The Role of Infrastructure in Land-use Dynamics and Rice Production in Viet Nam

Edmonds, Christopher M. Working Paper The Role of Infrastructure in Land-use Dynamics and Rice Production in Viet Nam ERD Working Paper Series, No. 16 Provided in Cooperation with: Asian Development Bank (ADB), Manila Suggested Citation: Edmonds, Christopher M. (2002) : The Role of Infrastructure in Land-use Dynamics and Rice Production in Viet Nam, ERD Working Paper Series, No. 16, Asian Development Bank (ADB), Manila, https://hdl.handle.net/11540/1933 This Version is available at: https://hdl.handle.net/10419/109237 Standard-Nutzungsbedingungen: Die Dokumente auf EconStor dürfen zu eigenen wissenschaftlichen Zwecken und zum Privatgebrauch gespeichert und kopiert werden. Sie dürfen die Dokumente nicht für öffentliche oder kommerzielle Zwecke vervielfältigen, öffentlich ausstellen, öffentlich zugänglich machen, vertreiben oder anderweitig nutzen. Sofern die Verfasser die Dokumente unter Open-Content-Lizenzen (insbesondere CC-Lizenzen) zur Verfügung gestellt haben sollten, gelten abweichend von diesen Nutzungsbedingungen die in der dort genannten Lizenz gewährten Nutzungsrechte. Terms of use: Documents in EconStor may be saved and copied for your personal and scholarly purposes. You are not to copy documents for public or commercial purposes, to exhibit the documents publicly, to make them publicly available on the internet, or to distribute or otherwise use the documents in public. If the documents have been made available under an Open Content Licence (especially Creative Commons Licences), you may exercise further usage rights as specified in the indicated licence. http://creativecommons.org/licenses/by/3.0/igo ECONOMICS AND RESEARCH DEPARTMENT ERD WORKING PAPER SERIES NO. 16 Christopher Edmonds July 2002 Asian Development Bank The Role of Infrastructure in Land-use Dynamics and Rice Production in Viet Nam’s Mekong River Delta 19 ERD Working Paper No. 16 THE ROLE OF INFRASTRUCTURE IN LAND-USE DYNAMICS AND RICE PRODUCTION IN VIET NAM’S MEKONG RIVER DELTA July 2002 Christopher Edmonds is an Economist with the Development Indicators and Policy Research Division of the Economics and Research Department, Asian Development Bank. The author acknowledges the close collaboration of S.P. Kam (GIS Specialist at the International Rice Research Institute) in GIS aspects of the research; H.C. Viet (Institute of Agricultural Sciences, Ho Chi Minh City) in the data collection and interpretation; and L. Villano (IRRI) in programming and research assistance. The support and assistance of several other individuals were important in enabling this reesearch to be carried out, namely, from the Institute of Agricultural Sciences of Vietnam: Professors P.V. Bien, H.T. Quoc, and T.T. Khai; from IRRI: C.T. Hoanh and T.P. Tuong; from Can Tho University: V.Q. Minh; and from the Sub-Institute for Agricultural Planning and Projection: Dr. N.V. Nhan. Any errors are the sole the responsibility of the author. The research was funded in part by the Rockefeller Foundation Social Science Research in Agriculture Program. ERD Working Paper No. 16 THE ROLE OF INFRASTRUCTURE IN LAND-USE DYNAMICS AND RICE PRODUCTION IN VIET NAM’S MEKONG RIVER DELTA 20 Asian Development Bank P.O. Box 789 0980 Manila Philippines 2002 by Asian Development Bank July 2002 ISSN 1655-5252 The views expressed in this paper are those of the author(s) and do not necessarily reflect the views or policies of the Asian Development Bank. 21 Foreword The ERD Working Paper Series is a forum for ongoing and recently completed research and policy studies undertaken in the Asian Development Bank or on its behalf. The Series is a quick-disseminating, informal publication meant to stimulate discussion and elicit feedback. Papers published under this Series could subsequently be revised for publication as articles in professional journals or chapters in books. ERD Working Paper No. 16 THE ROLE OF INFRASTRUCTURE IN LAND-USE DYNAMICS AND RICE PRODUCTION IN VIET NAM’S MEKONG RIVER DELTA 22 Contents Abstract vii I. Introduction 1 II. Description of the Study Area 2 III. Land Use Model 7 IV. Estimation Strategy and Results 8 V. Simulation Model for Evaluation of Investments 15 VI. Conclusion 16 References 17 23 Abstract This study examines the role of infrastructure development and technical change in explaining increases in agricultural production and changes in land use in the Mekong Delta Region of Viet Nam during the mid-1990s. The study relies on econometric analysis of household-level longitudinal farm survey data covering about 150 farms from eight villages in the Mekong River Delta from1994 to 1998. A model is developed that combines spatial factors in a neoclassical production framework to examine changes in land use and agricultural technology. Estimates make use of panel data estimation procedures that control for the effect of unobserved variables. Major findings emerging from the study are that the transportation costs involved in moving agricultural input and output between farms and markets significantly effect farm land use and production decisions. Greater transport costs reduced the likelihood that farms adopt intensive cropping patterns or cultivate nonrice crops. Improvements in roads and waterways both reduce transport costs in the area. Results suggest the quality of local water management infrastructure is much more important than transport costs in explaining the increased intensity of land use and level of production observed in the Mekong Delta during the 1990s. A simulation model is developed to highlight the implications of findings for policy aiming to increase rice production or alter land use in the Mekong Delta in the future. Unfortunately, lack of information on the costs of alternative infrastructure investments limits the policy conclusions that can be drawn from the study. 1 I. INTRODUCTION The increase in rice production in Viet Nam during the 1990s represents one of the recent success stories of Asian agricultural development. The increase in national production took the country from having a large deficit between rice demand and supply to becoming the third largest rice exporter worldwide. This expansion contributed to the country’s high rate of GNP growth by providing urban areas with cheap food and generating foreign exchange earnings. Increases in rice production in the Mekong River Delta, which supplies about half of Viet Nam’s total rice production, averaged about 6.3 percent per year during the 1990s according to official statistics. Although the rapid growth in rice production in Viet Nam is widely known, there have been few studies of the changes in market and physical infrastructure that prompted farm-level changes in rice production techniques and land use, and led to the production increases. Both biophysical and socioeconomic constraints influence land use decisions and limit the production activities of farming families in the Mekong River Delta. Infrastructure development and changes in economic policies modify both types of constraints. This makes understanding these constraints essential to developing technologies and advising on policies to increase agricultural production and spur economic development in the region. Integration of traditional econometric techniques with data organized in a geographic information system (GIS) offers a promising method for modeling constraints. This paper reviews a microeconomic model developed to explore the relationship between biophysical and socioeconomic characteristics and to derive hypotheses concerning the importance of local infrastructure development, market expansion, new technology adoption, and changes in input application in the mid-1990s in explaining production changes observed in the Mekong Delta. Hypotheses are examined using available data. Different areas in the Delta can be understood as being emblematic of different levels of agricultural development in the transition from rainfed to irrigated rice agriculture. This makes it a useful case to study, and findings carry implications for other areas in Asia making the transition between rainfed and irrigated agriculture. This paper begins by characterizing the changes in the agricultural environment and the household-level responses to these changes as captured in farm survey, GIS, and provincial level statistics. Two important developments in the study area during the 1990s were the “deepening” and geographic extension of market reforms started in 1988, and the installation of new water control and transport infrastructure. This latter development increased both the area protected from saline water intrusion and the reach of irrigation for dry season rice cultivation. The major changes in policies, institutions, and infrastructure relevant to rice agriculture during the 1990s are also briefly considered. Our review of the biophysical characteristics of surveyed villages relied on GIS data compiled by the International Rice Research Institute (IRRI) and collaborating research ERD Working Paper No. 16 THE ROLE OF INFRASTRUCTURE IN LAND-USE DYNAMICS AND RICE PRODUCTION IN VIET NAM’S MEKONG RIVER DELTA 2 institutions in Viet Nam. We capture farm-level changes in rice output and land use from a longitudinal household survey (1994 to 1997). The survey data were collected by the Institute of Agricultural Sciences (IAS) of Viet Nam and Unité d’Economie Générale, Faculté Universitaire des Sciences Agronomiques de Gembloux, Belgique, for a separate study of rice marketing channels in southern Viet Nam. In our estimates, we use data covering 149 farms from eight villages in three Mekong River Delta provinces. Because of nonreporting of some villages and to a lesser extent farm attrition from the survey, the sample size varies over time. Sampled villages represent a range of agroecological and production situations. The paper presents a model that combines spatial factors in a neoclassical production framework to examine changes in land use and agricultural technology that led to the increased output. Estimable forms of the production, land use, and revenue functions implied by the model are derived. Econometric models make use of panel data estimation procedures that control for the effect of unobserved variables. Estimations on single years of the survey use instrumental variable and system of equation estimators to correct for endogeneity bias in estimates of the effect of variables that are simultaneously determined with the outcomes of interest (e.g., cropping intensity, choice, and production level). The paper concludes by discussing estimation results. A simulation model is developed to highlight policy implications of findings. II. DESCRIPTION OF THE STUDY AREA The study area was characterizied according to the following biophysical variables: location of rice producers and accessibility of markets, soils, rainfall and temperatures, and seasonal flooding and saltwater intrusion on farmland. Figure 1 in the Appendix superimposes land use as reported by farms in the eight surveyed villages on a land use map for the Mekong Delta (circa 1996). The figure indicates the high correspondence between farm land use captured from remote sensing presented on the map and that reported by farms completing the longitudinal survey. The map also describes how land use in the surveyed villages changed over time. Beginning in 1988 with the adoption of Resolution 10 by the Politburo, Viet Nam undertook an ambitious program of decollectivizing its agriculture and liberalizing agricultural markets. Resolution 10 established farm households as autonomous economic entities in rural areas, and farms were permitted to own capital and land. Land formerly held in agricultural cooperatives was assigned to individual farms under long-term lease agreements. Because agricultural cooperatives functioned largely as a legal formality in the Mekong River Delta region where household farms were the ex facto productive unit earlier, the effect of Resolution 10 in that region was lesser here than in other regions of Viet Nam. In 1993, the Seventh Party Congress adopted Resolution 5 and the Road to Industrialization, which strengthen earlier reforms and adopted measures to promote rural industry and migration of workers out of employment in traditional agriculture. Investments in technology transfer (particularly in dissemination of higher yielding varieties) and water management infrastructure 9 analysis also guides the selection of variables and our expectations regarding their signs, but these are not reviewed in the interest of brevity. Different sets of right hand side variables are employed in estimates, depending upon the relevance of variables to the left hand side variable. In some estimates, the number of right hand side variables had to be reduced in order for the estimator to solve. These difficulties resulted from missing data and the relatively small sample size of the panel survey. Estimates use both cross-section- and panel data-based estimation procedures. Panel data estimation procedures provide more robust estimates because they can account for the effect of unobserved variables and have the potential to measure more precisely the effect of changes in explanatory variables. The empirical analysis also uses cross-sectional data-based estimators for two reasons. Panel data estimators cannot accommodate the use of time invariant right hand side variables in estimation equations, and many of the right hand side variables of interest were invariant or observed only a single time during the years of the survey. In the estimates, cropping patterns and land uses are defined by cardinal rankings (e.g., monocropping, double cropping) and according to the type of crop cultivated. Crops are divided into broad categories: (i) rice; (ii) upland row crops (e.g., sugarcane, potato, vegetables); and (iii) fruit trees or perennial fruit crops (e.g., dragon fruit) or trees maintained by farms for wood (e.g., eucalyptus). In order to apply panel data estimators, it is necessary to define cropping patterns and land use intensity as binary outcomes. Table 2 reports the results of three estimations that used a random effects probit estimation procedure: (i), farm cultivation of nonrice crops, (ii) farm cultivation of fruit trees or other perennial crops on its land, and (iii) cultivation of two or three rice crops per year. Because household-specific error terms are included in the models, the number of right hand side variables that could be considered in panel estimates was limited. The variables considered are: the on-farm land-labor ratio (acres per full-time equivalent family worker), age of the head of household, rice variety cultivated, and farm investment in dikes or land leveling. It is expected that households with lower land-labor ratios are more likely to farm land more intensively. Older farm operators and farmers with lower levels of educational attainment are expected to be more traditional and hesitant to adopt new technologies. The rice variety planted by farms clearly influences the feasible cropping intensity. Dummy variables are used to define farms growing short-duration, modern varieties and medium- or long-duration varieties. Because rice variety choice is endogenous with the choice of cropping pattern, estimates are open to endogeneity bias under the present specification. Unfortunately, data needed for suitable estimation procedures to control for endogeneity could not be identified. The parameter Rho indicates the significance of farm specific error estimates. The three models were each highly statistically significant. Several measures of the overall performance of the models in explaining land use are shown at the bottom of Table 2. Psuedo- R2measures vary between 44.7 and 6.4 percent across measures and models. Lastly, the table reports the share of land use categories correctly predicted by each model, and the distribution of actual versus predicted land use. This shows all three models performed well, predicting farm’s land use decisions correctly. Section IV Estimation Strategy and Results ERD Working Paper No. 16 THE ROLE OF INFRASTRUCTURE IN LAND-USE DYNAMICS AND RICE PRODUCTION IN VIET NAM’S MEKONG RIVER DELTA 10 The estimates of whether the farm cultivated a nonrice crop show that the land-labor ratio and the head of household’s age both had statistically significant negative affects on the probability that the farm cultivated a crop besides rice. Use of medium- or long-duration rice varieties and farm investments in water management infrastructure were found to increase significantly the likelihood of farm cultivation of a nonrice crop.1The estimated marginal effect of a one percent Table 2. Summary of Estimates of Land Use (Panel Data Estimators) LHS/Dependent Variable Cultivates Farm cultiv. Farm triple nonrice fruit/other crops rice Estimation coefficient crop(s) trees (vs. 2 rice) (Standard error of estimate) 1994-97 1994-97 1994-97 (N=436) (N=436) (N=354) Land-labor ratio on farm -1.058 *0.056 -0.365 *** (hectares per HH laborer) 0.578 0.591 0.153 Age of the household head -0.050 *** 0.014 ** -0.003 0.014 0.007 0.002 Cultivated short-duration -0.131 -0.018 0.088 modern varieties of rice 0.670 0.452 0.156 Cultivated medium- or long- 2.221 *** -0.334 0.846 *** duration modern rice varieties 0.775 0.364 0.172 Farm invested in land leveling or -0.116 -0.902 ** 0.068 other soil improvement 0.467 0.450 0.148 Farm invested in dike construct- 0.646 *-0.769 *0.080 ion or other water management 0.381 0.449 0.149 Rho 0.787 *** 0.970 *** 0.812 *** 0.225 0.081 0.125 Goodness of fit diagnostics: Pseudo R2: Cragg-Uhler 0.130 0.447 0.117 Maddela 0.064 0.317 0.079 McFadden 0.098 0.309 0.073 Likelihood ratio (X2) test 28.693 *** 166.549 *** 35.846 *** [degrees of freedom] 1 1 1 Pct. correctly predicted 0.842 0.672 0.624 Actual/Predicted 0 1 tot. 0 1 tot. 0 1 tot. 0 342 17 359 107 126 233 136 39 175 1 52 25 77 66 137 203 94 85 179 total 394 42 436 173 263 436 230 124 354 Notes: ***estimated coefficient is statistically significant at a 99% confidence level ** estimated coefficient is statistically significant at a 95% confidence level *estimated coefficient is statistically significant at a 90% confidence level Estimates used the random effects probit estimator for panel data. 1The random effects probit estimator is nonlinear, so estimation coefficients cannot be interpreted directly. The marginal effect of a change in a right hand side variable on the probability that a farm chose a particular land use at the mean values of the right hand side variables must be estimated using an approximation algorithm (see Greene 2000). 11 increase in the land-labor ratio of farms is a reduction of 4.0 percent in the likelihood that the farm cultivated more than a single crop per year. An increase of ten years in the age of the household head was associated with only a 0.2 percent decrease in the likelihood the farm cultivated a nonrice crop. Farm use of medium- or long-duration modern rice was associated with an 8.4 percent increase in the likelihood the farm grew a crop besides rice. The signs of the estimation coefficients are consistent with the expected signs outline in the previous section of the report. Farm-level investments in land leveling and water management were estimated to have a statistically significant effect on the likelihood that the farm cultivated tree crops, while older farm operators where significantly more likely to cultivate tree crops. Farms that invested in land leveling or other soil improvement or in water management infrastructure were, respectively, 35.9 and 30.6 percent less likely to cultivate a tree crop. The negative effect of the investments to improve the farm on tree crop cultivation is consistent with the understanding that such investments act as substitute responses to tree crop cultivation in addressing water scarcity and poor soil quality. Farms with a large amount of land per family worker were significantly less likely to cultivate three rice crops. A one percent increase in the land-to-labor ratio was associated with a 14.6 percent decrease in the likelihood of triple cropping. Farm use of medium- or long-duration varieties of rice was also found to have a positive statistically significant effect on the likelihood of triple cropping, although—surprisinging—use of short-duration varieties did not. Farms planting medium- or long-duration varieties were 33.7 percent more likely to grow three crops of rice a year. To summarize the discussion of Table 2, across these estimates it was found that farm size, particularly the relative abundance or scarcity of family agricultural labor in relation to the land operated by the farm, plays an important role in driving farm land use as expected. Farms with scarce labor relative to their farm size are less likely to cultivate land intensively. The choice of rice variety and corresponding crop maturation period of chosen varieties is closely related to broader land use choices of farms. Finally, investments in farm- or plot-level improvements in water management were also clearly linked to land use choices. One of the benefits of dike construction appears to be the opportunities it creates for farms to cultivate nonrice crops. In the absence of such investments, farms appeared to adopt land use options (i.e., fruit trees and other perennial crops) with greater immunity to the effects of poor water management. Lastly, the statistical significance of the estimation parameter Rho suggests that unobserved farm characteristics significantly influence land use choices, which underscores the complexity and idiosyncrasy of the land use choices of farms. Measures of market accessibility and variables characterizing biophysical conditions in the surveyed villages used in estimates were fixed over time or observed at only a single point in time. This makes it impossible to examine the principal hypotheses of the model related to these variables using the panel estimators. Instead, cross-sectional estimates of cropping patterns and rice cropping intensity are used to estimate the effect of time invariant regressors. Rice cropping intensity is a categorical variable where the categories have a natural ordering, so an ordered probit estimator is used. Section IV Estimation Strategy and Results ERD Working Paper No. 16 THE ROLE OF INFRASTRUCTURE IN LAND-USE DYNAMICS AND RICE PRODUCTION IN VIET NAM’S MEKONG RIVER DELTA 12 Rice cropping intensity estimates are significant overall in each of the four years, according to the goodness of fit measures reported on Table 3. Variables of particular interest in estimates are the measures of the distance between farming villages and the average travel time to all local markets, and the distance between homesteads and the plot or plots cultivated. The greater these distances, the lower the likely rice cropping intensity to be adopted by the farm. Estimation results Table 3. Summary of Estimates of Rice-cropping Intensity Rice- Rice- Rice- Rice- cropping cropping cropping cropping Left-hand side/dependent intensity intensity intensity intensity variable estimation coefficient in 1994ain 1995ain 1996ain 1997a (estimated standard error) (N = 60) (N = 114) (N = 114) (N = 77) Constant 7.2873 –17.884*** 16.599*** –34.870*** (5.5589) (4.217) (4.678) (11.088) Average distance between 0.0011 –0.131 –0.062 –0.025 homestead and plot or plots (0.1183) (0.138) (0.139) (0.185) Average travel time to all 0.0323 –0.041 0.020*** –0.060*** accessible local markets (0.0333) (0.011) (0.007) (0.027) Land-labor ratio on farm 0.0308 0.689 0.356 –0.100 (hectares/household laborer) (0.9918) (0.521) (0.464) (0.834) Years since family settled in 0.0064 –0.007 –0.003 –0.012 current place of residence (0.0110) (0.007) (0.007) (0.014) Maximum educational attainment 0.0020 0.311 0.071 0.526 of any family member (0.5324) (0.296) (0.294) (0.461) Whether farm served by good- 2.2076 2.053*** 4.266*** 4.704*** quality water control system (1.5133) (0.466) (0.843) (1.668) Annual precipitation at –0.0091 0.013*** –0.011*** 0.025*** locality where farm is located (0.0056) (0.003) (0.003) (0.008) Mu (1) 0.3267** 2.208*** 1.801*** 4.765** (0.1487) (0.330) (0.309) (2.182) Goodness of fit diagnostics: Pseudo R2: Cragg-Uhler 0.432 0.633 0.521 0.768 Maddela 0.361 0.555 0.460 0.654 McFadden 0.248 0.386 0.288 0.556 Likelihood ratio (X2) test 26.862*** 92.256*** 70.189*** 81.707*** (degrees of freedom) 7 7 7 7 % correctly predicted 0.800 0.719 0.632 0.805 Actual/predicted 0 1 2 Tot. 0 1 2 Tot. 0 1 2 Tot. 0 1 2 Tot. 0 37 0 0 37 21 12 0 33 18 13 0 31 6 6 0 12 16017 445655 5351151 238545 2 5 0 11 16 0 10 16 26 0 13 19 32 0 2 18 20 Total 48 0 12 60 25 67 30 114 23 61 30 114 8 46 23 77 aModel estimated using the ordered probit estimator. *** = estimated coefficient statistically significant at 99% confidence level, ** = estimated coefficient statistically significant at 95% confidence level, * = estimated coefficient statistically significant at 90% confidence level. 13 generally support the model’s hypotheses. Greater distances between farms and markets were associated with a reduced probability of intensive rice cultivation by the farm in 1995 and 1997, and estimated parameters were highly statistically significant. According to 1995 estimation results, a ten minute increase in the average travel time between the farm and available local markets was associated with 14 and 21 percent decreases in the probability of cultivating two and three crops during the year, respectively. The distance between farms and local markets in 1996 had a positive and statistically significant effect on rice cropping intensity. This result appears to be related to the heavy rains and the sample of villages surveyed that year. The distance between plots and homesteads had a negative, but not statistically significant effect on rice cropping intensity in 1995 through 1997. The availability of low-saline irrigation water to farms had a positive and statistically significant effect on the intensity of land use in all estimates. The magnitude of the effect of highquality irrigation on cropping intensity was much greater than the effects of other explanatory variables included in the model. Rainfall levels had mixed effects on the cropping intensity of surveyed farms. In years with normal to high rainfall, increased rain was associated with increased cropping intensity. Rains in 1996 were particularly heavy and higher rainfall in that year was associated with significantly reduced levels of cropping intensity among surveyed farms likely due to flooding problems associated with the heavy rains. Results show rice variety selection was clearly linked to cropping intensity, with the adoption of modern, short-duration rice varieties enabling more intensive rice cultivation by farm. Farm-level investments in land leveling or dike construction increased the likelihood that farms adopted intensive rice agriculture. Other variables such as the level of education in the household, the age of the household head, or the farming experience of the family did not have consistent statistically significant effects. Rice production estimates explained most of the observed variation in the levels of rice output across surveyed farms. Results of production function estimates, which are used in the simulation model discussed next, are summarized on Table 4. Adjusted R2coefficient estimates across the production models ranged between 0.76 and 0.89. All four models were highly statistically significant overall. The cropping intensity had consistent and statistically significant effect on annual production levels in all estimates. Monocropping was associated with significantly lower levels of output and triple-cropping was associated with significantly higher output levels compared to double-cropping. The land area cultivated and the amount of rice seed used were also associated with significantly higher levels of output in all estimates. The amount of hired labor applied on the farm had a positive and statistically significant effect on output in all the estimates except the 1994 cross-sectional estimate. The level of fertilizer applied on the farm had a positive and statistically significant effect on rice output in 1996 and 1997. The amount of family labor applied on farm was difficult to measure accurately from available data, but had a negative and significant effect on rice output in 1994 and a positive and significant effect in 1995. Pesticide application had a positive and statistically significant effect on output only in 1994. The signs of these estimated coefficients all conform to expectations. The one exception involved the use of modern varieties, which had inconsistent effects on rice production across estimates. It had statistically significant Section IV Estimation Strategy and Results ERD Working Paper No. 16 THE ROLE OF INFRASTRUCTURE IN LAND-USE DYNAMICS AND RICE PRODUCTION IN VIET NAM’S MEKONG RIVER DELTA 14 negative effects in estimates carried out using data from 1995 and 1997 and a positive effect in 1996. One explanation for this is that the variable was imprecisely defined due to aggregation across many distinct varieties. A second reason is that due to collinearity with rice cropping intensities, the principal effect of modern variety use seems to have been to enable farms to pursue more intensive rice production. Considered together, the various estimates provide a clear indication of the factors driving farm land use, production, and marketing decisions. Table 4. Rice Production Estimates (cross sectional estimators) Rice Rice Rice Rice Left-hand side/dependent production production production production variable estimation coefficient in 1995ain 1996ain 1997ain 1995-97b (estimated standard error) (N = 117) (N = 134) (N = 121) (N = 372) Constant(s) 7.493*** 4.642*** 6.295*** N.R.*** 0.322 0.765 0.350 N.R. Single-cropped rice -1.281*** -3.479*** -1.947*** -0.416 0.253 0.469 0.271 0.453 Triple-cropped rice 0.608*** 1.570*** 0.972*** -0.015 0.132 0.257 0.130 0.226 Largest area planted to rice 0.789*** 0.469*** 0.558*** 1.090*** in any season 0.075 0.194 0.070 0.254 Total household expenditure 0.067 0.483*** 0.098*** 0.210*** on hired labor 0.081 0.077 0.029 0.052 Imputed value of family -0.194** 0.429*** 0.102 0.176 labor applied on farm 0.097 0.165 0.098 0.127 Expenditure on fertilizer 0.029 0.205*0.151** 0.033 0.060 0.108 0.068 0.072 Expenditure on pesticides 0.096*** 0.010 0.016 -0.320 or herbicides 0.038 0.075 0.034 0.048 Average quantity of rice seed 0.622*** 0.735*** 0.747*** -0.419** used per season cultivated 0.125 0.236 0.101 0.192 Use of any modern variety -0.123*0.226*-0.119** 0.295 of rice seed 0.067 0.134 0.058 0.092 Year 1995 – – – -0.010 – – – 0.069 Year 1996 – – – -0.103*** – – – 0.063 Goodness of fit diagnostics: Adjusted R20.892 0.755 0.888 0.835 F-ratio 107.710*** 46.450*** 106.220*** 12.210*** [degrees of freedom] [9, 107] [9, 124] [9, 111] [167, 204] Likelihood ratio (X2) test 270.097*** 197.658*** 273.827*** 891.838*** [degrees of freedom] [9] [9] [9] [167] Notes: *** estimated coefficient statistically significant at a 99% confidence level ** estimated coefficient statistically significant at a 95% confidence level *estimated coefficient statistically significant at a 90% confidence level aEstimated in logarithms using the ordinary least squares estimator. bEstimated in logarithms using the fixed effects estimator for panel data. The results of the Hausman test (46.950*** with 11 d.f.) supported use of the fixed effects specificat. N.R. means household-specific intercepts are not reported. 15 V. SIMULATION MODEL FOR EVALUATION OF INVESTMENTS The implications of model estimates for evaluating the effect of development of different types of infrastructure can be better understood by generating a simulation model using estimation parameters. The results of a simulation model derived from empirical estimates are summarized in Tables 5 and 6. Table 5 shows the distribution of rice cropping intensities among surveyed farms. The actual distribution of farms in each of the four years of the survey is shown, along with the projected distribution under alternative scenarios. One scenario involves improvements in travel networks between surveyed villages and local markets. The second considers the effect of land transport improvements or land consolidation that brings homesteads and farm plots closer. The third contemplates extension of water control infrastructure to an additional 10 percent of the Table 5. Simulation of Effects of Investments on Distribution of Farm Rice-cropping Intensity Simulated distribution of farms with improvements in Transportation system: Land consolidation: Water control Rice Actual Reducing travel Reducing distance infrastructure: Cropping distribution to market from home to plot Increasing area intensity of farms by 10 minutes by 1 kilometer covered by 10% 1994 1995 1996 1997 1994 1995 1996 1997 1994 1995 1996 1997 1994 1995 1996 1997 Monocropping 37 33 31 12 36 21 26 12 40 29 33 12 15 13 0 12 Double cropping 7 55 51 45 7 62 51 45 6 57 51 44 8 67 52 38 Triple cropping 16 26 32 20 17 31 37 20 14 28 30 21 37 34 62 27 Table 6. Simulation of Effects of Investments on Rice Production among Surveyed Farms (tons) Predicted production Predicted production Predicted production for travel for distance from for better water to market home to plot management Actual productiona(–10 min)areduced by 1 kmaextension +10%a 1994 1995 1996 1997 1994 1995 1996 1997 1994 1995 1996 1997 1994 1995 1996 1997 1 × rice farm production 25 22 11 7 24 14 10 7 27 20 12 7 10 9 0 7 2 × rice farm production 5 41 24 28 5 47 24 28 5 43 24 28 6 50 24 24 3 × rice farm production 13 22 26 16 14 26 30 16 12 23 25 17 31 28 51 21 Total rice production 44 85 61 51 44 86 64 51 43 86 60 51 47 87 75 52 % change in total production 0.4 3.0 5.2 0.0 –0.9 1.0 –2.1 0.4 7.6 4.9 31.5 2.6 aColumns may not sum to total rice production due to rounding error. Section V Simulation Model for Evaluation of Investments ERD Working Paper No. 16 THE ROLE OF INFRASTRUCTURE IN LAND-USE DYNAMICS AND RICE PRODUCTION IN VIET NAM’S MEKONG RIVER DELTA 16 surveyed farms. Using results of production function estimates, the implied changes in the share of farms that double- or triple-crop rice can be applied to calculate an implied increase in aggregate rice output across farms. The production estimates provide a measure of the average change in annual rice yield associated with mono, double, or triple cropping of rice. Table 6 details the changes in total rice production from the scenarios. The simulation model shows a large effect of investments in irrigation extension on rice production, and more moderate effects obtained from improvements in the transportation system or land consolidation. Incorporating the estimates obtained in this research with other linear programming or general simulation models would be an important extension of this research. VI. CONCLUSION These results generally support the hypothesis that the time and direct cost of transporting inputs and outputs between rural homesteads, farm plots, and markets influence the land use and production decisions of farming households. Estimation results confirm our expectation that greater transport distances reduce the cropping intensity and make the cultivation of nonrice crops less likely. However, results suggest the quality of the water management infrastructure is far more important in determining land use than transport infrastructure. The magnitude of the effect of having high quality water management infrastructure dwarfed the effect of other variables. Other variables including the use of modern seed varieties, the age of the farm operator, the land-to-labor ratio of the farm, and rainfall influenced farm land use as predicted. Results suggest that investments in water management offer more promise in improving farm land use options and increasing rice production than transport infrastructure investments in the Delta. However, information on the relative costs of extending road and water management infrastructure is necessary before it would be appropriate to offer policy conclusions in this regard. This study relied on existing sources of data originally collected for a cost-price accounting study, and as a result encountered data constraints in analyses. 17 References Chomitz, K., and D. Gray, 1995. Roads, Lands, Markets, and Deforestation. Policy Research Working Paper No. 1444, The World Bank, Washington, D.C. Deichmann, U., 1998. “Spatial Scale and Resolution in the Analysis of Socioeconomic and Demographic Data.” Paper presented at the Workshop on Scaling Methodologies in Ecoregional Approaches for Natural Resources Management, Ho Chi Minh City, Viet Nam, 22-24 June. de Janvry, A., M. Fafchamps, and E. Sadoulet, 1991. “Peasant Household Behavior with Missing Markets: Some Paradoxes Explained.” Economics Journal 101:1400-17. Edmonds C.M., S. P. Kam, H. C. Viet, and L. Villano, 2001. “Land-use Dynamics and Changes in Rice Production in the Mekong Delta in the 1990s: An Econometric Analysis using Longitudinal Household Survey Data. In S. Peng and B. Hardy, eds., Rice Research for Food Security and Poverty Alleviation. Proceedings of the International Rice Research Conference, 31 March-3 April 2000, International Rice Research Institute, Los Baños, Philippines. Greene, W.H., 2000. Econometric Analysis.4 th ed. New Jersey:Prentice Hall. Hoanh C.T., R. Roetter, D.M. Jansen, P.K. Aggarwal, F.P. Lansigan, N.X. Lai, I.A. Bakar, and A. Tawang, 1998. “Generalizing SysNet Methodologies for Land Use Planning at the Subnational Level.” In R. Roetter, C.T. Hoanh, N.V. Luat, M.K. Van Ittersum, and H.H. Van Laar, eds., Exchange of Methodologies in Land Use Planning. SysNet Research Paper Series No. 1, International Rice Research InstituteLos Baños, Philippines. International Food Policy Research Institute (IFPRI), 1996. “Rice Monitoring and Policy Options Study: Final Report.” Paper prepared for TA No. 2224-VIE. Asian Development Bank, Manila. Institute of Agricultural Sciences, 1997. “Competitiveness of Rice Channel in Mekong Region: Study Report 1994, 1995, 1996.” Research report of the Competitiveness of Rice Production in the Mekong River Delta Project, Institute of Agricultural Sciences, Ho Chi Minh City, Viet Nam. ————, 1998. “Annual Report: Comparative Analysis of Economic Efficiency in Rice Production 1994-1997.” Research report of the Competitiveness of Rice Production in the Mekong River Delta Project, Institute of Agricultural Sciences, Ho Chi Minh City, Viet Nam. Government Statistical Office, 1998. Socio-economic Statistical Data of 61 Provinces and Cities in Vietnam. Hanoi: Statistical Publishing House. Kam, S.P., V.Q. Minh, T.P. Tuong, C.T. Hoanh, S.C. Liew, and P. Chen, 1998. “Remote Sensing and GIS Approaches to Studying Changes in Rice Cropping Systems in the Mekong River Delta, Vietnam.” Paper presented at the GISDECO 1998 Meeting, Pretoria, South Africa. Minh, V.Q., 1995. Use of Soil and Agrohydrological Characteristics in Developing Technology Extrapolation Methodology: A Case Study of the Mekong Delta, Vietnam. M.S. thesis. University of the Philippines Los Baños, Los Baños, Laguna, Philippines. Minot, N., and F. Goletti, 1998. “Export Liberalization and Household Welfare: The Case of Rice in Vietnam.” American Journal of Agricultural Economics 80:738-49. References ERD Working Paper No. 16 THE ROLE OF INFRASTRUCTURE IN LAND-USE DYNAMICS AND RICE PRODUCTION IN VIET NAM’S MEKONG RIVER DELTA 18 Pingali, P.L., and V. T. Xuan, 1990. Vietnam: Decollectivization and Rice Productivity Growth. IRRI Social Sciences Division Papers No. 89-16, International Rice Research Institute, Los Baños, Philippines. Thanh, D.N., L. T. Duong, and F. B. Gascon, 1995. “Changes from Deepwater to Irrigated Ecosystem in Mekong Delta of Vietnam: Impact on Productivity and Efficiency in Resource Use.” Journal of Asian Farming Systems Association 1:187-214. Tuong, T.P., C. T. Hoanh, and N. T. Khiem, 1991. “Agro-hydrological Factors as Land Qualities in Land Evaluation for Rice Cropping Patterns in the Mekong Delta of Vietnam.” In P. Deturck and F. M. Ponnamperuma, eds., International Rice Production on Acid Soils in the Tropics. Institute of Fundamental Studies, Kandy, Sri Lanka. United Nations Statistics Division, 1997. Accessibility Indicators in GIS. Report of the Department for Economic and Social Information and Policy Analysis. Geneva. Ut, T. T., and M. Hossain, 1999. “Effects of Improved Technology on Rice Production and Impact on Income Distribution and Poverty Alleviation: A Case Study of Vietnam.” Paper presented at the International Workshop on the Role of Agricultural Research on Poverty Alleviation, San José, Costa Rica, 12-15 September. 25 1. Informal Finance: Some Findings from Asia Prabhu Ghate et. al., 1992 $15.00 (paperback) 2. Mongolia: A Centrally Planned Economy in Transition Asian Development Bank, 1992 $15.00 (paperback) 3. Rural Poverty in Asia, Priority Issues and Policy Options Edited by M.G. Quibria, 1994 $25.00 (paperback) 4. Growth Triangles in Asia: A New Approach to Regional Economic Cooperation Edited by Myo Thant, Min Tang, and Hiroshi Kakazu 1st ed., 1994 $36.00 (hardbound) Revised ed., 1998 $55.00 (hardbound) 5. Urban Poverty in Asia: A Survey of Critical Issues Edited by Ernesto Pernia, 1994 $18.00 (paperback) 6. Critical Issues in Asian Development: Theories, Experiences, and Policies Edited by M.G. Quibria, 1995 $15.00 (paperback) $36.00 (hardbound) 7. From Centrally Planned to Market Economies: The Asian Approach Edited by Pradumna B. Rana and Naved Hamid, 1995 Vol. 1: Overview $36.00 (hardbound) Vol. 2: People’s Republic of China and Mongolia $50.00 (hardbound) Vol. 3: Lao PDR, Myanmar, and Viet Nam $50.00 (hardbound) 8. Financial Sector Development in Asia Edited by Shahid N. Zahid, 1995 $50.00 (hardbound) 9. Financial Sector Development in Asia: Country Studies Edited by Shahid N. Zahid, 1995 $55.00 (hardbound) 10. Fiscal Management and Economic Reform in the People’s Republic of China Christine P.W. Wong, Christopher Heady, and Wing T. Woo, 1995 $15.00 (paperback) 11. Current Issues in Economic Development: An Asian Perspective Edited by M.G. Quibria and J. Malcolm Dowling, 1996 $50.00 (hardbound) 12. The Bangladesh Economy in Transition Edited by M.G. Quibria, 1997 $20.00 (hardbound) 13. The Global Trading System and Developing Asia Edited by Arvind Panagariya, M.G. Quibria, and Narhari Rao, 1997 $55.00 (hardbound) 14. Rising to the Challenge in Asia: A Study of Financial Markets Asian Development Bank, 1999 Vol. 1 $20.00 (paperback) Vol. 2 $15.00 (paperback) Vol. 3 $25.00 (paperback) Vols. 4-12 $20.00 (paperback) SPECIAL STUDIES, OUP (SS,OUP) (Co-published with Oxford University Press; Available commercially through Oxford University Press Offices, Associated Companies, and Agents) SERIALS (Co-published with Oxford University Press; Available commercially through Oxford University Press Offices, Associated Companies, and Agents) 1. Asian Development Outlook (ADO; annual) $36.00 (paperback) 2. Key Indicators of Developing Asian and Pacific Countries (KI; annual) $35.00 (paperback) JOURNAL (Published in-house; Available commercially through ADB Office of External Relations) 1. Asian Development Review (ADR; semiannual) $5.00 per issue; $8.00 per year (2 issues)