ) large farms (> 575 ha SUG4). According to the results listed in table 5.6, for both models, farm sizes medium-large had higher efficiency than the average farm size at a high level of significance (95%), while farms of medium-small size were less efficient than the average farm size, again with high significance. Large farms were not statistically less efficient than the average size. Comparative analysis of different levels of this variable (Tables 5.8 and 5.9) indicat ed that medium-large farms were significantly more efficient at the 90% or higher level than small and large in both models. However, small farm were more efficient than farm sizes between 300 and 400 ha, and more than 575 ha. This result suggests that the optimal farm size is between 400 and 575 ha. This finding is contrary to the results found in empirical works where positive and negative relationships to farm size have been reported for developed and developing countries,

PAGE 126

111 respectively. An explanation for this behavior could be that the available technology to improve the efficiency of the dual-purpose cattle system has been based in a high investment cost and a high level of operation, which is not profitable for small farms because of economies of size, and not too attractive for larger farmers who tend to maximize leisure over profit. Location (Z) Farms located in Rosario and Machiques de Perija counties (Z31) were statistically the most efficient at the 95% level in both models when they were compared to the average zone (Table 5.6) and to other regions (Table 5.8 and 5.9). No statistical significance was found for the other regions when they were compared to the average, or when they where compared to each other despite the fact that region 1 (Z1) presented the best agro-ecological conditions. Empirical evidence suggests that productivity indexes can be at least doubled in Z1 area compared with Z2, Z3, and Z4 areas where productivity indexes can be improved but not in the same proportion. Production system (PSYST) The dual-purpose cattle system has been classified in three main systems: cow-calf (PSYST1), cow-yearling (PSYST2), and cow-st eer (PSYST3). Producers oriented to the cow-steer system (PSYST31) were significantly (99%) more efficient than the others when compared to the average farm (Table 5.6). Similarly, farmers oriented to steer production were more efficient than farmers more inclined to the milk production (PSYST1 and. PSYST2) (Table 5.8 and 5.9). This result may explain the prevalence of the DPCS in tropical areas. Milk and beef production in tropical areas is based on grazing of natural and improved pastures where nutritive value is not as high as for temperate pastures. Empirical evidence suggests that the nutritive value of tropical

PAGE 127

112 pasture makes beef production in tropical areas more efficient than milk production because of the higher nutritional requirement for milk production and environmental conditions. Policies and managerial decisions addressing farms for cow-steer production will have greater impact on efficiency of the system, especially considering that the cowsteer production system is the one of the three systems that most affects farm efficiency (Table 5.7). However, milk and beef price is a factor to take into account because farmer decisions are based on these prices. Land tenure (DTEN) This variable shows no impact on technical efficiency (Tables 5.6, 5.8, and 5.9). The relationship between efficiency and tenure has not been clear in empirical work. Some authors suggest that tenants compensate for the use of high capital technologies with technologies that require less fixed capital in order to increase their efficiency. Technological Variables Breeding system (DBRED) The effect of the breeding system on average technical efficiency was not statistically significant in either of the two models. It showed an unexpected negative sign. (Tables 5.6, 5.8, and 5.9). Farmers using artificial insemination were not more efficient than farmers using natural breeding. Artificial insemination has been a common practice introduced in this system in the last 15 years to improve animal genetics and milk production per cow. A weakness in the information collected in the survey is that it does not indicate whether farmers who were not using AI at the time of the survey had used it in previous years. Previous use information would assist in making a more complete analysis of this variable.

PAGE 128

113Frequency of technical assistance (DTECHN) DTECHN variable had the expected sign and it was significant at the 95% level for a one-tail test in both models when it was compared to average technical efficiency (Table 5.6). A higher frequency (once a month or higher) of technical assistance increases the technical efficiency of the farms. Comparison between the two levels of this variable (DTECHN1, for less than once month or no assistance, and DTECHN2 for once a month visit or more) indicates also that the level of significance varied depending on the model used (Tables 5.8 and 5.9). The half normal model had a higher level of significance (95%) for a one-tail test than the exponential model. Higher frequency of technical assistance is generally associated with the use of a higher level of technology such as artificial insemination, embryo transplant, rotational stocking, sanitary program, and weed control which required a high level of knowledge and increased major supervision. Cow productivity (PROD) Five different levels of total milk production per cow-year were established: i) less than 1000 l (PROD1), between 1000 and 1500 l (PROD2), 1500 and 2000 l (PROD3), 2000 and 2500 l (PROD4), and greater than 2500 l (PROD5). Farms with production levels greater than 2000 l per cow were more efficient than the average production levels at 95% in both models (half-normal and exponential) (Table 5.6). Comparison among the different levels of production indicates that efficiency of farms with production levels greater than 1500 l were significantly more efficient than farms with production levels less than 1500 (PROD1 and PROD2). Statistical differences in technical efficiency were not found among the farms with production levels higher than 1500 (PROD3, PROD4, and PROD5).

PAGE 129

114 Total liters of milk per cow-year depend upon the number of cows in production and production per lactation. The number of cows in production in turn depends on the length of the calving interval, while production per lactation is a function of the length of period and genetic factors. These elements show the importance of reproductive and genetic factors in milk production and therefore the efficiency of the farms. Unfortunately, information about these variables was not available in the survey. If milk production per cow is to be improved, programs targeted to improve reproduction indices and animal genetics should be implemented. A balance between milk and beef production should be established because, as was indicated in previous discussion, a production system oriented to beef resulted in more farm efficiency than were farms oriented to milk production. Labor productivity (LTMILKKER and LTMILKSQ) This variable had the highest positive impact on technical efficiency of the farm (Table 5.6). A significant quadratic relationship between this variable and the average technical efficiency of farms was found. As the productivity per milker increased, technical efficiency increased to a certain level and then it started to decrease. This important resource, labor in the form of milker, was the main item in the cost structure of this system (33%). Achieving the optimal level of productivity per laborer will help to improve technical efficiency of the farms. Stocking rate (CARGANEF) Contrary to production per cow, stocking rate did not have significant effect on technical efficiency of the farms. It seems that in the dual-purpose cattle system animal productivity has a greater impact on efficiency of the farm than land productivity. Farmers utilizing a higher stocking rate were as efficient as farmers with lower rates.

PAGE 130

115 Increases in stocking rate require not only fertilizer, which also was not a significant input in the production frontier of this system, but also more animals. Similarly, higher stocking rates require spatial arrangements of the pastures (fences and water places) resulting in great amounts of fixed investment. Maybe, this situation combined with the credit conditions, has driven farmers to place more attention to animal productivity which requires fewer fixed investments to improve farm efficiency.

PAGE 131

116Table 5.8. t-statistic for the dummy variables of the determ inants of technical efficiency (half-normal distribution) Bases DPEDU2 DPEXP2 DPPER2CRED2DS UG2 DSUG3 DSUG4 Z2 Z3 Z4 PSYST2 PSYS T3 PROD2 PROD3 PROD4 PROD5 DBRED2DTEN2DTECHN2 DPEDU1 0.44 DPEXP1 -2.01b DPPER1 -2.15b CRED1 1.52 DSUG1 2.33b-2.10b2.18b DSUG2 -4.17 a -0.71 DSUG3 3.21a Z1 -0.17-2.08b-0.79 Z2 -1.71c-0.56 Z3 1.50 PSYST1 -0.96-2.62a PSYST2 -2.54b PROD1 -1.36-3.03 a -3.01 a -2.48b PROD2 -2.47b-2.65 a -2.18b PROD3 -1.03-1.13 PROD4 -0.51 DBRED1 0.69 DTEN1 1.05 DTECHN1 -1.71c a=P 0.01; b= 0.01


PAGE 132

117Table 5.9. t-statistic for the dummy variables of the dete rminants of technical efficiency (Exponential distribution) Bases DPEDU2 DPEXP2 DPPER2CRED2DS UG2 DSUG3 DSUG4 Z2 Z3 Z4 PSYST2 PSYS T3 PROD2 PROD3 PROD4 PROD5 DBRED2DTEN2DTECHN2 DPEDU1 0.56 DPEXP1 -1.81c DPPER1 -2.08b CRED1 1.79c DSUG1 2.31b-2.25b1.88c DSUG2 -4.08 a -0.72 DSUG3 3.26 Z1 -0.19-2.26b-0.93 Z2 -1.86c-0.72 Z3 1.52 PSYST1 -0.84-2.60b PSYST2 -2.53b PROD1 -1.47-2.99 a -3.11 a -2.65 a PROD2 -2.32b-2.63 a -2.24b PROD3 -0.96-1.19 PROD4 -0.63 DBRED1 0.93 DTEN1 0.76 DTECHN1 -1.57 a=P 0.01; b= 0.01


PAGE 133

118 CHAPTER 6 SIMULATION OF THE DETERMINANTS OF TECHNICAL EFFICIENCY Simulation as defined by Pindyck and Rubinf eld (1998) is a mathematical solution to a simultaneous set of difference equations, where difference equation relates the current value of one variable to current and past values of other variables. Simulation models have been used not only for the analysis and prediction of public policy but also for management decisions. In this chapter a simulation model is constructed and used to measure the impact of farmer and farm characteristics, and technological variables on the technical efficiency of DPCS. The chapter is divided into three main sections, the first section related to the construction of the simulation model, the second deals with the analysis of the simple effect of the statistical significant socioeconomic variables that affect efficiency, and in the last section a combined effect of the main socio-economic variables on the efficiency of the farm is studied and analyzed. Simulation Model The coefficients obtained from equation 4.32 and shown in table 5.5 were used to develop the following simulation model: x iTE exp 1 1 where, iCARGANEF LTMILKSQ LTMILKER DTECHN DTEN DBRED PROD PROD PROD PROD PSYST PSYST Z Z Z DSUG DSUG DSUG CRED DPPER DPEXP DPEDU X 33 32 31 30 28 26 24 23 22 21 19 18 16 15 14 12 11 10 8 6 4 2 051 41 31 21 31 2 41 31 21 41 31 21

PAGE 134

119 Simulations of the technical efficiency values were determined at the mean values of liters per milker, liters per milker-squared, and stocking rate for all the independent variables. The results from the simulation model are listed in Appendix C and D for the half-normal and exponential distribution respectively, where 0.7652 and 0.9037 represent the average technical efficiency of the farms for each distribution. These values were obtained by the addition of the intercept term and the mean values of liters per milker, liters per milker squared, and stocking rate multiplied by their respective coefficient for the half-normal and exponential model respectively. A discussion of the impact of the different variables on technical efficiency is presented through the analysis of different graphs. Single Effect of Socio-economic Variables Farmer Characteristics According to the results obtained in chapter 5, producer experience and farmer presence on the farm were the only farmer characteristic that make a statistically significant impact on technical efficiency. Figures 6.1 and 6.2 show how the experience of the farmers increases the technical efficiency of the farms. Farmers with more than 5 year of experience were around 14% more efficient than farmers with less than five year of experience, and 6% more efficient than the average experience when the half-normal model was analyzed.

PAGE 135

120 0.60 0.70 0.80 0.90 1.00 Less than 5 years More than 5 years Less than 5 years More than 5 yearsLevels Technical efficiency Half-normal Exponential Figure 6.1. Impact of producer experience on technical efficiency -8.00 -6.00 -4.00 -2.00 0.00 2.00 4.00 6.00 8.00 Less than 5 years More than 5 years Less than 5 years More than 5 yearsLevelsPercentage of change respect to the average farm Half-normal Exponential Figure 6.2. Impact of producer experience on average technical efficiency

PAGE 136

121 The same tendency but with minor impact was found when the exponential model was used to calculate the technical efficiency of the farms. However, this factor (producer experience) could have a negative impact on efficiency in the short term if we consider that more than 30% of farmers are at the age of retirement (> 60years old) and they could be replaced by inexperienced persons. In relation to the owner presence on the farm, a positive relationship was found (Figure 6.3). Farmers that stayed on the farm more than twice a week captured about 8% more of the technical efficiency increment with respect to farmers that stayed on the farm less than twice a week. Likewise, the same farmers experienced around 4% of efficiency increment with respect to the average technical efficiency (Figure 6.4). The same behavior was found with the exponential values of technical efficiency however, the percentage of increment in technical efficiency was lower than the half-normal model. A policy oriented to promoting the presence of farmers in the work place will help to increase the efficiency of this sector if we considered that around 20% of farmers stay on the farm less than 2 days per week. Farm Characteristics Credit The credit variable only displayed significance in the exponential model at the 95% level for a one-tail test. Credit decreased the efficiency around 1% with respect to the farmers who did not use credit and around 0.6% respect to the average farm (Figure 6.5 and 6.6). Considering that 35% of farmers requested loans and were affected negatively, better financial policy address to this sector could improve the efficiency of the DPCS.

PAGE 137

122 0.60 0.70 0.80 0.90 1.00 Less than twice a week Twice a week or higher Less than twice a week Twice a week or higherLevelsTechnical efficiency Half-normal Exponential Figure 6.3. Impact of producer presence on technical efficiency -6.00 -4.00 -2.00 0.00 2.00 4.00 6.00 Less than twice a week Twice a week or higher Less than twice a week Twice a week or higherLevelsPercentage of change respect to the average farm Half-normal Exponential Figure 6.4. Impact of producer presence on average technical efficiency

PAGE 138

123 0.60 0.70 0.80 0.90 1.00 No creditCreditLevelsTechnical efficiency Exponential Figure 6.5. Impact of credit on technical efficiency -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00 No creditCreditLevelsPercentage of change respect to the average farm Exponential Figure 6.6. Impact of credit on average technical efficiency

PAGE 139

124 Farm size The effect of the farm size on technical effi ciency is shown in figures 6.7 and 6.8. Farm size between 400 and 575 ha (SUG3) had the highest efficiency values. This size of farm contributed to the efficiency increment by around 10%, 28%, and 20% for halfnormal, and 3%, 6%, and 5% for exponential model when for compare to farm sizes less than 300 ha (SUG1), 300 to 400 (SUG2), and greater than 575 (SUG4) ha respectively (Figure 6.7). 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 SUG1SUG2SUG3SUG4SUG1SUG2SUG3SUG4LevelsTechnical efficiency Half-normal Exponential Figure 6.7. Impact of farm size on technical efficiency In relation to the average farm, the increment on efficiency was around 12% and 3% for the half-normal and exponential models respectively; while farm size between 300 and 400 ha (SUG2) and greater than 575 ha (SUG4) were more inefficient (Figure 6.8). The relation between farm size and efficiency or productivity has been widely studied mainly in developing countries due to the land reform implications. Empirical

PAGE 140

125 evidence supports the existence of an inverse relationship between farm size and productivity (Berry and Cline, 1979; Cornia, 1985; Ghose, 1979; Taslim, 1989; all cited by Langedyk, 2001). However partial productivity indexes were used to measure this relationship. In this study a total factor productivity index is used to avoid the shortcoming of partial productivity indices. The results obtained in this study do not support the inverse relationship. This study indicates a quadratic relationship between farm size and efficiency. Efficiency increased to a certain farm size (SUG3) where it started to decline. This behavior could be partially explained by the economies of size for medium-large farm, and by the maximization of leisure for the larger farmers. -15.00 -10.00 -5.00 0.00 5.00 10.00 15.00 SUG1SUG2SUG3SUG4SUG1SUG2SUG3SUG4LevelsPercentage of change respect to the average farm Half-normal Exponential Figure 6.8. Impact of farm size on average technical efficiency

PAGE 141

126 Location Four different geo-political areas are found in Zulia state. Farms located in zone 3, which include Machiques and Rosario de Perija counties, were at least 6% and 2% more efficient than the farms located in the other regions for the half-normal and exponential models respectively; and 5.6% and 1.5% more efficient than the average farm respectively (Figures 6.9 and 6.10). A detail study about the specific characteristic of the different locations is necessary to explain this behavior because the south part of the state (Z1), which is the agro-ecological area with more potential for agricultural production according to empirical evidence, was less efficient. Factors such as managerial capacity of the farmers and agribusiness development in the location could help to explain this behavior. 0.60 0.70 0.80 0.90 1.00 Z1Z2Z3Z4Z1Z2Z3Z4ZonesTechnical efficiency Half-normal Exponential Figure 6.9. Impact of location on technical efficiency

PAGE 142

127 -4.00 -2.00 0.00 2.00 4.00 6.00 8.00 Z1Z2Z3Z4Z1Z2Z3Z4ZonesPercentage of change respect to the average farm Half-normal Exponential Figure 6.10. Impact of location on average technical efficiency Production system Technical efficiency of the farms increases as the beef production increases. Around 12% and 8%; and 3%, and 2% of the efficiency increment were obtained when the production system moved from cow-calf, and cow-yearling to cow-steer in the halfnormal and exponential models respectively (Figure 6.11). Likewise, increments in technical efficiency related to the average fa rm of about 6% and 2% were obtained when the cow-steer system was implemented in both models respectively (Figure 6.12). The results obtained in this study support the findings that beef production in tropical areas is more efficient than milk production. The cow-steer system was more efficient than the other two systems (cow-calf and cow-steer). Considering that this system is implemented by only 12% of the farmers, managerial decisions and public policy address to beef production could increase significantly the effici ency of DPCS. However, DPCS farmers

PAGE 143

128 mainly use the cow-yearling system (84%). They consider this system more flexible than the other two systems because due to market conditions (milk and beef price) transitions between producing either more beef or milk will be easier and shorter. Possibly this conclusion is related indirectly introducing price or allocative efficiency into the study by considering gross revenue as a dependent variable. This relationship, however, could change through the time as milk and beef price changes. Time series date was not available to make a more complete analysis. 0.60 0.70 0.80 0.90 1.00 Cow-calfCowyearling Cow-steerCow-calfCowyearling Cow-steerTechnical efficiency Half-normal Exponential Figure 6.11. Impact of production system on technical efficiency

PAGE 144

129 -6.00 -4.00 -2.00 0.00 2.00 4.00 6.00 8.00 Cow-calfCowyearling Cow-steerCow-calfCowyearling Cow-steerPercentage of change respect to the average farm Half-normal Exponential Figure 6.12. Impact of production syst em on average technical efficiency Technological Variables Cow productivity Technical efficiency of the farm increased as milk production per cow rose. However, as milk production per cow increased the differences in technical efficiency among the levels of production narrowed. At least 9% of the efficiency increment can be achieved when the level of production per cow is greater than 1500 l. (Figure 6.13). Percentage of change of 2%, 6%, and 8% in technical efficiency with respect to the average farm were reached for levels of production of 1500 to 2000, 2000 to 2500, and more than 2500 l. respectively (Figure 6.14). The same tendency is observed in the exponential model but with lower percentage of change on technical efficiency. A frequent question directed by farmers to technical persons is what kind of animal is more efficient for this system in terms of production and adaptability. According to

PAGE 145

130 this study no statistical difference was found among levels of production greater than 1500; however, a significant economic difference could be found among these levels. This finding indicates that well adapted cows with production levels higher than 2500 l and less than 3260 l, which represent the upper bound in this study, could be more appropriate for this system. Managerial decisions and public policy addressed to achieve this objective will improve the efficiency of DPCS considering that around 50% of farmers have production levels per cow of less than 1500 l. (Table 4.3). 0.60 0.70 0.80 0.90 1.00Equal or less than 1000 l Higher than 1000 to 1500 l Higher than 1500 to 2000 l Higher than 2000 to 2500 l Higher than 2500 l Equal or less than 1000 l Higher than 1000 to 1500 l Higher than 1500 to 2000 l Higher than 2000 to 2500 l Higher than 2500 lLiters per yearTechnical efficiency Half-normal Exponential Figure 6.13. Impact of production per cow on technical efficiency

PAGE 146

131 -16.00 -12.00 -8.00 -4.00 0.00 4.00 8.00 12.00Equal or less than 1000 l Higher than 1000 to 1500 l Higher than 1500 to 2000 l Higher than 2000 to 2500 l Higher than 2500 l Equal or less than 1000 l Higher than 1000 to 1500 l Higher than 1500 to 2000 l Higher than 2000 to 2500 l Higher than 2500 lLiters per yearPercentage of change respect to the average farm Half-normal Exponential Figure 6.14. Impact of production per cow on average technical efficiency Frequency of technical assistance This variable only displayed significance in the half normal model at the 95% level for a one-tail test. Farmers that received technical assistance with a frequency equal or higher than once a month were around 5%more efficient than farmers with less frequency or without this service for the half-norma l (Figure 6.15). Likewise, an efficiency increment of about 2% for the average farm was reported (Figure 6.16). Farmers without or receiving technical assistance less than once a month represent around 56% of total farmers. This finding implies that efficiency of these farms can improve substantially if farmers hire a technical assistance service or government implements some kind of policy to offer these services mainly to small producers who cannot afford to buy service.

PAGE 147

132 0.60 0.70 0.80 Less than once monthsOnce a months or higherLevelsTechnical efficiency Half-normal Figure 6.15. Impact of frequency of t echnical assistance on technical efficiency -3.00 -2.00 -1.00 0.00 1.00 2.00 3.00Less than once monthsOnce a months or higherLevelsPercentage of change respect to the average farm Half-normal Figure 6.16. Impact of frequency of techni cal assistance on average technical efficiency

PAGE 148

133 Labor productivity According to the standardized coefficients presented in table 5.7 Chapter 5, this variable had the greatest impact on technical e fficiency. The average technical efficiency 0.7652 and 0.9037 for the half-normal and the exponential models was calculated using the average liters per milker equal to 43012 liters and the average stocking rate equals 120.76 cows per 100 ha. Efficiency of the farms rises as labor productivity increases, but decreases some point (Figure 6.17). The highe st value of technical efficiency 0.86 and 0.93 was obtained when labor productivity reached approximately 104770 and 103742 thousand of liters per year for the half-normal and exponential models respectively after which efficiency started to decline. These values represent efficiency increments of 12% and 3% with respect to the average farm for the half-normal and exponential models respectively. However, this level of labor productivity only can be reached with milking machines. Empirical results and experience suggest that use of milking machines is conditioned by the degree of Bos indicus (Zebu) present in the blood of the dual-purpose cows. Cows with high degree of Bos indicus decrease milk production and present shorter lactation periods when milked without the calf present, a situation that usually occurs when milking machines are employed. In any case the use of this equipment should be conditioned by a cost-benefit analysis.

PAGE 149

134 0.60 0.70 0.80 0.90 1.00 020406080100120140160180Thousands of liters per milkerTechnical efficiency Half-normal Exponential Figure 6.17. Impact of labor productivity on technical efficiency Analysis of Alternative two Variables Simulations In this section, the impact of the combination of the main variables including labor and cow productivity, production system, producer presence, and technical assistance is analyzed using farm size as base. Only the half-normal model will be used for the discussion of the results because as was seen for the discussion of the simple effects, the exponential model shows the same behavior. Primarily, it will start with a discussion of the combination of farm size and labor productivity and it will continue with the other statistically significant variables. Farm Size and Labor Productivity The combined effect of farm size and labor productivity on technical efficiency is shown in figure 6.18. As the productivity of labor increases around 100 thousands liters per milker the impact on efficiency of the farm is very significant. With this level of

PAGE 150

135 productivity all farm sizes presented technical efficiency values above the average farm. These levels varied from 3% to 20% for the farm size from 300 to 400 ha (SUG2) and for 400 to 575 ha (SUG3) respectively. This result indicates that farmers should try to use the labor as efficiently as they can in order to increase the farm efficiency. -32 -24 -16 -8 0 8 16 24S U G 1 S U G 2 S U G 3 S U G 4 S U G 1 S U G 2 S U G 3 S U G 4 S U G 1 S U G 2 S U G 3 S U G 4 S U G 1 S U G 2 S U G 3 S U G 4Farm size Percentage of change respect to the average farm 22 65 108 151 Figure 6.18. Impact of farm size and labor productivity on average technical efficiency Farm Size and Production System Implementation of the cow-steer production system by farmers is another practice that helps to increase the efficiency of the farm as can be seen in figure 6.19. This system increased the efficiency of all the farm sizes by at least 10%. The major impact is obtained for farm sizes between 300 and 400 ha (SUG2) and cow-calf production systems where efficiency increments around 16% were obtained with the implementation of the cow-steer system.

PAGE 151

136 0.5 0.6 0.7 0.8 0.9 1S U G 1 S U G 2 S U G 3 S U G 4 S U G 1 S U G 2 S U G 3 S U G 4 S U G 1 S U G 2 S U G 3 S U G 4Farm size Technical efficiency Cow-calf Cow-yearling Cow-steer Figure 6.19. Impact of farm size and produc tion system on average technical efficiency Farm Size and Cow Productivity Although no statistical differences were found among the production level greater than 1500 l on the technical efficiency, levels of production greater than 2500 l improved the technical efficiency of the farm above the average farm for all farm sizes except for the 300 to 400 ha (SUG2). However, increments of 7% for farm size between 300 and 400 ha were achieved when the production per cows increases from 1500 to more than 2500 liters per cow. Likewise, increments of around 10% with respect to the average farm were obtained for the farm size SUG1 (Less than 300 ha) when the production level is increased more than 2500 liters per cow. Policies to address increases in the adoption of technologies that directly or indirectly increase the productivity per cow such as strategic supplementation, improved breeding, etc. will help to improve the efficiency of the farm considering that about 70% of the farms have less than 300 ha.

PAGE 152

137 -32.00 -24.00 -16.00 -8.00 0.00 8.00 16.00 24.00S U G 1 S U G 2 S U G 3 S U G 4 S U G 1 S U G 2 S U G 3 S U G 4 S U G 1 S U G 2 S U G 3 S U G 4 S U G 1 S U G 2 S U G 3 S U G 4 S U G 1 S U G 2 S U G 3 S U G 4Farm size Percentage of change respect to the average farm Less than 1000 1000 to 1500 1500 to 2000 2000 to 2500 Greater than 2500 Figure 6.20. Impact of farm size and cow productivity on average technical efficiency Farm Size and Producer Presence Presence of the owners on the farm is another factor that increased the efficiency of the farm independent of farm size. This aspect is very important as the efficiency of the system decreased in the recent years due to the absence of or less supervision of farms by the owners. Reduced supervision is due to insecurity in the countryside caused by the presence of Colombian guerrillas in Venezuelan. Policies to allow or to induce producers to stay on the farm twice a week or more will increase efficiency by at least 5% with respect to the average farm (Figure 6.21).

PAGE 153

138 -20 -10 0 10 20 SUG1 SUG2SUG3SUG4SUG1 SUG2SUG3SUG4Farm size Percentage of change respect to the average farm Less than twice a week Twice a week or higher Figure 6.21. Impact of farm size and produ cer presence on average technical efficiency Farm Size and Frequency of Technical Assistance The impact of frequency of technical assistance and farm size on technical efficiency was lower than the combination of the other variables discussed before. However, a higher frequency of technical assi stance increased the efficiency of the farm by at least 3% for all farm sizes with respect to the average farm (Figure 6.22). This result indicates that policies addresses to promote the extension services, public or private, will have an impact on the efficiency of this system if this service is given on a regular basis. Many other combinations could be analyzed. The combinations analyzed in this study give a good idea of where the managerial decisions and public policies should be focused and implemented to increase efficiency of dual-purpose cattle system. Demographic variables like producer experience and producer presence; technological variables such as productivity per cow and labor, and technical assistance; and policy to

PAGE 154

139 address the financial sector are the main factors to take into account as mean for increasing efficiency of this system. -20 -10 0 10 20 SUG1 SUG2SUG3SUG4SUG1 SUG2SUG3SUG4Farm sizePercentage of change respect to the average farm Less than once months Once a months or higher Figure 6.22. Impact of farm size and techni cal assistance on average technical efficiency

PAGE 155

140 CHAPTER 7 SUMMARY AND CONCLUSIONS The government of Venezuela has implemented several agricultural policies to increase the efficiency of agricultural businesses while helping achieve self-sufficiency and diminishing the loss of foreign currency due to food imports. Several policies have been addressed to increase efficiency in the milk industry. These policies promoted the establishment of intensive dairy systems similar to those used in developed countries based on the belief that the dual-purpose cattle system, the traditional system for producing milk in tropical areas, is inefficient. When protectionist policies in the agricultural sector were eliminated the intensive dairy systems collapsed due to excessive production costs. A milk deficit and increasing losses of foreign currency due to food imports followed. Few studies have been conducted to measure the efficiency of the dual-purpose cattle system (DPCS), using partial productivity indices. These indices do not consider the effect of total input on output as a measure of total efficiency. Likewise, these studies have not analyzed the determinants of DPCS efficiency to orient more effective public policy and managerial decisions and to save substantial financial resources and time for government and farmers. It is this gap that the present study tries close using the production frontier methodology. An objective of the research methodology in this study was to derive standards of efficiency based on the concept of total factor productivity in a first step, and to study the determinants of technical efficiency of the DPCS in a second step.

PAGE 156

141 Data used in this study came from a survey made in 1994 by La Unidad Coordinadora de Proyectos Conjuntos of La Universidad del Zulia, Zulia state, Venezuela, where DPCS is mainly located. Summary of the Findings Characterization of DPCS Descriptive statistics for DPCS show that more than 90% of farmers are literate persons with more than 20 years of experience (80%). Most of the farms are managed by the owners (66%), and more than 70% of farmers stay at or visit the farm daily during the week. This finding is contrary to the belief that most of owners live in big cities far away from the farm. The land where DPCS is settled belongs mostly to the government (64%). Generally, farm ownership and development spans more than 20 years (87%). Most of the farms (94%) are dedicated to cattle activity, with cow-yearling operation the main production system (84%). These farms present a development index of 93% indicating that most of the land has been dedicated to the production process. The type of animal used in this system is a breeding animal from Bos taurus, Bos indicus and Criollo . The structure of the herd varies depending on the system adopted by farmers, from cow-calf, cow-yearling, to cow –steer. Capital in land and forages, and capital in cattle represent the main investment (83%). Approximately 70% and 30% of the total revenue comes from milk and beef sales respectively. However this distribution changes depending on the production system employed by farmers. For example cow-calf system revenue comes mostly from milk production, as opposed to cow-steer where revenue from beef makes a significant contribution to farm earnings.

PAGE 157

142 In relation to partial productivity indices, the average stocking rate was about 1.2 animal units per ha, and liters per ha and per cow were approximately 1000 and 1500 per year, respectively. Labor represents the main input in the cost structure (33%) followed by supplement feed (17%). The ratio net income total capital, and revenue-cost is around 6% and 1.7:1 Bs respectively. The number of pastures per farm varies according to farm size, as does pasture size which ranged from 1.6 to 43 ha, with an average of 8.3 ha. The grazing method used is rotational stocking (92%) with utilization and rest periods less than 3 and 21 days respectively. Weed control is a common practice (98%) while fertilization is not. Herd cultural practices are dehorning (92%), belly bottom cure (97%), classification (84%), and branding of the he rd (93%). Herd classification by sex, production, or age is a common practice among farmers. In terms of breeding system, 55% of the farmers used natural breeding and only 26% artificial insemination. Milking is generally done manually (95%), with the calf present (96%), twice a day (98%) in a milking pen (96%). Herd feeding is based on grazing of improved pasture species, usually not fertilized. A health program is implemented by 95% of farmers, and about 88% enforce the complete regionally designed vaccination plan suggested by the Ministry of Agriculture. Control of external and internal parasites is a common practice ( 95%). Most farmers (60%) received technical assistance at least twice per month. The structure / organization of DPCS varies according to size and level of technology of the farms. Farmers tend to keep more technical records than accounting records (98% vs. 54%, respectively).

PAGE 158

143 Production Frontier and Technical Efficiency Production frontier models revealed that the main factors that positively affect milk and beef production were labor, capital in land, capital in machinery, capital in cattle, use of veterinary medicine, supplement feed, building maintenance, and taxes and insurances. Fertilizer, promoted by the extension service in order to improve the production of the DPCS, had no impact on production. A deterministic and two stochastic frontier models with based upon different distributions for the error term (half-normal and exponential) were used to estimate the technical efficiency of DPCS. The average t echnical efficiency for this system varied depending on the method used, and were 0.630, 0.819 and 0.922 for the deterministic, half-normal, and exponential models were obtained respectively. Higher values were obtained and expected for the stochastic frontiers models (half-normal and exponential) because the deterministic model considers all deviations from the frontier due to inefficiency. Comparing the half-normal and exponential models, the later presented higher efficiency values, because this model assumes a gap between observed and maximum output due to random effects. This is contrary to the half-normal distribution where the differences are assigned mainly to factors under control of farmers. The ordinal ranking of the farms according to technical efficiency estimates for the three models was similar. In conclusion, DPCS with average values of efficiency of 0.819 and 0.922 can be regarded as an efficient system to produce milk in tropical areas negating the statement that the dual-purpose cattle system is not efficient. Determinants of Technical Efficiency Producer experience and presence on the farm, farm size, location, production system, milk production per cow, liter per milker, credit and frequency of technical

PAGE 159

144 assistance were the main variables that explai ned variation in technical efficiency. The impact of these variables and their policy and managerial implications are summarized in the next section. Policy and Managerial Implications The impact of the socio-economic variables is summarized in this section only for the half-normal model because the exponential model followed the same behavior. Producer experience and producer presence played an important role in farm efficiency. Farmers with more than 5 years of experience were about 13% more efficient than farmers with less than 5 years of experience, and 6% more efficient than average experience, when the half-normal model was utilized. Similarly, producers staying on the farm more than twice a week experienced 7% increment in technical efficiency compared to farmers that stayed on the farm less than twice a week, and a 3% increment compared to the average technically efficient farm. In terms of farm characteristic variables, the results from this research suggest that the actual credit program available in this sector generated a negative impact on efficiency. Analyses shows a decrease of about 1% in efficiency for farmers who utilized loans compared to farmers who do not request credit as well a compared to the average farm. This outcome may be attributed to negative credit conditions, specifically to the high annual interest rate. Farm size was another factor that affected farm efficiency. A quadratic relationship was found in this study and farm sizes between 400-575 ha were the most efficient. This behavior could be explained by the size effect and leisure optimization of large farmers. Location was another farm characteristic that contributed to explain the efficiency variation among farms. According to this study farms located in region Z3 (Machiques

PAGE 160

145 and Rosario de Perija) were at least 7% more efficient than farms located in the other regions; and 6% more efficient than the average farm. A detailed study of the location characteristics including not only the agro-ecological conditions but also the agribusiness sector in each region along with managerial capacity of the farmers could help to explain why farmers located in the best agro-ecological conditions (Z1) were less efficient than farmers in the other regions. Cow-steer system was more efficient th an cow-calf and cow-yearling, augmenting farm efficiency at least 12% and 8% respectivel y. Likewise, increments in efficiency of about 6% with respect to the average farm were obtained when the cow-steer system was implemented. In relation to technological variables, cow and labor productivity, and frequency of technical assistance were the main factors in fluencing technical efficiency of farms. Levels of production per cow greater than 1500 liters increased farm efficiency at least 9%. Managerial and technical decisions dir ected to selection of animals with production levels greater than 1500 liters per year and well adapted to tropical conditions will have a great impact on the efficiency of the system in the medium and long term considering that 50% of farmers have production levels per cow less than 1500 liters. Labor productivity measured as liters of milk per milker was the variable with the greatest effect on efficiency. Increments in efficiency of 12% with respect to the average efficient farm were obtained when optimal productivity per labor was selected (104770 liters per milker). However, these productivity levels are obtained only using milking machines, which suggests that farmers should monitor their breeding animals to allow for

PAGE 161

146 effective use of this equipment. Empirical evidence indicates that animals with high degree of Bos indicus adapt badly to this equipment. Analysis of the combined effect of farm size and the other socio-economic variables was also simulated. Regardless of farm size, optimal levels of labor productivity increased the farm efficiency above the level of the average technically efficient farm. This finding suggests that im proved managerial decisions can improve the efficiency of this system. Similarly, the efficiency of the farm for all farm size increases at least 7% when the cow-steer system is implemented. Increments in milk production per cow, from less than 1000 liters to more than 2500 liters regardless of farm size, raise the efficiency of the farms about 20% compared to the average farm. Producer presence and frequency of technical assistance also incr ease farm efficiency at least 5% and 3% for all farm sizes respectively, and also with respect to the average farm. These simulation results show how the efficiency of DPCS can be improved if public policies and managerial decisions create and respond to a secure environment in rural areas, review or reformulate and employ an effective credit program, encourage and use optimal labor and cow productivity levels, provide and employ technical assistance, and analyze the intrinsic characteristics of specific location. Limitations and Future Studies The main constraints of this study stem from data limitations including the number of observations; a lack of time series data; and a lack of information on input price variables beside data was for 1994. These limitations also open further research avenues. More than a few hundred observations are needed to conduct production frontier models utilizing more flexible distribution of the random error (u). This would allow contrasting the two methodologies suggested in chapter 2 for analyzing the determinants of technical

PAGE 162

147 efficiency. The use of pooled data could allo w analysis not only of technical efficiency but also technological change in this syst em. Input price information would allow determination of economic efficiency for the system given by the interaction of technical efficiency and allocative or price efficiency. Some farms could be technically efficient but not price efficient. Others areas of research might character ize not only agro-ecological conditions but also the agribusiness structure of the different geo-political areas which will be necessary to determine what factors make one location more efficient than other. Studies based on farm size, or the level of technology used by the farmer could be conducted to determine efficiency according to these classificatory variables. The present study only measures technical efficiency based on an average of different technologies used by farmers.

PAGE 163

148 APPENDIX A QUESTIONNARIE I.INFORMACI”N GENERAL. 1.Identificacin del Productor. 1.1.Nombre (s): _________________________________________ 1.2.Nacionalidad: 1. Venezolano 2. Extranjero 1.3.Edad (es): 1. Menor de 21 aos 4. Entre 41 y 50 aos 2. Entre 21 y 30 aos 5. Entre 51 y 60 aos 8 3. Entre 31 y 40 aos 6. Ms de 60 aos 1.4.Grado de Instruccin: 1. Analfabeta 4. Secundaria 2. Lee y escribe 5. Universitaria 3. Primaria 6. Otros 1.5.Domicilio habitual. 1. La finca 3. Poblado lejano a la finca 2. Poblado cercano a la finca 4. Otros

PAGE 164

149 1.6.Permanencia en la finca. 1. Permanente 4. Dos veces por semana 2. Visita diaria 5. Otros 3. Cada dos das 1.7.Experiencia como productor 1. Ninguna 4. Entre 11 y 15 aos 2. Menos de 5 aos 5. Entre 16 y 20 aos 3. Entre 5 y 10 aos 6. Ms de 20 aos 1.8. Identific acin del Gerente. 1.8.1.Nombre (s):_____________________________ 1.8.2.Edad (es): 1. Menor de 21 aos 4. Entre 41 y 50 aos 2. Entre 21 y 30 aos 5. Entre 51 y 60 aos 3. Entre 31 y 40 aos 6. Ms de 60 aos 1.8.3.Nacionalidad 1. Venezolano 2. Extranjero 1.8.4.Grado de Instruccin: 1. Analfabeta 4. Secundaria 2. Lee y escribe 5. Universitaria 3. Primaria 6. Otros

PAGE 165

150 1.8.5.Domicilio habitual. 1. La finca 3. Poblado lejano a la finca 2. Poblado cercano a la finca 4. Otros 1.8.6.Experiencia como productor 1. Ninguna 4. Entre 11 y 15 aos 2. Menos de 5 aos 5. Entre 16 y 20 aos 3. Entre 5 y 10 aos 6. Ms de 20 aos 2. Identificacin de la finca. Nombre actual: ___________________________________________ Nombre anterior: __________________________________________ Personalidad: ____________________________________________ Razn social: ____________________________________________ 2.1.Ubicacin de la finca. Estado: _____________ Municipio: _________________ Sector: ________________ Zona:_____________________ 2.2.Tiempo de fundada la finca. 1. Entre 1 y 5 aos 4. Entre 16 y 20 aos 2. Entre 6 y 10 aos 5. Ms de 20 aos 3. Entre 10 y 15 aos 6. Otro

PAGE 166

151 2.3.Actividad Principal. 2.3.1.Ganadera Bovina. 1. Leche 2. Carne 3. Doble propsito 2.3.2.Mixta: 2.4.Sistema de Produccin predominante. 1. Vaca becerro 4. Recra 2. Vaca maute 5. Ceba 3. Vaca novillo 6. Otro 2.5.Topografa % % 1. Plana 3. Quebrada 2. Ondulada 4. Mixta 2.6.Drenaje % 1. Aparentemente bueno 2. Aparentemente deficiente 3. Mixto 2.7.Disponibilidad de agua 1. Lluvia 4. Acueducto 2. Ro, cao, arroyo, canal, etc. 5. Otro 3. Pozo 2.8.Vas de acceso. 1. Terrestre 1. Buena 2. Regular 3. Mala 4. No hay

PAGE 167

152 2. Areo ( Condiciones de la pista de Aterrizaje ) 1. Buena 2. Regular 3. Mala 4. No hay 3. Acceso Fluvial 1. Buena 2. Regular 3. Mala 4. No hay 2.10.Vas internas 1. Buena 2. Regular 3. Mala 4. No hay II.FACTORES DE PRODUCCI”N: 1.Tierras 1.1.Tenencia % Has. 1. Baldas Extensin 2. Propias Extensin 3. Ejidas Extensin 4. Arrendadas Extensin 5. Otras 1.2.Superficie utilizada: _____________________________ Has. 1.2.1.Superficie utilizada en Ganadera: ________ Has. 1.2.1.1.Pastos y Forrajes: TIPOS EXTENSI”N (Has.) TIPO EXTENSI”N (Has.) R S R S 1. GUINEA 5. SURVENOLA 2. BRACHIARIAS 6. OTRAS GRAMINEAS 7. PASTOS DE CORTE 3. ALEMN 8. LEGUMINOSAS 4. ESTRELLA

PAGE 168

153 1.2.1.2.Otros Cultivos Dedi cados a la alimentacin del Ganado: 1. Nombre Comn 2. Extensin (Has.) R S 1.2.2.Cultivos Comerciales: 1. Nombre Comn 2. Extensin (Has.) R S 1.3.2.Superficie en Asientos y Caminos:_______ Has. 1.3.Superficie no utilizada: TIPO EXTENSI”N ( Has.) Bosque Deforestada Ociosa Intil Otros 2.CONSTRUCCIONES, EDIFICACIONES E INSTALACIONES. 2.1.Vivienda: TIPO DIMENSIONES LxA EDAD 2.2.Vaqueras: TIPO DIMENSIONES LxA EDAD

PAGE 169

154 2.3.Cuarto de Leche: TIPO DIMENSIONES LxA EDAD 2.4.Galpones: TIPO DIMENSIONES LxA EDAD 2.5.Depsitos: TIPO DIMENSIONES LxA EDAD 2.6.Corrales: TIPO CARACTERSTICASDIMENSIONES L EDAD 2.7.Cercas: TIPO CARACTERSTICASDIMENSIONES L EDAD 2.8.Comederos TIPO CARACTERSTICASDIMENSIONES LxA h e EDAD

PAGE 170

155 2.9.Bebederos: TIPO CARACTERSTICASDIMENSIONES LxA h e EDAD 2.10.Saleros: TIPO CARACTERSTICASDIMENSIONES LxA h e EDAD 2.11.Tanques: TIPO CARACTERSTICAS Y/O CAPACIDAD DIMENSIONES LxA h e EDAD 2.12.Acometida Elctricas: TIPO DIMENSIONES LARGO EDAD 2.13.Caminos y Carreteras: TIPO DIMENSIONES LARGO X ANCHO X ESPESOR CONDICIONES EDAD 2.14.Jageyes, Lagunas y Represas: TIPO CARACTERSTICAS DIMENSIONES L A h EDAD

PAGE 171

156 2.15.Sistemas de Riego y Drenaje: TIPO CARACTERSTICAS DIMENSIONES L A h e EDAD 2.16.Pozos: TIPO CARACTERSTICAS DIMENSIONES P EDAD 2.17.Silos: TIPO CARACTERSTICAS DIMENSIONES LxA h e EDAD 2.18.Otras Instalaciones: TIPO CARACTERSTICAS DIMENSIONES EDAD 3.EQUIPOS, MAQUINARIAS, IMPLEMENTOS Y VEHCULOS. 3.1.Tanques de Enfriamiento: TIPO CARACTERSTICAS DIMENSIONES EDAD 3.2.Sistema de Ordeo Mecnico: TIPO CARACTERSTICAS DIMENSIONES EDAD

PAGE 172

157 3.3.Equipo para la Elaboracin de Queso: TIPO CARACTERSTICAS DIMENSIONES EDAD 3.4.Equipos de Riego: TIPO CARACTERSTICAS DIMENSIONES EDAD 3.5.Plantas Elctricas: TIPO MARCA MODELO CAPACIDAD EDAD 3.6.Bombas: TIPO MARCA MODELO CAPACIDADPOTENCIA EDAD 3.7.Motores: TIPO MARCA MODELO CAPACIDAD EDAD 3.8.Molinos: TIPO MARCA MODELO CAPACIDAD EDAD 3.9.Equipos de Taller: TIPO MARCA MODELO EDAD

PAGE 173

158 3.10.Tractores: TIPO MARCA MODELO CAPACIDAD EDAD 3.11.Implementos: TIPO MARCA MODELO CAPACIDAD EDAD 3.12.Vehculos: TIPO MARCA MODELO CAPACIDAD EDAD 3.13.Otros: TIPO MARCA MODELO CAPACIDAD EDAD 3.14.Inventario de Semovientes: Raza: ________________ GANADO BOVINO CABEZAS PRODUCCI”N LECHE ( LTS.) PRECIO PROMEDIO Kg. TOROS PUROS TOROS MESTIZO TOROS CRIOLLOS VACAS EN ORDEO VACAS SECAS NOVILLAS MAUTAS MAUTES BECERRAS BECERROS NOVILLOS TORETES

PAGE 174

159 OTROS SEMOVIENTES CABEZAS PRODUCCI”N DE LECHE PRECIO PROMEDIO OVINOS CAPRINOS PORCINOS OTROS 4.GASTOS DE EXPLOTACI”N 4.1.Semillas para Recuperar Potreros: NOMBRE DEL PRODUCTO MARCA UNIDAD DE MEDIDA CANTIDAD SUPERFICIE 4.2.Abonos y Enmiendas: NOMBRE DEL PRODUCTO MARCA UNIDAD DE MEDIDA CANTIDAD SUPERFICIE 4.3.Plaguicidas y Herbicidas: NOMBRE DEL PRODUCTO MARCA UNIDAD DE MEDIDA CANTIDAD 4.4.Alimentos y Suplementos: TIPO MARCA UNIDAD DE MEDIDA CANTIDAD

PAGE 175

160 4.5.Medicina Veterinaria: TIPO NOMBRE DEL PRODUCTO MARCA UNIDAD DE MEDIDA PRESENTACI”N CANTIDAD 4.6.Combustible y Lubricantes: TIPO NOMBRE DEL PRODUCTO MARCA UNIDAD DE MEDIDA CANTIDAD 4.7.Servicio de Maquinara para Mantenimiento de Potreros, Sistema de Riego, etc. ACTIVIDAD TIPO UNIDAD DE MEDIDA CANTIDAD PRECIO POR UNIDAD MONTO Bs. HRS. HAS. 4.8. Reparaciones y Repuestos para Mantenimiento de Maquinaria y Equipos: TIPO CANTIDAD MONTO TOTAL (Bs.) 4.9. Mantenimiento de Construcciones, Edificaciones e Instalaciones: TIPO CANTIDAD MONTO TOTAL (Bs.)

PAGE 176

161 4.10.Impuestos, Patentes y Seguros: TIPO MONTO (Bs.) 4.11.Servicios Bsicos: TIPO MONTO (Bs.) 4.12.Gastos Generales: TIPO MONTO (Bs.) 4.13.Otros Servicios: TIPO MONTO (Bs.) 4.14.Recursos Humanos: TIPO No. RELAC. DE FORMA DE CANTIDAD MONTO Bs. PRESTAC. PRIMAS Y BENEF. TRAB. PAGO SI NO MONT. MOTIV. 4.15.Otros Gastos: TIPO MONTO (Bs.) 5. Ingresos TIPO UNIDAD CANTIDAD MONTO Bs. PRECIO Bs. ASPECTOS FINANCIEROS 1. Ha solicitado crdito durante los ltimos diez aos? a)Si b)No

PAGE 177

162 2. Se le otorgo el crdito? a)Si b)No 3. En caso de habrsele negado la solicit ud cuales razones alegaron para rechazarlo? 4. En caso de habrsele otorgado el cr dito establezca las condiciones del mismo? AO INSTITUCI”N MONTO SOLICITADO (Bs.) MONTO CONCEDIDO (Bs.) PLAZO (AOS) TASA DE INTERS USO DEL CRDIT O 5. Considera positivas las condiciones del crdito a)Si b)No Explique: 6. Solicito usted refinanciamiento de algn prstamo bancario? a)Si b)No 7. Ante cual Banco hizo la solicitud 8. Le fue aprobado el refinanciamiento? a)Si b)No 9. Si no le aprobaron el refinanc iamiento que razones alegaron para rechazarlo? 10. Si le aprobaron cuanto fue el monto del refinanciamiento?

PAGE 178

163 11. Cul es el monto aproximado y el plazo de las deudas que su finca tiene actualmente por los siguientes conceptos? DESCRIPCI”N MONTO PLAZO INTERS VEHCULOS MAQUINARIAS Y EQUIPOS MATERIALES DE CONSTRUCCI”N COMPRA DE GANADO COMPRA DE TIERRA ALIMENTACI”N DEL REBAO MED. Y SERV. VETERINARIOS FERTILIZANTES Y HERBICIDAS REPARACI”N Y REPUESTOS ALIMENTACI”N AL PERSONAL OTROS INSUMOS ASPECTOS TCNICOS 1. PASTOS Y FORRAJES 1. Nmero de Potreros: _______________________________________________ 2. Tamao de los Potreros:___________________________________________ 3. Pastoreo: a) Permanente b)Rotativo Verano Invierno 4. Perodo de ocupacin (Das) 5. Perodo de descanso (Das) 6. Clasificacin del Rebao para Pastoreo: a)Si b)No Grupos:_______________________________________________________ 7. Fertilizacin: Tipo Dosis 8. Control de Plagas:a)PLaguicidas b)Pastoreo Intens.

PAGE 179

164 9. Control de Malezasa)Manual b)Mecnico c)Qumico d)Todas 10. Conservacin: a)Heno b)Silo c)Otros: 11. Riego: a)Si b)No Mtodo de Riego; 2. ASPECTOS ZOOTCNICOS, R EPRODUCTIVOS Y SANITARIOS 2.1. Criterios de seleccin de Vacas y Novillas: 1. Produccin: 2. Condiciones Fsicas: 3. Fenotipo: 4. Inf. de los Padres 5. Otras: 2.2. Prcticas: 1. descorn 2. Castracin 3. Pesaje de leche 4. Pesaje de Anim. 5. Cura de ombligo 6. Clasif. del rebao 7. Ident. del rebao 8. Separac. por sexo 9. Peso al nacer 10. Destete Edad: Peso: 11. Venta de los machos; Edad: Peso: 12. Servicio de hembras Edad: Peso: 13. Primer parto: Edad: Peso: 14. Duracin de la lactancia 15. Ordeo Frecuencia Hora a) Manual b) Mecnico c) En la Vaquera d) En el patio e) Con el becerro f) Sin becerro

PAGE 180

165 16. Sitio de pariciones a) Patio b) Potrero especial c) Cualquier lugar 17. Perodo de descanso pre-parto (secado) ______________________________ 18. Perodo parto primer servicio ______________________________________ 19. Intervalo de parto. __________________________________________________ 20. Reproduccin. a) Monta libre b) Monta controlada c) Inseminacin artificial 21. Deteccin de mastitis: a) Clnica b) Sub-clnica 22. Causas ms frecuentes de muerte: a) En adultos: _____________________________________________________ b) En becerros: ____________________________________________________ 23. Plan sanitario: a)Si b)No 24. Control de hematozoarios: a)Si b)No 25. Programa de desparasitaciones. Frecuencia:________________________

PAGE 181

166 2.3. Transferencia de tecnologa: 1. Recibe asistencia tcnica a) Si b) No 2. De organismos pblicos a) Si b) No 3. De organismos privados a) Si b) No 4. Tipo de profesional que presta la asistencia tcnica 5. Frecuencia de visitas: ________________________________________________ 6. Medios que utiliza: ___________________________________________________ 7. Ha aplicado algunas tcnicas aprendidas por algunos de estos medios a) Si b) No c) Cuales ASPECTOS GERENCIALES. A. ORGANIZACI”N 1.De que manera se les informa a los obreros de sus tareas y responsabilidades dentro de la finca? 2.Saben sus obreros quin da las ordenes en todo momento? a)Si b)No 3.Como es la organizacin adminis trativa de la finca (organigrama administrativo)? 4.Como dividira usted el trabajo en la finca.(clasificacin del trabajo)? 5.Posee plano de la finca? a)Si b)No Qu uso le da?_______________________________________________

PAGE 182

167 B. CONTROL. 1.Cuanto le cuesta producir un litro de leche? No sabe: 2.Tipos de registros que se llevan en la finca. 2.1. Contables 1. Libro diario a)Si b) No 2. Libro mayor a)Si b) No 3. Caja chica a)Si b) No 4. Estado de Ganancia y Perdidas a)Si b) No 5. Balance general a)Si b) No 6. Libro de inventario a)Si b) No 2.2. Operacionales: 1. Inventario ganado a)Si b) No 2. Control de la mano de obra a)Si b) No 3. Pesaje de leche a)Si b) No 4. Historia del rebao a)Si b) No 5. Control de Insumos a)Si b) No 6. Control de vacunaciones a)Si b) No 3.Que utilidad le ofrecen los registros (para que los usa)? 4.Se hacen comparaciones de los resultados obtenidos en un ao con otro para saber si hubo perdidas y ganancia?

PAGE 183

168 5.Que tipo de correctivo se toma cuando no se logra los resultados esperados? 6. Cuanto tiempo tardan sus tr abajadores en ordear una vaca? 7. Como controla el trabajo de los operadores y maquinarias? C. PLANIFICACI”N 1.Cuales son los objetivos a corto, mediano y largo plazo con relacin a la finca? 2.Tiene un plan escrito para desarrollar la finca para el futuro de su finca? a)Si b)No 2.1. Por escrito: ______________________________________________ 3.Que tipo de instrumento utiliza para programar sus actividades ? a) Notas informes b) Presupuesto c) Proyectos d) Otros (especifique) e) Nada 4.Cree usted que puede mejorar los niveles de produccin de su finca? a)Si b)No 4.1. Como?_________________________________________________ 4.2. D. DIRECCI”N 1.Quin y cmo resuelve los conflictos laborales? 2.Acepta usted sugerencia del personal? 3.En este momento se podran clasificar a sus obreros en relacin a la labor que desempea e importancia de la misma. (persona clave).

PAGE 184

169 APPENDIX B TSP PROGRAM OPTIONS CRT LIMWARN=0; FREQ N; SMPL 1,127; ? LEOMDL05.TSP; ? COBB-DOUGLASS TYPE PRODUCTION FUNCTION (MLTRUNCATED); ?IN 'C:\ZZMODEL\LEODAT03'; IN 'C:\Leo\ZZMODELA\LEODAT03'; ? in 'd:\ZSTUDENT\ORTEGA\LEODAT03'; ? dblist 'd:\ZSTUDENT\ORTEGA\LEODAT03'; ? production function frontier model; ? dependent variable = INGTOTAL; ? independent variable CAPITOTA, EQHOMBRE GASERTEC; ? I2 = seed, I3= fertilizer, I4=Herbicide&insect; ? I5=Suppl feed, I6=Med vet , I7=Gas&lube; ? J2=Mach rent, J3=Parts&repairs, J4=Build maintenance; ? J5=Insurance&Taxes, J6=Utilities J7=others; ? CAPPASTO CAPTIERR CAPEDIFI; ? CAPMAQUI CAPGANAD CAPITOTA GAPERTEC GAPEROBR GASPERS; ? OTRGASPE GTPERSON EQHOMBRE EQHOMORD ORDENAD I2 I3 I4; ? I5 I6 I7 J2 J3; ? J4 J5 J6 J7 GASTINSU GASERTEC GASINSUM DEPRECIA GASTOPER GASTOST ; ? CANLECHE INGLECHE CANCARNE INGCARNE CANOTRO; ? INGOTRO INGTOTAL; Y = INGTOTAL; CT = CAPITOTA; E1 = EQHOMBRE; GT = GASERTEC; CP = CAPPASTO; CL = CAPTIERR; CE = CAPEDIFI; CM = CAPMAQUI; CC = CAPGANAD; E2 = EQHOMORD; M1 = ORDENAD; DM=DEPRMAQE; DB=DEPRINST; SUF=SUFORRAJ; TAU=TOTALUAB; TCOW=CAVACASE+CAVACAOR; LEXP=GTPERSON; USEREC=MB3; SELECT CP>0; ?=================================================================; ? CHANGE VARIABLES FROM ZERO TO 1; ?=================================================================;

PAGE 185

170 DOT I2 I3 I4 I5 I6 I7 J2 J3 J4 J5 J6 J7 DM DB; .=(.=0)*0.01+(.>0)*.; ENDDOT; ?PRINT I2; ?==================================================================; ? CALCULATING THE LOG OF THE VARIABLES; ?==================================================================; DOT Y CT E1 GT CP CL CE CM CC E2 M1 I2 I3 I4 I5 I6 I7 J2 J3 J4 J5 J6 J7 ; L.=LOG(.); enddot; ?==================================================================; ? PRINCIPAL COMPONENT APPROACH TO AVOID MULTICOLINEARITY; ?==================================================================; PRIN(NAME=PZ) LE1 LCP LCL LCE LCM LCC LJ3 LI6; ?K=@EIGVAL(1)/@EIGVAL(8); ?PRINT K; DOT LE1 LCP LCL LCE LCM LCC LJ3 LI6; MSD(NOPRINT) .; N.= (. @MEAN)/@STDDEV; SET MN.=@MEAN; SET STD.=@STDDEV; SET V.=STD.**2; ENDDOT; MSD(COVA,NOPRINT)LE1 LCP LCL LCE LCM LCC LJ3 LI6; MAT VAR=@COVA; ?PRINT VAR; ?PRINT STDLE1 VLE1 VAR(1,1); ? STEP #1; ?=============================================================; ? CREATING THE LOADING FACTORS; ?=============================================================; CORR PZ1-PZ8 NLE1 NLCP NLCL NLCE NLCM NLCC NLJ3 NLI6; MAT NLOAD=@CORR; OLSQ(ROBUSTSE) LY C PZ1 PZ2 PZ3 PZ4 LI2 LI3 LI4 LI5 LI7 LJ2 LJ4 LJ5 LJ6 LJ7; RES1=@RES; ? LOADING FACTORS ARE CORRECTED FOR THE EIGENVALUE VALUES TO ? NORMALIZE THE W'W; DOT(VALUE=J) 1 2 3 4; SET CROOT.=NLOAD(9,J)**2 + NLOAD(10,J)**2 + NLOAD(11,J)**2 + NLOAD(12,J)**2 + NLOAD(13,J)**2 + NLOAD(14,J)**2 + NLOAD(15,J)**2 + NLOAD(16,J)**2; ENDDOT; ?PRINT CROOT1-CROOT4; MAT LOAD=NLOAD; ? PRINCIPAL COMPONENTS ARE NORMALIZED BY THE CHARACTERISTIC ROOTS; DOT(VALUE=J) 1 2 3 4; DO K=9 TO 16 BY 1;

PAGE 186

171 SET LOAD(K,J)=NLOAD(K,J)/CROOT.; SET IN=NLOAD(K,J)/LOAD(K,J); ?PRINT K J IN LOAD(K,J) NLOAD(K,J); ENDDO; ENDDOT; ? STEP #12; ?==========================================; ?STOCHASTIC PRODUCTION FRONTIER; ?==========================================; ?STEP #12-2; ?====================================================================; ? MAXIMUM LIKELIHOOD ESTIMATES OF ALL PARAMETERS; ?SEE TSP MANUAL PAG 77; STOCH FRONT ANALYSIS BOOK PAG 75-77; ?AND MADDALA (LIMITED DEPEND BOOK) PAG 195; ?====================================================================; LY=LOG(Y); MFORM(TYPE=GEN,NROW=1,NCOL=5) ZMLZ=0; ? STEP#ML1; ?====================================================================== ?CREATING THE VARIABLES CORRECTION FOR HETEROCEDASTICITY; ?====================================================================== OLSQ(SILENT) LY C PZ1 PZ2 PZ3 PZ4 LI2 LI3 LI4 LI5 LI7 LJ2 LJ4 LJ5 LJ6 LJ7; ERRSQ=@RES**2; ? SQUARED RESIDUALS; FLY=@FIT; SQFLY=FLY**2; OLSQ ERRSQ C FLY SQFLY; ? SHORTER FORM OF THE WHITE"S TEST; WTERR=SQRT(@FIT); ? WEIGHT FOR CORRECTING THE HETEROCEDASTICITY; ?STEP #ML2; ?====================================================================== ? CREATING THE WEIGHTED VARIABLES; ?====================================================================== ?II=1; ? SET WT=0; ? NO CORRECTION FOR HETEROCEDASTICITY; ? SET WT=1; ? WITH CORRECTION FOR HETEROCEDASTICITY; ?DOT LY II PZ1 PZ2 PZ3 PZ4 LI2 LI3 LI4 LI5 LI7 LJ2 LJ4 LJ5 LJ6 LJ7; ?W.=./ ( ( WT=0) + (WT=1)*WTERR); ? ENDDOT; ? STEP #ML3; ?====================================================================== ? PRODUCTION FRONTIER FUNCTION; ?====================================================================== FRML E (LY-BB0-BB1*PZ1-BB2*PZ2-BB3*PZ3-BB4*PZ4-BB5*LI2-BB6*LI3 -BB7*LI4-BB8*LI5-BB9*LI7-BB10*LJ2-BB11*LJ4-BB12*LJ5-BB13*LJ6BB14*LJ7); PARAM BB0-BB14;

PAGE 187

172 ?====================================================================== ? HALF NORMAL DISTRIBUTION; ?====================================================================== FRML EZ1 (E/SIGI); FRML SSS (-E*LAMBDA/SIGI); FRML PFUNCT LOGL=LOG(2)-LOG(SIGI)+LNORM(EZ1) +LCNORM(SSS); PARAM BB0-BB14 LAMBDA SIGI; EQSUB PFUNCT EZ1 SSS E; ?------------------------------------------------------------------?====================================================================== ? EXPONENTIAL DISTRIBUTION; ?====================================================================== ?THETA=1/SIGU; ?FRML PFUNCT LOGL=LOG(THETA)+ LCNORM( -((E/SIGV)+(THETA*SIGV)) ) ? + (E*THETA) + ( (1/2)*(THETA**2)*(SIGV**2) ); ?PARAM THETA SIGV; ?EQSUB PFUNCT E; ?--------------------------------------------------------------------OLSQ(SILENT)LY C PZ1 PZ2 PZ3 PZ4 LI2 LI3 LI4 LI5 LI7 LJ2 LJ4 LJ5 LJ6 LJ7; UNMAKE @COEF BB0-BB14; SET SIGI=@S; SET SIGU=0.1; SET SIGV=0.25; SET THETA=12; SET MSIGU=0.1; SET LAMBDA=0.1; SET M=0.1; SET T=0.1; ML (HCOV=N, HITER=N) PFUNCT; ?ML (HCOV=B, HITER=B, MAXIT=1500, TOL=1E-9, TOLS=0.01, SQZTOL=0.1) PFUNCT; ?ML PFUNCT; SET i=1; SET ZMLZ(I,1)=J; SET ZMLZ(I,2)=@LOGL; SET ZMLZ(I,3)=@IFCONV; SET ZMLZ(I,4)=@COEF(13); SET ZMLZ(I,5)=@COEF(14); PRINT ZMLZ; ?---------------------------------------------------------------; ? Writing the ML results in excel; ?---------------------------------------------------------------; COPY @COEF COEF; COPY @SES STDDEV; COPY @T TTEST; COPY %T PVALUE; MMAKE M COEF STDDEV TTEST PVALUE; WRITE(FORMAT=EXCEL,FILE=HALFN1.XLS)M; ?---------------------------------------------------------------;

PAGE 188

173 ? STEP #12-3; ?====================================================================; ? CALCULATING THE COEFFICIENTS FOR THE VARIABLES IN THE PRINCIPAL ? COMPONENT USING ML COEFFICIENT; ?====================================================================; SET MSB01=LOAD (9,1)*@COEF(2) + LOAD (9,2)*@COEF(3) + LOAD (9,3)*@COEF(4) + LOAD (9,4)*@COEF(5); SET MSB02=LOAD(10,1)*@COEF(2) + LOAD(10,2)*@COEF(3) + LOAD(10,3)*@COEF(4) + LOAD(10,4)*@COEF(5); SET MSB03=LOAD(11,1)*@COEF(2) + LOAD(11,2)*@COEF(3) + LOAD(11,3)*@COEF(4) + LOAD(11,4)*@COEF(5); SET MSB04=LOAD(12,1)*@COEF(2) + LOAD(12,2)*@COEF(3) + LOAD(12,3)*@COEF(4) + LOAD(12,4)*@COEF(5); SET MSB05=LOAD(13,1)*@COEF(2) + LOAD(13,2)*@COEF(3) + LOAD(13,3)*@COEF(4) + LOAD(13,4)*@COEF(5); SET MSB06=LOAD(14,1)*@COEF(2) + LOAD(14,2)*@COEF(3) + LOAD(14,3)*@COEF(4) + LOAD(14,4)*@COEF(5); SET MSB07=LOAD(15,1)*@COEF(2) + LOAD(15,2)*@COEF(3) + LOAD(15,3)*@COEF(4) + LOAD(15,4)*@COEF(5); SET MSB08=LOAD(16,1)*@COEF(2) + LOAD(16,2)*@COEF(3) + LOAD(16,3)*@COEF(4) + LOAD(16,4)*@COEF(5); SET MSB1=LOAD (9,1)*@COEF(2) + LOAD (9,2)*@COEF(3) + LOAD (9,3)*@COEF(4) + LOAD (9,4)*@COEF(5); SET MSB2=LOAD(10,1)*@COEF(2) + LOAD(10,2)*@COEF(3) + LOAD(10,3)*@COEF(4) + LOAD(10,4)*@COEF(5); SET MSB3=LOAD(11,1)*@COEF(2) + LOAD(11,2)*@COEF(3) + LOAD(11,3)*@COEF(4) + LOAD(11,4)*@COEF(5); SET MSB4=LOAD(12,1)*@COEF(2) + LOAD(12,2)*@COEF(3) + LOAD(12,3)*@COEF(4) + LOAD(12,4)*@COEF(5); SET MSB5=LOAD(13,1)*@COEF(2) + LOAD(13,2)*@COEF(3) + LOAD(13,3)*@COEF(4) +

PAGE 189

174 LOAD(13,4)*@COEF(5); SET MSB6=LOAD(14,1)*@COEF(2) + LOAD(14,2)*@COEF(3) + LOAD(14,3)*@COEF(4) + LOAD(14,4)*@COEF(5); SET MSB7=LOAD(15,1)*@COEF(2) + LOAD(15,2)*@COEF(3) + LOAD(15,3)*@COEF(4) + LOAD(15,4)*@COEF(5); SET MSB8=LOAD(16,1)*@COEF(2) + LOAD(16,2)*@COEF(3) + LOAD(16,3)*@COEF(4) + LOAD(16,4)*@COEF(5); PRINT MSB1 MSB2 MSB3 MSB4 MSB5 MSB6 MSB7 MSB8; ?STEP #12-4; ?==================================================================; ? DISNORMALIZING THE COEFFICIENTS FOR THE VARIABLES IN THE PRINCIPAL ? COMPONENT; ?===================================================================; SET MB1=MSB1/STDLE1; SET MB2=MSB2/STDLCP; SET MB3=MSB3/STDLCL; SET MB4=MSB4/STDLCE; SET MB5=MSB5/STDLCM; SET MB6=MSB6/STDLCC; SET MB7=MSB7/STDLJ3; SET MB8=MSB8/STDLI6; SET MB01=MSB01*MNLE1/STDLE1; SET MB02=MSB02*MNLCP/STDLCP; SET MB03=MSB03*MNLCL/STDLCL; SET MB04=MSB04*MNLCE/STDLCE; SET MB05=MSB05*MNLCM/STDLCM; SET MB06=MSB06*MNLCC/STDLCC; SET MB07=MSB07*MNLJ3/STDLJ3; SET MB08=MSB08*MNLI6/STDLI6; SET MB00=MB01+MB02+MB03+MB04+MB05+MB06+MB07+MB08; PRINT MB00 ; ? STEP #12-5; ?================================================================; ? CALCULATING THE INTERCEPT FROM THE PRINCIPAL COMPONENT; ?================================================================; SET MB0=@COEF(1)-MB00; PRINT MB0 MB1-MB8; ?STEP #12-6 ?====================================================================; ? CALCULATING THE PREDICTED LN PRODUCTION FUNCTION (HAT) MHLY; ?====================================================================; MHLY=MB0+MB1*LE1+MB2*LCP+MB3*LCL+MB4*LCE+MB5*LCM+MB6*LCC+MB7* LJ3+ MB8*LI6+@COEF(6)*LI2+@COEF(7)*LI3+@COEF(8)*LI4+@COEF(9)*LI5+ @COEF(10)*LI7+@COEF(11)*LJ2+@COEF(12)*LJ4+@COEF(13)*LJ5

PAGE 190

175 +@COEF(14)*LJ6+@COEF(15)*LJ7; ?PRINT MHLY; ? STEP #12-7; ?=====================================================================; ? CALCULATING THE PREDICTED PRODUCTION FUNCTION (HAT) MHY, AND ? RESIDUAL(MHE); ?=====================================================================; ?EE=LY-BB0-BB1*PZ1-BB2*PZ2-BB3*PZ3-BB4*PZ4-BB5*LI2-BB6*LI3 ? -BB7*LI4-BB8*LI5-BB9*LI7-BB10*LJ2-BB11*LJ4-BB12*LJ5-BB13*LJ6BB14*LJ7; E=LY-MHLY; MSD E ; PRINT E ; ?STEP #12-8; ?====================================================================; ? ESTIMATES OF EXPECTED U/E(HUU); ? STOCH FRONT ANALYSIS BOOK PAG 77-80; ? GREENE A GAMMA-DIST,1990,PAG143; ?====================================================================; ?HALF NORMAL DISTRIBUTION. ?U=(Sigma*)*((Norm(E*Lambda/Sig)/(1-Cnorm(E*Lambda/Sig)))((E*Lambda/Sig))); ?--------------------------------------------------------------------; SET SIG=SIGI; SET LAMBDA=@COEF(16); PRINT SIG LAMBDA; PRT1=SIG*LAMBDA/(1+(LAMBDA**2)); PRT2=norm(E*LAMBDA/SIG); PRT3=1-(cnorm(E*LAMBDA/SIG)); PRT4=E*LAMBDA /SIG; U=PRT1*((PRT2/PRT3)-PRT4); PRINT U; ?---------------------------------------------; ?EXPONENTIAL DISTRIBUTION; ?THETA=@COEF(16); ?SIGV=@COEF(17); ?A=-((-E/SIGV)-(SIGV*THETA) ); ?UEXP=SIGV*((NORM(A)/CNORM(-A))-A); ?PRINT UEXP; ?STEP #12-9; ?==================================================================; ? ESTIMATES OF THE TECHNICAL EFFICIENCY OF EACH PRODUCER(TE); ?===================================================================; TE=EXP(-U); ?TE=EXP(-UEXP); PRINT TE ; ?SORT TE IDD0 SUG; MSD TE; PRINT E U TE IDD0 SUG; MMAKE THALF E U TE IDD0; write(format=excel, file=HALFN2.xls)THALF;

PAGE 191

176 ?OLSQ TE C SUG; ? STEP #12-10; ?=============================================================; ? CALCULATING THE VARIANCE OF THE COEFFICIENT; ?=============================================================; SET VB1=(LOAD(9,1)**2)*@VCOV(2,2)/VLE1 +(LOAD(9,2)**2)*@VCOV(3,3)/VLE1 +(LOAD(9,3)**2)*@VCOV(4,4)/VLE1 + (LOAD(9,4)**2)*@VCOV(5,5)/VLE1; SET VB2=(LOAD(10,1)**2)*@VCOV(2,2)/VLCP+(LOAD(10,2)**2)*@VCOV(3,3)/VLCP +(LOAD(10,3)**2)*@VCOV(4,4)/VLCP+(LOAD(10,4)**2)*@VCOV(5,5)/VLCP; SET VB3=(LOAD(11,1)**2)*@VCOV(2,2)/VLCL+(LOAD(11,2)**2)*@VCOV(3,3)/VLCL +(LOAD(11,3)**2)*@VCOV(4,4)/VLCL+(LOAD(11,4)**2)*@VCOV(5,5)/VLCL; SET VB4=(LOAD(12,1)**2)*@VCOV(2,2)/VLCE+(LOAD(12,2)**2)*@VCOV(3,3)/VLCE +(LOAD(12,3)**2)*@VCOV(4,4)/VLCE+(LOAD(12,4)**2)*@VCOV(5,5)/VLCE; SET VB5=(LOAD(13,1)**2)*@VCOV(2,2)/VLCM+(LOAD(13,2)**2)*@VCOV(3,3)/VLCM +(LOAD(13,3)**2)*@VCOV(4,4)/VLCM+(LOAD(13,4)**2)*@VCOV(5,5)/VLCM; SET VB6=(LOAD(14,1)**2)*@VCOV(2,2)/VLCC+(LOAD(14,2)**2)*@VCOV(3,3)/VLCC +(LOAD(14,3)**2)*@VCOV(4,4)/VLCC+(LOAD(14,4)**2)*@VCOV(5,5)/VLCC; SET VB7=(LOAD(15,1)**2)*@VCOV(2,2)/VLJ3+(LOAD(15,2)**2)*@VCOV(3,3)/VLJ3 +(LOAD(15,3)**2)*@VCOV(4,4)/VLJ3+(LOAD(15,4)**2)*@VCOV(5,5)/VLJ3; SET VB8=(LOAD(16,1)**2)*@VCOV(2,2)/VLI6+(LOAD(16,2)**2)*@VCOV(3,3)/VLI6 +(LOAD(16,3)**2)*@VCOV(4,4)/VLI6+(LOAD(16,4)**2)*@VCOV(5,5)/VLI6; ?PRINT @VCOV(2,2) VLE1 VB1-VB8; ? STEP #12-11; ?=============================================================; ? TESTING HYPOTHESES THAT THE COEFFICIENTS ARE DIFFERENT FROM ZERO; ?=============================================================; SET TTEST1=MB1/SQRT(VB1); SET TTEST2=MB2/SQRT(VB2); SET TTEST3=MB3/SQRT(VB3); SET TTEST4=MB4/SQRT(VB4); SET TTEST5=MB5/SQRT(VB5); SET TTEST6=MB6/SQRT(VB6); SET TTEST7=MB7/SQRT(VB7); SET TTEST8=MB8/SQRT(VB8); SET DF1=123-19; DOT TTEST1 TTEST2 TTEST3 TTEST4 TTEST5 TTEST6 TTEST7 TTEST8; CDF(T,DF=DF1).; ENDDOT; PRINT MB1 MB2 MB3 MB4 MB5 MB6 MB7 MB8; ? STEP #12-12; ?====================================================================== ? STANDARIZED COEFFICIENTS; ?(It allows to compare the coefficient in the same scale) (Pindyck book pag 98); ?====================================================================== DOT LY LI2 LI3 LI4 LI5 LI7 LJ2 LJ4 LJ5 LJ6 LJ7; MSD(NOPRINT).; SET STD.=@STDDEV; ENDDOT; SET STB1=MB1*STDLE1/STDLY; SET STB2=MB2*STDLCP/STDLY; SET STB3=MB3*STDLCL/STDLY;

PAGE 192

177 SET STB4=MB4*STDLCE/STDLY; SET STB5=MB5*STDLCM/STDLY; SET STB6=MB6*STDLCC/STDLY; SET STB7=MB7*STDLJ3/STDLY; SET STB8=MB8*STDLI6/STDLY; SET STB9=@COEF(6)*STDLI2/STDLY; SET STB10=@COEF(7)*STDLI3/STDLY; SET STB11=@COEF(8)*STDLI4/STDLY; SET STB12=@COEF(9)*STDLI5/STDLY; SET STB13=@COEF(10)*STDLI7/STDLY; SET STB14=@COEF(11)*STDLJ2/STDLY; SET STB15=@COEF(12)*STDLJ4/STDLY; SET STB16=@COEF(13)*STDLJ5/STDLY; SET STB17=@COEF(14)*STDLJ6/STDLY; SET STB18=@COEF(15)*STDLJ7/STDLY; PRINT STB1 MB1 STB2 MB2 STB3 MB3 STB4 MB4 STB5 MB5 STB6 MB6 STB7 MB7 STB8 MB8 STB9 @COEF(6) STB10 @COEF(7) STB11 @COEF(8) STB12 @COEF(9) STB13 @COEF(10) STB14 @COEF(11) STB15 @COEF(12) STB16 @COEF(13) STB17 @COEF(14) STB18 @COEF(15); ?STEP #12-13; ?==============================================; ? DETERMINANTS OF TECHNICAL EFFICIENCY; ?==============================================; SMPL 1,127; ?==============================================; ? Human Capital variables; ? PEDU=producer education; PRES= producer residence; ? PEXP=producer experience; ?==============================================; PAGE=I13; DPAGE=PAGE>2; PEDU=I14; DPEDU=(I14>2); ?PRINT DPEDU I14; DPEDU1=(I14<=2); DPEDU2=(I14>= 3 & I14<=4); DPEDU3=(I14>=5); ?PRINT I14 DPEDU DPEDU1 DPEDU2 DPEDU3; PRES=I15; DPRES=I15<3; ?PRINT DPEDU I14 DPRES I15; PPER=I1O6; DPPER=I1O6<5; PRINT DPPER I1O6; PEXP=I1O7; DPEXP=I1O7>2; DPEXP1=(I1O7<=3); DPEXP2=(I1O7>=4 & I1O7<=5); DPEXP3=(I1O7>=6);

PAGE 193

178 DMAGE=(I18_2>2); DMEDU=(I18_4>4); DMEXP=(XX1>2); DMRES=(I18_5<3); ?WHO IS THE MANAGER; MANAG=(I18_1=1)*1 + (I18_1=2)*0; DMANAG=MANAG>0; ?PRINT I18_1 DMANAG; ?================================================; ? MANAGERIAL VARIABLES; ?================================================; ?FARMER OBJECTIVE(FOBJ); FOBJ=MC1>0; ?TECHNICAL ASISTANCE(TASIST); TASIST=T23_1; DTASIST=T23_1>0; ? PRINT DTASIST; ? FREQUENCY OF TECHNICAL ASISTANCE(TECHA); TECHA=T23_5; DUMMY TECHA; ?PRINT TECHA1 TECHA2 TECHA3 TECHA4 TECHA5 ?TECHA6 TECHA7 TECHA8 TECHA9 TECHA10; DDTECHN=(TECHA=12 | TECHA=3 | TECHA=0) + (TECHA=4 | TECHA=5 | TECHA=8)*2 + (TECHA=1 | TECHA=2 | TECHA=6 | TECHA=7)*3 ; ?DUMMY DDTECHN; DTECHN=DDTECHN>2; ? TECHNICAL SERVICE; TECSER=GASERTEC; DTECSER=TECSER>0; ?PRINT TECSER DTECSER; ? KEEPING ACCOUNT(RECACC) AND TECHNICAL(RECTEC)REDCORDS; RECACC=MB21_1+MB21_2+MB21_3+MB21_4+MB21_5+MB21_6; RECTEC=MB22_1+MB22_2+MB22_3+MB22_4+MB22_5+MB22_6; ?PRINT RECACC RECTEC; ?HIST(DISCRETE)RECACC RECTEC; DRECACC=RECACC>0; DRECTEC=RECTEC>2; ?PRINT DRECTEC RECTEC; ? CREDIT(CRED); CRED=C02; DUMMY CRED; ?===============================================; ? Farm Characteristic; ? SUG=livestock used area; Z1=South, Z2=Eastern; ? Z3=Western, Z4=Northwest; CRED=credit; ?===============================================;

PAGE 194

179 ?LAND VARIABLES; ?DEVELOPMENT INDEX (DI); ?-----------------------------------------------; DI=SUG/SUT; SUGSQ=SUG**2; SQSUG=SQRT(SUG); RPSUG=1/SUG; DSUG= (SUG<300)+ [(SUG>=300)&(SUG<=400)]*2 + [(SUG>400)&(SUG<=575)]*3 + (SUG>575)*4; HIST (DISCRETE)DSUG; DUMMY DSUG; ?PRINT DSUG; ? ZONE(Z); Z=G21_4; DUMMY Z; ? TENURE; TENURE=[(F11=1)|(F11=4)|(F11=5)]*0 +[(F11=2)|(F11=3)]*1; DTEN=TENURE>0; ?PRINT F11 TENURE DTEN; ? LT. MILK PER HA.(LTLECSUF); ?-----------------------------------------------; ? CATTLE VARIABLES; ?-----------------------------------------------; ? TOTAL COW(TCOW); TCOW=UAVACAO+UAVACAS; TCSQ=TCOW**2; ? MILK PRODUCTION PER MILKING COW (LTVACORD); VACORDSQ=LTVACORD**2; VACTOTSQ=LTVACTOT**2; HIST LTVACORD; ?LTCOW=(LTVACORD<=1500) + ? [(LTVACORD>1500) & (LTVACORD<=2500)]*2 + ? [(LTVACORD>2500) & (LTVACORD<=3000)]*3 + ? [(LTVACORD>3000) & (LTVACORD<=3500)]*4 + ? (LTVACORD>3500)*5; ?LTCOW=(LTVACORD<=1500) + ? [(LTVACORD>1500) & (LTVACORD<=2000)]*2 + ? [(LTVACORD>2000) & (LTVACORD<=2500)]*3 + ? [(LTVACORD>2500) & (LTVACORD<=3000)]*4 + ? [(LTVACORD>3000) & (LTVACORD<=3500)]*5 + ? (LTVACORD>3500)*6; LTCOW=(LTVACTOT<=1000) + [(LTVACTOT>1000) & (LTVACTOT<=1500)]*2 + [(LTVACTOT>1500) & (LTVACTOT<=2000)]*3 + [(LTVACTOT>2000) & (LTVACTOT<=2500)]*4 + (LTVACTOT>2500)*5;

PAGE 195

180 HIST(DISCRETE)LTCOW; DUMMY LTCOW; HIST (DISCRETE) LTCOW; ? TOTAL COW(TOTCOW); TOTCOW=CAVACAOR +CAVACASE; ? RATIO MILKING COW-TOTAL COW (RMCOWTCOW); MCOWTCOW=CAVACAOR/TOTCOW; ? PRODUCTION SYSTEM (COW-CALF, COW-YEARLING, AND COW-STEER); RINGLINGT=INGLECHE/INGTOTAL; HIST RINGLINGT; PSYST=(RINGLINGT>0.8)*1 + [(RINGLINGT>0.7)& (RINGLINGT<=0.8)]*2 + (RINGLINGT<=0.7)*3; DPSYST=(RINGLINGT<=0.7); ?HIST(DISCRETE)PSYST; DUMMY PSYST; ? BEEF PRODUCTION; KGCARSQ=KGCARVAC**2; ? WEIGHT OF MALE SALE; WSALE=T22_112; WWSALE=(T22_112<=150) + [(T22_112>150)&(T22_112<=300)]*2 + (T22_112>300)*3; HIST(DISCRETE)WWSALE; DUMMY WWSALE; ? MILK PRICE(MPRICE); MPRICE=INGLECHE/CANLECHE; ?-------------------------------------------------; ?AGRONOMY VARIABLES; ?-------------------------------------------------; DCONSER=(T110>0); ?DIRRIG=(T111>0); ?PRINT T110 T111; FERT=T107_1; DFERT=FERT>0; PSIZE=T102; PSSQ=T102**2; HIST PSIZE; DPSIZE=(PSIZE<=5)*1 + [(PSIZE>5) & (PSIZE<=10)]*2 + [(PSIZE>10) & (PSIZE<=15)]*3 + (PSIZE>15)*4; DDPSIZE=DPSIZE>2; DUMMY DPSIZE; HIST (DISCRETE)DPSIZE; PNUMBER=T101; PNSQ=T101**2; ?PRINT PNUMBER;

PAGE 196

181 ?STOCKING RATE (CARGANEF); ?-------------------------------------------------; ?ZOOTECNICAL VARIABLES; ?-------------------------------------------------; ? HERD CLASSIFICATION BY SEX; DGSEX=(T22_08); ?HERD CLASSIFICATION; HCLASS=T22_06; DHCLASS=HCLASS>0; HIST HCLASS; ?WEIGHT BREEDING; WBREED=T22_122; WBREEDSQ=WBREED**2; ?BREEDING SYSTEM; BREEDS=T22_20; DBRED=T22_20>3; ?PRINT T22_20 DBRED; ?CONC FEED; CFEED=T22_211; DCFEED=CFEED>0; ?PRINT CFEED; MINERAL=T22_213; DMINERAL=MINERAL>0; SALT=T22_214; DSALT=SALT>0; MOLASS=T22_215; DMOLASS=MOLASS>0; OTHFEED=T22_216; DOTHFEED=OTHFEED>0; MINSALT=DMINERAL + DSALT; DMINSALT=MINSALT>0; ?PRINT MINSALT; HIST MINSALT; ?-------------------------------------------------; ?LABOR VARIABLES; ?-------------------------------------------------; ?LT. PER MILKER; LTMILKER=CANLECHE/(ORDENAD*1000); LTMILKSQ=LTMILKER**2; ?PRINT LTLEORDE LTMILKER; ?COW PER MILKER; COWMILKER=UAVACAO/ORDENAD; ?MSD COWMILKER;

PAGE 197

182 ?LABOR PER HA.; LABORHA=SUG/EQHOMBRE; ?LABOR PER COW; LABCOW=EQHOMBRE/TOTALUAB; ?Milking type; Hist(Discrete)t22_151; Milktype=T22_151>1; ?-------------------------------------------------; ?CAPITAL INTENSITY VARIABLES; ?-------------------------------------------------; FCAPITAL=CAPTIERR+CAPPASTO; ECAPITAL=CAPMAQUI+CAPEDIFI+CAPGANAD; RFCAPTCAP=FCAPITAL/CAPITOTA; RECAPTCAP=ECAPITAL/CAPITOTA; RECAPFCAP=ECAPITAL/FCAPITAL; TRACT=E406_011; DTRACT=TRACT>0; ?================================================; ? Model LGTE=LOG( (1/TE)-1); ?================================================; LGTE=LOG( (1/TE)-1); PROD= (LTVACTOT<=1000)+ [(LTVACTOT>1000) & (LTVACTOT<=1500)]*2+ [(LTVACTOT>1500) & (LTVACTOT<=2000)]*3+ [(LTVACTOT>2000) & (LTVACTOT<=2500)]*4+ (LTVACTOT>2500)*5; ?PRINT PROD LTVACTOT; ?HIST PROD; DUMMY PROD; ?ITERACTION BETWEEN PRODUCTION AND STOCKING RATE; ?stockprod=ltvactot*carganef; OLSQ(ROBUSTSE) LGTE C DPEDU DPEXP DPPER CRED DSUG2 DSUG3 DSUG4 Z2 Z3 Z4 PSYST2 PSYST3 PROD2 PROD3 PROD4 PROD5 DBRED DTEN DTECHN LTMILKER LTMILKSQ CARGANEF; ?CORR DBRED DTECHN; ?CORR DPEXP DBRED ; ?COPY @COEF COEF; ?COPY @SES STDEV; ?COPY @T TTEST; ?COPY %T PVALUES; ?MMAKE NEWH COEF STDEV TTEST PVALUES ; ?WRITE(FORMAT=EXCEL, FILE='HALFN3.XLS')NEWH; ?--------------------------------------------------------------; ?AVERAGE MODEL; ?--------------------------------------------------------------; RPEDU=(DPEDU=1) + (DPEDU=0)*-1; RPEXP=(DPEXP=1) + (DPEXP=0)*-1;

PAGE 198

183 RPPER=(DPPER=1) + (DPPER=0)*-1; RCRED=(CRED=0)*-1 + (CRED=1); RSUG2=(DSUG2-DSUG1); RSUG3=(DSUG3-DSUG1); RSUG4=(DSUG4-DSUG1); RZ2=(Z2-Z1); RZ3=(Z3-Z1); RZ4=(Z4-Z1); RPSYST2=(PSYST2-PSYST1); RPSYST3=(PSYST3-PSYST1); RPROD2=(PROD2-PROD1); RPROD3=(PROD3-PROD1); RPROD4=(PROD4-PROD1); RPROD5=(PROD5-PROD1); RBRED=(DBRED=1) + (DBRED=0)*-1; RTEN=(DTEN=1) + (DTEN=0)*-1; RTECHN=(DTECHN=1) + (DTECHN=0)*-1; OLSQ(ROBUSTSE)LGTE C RPEDU RPEXP RPPER RCRED RSUG2 RSUG3 RSUG4 RZ2 RZ3 RZ4 RPSYST2 RPSYST3 RPROD2 RPROD3 RPROD4 RPROD5 RBRED RTEN RTECHN LTMILKER LTMILKSQ CARGANEF; COPY @COEF COEF; COPY @SES STDEV; COPY @T TTEST; COPY %T PVALUES; MMAKE NEWH COEF STDEV TTEST PVALUES ; WRITE(FORMAT=EXCEL, FILE='HALFN4.XLS')NEWH; ?------------------------------------------------------------; ?STANDARDIZED COEFFICIENTS; ?------------------------------------------------------------; DOT(VALUE=J) 1-23; SET B.=@COEF(J); PRINT B.; ENDDOT; DOT LGTE RPEDU RPEXP RPPER RCRED RSUG2 RSUG3 RSUG4 RZ2 RZ3 RZ4 RPSYST2 RPSYST3 RPROD2 RPROD3 RPROD4 RPROD5 RBRED RTEN RTECHN LTMILKER LTMILKSQ CARGANEF; MSD(NOPRINT).; SET STD.=@STDDEV; ENDDOT; SET STB2=B2*STDRPEDU/STDLGTE; SET STB3=B3*STDRPEXP/STDLGTE; SET STB4=B4*STDRPPER/STDLGTE; SET STB5=B5*STDRCRED/STDLGTE; SET STB6=B6*STDRSUG2/STDLGTE; SET STB7=B7*STDRSUG3/STDLGTE; SET STB8=B8*STDRSUG4/STDLGTE; SET STB9=B9*STDRZ2/STDLGTE; SET STB10=B10*STDRZ3/STDLGTE; SET STB11=B11*STDRZ4/STDLGTE;

PAGE 199

184 SET STB12=B12*STDRPSYST2/STDLGTE; SET STB13=B13*STDRPSYST3/STDLGTE; SET STB14=B14*STDRPROD2/STDLGTE; SET STB15=B15*STDRPROD3/STDLGTE; SET STB16=B16*STDRPROD4/STDLGTE; SET STB17=B17*STDRPROD5/STDLGTE; SET STB18=B18*STDRBRED/STDLGTE; SET STB19=B19*STDRTEN/STDLGTE; SET STB20=B20*STDRTECHN/STDLGTE; SET STB21=B21*STDLTMILKER/STDLGTE; SET STB22=B22*STDLTMILKSQ/STDLGTE; SET STB23=B23*STDCARGANEF/STDLGTE; PRINT STB2-STB23; ?---------------------------------------------------------------------; ?T-STATISTIC FOR THE coefficients; ?-------------------------------------------------------------------; ?Education; SET RPEDU12= 2*@COEF(2)/SQRT(4*@VCOV(2,2)); SET NDT=@NOB-@NCID; CDF(T,DF=NDT)RPEDU12; ?Experience; SET RPEXP12= 2*@COEF(3)/SQRT(4*@VCOV(3,3)); CDF(T,DF=NDT)RPEXP12; ?Presence; SET RPPER12= 2*@COEF(4)/SQRT(4*@VCOV(4,4)); CDF(T,DF=NDT)RPPER12; ?Credit; SET RCRED12= 2*@COEF(5)/SQRT(4*@VCOV(5,5)); CDF(T,DF=NDT)RCRED12; ?SUG; SET RSUG12= [(2*@COEF(6))+ @COEF(7)+ @COEF(8)]/SQRT[(4*@VCOV(6,6))+ @VCOV(7,7)+ @VCOV(8,8) + (4*@VCOV(6,7)) + (4*@VCOV(6,8)) + (2*@VCOV(7,8))]; SET NDT=@NOB-@NCID; CDF(T,DF=NDT)RSUG12; SET RSUG13= [(@COEF(6))+ 2*@COEF(7)+ @COEF(8)]/SQRT[(@VCOV(6,6))+ 4*@VCOV(7,7)+ @VCOV(8,8) + (4*@VCOV(6,7)) + (2*@VCOV(6,8)) + (4*@VCOV(7,8))]; CDF(T,DF=NDT)RSUG13; SET RSUG14= [(@COEF(6))+ @COEF(7)+ 2*@COEF(8)]/SQRT[(@VCOV(6,6))+ @VCOV(7,7)+ 4*@VCOV(8,8) + (2*@VCOV(6,7)) + (4*@VCOV(6,8)) + (4*@VCOV(7,8))]; CDF(T,DF=NDT)RSUG14; SET RSUG23 = (@COEF(7)-@COEF(6))/SQRT[@VCOV(6,6)+@VCOV(7,7)2*@VCOV(6,7)]; CDF(T,DF=NDT)RSUG23; SET RSUG24=(@COEF(8)-@COEF(6))/SQRT[@VCOV(6,6)+@VCOV(8,8)2*@VCOV(6,8)]; CDF(T,DF=NDT)RSUG24; SET RSUG34=(@COEF(8)-@COEF(7))/SQRT[@VCOV(7,7)+@VCOV(8,8)2*@VCOV(7,8)];

PAGE 200

185 CDF(T,DF=NDT)RSUG34; ?ZONE; SET RZ12= [(2*@COEF(9))+ @COEF(10)+ @COEF(11)]/SQRT[(4*@VCOV(9,9))+ @VCOV(10,10)+ @VCOV(11,11) + (4*@VCOV(9,10)) + (4*@VCOV(9,11)) + (2*@VCOV(10,11))]; SET NDT=@NOB-@NCID; CDF(T,DF=NDT)RZ12; SET RZ13= [(@COEF(9))+ 2*@COEF(10)+ @COEF(11)]/SQRT[(@VCOV(9,9))+ 4*@VCOV(10,10)+ @VCOV(11,11) + (4*@VCOV(9,10)) + (2*@VCOV(9,11)) + (4*@VCOV(10,11))]; CDF(T,DF=NDT)RZ13; SET RZ14= [(@COEF(9))+ @COEF(10)+ 2*@COEF(11)]/SQRT[(@VCOV(9,9))+ @VCOV(10,10)+ 4*@VCOV(11,11) + (2*@VCOV(9,10)) + (4*@VCOV(9,11)) + (4*@VCOV(10,11))]; CDF(T,DF=NDT)RZ14; SET RZ23 = (@COEF(10)-@COEF(9))/SQRT[@VCOV(9,9)+@VCOV(10,10)2*@VCOV(9,10)]; CDF(T,DF=NDT)RZ23; SET RZ24=(@COEF(11)-@COEF(9))/SQRT[@VCOV(9,9)+@VCOV(11,11)2*@VCOV(9,11)]; CDF(T,DF=NDT)RZ24; SET RZ34=(@COEF(11)-@COEF(10))/SQRT[@VCOV(10,10)+@VCOV(11,11)2*@VCOV(10,11)]; CDF(T,DF=NDT)RZ34; ?PRODUCTION SYSTEM; SET RPSYS12= [(2*@COEF(12))+ @COEF(13)]/SQRT[(4*@VCOV(12,12))+ @VCOV(13,13)+ + (4*@VCOV(12,13))]; CDF(T,DF=NDT)RPSYS12; SET RPSYS13= [(@COEF(12))+ 2*@COEF(13)]/SQRT[(@VCOV(12,12))+ 4*@VCOV(13,13)+ + (4*@VCOV(12,13))]; CDF(T,DF=NDT)RPSYS13; SET RPSYS23= [(@COEF(13))@COEF(12)]/SQRT[(@VCOV(12,12))+ @VCOV(13,13)+ (2*@VCOV(12,13))]; CDF(T,DF=NDT)RPSYS23; ?PRODUCTION PER COW; SET RPROD12= [(2*@COEF(14))+ @COEF(15)+ @COEF(16)+ @COEF(17)]/SQRT[(4*@VCOV(14,14)) + @VCOV(15,15)+ @VCOV(16,16) + @VCOV(17,17)+ (4*@VCOV(14,15)) + (4*@VCOV(14,16)) + (4*@VCOV(14,17))+ (2*@VCOV(15,16)) + (2*@VCOV(15,17))+ (2*@VCOV(16,17))]; SET NDT=@NOB-@NCID; CDF(T,DF=NDT)RPROD12; SET RPROD13= [(@COEF(14))+ 2*@COEF(15)+ @COEF(16)+ @COEF(17)]/SQRT[(@VCOV(14,14)) + 4*@VCOV(15,15)+ @VCOV(16,16) + @VCOV(17,17)+ (4*@VCOV(14,15)) + (2*@VCOV(14,16))

PAGE 201

186 + (2*@VCOV(14,17))+ (4*@VCOV(15,16)) + (4*@VCOV(15,17)) +(2*@VCOV(16,17))]; SET NDT=@NOB-@NCID; CDF(T,DF=NDT)RPROD13; SET RPROD14= [(@COEF(14))+ @COEF(15)+ 2*@COEF(16)+ @COEF(17)]/SQRT[(@VCOV(14,14)) + @VCOV(15,15)+ 4*@VCOV(16,16) + @VCOV(17,17)+ (2*@VCOV(14,15)) + (4*@VCOV(14,16)) + (2*@VCOV(14,17))+ (4*@VCOV(15,16)) + (2*@VCOV(15,17)) +(4*@VCOV(16,17))]; SET NDT=@NOB-@NCID; CDF(T,DF=NDT)RPROD14; SET RPROD15= [(@COEF(14))+ @COEF(15)+ @COEF(16)+ 2*@COEF(17)]/SQRT[(@VCOV(14,14)) + @VCOV(15,15)+ @VCOV(16,16) + 4*@VCOV(17,17)+ (2*@VCOV(14,15)) + (2*@VCOV(14,16)) + (4*@VCOV(14,17))+ (2*@VCOV(15,16)) + (4*@VCOV(15,17)) +(4*@VCOV(16,17))]; SET NDT=@NOB-@NCID; CDF(T,DF=NDT)RPROD15; SET RPROD23=(@COEF(15)-@COEF(14))/SQRT[@VCOV(14,14)+@VCOV(15,15)2*@VCOV(14,15)]; CDF(T,DF=NDT)RPROD23; SET RPROD24=(@COEF(16)-@COEF(14))/SQRT[@VCOV(14,14)+@VCOV(16,16)2*@VCOV(14,16)]; CDF(T,DF=NDT)RPROD24; SET RPROD25=(@COEF(17)-@COEF(14))/SQRT[@VCOV(14,14)+@VCOV(17,17)2*@VCOV(14,17)]; CDF(T,DF=NDT)RPROD25; SET RPROD34=(@COEF(16)-@COEF(15))/SQRT[@VCOV(15,15)+@VCOV(16,16)2*@VCOV(15,16)]; CDF(T,DF=NDT)RPROD34; SET RPROD35=(@COEF(17)-@COEF(15))/SQRT[@VCOV(15,15)+@VCOV(17,17)2*@VCOV(15,17)]; CDF(T,DF=NDT)RPROD35; SET RPROD45=(@COEF(17)-@COEF(16))/SQRT[@VCOV(16,16)+@VCOV(17,17)2*@VCOV(16,17)]; CDF(T,DF=NDT)RPROD45; ?Breed; SET RBRED12= 2*@COEF(18)/SQRT(4*@VCOV(18,18)); CDF(T,DF=NDT)RBRED12; ?Tenure; SET RTEN12= 2*@COEF(19)/SQRT(4*@VCOV(19,19)); CDF(T,DF=NDT)RTEN12; ?technical assistanse frequency; SET RTECHN12= 2*@COEF(20)/SQRT(4*@VCOV(20,20)); CDF(T,DF=NDT)RTECHN12; ?---------------------------------------------------------------------? SIMULATION PREPARING THE VARIABLES? ?---------------------------------------------------------------------;

PAGE 202

187 DOT(VALUE=J) 1-23; SET B.=@COEF(J); PRINT B.; ENDDOT; ? DEFINING THE BETA COEFFICIENTS; DOT RPEDU RPEXP RPPER RCRED RSUG2 RSUG3 RSUG4 RZ2 RZ3 RZ4 RPSYST2 RPSYST3 RPROD2 RPROD3 RPROD4 RPROD5 RBRED RTEN RTECHN LTMILKER LTMILKSQ CARGANEF; SET X.=0; SET MN.=0; ENDDOT; ? THIS INITIALIZES THE VARIABLES BEFORE USING THE PROCEDURES. SET TESIM=0; SET I=0; MFORM(TYPE=GEN,NROW=500,NCOL=24) ZTECHZ=0; ? PROCEDURE FOR CLEARING THE VARIABLE VALUES BEFORE EACH SIMULATION; PROC ZINZ; DOT LTMILKER LTMILKSQ CARGANEF; MSD(NOPRINT) .; SET MN.=@MEAN(1); SET X.=MN.; ENDDOT; DOT RPEDU RPEXP RPPER RCRED RSUG2 RSUG3 RSUG4 RZ2 RZ3 RZ4 RPSYST2 RPSYST3 RPROD2 RPROD3 RPROD4 RPROD5 RBRED RTEN RTECHN; SET X.=0; ENDDOT; ENDPROC ZINZ; PROC ZSIMZ; ? THIS IS JUST THE SIMULATION NAME CAN BE ANYTHING; SET XB= B1 + B2*XRPEDU + B3*XRPEXP + B4*XRPPER + B5*XRCRED + B6*XRSUG2 + B7*XRSUG3 + B8*XRSUG4 + B9*XRZ2 + B10*XRZ3 + B11*XRZ4 + B12*XRPSYST2 + B13*XRPSYST3 + B14*XRPROD2 + B15*XRPROD3 + B16*XRPROD4 + B17*XRPROD5 + B18*XRBRED + B19*XRTEN + B20*XRTECHN + B21*XLTMILKER + B22*XLTMILKSQ + B23*XCARGANEF; SET TESIM = 1 / ( 1 + EXP(XB)); ? THIS IS THE SIMULATED VALUE OF THE TECHNICAL EFFICIENCY; SET I=i+1; SET J=1; SET ZTECHZ(I,J)=NOSIM; SET J=2; SET ZTECHZ(I,J)=TESIM; DOT XRPEDU XRPEXP XRPPER XRCRED XRSUG2 XRSUG3 XRSUG4 XRZ2 XRZ3 XRZ4 XRPSYST2 XRPSYST3 XRPROD2 XRPROD3 XRPROD4 XRPROD5 XRBRED XRTEN XRTECHN XLTMILKER XLTMILKSQ XCARGANEF; SET J=J+1; SET ZTECHZ(I,J)=.; ENDDOT; ENDPROC ZSIMZ; ?=====================================================================; ? STARTING THE SIMULATIONS ?=====================================================================; ? EDUCATION #1; SET NOSIM=1; ZINZ; DOT PEDU; SET XR.= 0; ZSIMZ; ENDDOT; ? HAD CREDIT; DOT PEDU; SET XR.=-1; ZSIMZ; ENDDOT; ? NO CREDIT;

PAGE 203

188 DOT PEDU; SET XR.= 1; ZSIMZ; ENDDOT; ? HAD CREDIT; ? EXPERIENCE #2; SET NOSIM=2; ZINZ; DOT PEXP; SET XR.= 0; ZSIMZ; ENDDOT; ? HAD CREDIT; DOT PEXP; SET XR.=-1; ZSIMZ; ENDDOT; ? NO CREDIT; DOT PEXP; SET XR.= 1; ZSIMZ; ENDDOT; ? HAD CREDIT; ? PRESENCE #3; SET NOSIM=3; ZINZ; DOT PPER; SET XR.= 0; ZSIMZ; ENDDOT; ? HAD CREDIT; DOT PPER; SET XR.=-1; ZSIMZ; ENDDOT; ? NO CREDIT; DOT PPER; SET XR.= 1; ZSIMZ; ENDDOT; ? HAD CREDIT; ? CREDIT ACTIVITIES #4; SET NOSIM=4; ZINZ; DOT CRED; SET XR.= 0; ZSIMZ; ENDDOT; ? HAD CREDIT; DOT CRED; SET XR.=-1; ZSIMZ; ENDDOT; ? NO CREDIT; DOT CRED; SET XR.= 1; ZSIMZ; ENDDOT; ? HAD CREDIT; ? SIZE OF FARM #5; SET NOSIM=5; ZINZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ? ZONES #6; SET NOSIM=6; ZINZ; DOT Z; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT Z; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT Z; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT Z; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT Z; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ? PSYST SYSTEM ACTIVITIES #7; SET NOSIM=7; ZINZ; DOT PSYST; SET XR.2= 0; SET XR.3= 0; ZSIMZ; ENDDOT; DOT PSYST; SET XR.2=-1; SET XR.3=-1; ZSIMZ; ENDDOT; DOT PSYST; SET XR.2= 1; SET XR.3= 0; ZSIMZ; ENDDOT; DOT PSYST; SET XR.2= 0; SET XR.3= 1; ZSIMZ; ENDDOT; ? COW PRODUCTION #8; SET NOSIM=8; ZINZ; DOT PROD; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; SET XR.5=0 ; ZSIMZ; ENDDOT; DOT PROD; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; SET XR.5=-1 ; ZSIMZ; ENDDOT;

PAGE 204

189 DOT PROD; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; SET XR.5=0 ; ZSIMZ; ENDDOT; DOT PROD; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; SET XR.5=0 ; ZSIMZ; ENDDOT; DOT PROD; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; SET XR.5=0 ; ZSIMZ; ENDDOT; DOT PROD; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; SET XR.5=1 ; ZSIMZ; ENDDOT; ? BREEDING SYSTEM #9; SET NOSIM=9; ZINZ; DOT BRED; SET XR.= 0; ZSIMZ; ENDDOT; ? HAD CREDIT; DOT BRED; SET XR.=-1; ZSIMZ; ENDDOT; ? NO CREDIT; DOT BRED; SET XR.= 1; ZSIMZ; ENDDOT; ? HAD CREDIT; ? TENURE #10; SET NOSIM=10; ZINZ; DOT TEN; SET XR.= 0; ZSIMZ; ENDDOT; ? HAD TENURE; DOT TEN; SET XR.=-1; ZSIMZ; ENDDOT; ? NO TENURE; DOT TEN; SET XR.= 1; ZSIMZ; ENDDOT; ? HAD TENURE; ? TECHNICAL ASISTANCE #11; SET NOSIM=11; ZINZ; DOT TECHN; SET XR.= 0; ZSIMZ; ENDDOT; ? HAD TECHN ASIST; DOT TECHN; SET XR.=-1; ZSIMZ; ENDDOT; ? NO TECHN ASIST; DOT TECHN; SET XR.= 1; ZSIMZ; ENDDOT; ? HAD TECHN ASIST; ? CHANGING MILKER PRODUCTIVITY; SET NOSIM=12; ZINZ; DO ADJ=.50 TO 4 BY .10; DOT LTMILK ; SET X.ER= MN.ER * ADJ; SET X.SQ=X.ER**2; ZSIMZ; ENDDOT; ENDDO; ? CHANGING STOCKING RATE; SET NOSIM=13; ZINZ; DO ADJ=0.5 TO 4 BY .10; DOT CARGANEF ; SET X.= MN.* ADJ; ZSIMZ; ENDDOT; ENDDO; ? EXAMPLE OF A COMBINATION OF TWO CONDITIONS SAY SIZE AND MILKERS; SET NOSIM=14; ZINZ; DO ADJ=.50 TO 4 BY 0.5; DOT LTMILK ; SET X.ER= MN.ER * ADJ; SET X.SQ=X.ER**2; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; ENDDO;

PAGE 205

190 ? EXAMPLE OF A COMBINATION OF TWO CONDITIONS SAY SIZE AND PSYST; SET NOSIM=15; ZINZ; DOT PSYST; SET XR.2= 0; SET XR.3= 0; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; DOT PSYST; SET XR.2=-1; SET XR.3=-1; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; DOT PSYST; SET XR.2= 1; SET XR.3= 0; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; DOT PSYST; SET XR.2= 0; SET XR.3= 1; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; ? EXAMPLE OF A COMBINATION OF TWO CONDITIONS SAY SIZE AND PROD; SET NOSIM=16; ZINZ; DOT PROD; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; SET XR.5=0 ; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; DOT PROD; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; SET XR.5=-1 ; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT;

PAGE 206

191 ENDDOT; DOT PROD; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; SET XR.5=0 ; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; DOT PROD; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; SET XR.5=0 ; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; DOT PROD; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; SET XR.5=0 ; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; DOT PROD; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; SET XR.5=1 ; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; ? EXAMPLE OF A COMBINATION OF TWO CONDITIONS SAY SIZE AND PPER; SET NOSIM=17; ZINZ; DOT PPER; SET XR.=-1; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; DOT PPER; SET XR.= 1; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT;

PAGE 207

192 ENDDOT; ? EXAMPLE OF A COMBINATION OF TWO CONDITIONS SAY SIZE AND TECHNICAL ASSISTANCE; SET NOSIM=18; ZINZ; DOT TECHN; SET XR.=-1; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; DOT TECHN; SET XR.= 1; ZSIMZ; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2=-1; SET XR.3=-1; SET XR.4=-1; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 1; SET XR.3= 0; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 1; SET XR.4= 0; ZSIMZ; ENDDOT; DOT SUG; SET XR.2= 0; SET XR.3= 0; SET XR.4= 1; ZSIMZ; ENDDOT; ENDDOT; ? NOW WRITE THE MATRIX OF SIMULATIONS OUT TO YOUR SPREADSHEET; WRITE(FORMAT=EXCEL,FILE='HALFN5.XLS') ZTECHZ; FREQ N; SMPL 1,127; SELECT.NOT.MISS(DPEDU).AND..NOT.MISS(DPEXP).AND..NOT.MISS(DPPER) .AND..NOT.MISS(CRED).AND..NOT.MISS(DSUG2).AND..NOT.MISS(DSUG3) .AND..NOT.MISS(DSUG4).AND..NOT.MISS(Z2).AND..NOT.MISS(Z3) .AND..NOT.MISS(Z4).AND..NOT.MISS(PSYST2).AND..NOT.MISS(PSYST3) .AND..NOT.MISS(PROD2).AND..NOT.MISS(PROD3).AND..NOT.MISS(PROD4) .AND..NOT.MISS(PROD5).AND..NOT.MISS(DBRED).AND..NOT.MISS(DTEN) .AND..NOT.MISS(DTECHN).AND..NOT.MISS(LTMILKER).AND..NOT.MISS(CARGANEF); ?PRINT DPEDU DPEXP DPPER CRED DSUG2 DSUG3 DSUG4 Z2 Z3 Z4 PSYST2 ?PSYST3 PROD2 PROD3 PROD4 PROD5 DBRED DTEN DTECHN LTMILKER CARGANEF; IDENT EQ1 TE=1/(1+EXP(0.803325+0.076724*DPEDU-0.565172*DPEXP0.339055*DPPER +0.161559*CRED+0.580120*DSUG2-0.514609*DSUG3+0.380311*DSUG4-0.029128*Z2 -0.395596*Z3-0.117368*Z4-0.132139*PSYST2-0.499804*PSYST3-0.219826*PROD2 -0.563285*PROD3-0.728173*PROD40.858165*PROD5+0.097898*DBRED+0.110738*DTEN -0.191263*DTECHN-0.028868*LTMILKER+0.000137768*LTMILKER**20.366656*CARGANEF)); ?DIFFER(PRINT) EQ1 LTMILKER; IDENT EQ2 LTMILKER=-(-0.028868/(2*(0.000137768))); ?IDENT EQ2 LTMILKER=LTMILKER+(-0.028868+2*0.000137768*LTMILKER)*Y*(Y1); EQSUB EQ2 EQ1; SIML(ENDOG=(LTMILKER,TE))EQ1 EQ2;END;

PAGE 208

APPENDIX C SIMULATION OUTPUT / TECHNICAL EFFICIENCY / HALF-NORMAL DISTRIBUTION

PAGE 209

194 Table C.1. Simulation of the determinants of technical efficiency (half-normal distribution) Concept TELTMILKERLTMILKSQCARGANEF Base (Average) 0.765243.012425.55120.75 Education (Illiterate or write and read) 0.772043.012425.55120.75 Education (elementary school or hi gher education) 0.758243.012425.55120.75 Experience (Less than or equal 5 years) 0.710743.012425.55120.75 Experience (More than 5 years) 0.812143.012425.55120.75 Presence (less than twice a week) 0.733443.012425.55120.75 Presence (Twice a week or more) 0.794343.012425.55120.75 Credit (No credit) 0.779443.012425.55120.75 Credit (Credit) 0.750443.012425.55120.75 F. size (< 300 ha) 0.784643.012425.55120.75 F. size (>= 300 and <= 400 ha) 0.671043.012425.55120.75 F. size (> 400 and <= 575 ha) 0.859143.012425.55120.75 F. size (> 575 ha) 0.713543.012425.55120.75 Location (Z1) 0.740043.012425.55120.75 Location (Z2) 0.745543.012425.55120.75 Location (Z3) 0.808743.012425.55120.75 Location (Z4) 0.761943.012425.55120.75 Prod. System (Cow-calf) 0.725343.012425.55120.75 Prod. System (Cow-yearling) 0.750843.012425.55120.75 Prod. System (Cow-steer) 0.813143.012425.55120.75 Prod. per cow (<= 1000 l) 0.669843.012425.55120.75

PAGE 210

195Table C.1. Continued. Concept TELTMILKERLTMILKSQCARGANEF Prod. per cow (> 1000 and <= 1500 l) 0.716543.012425.55120.75 Prod. per cow (> 1500 and <= 2000 l) 0.780943.012425.55120.75 Prod. per cow (> 2000 and <= 2500 l) 0.807843.012425.55120.75 Prod. per cow (> 2500 l) 0.827243.012425.55120.75 Breeding sys. (Natural breeding) 0.773943.012425.55120.75 Breeding sys. (Artificial insemination) 0.756343.012425.55120.75 Land tenure (Government land) 0.775043.012425.55120.75 Land tenure (Private land) 0.755143.012425.55120.75 Tech. Assintance (Frequency = Less than once a month) 0.747643.012425.55120.75 Tech. Assintance (Frequency = Once a month or higher) 0.781943.012425.55120.75 Labor productivity 0.696621.51462.52120.75 Labor productivity 0.716525.81666.02120.75 Labor productivity 0.734630.11906.53120.75 Labor productivity 0.751034.411184.04120.75 Labor productivity 0.765838.711498.56120.75 Labor productivity 0.779143.011850.07120.75 Labor productivity 0.791147.312238.58120.75 Labor productivity 0.801751.612664.10120.75 Labor productivity 0.811155.923126.62120.75 Labor productivity 0.819560.223626.14120.75 Labor productivity 0.826864.524162.66120.75 Labor productivity 0.833168.824736.18120.75 Labor productivity 0.838673.125346.70120.75 Labor productivity 0.843377.425994.22120.75 Labor productivity 0.847281.726678.75120.75 Labor productivity 0.850486.027400.28120.75 Labor productivity 0.852990.338158.81120.75 Labor productivity 0.854794.638954.34120.75 Labor productivity 0.855998.939786.87120.75

PAGE 211

196Table C.1. Continued. Concept TELTMILKERLTMILKSQCARGANEF Labor productivity 0.8564103.2310656.40120.75 Labor productivity 0.8563107.5311562.93120.75 Labor productivity 0.8556111.8312506.47120.75 Labor productivity 0.8542116.1313487.01120.75 Labor productivity 0.8522120.4314504.54120.75 Labor productivity 0.8496124.7415559.08120.75 Labor productivity 0.8462129.0416650.62120.75 Labor productivity 0.8421133.3417779.17120.75 Labor productivity 0.8372137.6418944.71120.75 Labor productivity 0.8314141.9420147.26120.75 Labor productivity 0.8248146.2421386.80120.75 Labor productivity 0.8172150.5422663.35120.75 Labor productivity 0.8086154.8423976.90120.75 Labor productivity 0.7988159.1525327.45120.75 Labor productivity 0.7878163.4526715.00120.75 Labor productivity 0.7755167.7528139.56120.75 Stocking rate 0.761243.012425.5560.38 Stocking rate 0.762043.012425.5572.45 Stocking rate 0.762843.012425.5584.53 Stocking rate 0.763643.012425.5596.60 Stocking rate 0.764443.012425.55108.68 Stocking rate 0.765243.012425.55120.75 Stocking rate 0.766043.012425.55132.83 Stocking rate 0.766843.012425.55144.91 Stocking rate 0.767643.012425.55156.98 Stocking rate 0.768443.012425.55169.06 Stocking rate 0.769143.012425.55181.13 Stocking rate 0.769943.012425.55193.21

PAGE 212

197Table C.1. Continued. Concept TELTMILKERLTMILKSQCARGANEF Stocking rate 0.770743.012425.55205.28 Stocking rate 0.771543.012425.55217.36 Stocking rate 0.772343.012425.55229.43 Stocking rate 0.773143.012425.55241.51 Stocking rate 0.773843.012425.55253.59 Stocking rate 0.774643.012425.55265.66 Stocking rate 0.775443.012425.55277.74 Stocking rate 0.776143.012425.55289.81 Stocking rate 0.776943.012425.55301.89 Stocking rate 0.777743.012425.55313.96 Stocking rate 0.778443.012425.55326.04 Stocking rate 0.779243.012425.55338.11 Stocking rate 0.780043.012425.55350.19 Stocking rate 0.780743.012425.55362.26 Stocking rate 0.781543.012425.55374.34 Stocking rate 0.782243.012425.55386.42 Stocking rate 0.783043.012425.55398.49 Stocking rate 0.783743.012425.55410.57 Stocking rate 0.784543.012425.55422.64 Stocking rate 0.785243.012425.55434.72 Stocking rate 0.786043.012425.55446.79 Stocking rate 0.786743.012425.55458.87 Stocking rate 0.787543.012425.55470.94 F. size (< 300 ha) & labor prod. 0.719621.51462.52120.75 F. size (>= 300 and <= 400 ha) & labor prod . 0.589621.51462.52120.75 F. size (> 400 and <= 575 ha) & labor prod. 0.811121.51462.52120.75 F. size (> 575 ha) & labor prod. 0.636921.51462.52120.75 F. size (< 300 ha) & labor prod. 0.797743.011850.07120.75

PAGE 213

198Table C.1. Continued. Concept TELTMILKERLTMILKSQCARGANEF F. size (>= 300 and <= 400 ha) & labor prod . 0.688343.011850.07120.75 F. size (> 400 and <= 575 ha) & labor prod. 0.868443.011850.07120.75 F. size (> 575 ha) & labor prod. 0.729443.011850.07120.75 F. size (< 300 ha) & labor prod. 0.842264.524162.66120.75 F. size (>= 300 and <= 400 ha) & labor prod . 0.749264.524162.66120.75 F. size (> 400 and <= 575 ha) & labor prod. 0.899364.524162.66120.75 F. size (> 575 ha) & labor prod. 0.784864.524162.66120.75 F. size (< 300 ha) & labor prod. 0.864086.027400.28120.75 F. size (>= 300 and <= 400 ha) & labor prod . 0.780686.027400.28120.75 F. size (> 400 and <= 575 ha) & labor prod. 0.914086.027400.28120.75 F. size (> 575 ha) & labor prod. 0.812986.027400.28120.75 F. size (< 300 ha) & labor prod. 0.8695107.5311562.93120.75 F. size (>= 300 and <= 400 ha) & labor prod . 0.7886107.5311562.93120.75 F. size (> 400 and <= 575 ha) & labor prod. 0.9177107.5311562.93120.75 F. size (> 575 ha) & labor prod. 0.8200107.5311562.93120.75 F. size (< 300 ha) & labor prod. 0.8601129.0416650.62120.75 F. size (>= 300 and <= 400 ha) & labor prod . 0.7749129.0416650.62120.75 F. size (> 400 and <= 575 ha) & labor prod. 0.9114129.0416650.62120.75 F. size (> 575 ha) & labor prod. 0.8079129.0416650.62120.75 F. size (< 300 ha) & labor prod. 0.8333150.5422663.35120.75 F. size (>= 300 and <= 400 ha) & labor prod . 0.7367150.5422663.35120.75 F. size (> 400 and <= 575 ha) & labor prod. 0.8932150.5422663.35120.75 F. size (> 575 ha) & labor prod. 0.7736150.5422663.35120.75 F. size (< 300 ha) & labor prod. 0.7814172.0529601.11120.75 F. size (>= 300 and <= 400 ha) & labor prod . 0.6668172.0529601.11120.75 F. size (> 400 and <= 575 ha) & labor prod. 0.8568172.0529601.11120.75 F. size (> 575 ha) & labor prod. 0.7096172.0529601.11120.75 F. size (< 300 ha) & prod. Sys. (Cow-calf) 0.746943.012425.55120.75

PAGE 214

199Table C.1. Continued. Concept TELTMILKERLTMILKSQCARGANEF F. size (>= 300 and <= 400 ha) & prod. Sys. (Cow-calf) 0.622943.012425.55120.75 F. size (> 400 and <= 575 ha) & prod. Sys. (Cow-calf) 0.831643.012425.55120.75 F. size (> 575 ha) & prod. Sys. (Cow-calf) 0.668643.012425.55120.75 F. size (< 300 ha) & prod. Sys. (Cow-yearling) 0.771143.012425.55120.75 F. size (>= 300 and <= 400 ha) & prod. Sys. (Cow-yearling) 0.653443.012425.55120.75 F. size (> 400 and <= 575 ha) & prod. Sys. (Cow-yearling) 0.849343.012425.55120.75 F. size (> 575 ha) & prod. Sys. (Cow-yearling) 0.697243.012425.55120.75 F. size (< 300 ha) & prod. Sys. (Cow-steer) 0.829543.012425.55120.75 F. size (>= 300 and <= 400 ha) & prod. Sys. (Cow-steer) 0.731443.012425.55120.75 F. size (> 400 and <= 575 ha) & prod. Sys. (Cow-steer) 0.890643.012425.55120.75 F. size (> 575 ha) & prod. Sys. (Cow-steer) 0.768843.012425.55120.75 F.size (< 300 ha) & prod. per cow (<= 1000 l) 0.694043.012425.55120.75 F.size (>= 300 and <= 400 ha) & prod. per cow (<= 1000 l) 0.559443.012425.55120.75 F.size (> 400 and <= 575 ha) & prod. per cow (<= 1000 l) 0.791443.012425.55120.75 F.size (> 575 ha) & prod. per cow (<= 1000 l) 0.607943.012425.55120.75 F.size (< 300 ha) & prod. per cow (> 1000 and <= 1500 l) 0.738643.012425.55120.75 F.size (>= 300 and <= 400 ha) & prod. per cow (> 1001 and <=1500 l) 0.612743.012425.55120.75 F.size (> 400 and <= 575 ha) & prod. per cow (> 1001 and <=1500 l) 0.825443.012425.55120.75 F.size (> 575 ha) & prod. per cow (> 1001 and <=1500 l) 0.658943.012425.55120.75 F.size (< 300 ha) & prod. per cow (> 1500 and <=2000 l) 0.799343.012425.55120.75 F.size (> 300 and <= 400 ha) & prod. per cow (> 1500 and <=2000 l) 0.690443.012425.55120.75 F.size (> 400 and <= 575 ha) & prod. per cow (> 1500 and <=2000 l) 0.869543.012425.55120.75 F.size (> 575 ha) & prod. per cow (> 1500 and <=2000 l) 0.731443.012425.55120.75 F.size (< 300 ha) & prod. per cow (> 2000 and <=2500 l) 0.824543.012425.55120.75 F.size (>= 300 and <= 400 ha) & prod. per cow (> 2000 and <=2500 l) 0.724543.012425.55120.75 F.size (> 400 and <= 575 ha) & prod. per cow (> 2000 and <=2500 l) 0.887143.012425.55120.75 F.size (> 575 ha) & prod. per cow (> 2000 and <=2500 l) 0.762643.012425.55120.75 F.size (< 300 ha) & prod. per cow (> 2500 l) 0.842543.012425.55120.75

PAGE 215

200Table C.1. Continued. Concept TELTMILKERLTMILKSQCARGANEF F.size (>= 300 and <= 400 ha) & prod. per cow (> 2500 l) 0.749743.012425.55120.75 F.size (> 400 and <= 575 ha) & prod. per cow (> 2500 l) 0.899543.012425.55120.75 F.size (> 575 ha) & prod. per cow (> 2500 l) 0.785343.012425.55120.75 F.size (< 300 ha) & producer presence (Less than twice a week) 0.754643.012425.55120.75 F.size (>= 300 and <= 400 ha) & producer presence (Less than twice a week) 0.632643.012425.55120.75 F.size (> 400 and <= 575 ha) & producer presence (Less than twice a week) 0.837243.012425.55120.75 F.size (> 575 ha) & producer presence (Less than twice a week) 0.677643.012425.55120.75 F.size (< 300 ha) & producer presence (Twice a week or higher) 0.811943.012425.55120.75 F.size (>= 300 and <= 400 ha) & producer presence (Twice a week or higher) 0.707343.012425.55120.75 F.size (> 400 and <= 575 ha) & producer presence (Twice a week or higher) 0.878443.012425.55120.75 F.size (> 575 ha) & producer presence (Twice a week or higher) 0.746943.012425.55120.75 F.size (< 300 ha) & tech. Assistance (Less than once a month) 0.768043.012425.55120.75 F.size (>= 300 and <= 400 ha) & tech. Assistance (Less than once a month) 0.649643.012425.55120.75 F.size (> 400 and <= 575 ha) & tech. Assistance (Less than once a month) 0.847143.012425.55120.75 F.size (> 575 ha) & tech. Assistance (Less than once a month) 0.693643.012425.55120.75 F.size (< 300 ha) & tech. Assistance (Once a month or higher) 0.800343.012425.55120.75 F.size (>= 300 and <= 400 ha) & tech. Assistance (Once a month or higher) 0.691843.012425.55120.75 F.size (> 400 and <= 575 ha) & tech. Assistance (Once a month or higher) 0.870243.012425.55120.75 F.size (> 575 ha) & tech. Assistance (Once a month or higher) 0.732743.012425.55120.75

PAGE 216

201 APPENDIX D SIMULATION OUTPUT / TECHNICAL EFFICIENCY / EXPONENTIAL DISTRIBUTION Table D.1. Simulation of the determinants of technical efficiency (Exponential distribution)Concept TE LTMILKER LTMILKSQ CARGANEF Base (Average) 0.903743.01 2425.55 120.75 Education (Illiterate or write and read) 0.906243.01 2425.55 120.75 Education (elementary school or hi gher education) 0.901243.01 2425.55 120.75 Experience (Less than or equal 5 years) 0.887643.01 2425.55 120.75 Experience (More than 5 years) 0.917743.01 2425.55 120.75 Presence (less than twice a week) 0.892243.01 2425.55 120.75 Presence (Twice a week or more) 0.914143.01 2425.55 120.75 Credit (No credit) 0.908943.01 2425.55 120.75 Credit (Credit) 0.898343.01 2425.55 120.75 F. size (< 300 ha) 0.908643.01 2425.55 120.75 F. size (>= 300 and <= 400 ha) 0.877243.01 2425.55 120.75 F. size (> 400 and <= 575 ha) 0.931343.01 2425.55 120.75 F. size (> 575 ha) 0.889543.01 2425.55 120.75 Location (Z1) 0.895743.01 2425.55 120.75 Location (Z2) 0.897543.01 2425.55 120.75 Location (Z3) 0.916743.01 2425.55 120.75 Location (Z4) 0.903543.01 2425.55 120.75 Prod. System (Cow-calf) 0.892043.01 2425.55 120.75 Prod. System (Cow-yearling) 0.898643.01 2425.55 120.75 Prod. System (Cow-steer) 0.918743.01 2425.55 120.75 Prod. per cow (<= 1000 l) 0.873643.01 2425.55 120.75 Prod. per cow (> 1000 and <= 1500 l) 0.890043.01 2425.55 120.75 Prod. per cow (> 1500 and <= 2000 l) 0.909243.01 2425.55 120.75 Prod. per cow (> 2000 and <= 2500 l) 0.916143.01 2425.55 120.75 Prod. per cow (> 2500 l) 0.922443.01 2425.55 120.75 Breeding sys. (Natural breeding) 0.907143.01 2425.55 120.75 Breeding sys. (Artificial insemination) 0.900243.01 2425.55 120.75 Land tenure (Government land) 0.905843.01 2425.55 120.75 Land tenure (Private land) 0.901543.01 2425.55 120.75 Tech. Assintance (Frequency = Less than once a month) 0.898743.01 2425.55 120.75 Tech. Assintance (Frequency = Once a month or higher) 0.908543.01 2425.55 120.75 Labor productivity 0.884021.51 462.52 120.75 Labor productivity 0.889825.81 666.02 120.75 Labor productivity 0.895030.11 906.53 120.75 Labor productivity 0.899734.41 1184.04 120.75

PAGE 217

202 Table D.1. Continued. Concept TE LTMILKER LTMILKSQ CARGANEF Labor productivity 0.904038.71 1498.56 120.75 Labor productivity 0.907843.01 1850.07 120.75 Labor productivity 0.911347.31 2238.58 120.75 Labor productivity 0.914451.61 2664.10 120.75 Labor productivity 0.917155.92 3126.62 120.75 Labor productivity 0.919660.22 3626.14 120.75 Labor productivity 0.921764.52 4162.66 120.75 Labor productivity 0.923668.82 4736.18 120.75 Labor productivity 0.925373.12 5346.70 120.75 Labor productivity 0.926777.42 5994.22 120.75 Labor productivity 0.927881.72 6678.75 120.75 Labor productivity 0.928886.02 7400.28 120.75 Labor productivity 0.929590.33 8158.81 120.75 Labor productivity 0.930194.63 8954.34 120.75 Labor productivity 0.930498.93 9786.87 120.75 Labor productivity 0.9305103.23 10656.40 120.75 Labor productivity 0.9304107.53 11562.93 120.75 Labor productivity 0.9302111.83 12506.47 120.75 Labor productivity 0.9297116.13 13487.01 120.75 Labor productivity 0.9290120.43 14504.54 120.75 Labor productivity 0.9281124.74 15559.08 120.75 Labor productivity 0.9270129.04 16650.62 120.75 Labor productivity 0.9256133.34 17779.17 120.75 Labor productivity 0.9240137.64 18944.71 120.75 Labor productivity 0.9222141.94 20147.26 120.75 Labor productivity 0.9201146.24 21386.80 120.75 Labor productivity 0.9177150.54 22663.35 120.75 Labor productivity 0.9150154.84 23976.90 120.75 Labor productivity 0.9120159.15 25327.45 120.75 Labor productivity 0.9087163.45 26715.00 120.75 Labor productivity 0.9049167.75 28139.56 120.75 Stocking rate 0.902043.01 2425.55 60.38 Stocking rate 0.902343.01 2425.55 72.45 Stocking rate 0.902743.01 2425.55 84.53 Stocking rate 0.903043.01 2425.55 96.60 Stocking rate 0.903443.01 2425.55 108.68 Stocking rate 0.903743.01 2425.55 120.75 Stocking rate 0.904043.01 2425.55 132.83 Stocking rate 0.904443.01 2425.55 144.91 Stocking rate 0.904743.01 2425.55 156.98 Stocking rate 0.905043.01 2425.55 169.06 Stocking rate 0.905443.01 2425.55 181.13 Stocking rate 0.905743.01 2425.55 193.21 Stocking rate 0.906043.01 2425.55 205.28

PAGE 218

203 Table D.1. Continued. Concept TELTMILKER LTMILKSQ CARGANEF Stocking rate 0.906443.01 2425.55 217.36 Stocking rate 0.906743.01 2425.55 229.43 Stocking rate 0.907043.01 2425.55 241.51 Stocking rate 0.907343.01 2425.55 253.59 Stocking rate 0.907743.01 2425.55 265.66 Stocking rate 0.908043.01 2425.55 277.74 Stocking rate 0.908343.01 2425.55 289.81 Stocking rate 0.908643.01 2425.55 301.89 Stocking rate 0.909043.01 2425.55 313.96 Stocking rate 0.909343.01 2425.55 326.04 Stocking rate 0.909643.01 2425.55 338.11 Stocking rate 0.909943.01 2425.55 350.19 Stocking rate 0.910243.01 2425.55 362.26 Stocking rate 0.910643.01 2425.55 374.34 Stocking rate 0.910943.01 2425.55 386.42 Stocking rate 0.911243.01 2425.55 398.49 Stocking rate 0.911543.01 2425.55 410.57 Stocking rate 0.911843.01 2425.55 422.64 Stocking rate 0.912143.01 2425.55 434.72 Stocking rate 0.912443.01 2425.55 446.79 Stocking rate 0.912743.01 2425.55 458.87 Stocking rate 0.913043.01 2425.55 470.94

PAGE 219

204 LIST OF REFERENCES Aigner, D., C.A.K. Lovell, and P. Schmidt. “Formulation and Estimation of Stochastic Frontier Production Function Models.” Journal of Econometric 6, 1977: 21-37. Battesse, G.E., and G. S. Corra. “Estimation of a Production Frontier Model: With Application to the Pastoral Zone of Eastern Australia.” Australian Journal of Agricultural Economics 21 (3), 1977: 169-179 Bauer, P.W. “Recent Developments in the Econometric Estimation of Frontiers.” Journal of Econometrics 46 (1/2), 1990:39-56. Bravo-Ureta, B.E., and A.E. Pinheiro. “E fficiency Analysis of Developing Country Agriculture: A Review of the Frontier Function Literature.” Agricultural and Resource Economics Review 22 (1) , 1993: 88-101. Bravo-Ureta, B.E., and L. Rieger. “Dai ry Farm Efficiency Measurement Using Stochastic Frontiers and Neoclassical Duality.” American Journal of Agricultural Economics 73, 1991: 421-428. Bravo-Ureta, B.E., and L. Rieger, “Alte rnative Production Frontier Methodologies and Dairy Farm Efficiency.” Journal of Agricultural Economics 41, 1990: 215-226 Castillo, J. “Los Sistemas the Produccion.” En Gonzlez-Stagnaro, C. (ed.), Ganadera Mestiza de Doble Propsito, Ed. Astro Data S.A., Maracaibo, Venezuela, Cap. I, 1992: 26-40. Delgado de S.H. “Descripcin del Sistema de Produccin de Leche y Carne en la Cuenca del Lago de Maracaibo.” En Foro Sistema de Produccin Bovina de Leche y Carne en la Cuenca del Lago de Maracaibo, Corporacin de Desarrollo de la Regin Zulia ( CORPOZULIA), La Universidad del Zulia, Programa Laberinto (Convenio LUZ CORPOZULIA INVERSORA LABERINTO), Mimeografiado, 1989: 21. Estrada, R.D. “Ventajas Economicas Comparativas de los Sistemas Doble Proposito.” En Lopez, A.C. (ed.), Ganaderiade Doble Proposito, Memorias Seminario International, Ed. ICA-GTZ, Cartagenas de Indias, Colombia, 1993: 32-55. Farrell, M.J. “The Measurement of Productivity Efficiency.” Journal of the Royal Statistical Society, Series A, General, 120, 1957: 253-281

PAGE 220

205 Feder, G., R.E. Just, and D. Zilberman. “Adoption of Agricultural Innovations in Developing Countries: A Survey.” Economic Development and Cultural Change 33, 1985: 255-298. Fernandez, N. “Diagnostico Tecnico-Economico de la Ganaderia Bovina de Doble Proposito del Estado Zulia.” Unidad Coordinadora de Proyectos Conjuntos, Universidad Del Zulia,Venezuela, Mimeografiado, 1996: 64. Fernandez, N. “Aspectos Tcnico-Econmicos de la Ganadera Bovina de Doble Propsito de la Cuenca del Lago de Maracaibo.” En Gonzlez-Stagnaro, C. (ed.), Ganadera Mestiza de Doble Propsito, Ed. Astro Data S.A., Maracaibo, Venezuela, Cap. XXV, 1992: 535-552. Fernandez-Baca, S. “Desafios de la Producion Bovina de Doble Proposito en la America Tropical.” En Madrid, N y Eleazar Soto B. (eds.), Manejo de la Ganaderia Meztiza de Doble Proposito, Ed. Astro Data S.A., Maracaibo, Venezuela, Cap. Introductorio, 1995: 3-19. Frsund, F.R., C.A.K. Lovell, and P. Schmidt. “A Survey of Frontier Production Functions and their Relationship to Efficiency Measurement.” Journal of Econometrics 13, 1980: 5-25. Garcia, O. “Desarrollo de Sistemas de Produccion de Ganado de Doble Proposito en Pastoreo Bajo Condiciones de la Orinoquia y Amazonia Colombianas.” En Lopez, A.C. (ed.), Ganaderiade Doble Proposito, Memorias Seminario International, Ed. ICA-GTZ, Cartagenas de Indias, Colombia, 1993: 125-139. Gonzlez B. “Ganadera Mestiza a Base de Pastos en Condiciones Hmedas y Subhmedas de la Cuenca del Lago de Maracaibo.” En Gonzlez-Stagnaro, C. (ed.), Ganadera Mestiza de Doble Propsito, Ed. Astro Data S.A, Maracaibo, Venezuela, Cap. XVII, 1992: 365-379 pp. Graterol, J; E.Fuenmayor; A. Gmez; O. Rodrguez, y R. Acosta. “Consideraciones Sobre la Identificacin y Clasificacin de los Sistemas de Produccin de Ganadera de Doble Propsito en el Estado Zulia.” FONAIAP, Estacin Experimental Zulia, Maracaibo, Venezuela, Mimeografiado Serie C No 6-21, 1987: 35. Greene, W.H. “Frontier Production Functions.” In Pesaran, H. and P.Schmidt (eds.), Handbook of Applied Econometrics , Vol. II-Microeconomics, Cambridge, Mass., USA: Blackwell, 1997. Greene, W.H. Limdep 6 . Bellport, NY: Econometric Software, Inc., 1991. Greene, W.H. “A Gamma-Distributed Stochastic Frontier Model.” Journal of Econo metrics 46, 1990: 141-163.

PAGE 221

206 Greene, W.H. “Maximum Likelihood Estimation of Econometric Frontier Functions.” Journal of Econometrics 13, 1980:27-56. Grosskopf, S. “Efficiency and Productivity.” In Fried, H.O., C.A.K. Lovell, and S.S. Schmidt (eds), The Measurement of Productive Efficiency: Techniques and Applications , New York: Oxford University Press, 1993:160-194. Guerra, G. Manual de Administracin de Empresas Agropecuarias . 2nd ed., San Jos, Costa Rica: Instituto Interamericano de Cooperacin para la Agricultura (IICA), 1992. Gujarati, D.N. Basic Econometrics. Third edition, New York: McGraw-Hill, Inc., 1995. Guzman, S. “Situacion Actual del Sistema de Ganaderia Bovina de Doble Proposito en Colombia.” En Navarro, V.R., H.J. Anzola, and G. A. Ossa (eds.), Ganaderia de Doble Proposito, Ed. ICA-PRONATA, Instituto Colombiano Agropecuario, Bogota, Colombia, 1995: 1-9. Hall, B.H., and C. Cummins. Reference Manual, Time Series Processor, Version 4.4, Palo Alto, California, 1988. Holmann, F., y C. Lascano. “Una Nueva Estrategia para Mejorar los Sistemas de Produccion de Doble Proposito en los Tropicos: El Consorcio Tropileche.” En Gonzalez-Stagnaro C., N. Madrid-Bury, and E. Soto (eds.), Mejora de la Ganaderia Mestiza de Doble Proposito, Ed. Astro Data S.A., Maracaibo, Venezuela, Cap.II, 1998: 33-58. Holmann, F., R.W. Blake, M.V. Hahn, R. Barker, R.A. Milligan, P.A. Oltenacu, and T.L. Stanton. “Comparative Profitability of Purebred and Crossbred Holstein Herds in Venezuela.” Journal of Dairy Science 73, 1990: 2190-2205. Isea, W., y E. Rincon. “Produccion de Leche y Crecimiento en la Ganaderia Mestiza de Doble Proposito 1992.” En Gonzlez-Stagnaro, C. (ed.), Ganadera Mestiza de Doble Propsito, Ed. Astro Data S.A., Maracaibo, Venezuela, Cap. VI, 1992: 113-139. Jondrow, J., C.A.K. Lovell, I.S. Materov, and P. Schmidt. “On the Estimation of Technical Efficiency in the Stochastic Frontier Production Function Model.” Journal of Econometrics 19, 1982: 233-238. Kalirajan, K.P. “The Importance of Efficient Use in the Adoption of Technology: A Micro Panel Data Analysis.” Journal of Productivity Analysis 2, 1991: 113-126. Kay, R.D. Farm Management. Planning, Control, and Implementation . New York: McGraw-Hill, Inc., 1981.

PAGE 222

207 Kopp, R.J., and W.E. Diewert. “The Decomposition of Frontier Cost Function Deviation Into Measures of Technical and Allocative Efficiency.” Journal of Econometrics 19, 1982: 319-331. Kopp, R. J., and V. K. Smith. “Frontier Production Function Estimates for Steam Electric Generation: A Comparative Analysis.” Southern Economic Journal 47, 1980: 1049-1059. Kumbhakar, S.C., and C.A.K. Lovell. Stochastic Frontier Analysis . Cambridge, United Kingdom: Cambridge University Press, 2000. Kumbhakar, S.C., S. Ghosh, and J.T. McGuckin. “A Generalized Production Frontier Approach for Estimating Determinants of Inefficiency in U.S. Dairy Farms.” Journal of Business and Economics Statistic 9 (3), 1991: 279-286. Langedyk, K. “Farm Size and Productivity: An Empirical Analysis of the Farm SizeProductivity Relationship in Ecuador.” Ph.D . Dissertation, University of Florida, Gainesville, 2001. McCarty, T.A., and S. Yaiswarng. “Techni cal Efficiency in New Jersey School Districts.” In Fried, H. O., Lovell, C. A. K., and S. S. Schmidt (eds), The Measurement of Productive Efficiency: Techniques and Applications, New York: Oxford University Press, 1993: 271-286. McDermontt, J.K., and C.O. Andrew. “Agricultural Program Management.” Working paper, Food and Resource Economics Dept., University of Florida, Gainesville, 1999. Meeusen, W., and Van Den Broeck. “Efficiency Estimation from Cobb-Douglass Production Function with Composed Error.” International Economic Review 18, 1977: 435-444 Morillo, F.J., y Urdaneta, F. “Sistemas de Produccion de Doble Proposito con Bovinos para los Tropicos Americanos.” Memorias de la Conferencia International sobre Ganaderia en los Tropicos, Institute of Food and Agricultural Science, University of Florida, Gainesville, Florida, U.S., 1998: 81-104. Neff, D.L., P. Garcia, and R. Hornbaker. “Efficiency Measures Using the RayHomothetic Function: A Multiperiod Analysis.” Southern Journal of Agricultural Economics 23(2), 1991: 113-121. Nicholson, C.F., R.W. Blake, C.I. Urbina, D.R. Lee, D.G. Fox, and P.J. Van Soest. “Economic Comparison of Nutritional Management Strategies for Venezuelan Dual-Purpose Cattle System.” Journal of Animal Science 72, 1994: 1680-1696.

PAGE 223

208 Ortega, L. “Diagnostico Tecnico_Economico de las Fincas Bovinas de la Zona Sur del Lago del Estado Zulia.” Facultad de Agronomia, Universidad del Zulia, Maracaibo, Venezuela, Mimeografiado, 1996: 111. Pearson de Vaccaro, L. “Sistemas de Produccin Bovina Predominantes en el Trpico Latinoamericano.” En Arango-Nieto, L.; A. Charry y R. Vera (eds), Panorama de la Ganadera de Doble Propsito en la Amrica Tropical, Ed. ICA-CIAT, Bogot, Colombia, 1986: 29-44. Peterson, W., and Y. Hayami. “Technical change in Agriculture.” In Martin, Lee R. (ed.), A Survey of Agricultural Economics Literature Vol. 1, Minneapolis: University of Minnesota Press, 1977: 497-540. Pindyck, R.S., and D.L. Rubinfeld. Econometric Models and Economic Forescasts . 4th ed., Boston: Irwin/McGraw-Hill, 1998 Plasse, D. “Presente y Futuro de la Produccin Bovina en Venezuela.” En GonzlezStagnaro, C. (ed.), Ganadera Mestiza de Doble Propsito, Ed. Astro Data S.A., Maracaibo, Venezuela, Cap. Introductorio, 1992: 1-24. Ritter, C., and L. Simar. “Pitfalls of Normal-Gamma Stochastic Frontier Models.” Journal of Productivity Analysis 8 (2), 1997: 167-82. Romero, O. “Productividad and Tecnologia: Claves de La Ganaderia de Doble Proposito.” En Madrid, N y Eleazar Soto B (eds.), Manejo de la Ganaderia Meztiza de DobleProposito, Ed. Astro Data S.A., Maracaibo, Venezuela, Cap. III, 1995: 57-89. Schmidt, P. “Frontier Production Functions.” Econometrics Reviews 4, 1985-86: 289328. Sere, C., and de Vaccaro, L. “Milk Production from Dual-Purpose Systems in Tropical Latin America.” In Smith, A.J. (ed.), Milk production in Developing Countries, Univ. Edinburgh, Scotland, Great Britai n, Trowbridge: Redwood Burn Ltd., 1985: 459-475. Sposito, F. La Investigacion de Fincas en la Transferencia de Tecnologa Agrcola. Facultad de Agronomia,. Universidad Central de Venezuela, Maracay: Imprenta Universitaria, 1994. Stevenson, R.E. “Likelihood Functions for Generalized Stochastic Frontier Estimation.” Journal of Econometrics 13, 1980: 57-66. Stobbs, T.H. “Quality of Pasture and Forage Crops for Dairy Production in the Tropical Regions of Australia.” Tropical Grassland 5(3), 1971: 159-70.

PAGE 224

209 TSP International. TSP 4.4. Palo Alto, CA., Econometric Software, 1998. Tauer, L.W., and K.P. Belbase. “Technical Efficiency of New York Dairy Farms.” Northeastern Journal of Agricultural Research Economics 16, 1987: 10-16. Urdaneta, F., E. Martinez, H. Delgado, Z. Chirinos, D. Osuna, y L. Ortega. “Caracterizacion de los Sistemas de Pr oduccion de Ganaderia Bovina de Doble Proposito de la Cuenca del Lago de Maracaibo.” En Madrid-Bury, N. and E. Soto (eds.), Manejo de la Ganaderia Mestiza de Doble Proposito, Ed. Astro Data S.A., Maracaibo, Venezuela, Cap. I, 1995: 21-44. Urdaneta, M., H. Delgado, y D. Osuna. “Gan aderia Bovina a Base de Pastos en la Altiplanicie de Maracaibo.” En Gonzlez-Stagnaro, C. (ed.), Ganadera Mestiza de Doble Propsito, Ed. Astro Data S.A., Maracaibo, Venezuela, Cap. XVIII, 1992: 381-406. Vaccaro, R., L. Vaccaro, O. Verde, R. Alvarez, H. Mejias, L. Rios, y E. Romero. “Experiencias del Proyecto UCV-IICA: Mejoramiento Genetico en Ganaderias de Doble Proposito.” En Lopez, A.C. (ed.), Ganaderia de Doble Proposito, Memorias Seminario International, Ed. ICA-PRONATA, Cartagenas de Indias, Colombia, 1993: 79-98 Vargas, H.E. “Experiencias del Proyecto Mejoramiento de Sistemas de Produccion Bovina de Doble Proposito en Guatemala.” En Lopez, A.C. (ed.), Ganaderia de Doble Proposito, Memorias Seminario International, Ed. ICA-PRONATA, Cartagenas de Indias, Colombia, 1993: 57-66. Zepeda, L. “Simultaneity of Technology Adoption and Productivity.” Journal of Agricultural and Resource Economics 19(1), 1994: 46-57.

PAGE 225

210 BIOGRAPHICAL SKETCH Leonardo E. Ortega Soto was born on 16 Jan. 1959, in Maracaibo, Zulia State, Venezuela. He spent many of his teenage years working craftsmanship in a poultry farm where he later became a technical manager. He received a B.A. degree in agronomy (specialization in irrigation systems) in 1981, and in 1990 his Master of Science degree in animal production from La Universidad del Zulia (LUZ). Concurrently for two years, he worked for a water wells and irrigation design consultant company calculating deep wells pumps and irrigation systems. From 1985-1989 he worked as a technical manager of an intensive production livestock farm in Zulia State. In 1990 he joined LUZ as an assistant professor where he teaches animal production systems and project evaluation, and has participated in some administrative assignments (management program of LUZ’s livestock farm, agronomy research institute’s rural section program and resettlement project of Agricultural Community “Curva del Pato” financed by Petroquimica de Venezuela). In 1994, he was recipient of the Cochran Fellowship from USDA in the area of agricultural program management at the University of Florida (UF). In summer 1997, he enrolled at UF to pursue the Doctor of Philosophy degree in the Food and Resource Economics Department. Leonardo is currently a candidate for the Doctor of Philosophy degree. He is married to Yoana Newman (currently a postdoctoral research associate at UF), and they have two children, David Leonardo and Yoana Victoria.